First Concurrent Measurement of Primary Production in the Yellow Sea, the South Sea of Korea, and the East/Japan Sea, 2018

: Dramatic environmental changes have been recently reported in the Yellow Sea (YS), the South Sea of Korea (SS), and the East/Japan Sea (EJS), but little information on the regional primary productions is currently available. Using the 13 C- 15 N tracer method, we measured primary productions in the YS, the SS, and the EJS for the ﬁrst time in 2018 to understand the current status of marine ecosystems in the three distinct seas. The mean daily primary productions during the observation period ranged from 25.8 to 607.5 mg C m − 2 d − 1 in the YS, 68.5 to 487.3 mg C m − 2 d − 1 in the SS, and 106.4 to 490.5 mg C m − 2 d − 1 in the EJS, respectively. In comparison with previous studies, signiﬁcantly lower ( t -test, p < 0.05) spring and summer productions and consequently lower annual primary productions were observed in this study. Based on PCA analysis, we found that small-sized (pico- and nano-) phytoplankton had strongly negative effects on the primary productions. Their ecological roles should be further investigated in the YS, the SS, and the EJS under warming ocean conditions within small


Introduction
Marine phytoplankton as primary producers play an important role as the base of the ecological pyramid in the ocean and are responsible for nearly a half of global primary production [1,2]. The primary production of phytoplankton is widely used as an important indicator to predict annual fishery yield in various oceanic regions [3][4][5], because it is one of key factors in determining amount of food source for upper-trophic-level consumers [6,7]. Lee et al. [8,9] also reported that an algorithm for estimation of the habitat suitability index for the mackerels and squids around the Korean peninsula was largely improved by including a primary production term. The physiological conditions and community structures of phytoplankton are closely related to physical and chemical factors (e.g., light regime, nutrients, and temperature) [10][11][12], which induce greatly different phytoplankton productions in various marine ecosystems [3,13,14]. Thus, the primary production measurements can provide fundamental backgrounds for better understanding marine ecosystems with different environmental conditions and detecting current potential ecosystem changes.
The Yellow Sea (hereafter YS), the South Sea of Korea (SS), and the East/Japan Sea (EJS), belonging to the East Asian marginal seas, have experienced 2-4 times faster increase (0.7-1.2 • C) in seawater temperature than that in global mean water temperature (0.4 • C) for 20 years   [15]. Moreover, some notable changes in physicochemical conditions were reported, such as increasing limitation of nutrients in the YS and rapid Table 1. Description of sampling sites in the YS, the SS, and the EJS for each cruise period, in 2018. (o) means investigation was conducted, while (-) means investigation was not conducted. The bottom depths at our sampling stations in the YS and the SS had relatively narrow range, whereas the EJS had a wide range of bottom depths (48-2340 m) in this study ( Table 1). The six water depths were determined at each station by converting Secchi disc depth to 6 corresponding light depths (100, 50, 30, 12, 5, and 1% of surface photosynthetic active radiation; (PAR)). Then, each water sample was collected from 6 different depths using Niskin bottles (8 L) equipped with a conductivity, temperature, and depth (CTD)rosette. The water temperature and salinity were obtained from SBE9/11 CTD (Sea-Bird Electronics, Bellevue, WA, USA). The mixed-layer depth (MLD) was defined as the depth at which the density is increased by 0.125 density units from the sea surface density [27,28]. Water samples for dissolved inorganic nutrients (NH 4 , NO 2 + NO 3 , PO 4 , and SiO 2 ) and chl-a (total and size-fractionated) concentrations were collected at three light depths (100, 30, and 1% of PAR). Water samples for measuring the particle organic carbon (POC) and particle organic nitrogen (PON) concentrations and total carbon uptake rates (primary production) of phytoplankton were collected at six light depths (100, 50, 30, 12, 5, and 1% of PAR). The euphotic zone is defined as the depth from 100 to 1% of PAR.

Inorganic Nutrients Concentrations
To measure concentrations of dissolved inorganic nutrients (NH 4 , NO 2 + NO 3 , PO 4 , and SiO 2 ), 0.1 L water samples were filtered onto Whatman GF/F filters (ø = 47 mm) at a vacuum pressure lower than 150 mmHg. Filtered water samples were immediately frozen at −20 • C for further analysis in our laboratory. An auto-analyzer (Quattro, Seal Analytical, Norderstedt, Germany) in the NIFS was used for the analysis of dissolved inorganic nutrients according to the manufacturer's instruction.

Chl-a Concentration
The primary method and calculation for determining the chl-a concentrations were conducted according to Parsons et al. [29]. Water samples (0.1-0.4 L) for total chl-a concentration were filtered through Whatman GF/F filters (ø = 25 mm), and samples (0.3-1 L) for three different size-fractionated chl-a concentrations were passed sequentially through 20 µm and 2 µm membrane filters (ø = 47 mm) and GF/F filters (ø = 47 mm) at low vacuum pressure. The filtered samples were then placed in a 15 mL conical tube, immediately stored in −20 • C freezer until the analysis. In the laboratory, the frozen filters were extracted with 90% acetone at 4 • C for 20-24 h, and chl-a concentrations were then measured using a fluorometer (Turner Designs, 10-AU, San Jose, CA, USA) calibrated based on commercially available reference material for chl-a.

Measurements of Phyoplankton Carbon and Nitrogen Uptake Rate
The 13 C-15 N dual stable isotope tracer technique was used for simultaneously measuring the carbon and nitrogen uptake rates of the phytoplankton as described by Dugdale and Goering [30] and Hama et al. [31]. In brief, water samples from each light depth (100%, 50%, 30%, 12%, 5%, and 1% of PAR) were immediately transferred to acid-rinsed polycarbonate incubation bottles (1 L) covered with neutral density screens (Lee Filters) [32] after passing through 333 µm sieves to eliminate the large zooplankton. The incubation bottles filled with seawater at each light depth were inoculated with the labeled carbon (NaH 13 CO 3 ) and nitrate (K 15 NO 3 ) or ammonium ( 15 NH 4 Cl), which correspond to 10-15% of the concentrations in the ambient water [30,31]. Then, the tracer-injected bottles were incubated in a large polycarbonate incubator at a constant temperature maintained by continuously circulating sea surface water under natural surface light for 4-5 h. The incubated water samples (0.1-0.4 L) were filtered onto Whatman GF/F filters (ø = 25 mm) precombusted at 450 • C, and the filters were then kept in a freezer (−20 • C) until mass spectrometer analysis. At the laboratory of Pusan National University, the filters were fumed with a strong hydro acid in a desiccator to remove the carbonate overnight and dried with a freeze drier for 2 h. Then, POC and PON concentrations and atom % of 13 C were analyzed by Finnigan Delta+XL mass spectrometer at the stable isotope laboratory of the University of Alaska (Fairbanks, AK, USA). The carbon uptake rates of the phytoplankton were estimated as described by Dugdale and Goering [30] and Hama et al. [31]. The final values of the carbon uptake rates of phytoplankton were then calculated by subtracting the carbon uptake rates of dark bottles to eliminate the heterotrophic bacterial production [33][34][35]. The daily primary productions of phytoplankton were calculated from the hourly primary productions observed in this study and 10-h photoperiod per day reported previously in the YS and EJS [22,24].

Statistical Analysis
The statistical analyses for Pearson's correlation, t-test, and one-way analysis of variance (one-way ANOVA) were performed using SPSS (version 12.0, SPSS Inc., Chicago, IL, USA). In the one-way ANOVA, a test to certify the homoscedasticity of variables was conducted by using Levene's test. To compare pairwise differences for the variables, Scheffe's (homogeneity) and Dunnett's (heteroscedasticity) post hoc tests were used, based on homogeneity of variances.
Principal component analysis (PCA) with the Varimax method with Kaiser normalization using the XLSTAT software (Addinsoft, Boston, MA, USA) was used to identify relatively significant factors affecting the total carbon uptake rates of phytoplankton in each sea during our observation time. Fourteen variables for PCA included physical (water temperature and salinity and euphotic and mixed-layer depths), chemical (NH 4 , NO 2 +NO 3 , PO 4 , and SiO 2 concentrations), and biological (total and size-fractionated chl-a and POC concentration) factors and carbon uptake rates of phytoplankton.

Physicochemical Environmental Conditions
Seasonal vertical profiles of the mean temperatures and salinities at each light depth in the YS, the SS, and the EJS are presented in Figure 2. Seasonal water temperatures and salinities in the YS, the SS, and the EJS were evenly distributed within the euphotic zone except in August. The mean temperatures within the euphotic zone in the YS, the SS, and the EJS were lowest in February, with means of 5.9 (S.D. = ± 2.3), 13.6 (± 1.3), and 9.9 (± 1.7) • C, respectively, and gradually increased to their highest in August, with means of 23.2 (± 1.4), 23.8 (± 1.6), and 20.9 (± 2.9) • C, respectively ( Figure 2). The average water temperature in the YS was significantly lower than those in the SS and EJS during February and April (one-way ANOVA, p < 0.05). The highest mean salinities in the YS and EJS were observed in April (32.8 ± 0.8 and 34.4 ± 0.1 psu), whereas salinity in the SS was highest in February at 34.6 ± 0.0 psu ( Figure 2). Overall, lower salinities were found in the YS than in the SS and the EJS throughout the observation period.
˚C, respectively, and gradually increased to their highest in August, with means of 23.2 (± 1.4), 23.8 (± 1.6), and 20.9 (± 2.9) ˚C, respectively ( Figure 2). The average water temperature in the YS was significantly lower than those in the SS and EJS during February and April (one-way ANOVA, p < 0.05). The highest mean salinities in the YS and EJS were observed in April (32.8 ± 0.8 and 34.4 ± 0.1 psu), whereas salinity in the SS was highest in February at 34.6 ± 0.0 psu ( Figure 2). Overall, lower salinities were found in the YS than in the SS and the EJS throughout the observation period.  The mean euphotic depths in the YS, the SS, and the EJS were deepest in August at 37.6 ± 15.6, 49.8 ± 11.3, and 54.4 ± 10.7 m, respectively ( Figure 3). In particular, the euphotic depth in the EJS in February (51.0 ± 5.8 m) was significantly deeper (one-way ANOVA, p < 0.01) than those in the YS (12.8 ± 6.2 m) and the SS (28.1 ± 4.7 m). The deepest MLDs in the YS, the SS, and the EJS were observed in February, with means of 68.7 ± 15.7, 59.0 ± 40.5, and 80.6 ± 57.4 m, respectively ( Figure 3). The MLDs in the YS, the SS, and the EJS became continuously shallow until August at 12.0 ± 14.2, 13.7 ± 6.6, and 13.2 ± 6.2 m, respectively, and then deepened in October to 26.3 ± 13.7, 30.2 ± 16.1, and 37.9 ± 14.2 m, respectively. In all regions, the differences between the MLDs and euphotic depths were greatest in February, decreased toward April, and then reversed in August when MLDs were significantly shallower than the euphotic depths (t-test, p < 0.01) ( Figure 3). These results indicate that the euphotic zone was vertically well-mixed in all study regions during February and April, whereas strong stratifications were developed in the euphotic water columns during August.
respectively. In all regions, the differences between the MLDs and euphotic depths were greatest in February, decreased toward April, and then reversed in August when MLDs were significantly shallower than the euphotic depths (t-test, p < 0.01) ( Figure 3). These results indicate that the euphotic zone was vertically well-mixed in all study regions during February and April, whereas strong stratifications were developed in the euphotic water columns during August. Major dissolved inorganic nutrient concentrations at each light depth (100%, 30%, and 1%) in the YS, the SS, and the EJS for each cruise are summarized in Table 2. The ranges of NO2+NO3, PO4, and SiO2 concentrations during the study period were 0.5-9.9, <0.1-0.6, and 2.4-10.0 μM in the YS; 0.9-8.1, 0.1-0.4, and 5.1-11.3 μM in the SS; and 0.2-8.7, 0.1-0.5, and 2.4-11.0 μM in the EJS, respectively. Ranges of nutrient concentrations except for NH4 varied significantly in all regions during the study period, being generally high in February and low in other seasons except NO2+NO3 concentrations in the YS in April. The nutrient concentrations, except for NH4 at 1% light depths in the YS and the EJS, were higher (one-way ANOVA, p < 0.01) than those at 100% and 30% light depths during August and October, whereas vertical differences in the SS were only detected in August. NH4 concentrations ranged from 0.5 to 1.2 μM in the YS, 0.1 to 0.6 μM in the SS, and 0.4 to 0.9 μM in the EJS, respectively, during the observation period. Unlike other nutrients, NH4 concentrations had no distinct seasonal and vertical characteristics in all study regions. Major dissolved inorganic nutrient concentrations at each light depth (100%, 30%, and 1%) in the YS, the SS, and the EJS for each cruise are summarized in Table 2. The ranges of NO 2 +NO 3 , PO 4 , and SiO 2 concentrations during the study period were 0.5-9. The nutrient concentrations, except for NH 4 at 1% light depths in the YS and the EJS, were higher (one-way ANOVA, p < 0.01) than those at 100% and 30% light depths during August and October, whereas vertical differences in the SS were only detected in August. NH 4 concentrations ranged from 0.5 to 1.2 µM in the YS, 0.1 to 0.6 µM in the SS, and 0.4 to 0.9 µM in the EJS, respectively, during the observation period. Unlike other nutrients, NH 4 concentrations had no distinct seasonal and vertical characteristics in all study regions. Apr. Aug. Oct. Oct. Apr.

POC and PON Concentration
The mean POC concentrations integrated in the euphotic zone in the YS, the SS, and the EJS showed different seasonal patterns in comparison to the chl-a concentrations (Figure 6b). The POC concentrations in the YS and SS gradually increased from February, at

POC and PON Concentration
The mean POC concentrations integrated in the euphotic zone in the YS, the SS, and the EJS showed different seasonal patterns in comparison to the chl-a concentrations (Figure 6b). The POC concentrations in the YS and SS gradually increased from February, at 1.7 ± 0.5 and 2.7 ± 1.0 g C m −2 , to October, with 10.4 ± 3.7 and 7.5 ± 3.1 g C m −2 , respectively. In comparison, the POC concentrations in the EJS were the highest during August at 8.9 ± 1.5 g C m −2 but remained constant at an average of~4 g C m −2 during other seasons.

POC and PON Concentration
The mean POC concentrations integrated in the euphotic zone in the YS, the SS, and the EJS showed different seasonal patterns in comparison to the chl-a concentrations (Figure 6b). The POC concentrations in the YS and SS gradually increased from February, at 1.7 ± 0.5 and 2.7 ± 1.0 g C m −2 , to October, with 10.4 ± 3.7 and 7.5 ± 3.1 g C m −2 , respectively. In comparison, the POC concentrations in the EJS were the highest during August at 8.9 ± 1.5 g C m −2 but remained constant at an average of ~4 g C m −2 during other seasons.

Primary Production of Phytoplankton
The primary productions of phytoplankton integrated from different six-light depths (100, 50, 30, 12, 5, and 1%) ranged from 1.0 to 135.1 (YS), 1.8 to 63.7 (SS), and 2.3 to 119.3 (EJS) mg C m −2 h −1 , respectively (Figure 7). The ranges of the primary productions in the YS and the EJS were more variable than in the SS in this study. High mean primary productions in the YS, the SS, and the EJS were observed during April (29.3 ± 39.4, 42.6 ± 7.8, and 49.1 ± 25.2 mg C m −2 h −1 ) and October (60.6 ± 17.8, 48.4 ± 15.4, and 43.3 ± 31.1 mg C m −2 h −1 ) (Figure 6c). In comparison, the mean primary productions during February and August were low in the YS (2.6 ± 1.2, 9.3 ± 1.0 mg C m −2 h −1 ), the SS (6.8 ± 3.5 and 19.5 ± 12.5 mg C m −2 h −1 ), and the EJS (10.6 ± 7.7 and 28.4 ± 20.4 mg C m −2 h −1 ) (Figure 6c). Overall, there were distinct seasonal variations in the primary productions, which were higher in spring and autumn than those in winter and summer in all waters of the littoral sea in Korea in 2018. and 49.1 ± 25.2 mg C m −2 h −1 ) and October (60.6 ± 17.8, 48.4 ± 15.4, and 43.3 ±31.1 mg C m −2 h −1 ) (Figure 6c). In comparison, the mean primary productions during February and August were low in the YS (2.6 ± 1.2, 9.3 ± 1.0 mg C m −2 h −1 ), the SS (6.8 ± 3.5 and 19.5 ± 12.5 mg C m −2 h −1 ), and the EJS (10.6 ± 7.7 and 28.4 ± 20.4 mg C m −2 h −1 ) (Figure 6c). Overall, there were distinct seasonal variations in the primary productions, which were higher in spring and autumn than those in winter and summer in all waters of the littoral sea in Korea in 2018.  The results of PCA to determine major environmental and biological factors affecting the primary productions of phytoplankton in the YS, the SS, and the EJS throughout the observation period are shown in Figure 8a-c. The two ordination axes (PC1 and PC2) of principal components (PC) accounted for the cumulative variances of 61.6, 66.9, and 54.7% in the YS, the SS, and the EJS, respectively. Primary production in the YS was positively correlated with the chl-a and POC concentrations and temperature but negatively correlated with the MLD and compositions of nano-sized phytoplankton (Figure 8a). The positive relations between primary production and total chl-a concentrations and compositions of micro-sized phytoplankton were observed in the SS (Figure 8b). In contrast, picosized phytoplankton compositions and nutrients except for NH 4 were negatively related to primary production in the SS (Figure 8b). For the EJS, the total chl-a concentrations, compositions of the micro-sized plankton, and salinity had positive effects, whereas the pico-sized plankton and water temperature had negative effects on the primary production ( Figure 8c).
No strong correlation (R 2 = 0.1225 p > 0.05) was found between the biomass contributions of pico-sized phytoplankton and the primary production of phytoplankton in the YS (Figure 9a). In contrast, significantly negative correlations between the biomass contributions of pico-sized phytoplankton and the primary production were observed in the SS (R 2 = 0.791, p < 0.01) and the EJS (R 2 = 0.801, p < 0.01) (Figure 9b,c).
itive relations between primary production and total chl-a concentrations and compositions of micro-sized phytoplankton were observed in the SS (Figure 8b). In contrast, picosized phytoplankton compositions and nutrients except for NH4 were negatively related to primary production in the SS (Figure 8b). For the EJS, the total chl-a concentrations, compositions of the micro-sized plankton, and salinity had positive effects, whereas the pico-sized plankton and water temperature had negative effects on the primary production (Figure 8c). No strong correlation (R 2 = 0.1225 p > 0.05) was found between the biomass contributions of pico-sized phytoplankton and the primary production of phytoplankton in the YS (Figure 9a). In contrast, significantly negative correlations between the biomass contributions of pico-sized phytoplankton and the primary production were observed in the SS (R 2 = 0.791, p < 0.01) and the EJS (R 2 = 0.801, p < 0.01) (Figure 9b,c).   No strong correlation (R 2 = 0.1225 p > 0.05) was found between the biomass contributions of pico-sized phytoplankton and the primary production of phytoplankton in the YS (Figure 9a). In contrast, significantly negative correlations between the biomass contributions of pico-sized phytoplankton and the primary production were observed in the SS (R 2 = 0.791, p < 0.01) and the EJS (R 2 = 0.801, p < 0.01) (Figure 9b,c).

Comparisons of Primary Production between This and Previous Studies
Based on a 10-h photoperiod and the hourly primary productions obtained in this study (Figure 6), the mean daily primary productions in the YS were 25.8 ± 11.9, 292.7 ± 393.9, 139.0 ± 66.9, and 607.5 ± 172.6 mg C m −2 d −1 during winter, spring, summer, and autumn, respectively. Our values obtained in this study were slightly lower than the ranges (56-947 mg C m −2 d −1 ) of the values reported previously in adjacent or nearly identical regions to our sites in the YS (Table 3). In particular, our spring (293 mg C m −2 d −1 ) and summer (139 mg C m −2 d −1 ) values in this study were significantly lower (t-test, p < 0.05) than the spring (851 ± 108 mg C m −2 d −1 ) and summer (555 ± 231 mg C m −2 d −1 ) values averaged from previous studies. These lower seasonal productions in 2018 might be explained by a recent change in the nutrient budgets in the YS. An increasing trend in dissolved inorganic nitrogen (DIN) concentration since the 1980s was reported, whereas a decreasing trend from the 1980s to 2000 followed by a slight increase in PO 4 concentration was observed in the YS [16,36]. These changes in DIN and PO 4 have induced a gradual increase in the N/P ratio and a shift from N-limitation to P-limitation in the YS [36]. The P-limited condition could convert dominant species of phytoplankton from diatoms to small-sized non-diatoms with higher growth rates in P-limited waters but lower photosynthetic efficiencies [18,24,37]. Lin et al. [16] reported that a dramatic decrease in primary production in the YS during all seasons between 1983-1986 and 1996-1998 periods could be one of the ecological responses caused by the increase in the N/P ratio. In this study, the N/P ratios (32 ± 14) during the spring period were significantly higher (one-sample t-test, p < 0.01) than the Redfield ratios (16) [38], which could have resulted in a limitation for diatom growth [39,40]. Indeed, the diatom compositions (approximately 50%) in the YS in spring based on the results from our parallel study (non-published data) were distinctly lower than those reported previously in 1986 (89%) and 1998 (70%) [23]. This shift in dominant species could have caused the low primary production in spring 2018. Jang et al. [24] reported that the high contribution of pico-sized (<2 µm) phytoplankton to the total primary production could induce a lower total primary production in the YS when the N/P ratio is higher than 30 during the summer period. We did not measure the production of pico-sized phytoplankton in this study, but the higher N/P ratio (54 ± 78) at upper euphotic depth (100 and 30%) accounted for about 75% of integrated primary production and could explain the lower primary production in the YS during summer 2018.
Since the primary production measurements have rarely been conducted in the SS section belonging to the northern part of the East China Sea, we compared our results with those measured previously in the entire East China Sea ( Table 4). The average daily primary productions in the SS during this observation are within the range (102-1727 mg C m −2 d −1 ) reported previously in the East China Sea (Table 4). However, the winter and summer values in this study were significantly lower (t-test, p < 0.05) than the mean winter (206 ± 93 mg C m −2 d −1 ) and summer (621 ± 179 mg C m −2 d −1 ) productions reported previously. In comparison, the autumn value (487 mg C m −2 d −1 ) in this study was consistent with the previous findings (503 ± 186 mg C m −2 d −1 ). For the springtime, our daily production (426 mg C m −2 d −1 ) was not statistically different (t-test, p > 0.05) from the mean production (350 ± 161 mg C m −2 d −1 ) in early spring (March), but our spring value was considerably lower than those reported previously in April (1727 mg C m −2 d −1 ) and May (1375 mg C m −2 d −1 ). Table 3. Comparisons of daily primary production in the YS. PP represents daily primary production.

Resion
Method  The daily primary productions measured in this study during four seasons are within the range (44-1505 mg C m −2 d −1 ) obtained previously from the various regions in the EJS in different seasons (Table 5). However, our value (284 mg C m −2 d −1 ) during the winter period was significantly higher (t-test, p < 0.05) than the winter mean value (75 ± 44 mg C m −2 d −1 ) reported by Nagata [51] and Yoshie et al. [52], whereas our spring (491 mg C m −2 d −1 ) and summer (106 mg C m −2 d −1 ) rates were significantly lower (t-test, p < 0.05) than the spring (858 ± 376 mg C m −2 d −1 ) and summer (519 ± 184 mg C m −2 d −1 ) values averaged from various previous studies. A plausible mechanism for the difference might be related to the development of the MLD in the EJS during the wintertime. A vigorous vertical mixing driven by the Asian winter monsoon can limit the availability of light to phytoplankton in winter [53,54] but induces an increase in the nutrient availability in the upper euphotic layer from spring to summer [55,56]. However, the MLD has been gradually decreased by an increase in water temperature and weakened wind stress in the EJS [17,57,58], which could offer better light conditions for phytoplankton growth in winter but fewer nutrients for the spring phytoplankton bloom. In this way, the difference in seasonal primary production in the EJS mentioned above could be explained by the recent change in the MLD. However, because our surveys in the EJS were restricted to only 2018, this mechanism needs to be verified by a long-term observation. Another reason for the low primary production, especially in spring 2018, could be potentially having missed the bloom timing of the phytoplankton during our observation period. In general, the spring bloom in the EJS is mainly driven by the massive growth of diatoms, which account for the majority of large-sized (> 20 µm) phytoplankton [59][60][61]. Indeed, Kwak et al. [62] observed a significantly higher contribution (approximately 60%) of diatoms during the spring bloom period than in other seasons. In this study, the contribution of the large-sized phytoplankton was rather lower during the spring (Figure 5c). However, much lower diatom contributions were detected based on our parallel stud, showing that diatoms accounted for only 23.1% (± 9.9%) of total phytoplankton communities in the EJS in spring (non-published data). The other reason might be conspicuously low phytoplankton biomass in the EJS in April 2018. Based on MODIS (Moderate Resolution Imaging Spectrometer)-Aqua monthly level-3 datasets regarding chl-a (https://oceandata.sci.gsfc.nasa.gov/MODIS-Aqua/, accessed on 3 August 2021), the surface chl-a showed strong negative anomalies in the southwestern part of the EJS during April between 2003-2015 and 2018 (data not shown). As the chl-a concentrations in the EJS were one of the major factors controlling the primary production (Figure 8c), a noticeable low chl-a concentration could cause lower primary production in the EJS during the springtime in 2018. At the current stage, it is difficult to find out a solid reason for the low chl-a concentration in the EJS during April 2018, which should be further resolved for a better understanding of the EJS ecosystem.

Main Factors Affecting the Primary Production in the YS, SS, and EJS in 2018
Based on the PCA results (Figure 8), the major factors controlling the phytoplankton productions were different among the three seas. Total chl-a concentrations (positively; +), temperature (+), MLD (negatively; −), and nano-sized phytoplankton contribution (−) are found to be major controlling factors in the YS. In comparison, total chl-a concentrations (+), pico-(−) and micro-sized (+) phytoplankton contributions, and nutrients (−) except for NH 4 can greatlyaffect the primary production in the SS. For the EJS, the primary production of phytoplankton can greatly vary due to total chl-a concentrations (+), micro-sized phytoplankton contribution (+), salinity (+), pico-sized phytoplankton (−), and water temperature (−). The effects of physical (temperature, salinity, and MLD) and chemical (nutrients) factors are different in the YS, the SS, and the EJS. Given the positive relationships between the primary productions and the total chl-a concentrations in this study, biomass-driven primary productions are characteristics in the YS, the SS, and the EJS ecosystems, at least in 2018. However, the effects of the three size groups of phytoplankton can be different among the three seas. The contribution of nano-sized phytoplankton in the YS and the contributions of pico-sized phytoplankton in the SS and the EJS are negatively correlated with the primary productions in this study. Choi et al. [42] reported nano-phytoplankton contributed greatly to the primary production in the YS, based on the large biomass contribution of nano-phytoplankton (approximately 60%). In this study, the negative relationship between the nano-sized phytoplankton contribution and the primary production indicates that increasing contributions of the nano-sized phytoplankton could decrease the primary production in the YS. In the EJS, several previous studies reported higher contributions of pico-sized phytoplankton could cause a decrease in the primary production [12,22,69]. Indeed, marked decreasing trends in the primary productions with increasing pico-sized phytoplankton biomass were observed in the SS and EJS during our observation period in 2018 (Figure 9b.c). This could be caused by the different primary productivities between pico-and large-sized (>2 µm) phytoplankton [22]. Generally, pico-sized phytoplankton have a lower primary productivity than large phytoplankton [14,22,70]. Therefore, the total primary production can be decreased by increasing contribution of pico-sized phytoplankton, with their lower productivity traits. Under ongoing warming ocean conditions, pico-sized phytoplankton are expected to be predominant in phytoplankton communities [71][72][73][74]. In this pico-sized-phytoplanktondominated ecosystem, a lower total primary production could be expected in the SS and the EJS based on the negative relationships between the primary production and pico-sized phytoplankton observed in this study. The ecological roles of pico-sized phytoplankton in regional marine ecosystems should be further investigated in the YS, the SS, and the EJS with current environmental changes.