Middle to Late Pleistocene radiolarian biostratigraphy in the water-mixed region of the Kuroshio and Oyashio currents, northeastern margin of Japan (JAMSTEC Hole 902-C9001C)

A c ontinuous Quaternary sediment sequence was recovered from Hole 902-C9001C during the D/V Chikyu 2006 mission along the northeastern margin of Japan. The age and rate of deposition of this core were estimated using calcareous nannofossil biostratigraphy and oxygen isotope curves measured from benthic foraminifera ( Uvigerina akitaensis ) and dated from 740 ka to the present, a period that spanned the Brunhes normal polarity epoch. Sediment consisted of diatomaceous siltstone and contained an abundance of radiolarians. A total of 91 radiolarian species was found in the core, of which 74 were analysed. Of these radiolarian species, 36 demonstrated continuous stratigraphical distribution over the past 740 ka and 38 had shorter ranges of biostratigraphical interest. Three of the 38 species were determined to be novel and are described in the present study ( Amphisphaera tanzhiyuani sp. nov., Schizodiscus japonicus sp. nov. and Siphonosphaera ? paraphoros sp. nov.). Based on 17 radiolarian bioevents, including four datums which have been commonly used across a wide area of the North Pacific, the radiolarian sequence of this core was divided into 8 zones: Amphirhopalum virchowii Zone (613–740 ka), Spongaster tetras irregularis Zone (516–613 ka), Cyrtidosphaera reticulata Zone (357–516 ka), Spongurus cylindricus Zone (238–357 ka), Pterocanium depressum Zone (209–238 ka), Spongoliva ellipsoides Zone (131–209 ka), Ceratospyris problematica Zone (33–131 ka), and the Acanthodesmia vinculata Zone (0–33 ka).


systeMatIc Palaeontology
Thorough examination of species composition within the 163 samples revealed a total of 91 species (3 collodarians, 51 spumellarians, and 37 nassellarians). Among these species, the stratigraphical distributions of 74 are presented in Figure 3. Forty-eight important species are illustrated (Pls 1-3) and their taxonomic references are listed in Table 1. Of these 48 species, 38 demonstrated sufficiently discontinuous distributions to be biostratigraphically useful. Among these 38 species, three are newly described in detail below.

diagnosis.
A single cortical shell relatively irregular in shape, platy surface with two types of pores.

description.
A single irregular and platy surface cortical shell bearing two types of pores, the first group comprises several rounded polygonal pores of relatively larger size with a short centrifugal tube, and the second group a small number of irregular pores throughout the platy cortical shell. A hook-like spine is present on some of the larger rounded polygonal pores. No spines or significant projections are present on the cortical shell.
occurrence. Continuous for the past 740 ka, living and fossil (this study). Haeckel, 1887 is the presence of a short centrifugal tube on each larger pore. This species is morphologically similar to members of the genus Odontosphaera Haeckel, 1887 in that a hook-like spine on some larger pores is present in this new species. However, the characteristic centrifugal tube is present on each larger pore, and thus we tentatively regard this new species as a member of Siphonosphaera.
(Pl. 2, figs 21-24) 1974 Stylatractus pyriformis (Bailey); Kruglikova: 188,190,figs 2.2,2.3 [only]. 1982 Amphistylus sp. Tan & Su: 141-142, pl. 3, fig. 10. 1984 ?Stylatractus sp. Nishimura & Yamauchi: 34, pl. 5, fig. 11.  Modified from Domitsu et al. (2011) Age (ka) Fig. 2. Age model of core 902-9001C. The depositional age of core 902-9001C was established using oxygen isotope measured from benthic foraminifera (Domitsu et al. 2011), calcareous nannofossil datums (FO of Emiliania huxleyi (270 ka) and Pseudoemiliania lacunosa (450 ka)) and tephrostratigraphy datums (Spfa-1 (43 ka) and Aso-4 (87 ka)). description. Three concentric shells with two long, cylindrical, bipolar spines. Innermost shell is a spherical microsphere with numerous radial beams. The second shell is a spherical macrosphere with numerous radial beams. The third shell is the first cortical shell with a spherical or rather oblong spherical shape. The surface of the third shell is neither rough nor smooth. The wall of the first cortical shell is thick in the mature form and the wall thickness is equal everywhere. Pores are hexagonal in shape and are arranged as six to seven pores in both the longitudinal and equatorial axes of the cortical shell. The pore frame tends to be robust, but pores always appear to be visible even in mature forms. Bipolar spines are thin, generally equal in length and are cylindrical or subcylindrical in cross-section. remarks. Amphisphaera gracilis Campbell & Clark, 1944 is similar to this morphotype, but the former differs from the latter by having significantly bladed bipolar spines. This morphotype has no similarity to Stylosphaera hispida Ehrenberg. S. hispida has only two shells, a pear-shaped macrosphere and an ellipsoid cortical shell. The pear-shaped macrosphere is connected to the cortical shell by six radial beams. This morphotype is occasionally misidentified as Stylosphaera pyriformis (Bailey), but is easily distinguished from the latter by having numerous thick radial beams between the cortical shell and the macrosphere and the presence of an ellipsoid macrosphere instead of a heteropolar macrosphere as in S. pyriformis.  Ling: 778, pl. 1, figs 9, 10. 1975 Spongodiscus sp. Ling: pl. 4, fig. 5;Ling, 1980: 368, pl. 1, fig. 7. 1980Spongodiscus sp. Sakai: 709, pl. 6, fig. 5. 2002 diagnosis. The diagnosis of this new species is fully cited in Ling (1973, p. 778). The morphological characters of this morphotype are concordant with the description of Spongodiscus sp. in Ling (1973): discoidal biconvex shell whose surface consists of an irregular network with circular to subcircular pores approximately uniform in size and a darker central part. description. Large inflated, opaque discoidal skeleton with flat sides presenting a porous structure. The pores are circular to subcircular and their sizes fluctuate between 8 and 11 µm. The central part of the disk is one-half to two-thirds of the diameter of the disc and appears darker than the rest of the shell, suggesting a thickening of the test. The diameter of this central part fluctuates from 70 to 100 µm and is also finely pored. The observed specimens were marked by an absence of spines while the siliceous microfossil preservation was good down-core. The absence of spines in the periphery of the skeleton is a strong taxonomic feature of Schizodiscus japonicus. The skeleton periphery is also marked by an absence of pylome for most of the encountered specimens; however, some specimens present a rounded tube-like aperture continuing inside the disk, close to a pylome, illustrated in Plate 2, figs 29-30.
occurrence. Middle Pleistocene only in this site.
Schizodiscus japonicus differs from S. stylotrochoides Dogiel by the absence of the short, main spine system and coronet, and differs from S. spatangus Dogiel by the absence of numerous short spines distributed in the disc periphery. S. stylotrochoides Dogiel is marked by a coronet, several relatively thick and short spines, while S. spatangus Dogiel possesses numerous fine, short spines and an absence of coronet. Schizodiscus sp. A (Pl. 2, fig.  26) is distinguishable from S. japonicus by the presence of many equatorial radial spines originating from the central part of the test, a finer test pore frame, and the presence of a thin cover over the central to medial regions of the disk. S. japonicus is completely different from Spongodiscus biconcavus Haeckel, 1887 in having very coarse pores.
The relative abundance of L. sakaii changed significantly throughout this core and provided five additional datums: three abundance peaks at 229, 207 and 61ka, a rapid increase datum at 240 ka and a rapid decrease datum at 55 ka.

radIolarIan bIoevents
In the present study 12 bioevents were identified over the past 740 ka. Of these events, nine were identified for the first time in core 902-C9001C (Fig. 5), while the remaining three events had been identified previously as useful datums within the northwestern Pacific. Table 2 summarizes the stratigraphical horizons of the 12 bioevents and their correlative numerical ages with oxygen isotope stratigraphy. The correlation of datums with other regions is described in detail below.

last occurrence datum of Axoprunum acquilonium (357 ka)
A. acquilonium has been recorded only in North Pacific sediment (e.g. Hays, 1970;, including the Bering and Japan seas (Ling, 1973) and the Sea of Okhotsk (Matul et al. 2002). Hays (1970) and Ling (1973) noted that the relative abundance of this taxon never exceeded 5% in the Lower to Middle Pleistocene intervals. The LO datum of this taxon was identified at 310 ka in the North Pacific (Hays, 1970), at around 330 ka in the northwestern Pacific  and at 329 ka in the Sea of Okhotsk (Matul et al., 2002). The LO of A. acquilonium in the present study was located between sample 23H-5, 54-60 cm (208.97 mbsf) and sample 24H1-1, 25-31 cm (213.16 mbsf) at 357 ka ( Table 2). The LO of A. acquilonium identified in the present study appeared to be roughly synchronous with the LO previously identified in the northwestern Pacific, revealing a time gap of less than 27 ka. last occurrence datum of Schizodiscus japonicus (238 ka) Schizodiscus japonicus sp. nov. has long been referred to as Spongodiscus sp. (e.g. Ling, 1973) and its LO has been recognized in the Middle Pleistocene at sites in the northwestern Pacific (Ling, 1973;Sakai, 1980;Matul et al. 2002). The numerical age of this datum was roughly estimated at 290 ka by Matul et al. (2002) in the Sea of Okhotsk. Based on benthic foraminiferal oxygen isotopic stratigraphy, the LO age at the site examined in the present study was estimated at 238 ka (Fig. 5), which revealed a time gap of 62 ka between this site and the site examined by Matul et al. (2002). These results left no doubt that the LO of S. japonicus was located in the upper Middle Pleistocene at sites in the North Pacific.

stratigraphical events of Lychnocanoma sakaii
The LO of L. sakaii was the most recent of the radiolarian bioevents identified from larger radiolarian fractions (>63 µm) in the North Pacific (Sakai, 1980). According to Sachs (unpublished thesis, Brown University, 1973), this datum served as a useful bioevent in Pleistocene deep-sea sediments. In this section the L. sakaii curve will be presented as a stratigraphical proxy based on Figure 4.

Rapid decrease datum (50 ka) and last occurrence datum (33 ka) of L. sakaii.
The FO of L. sakaii was not detected in core 902-C9001C as the base of the core is only 740 ka, and the FO of L. sakaii is recognized at 1.6 Ma by both Morley & Nigrini (1995) and Kamikuri (2010). The LO of L. sakaii was detected here in the interval between sample 3H-4, 42-48 cm (20.91 mbsf) and sample 3H-6, 50-56 cm (23.74 mbsf), which corresponds to 33 ka (Table 1). According to several previous studies (robertson, unpublished thesis, Columbia University, 1975;Morley & Nigrini, 1995;Kamikuri, 2010), the LO of this species was dated at 50 ka, while Matul et al. (2002) dated the LO of L. sakaii at 29 ka in the Sea of Okhotsk, and Morley et al. (1982) dated the LO between 16.7 and 34 ka. The LO of the species identified in the present study appeared to be similar to the LOs identified by Morley et al. (1982) and Matul et al. (2002). The long time gap between our L. sakaii LO datum of 33 ka and that identified in the records of robertson (unpublished thesis, Columbia University, 1975), Morley & Nigrini (1995) and Kamikuri (2010) at around 50 ka requires examination. This date is almost the same as that at which the present study identified a rapid decrease datum (rD, Fig. 4). This phenomenon could have been plausibly interpreted as a LO in a low resolution study, while in a high resolution study -as the present study and that of Matul et al. (2002) -the rD datum and LO appear as two distinct events.

Rapid increase datum (240 ka) and abundance peaks (61, 207 and 229 ka) of Lychnocanoma sakaii.
As shown in Figure 4, the change in relative abundance of L. sakaii was so significant that the rapid increase (rI) datum and abundance peaks (AP) of this species were potentially applicable to biostratigraphical correlation with adjacent regions. This significance had already been noted for the North Pacific by Sachs (unpublished thesis, Brown University, 1973); however, in that study, L. sakaii was identified as Lychnocanoma grande Campbell & Clark, 1944. In the present study, one rID and three APs higher than 10% of the total assemblages were identified at 240, 229, 207 and 61 ka, respectively (Fig. 4, Table 3). The highest relative abundance peak of L. sakaii during the last 740 ka (AP3) was located at 61 ka during the early MIS 4. This event was detected at 60 ka in the northwestern Pacific by robertson (unpublished thesis, Columbia University, 1975) and Sachs (unpublished thesis, Brown University, 1973), while Matul et al. (2002) recorded a high abundance peak for L. sakaii in the Sea of Okhotsk at 72 ka. L. sakaii AP3 appears to be a suitable age tie point in the northwestern Pacific.

radIolarIan ZonatIon
As documented in the previous sections, 12 bioevents were detected in 902-C9001C (Fig. 3). Among these bioevents, nine were recognized for the first time in core 902-C9001C (FOs of Spongaster tetras irregularis, Cyrtidosphaera reticulata, Amphisphaera tanzhiyuani, Pterocanium depressum, Spongoliva ellipsoides and Ceratospyris problematica; FCO of Acanthodesmia vinculata; LOs of Axoprunum acquilonium and Schizodiscus sp. A). Among these nine bioevents, seven define eight new radiolarian interval zones at the Shimokita site (Fig. 3). The LO of Schizodiscus sp. A was not selected as a zonal marker since identification proved difficult due to its strong morphological similarities with other species, although Schizodiscus japonicus is separated from other Schizodiscus species by the absence of radial spines around the disk and significant light contrast between the central part and its adjacent exterior thinner part of the disk (Dogiel & reshetnyak, 1952;Petrushevskaya, 1968). The remaining bioevents could potentially be used subordinately for refinement and confirmation of age determinations using the zones proposed in the present study.  Interval and age. The stratigraphical interval between sample 40H-10 at 55-61 cm (362 mbsf) and 35X-5 at 47-53 cm (316 mbsf) in the 902-C9001C core. This zone covers the period from the core base to 613 ka.

Spongaster tetras irregularis Interval Zone (613-516 ka)
Definition. Base of zone defined by FO of Spongaster tetras irregularis. Top of zone defined by base of Cyrtidosphaera reticulate Interval zone.
Faunal character. Assemblage is marked by the continuous occurrence of Tetrapyle octacantha, Larcopyle buetschlii, Stylochlamydium? venustum and Cycladophora davisiana, as in the previous zone. The FOs of Spongosphaera streptacantha and Eucecryphalus cervus are placed in this zone.

Cyrtidosphaera reticulata Interval Zone (516-357 ka)
Definition. Base of zone defined by the FO of Cyrtidosphaera reticulata. Top of zone defined by the base of the Spongurus cylindricus Interval zone.
Faunal character. The assemblage is marked by the continuous occurrence of Tetrapyle octacantha, Larcopyle buetschlii, Stylochlamydium? venustum and Cycladophora davisiana from the Amphirhopalum virchowii zone. The FO of Amphisphaera tanzhiyuani sp. nov (451 ka) is placed in this zone, but it cannot be used as a biostratigraphical marker because of limited geographical coverage.

Spongurus cylindricus Interval Zone (357-238 ka)
Definition. Base of zone defined by the LO of Axoprunum acquilonium. Top of zone defined by the base of Pterocanium depressum Interval zone.
Faunal character. The characteristic species in this zone are the same as in the Spongaster tetras irregularis zone. In addition, the first occurrence (FO) of Pterocanium depressum (259 ka) is the key marker of this zone.

Pterocanium depressum
Remarks. The last occurrence of Amphirhopalum virchowii is not the true extinction event world-wide, because this species is extant based on the type locality of specimens collected from modern seawater ).

Spongoliva ellipsoides Interval Zone (209-131 ka)
Definition. Base of zone defined by the FO of Spongoliva ellipsoides. Top of zone defined by base of Ceratospyris problematica Interval zone.
Faunal character. The assemblage is characterized by the same species as in the previous zones, except for the occurrence of Cleveiplegma boreale, Spongoliva ellipsoides and Pylodiscus triangularis (Fig. 3).

Ceratospyris problematica Interval Zone (131-33 ka)
Definition. Base of zone is defined by the FO of Ceratospyris problematica. Top of zone defined by the base of the Acanthodesmia vinculata Interval zone. Interval and age. The stratigraphical interval between sample 8H-6 at 42-48 cm (70.85 mbsf) and 3H-4 at 42-48 cm (20.91 mbsf) of the 902-C9001C core. This zone covers the period from 131 to 33 ka.
Remarks. Ceratospyris problematica has not been identified from other regions previously, except for the original description by Petrushevskaya (1969). we confirm this species here and its stratigraphical range (Fig. 5).

Acanthodesmia vinculata Interval Zone (33-0 ka)
Definition. Base of zone defined by the LO of Lychnocanoma sakaii. Top of zone not defined -top of the core (0 mbsf).
Faunal character. All the species that occur in the previous zone are found in this zone. The FCO of Acanthodesmia vinculata (17 ka) and the reappearance of Amphisphaera tanzhiyuani sp. nov. occur in this interval zone.
Interval and age. The stratigraphical interval between sample 3H-4 at 42-48 cm (20.91 mbsf) and the top of core 902-C9001C (0 mbsf). This zone covers the period from 33 to 0 ka.

correlatIon oF radIolarIan bIoevents In the northwestern PacIFIc sInce the MIddle PleIstocene
Based on thorough examination of the distribution of radiolarian species within the 902-C9001C core, eight radiolarian regional zones are proposed for the past 740 ka within the Shimokita region (northwestern Pacific off the east coast of northern Japan). As seven bioevents in this region were first recognized in the present study, direct comparison with previously established zones in the North Pacific was necessary to locate these new radiolarian zones against the North Pacific standard radiolarian zonation for the Quaternary period established by Kamikuri et al. (2004Kamikuri et al. ( , 2007 and Motoyama et al. (2004). These studies proposed a Neogene radiolarian biostratigraphy for the North Pacific in which the Middle to Upper Pleistocene comprised only two radiolarian zones: the Stylatractus universus zone and the Botryostrobus aquilonaris zone, which were defined by the LO of Eucyrtidium matuyamai at 1050 ka and by the LO of S. universus at 430 ka, respectively (e.g. Motoyama et al. 2004;Fig. 5). In the present study, a major problem regarding these last datums was identified, as S. universus was not recorded at the study site, despite the fact that the LO of S. universus has been established as the tie point in the Middle Pleistocene within the North Pacific. This absence indicates that this datum is not always applicable in regions within the North Pacific, and that updates to the North Pacific radiolarian zonation are necessary in order to determine the regional suitability of each datum. Figure 5 shows the updates in which the radiolarian zones identified in the present study have been compared to the East Japan radiolarian zonations (after Motoyama et al. 2004;Kamikuri et al. 2007) and the Sea of Okhotsk radiolarian zonation (after Matul et al. 2002).
Sea of Okhotsk. The A. acquilonium zone correlated with three of the radiolarian zones identified in the present study. An A. virchowii zone (c. 613 ka), which would cover the lower part of the A. acquilonium zone, as defined by Matul et al. (2002), could serve to cover this interval. The middle part of the A. acquilonium zone established by Matul et al. (2002) correlates with the S. tetras irregularis zone (613-516 ka) herein (Fig. 5). The C. reticulata zone (516-357 ka) covers the top of the A. acquilonium zone of Matul et al. (2002). The LO of A. acquilonium determined at our study site (357 ka) was relatively synchronous with the LO of the same species observed in the Sea of Okhotsk (330 ka). The B. aquilonaris zone of Motoyama et al. (2004) and defined by the LO of S. universus at 430 ka correlates with four radiolarian zones in the Sea of Okhotsk (Matul et al. 2002); in the present study the B. acquilonaris zone correlates with five radiolarian zones. The base of the B. acquilonaris zone correlated with the Spongodiscus sp. zone established by Matul et al. (2002), which was defined by the LO of A. acquilonium (330 ka) (Base) and the LO of Spongodiscus sp. (Schizodiscus japonica sp. nov.) at 290 ka (Top). The Spongodiscus sp. zone of Matul et al. (2002) correlates with the Spongurus cyclindrica zone (357-259 ka) newly established in the present study, which was defined by the interval between the LO of A. acquilonium (357 ka) and the base of the P. depressum zone (238 ka) (Fig. 5). The middle part of the B. acquilonaris zone correlates with the Amphimelissa setosa zone established by Matul et al. (2002) in the Sea of Okhotsk as defined by the LO of Spongodiscus sp. at 290 ka and the LO of A. setosa at 72 ka (Fig. 5). This only correlated with the zones identified in the present study in an indirect manner, as A. setosa was not included in our analysis due to its small size (<60 µm). Based on the numerical ages of Domitsu et al. (2011), it was determined that the A. setosa zone would correlate with the P. depressum zone (238-209 ka) in C9001C. The base of the P. depressum zone was defined by the LO of Schizodiscus japonicus (238 ka). A time gap of nearly 50 ka existed between the LOs of S. japonicus sp. nov. (Spongodiscus sp. in Matul et al. 2002) determined at the present study site and in the Sea of Okhotsk (Matul et al. 2002). This time gap demonstrates that the LOs of S. japonicus sp. nov. at these two sites were not synchronous. The middle part of the A. setosa zone in the Sea of Okhotsk was correlated with the S. ellipsoides zone (209-131 ka) identified in the present study as defined by the FO of S. ellipsoides (209 ka).
The upper part of the B. acquilonaris zone  correlates with the L. sakaii zone in the Sea of Okhotsk, which was defined by the LOs of A. setosa at 72 ka and L. sakaii at 28 ka (Matul et al. 2002),. This zone also correlates with the upper part of the C. problematica zone (131-33 ka), which was defined in the present study by the FO of C. problematica at 131 ka and LO of L. sakaii at 33 ka. These results demonstrate that the LO of L. sakaii appears to be synchronous in both the Shimokita region and the Sea of Okhotsk within a time gap of 5 ka.

conclusIons
In the present study, a high-resolution Middle to Upper Pleistocene radiolarian biostratigraphical scheme is established for the northwestern Pacific. we confirmed continuous stratigraphical occurrences of 74 radiolarian species and identified 38 of these as useful for biostratigraphical purposes. Furthermore, three new species were described (Amphisphaera tanzhiyuani sp. nov., Schizodiscus japonicus sp. nov. and Siphonosphaera? paraphoros sp. nov.). A total of 12 bioevents were identified during the past 740 ka and were used in the establishment of eight new Pleistocene radiolarian zones. Dates were based on the age model of Domitsu et al. (2011). Of the 12 events identified, 12 had never been previously identified in the northwestern Pacific. The newly established radiolarian zones were compared with zonations previously proposed in the northwestern Pacific and the Sea of Okhotsk by Matul et al. (2002), Motoyama et al. (2004) and Kamikuri et al. (2007). It was noted that the well-known Stylatractus universus LO, which has served as one of the most synchronous datum levels throughout the North Pacific, could not be used, as the species was absent in both the Shimokita site and the Sea of Okhotsk. This suggested that these sites were located outside the northwestern distribution limit of S. universus. Therefore, it was determined that the LO of S. universus was not a suitable datum for use in the western margin of the northwestern Pacific in North Japan and the Sea of Okhotsk. Instead of the LO of S. universus, the following three datums were determined to be important with respect to the Pleistocene: the LOs of A. aquilonium (357 ka), S. japonicus (238 ka) and L. sakaii (33 ka). Additionally, the abundance curve of L. sakaii was identified as a potentially useful stratigraphical tool within Middle Pleistocene sediments in the northwestern Pacific due to its rapid decrease datum (rD; 55 ka), rapid increase datum (rI; 240 ka) and three abundance peaks (229 ka (AP1), 207 ka (AP2) and 61 ka (AP3)), which could serve as synchronous events within the region.

acKnowledgeMents
The authors thank Drs K. r. Bjørklund and J. Gregory for their helpful reviews and the Japan Agency for Marine-Earth Science and Technology (JAMSTEC) for providing the samples. This research was supported by the Japan Society for the Promotion of Science, Tohoku University International Advanced research and Education Organization and Global Center of Excellence Program on Global Education and research Center for Earth and Planetary Dynamics at Tohoku University (Leader E. Ohtani) financed by the Ministry of Education, Culture, Sports, Science, and Technology of Japan.