Diatom sampling was performed at 114 sites along the coast of R. O. Korea (Fig. 1, Table S1) during January and February 2010. We chose sites accessible from the coast for phytoplankton netting, and that were <40 km apart in order to achieve full coverage of Korean waters. One-liter water samples (sampling depth: 1 m under sea surface) were collected in clean polyethylene bottles for quantitative analysis, and phytoplankton samples were taken using a 20 μm mesh net (horizontal netting for 10 m) for qualitative analysis. Collected samples were immediately fixed with 5% Lugol’s solution (Sigma, USA) and transported to the laboratory. Environmental data, including temperature, salinity, and pH were measured in situ, using a YSI-6600 portable meter (YSI, USA).

The Lugol’s fixed water samples were settled for 1 d, and then the supernatant was removed to concentrate the inorganic and organic particles including phytoplankton communities at approximately 50 mL. The counting of total phytoplankton communities focused on diatoms in each concentrated sample under a light microscope (Axioskop 40, Zeiss, Germany) was continued until the minimum 600 cells per sample using a Sedgwick-Rafter counting chamber (Kim et al., 2016; Kang et al., 2021; Jung et al., 2021). To analyze the fine structure of diatom species, we cleaned the diatom sample. Briefly, cellular organic material in a diatom cell was removed using equal amounts of KMnO₄ and HCl in a 70°C water bath until the sample became clear, and then the acid was rinsed using distilled water in five time. Selected cleaned samples were mounted in a Pleurax (Cat. No.139-06682, Wako, Japan) and observed under the light microscope equipped with a CCD camera (AxioCamMRc5, Zeiss, German). In addition, we observed using a scanning electron microscope (SEM) (JSM7600F, Jeol, Tokyo, Japan) by cleaning sample (Jung and Park, 2019; Lim et al., 2012). For examination using the SEM, the rest of the cleaned samples were filtered onto a polycarbonate membrane (3.0-μm pore size; TSTP02500, Millipore, USA), which was then dried in air. The filtrated membranes were attached to an aluminum stub using carbon tape and then sputter-coated with gold. The SEM was operated at accelerating voltage of 5 kV using a 10 mm working distance. Diatom identification is based on the authoritative literatures such as Schmidt’s Atlas der Diatomaceen-kunde (Schmidt, 1874–1959), Hustedt’s Die Kieselalgen series (Hustedt, 1927–1966), (Round et al., 1990), and (Halse and Syvertsen, 1996). In particular, the highly diverse genera such as Chaetoceros and Thalassiosira are identified based on Lee et al. (2014a, 2014b) and (Park et al., 2016).

Species that contributed ≥1% (relative abundance) of total diatomic assemblage of each sample were selected for numerical analysis resulting in 156 diatom taxa being used. Diatom assemblage diversity was calculated using the Shannon-Wiener diversity index. The absolute abundance of each species was transformed by its fourth root into normalizing skewed composition, data and pairwise distances between sampling sites were calculated using the Bray-Curtis similarity algorithm.

To identify spatial similarity between sampling sites, we performed eight hierarchical clustering methods based on the pairwise distance matrix. These were as follows: the single linkage method, the complete linkage method, the unweighted pair-group method using arithmetic averages (UPGMA), the weighted pair-group method using arithmetic averages, the unweighted pair-group method using centroids, the weighted pair-group method using centroids, and two variants of ward’s minimum variance method (ward.D and ward.D2). The degrees of data distortion from the eight methods were then assessed based on cophenetic correlation coefficients (Sokal and Rohlf, 1962). Pairwise distances between sampling sites were calculated in the “vegan” package (Oksanen et al., 2013), and clustering was visualized using the “factoextra” package (Kassambara and Mundt, 2017), both in R (the R Project for Statistical Computing, supported by the R Foundation for Statistical Computing).

The indicator value method (IndVal) was applied to identify indicator species among the groups of sites using the “indicspecies” method in R (De Cáceres, 2013). The IndVal values ranged from zero for “not an indicator species” to one for “maximum indicator ability”.

Based on the cophenetic correlation coefficient values (r=0.73), the UPGMA clustering algorithm provided the best dissimilarity approximation compared with the other seven methods, and so we adopted the groupings achieved by this method to establish diatom-based regions. The UPGMA clustering, based on the Bray-Curtis similarity matrices, clustered the data into four major groups and three additional unique groups (Fig. 2). Group 1 consisted of 22 sites, mostly in the YS but including one in the SS (Site 64) and two in the ES (Sites 75 and 86). Group 2 included 51 sites located in the SS with four sites from the YS (Sites 3, 10, 12 and 13) also included. Group 3 included 25 sites in the SES, while Group 4 had 11 sites from the NES, as well as solitary site (Site 82) from the SES region. Three unique groups with one to two sites were separated from these four major groups: Unique Group 1 and Unique Group 2 included two sites each: Sites 47 and 48, and Sites 102 and 104, respectively, while Unique Group 3 included just Site 105.

Abiotic variability for water temperature, salinity, and pH, was relatively low in the YS, high in the SS, and moderate in the ES (Table 1, Figs 1a–c). Mean diatom abundance across the entire study area was 2.89×10⁵ cells/L, with means in the YS, SS, SES and NES of 2.21×10⁵ cells/L, 4.80×10⁵ cells/L 0.24×10⁵ cells/L, and 0.22×10⁵ cells/L, respectively (Table 1, Fig. 1d). The changes in Shannon diversity were similar to the pattern of change of the richness, while the Simpson evenness followed the opposite trend (Table 1, Fig. 1e). Shannon diversity indices were variable (0.25–2.82) site-by-site with low values in the western SS, dominated by Thalassiosira nordenskioeldii (0.42–0.79 in Sites 25–30), and Eucampia zodiacus (0.42 at Site 31). Diatom richness was 7–36 with low values in the SES (7–9 at Sites 89–91, 93, 100 and 101) (Fig. 1f).

The indicator species for the four major regions were identified using calculated indicator values (p<0.05), and 7, 17, 6 and 7 indicator species were found for the four regions, respectively (Table 2, Fig. 3). Among the 37 indicator species, Actinoptychus senarius (Fig. 4i), Cyclotella litoralis (Fig. 4ii), Melosira nummuloides (Fig. 4iii), Paralia sulcata (Fig. 4iv), Thalassiosira eccentrica (Fig. 4v) and Pleurosigma angulatum (Fig. 4vi) were identified as being significant for the YS, while Asterionellopsis glacialis (Figs 4vii, viii), Ch. affinis (Fig. 4xii), Ch. brevis (Fig. 4xii), Ch. constrictus (Fig. 4xii), Ch. contortus (Fig. 4ix), Ch. curvisetus (Fig. 4x), Ch. debilis (Fig. 4xi), Ch. laciniosus (Fig. 4xiii), Chaetoceros spp. (Fig. 4xiii), Detonula pumila (Fig. 4xiii), Skeletonema dohrnii-marinoii complex (Fig. 4xiv), Eucampia zodiacus (Fig. 4xvii), Pseudo-nitzschia pungens (Fig. 4xviii), Thalassionema curviseriata (Fig. 4xv), Thalassiosira curviseriata (Fig. 4xv), and T. nordenskioeldii (Fig. 4xvi) were identified as being significant for the SS. Achnanthes spp. (Figs 4xix, xx), Entomoneis paludosa (Fig. 4xxiv), Licmophora grandis (Fig. 4xxi), L. paradoxa (Fig. 4xxii), L. paradoxa and Odontella aurita (Fig. 4xxiii) were identified as being significant for the SES, with Ch. radicans (Fig. 4xxv), Corethron pennatum (Fig. 4xxvi), Coscinodiscus centralis (Fig. 4xxvii), L. ehrenbergii (Fig. 4xxx), Porosira glacialis (Fig. 4xxiv), Thalassiosira pacifica. (Fig. 4xxviii) and Thalassiosira pacifica (Fig. 4xxviii) identified as being significant for the NES.

The three unique groups also had indicator species: Amphora spp., Bacillaria paxillifera, Grammatophora marina, Lauderia annulata, L. flabellata, Navicula elegans, and Tabularia fasciculata species, were identified for Unique Group 1 and L. debilis and Licmophora spp., representing Unique Group 2. An indicator value was not calculated for Unique Group 3 as only one sample (Site 105) was included, but although an indicator was not calculated here, we noted that the nearby Site 105 was strongly dominated by Ch. curvisetus.

The Korean coastal waters were able to be divided into four groups based on the diatom assemblage data by clustering analysis, and each group is geographically separated, namely YS, SS, SES, and NES. Overall, the data indicated that differences in the geographic, physical, and environmental factors affected diatom assemblages in each coastal area. In other words, this result implies that the diatom assemblages reflect the characteristics of Korean coastal waters, and can be used as a bioindicator for defining the eco-regions of Korean water based on the diatoms. We sampled during winter, and the water temperature did not exceed the 13°C seasonal average for Korean waters. Despite lower water temperature of the sampling sites, the diatom species were comprised to the common coastal species in temperate regions, except for northern part of the East Sea (Table S2). The diatom occurrences in Korean coastal waters seem to be influenced by the salinity and tidal current rather than the water temperature. For example, although the water temperatures in the YS were low (0.2–5.7°C), the diatom species found were comprised to the frequently occur year-round species such as Actinoptychus senarius, Paralia sulcata and Cyclotella litoralis. Although we did not analyze inorganic nutrients in the present study, previous studies reported that rivers supply inorganic nutrients to both coastal upwelling and tidal fronts in the YS and SS, and the nutrient concentrations are associated with increased diatom community abundance (Lim et al., 2012; Yeo and Kang, 1998). Warm currents from Tsushima (a branch of the Kuroshio warm currents) continuously flow into the SS which also experiences coastal upwelling and front structures. Fronts in the YS and SS occur at the boundary between turbid coastal waters and stratified offshore waters leading to high levels of phytoplankton abundance and primary production (Choi, 1991; Seung et al., 1990). Jung et al. (2013) reported that in the YS and SS, inorganic nutrient concentrations (dissolved nitrogen, dissolved phosphorus, and dissolved silica) were three times higher than those measured for the ES, and such differences probably played a major role in increasing diatom abundance. The low diatom abundance for the ES, which has the characteristics of the open sea, may be related to a low inorganic nutrient flux caused by the area’s deep water and lack of river mixing inputs (Kang et al., 2004; Kim et al., 2001; Yun et al., 2004). The patterns in these results with lower abundance in the ES reflected previous studies (e.g., Kang et al., 2004, 2005). Most diversity studies in marine ecosystems have focused on revealing the community composition, in relation to habitats, depth within the water column or biogeographical patterns (DeLong et al., 2006; Galand et al., 2009). In the present study, the diversity and evenness of diatom community in the NES, suggesting a comparatively high diversity (Table 1): this may be affected by mixing between cold and warm water currents. In the present study, diatom assemblage may be most influenced by current, such as Porosira glacialis and Coscinodiscus centralis known to be typical polar diatom (Villareal and Fryxell, 1983) and cosmopolitan (Halse and Syvertsen, 1996), respectively.

We performed diatom indicator species analyses to identify the meaningful species from four diatom-based groups in the coastal waters of R. O. Korea. This analysis was used to identify diatom species that reflected the geographic and seasonal characteristics of each group. In the YS group, the indicator species included tychoplanktonic diatoms known to be typically benthic and brackish water species (Table 2). Paralia sulcata, Actinoptychus senarius, Pleurosigma angulatum, Cyclotella litoralis, and Asteroplanus karianus were selected as indicator species here. Paralia sulcata is an environmental indicator for vertically well-mixed water due to its tychoplanktonic nature (McQuoid and Nordberg, 2003); this species is typically found in regions with frequent upwelling, and its abundance is correlated with high nutrient concentrations (Abrantes, 1988). Pleurosigma angulatum is a typical epipelic intertidal diatom which overcomes its severe environment by undergoing periodic vertical migrations coincident with the tides and light (Happey-Wood and Jones, 1988). Cyclotella litoralis is a euryhaline diatom frequently found in the estuaries of YS (Park et al., 2013), while Asteroplanus karianus can rapidly uptake nutrients and form very large blooms (Yamaguchi et al., 2014), and so nutrient inputs from sediments and rivers draining into YS provide a suitable habitat for this species. In summary, the characteristics of the YS indicator species have straightforward explanations including a liking for areas exhibiting well-established vertical mixing of the water column, freshwater inputs, and nutrient enrichment. The water column in winter is well-mixed vertically by tidal currents and winds in the YS; the NW monsoon cools the surface water and the water temperature vertical gradient becomes more uniform (Roelofs, 1984; Hobson and McQuoid, 1997; Gebühr et al., 2009). Vertical mixing in the water column provides nutrient enrichment from the sediment to the surface and gives benthic and tychoplanktonic diatoms a chance to access both nutrients and light. The Han River and Keum River (two of the five largest rivers in R. O. Korea) drain into the YS region providing additional nutrient enrichment (Koh and Khim, 2014; Wang et al., 2003).

The indicator species in the SS group were colony and chain-forming diatoms, such as Asterionellopsis glacialis, Chaetoceros spp. and Eucampia zodiacus, and Thalassiosirales such as Detonula pumila, Skeletonema dohrnii-marinoii complex, Thalassiosira curviseriata, and T. nordenskioeldii. The SS coastline of R. O. Korea is geologically a ria with many bays and islands, and is physically influenced by south-western, wind-driven currents, tidal currents, and the Tsushima Warm Current. These physical factors encourage the SS sediment re-suspension and the geologically complex coastline emphasizes tidal effects in bays (Bae and Kim, 2012). Several theories have been advanced to explain the success of chain-forming diatoms with the advantages of chain formation being reported as including beneficial responses to physical, chemical and biological constraints (Musielak et al., 2009; Peters et al., 2006; Bjærke et al., 2015). Recently, turbulence shear has been reported as enhancing nutrient uptake in chain-forming diatoms (Bergkvist et al., 2018), and although the type of turbulence in the SS was not studied, its physical and geological characteristics cause continuous turbulence, which may well be increasing the nutrient availability for chain-forming diatoms.

The ES has a coastal terrace with a simple, linear coastline, where diatom distribution is mainly affected by two major currents—the S-trending Liman Current, and the N-trending Tsushima Current. These currents meet and form the subpolar fronts and mesoscale eddies that influence phytoplankton community structure and distribution in this area (Choi et al., 2016). In our study, the ES diatom-based regions were divided into distinct southern and northern groups. In the SES, the indicator species were estuarine, stalk-forming diatoms, such as L. grandis, L. paradoxa, Achnanthes spp, and Odontella aurita. Licmophora is a diatom that is usually found submerged in rock pools throughout the littoral zone (Honeywill, 1998). Its species are known to survive on various substrates, including sediments, rocks, microalgae, vertebrates, and ice. The stalk attachment has to be strong to survive intertidal forces and wave action, as well as being pounded against rocks and macroalgae, although in sub-optimal conditions cells can easily become dislodged (Honeywill, 1998). In a recent study, Licmophora grew well and successfully formed colonies under experimental high light intensities and low turbulence, and their growth rate showed no relationship to nutrients (Ravizza and Hallegraeff, 2015). Currently, no clear explanation exists for the significant presence of attached diatoms such as Licmophora species, including L. grandis and L. paradoxa, in the water column. The occurrence of Licmophora species and Odontella aurita as indicator species in the SES might be related to the extensive presence of the massive macroalgae habitats which are preferred by attached diatoms (Jeong et al., 2014). It is also likely that the continuous effect of the Tsushima current may cause diatoms to become detached from substrates and suspended in the water column (Kooistra et al., 2009).

The selected indicator species in the NES included psychrophilic diatoms such as Corethron pennatum, Coscinodiscus centralis, Porosira glacialis, and Thalassiosira pacifica. Coscinodiscus centralis is a large, centric diatom known to be cosmopolitan (Halse and Syvertsen, 1996) even occurring in the Arctic region (Lovejoy et al., 2002; Duerksen et al., 2014). Porosira glacialis is a typical polar diatom (Villareal and Fryxell, 1983), while Thalassiosira pacifica has been reported from cold to tropical regions (Park et al., 2016a), although most frequently occurring in the Arctic (Joo et al., 2012). In the NES, the Liman Cold Current has a more undiluted impact, while cold water and warm water probably differ in their effects. In winter, the Liman Cold Current flows strongly southward, and the indicator species are those related to the temperature transitions associated with the Liman Cold Current (Yun et al., 2004). Specifically, our work indicated that the NES region was probably characterized by the action of the Liman Cold Current weakening the Tsushima Warm Current (Kim and Min, 2008) suggesting that this region’s diatom assemblage may be most influenced by current, rather than by geography.