The study region encompassed an area between 21.0°– 42.0°N and 118.0°–156.0°E. The 45 phytoplankton sampling stations included Sections A, B and C in different water currents and Section D along 146.5°E (Fig. 1). Two seasonal cruises were carried out in the NWPO during 11 to 30 May and 27 August to 24 September 2017, representing spring and summer, respectively. A total of 720 phytoplankton samples were collected, including 630 water samples and 90 net samples. The net samples were collected from 0 to 200 m by vertical tows with a conical plankton net (mouth opening 0.1 m², mesh size 0.076 mm). The samples were concentrated to 500 mL and preserved with formaldehyde (final concentration of 2%). A total of 0.5 mL subsamples were identified under a Carl Zeiss microscope. The water samples were collected using 10 L Niskin bottles at discrete depths (2, 30, 50, 75, 100, 150 and 200 m). All samples were immediately fixed with Lugol’s solution (final concentration of 2%). After 24 h, buffered formaldehyde was added (final concentration 2%). Determinations and counts of 10 mL subsamples were analyzed under an inverted Zeiss-Z1 microscope (Carl Zeiss MicroImaging, Göttingen, Germany) at 200× or 400× magnification using the Utermöhl method (Utermöhl, 1958). Identification of algal species was based on Round et al. (1990), Tomas (1997), Sun and Liu (2002) and Lee (2008). Temperature and salinity data were collected using a Sea-Bird 911 CTD probe (Seabird Electronics, Inc., Bellevue, WA, USA). Nutrient samples were collected using Niskin water bottles and processed using a Model 300 automated nutrient analyzer (Alpkem, Clackamas, OR, USA; Grasshoff et al., 1999). The size limit of resolution for this analysis was ~5 μm.

Community diversity is described by the Shannon-Wiener diversity index (H′), following the formulation given by Sun and Liu (2004): H’=–Σ p_i log₂ p_i, J=H′/log₂S, where p_i is defined as n_i/N, n_i is the sum of cell abundance for species i in all samples, N is the sum of cell abundance for all species, and S is the species number in the sample.

Relationships between the phytoplankton community and environmental parameters were analyzed by SPSS 14.0 and CANOCO version 4.5. Only those taxa that were observed in more than 10% of the samples and with more than 1% in the total abundance were included in analyses. Both of the species data and the environmental variables were log₁₀(x+1) transformed before analyses. Detrended correspondence analysis (DCA) for the species data was used to determine the methods that would be applied. As was suitable for unimodal ordination, redundancy analysis (RDA) was chosen since that the maximum gradient length of the four axes was less than 3.

The distribution of temperature, salinity and nutrients in the surface water are shown in Fig. 2 and Table 1. SST was relatively low in spring. The average temperature of the whole region was (21.46±6.58)°C, varying from 6.62 to 29.37°C. The water temperature increased significantly from north to south, the values of Transect A in the Transition Region was below 10°C, and that of Transect B in the Kuroshio Extension region increased to nearly 20°C. The water temperature of Transect C in the southern region further increased to above 28°C, whereas a small cold eddy appeared at Sta. C4 in the south of Taiwan with water temperature below 27.8°C due to the upwelling. In summer, the water temperature increased markedly, with an average of (26.70±3.03)°C, ranging from 16.72°C to 30.25°C. The temperature in the Transition Region was still the lowest, with the minimum value (16.72°C) recorded at Sta. A1. The distribution pattern of temperature was consistent with that in spring.

In spring, the average of sea surface salinity (SSS) was 34.46±0.56, varying from 32.98 to 35.18. The salinity of Transect A was the lowest, and that at Stas A1 and A2 was below 33 due to the influence of the Oyashio current. While the salinity in the southern area was the highest, with levels at Stas D14–D18 was above 35. In summer, the average salinity of the whole region was 34.35±0.54, varying between 32.27 and 34.87. Because the Oyashio current became weaker and retreated northward, only the salinity at Stas A1 (32.27) and D1 (32.40) in the northern area was below 33.0. Similarly, since the influence of coastal current in the south of Taiwan, the salinity from Stas C1 to C5 in Transect C was lower than 34.0, with an average of 33.72±0.15, whereas that from Stas C6 to C12 in Transect C was higher than 34.0, with an average of 34.56±0.25.

In spring, nitrogen and silicon were abundant, while phosphorus was relatively insufficient. The average of dissolved inorganic nitrogen (DIN) concentration was (2.29±4.81) μmol/L, varying from 0.27 to 22.03 μmol/L. Affected by the nutrient-enriched Oyashio Current, the average DIN concentration of Transect A in the northern area was the highest (mean 14.48 μmol/L), and the values at Stas A2, A4 and A5 were as high as 16.18 μmol/L, 22.03 μmol/L and 18.56 μmol/L, respectively. Due to the influence of the oligotrophic Kuroshio, the DIN concentrations in the south-central area declined significantly, and the average of Transects B, C and D were all below 1.0 μmol/L. The average of phosphate concentration was (0.19±0.97) μmol/L, the levels at several stations were below the detection limit. Whereas, the level of Transect A was also the highest, with maximum value (1.18 μmol/L) recorded at Sta. A5. The average of silicate concentration was (2.91±3.69) μmol/L, and level of Transect A was also the highest, with an average of (12.20±1.92) μmol/L, which was 6.35, 6.74 and 7.44 times higher than that of Transect B, C and D, respectively. In summer, the nutrient concentrations decreased drastically. Especially, the phosphate and silicate at about 25% of the stations were lower than the detected value. The distribution pattern of nutrients was similar to that in spring, which the nutrient concentrations of Transect A were the highest. The average of DIN concentration was (0.30±0.34) μmol/L, varying from 0.08 to 1.87 μmol/L. The average DIN concentration of Transect A was still the highest, with an average of (0.83±0.56) μmol/L, which was 3.95, 2.24 and 3.19 times higher than that of Sections B, C and D, respectively. The average of phosphate concentration was (0.10±0.09) μmol/L, and that at Stas A1, A2 and D1 in the Transition Region were 0.21, 0.14 and 0.16 μmol/L, respectively. Whereas, the values of Transect C in the southern area was the lowest, and that at Stas C9–C12 were lower than the detection value. The average of silicate concentration was (0.61±0.52) μmol/L. The concentrations of most stations were less than 1.00 μmol/L, while the concentration of Transect A was relatively high, and the maximum value (2.53 μmol/L) was observed at Sta. A2.

A total of 281 phytoplankton taxa belonging to 5 phyla and 61 genera were identified in the two seasons. The phytoplankton community in the NWPO was mainly composed of diatoms and dinoflagellates, with 165 taxa in 44 genera of diatoms and 110 taxa in 13 genera of dinoflagellates, respectively. Most of the phytoplankton communities were cosmopolitan species and warm-water species, a few warm-temperate species such as Thalassiosira nordenskioldii, Chaetoceros concavicornis, C. convolutus and C. furcellatus were mainly found in the Transition Region due to the influence of the Oyashio Current. Warm-water species such as Planktoniella sol, Ceratium vultur var. sumatranum, Ornithocercus splendidus and Pyrocystis hamulus var. inaequalis mainly appeared in the south-central area affected by the Kuroshio Current. Trichodesmium erythraeum, T. hildebrandtii and T. thiebauti of Cyanophyta were mainly distributed in the south of Taiwan. The most species-rich genera were Chaetoceros spp., Coscinodiscus spp. and Rhizosolenia spp. for diatoms, and Ceratium spp. for dinoflagellates. The number of phytoplankton species in spring was more than that in summer, with 248 species and 235 species, respectively. In spring, diatom was the largest group, accounting for 58.47% of the total number of species and 97.45% of the total phytoplankton abundance. The dominant species were C. compressus, R. alata and T. subtilis, accounting for 14.52%, 12.86% and 11.1% of the total phytoplankton abundance, respectively. Dinoflagellate was the second most common group, accounting for only 39.11% of the total number of species and 0.44% of the total phytoplankton abundance. The frequently encountered species were C. fusus var. seta, C. fusus and C. trichoceros, which accounted for less than 0.10% of the total phytoplankton abundance. In summer, the species composition of phytoplankton community diversified, the dominant species were not prominent, with individual distribution among species was relatively uniform. The dominant genera of diatom was Chaetoceros spp., accounting for 41.10% of the total phytoplankton abundance, and the dominant species were Thalassionema frauenfeldii, Pseudo-nitzschia delicatissima, C. messanensis and C. lorenzianus, which accounted for 13.68%, 7.40%, 7.02% and 5.69% of the total phytoplankton abundance, respectively. The dominant genera of dinoflagellate was still Ceratium spp., and the dominant species were C. trichoceros, C. tripos, C. fusus and C. fusus var. seta, which accounted for 0.27%, 0.23%, 0.19% and 0.17% of the total phytoplankton abundance, respectively.

The horizontal distributions of surface phytoplankton abundance are shown in Fig. 3a, b. In spring, the average of surface phytoplankton abundance was (227.78±366.86)×10² cells/L, ranging from 38.0×10² to 1 860.0×10² cells/L. Phytoplankton abundance was mainly concentrated in the northeastern area affected by the Oyashio Current, and high phytoplankton abundances were observed at Stas A1 to A5 of Transect A as well as Stas D1 and D2 of Transect B. The maximum abundance appeared at Sta. A2, and the predominant genera were Chaetoceros spp., Rhizosolenia spp. and Thalassiosira spp., accounting for 63.90%, 27.18% and 2.40% of the station abundance, respectively. The predominant species mainly consisted of T. nordenskioldii (17.52%), C. concavicornis (17.07%), P. delicatissima (15.42%), R. alata (7.49%), C. diadema (4.87%) and R. hebetata f. semispina (3.24%). Elevated abundances (100.0×10² cells/L) were observed at Stas B1–B4 in Transect B, with dominant species being chain-forming diatoms such as Pseudoeunotia doliolus and T. nitzschioides. Along Transect C, patches of high abundance (225.0×10² cells/L) occurred at Sta. C4 in the south of Taiwan, and the predominant species was T. subtilis being clump-forming diatoms, which was brought to the surface by upwelling and contributed more than 67.76% of cell abundance at this station.

In summer, phytoplankton abundance declined significantly, with average of (83.62±109.71)×10² cells/L, ranging from 15.0×10² to 560.0×10² cells/L. High abundances (>500.0×10² cells/L) still appeared in the northern part of the survey area, but the range was reduced to Stas A1 and D1. The maximum abundance appeared at Sta. A1, which mainly consisted of T. subtilis (12.28%), C. affinis (9.70%), T. nitzschioides (6.48%), P. delicatissima (5.51%), Leptocylindrus mediterraneus (2.37%), C. lorenzianus (1.52%). A relatively high abundance was observed at Stas D6 and B1 in the Kuroshio Extension region, which were mainly composed of cosmopolitan species and warm-water species. Phytoplankton abundance in Transect C was the lowest. However, due to the influence of coastal current, the abundance at Stas C1–C5 in the south of Taiwan was significantly higher than other stations in this transect. Some neritic species such as Skeletonema costatum, P. delicatissima and Paralia sulcata were relatively abundant in the adjacent water of Taiwan.

For comparison of the two sampling methods, the horizontal distributions of net-sampling phytoplankton were shown in Figs 3c, d. In spring, the mean phytoplankton abundance of net-sampling was (959.06±1 539.22)×10³ cells/m³, ranging from 60.0×10³ to 8 600.0×10³ cells/m³. The abundance at Stas A1 to A5 in Transect A and D1 to D3 in Transect D were all above 1 000.0×10³ cells/m³, and the maximum occurred at Sta. A2, up to 8 600.0×10³ cells/m³. In summer, the mean phytoplankton abundance of net-sampling was (143.77±269.91)×10³ cells/m³, ranging from 3.07×10³ to 1 767.0×10³ cells/m³. The high value areas with abundance more than 1 000×10³ cells/m³ only appeared at Stas A2, A3 and D3 in the northern Transition Region. The maximum abundance appeared at Sta. A3, but decreased to 1 767.0×10³ cells/m³. Distribution of phytoplankton abundance of net-sampling was similar to that in the surface layer. In summary, in both water-sampling and net-sampling, the maximum phytoplankton abundance in the NWPO always appeared in the northern transition region affected by the nutrient-rich Oyashio Current, while the minimum always appeared in the southern area affected by oligotrophic Kuroshio Current. The horizontal distribution of phytoplankton abundance increased from low to high latitudes, which was consistent with the trend of nutrient distributions, but contrary to that of water temperature and salinity.

The vertical distribution of phytoplankton in the NWPO showed regional differences and seasonal variations. The vertical distribution of phytoplankton varied significantly in spring. The phytoplankton abundance at depths of 2, 30, 50, 75, 100, 150 and 200 m were 227.78, 231.22, 138.36, 57.83, 18.96, 8.79 and 5.08×10² cells/L, respectively. However, the vertical variation of phytoplankton abundance was not obvious in summer. The phytoplankton abundance at depths of 2, 30, 50, 75, 100, 150 and 200 m were 83.62, 64.36, 35.33, 12.89, 5.12, 2.95 and 1.54×10² cells/L, respectively. Because the stations in Transect D were distributed from low latitude to high latitude, the distribution pattern of phytoplankton in Transect D was basically consistent with that in the whole survey area, which could reflect the distribution trend of phytoplankton in the NWPO. Therefore, this study only took Transect D as an example to analyze the regional differences of vertical distribution of phytoplankton. The vertical distributions of temperature, salinity, nutrients and phytoplankton abundance along Transect D are shown in Fig. 4. The vertical distribution of phytoplankton abundance was notably different among regions. In the northern area affected by the Oyashio Current, the phytoplankton abundance was mainly concentrated in the upper 30 m of water column, and was mainly composed of diatoms. In the south-central area affected by the Kuroshio Current, the maximum abundance often occurred at depths of 50–75 m, which was mainly composed of dinoflagellates. For example, at Stas D1 and D2 affected by the Oyashio Current in spring, the phytoplankton abundance was mainly concentrated in the surface layer. While the phytoplankton abundance at Stas D3–D7 peaked at depth of 30 m. The maximum abundance at Stas D8–D14 appeared at depths of 30–50 m and decreased rapidly with increased depth. At Stas D15–D19 in the southern Kuroshio region, the maximum abundance was observed at depths of 50–75 m. In summer, the water column was homogeneously mixed except for the northern region. At Sta. D1 affected by the Oyashio Current, the phytoplankton abundance was mainly concentrated in the super layer, while the phytoplankton abundance at Stas D2–D5 peaked at depth of 30 m. At Stas D6–D14 in the central region, the maximum abundance were recorded at depths of 50–75 m and decreased gradually with increased depth. At Stas D15–D19 in southern region, the phytoplankton abundance was distributed homogenously throughout the water column.

The survey area is strongly influenced by two currents with different hydrographic properties, and their influence ranges and degrees vary with seasons. The hydrography in the Oyashio region is often complex, with two or three dominant water masses with different chemical and physical properties moving through during spring (Kono and Sato, 2010). The Oyashio Current and coastal waters are both enriched with iron, a critical nutrient for phytoplankton growth. Blooms off the coast of Hokkaido are a yearly spring-time occurrence (Liu et al., 2004; Nakayama et al., 2010). The high productivity of higher trophic levels, including fisheries production, in the Oyashio region is often related to this conspicuous event (spring phytoplankton bloom) (Taniguchi and Kawamura, 1972; Taniguchi, 1999). In spring, the Oyashio Current flowing south-westward was relatively strong and had a wide range of influence, covering five stations of Transect A and two stations of Transect D. Since the Oyashio Current was characterized by low temperature and salinity and high nutrient and chlorophyll a concentrations, the temperature and salinity at these stations were obviously lower than that at other stations, with temperature and salinity lower than 12.5°C and 34.0, respectively. Whereas the nutrient concentrations of Transect A were the highest, with the concentrations of DIN, phosphate and silicate were 6.32, 4.32 and 4.19 times as high as that of the whole region, respectively. Although consumed by phytoplankton, the concentrations of DIN, phosphate and silicate were still as high as 14.48 μmol/L, 0.83 μmol/L and 0.82 μmol/L, respectively. Therefore, the phytoplankton abundance in spring was 2.72 times higher than that in summer. In addition, the phytoplankton abundance of Transect A was the highest in the four transects, which was 8.65, 11.75, 8.12 and 4.79 times higher than that of Transects B, C, D and the whole area, respectively (Figs 3a, c). In summer, the Oyashio Current weakened, and the influence range retreated northward. The Oyashio Current only affected Stas A1 and D1. The water temperature and salinity of these two stations were obviously lower than those of other stations, with the water temperature below 20°C and the salinity below 33.48. Conversely, the Kuroshio Current flowing north-eastward was stronger and the scope of influence expanded northward. For example, the surface water temperature increased markedly from 19.88°C at Sta. D1 to 23.66°C at Sta. D2 and salinity increased significantly from 32.27 to 34.45. Because of the low nutrient and chlorophyll a concentrations of Kuroshio Current, the nutrient concentrations decreased greatly, and the decrease of Transect A was the largest. The concentrations of phosphate and silicate at many stations were below the detection values. The average concentration of DIN decreased from 2.29 μmol/L in spring to 0.30 μmol/L in summer. Similarly, the concentration of phosphate decreased from 0.19 μmol/L to 0.10 μmol/L, and that of silicate decreased from 2.91 μmol/L to 0.61 μmol/L. Thus the average abundance of phytoplankton declined sharply from 227.78×10² cells/L in spring to 83.62×10² cells/L in summer, and the relatively high abundance appeared only at Stas A1 and D1(Figs 3b, d). Furthermore, the seasonal variation of phytoplankton abundance in Transect A was the largest among the four transects, and the decrease magnitude of Transects A, B, C and D were 4.81, 2.13, 1.87 and 1.70 times, respectively. In contrast, the nutrients in the south-central area directly affected by the Kuroshio were quite low throughout the year, especially phosphate, the concentrations at many stations was below the detection value. Both nitrogen and phosphorus could limit phytoplankton growth in this area. Therefore, phytoplankton abundance was correspondingly the lowest in the whole region.

Several previous studies have shown that small changes in the nutrient supply can have profound impacts on the plankton communities in oligotrophic regions (Richardson, 2008). The magnitude and duration of the spring bloom are different between the coastal and offshore regions because they can have different strengths of stratification and nutrient concentrations (Kasai et al., 1997). Primary productivity was strongly affected by advection, which causes upward and downward transport of nutrients and phytoplankton blooms resulted from the cross-frontal flow along the meander of the Kuroshio Extension (Ito et al., 2000). The present study found that relatively high phytoplankton abundance was observed in the Kuroshio Extension. The common dominant species were mainly composed of cosmopolitan species such as T. nitzschioides, T. frauenfeldii, C. lorenzianus, R. styliformis, P. seriata, and warm-water species such as C. denticulatus, R. castracanei, P. sol, O. splendidus. The species richness and community diversity of phytoplankton in this region were relatively higher than those in the adjacent regions, implying significant frontal effect on phytoplankton community structure. Numerous studies have found that primary productivity is often high in estuaries and upwelling regions around the world. Large amounts of nutrients are supplied into the East China Sea (ECS) by runoff from the Changjiang River (Yangtze River) (Edmond et al., 1985; Shen et al., 2008) and upwelling of Taiwan Warm Current along the Chinese coast (Lee-Chen, 2000), and a significant amount of nutrients is supplied by the persistent upwelling of Kuroshio subsurface water in the southern ECS northeast of Taiwan (Gong et al., 1996). With such fertile hydrographic conditions, the ECS is famous for its high biological productivity (Chen and Wang, 1999). Similarly, due to the superimposing influence of coastal current and upwelling, the nutrient concentrations in the south of Taiwan were higher than that of other stations in Transect C. In spring, from Stas C1 to C5, the concentrations of DIN, phosphate and silicate in surface layer were 0.80, 0.19 and 2.55 μmol/L, respectively. However, from Stas C6 to C12, the average concentrations of DIN, phosphate and silicate in surface layer decreased to 0.55, 0.11 and 1.27 μmol/L, respectively. Thus the average abundance of Stas C1–C5 was nearly twice as high as that of Stas C6–C12. In summer, the distribution patterns of phytoplankton abundance and nutrient concentrations in Transect C was similar to that in spring.

In a word, the Oyashio Current flowing south-westward was characterized by low temperature and salinity and high nutrients. Conversely, the Kuroshio Current flowing north-eastward was characterized by high temperature and salinity and low nutrients. As a result, significant negative correlations were observed between phytoplankton abundance and temperature and salinity in all water layers both in spring and summer (p<0.01). Whereas, phytoplankton abundances were positively correlated with DIN, phosphate and silicate concentrations, especially the surface layer in spring (p<0.01, Table 2).

The relationship between phytoplankton and environmental factors was analyzed by RDA. We obtained a two-dimensional distribution figure of species, sample distribution and environmental factors (Fig. 5). The result showed that there were three distinct communities of phytoplankton in the NWPO in spring: community I distributed in the northern Transition Region which was directly affected by nutrient-enriched Oyashio current, and encompassed samples collected at Stas A1–A5 in Transect A as well as D1 and D2 in Transect D. The community was characteristic of high abundance, and primarily dominated by warm-temperate species such as R. alata, P. delicatissima, C. concavicornis and T. nordenskioldii, which were positively correlated with high concentrations of DIN, phosphate and silicate. Community II situated in the southern region, including Stas C1–C12 in Transect C as well as D19. The assemblages of phytoplankton were represented by warm-water species such as C. atlanticus which was positively correlated with water temperature. Community III mainly distributed in the south-central area, encompassing samples collected at Stas D3–D19 in Transect D and B1–B9 in the Kuroshio Extension, where was directly affected by the Kuroshio Current. The assemblages of phytoplankton were represented by cosmopolitan species such as C. compressus and C. lorenzianus which was positively correlated with salinity. In summer, community I included only Stas A1 and D1 which were still affected by the Oyashio, and represented the coastal water with high nutrient concentrations. The phytoplankton assemblage was mainly composed of neritic species such as C. affinis and P. seriata and cosmopolitan species such as T. nitzschioides, which were positively correlated with silicate concentration. Community II situated in the south of Taiwan Island and included samples collected at Stas C1–C5 in Transect C. The main dominant species were T. subtilis, C. lorenzianus and L. mediterraneus, which were brought by upwelling and coastal water and correlated with phosphate concentration. Community III was distributed in the Transition Region and Kuroshio Extension. The phytoplankton community was mainly composed of P. delicatissima and R. alata, which was correlated with DIN concentrations. Community IV situated in the south-central area, affected by the offshore Kuroshio water with low nutrient concentrations and high temperature and salinity. The community was characteristic of high species diversity and low abundance. The phytoplankton assemblage was represented by warm-water species such as C. coarctatus and C. atlanticus, which were positively correlated with water temperature and salinity.

In a word, the availability of nutrients, water temperature and salinity differed among water types. The northern Transition Region was always nutrient-enriched, while southern Kuroshio water was characterized by high temperature, high salinity and low nutrients. The seasonal variation in environmental parameters mentioned above should greatly influence the phytoplankton community. In the south-central area, the phytoplankton abundance was significant correlated with temperature and salinity regardless of spring and summer, while no obvious correlation was observed between phytoplankton abundances and any nutrient concentration. In the northern Transition Region, the phytoplankton abundances were positively correlated with nutrient concentrations, especially in spring. It embodied the unification between ecological characterization of phytoplankton and its inhabiting environment.

Studies of the factors regulating phytoplankton communities have been carried out for several decades, but the role of nutrient limitation and the resource–ratio hypothesis in influencing the phytoplankton community is still not fully understood. As diatoms and dinoflagellates show great differences in cell morphology, structure, and nutrition mode, they differ greatly in their acquisition of nutrients strategy. Many studies have revealed that dinoflagellates have the ability of mixotrophy, and the mixotrophic modes of dinoflagellates include direct engulfment of prey, peduncle feeding, and pallium feeding (Smayda, 1997; Huang et al., 2005). Stoecker (1999) found that phosphorus limitation was a common factor stimulating dinoflagellates to ingest particulate nutrients, which may be another reason for the dominance of dinoflagellates in phosphorus-limited conditions. The variation of phytoplankton community structure was always correlated with the fluctuation of physico-chemical environmental parameters. The nutrients in the NWPO were deficient except the Transition Region in the northern area. The nutrient concentrations in Transect A were the highest in the whole region, and the N:P ratio was 17.66, which was nearly to the Redfield ratio of 16:1 (Redfield et al., 1963). Moreover, the concentration of silicate was still as high as 12.20 μmol/L despite absorbed largely by phytoplankton (Table 1). Smayda (1997) and Egge (1998) considered that when silicate concentration in the environment was >2 μmol/L, diatoms could dominate the phytoplankton community if other nutrients were sufficient. Therefore, the growth of diatom in Transect A was not limited by nutrient limitation, the phytoplankton abundance was the highest in the whole region, and diatom dominated the phytoplankton community. However, the nutrient concentrations were very low in the offshore KW. The N:P ratios in Transects B, C and D were all less than 10, which implied nitrogen limitation (Justic et al., 1995). In addition, the phosphate concentrations of almost a quarter of stations were below the detection values (24.44% in spring and 26.67% in summer), and that of nearly two-thirds of the stations were below 0.1 μmol/L (62.22% in spring and 64.44% in summer), which indicated a serious lack of phosphate (Fisher et al., 1992). Egge (1998) found that diatoms are more easily affected than dinoflagellates when nutrients are limited, especially under phosphorus limitation. Thus the growth of diatom in the offshore KW was inhibited by both nitrogen and phosphorus limitation, and the phytoplankton abundance declined sharply. However, due to dinoflagellates possessing a survival advantage over diatoms by utilizing dissolved organic phosphate, the proportion of dinoflagellates in total abundance increased significantly. Whether in spring or summer, the abundance ratio of dinoflagellates to total phytoplankton increased, while that of diatoms to total phytoplankton decreased with N:P ratio decreased. On the other hand, the Si:N ratios were more than 1.0 except for Transect A in spring, which indicates that silicate was relatively abundant in the investigated sea. Therefore, no obvious correlation was observed between phytoplankton composition and Si:N ratio (Table 3). Further analysis revealed that the relative abundance of diatoms to dinoflagellates was significantly positive correlated with the N:P ratio (R=0.942^**, ^**p<0.01), and negatively correlated with the Si:N ratios (R=–0.693^*, ^*p<0.05). The present study indicates that N: P ratio rather than Si:N ratio played a significant role in regulating the phytoplankton community structure in the NWPO. Yin et al. (2000) reported a similar phenomenon in the Zhujiang River (Pearl River) Estuary, where phosphorus limitation favored the dominance of dinoflagellates. Hashihama et al. (2008) also reported that the dominance of flagellates was related to the influence of Kuroshio water on the Sagami Bay. Guo et al. (2014) revealed a relatively good correlation between the relative abundance of diatoms and dinoflagellates with N:P ratios in the East China Sea. In contrast, diatom dominance was mainly controlled by the horizontal advection of coastal water.