Interactions Between the Kleptoplastidic Dinoflagellate Shimiella gracilenta and Several Common Heterotrophic Protists

The newly described dinoflagellate, Shimiella gracilenta, is known to survive for approximately 1 month on the plastids of ingested prey cells during starvation, indicating kleptoplastidy. To understand the population dynamics of this dinoflagellate in marine planktonic food webs, its growth and mortality rate due to predation should be assessed. Thus, we investigated the feeding occurrence of eight common heterotrophic protists on S. gracilenta. We also determined the growth and ingestion rates of Oxyrrhis marina and the naked ciliate, Rimostrombidium sp. on S. gracilenta as a function of the prey concentration. The common heterotrophic dinoflagellates (HTDs) Gyrodinium dominans, O. marina, and Pfiesteria piscicida and a naked ciliate Rimostrombidium sp. were able to feed on S. gracilenta; whereas the HTDs Aduncodinium glandula, Gyrodinium jinhaense, Oblea rotunda, and Polykrikos kofoidii were not. Shimiella gracilenta supported positive growth of O. marina and Rimostrombidium sp. but did not support that of G. dominans and P. piscicida. With increasing prey concentrations, the growth and ingestion rates of O. marina and Rimostrombidium sp. on S. gracilenta increased and became saturated. The maximum growth rates of O. marina and Rimostrombidium sp. on S. gracilenta were 0.645 and 0.903 day−1, respectively. Furthermore, the maximum ingestion rates of O. marina and Rimostrombidium sp. on S. gracilenta were 0.11 ng C predator day−1 (1.6 cells predator−1 day−1) and 35 ng C predator day−1 (500 cells predator−1 day−1), respectively. The maximum ingestion rate of O. marina on S. gracilenta was lower than that on any other algal prey reported to date, although its maximum growth rate was moderate. In conclusion, S. gracilenta had only a few common heterotrophic protist predators but could support moderate growth rates of the predators. Thus, S. gracilenta may not be a common prey species for diverse heterotrophic protists but may be a suitable prey for a few heterotrophic protists.

The dinoflagellate Shimiella gracilenta SGJH1904 that was isolated from the waters of Jinhae Bay, Korea, has been recently described . This dinoflagellate is known to survive without added prey for approximately 1 month on chloroplasts obtained from the prey, a cryptophyte called Teleaulax amphioxeia, indicating its "kleptoplastidy" . To understand the ecological roles of a newly described heterotrophic, kleptoplastic, or mixotrophic dinoflagellate, its prey and predators should be first identified (Potvin et al., 2013;Jeong et al., 2018a,b;Kang et al., 2019). Based on the results of the studies on the prey and predators, the transfer of elements from prey (i.e., the target dinoflagellate) to the predators in cycling of elements and food webs can be understood (Jeong et al., 2010b). To explore the ecological role of S. gracilenta as a predator, the kind of prey that S. gracilenta is able to feed on, growth and ingestion rates of S. gracilenta on prey, and grazing impact by S. gracilenta on prey populations should be assessed. On the other hand, to explore the ecological role of S. gracilenta as prey, the kind of predators that are able to feed on S. gracilenta, growth and ingestion rates of the predator on S. gracilenta, and predation impact by the predator on S. gracilenta populations should be assessed. Furthermore, to understand the population dynamics of the dinoflagellate, its growth rate (k) and mortality rate due to predation (g) under a given condition should be determined . To determine the mortality rate due to predation, first, predators that are able to feed on the dinoflagellate prey should be identified (Berge et al., 2008;Lim et al., 2015). Second, the ingestion and growth rates of the predators on the dinoflagellate prey should be determined Anderson and Menden-Deuer, 2017). The ingestion rate of a predator on dinoflagellate prey is proportional to the clearance rate that, in turn, is proportional to g in bottle incubation experiments Kang et al., 2011). When data on the abundances of the target dinoflagellate and its co-occurring predator species in marine environments are available, the mortality rate due to predation by multiple predators can be estimated (Yoo et al., 2013a;Lim et al., 2017). Thus, revealing predators that are able to feed on S. gracilenta and determining the ingestion and growth rates of the predators on S. gracilenta are needed.
Typically, to identify predators of specific dinoflagellates, mixotrophic and/or heterotrophic organisms, which are abundant during or after the bloom of the target dinoflagellate (Yoo et al., 2013a;Lim et al., 2017;Eom et al., 2021), are selected as the potential predators. Alternatively, predators of the target dinoflagellate can be identified by observing the feeding behavior under a microscope after mixing the target dinoflagellate cells with potential predators commonly found in many marine environments (Yoo et al., 2010(Yoo et al., , 2015Potvin et al., 2013;Ok et al., 2017;Kim et al., 2019). The heterotrophic dinoflagellates (HTDs) Gyrodinium dominans, Oblea rotunda, Oxyrrhis marina, Pfiesteria piscicida, and Polykrikos kofoidii and the naked ciliate Rimostrombidium sp. are commonly present in the waters of many countries (e.g., Costello et al., 2001;Jakobsen et al., 2002;Rhodes et al., 2006;Rublee et al., 2006;Sherr and Sherr, 2007;Felder and Camp, 2009;Agatha, 2011;Watts et al., 2011;Kilias et al., 2013;McCarthy, 2013). They have also been regarded as potential heterotrophic protist predators on many mixotrophic or kleptoplastidic dinoflagellates Anderson, 1986, 1996;Jeong et al., 2001aJeong et al., , 2007aJeong et al., , 2017Jeong et al., , 2018aOk et al., 2017). The newly described HTD, Gyrodinium jinhaense, is commonly found in Korean coastal waters . The HTD Aduncodinium glandula has been used as a potential predator in several studies Jeong et al., 2018a). These heterotrophic protists have been reported to be present in the eastern South Sea of Korea, where S. gracilenta SGJH1904 was isolated (Jeong et al., 2006Kang et al., 2015Kang et al., , 2019Lim et al., 2017;Jung et al., 2021). Thus, S. gracilenta is likely to encounter these heterotrophic protists in natural environments. Moreover, these protists exhibit diverse sizes and shapes and adopt various feeding mechanisms, such as engulfment, peduncle, and pallium feeding.
In the present study, using the clonal cultures of S. gracilenta SGJH1904, we investigated the feeding occurrence of these potential heterotrophic protist predators on S. gracilenta and also determined the growth and ingestion rates of O. marina and Rimostrombidium sp. on S. gracilenta as a function of prey concentration. This was done because S. gracilenta supported the growth of only these two predators. Additionally, we compared the growth rate of O. marina feeding on S. gracilenta to that on other algal prey species. The results of this study can contribute to the understanding of interactions between S. gracilenta and common heterotrophic protists, as well as the understanding of ecological roles of S. gracilenta in marine planktonic food webs.

Preparation of Experimental Organisms
Shimiella gracilenta SGJH1904 cells were isolated from surface waters off the coast of Jinhae, Korea on April 6, 2019 . A clonal culture of S. gracilenta was established using two serial single cell isolations, and a culture of S. gracilenta was grown on T. amphioxeia (30,000-40,000 cells ml −1 ). Teleaulax amphioxeia was originally isolated from the coastal waters of Gomso Bay, Korea.
The HTDs A. glandula, G. dominans, G. jinhaense, O. rotunda, O. marina, and P. kofoidii, isolated from plankton samples collected from the coastal waters of Masan, Jeongok, Jinhae, Kunsan, and Jangheung during 2001-2019, were used in this study ( Table 1). The clonal culture of P. piscicida was obtained from the National Center for Marine Algae and Microbiota (NCMA) in the USA. To isolate and cultivate the naked ciliate Rimostrombidium sp., plankton samples were collected from the coastal waters of Saemankeum, Korea, in July 2020 ( Table 1).

Determination of Carbon Content and Cell Volume
The carbon contents of the potential predators and S. gracilenta were estimated based on cell volume (Menden-Deuer and Lessard, 2000). The cell volumes of A. glandula, G. jinhaense, G. dominans, O. rotunda, O. marina, P. kofoidii, P. piscicida, and Rimostrombidium sp. were estimated following the method described by Jang et al. (2016), Kang et al. (2020), Jeong et al. (2001bJeong et al. ( , 2007aJeong et al. ( , 2008, Ok et al. (2017), and Eom et al. (2021). The cell volume of S. gracilenta was also measured in the present study.

Determination of Feeding Capability
Experiment 1 was designed to examine whether the selected potential heterotrophic protists could feed on S. gracilenta after mixing the two species. Shimiella gracilenta cells were added to 42-ml polycarbonate (PC) bottles at a concentration of 15,000 cells ml −1 . Next, each of the potential predators was added. A predator control bottle (no prey) and a prey control bottle (no predator) were also set up for each experiment. The initial concentration of each target heterotrophic protist was set up, considering its carbon contents or the maximum cell abundances as described in Supplementary Table 1. The bottles were placed on a plankton wheel rotating at 0.9 rpm (0.00017 × g) and incubated at 20 • C under 20-µE m −2 s −1 illumination with cool white fluorescent light and a 14:10-h light-and-dark cycle.
After 1, 2, 6, 24, and 48 h of incubation, 5-ml aliquots were collected from each bottle using an autopipette and transferred into six-well plate chambers. When prey cells were not eaten by a target predator, >40 of the cells of the predator were observed under a dissecting microscope at a magnification of × 10-63 to determine the feeding occurrence of the potential predators on S. gracilenta. However, when a prey cell was eaten by a target predator, >15 cells of the predator were observed. The attack ratio (i.e., number of attempted captures relative to the number of physical contacts between a predator and prey) and capture success (i.e., number of prey ingested relative to the number of attempted captures) were determined by monitoring the behavior of the potential dinoflagellate predators in the presence of S. gracilenta. The ratios for Rimostrombidium sp. were not measured because this predator quickly jumped.
To explore the process of feeding, wells containing both a predator and S. gracilenta were placed under a light microscope (Zeiss-Axiovert 200M; Carl Zeiss Ltd., Göttingen, Germany), equipped with a video-analyzing system (Sony DXC-C33; Sony Co., Tokyo, Japan). The feeding process of the target species was recorded at a magnification of × 200-630. Cells of each target predator species and S. gracilenta were photographed on a confocal dish with cover glasses using a digital camera (Zeiss AxioCam 506; Carl Zeiss Ltd.), attached to a light and an epifluorescence microscope at a magnification of × 400-1,000.

Measurement of Growth and Ingestion Rates as a Function of Prey Concentration
In the preliminary test, S. gracilenta supported the positive growth rate of only two of the investigated potential predators, namely O. marina and Rimostrombidium sp. Thus, Experiments 2 and 3 were designed to determine the growth and ingestion rates of O. marina and Rimostrombidium sp. on S. gracilenta.
In Experiment 2, when prey cells were undetected under 1-ml Sedgewick-Rafter chamber (SRC), a dense culture of O. marina was transferred into a round 1-L PC bottle (Triforest R PC Centrifuge/Wide-mouth bottles) containing S. gracilenta (approximately 2,000 S. gracilenta cells ml −1 and 2,000 O. marina cells ml −1 ). The bottle was filled to the capacity with freshly filtered seawater to decrease the damage to the organisms, capped and placed on a plankton wheel rotating at 0.9 rpm (0.00017 × g). It was then incubated at 20 • C under an illumination of 20 µE m −2 s −1 and placed in a 14:10-h light-and-dark cycle. Triplicate 1-ml aliquots were removed from the bottle every day, and cells of S. gracilenta and O. marina were counted under a light microscope to determine complete consumption of S. gracilenta and cell concentration and residual growth of O. marina (i.e., no difference in the growth rates of O. marina with and without added prey). After subsampling, the bottle was refilled with freshly filtered seawater and placed on the plankton wheel once again. The predetermined volumes of O. marina and S. gracilenta were transferred to PC bottles using an autopipette (Supplementary Table 1). Triplicate 42ml PC experimental bottles (containing mixtures of predators and prey) and triplicate control bottles (containing only prey) were set up for each predator-prey combination. Triplicate control bottles containing only predators were also set up at a single predator concentration. Cultures of predator cells were passed through a 0.2-µm disposable syringe filter (DISMIC-25CS type, 25 mm; Advantec, Toyo Roshi Kaisha Ltd., Chiba, Japan), and then the filtrates were added to the prey control bottles as the same volume as that of the predator culture added to the experimental bottles for each predator-prey combination to ensure similar water conditions. All the bottles were then filled to capacity with freshly filtered seawater and capped. At the start of the experiment, a 5-ml aliquot was collected from each bottle, fixed in 5% Lugol's solution, and enumerated under a light microscope to determine the actual predator and prey abundances. The bottles were refilled to the capacity with freshly filtered seawater and placed on rotating wheels under the conditions described above. Dilution of the cultures by refilling the bottles was considered when calculating the growth and ingestion rates. After 48 h of incubation, a 10-ml aliquot was collected from each bottle and fixed in 5% Lugol's solution. The predator and prey abundance sampled at 0 and 48 h after setting up the experiment was determined by counting all (or over 200) cells using 1-ml SRC in triplicate.
In Experiment 3, when the prey cells were completely consumed, a dense culture of Rimostrombidium sp. was transferred to a 500-ml PC bottle containing S. gracilenta (approximately 600 S. gracilenta cells ml −1 and 20 Rimostrombidium sp. cells ml −1 ), and the bottle was filled with filtered seawater. The bottles were incubated under the same conditions as described above. Every 12 h of incubation, triplicate 1-ml aliquots were collected from the bottle, and the cell concentration and residual growth of Rimostrombidium sp. were determined under a light microscope. The initial concentration of Rimostrombidium sp. and S. gracilenta was established using an autopipette to deliver predetermined volumes of known cell concentrations to the bottles (Supplementary Table 1). The experiment was conducted using the methodology described above. After 24 h of incubation, a 10-ml aliquot was collected from each bottle, fixed in 5% Lugol's solution, and enumerated under an inverted microscope (BX51, Olympus, Japan). Triplicate 10-ml aliquots were collected from the bottles with the three lowest concentrations and all predator control bottles and enumerated in six-well plate chambers under a dissecting microscope (SZX2-ILLB, Olympus, Japan) without fixation.
The specific growth rate of each heterotrophic protist predator, µ (day −1 ), was calculated as follows: where µ max is the maximum growth rate (day −1 ); x is the prey concentration (cells ml −1 or ng C ml −1 ); x ′ is the threshold prey concentration (prey concentration where µ = 0); and K GR is the prey concentration sustaining 1/2 µ max . Data were iteratively fitted to the model using Delta Graph (IBM-SPSS Inc., Armonk, NY, USA).
The ingestion and clearance rates were calculated using the modified equations described by Frost (1972) and Heinbokel (1978). The incubation time for calculating the ingestion and clearance rates was the same as that for estimating the growth rate. The calculated ingestion rate (IR; cells predator −1 day −1 ) of each heterotrophic protist was fitted to the Michaelis-Menten equation as follows: where I max is the maximum ingestion rate (cells predator −1 day −1 or ng C predator −1 day −1 ); x is the prey concentration (cells ml −1 or ng C ml −1 ); and K IR is the prey concentration sustaining 1/2 I max .

Measurement of Swimming Speed
A culture of S. gracilenta with approximately 40,000 cells ml −1 growing on T. amphioxeia was transferred into a 1-L PC bottle. To determine the presence of any prey cells in the S. gracilenta culture, a 5 ml aliquot was removed from the bottle, fixed with 5% Lugol's solution, and examined under a light microscope. When no prey cells were detected, an  Kang et al. (2020); (2)  aliquot of S. gracilenta culture was transferred to a 38-ml cell culture flask (BD Biosciences, USA) and allowed to acclimate for 30 min to avoid the interference of water flow. A video camera was focused on a single frame on the cell culture flask under a stereomicroscope (SZX2-ILLK, Olympus, Japan) at 20 • C. Swimming of S. gracilenta cells was recorded at a magnification of × 25 using a video-analyzing system (SRD-1673 DN, Samsung, Korea) and a CCD camera (DXC-C33, SONY, Japan). The speed of all swimming cells viewed within 10-70 min was measured, and the average swimming speed (n = 30) was calculated based on the straight line of the cells during single-frame playback.

Statistical Analysis
Pearson's correlation analysis was used to examine the association among the variables [i.e., equivalent spherical diameter (ESD) of prey species, µ max and I max of O. marina feeding on dinoflagellate prey species, maximum swimming speed (MSS) of prey species, and µ max and I max of O. marina feeding on dinoflagellate prey species]. Statistical analyses were performed using SPSS ver. 25.0 (IBM-SPSS Inc., Armonk, NY, USA).

Feeding Occurrence by Heterotrophic Protists on Shimiella gracilenta
The HTDs O. marina, P. piscicida, and G. dominans, as well as the naked ciliate Rimostrombidium sp., were able to feed on S. gracilenta SGJH1904 (  Figures 1, 2). The attack ratios of G. dominans, P. piscicida, and O. marina to S. gracilenta were 6, 33, and 14%, respectively ( Table 2). Once these predators attacked S. gracilenta, they successfully fed on S. gracilenta (i.e., 100%). The HTDs A. glandula, G. jinhaense, O. rotunda, and P. kofoidii did not feed on S. gracilenta (Table 2; Supplementary Figure 3). Cells of A. glandula, G. jinhaense, O. rotunda, and P. kofoidii did not attack S. gracilenta cells. Furthermore, S. gracilenta cells were observed quickly jumping backward when A. glandula, G. jinhaense, O. rotunda, and P. kofoidii touched them. However, it was observed that O. marina spun very quickly around an S. gracilenta cell and engulfed the prey cell (Supplementary Video 1). Cells of Rimostrombidium sp. made feeding currents to draw prey cells and intercepted prey cells inside the feeding currents. Often, several cells of P. piscicida attacked an S. gracilenta cell together.

Growth Rates of Rimostrombidium sp. and Oxyrrhis marina Feeding on Shimiella gracilenta as a Function of Prey Concentration
With increasing mean prey concentrations, the specific growth rate of Rimostrombidium sp. feeding on S. gracilenta increased at mean prey concentrations ≤204 ng C ml −1 (2,919 cells ml −1 ), but this became saturated at higher mean prey concentrations ( Figure 1A). When the data were fitted to Equation (2), the calculated maximum growth rate (µ max ) of Rimostrombidium sp. feeding on S. gracilenta was 0.903 day −1 .
The specific growth rate of O. marina feeding on S. gracilenta rapidly increased with the increase of mean prey concentrations, with the highest growth rate at a mean prey concentration of 9.4 ng C ml −1 (134 cells ml −1 ; Figure 1B). However, this slightly decreased at higher mean prey concentrations. When the data were fitted to Equation (2), the calculated µ max of O. marina on S. gracilenta was 0.645 day −1 .

Ingestion Rates of Rimostrombidium sp. and Oxyrrhis marina Feeding on Shimiella gracilenta as a Function of Prey Concentration
With increasing mean prey concentrations, the ingestion rate of Rimostrombidium sp. feeding on S. gracilenta rapidly increased at mean prey concentrations ≤204 ng C ml −1 (2,919 cells ml −1 ) but slowly increased at higher mean prey concentrations (Figure 2A). When the data were fitted to Equation (3), the calculated maximum ingestion rate (I max ) of Rimostrombidium sp. feeding on S. gracilenta was 35 ng C predator −1 day −1 (500 cells predator −1 day −1 ).
With increasing mean prey concentrations, the ingestion rate of O. marina feeding on S. gracilenta increased at mean prey concentrations ≤62.3 ng C ml −1 (890 cells ml −1 ) but became almost saturated at higher mean prey concentrations ( Figure 2B). When the data were fitted to Equation (3), the calculated I max of O. marina on S. gracilenta was 0.11 ng C predator −1 day −1 (1.6 cells predator −1 day −1 ).

Feeding Ability of the Predators
Among the eight common heterotrophic protists tested in the present study, four species were able to feed on S. gracilenta SGJH1904, whereas the other four species did not feed at all. The feeding mechanism, behavior, size, and/or shape of protist predators, as well as the biochemistry, behavior, size, and/or shape of the prey, have been known to affect the feeding abilities of predators (Scharf et al., 2000;Strom et al., 2003;Evans and Wilson, 2008;Jeong et al., 2010b;Roberts et al., 2011). The engulfment feeders G. dominans, O. marina, and Rimostrombidium sp. ingested S. gracilenta, whereas other engulfment feeders, G. jinhaense and P. kofoidii, did not feed on the prey species. Furthermore, the peduncle feeder P. piscicida fed on S. gracilenta, whereas another peduncle feeder, A. glandula, did not feed on the prey species. Therefore, feeding mechanisms are likely not to affect the feeding occurrence of these heterotrophic protists on S. gracilenta. Meanwhile, among the protists G. dominans, P. piscicida, O. marina, and Rimostrombidium sp., which were able to feed on S. gracilenta, the sizes were larger than those of G. jinhaense, which was not able to feed on S. gracilenta but smaller than that of A. glandula, O. rotunda, and P. kofoidii, which were not able to feed on the prey (Table 2). Therefore, the predator size may not affect the feeding occurrence of these heterotrophic protists on S. gracilenta.
Although G. jinhaense and G. dominans are included in the same genus and have similar ESD values, G. dominans is known to feed on Mesodinium rubrum, which has jumping behavior, whereas G. jinhaense does not (Lee et al., 2014b;Kang et al., 2020). Thus, the rapid jumping behavior of S. gracilenta may be an effective anti-predation tool against G. jinhaense, but not against G. dominans. Moreover, the maximum swimming speed of G. dominans (440 µm s −1 ) is considerably greater than that of G. jinhaense (331 µm s −1 , Table 2). Therefore, the swimming speed of these two Gyrodinium species may affect this differential feeding occurrence on S. gracilenta.
Cells of P. kofoidii are known to deploy a nematocysttaeniocyst complex to capture a prey cell (Matsuoka et al., 2000;Tillmann and Hoppenrath, 2013). However, cells of P. kofoidii were not observed deploying a nematocyst-taeniocyst complex on the surface of an S. gracilenta cell. The minimum prey size that P. kofoidii was able to feed on is known to be approximately 10 µm (Jeong et al., 2001b). Therefore, P. kofoidii may have difficulty deploying a nematocyst-taeniocyst complex on the surface of a small-sized S. gracilenta cell.
The pallium feeder, O. rotunda, was able to feed on smaller prey than S. gracilenta . Ok et al. (2021) reported that many trichocysts were observed inside the cytoplasm of S. gracilenta. It is known that trichocysts of Fibrocapsa japonica may substantially complicate the capturing process involving the tow filament of O. rotunda (Tillmann and Reckermann, 2002). Thus, trichocysts in S. gracilenta may act mechanically as a grazer deterrent to O. rotunda.

Growth and Ingestion Rates of the Predators
Both the growth and ingestion rates of O. marina and Rimostrombidium sp. on S. gracilenta are affected by prey concentrations (Figures 1, 2). The half-saturation constant for the growth (K GR ) of O. marina feeding on S. gracilenta (0.14 ng C ml −1 ; 2 cells ml −1 ) was lower than that of Rimostrombidium sp. (94.5 ng C ml −1 ; 1,350 cells ml −1 ). Additionally, the half-saturation constant for the ingestion (K IR ) of O. marina (47.1 ng C ml −1 ; 673 cells ml −1 ) was also lower than that of Rimostrombidium sp. (146 ng C ml −1 ; 2,086 cells ml −1 ). Thus, at low prey concentrations, the growth and ingestion rates of O. marina would respond more readily to changes in prey concentrations than those of Rimostrombidium sp. Until now, the maximum cell abundance of S. gracilenta in natural environments has been reported as 29.9 cells ml −1 (Back et al., 2021). Without considering spatial dependence and other physical effects, O. marina may grow under this S. gracilenta concentration, while Rimostrombidium sp. may not.
The threshold concentration of S. gracilenta for the growth of Rimostrombidium sp. was 69.1 ng C ml −1 (987 cells ml −1 ). At this prey concentration, the ingestion rate of Rimostrombidium sp. on S. gracilenta was 11.2 ng C predator −1 day −1 (160 cells ml −1 ). Thus, this ingestion rate corresponds to the threshold prey concentration for basic maintenance in the abundance of Rimostrombidium sp. Meanwhile, the growth rate of O. marina without added S. gracilenta cells (i.e., predator control) was positive. Oxyrrhis marina is known to feed on bacteria and grow on dissolved nutrients (Jeong et al., 2008;Lowe et al., 2011). Thus, this physiological characteristic of O. marina may be the reason why it had a positive growth without added S. gracilenta cells in this study.
There have been three kleptoplastidic dinoflagellates whose predators were identified and on which the growth and ingestion rates of the predators were determined; Gymnodinium smaydae, P. piscicida, and S. gracilenta (Supplementary Table 2; Jeong et al., 2007aJeong et al., , 2018b. These three kleptoplastidic dinoflagellates were ingested by O. marina. The ranges of µ max and I max of O. marina on the three dinoflagellates were 0.41-0.66 day −1 and 0.11-0.33 ng C predator −1 day −1 , respectively (Supplementary Table 2). These ranges of µ max and I max of O. marina on kleptoplastic dinoflagellates showed no distinct values compared with those of O. marina on phototrophic and heterotrophic dinoflagellates (0.16-1.20 day −1 and 0.07-6.36 ng C predator −1 day −1 , respectively). Moreover, the K GR of O. marina feeding on G. smaydae and S. gracilenta (1.4 and 0.14 ng C ml −1 , respectively) showed relatively low values, while that on P. piscicida (21 ng C ml −1 ) was much higher (Jeong et al., 2007a(Jeong et al., , 2018b. Thus, the nutritional properties of each kleptoplastidic dinoflagellate as prey for a predator species may vary from species to species. More studies are needed to understand the nutritional values of kleptoplastidic dinoflagellates as prey in marine ecosystems. Neither the I max nor the µ max of O. marina on target prey species was significantly correlated with prey size (>0.05, Pearson's correlation analysis; Figures 3A,B), and the µ max was also not significantly correlated with the I max (>0.05; Figure 3C). However, the maximum swimming speed (MSS) of the prey species was significantly and negatively correlated with the µ max of O. marina on prey species (<0.05), but it was not correlated with the I max (>0.05; Figure 4). The higher the MSS of prey species, the more the predator O. marina may spend energy in catching the prey cells. Thus, the growth rates of O. marina are likely to be affected by the MSS of prey species. Among the reported dinoflagellate prey species of O. marina having the size of 9-10 µm, the µ max of O. marina on S. gracilenta was higher than those on A. cf. poporum, B. adriatica, G. smaydae, and a toxic strain of K. veneficum CCMP2064. Therefore, O. marina may spend less energy in capturing, ingesting, and digesting S. gracilenta than A. cf. poporum, B. adriatica, and G. smaydae. Among the 16 dinoflagellate prey species of O. marina, the I max of O. marina on S. gracilenta was lower than that on the other prey FIGURE 3 | The maximum growth rate (µ max ) and the ingestion rate (I max ) of Oxyrrhis marina on prey as a function of prey size (ESD, equivalent spherical diameter; µm) and µ max as a function of I max , as in Supplementary Table 2, respectively. The µ max (A; r = 0.15, p = 0.53) and I max (B; r = 0.13, p = 0.57) of Oxyrrhis marina on prey (black square) and Shimiella gracilenta (red circle), as a function of prey size (ESD  Lee (1998), Jeong et al. (2001a, 2007a,b, 2011, 2014, 2018a, Tillmann and Reckermann (2002), Strom et al. (2003), Kimmance et al. (2006), Adolf et al. (2007), Yoo et al. (2010Yoo et al. ( , 2013b, Potvin et al. (2013), Anderson and Menden-Deuer (2017), and Kang et al. (2019). species, except for Yihiella yeosuensis and Luciella masanensis (Supplementary Table 2). The fast swimming speed and quick jumping behavior of Y. yeosuensis might be responsible for the FIGURE 4 | The maximum growth rate (µ max ) and the ingestion rate (I max ) of Oxyrrhis marina on prey as a function of maximum swimming speed (MSS), as described in Supplementary Table 2 Throndsen (1973), Kamykowski and McCollum (1986), Lee (1998), Jeong et al. (1999, 2001a, 2007a,b, 2010a, 2011, 2014, 2018a, Kimmance et al. (2006), Yoo et al. (2010Yoo et al. ( , 2013b, Kang et al. (2011Kang et al. ( , 2019, Potvin et al. (2013), Lee et al. (2014a), Jang et al. (2015Jang et al. ( , 2016Jang et al. ( , 2017, Anderson and Menden-Deuer (2017), and Lim et al. (2018).
lack of predation (Jeong et al., 2018a). Therefore, the quick jumping behavior of S. gracilenta may lower the I max of O. marina on itself. However, the jumping speed of S. gracilenta is lower than that of Y. yeosuensis, and this may partially be responsible for the higher I max of O. marina on S. gracilenta. Cells of L. masanensis do not have the quick jumping behavior, but it was suggested that they excrete chemicals to protect themselves from predation (Jeong et al., 2007a).
This study clearly showed that S. gracilenta supported the positive growth of only O. marina and Rimostrombidium sp. among the potential predator species tested. Notably, these two predators have cosmopolitan distributions (Agatha, 2011;Watts et al., 2011). Shimiella gracilenta has been found in the regions where O. marina and Rimostrombidium sp. have been found: the east coast of the United States (Campbell, 1973;Johnson et al., 2003;Tucker et al., 2017), the Baltic Sea (Pedersen and Hansen, 2003;Hällfors, 2004;Watts et al., 2011), and Jinhae Bay, Korea (Choi et al., 2016;Yoon, 2018;Ok et al., 2021). Therefore, O. marina and Rimostrombidium sp. may grow on S. gracilenta if both predators and prey coexist in the regions.
To investigate the impact of predation by O. marina or Rimostrombidium sp. on the populations of S. gracilenta, ingestion rates of O. marina or Rimostrombidium sp. on S. gracilenta and their abundances should be quantified. The results of the present study provided the ingestion rates of O. marina and Rimostrombidium sp. on S. gracilenta as a function of the prey concentration. Thus, if the abundances of S. gracilenta and co-occurring O. marina or Rimostrombidium sp. are available, the predation impact by O. marina or Rimostrombidium sp. on the populations of S. gracilenta can be estimated at any prey concentration. Depending on the degree of predation impact by O. marina or Rimostrombidium sp. on the populations of S. gracilenta, the ecological roles of S. gracilenta as prey for the predator can be evaluated. Moreover, the results of the present study provide the threshold S. gracilenta concentration for the growth of O. marina or Rimostrombidium sp. and also, growth rates of the predators as a function of the prey concentration. Thus, the ecological roles of S. gracilenta as prey for O. marina or Rimostrombidium sp. can be determined. This study suggested that S. gracilenta may have an advantage for survival due to only a few common heterotrophic protist predators and low ingestion rates of the predators in marine ecosystems.

DATA AVAILABILITY STATEMENT
The original contributions presented in the study are included in the article/Supplementary Material, further inquiries can be directed to the corresponding author/s.

AUTHOR CONTRIBUTIONS
SP and HJ designed the study conception and drafted the manuscript. SP, HJ, JO, HK, JY, SE, and EP obtained the data and conducted the experiments. SP, HJ, and JO performed data analyses. All authors contributed to the article and approved the submitted version.