Sample collection and preparation

The samples used for this study were collected during marine geological surveys (GH cruises) conducted during 1977–1996 by the Geological Survey of Japan. Numerous surface sediment samples were collected during these cruises, mostly from the nearshore shelf to slope regions extending seaward from the Japanese island masses. For this study, we selected fine-grained sediments from 69 sites, eliminating sand and gravel samples, covering a wide depth range from 122 m to 3658 m (Figure 1c, Table 1). All sites were within the region where the annual mean SSTs are between 9°C and 20°C. The selected samples consisted of clay, silt, and sandy silt from the uppermost 2 cm of the subsurface sediments, and they had been recovered by using piston/gravity corers or grab/box samplers. In the laboratory, all samples were first treated with a 10–30% solution of H2O2 to remove organic matter, and then HCl was added to dissolve calcareous components. Disaggregated particles were wet-sieved through a 63 m mesh sieve. The residue on the sieve was air dried and then strewn on a glass slide and mounted with Entellan new or Canada balsam medium. At least 300 radiolarian specimens were counted in each sample. Samples were not measured for volume or weight, so absolute abundances are not known.

Figure 1.

Index map (a) of the Japan Sea showing seafloor topography and oceanic currents. Vertical water structure (b). Surface sediment sampling locations (c) indicated by dots labeled 1 to 69 and annual mean sea surface temperature (°C: contours) (Japan Oceanographic Data Center, 1975).

Table 1.

List of surface sediment samples. References: 1, Honza (1978a); 2, Honza (1978b); 3, Honza (1979); 4, Ikehara and Kawahata (1986); 5, Ikehara et al. (1987); 6, Katayama and Ikehara (1988); 7, Katayama (1989); 8, Katayama et al. (1991); 9, Nakajima and Katayama (1992); 10, Nakajima et al. (1993); 11, Inouchi et al. (1995); 12, Inouchi et al. (1996); 13, Katayama et al. (1997).

Continued.

Taxonomy

About 100 taxa were recorded in the 69 studied samples (listed in the Appendix). Most of these were previously described and illustrated by Nigrini and Moore (1979), Itaki (2003, 2009), Motoyama and Nishimura (2005), Itaki et al. (2008), or Kamikuri et al. (2008) (Table 2) and we used these studies as the basis of our species identification. Although Ogane et al. (2009) have synonymized Euchitonia with Dictyocoryne, we herein distinguish Dictyocoryne spp. and Euchitonia spp. In addition, we were unable to consider the taxonomy of Matsuzaki et al. (2015), based on Pleistocene radiolarians from a deep-sea core collected near northeastern Japan, because their monograph was published after our observations were completed. Thus, in this paper, we essentially follow the high-level classification system for radiolarians proposed by De Wever et al. (2001) (Table 2).

Cluster analysis

Cluster analysis is a powerful tool for summarizing multivariate microfossil raw abundance data. We used the SPSS statistical software package (SPSS Inc., 2007) and ran R-mode and Q-mode cluster analyses of the census data in order to classify species and samples, respectively, into groups. R-mode analysis classifies variables (species in this case) into groups on the basis of their similarities, and the Q-mode technique focuses on interrelations between objects (samples in this case).

Before computation, we grouped Dictyocoryne profunda, D. truncatum and Dictyocoryne spp. together into a Dictyocoryne group, because Dictyocoryne spp. included juvenile forms of the former two species. Likewise, we grouped Euchitonia elegans, E. furcata and Euchitonia spp. into a Euchitonia group due to taxonomic difficulties. Adult and juvenile forms of Actinomma leptodermum were also grouped. In addition, we omitted from our analyses rare taxa (those whose relative abundance was never greater than 2%), as well as radiolarian taxa that could not be classified to the species or genus level. Then we used the remaining 32 taxa (Figures 2, 3) for the statistical cluster analyses.

In the R- and Q-mode cluster analyses of the population data (69 samples × 32 taxa), we used the cosine theta coefficient as the similarity measure and applied the average linkage method.

Regional distribution of taxa

The R-mode computation produced four distinct clusters, A, B, C, and D, when a cosine theta coefficient of 0.4 was used as the cutoff value (Figure 4). These clusters and their distributions are described below, in the order shown in Figure 4.

Cluster A (Tsushima Current assemblage).—This cluster consists of 24 taxa with high abundances in the southern part of the Japan Sea and its eastern coastal area (Figure 5a–p). The abundance of most taxa tended to decrease from south to north or from east to west, and at some northwestern sample sites most taxa belonging to this cluster were absent. Cluster A is subdivided into subclusters A1, A2 and A3 (Figure 4). Subcluster A1 comprises seven taxa, all of which are spumellarians/ entactinarians. Subcluster A2 consists of 12 taxa, of which 8 belong to the plagiacanthids (3 spicular forms and 5 two-segmented forms). We found no obvious differences in the distribution patterns of subclusters A1 (Figure 5a–e) and A2 (Figure 5f–k). However, subcluster A1 taxa showed relatively stronger abundance gradients from south to north, whereas in the components of subcluster A2, east-west gradients and nearshore increases in the northern area were more apparent. Subcluster A3 was characterized by higher scores in the northeastern margin of the studied area (Figure 5l–p).

Table 2.

Species list. The high-level taxonomy basically follows De Wever et al. (2001).

Continued.

Continued.

Continued.

Most taxa of subcluster A1 were restricted mainly to the Tsushima Current region and they were absent from the northwestern part of the Japan Sea, north of the polar front. The relative abundance of the Tetrapyle octacantha group was 5% to 30% in the southern part of the Japan Sea (Figure 5a), with a maximum abundance of 43.9% in the Tsushima Strait. This group is known to comprise tropical to subtropical species living in the surface layer (Kling, 1979; Ishitani et al., 2008), and it never exceeds 20% in relative abundance in open ocean sediments (Lombari and Boden, 1985; Motoyama and Nishimura, 2005; Kamikuri et al., 2008). Its exceptionally high abundance in the Tsushima Strait probably reflects the existence, in this region, of a shallow water assemblage from which subsurface and deeper dwelling species have been mainly excluded. In fact, T. octacantha accounts for ~30% of the radiolarian plankton fauna in the Tsushima Strait (Itaki et al., 2010). The Dictyocoryne group also reached its highest abundance (15.6%) at the southernmost sampling site (Figure 5b). Peak abundances of Didymocyrtis tetrathalamus, Euchitonia group, and juvenile L. buetschlii occurred in shallow areas close to the Noto Peninsula and the Niigata and Akita districts (Figure 5c–e), a result concordant with the findings of plankton tow studies showing that these species are surface dwellers in the North Pacific (Kling, 1979; Ishitani and Takahashi, 2007; Ishitani et al., 2008) as well as in the Japan Sea (Itaki, 2003; Kurihara et al., 2006; Ishitani et al., 2008; Kurihara and Matsuoka, 2010).

Figure 2.

a, b, Acanthosphaera spp.; a, Sample no. 41; b, Sample no. 31; c, Actinomma boreale Cleve, Sample no. 14; d–g, Actinomma leptodermum (Jørgensen); d, Sample no. 31, adult form; e, Sample no. 2, adult form; f, Sample no. 14, juvenile form; g, Sample no. 15, juvenile form; h, Actinomma langii (Dreyer), Sample no. 15; i, j, Cyrtidosphaera reticulata Haeckel; i, Sample no. 26; j, Sample no. 31; k–m, Hexacontium spp.; k, Sample no. 31; l, Sample no. 31; m, Sample no. 31; n, Spongosphaera streptacantha Haeckel, Sample no. 15; o, Didymocyrtis tetrathalamus (Haeckel), Sample no. 39; p, Spongoliva ellipsoides Popofsky, Sample no. 31; q, Euchitonia furcata Ehrenberg, Sample no. 32; r, Dictyocoryne profunda Ehrenberg, Sample no. 39; s, Euchitonia sp., Sample no. 49; t, Stylodictya multispina Haeckel, Sample no. 40; u, Dictyocoryne truncatum (Ehrenberg), Sample no. 15; v, Spongaster tetras tetras Ehrenberg, Sample no. 33; w, Dictyocoryne sp., Sample no. 26; x, Spongodiscus resurgens Ehrenberg, Sample no. 30; y–ab, Tetrapyle octacantha Müller group; y, Sample no. 41; z, Sample no. 31; aa, Sample no. 31; ab, Sample no. 15.

Figure 3.

a, b, adult forms of Larcopyle buetschlii Dreyer; a, Sample no. 15; b, Sample no. 33; c, d, juvenile forms of L. buetschlii; c, Sample no. 39; d, Sample no. 30; e, f, Lithomelissa spp.; e, Sample no. 33; f, Sample no. 15; g–j, Lophophaena spp.; g, Sample no. 31; h, Sample no. 26; i, Sample no. 31; j, Sample no. 15; k, Acanthocorys castanoides Tan and Tchang, Sample no. 33; l, Plectacantha oikiskos Jørgensen, Sample no. 11; m, Phormacantha histrix Jørgensen, Sample no. 66; n, o, Pseudocubus obeliscus Haeckel; n, Sample no. 26; o, Sample no. 15; p, q, Pseudodictyophimus gracilipes (Bailey); p, Sample no. 39; q, Sample no. 26; r, Neosemantis distephanus (Haeckel), Sample no. 15; s, Ceratocyrtis galeus (Cleve), Sample no. 39; t, u, Ceratocyrtis spp.; t, Sample no. 15; u, Sample no. 5; v, w, Lipmanella spp.; v, Sample no. 31; w, Sample no. 15; x, Acanthodesmia vinculata Müller, Sample no. 39; y, Cycladophora davisiana Ehrenberg, Sample no. 15; z, Spirocyrtis scalaris Haeckel, Sample no. 49; aa, ab, Spirocyrtis seriata (Jørgensen); aa, Sample no. 64; ab, Sample no. 15.

Chang et al. (2003) recognized three radiolarian assemblages in surface sediments of the northern East China Sea and correlated them with three surface water masses, namely, Kuroshio water, Tsushima Warm Current water (a water mass different from that of the Tsushima Current in the Japan Sea), and a mixture of shelf water and Tsushima Warm Current water. They reported that the Kuroshio water assemblage, that is, the warmest of the three assemblages recognized in that study, was characterized by Acrosphaera spinosa, Dictyocoryne profunda, D. truncatum, Lithelius minor, D. tetrathalamus (Ommatartus tetrathalamus tetrathalamus), Spongaster tetras, and Stylodictya multispina. Major components of the Tsushima Warm Current radiolarian fauna were Monozonium pachystylum and a member of the Tetrapyle octacantha group (T. quadriloba), and the mixedwater assemblage was dominated by D. tetrathalamus and the T. octacantha group (T. circularis, T. quadriloba). Most of these species are in our subcluster A1 assemblage, suggesting a strong relationship between the A1 assemblage and the East China Sea assemblages. Thus, we inferred the T. octacantha group and D. tetrathalamus (Figure 5a, c) to be of East China Sea origin. Components of the Dictyocoryne group showed similar distribution patterns, and the distribution of the group as a whole (Figure 5b) also suggests that it derived from the East China Sea because Chang et al. (2003, their table 3) reported moderately strong positive correlations of D. profunda and D. truncatum with Tsushima Warm Current water in the East China Sea, as well as because these two species are both common components of the plankton fauna in the Tsushima Strait (Itaki et al., 2010).

Figure 4.

Results of the R-mode cluster analysis of radiolarian taxa from the surface sediment samples.

The distributions of subcluster A2 components also largely coincided with the Tsushima Current region (Figures 1, 5f–k). At some sites where subcluster A1 taxa were common to abundant, however, most subcluster A2 components were absent or rare; these sites, in the Tsushima Strait (sample 40) and the nearshore area around the Noto Peninsula (samples 51, 52), are the three shallowest of the studied sites (Table 1). For example, the three subcluster A2 taxa Spongosphaera streptacantha, Pseudodictyophimus gracilipes, and Acanthocorys castanoides are mainly absent from these three sites but present at deeper sites. This distribution suggests that the preferred habitats of these species are at depths beneath the warm surface layer along the upstream to middle part of the Tsushima Current. This result is also concordant with plankton tow results from the Tsushima Current region reported by Ishitani and Takahashi (2007), who recorded that P. gracilipes and A. castanoides were rare in the upper surface layer but increased in abundance in the subsurface layer. The distributions of Pseudocubus obeliscus, P. gracilipes, and Lophophaena spp. (Figure 5g, i, k, respectively) extended northward to the nearshore Hokkaido area, where they were present with relatively high frequencies. These species seemed to be characteristic of the middle to downstream parts of the Tsushima Current. Pseudocubus obeliscus is known to be a warm surface water dweller in both the equatorial Pacific (Yamashita et al., 2002) and the Kuroshio Current region (Ishitani and Takahashi, 2007; Ishitani et al., 2008), and it occasionally reaches great abundance (ca. 50%) in plankton from the surface layer (0–35 m) of the Japan Sea during the summer (Kurihara et al., 2007). This species was not recorded in sample 40, but it is abundant in planktonic assemblages in the eastern channel of the Tsushima Strait (Itaki et al., 2010). Therefore, we inferred that P. obeliscus derives from the Kuroshio Current, prospers in the midstream of the Tsushima Current, and is transported to the far northern Japan Sea by the enhanced Tsushima Current during summer.

Figure 5.

Distribution maps of the radiolarian species in surface sediments. The contours indicate relative abundances (%). a–e, subcluster A1; f–k, subcluster A2; l–p, subcluster A3; q–v, cluster B; w, cluster C; x, cluster D.

Figure 5. Continued.

Distribution maps of the radiolarian species in surface sediments. The contours indicate relative abundances (%). a–e, subcluster A1; f–k, subcluster A2; l–p, subcluster A3; q–v, cluster B; w, cluster C; x, cluster D.

Figure 5. Continued.

Distribution maps of the radiolarian species in surface sediments. The contours indicate relative abundances (%). a–e, subcluster A1; f–k, subcluster A2; l–p, subcluster A3; q–v, cluster B; w, cluster C; x, cluster D.

Figure 5. Continued.

Distribution maps of the radiolarian species in surface sediments. The contours indicate relative abundances (%). a–e, subcluster A1; f–k, subcluster A2; l–p, subcluster A3; q–v, cluster B; w, cluster C; x, cluster D.

Figure 5. Continued.

Distribution maps of the radiolarian species in surface sediments. The contours indicate relative abundances (%). a–e, subcluster A1; f–k, subcluster A2; l–p, subcluster A3; q–v, cluster B; w, cluster C; x, cluster D.

Figure 5. Continued.

Distribution maps of the radiolarian species in surface sediments. The contours indicate relative abundances (%). a–e, subcluster A1; f–k, subcluster A2; l–p, subcluster A3; q–v, cluster B; w, cluster C; x, cluster D.

The five taxa belonging to subcluster A3, Cyrtidosphaera reticulata, Spirocyrtis scalaris, S. seriata, Lipmanella spp., and Ceratocyrtis spp. (Figure 4), were absent or present in low abundance in most of the Japan Sea (Figure 5l–p). Their abundance was greater than 5% only in its northeastern part off Hokkaido; thus, as a group, they seemed to prefer the most downstream part of the Tsushima Current. Even though S. seriata was not found in sample 40, this species probably enters the Japan Sea from the East China Sea, because it is rare to common in the water column of the Tsushima Strait (Itaki et al., 2010). Kurihara et al. (2006) and Kurihara and Matsuoka (2010) considered C. reticulata to be a deep cold water species owing to its absence in surface plankton collected in summer near Sado Island. Similarly, a sediment trap experiment revealed that C. reticulata increased in spring and decreased to near absence during summer to autumn in the Tsugaru Current, which is a downstream extension of the Tsushima Current (Itaki et al., 2008). In our results (Figure 5l), the distribution of this species showed a strong relationship to the warm Tsushima Current and seemed to prefer the downstream portion of the current. These observations suggest that C. reticulata prefers moderately chilled warm water rather than cold water. Further, this species probably inhabits the cooler subsurface layer, beneath the relatively warmer surface layer, in the upper to midstream parts of the Tsushima Current (Kurihara et al., 2006).

Cluster B (cool-water/deep-water assemblage).—Five of the six species in this assemblage were distributed dominantly in central to north-central parts of the Japan Sea (Figure 5q–u), and the assemblage appeared to be characteristic of deep-water environments (Figure 6). Actinomma leptodermum (adult form) was mainly absent from samples collected from sites at water depths shallower than 1000 m, and its abundance gradually increased with the depth of the sampling site, up to 30% at 3300 m depth (Figure 6a). A similar trend is seen in the case of juvenile A. leptodermum (Figure 6b). A. boreale was limited to a depth range below 1500 m, and its relative abundance increased to 30% at 3600 m (Figure 6c). Itaki (2003), who investigated the depth-related distributions of radiolarian species in the Japan Sea by using plankton-net sampling (0–2000 m water depth) off Hokkaido and surface sediments (60–3653 m), reported that A. boreale (his A. boreale is equivalent to our A. boreale and A. leptodermum) was completely absent in plankton collected from 0–1000 m water depth and was mainly found in surface sediments below 1000 m water depth while dominating assemblages below 2000 m (up to 40% at the deepest site). Itaki's findings together with our results indicate that A. leptodermum and A. boreale definitely prefer the deep to bottom waters of the JSPW. A. langii also seems to be a deep dweller because it occurred mainly in bottom water below 2000 m (Figure 6d).

The relative abundance of adult Larcopyle buetschlii was high at most sites, accounting for more than 10% of the radiolarians, and this species was dominant in the northern portion of the study area (Figure 5u). Its abundance clearly increased in deeper zones, often exceeding 50% at depths below 1500 m, and was never lower than 20% at bottom-water depths (Figure 6e). The abundance of juvenile L. buetschlii, however, did not show any marked vertical distribution pattern, and it occurred at frequencies of 2–18% in all but two of the samples (Figure 6f). Depth profiles of plankton absolute abundance reported by Itaki (2003) show that the abundance mode of living juvenile L. buetschlii is at subsurface depths (40–120 m), and that of living adult L. buetschlii is at subsurface to intermediate depths (40–300 m). At deeper depths down to 2000 m the juveniles are very rare and adults are rare to very rare. Thus, juvenile and adult L. buetschlii may prefer subsurface to intermediate depths (40–300 m), whereas adult L. buetschlii is predominant among plankton radiolarian fauna in the upper JSPW (300–1000 m), where Itaki (2003) reported relative abundances of 60 to 80%. These findings of Itaki (2003) indicate that in the Japan Sea adult L. buetschlii is both a shallow-water dweller (<300 m) and a characteristic component of deep assemblages. Moreover, they are consistent with our data showing an overall prevalence in the Japan Sea, and frequent dominance in deep-water assemblages (Figure 5u, 6e).

Cycladophora davisiana was rare or absent in the samples collected at depths above 500 m, and its relative abundance was greater than 10% at most sites below 2000 m (Figure 6g). In Itaki's (2003) plankton-net survey, living specimens of this species were restricted to depths below 500 m in the northern Japan Sea and its abundance was high (80% of the living assemblage) at depths of 1000–2000 m. Our results are consistent with these observations and suggest that this species indeed prefers deep-water habitats.

Figure 6.

Vertical abundance profiles of radiolarian species in surface sediments. a, Actinomma leptodermum (adult); b, A. leptodermum (juvenile); c, A. boreale; d, A. langii; e, Larcopyle buetschlii (adult); f, L. buetschlii (juvenile); g, Cycladophora davisiana. The horizontal axis represents the relative abundance (%) of species. The vertical axis shows the water depth of the sampling site.

In our R-mode cluster analysis results, Spongodiscus resurgens is associated with adult Larcopyle buetschlii (Figure 4). However, this species was most abundant in the rather shallow area off Hokkaido (Figure 5v). S. resurgens is certainly a surface to subsurface dweller (Itaki, 2003), and in the North Pacific it is a subarctic element (Kamikuri et al., 2008). Therefore, the higher abundances that we observed at relatively shallow sites in the northern Japan Sea are plausible. Because S. resurgens was relatively rare at some stations where adult L. buetschlii was most abundant (>50%) (compare Figure 5u and 5v), we are a little dubious about the cluster result that associates the two species. Both S. resurgens and adult L. buetschlii were found at most of the stations in the Japan Sea, unlike most of the other nominated taxa which were absent from many stations (Figure 5d, f, g–p, r, w, x). These similar/dissimilar distributions might explain the computation result. The rarity of S. resurgens in the Tsushima Current region south of the polar front is compatible with the reported small abundances of this species in subtropical to tropical regions of the North Pacific (Kamikuri et al., 2008).

These observations suggest that cluster B species, except Spongodiscus resurgens, are particularly well adapted to deep-water conditions in the Japan Sea. Even though adult Larcopyle buetschlii and Cycladophora davisiana prefer deep water below 500 m, they also occur in low abundance (<1 specimen per liter of water) in the plankton fauna (Itaki, 2003). Therefore, the high relative abundances of these deep dwellers that we observed in the sediments were not due to their high productivity in deep water but were likely due to the accumulation in the sediments of specimens living at low density throughout the water column. Thus, the integration of these low abundances over a water depth range of several hundreds to about 3000 m caused the observed high abundances in the sediments.

Figure 7.

Results of the Q-mode cluster analysis of the samples, carried out using 32 selected taxa (left), and the faunal composition (R-mode clusters) of each sample (right).

Cluster C.—This cluster consists of one species, Plectacantha oikiskos, which was absent or very rare in almost all of the studied region. Its relative abundance was highest (3.1%) in the southern Japan Sea off Korea (Figure 5w). This species was present with low abundance in the plankton collected from surface waters near Sado Island in September 2001 (Kurihara and Matsuoka, 2005), June 2005 (Kurihara et al., 2006), September 2005 (Kurihara et al., 2007), and June 2007 (Kurihara et al., 2008). P. oikiskos has a thin spicular form that would easily dissolve in most settings, so it is likely that this species lives in the Tsushima Current region at least during spring and summer, but that most individuals are not preserved in the sediments. Another possible reason for its near absence in sediments is mechanical destruction of the skeletons during sample preparation, which would result in their being washed through the 63 µm sieve openings. In addition, we cannot explain the occurrence of P. oikiskos in sediments of the southern and northern Japan Sea, because there are no plankton data for this species off Korea or west of Hokkaido. For these reasons, we were unable to characterize the ecology of this cluster.

Cluster D (Tsushima Strait assemblage).—Stylodictya multispina is the sole species in this cluster. It reached its highest abundance (3.7%) in sample 40 and was mostly absent from the rest of the studied samples (Figure 5x).

Regional distribution of assemblages

Q-mode cluster analysis produced three distinct clusters: S, T, and U (Figure 7). The distribution map of the Q-mode clusters shows that cluster S samples were restricted to the northern part of the Japan Sea, cluster U samples were from the southern area and along the Japanese coast, and the single sample in cluster T was collected in the central Japan Sea (Figure 8). Therefore, we identified two large radiolarian biogeographic provinces in the Japan Sea: a warm Tsushima Current province, represented by cluster U, and a cold province, represented by cluster S.

In terms of their geographical distribution, the assemblages represented by R-mode clusters A to D appear to be closely related to Q-mode clusters S to U (Figure 7). Samples belonging to cluster S are characterized by relatively low abundances of subclusters A1 and A2 assemblages (mostly <50%) and high abundances of the cluster B assemblage (mostly >40%). Within cluster S, the abundance of the subcluster A3 assemblage is moderately high (ca. 35%) only in samples 5 and 6, which were collected off Hokkaido. The dominance of the deepdwelling cluster B assemblage in cluster S suggests that the Q-mode results reflect not only surface environments but also deep environments. For example, the distribution of cluster S closely overlaps the areas in which Actinomma leptodermum and adult Larcopyle buetschlii occurred at frequencies greater than 2% and 20%, respectively (Figures 5s, u, 8).

Figure 8.

Distribution of clusters S to U obtained by the Q-mode analysis performed using the data set of the 32 selected radiolarian taxa.

Cluster U contains samples with high abundances of cluster A species (mostly >70%) and relatively low abundances of cluster B species (<30%). Cluster U is distributed within the Tsushima Current region, especially in the near-shore area.

Cluster T includes only sample 14, collected from the central Japan Sea. This sample consisted dominantly of cluster B components (92%), perhaps reflecting the great depth (3230 m) of the sample site and the overlying cold surface water.

As we have suggested above, the Q-mode analysis clusters probably reflect not only surface water masses but also deep-water environments. To investigate the relationships between radiolarian assemblages and surface water masses, we performed an additional Q-mode analysis after excluding deep dwellers (all taxa of cluster B except Spongodiscus resurgens) from the data set and using the same methods as before (cosine similarity, average linkage). This second analysis produced three distinct clusters, V, W, and X (Figure 9). Cluster W samples were distributed in the northern Japan Sea and cluster X samples were dominantly from the southern Japan Sea (Figure 10). Cluster V consists of only two samples (samples 5 and 6), collected off Hokkaido. The northern limit of the cluster X distribution is approximated by the 40°N parallel, and, thus, seems to demarcate well the border between the Tsushima Current and the Liman Current regions. It is noteworthy that sample 14 no longer stands alone as an independent cluster but is part of cluster W. Cluster V clearly represents inshore assemblages in the northern part of the Japan Sea (Figure 10), and it also corresponds to the most downstream part of the Tsushima Current. This second Q-mode analysis result clearly shows that there are three distinct faunal provinces in the Japan Sea, and that the radiolarian surface water assemblages represent well the surface water masses in the sea.

Figure 9.

Results of the Q-mode cluster analysis performed using 27 taxa (after exclusion of the deep dwellers).

Figure 10.

Distribution of clusters V to X obtained by the Q-mode analysis carried out with 27 taxa (excluding the deep dwellers).

Comparison with North Pacific radiolarian assemblages

The Japan Sea is located in the temperate climate region of the North Pacific. Offshore of Japan in the North Pacific, the warm Kuroshio and the cool Oyashio Currents meet and merge, forming a transitional area in the confluence zone. The surface water structure in the Japan Sea is similar. The warm Tsushima Current flows northward and meets the southward flowing cold Liman Current in a frontal zone along 40°N. The surface waters above all of our sampling sites in the Japan Sea have annual mean SSTs between 9°C and 20°C (Figure 1c). This SST range is largely equivalent to that of the transitional zone in the North Pacific. This oceanographic similarity on both sides of the Japanese islands is also interesting, because it allows us to make some inferences to explain faunal similarities and dissimilarities between the Japan Sea and the Pacific Ocean.

The 69 studied samples were collected between 35°N and 45°N latitude, a latitudinal range that corresponds to the transitional zone and the southern part of the subarctic zone in the western North Pacific (Kamikuri et al., 2008). Radiolarian assemblages from the surface sediments of the transitional zone are characterized by rare subtropical elements, including many collosphaerid species, Didymocyrtis tetrathalamus, Axoprunum stauraxonium, Ellipsoxiphus atractus, Liriospyris spp., and Stylochlamydium asteriscus (Motoyama and Nishimura, 2005; Kamikuri et al., 2008). Similarly, these species are rare in the Japan Sea; only one, D. tetrathalamus ever exceeds 2% in relative abundance. This loss of subtropical elements in the Japan Sea can be attributed to the alteration of the subtropical waters that derive from the Kuroshio Current as they pass through the northern East China Sea and the Tsushima Strait, where their temperature decreases and they become mixed with shelf waters. For example, the subtropical species Acrosphaera spinosa is abundant in Kuroshio Current water in the East China Sea, but after the bifurcation of the Kuroshio its abundance decreases northward in the Tsushima Current branch (Chang et al., 2003), and it finally disappears in the Japan Sea, being last recorded in the Tsushima Strait (sample 40) (Appendix). The case of Stylodictya multispina is similar. These findings are concordant with those of Matsuoka et al. (2001), who suggested that the lower diversity of the surface radiolarian fauna in the Japan Sea near Sado Island, in comparison with the diversified plankton fauna of the Kuroshio region near the Okinawa Islands (Matsuoka, 2009), might have resulted from a sorting effect associated with the branching off of the Tsuhima Current from the Kuroshio Current.

Some warm-water species, however, are not affected in the same way. For example, the Tetrapyle octacantha group, Didymocyrtis tetrathalamus and the Euchitonia group, which were distributed north of 40°N in the Japan Sea (Figure 5a, c, d), are distributed over a relatively wide latitudinal range north of 40°N in the North Pacific (Lombari and Boden, 1985; Kamikuri et al., 2008). These distributions suggest that these species probably have high ecological adaptability. Dictyocoryne truncatum is restricted to the tropical to subtropical province south of 30°N in the North Pacific (Lombari and Boden, 1985), but this species is found in the northern East China Sea (Chang et al., 2003), and Kurihara et al. (2006) reported that it was rare in the plankton fauna in the surface waters near Sado Island (38°N). In the present study, this species was found to be distributed as far north as 44°N in the Japan Sea (Table 2, Figure 5b), presumably as a result of transport by the Tsushima Current. Species in subcluster A2 appear to prefer coastal environments (Figure 5f–k), a result that is supported by the fact that most subcluster A2 species have not been reported to be major components in surface sediments from the open ocean (Motoyama and Nishimura, 2005; Kamikuri et al., 2008).

In the North Pacific, the Pterocorys zancleus group (6% of the total radiolarian assemblage), Styptosphaera spumacea (5%), and Actinomma medianum (4%) thrive in the transitional zone (Motoyama and Nishimura, 2005; Kamikuri et al., 2008) and representatives of the first two groups (i.e., Pterocorys macroceras and Styptosphaera(?) spumacea, respectively) are characteristic of the mixed-water fauna of the near-shore area off northeastern Japan (Yamauchi, 1986), but these species were very rare to absent in the Japan Sea assemblages (Appendix). We attribute their rarity in the Japan Sea, in spite of the similarity of the hydrographic conditions to those in the North Pacific at similar latitudes, to Japan's two main islands acting as a barrier to obstruct migration of plankton directly from the Pacific Ocean to the Japan Sea.

Many North Pacific subarctic components are also absent or occur with very low abundance in the Japan Sea. Pterocanium korotnevi, Spongopyle osculosa, and Actinomma delicatulum were not found in this study, and Stylochlamydium venustum, Spongotrochus glacialis, and Stylodictya validispina were scarce. These six species are major elements of subarctic cool water environments of the North Pacific (Nigrini, 1970; Lombari and Boden, 1985; Motoyama and Nishimura, 2005; Kamikuri et al., 2008), and most of them are known to be surface/ subsurface dwellers (Tanaka and Takahashi, 2008). The absence of cool surface water radiolarian species in the Japan Sea might be attributable to the unsuitability of the higher annual SSTs there (10–20°C versus <10°C and mainly <5°C in the subarctic Pacific) for these species. A species that lives at intermediate depth in the Oyashio cool current region, Ceratospyris borealis, is also very rare in the Japan Sea.

Lithelius minor, Cycladophora cornutoides, Botryostrobus aquilonaris, C. davisiana, and Actinomma leptodermum prefer depths deeper than 500 m in the subarctic Pacific (Tanaka and Takahashi, 2008). The first three are almost completely absent from the Japan Sea (Appendix). Their scarcity is probably attributable to the island barrier. However, the last two species prosper in the Japan Sea (Figure 5s, t), possibly because the deep waters of the Japan Sea are oxygen rich, and among deep dwellers, C. davisiana and A. leptodermum (probably also A. boreale and A. langii) prefer oxygen-rich waters. C. davisiana and A. leptodermum live also in the subsurface layer in the subarctic Pacific (Tanaka and Takahashi, 2008), and C. davisiana is reported to intrude into shallow water in association with upwelling water in the Tsushima Strait (Itaki et al., 2010). These species may have intruded from outside the Japan Sea and become adapted to the oxygen-rich, cold water there. We cannot determine, however, whether their intrusion is ongoing or their presence is due to a past event. This is an interesting problem in terms of the historical development of the present fauna of the Japan Sea (Itaki, 2007, 2011), and it can be resolved by examining the fossil record (Amano et al., 2007) and the molecular phylogeny of living marine organisms around Japan.