INTRODUCTION
The Sea of Japan (also termed the East Sea) is a marginal sea of the northwestern Pacific Ocean and has a circulation system that is isolated from those of the Pacific Ocean and East China Sea. The Tsushima Warm Current (TWC), which is a branch of the subtropical western boundary current (Kuroshio Warm Current), flows into the Sea of Japan through the Tsushima Strait (120 m in depth). The water of the TWC cools off Khabarovsk, in the northern part of the Sea of Japan, and sinks to form the intermediate and bottom waters. The cooling of the water is intensified by the Asian winter monsoon (Gamo et al., Reference Gamo, Momoshima and Tolmachyov2001). The intermediate and bottom waters are referred to as the Japan Sea Intermediate–Proper Water (JSIPW) and are characterized by a high oxygen content (>5 mL O2/L) and a homogeneous density with a temperature of 0–2°C and a salinity of less than 34.06 psu (e.g., Senjyu, Reference Senjyu1999). The turnover periods of the intermediate and bottom waters are estimated to be ~100 yr (e.g., Kumamoto et al., Reference Kumamoto, Yoneda, Shibata, Kume, Tanaka, Uehiro, Morita and Shitashima1998). The circulation changed quickly in response to the Plio-Pleistocene glacial–interglacial climate changes (e.g., Tada, Reference Tada1994; Tada et al., Reference Tada, Irino and Koizumi1999; Itaki et al., Reference Itaki, Ikehara, Motoyama and Hasegawa2004; Watanabe et al., Reference Watanabe, Tada, Ikehara, Fujine and Kido2007).
Tada (Reference Tada1994) outlined the Neogene and Quaternary oceanography of the Sea of Japan. The structures of the deep water mass altered in response to the interglacial and glacial sea level changes after 2.5 Ma. During the interglacial periods, the TWC flowed, the JSIPW was formed, and below the TWC the water masses mixed. During the glacial periods, the low-stand sea level led to the TWC being blocked at the Tsushima Strait. Low-salinity water was input from rivers and formed a cap on the sea surface. The structure of the water mass was stratified (Tada, Reference Tada1994). Anoxic conditions ceased in the deep water and caused dark-colored sediments with less or no bioturbation. The sediments are called the “dark layers” and used as marker layers to correlate with core sediments in the Sea of Japan (e.g., Ikehara, Reference Ikehara2015; Tada, R., Irino, T., Expedition 346 Scientists, unpublished data).
Preliminary conclusions of Plio-Pleistocene oceanography indicate repeated periods of circulation with convection of oxygen during the interglacial periods and a stagnant water mass with a lack of oxygen during the glacial periods. Recent works have revealed variations in circulation changes and seafloor environments in response to the changes in climate (e.g., Watanabe et al., Reference Watanabe, Tada, Ikehara, Fujine and Kido2007; Usami et al., Reference Usami, Ohi, Hasegawa and Ikehara2013; Ortakand et al., Reference Ortakand, Hasegawa and Matsumoto2015). Watanabe et al. (Reference Watanabe, Tada, Ikehara, Fujine and Kido2007) studied the texture and fabric of sediment and inferred changes between five oceanographic modes after 160 ka. Usami et al. (Reference Usami, Ohi, Hasegawa and Ikehara2013) examined benthic foraminifera from deep-sea cores studied by Watanabe et al. (Reference Watanabe, Tada, Ikehara, Fujine and Kido2007). Watanabe et al. (Reference Watanabe, Tada, Ikehara, Fujine and Kido2007) and Usami et al. (Reference Usami, Ohi, Hasegawa and Ikehara2013) indicated variations in circulation, surface productivity, and oxygen content on the seafloors and correlated the variations with changes in the intensity of the Asian winter monsoon.
Little is known about the details of circulation changes and seafloor environments before 160 ka. Itaki (Reference Itaki2016) summarized geochemical and microfossil data from ocean drilling cores and outcrops on land in Japan and revised the outline of paleoceanography after 3.5 Ma. He suggested that the pattern of the circulation clearly changed at ~1.7 Ma, as Gallagher et al. (Reference Gallagher, Kitamura, Iryu, Itaki, Koizumi and Hoiles2015) pointed out. There are few data on benthic microfossils that indicate conditions in the deep-sea water mass through 1.7 Ma. Kato (Reference Kato1992) examined benthic foraminiferal faunas after ~4 Ma at Ocean Drilling Program (ODP) Site 798, Oki Ridge. According to the data compiled by Itaki (Reference Itaki2016), samples with benthic foraminifera are sporadic before 1.7 Ma. Ostracods are used as tracers of water masses (e.g., Cronin et al., Reference Cronin, Boomer, Dwyer and Rodriguez-Lazaro2002). In the Sea of Japan, Ozawa (Reference Ozawa2003) investigated the bathymetric distribution of ostracod species from the modern sediments and correlated the ostracod assemblages with the water masses. Irizuki et al. (Reference Irizuki, Kusumoto, Ishida and Tanaka2007) studied Pliocene (2.7–3.5 Ma) deep-sea ostracods from the Kuwae Formation in Niigata, central Japan, and inferred developments in the Temperate Intermediate Water (TIW) during the interglacial periods and in the coastal water during the glacial periods. They consider that the TIW was above seafloors in the lower sublittoral and upper bathyal zones and warmer than the JSIPW.
No studies have ever dealt with ostracods from deep-sea drilling cores in the Sea of Japan, although many studies on ostracods have addressed the shallow-marine faunas from the Plio-Pleistocene strata on the land at the coast of the Sea of Japan (e.g., Cronin et al., Reference Cronin, Kitamura, Ikeya, Watanabe and Kamiya1994; Ozawa and Kamiya, Reference Ozawa and Kamiya2001; Yamada et al., Reference Yamada, Irizuki and Tanaka2002, Reference Yamada, Tanaka and Irizuki2005). Ostracods are more sensitive to the oxygen content than benthic foraminifera (Moodley et al., Reference Moodley, van der Zwaan, Herman, Kempers and van Breugel1997; Braeckman et al., Reference Braeckman, Vanaverbeke, Vincx, van Oevelen and Soetaert2013). They cannot survive in suboxic and anoxic conditions. Moffitt et al. (Reference Moffitt, Hill, Roopnarine and Kennett2015) discussed changes in the oxygen content in the Santa Barbara basin after the last glacial maximum (15–16 ka) in terms of the abundance of ostracods. According to this study, the absence of ostracods indicates a low oxygen content (<0.4 mL O2/L). Analyzing ostracods allows us to discuss changes in oxygen content on the seafloor.
During the Integrated Ocean Drilling Program (IODP) Expedition 346, drilling was carried out on the seafloor at a depth of 903 m on the Oki Ridge—namely, at IODP Site U1426 (Tada et al., Reference Tada, Murray and Alvarez Zarikian2015). This site comprises successive sediments after 7 Ma. Our aims were to reveal the deep-sea ostracod faunas through 1.7 Ma at Site U1426 and to discuss changes in the seafloor environments, water mass, and benthic community on the Oki Ridge during interglacial and glacial periods.
LOCATION, LITHOLOGY, AND AGE MODEL OF U1426 CORES
At Site U1426 (37°2.00′N, 134°48.00′E), four cores (holes A–D) were drilled on the seafloor of the Oki Ridge at a depth of 903 m (Tada et al., Reference Tada, Murray and Alvarez Zarikian2015; Fig. 1). The seafloor is in contact with the JSIPW, which has a temperature of less than 2°C and an oxygen content of 4–5 mL O2/L (e.g., Ikehara, Reference Ikehara1991).
(A) Study and discussion sites and dominant wind vectors in winter during the East Asian winter monsoon. (B) Study and discussion areas and modern paths of ocean currents. F, formation; IODP, Integrated Ocean Drilling Program; ODP, Ocean Drilling Program.
The core sediments consist of intermediate-frequency alternating layers (on a scale of 3–5 m) of brown and olive-gray clay; brownish, olive-green, and greenish-gray biosiliceous-rich nannofossil ooze; biosiliceous-rich clay; and diatomaceous ooze (Tada et al., Reference Tada, Murray and Alvarez Zarikian2015). Eight types of sediment layer were observed. In the study, we unified seven of these types (with the exception of a tephra layer) into three types of sediment: biogenic siliceous ooze, biogenic calcareous ooze, and terrigenous sediments (Fig. 2). In our measurements, most of the sediments had sand contents of less than 5%. Among the physical property parameters (Tada et al., Reference Tada, Murray and Alvarez Zarikian2015), the color reflectance parameters L* and b* fluctuated between 17.1 and 63.5 and between −9.8 and 7.2, respectively. The gamma-ray attenuation bulk density ranged from −0.329 to 1.697 g/cm3. These physical property parameters exhibit cyclic fluctuations.
Images and types of lithology, stratigraphic position of samples, color reflectance parameters (L* and b*), gamma-ray attenuation (GRA) bulk density, and magnetostratigraphy. All the data are sourced from Tada et al. (Reference Tada, Murray and Alvarez Zarikian2015). For the color reflectance and GRA bulk density data, the black curves represent 10-point moving averages. In the L*-depth plot, arrows indicate the dark layers (Tada, R., Irino, T., Expedition 346 Scientists, unpublished data). CCSF, composite core depth below seafloor.
For the core depth scale, we use “m composite core depth below seafloor (CCSF)–D patched Ver1,” which consists of revised splice intervals (Tada, R., Irino, T., Expedition 346 Scientists, unpublished data) and is shown at the website (http://geos45.ees.hokudai.ac.jp/~irino/exp346stratigraphy/). The geologic age of the core sediments was determined by magnetostratigraphy, a volcanic ash layer, and marker layers. Tada et al. (Reference Tada, Murray and Alvarez Zarikian2015) indicated the chron boundaries of the magnetostratigraphy at Site U1426. Using the “dark layers” and volcanic ash layers, the sediment cores between the IODP Expedition 346 sites are being correlated (Tada, R., Irino, T., Expedition 346 Scientists, unpublished data). They calculated the ages of the “dark layers,” using radiometric ages of volcanic ash layers, magnetostratigraphy, and microfossil biostratigraphy. In the interval of 123–225 m CCSF, 14 dark layers were recognized (Table 1). The Ebisutoge-Fukuda tephra layer is recognized at 187.156–187.196 m CCSF-D patched Ver1 in hole A. Chron C2n and the top of C2An.1n are correlated with the intervals of 195.31 to 218.308 m CCSF-D patched Ver1 and 283.557 m CCSF-D patched Ver1, respectively. The datum events of the layers and magnetostratigraphy allow us to give dating of the sediment samples. The age model allows us to calculate the sedimentation rates, which were 6.29–29.12 cm/ka (Fig. 3).
Age–depth model. Each circle corresponds to the depth and age of a discrete sample. The numerical values indicate the sedimentation rates. CCSF, composite core depth below seafloor.
Datum events for the age model. CCSF, composite core depth below seafloor.
MATERIALS AND METHODS
Sediment samples and extraction of ostracods
One hundred ninety-nine discrete samples with a volume of 34 cm3 were taken at 50-cm intervals at a CCSF of 124.72–224.7 m in holes U1426A and U1426C. Applying the age model, the median of the time interval between the samples is estimated to be 3.0 ka. They were frozen under low pressure to dry before being washed with tap water through a 63 µm sieve. After the freeze-dried samples were washed, the samples from hole U1426A and hole U1426C were dried at room temperature and at less than 50°C, respectively, over two nights in an oven. Ostracods were obtained from the >150 µm fraction. To identify species, we used binocular microscopes and scanning electron microscopes at the Center for Advanced Marine Core Research, Kochi University and Shinshu University, Japan. We counted their abundance as the sum of the numbers of valves and carapaces. We obtained 414 carapaces and valve specimens from 87 of the 199 sediment samples and identified 26 taxa (Appendix A; Yamada et al., Reference Yamada, Kuroki and Yamaguchiin press). Detailed data on the ostracods are shown in Yamada et al. (Reference Yamada, Kuroki and Yamaguchiin press), including a list and scanning electron microscopy images of the ostracod taxa.
Ostracod taxa
The taxa include indicator species of two water masses: TIW taxa and JSIPW taxa. The TIW taxa, as proposed by Irizuki et al. (Reference Irizuki, Kusumoto, Ishida and Tanaka2007), consist of Krithe antisawanensis and Krithe dolichodeira [=K. hemideclivata of Zhao and Whatley (Reference Zhao and Whatley1997) and Irizuki et al. (Reference Irizuki, Kusumoto, Ishida and Tanaka2007); Ayress et al., Reference Ayress, Barrows, Passlow and Whatley1999]. These taxa live in a water mass with a temperature of 6–20°C and a depth of 0–800 m in the East China Sea but do not inhabit the modern Sea of Japan. The TIW was warmer than the JSIPW and developed on seafloors in the lower sublittoral and upper bathyal zones (Irizuki et al., Reference Irizuki, Kusumoto, Ishida and Tanaka2007). The JSIPW taxa, as defined by Ozawa (Reference Ozawa2003), comprise eight species: Acanthocythereis dunelmensis, Argilloecia toyamaensis, Krithe reversa [=K. sawanensis of Ozawa (Reference Ozawa2003); Coles et al., Reference Coles, Whatley and Moguilevsky1994], Palmoconcha saboyamensis, Propontocypris uranipponica, Robertsonites hanaii, and R. tabukii [=R. reticuliforma of Ozawa (Reference Ozawa2003); Irizuki et al. (Reference Irizuki, Kusumoto, Ishida and Tanaka2007)]. These taxa live in the JSIPW at a temperature of <5°C and a depth of 150–1500 m. Juveniles of Krithe species are regarded as Krithe spp. because it is difficult to identify them at the species level (Yamada et al., Reference Yamada, Kuroki and Yamaguchiin press). Some specimens of Krithe spp. could belong to the TIW taxa. Using the sedimentation rates (cm/ka) and a sample volume of 34 cm3, we calculated the benthic Ostracoda accumulation rates (BOARs) (ind./cm2/ka), including those of the TIW taxa, the JSIPW taxa, and Krithe spp.
Statistics
To examine the relationship between the occurrence of ostracods, the oxygen content, and the export productivity, we subjected the L* values of the samples with and without ostracods and the BOAR values in the three types of sediment to statistical analyses. For a comparison of the brightness of the sediment color, the L* values of the two types of sample were subjected to a permutation test. By performing a permutation test with 1000 iterations, the P value was calculated. To determine differences in the BOAR between the sediment types, the Steel-Dwass test with the Monte Carlo method was employed. A total of 56% of the discrete samples did not contain ostracods. The data for BOAR values were skewed and characterized by an abundance of zero values. For the assessment of data with abundant zero values, parametric methods such as the Student’s t-test and Tukey’s range test perform poorly because the mean is easily influenced by extreme values (Delucchi and Bostrom, Reference Delucchi and Bostrom2004). Delucchi and Bostrom (Reference Delucchi and Bostrom2004) recommend nonparametric and resampling methods to avoid type I errors. We applied the Steel-Dwass test, which is a nonparametric method for multiple comparisons, to the BOAR values in the three types of sediment. Using the Monte Carlo method, the Steel-Dwass test was conducted 1000 times, and P values were calculated. The statistical analysis was performed using the free software R (R Core Team, 2015) and its package RcmdrPlugin.EZR (Kanda, Reference Kanda2013).
To illustrate the long-term changes in the BOAR values, the LOWESS (locally weighted scatterplot smoothing) filter was applied to the time series data of the values.
To identify periodicity in the ostracod abundance, we subjected the BOAR values and accumulation rates of the TIW and JSIPW taxa to the multitaper method (MTM) of spectrum analysis (Thomson, Reference Thomson1982; Mann and Lee, Reference Mann and Lees1996). Meyers (Reference Meyers2012) developed the MTM of Mann and Lee (Reference Mann and Lees1996) to minimize its “edge effect” problem and simulated the method’s suitability for time series data with noise. The modified MTM isolated and identified significant frequencies using the autoregressive-1 (AR1) “red noise” confidence interval and the MTM harmonic F-test confidence interval. According to Meyers (Reference Meyers2012), we considered periodicities with probabilities of greater than 90% in the robust AR1 model and the MTM harmonic F-test. The analysis used a time–bandwidth product of 2, and the data were linearly detrended prior to analysis. Using the piecewise linear interpolation method, we interpolated the time series data of the BOAR, TIW, and JSIPW values at an even interval of 3.0 ka. The time series data were padded to 1260 points. The MTM of Mann and Lee (Reference Mann and Lees1996) was performed using the R package “astrochron” (Meyers, Reference Meyers2014).
Radiolarian subtropical water taxa
To recognize the inflow of the TWC, we examined the relative abundance of radiolarian taxa that are diagnostic for the TWC. These diagnostic taxa consist of the Tetrapyle octacantha group, the Dictyocoryne group, Didymocyrtis tetrathalamus, the Euchitonia group, and Spongaster tetras and are referred to as the subtropical water taxa. Today, these five taxa are distributed along the Kuroshio Warm Current in the tropical–subtropical Pacific (e.g., Lombari and Boden, Reference Lombari and Boden1985) and the East China Sea (Chang et al., Reference Chang, Zhouang, Li, Yan, Cao and Cang2003). In the southwestern part of the Sea of Japan, the relative abundances of the five taxa have higher values than anywhere else. They are correlated with the TWC (Motoyama et al., Reference Motoyama, Yamada, Hoshiba and Itaki2016). The relative abundance of the subtropical water taxa has been used to assess the flow of the TWC in the Sea of Japan during the late Pleistocene and Holocene (Itaki, Reference Itaki2007; Itaki et al., Reference Itaki, Komatsu and Motoyama2007; Itaki, T., Sagawa, T., Kubota Y., unpublished data). We made microscope slides from 95 sediment samples from hole 1426A, identified the five taxa among radiolarian assemblages with more than 200 specimens, and counted the five taxa and all the radiolarians (>45 μm).
Foraminifer and geochemical data for the Oki Ridge
To discuss the paleoenvironments on the Oki Ridge, we compared the ostracod data with the geochemical and benthic foraminifer data at ODP Site 798, Oki Ridge of Kheradyar (Reference Kheradyar1992), Kato (Reference Kato1992), and Piper and Isaacs (Reference Piper and Isaacs1995, Reference Piper and Isaacs1996). Site 798 is located at 37°2.310′N, 134°47.976′E and at a depth of 900 m. The site is located 0.6 km northwest of Site U1426. Piper and Isaacs (Reference Piper and Isaacs1995, Reference Piper and Isaacs1996) measured trace elements such as molybdenum (Mo), uranium (U), vanadium (V), and chromium (Cr) in sediments from Site 798 and inferred that redox conditions were present on seafloors and in water columns between 1.40 and 1.28 Ma. These elements are sensitive to redox conditions and are correlated with the amounts of organic matter and the formation of pyrite (e.g., Tribovillard et al., Reference Tribovillard, Algeo, Lyons and Riboulleau2006). Using their measurements, we calculated the V/Cr ratio, which is a proxy for redox conditions (Jones and Manning, Reference Jones and Manning1994; Rimmer, Reference Rimmer2004; Tribovillard et al., Reference Tribovillard, Algeo, Lyons and Riboulleau2006). In our study, we used the classification of Tyson and Pearson (Reference Tyson and Pearson1991) of environments with different oxygen concentrations: 2–8 mL O2/L for oxic, 0.2–2 mL O2/L for dysoxic, and 0–0.2 mL O2/L for suboxic conditions. Jones and Manning (Reference Jones and Manning1994) proposed values of the V/Cr ratio (ppm/ppm) for redox conditions: <2.0 for oxic, 2.0–4.25 for dysoxic, and >4.25 for suboxic conditions. The ratio may also indicate relative differences in redox conditions (Rimmer, Reference Rimmer2004). The age model is based on magnetostratigraphy (Ingle et al., Reference Ingle, Suyehiro and von Breymann1990; Hamano et al., Reference Hamano, Krumsiek, Vigliotti and Wippern1992). Using the age model, the sedimentation rates were calculated to be between 12.82 and 13.15 cm/ka. Kheradyar (Reference Kheradyar1992) and Kato (Reference Kato1992) studied planktic and benthic foraminifer assemblages, respectively. We plotted the relative abundance of Globorotalia inflata and Globigerina ruber in planktic foraminifer assemblages with more than 100 specimens in Kheradyar (Reference Kheradyar1992). These species flourish in the channel of the Tsushima Strait (Domitsu and Oda, Reference Domitsu and Oda2005). G. inflata and G. ruber indicate a warm intermediate water mass and the TWC influx, respectively (Kitamura and Kimoto, Reference Kitamura and Kimoto2006; Kitamura, Reference Kitamura2009). Using the relative abundance of this species, Kitamura (Reference Kitamura2009) has already discussed changes in the temperature of the intermediate water at the Oki Ridge. Using data from Kato (Reference Kato1992), we calculated the benthic foraminifer accumulation rate (BFAR), which is an indicator of export productivity (Herguera and Berger, Reference Herguera and Berger1991). Kato (Reference Kato1992) collected benthic foraminifera with a size of greater than 125 µm and measured the number of benthic foraminifer individuals per 1 g sediment (i.e., the foraminifer density). By multiplying by the foraminifer density, the dry bulk density in each section (Ingle et al., Reference Ingle, Suyehiro and von Breymann1990), and the sedimentation rates, the BFAR values were calculated (Appendix C in the Supplementary Materials).
RESULTS
Ostracod taxa
Ostracods were present in series of 2–9 successive sediment samples and absent in series of 2–13 successive samples (Appendix B in the Supplementary Materials). In 56% of the samples, the abundance and BOAR value were zero. The remaining samples, which contained ostracods, had an abundance of 1–50 and a BOAR of 0.19–22.41. At a 186.306 m CCSF-D patched Ver1 (346-1426A-18H-5W, 75–77 cm), the abundance and BOAR reached maxima. Regarding the pattern of ostracod occurrence, ostracods were found in 30 series with 1–8 samples (1–35.5 ka in duration) but were not found in 31 series with 1–16 samples (1.5–104.5 ka in duration). A. dunelmensis, R. tabukii, and Krithe species were found frequently and were present in 78 samples. The JSIPW taxa occurred successively (Fig. 4) and were found in 29 series of the samples. Their presence continued for 3–355 ka. The TIW taxa appeared discontinuously in 16 samples, and their presence persisted for 1–8 ka (Table 2). The JSIPW taxa were not often associated with the TIW taxa. Krithe spp. were present in 30 discrete samples and 23 series with 1–3 samples (1–15 ka in duration).
Lithology, color reflectance parameter L*, gamma-ray attenuation (GRA) bulk density, sand content, ostracod abundance, benthic Ostracoda accumulation rate (BOAR), species richness, and occurrence of six dominant taxa. The lithology, L*, and GRA data were sourced from Tada et al. (Reference Tada, Murray and Alvarez Zarikian2015). For the color reflectance and GRA bulk density data, the black curves represent 10-point moving averages. In the L*-depth plot, arrows indicate the dark layers (Tada, R., Irino, T., Expedition 346 Scientists, unpublished data).
Statistics
In a comparison of L* values between samples with and without ostracods, the permutation test rejected the null hypothesis that the L* values were equivalent for both types of sample. The median L* values were 35.25 for samples with ostracods and 34.0 for samples without ostracods (Fig. 5A). The P value from the permutation test was calculated to be 8.9×10–6. The ostracod-bearing samples displayed significantly higher L* values than the other samples. In a comparison of the BOAR between biogenic siliceous ooze, biogenic calcareous ooze, and terrigenous sediment, we recognized significant differences in the BOAR between biogenic siliceous ooze and terrigenous sediment and between biogenic calcareous ooze and terrigenous sediment at a significance level of P<0.05 (Table 2). The biogenic sediments contained significantly more ostracods than the terrigenous sediments. The median BOAR values were zero for biogenic siliceous ooze, 0.45 for biogenic calcareous ooze, and zero for terrigenous sediment (Fig. 5B). The 75th percentiles were 1.26 for biogenic siliceous ooze, 1.34 for biogenic calcareous ooze, and 0.45 for terrigenous sediment.
(A) Color reflectance parameter L* values of samples with and without ostracods. (B) Benthic Ostracoda accumulation rate (BOAR) values of biogenic siliceous and calcareous oozes and terrigenous sediment; n represents the number of samples.
Multitaper method (MTM) spectra and confidence level estimates for the time series data of the Temperate Intermediate water taxa. (A) MTM power spectrum shown as a black curve. The thick red curve represents the background using the robust AR1 model. The dotted curves correspond to confidence levels of 90%, 95%, and 99% using the robust AR1 model. (B) Confidence level estimates using the robust AR1 model. (C) Confidence level estimates using the MTM harmonic F-test. The numerals are used to identify the significant frequencies. The significant frequencies sand their probabilities are shown in Table 4. AR1, autoregressive-1. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
Duration of the Temperate Intermediate Water (TIW) during the early Pleistocene and late Pliocene. The bottom and top of the TIW taxa were the geologic ages of the midpoint between samples with and without the taxa. Irizuki et al. (Reference Irizuki, Kusumoto, Ishida and Tanaka2007) estimated the age of samples using two types of the age model. In the Kuwae data, the time ranges are shown in the bottom, top, and duration of factor 1, which indicates the TIW.
The LOWESS filter indicated declines in the BOAR from 1.90 to 1.86 Ma and from 1.70 to 1.67 Ma. The combination of the robust AR1 model and the MTM harmonic F-test identified a significant peak at confidence levels of 90% in the power spectra TIW values, which occurred at 15.18 ka (Fig. 6; Table 4). However, any peaks for the accumulation rates of the BOAR and JSIPW taxa were not significant.
Foraminifer, radiolarian, and geochemical data for the Oki Ridge
In the U1426 sediments, we observed calcareous benthic foraminifera in all the samples used for the analysis of ostracods. From the results for the radiolarians, the subtropical water taxa were found in 58 samples. The relative abundances of subtropical water taxa were less than 9.7%. At 1.2–1.4 Ma, the subtropical water taxa were not found. At 1.485–1.520, 1.619–1.660, and 1.712–1.775 Ma, the relative abundances often exceeded 1.0%. At 1.686 Ma, the relative abundance reached a maximum of 9.7%. Between 1.8 and 2.0 Ma, the subtropical water taxa often appeared and accounted for 0.3–1.0% of radiolarian assemblages.
From the geochemical data for ODP Site 798, Mo, U, and V displayed high values at 1.34–1.35 and 1.39–1.40 Ma. The peak values for Mo exceeded 40 ppm. The maximum values for U and V exceeded 5 ppm and 20 ppm, respectively (Piper and Isaacs, Reference Piper and Isaacs1996; Fig. 7). The V/Cr ratio (ppm/ppm) ranged between 0.38 and 5.56 and exhibited plateaus with values of greater than 2.0 at 1.349–1.356 and 1.389–1.396 Ma. Peaks with values of greater than 4.25 were observed at 1.393 and 1.395 Ma. In the intervals between the plateaus, Mo, U, and V also exhibited high values. The BFAR fluctuated between 2 and 8852 (Fig. 8). The maximum BFAR appeared at 1.37 Ma. G. inflata appeared at ~1.55 Ma, exhibited two peaks in its relative abundance at 1.44 and 1.35 Ma, and disappeared at ~1.31 Ma (Fig. 8).
Plots of color reflectance parameter L*, benthic Ostracoda accumulation rate (BOAR), and dominant ostracod taxa at Site U1426 and of the benthic foraminifer accumulation rate (BFAR), species richness, geochemical elements (Mo, U, and V; ppm), V/Cr ratio (ppm/ppm), and distribution of the dark layers at Hole 798B between 1.4 and 1.2 Ma. The curve for L* represents a 10-point moving average. The dashed line indicates the median value of 34.6. The L* values were sourced from Tada et al. (Reference Tada, Murray and Alvarez Zarikian2015). For the V/Cr ratio, the dashed lines indicate values of 2.0 and 4.25. The benthic foraminifer data are from Kato (Reference Kato1992), whereas the geochemical and dark-layer data are from Piper and Isaacs (Reference Piper and Isaacs1996).
Plots of benthic foraminiferal δ18O of the LR04 stack (Lisiecki and Raymo, Reference Lisiecki and Raymo2005), the radiolarian subtropical taxa, benthic Ostracoda accumulation rate (BOAR), ostracod taxa, and accumulation rates of Krithe spp., the Temperate Intermediate Water (TIW) taxa, and Japan Sea Intermediate–Proper Water (JSIPW) taxa at Site U1426 and of the benthic foraminifer accumulation rate (BFAR) and relative abundances of Globorotalia inflata and Globigerina ruber at Hole 798B. IAR indicates the individual accumulation rate. The gray areas indicate interglacial stages. The BFAR and planktic foraminiferal data were sourced from Kato (Reference Kato1992) and Kheradyar (Reference Kheradyar1992), respectively. For the BOAR data, the thick gray line indicates the LOWESS (locally weighted scatterplot smoothing)-filtered curve. VPDB, Vienna Pee Dee belemnite.
DISCUSSION
Ostracod responses to oxygen content and export productivity
All the sediments contained calcareous benthic foraminifera, although 56% of them lacked ostracods. The absence of ostracods was not caused by the dissolution of carbonate during diageneses.
Ostracods occurred in brightly colored sediments with higher L* values (Fig. 5A). Because the brightness of sediments is related to their biological productivity and oxygen content (Watanabe et al., Reference Watanabe, Tada, Ikehara, Fujine and Kido2007; Usami et al., Reference Usami, Ohi, Hasegawa and Ikehara2013), the ostracod occurrence was linked to these environmental parameters. In the Sea of Japan, brightly colored sediments indicate oxic conditions and a high biogenic content (e.g., Tada et al., Reference Tada, Irino and Koizumi1999; Watanabe et al., Reference Watanabe, Tada, Ikehara, Fujine and Kido2007). The occurrence of ostracods was correlated with high biogenic productivity and oxic conditions.
At Site 798, the V/Cr ratios of greater than 2.0 suggest anoxic to suboxic conditions at 1.349–1.356 and 1.389–1.396 Ma (Fig. 7), according to Jones and Manning (Reference Jones and Manning1994).
In the intervals of 1.349–1.356 and 1.389–1.396 Ma, the benthic faunas indicated oxic conditions. At 1.356 Ma (135.95 meters below the seafloor [mbsf]; 128-798B-15H-3W, 39–41cm), the foraminifer assemblage was dominated by Brizalina pacifica [=Bolivina pacifica of Kato (Reference Kato1992); Usami et al., Reference Usami, Ohi, Hasegawa and Ikehara2013], which accounted for 34% of all the foraminifera (Kato, Reference Kato1992). Epistominella pulchella, which was the subordinate species, composed 33% of the assemblage. Because the relative abundance of the dominant species was not markedly higher than that of the subordinate species, we describe this as an assemblage with an indistinct dominant species, as defined by Usami et al. (Reference Usami, Ohi, Hasegawa and Ikehara2013). The foraminifer assemblage indicated oxic conditions and relatively low productivity (Usami et al., Reference Usami, Ohi, Hasegawa and Ikehara2013). At 1.392 Ma (140.46 m bsf; 128-798B-15H-6W, 40–42cm), an assemblage dominated by Angulogerina ikebei [=A. kokozuraensis of Kato (Reference Kato1992); Usami et al., Reference Usami, Ohi, Hasegawa and Ikehara2013] was found (Kato, Reference Kato1992), which suggests oxic conditions (Usami et al., Reference Usami, Ohi, Hasegawa and Ikehara2013). Ostracods were found at 1.326 and 1.34–1.37 Ma. Living ostracods are found at an oxygen content of greater than 0.4 mL O2/L off Benguela in the southeastern Atlantic Ocean (Dingle, Reference Dingle1995). Zhou and Ikeya (Reference Zhou and Ikeya2002) and Ozawa (Reference Ozawa2004) reported the absence of ostracods in the oxygen minimum zone with an oxygen content of less than 1.8 and 1.0 mL O2/L in Suruga Bay and the Sea of Okhotsk, respectively. We consider that the occurrence of ostracods indicates an oxygen content of greater than 1.0 mL O2/L around the Japanese Islands. The oxygen contents were estimated to be greater than 1.0 mL O2/L. The L* values were higher than the median, which reflects the oxidizing conditions. The high V/Cr ratios are less than 2.0, indicating the oxic condition. As Piper and Isaacs (Reference Piper and Isaacs1995, Reference Piper and Isaacs1996) considered, the high V/Cr ratios and Mo, U, and V contents were caused by sulfate-reducing and less denitrified conditions. Alternatively, as Rimmer (Reference Rimmer2004) suggested, the high contents of the chemical elements may indicate relatively low-oxygen conditions.
The occurrence of ostracods is linked to the blooming of phytoplankton. The biogenic oozes—namely, both biogenic siliceous and biogenic calcareous oozes—had significantly higher BOAR values than the terrigenous sediments, which contained fewer fossil phytoplankton (Fig. 5B; Table 3). In general, the blooming of phytoplankton leads to an increase in meiobenthic productivity (e.g., Levin et al., Reference Levin, Etter, Rex, Gooday, Smith, Pineda, Stuart, Hessler and Pawson2001). The results of statistical analysis indicate that the type of phytoplankton (diatoms or coccoliths) does not clearly affect BOAR values. Hyun et al. (Reference Hyun, Bahk, Suk and Park2007) investigated the carbonate, opal, and total organic carbon (TOC) contents in the Quaternary sediments of the Sea of Japan and inferred that the blooming of diatoms provides more nutrients on the seafloor than that of coccoliths. The export productivity could vary between the blooming of diatoms and coccoliths. The BOAR values for biogenic siliceous ooze were as high as those for biogenic calcareous ooze (Fig. 5B, Table 3), which suggests that the BOAR values do not clearly depend on the type of phytoplankton. At the Oki Ridge, the changes in BOAR values at Site U1426 were not consistent with those in BFAR values at ODP Site 798 (Fig. 8). The relationship between the BOAR and food supply has been disputed (Yasuhara et al., Reference Yasuhara, Hunt, Cronin, Hokanishi, Kawahata, Tsujimoto and Ishitake2012; Stepanova and Lyle, Reference Stepanova and Lyle2014). Yasuhara et al. (Reference Yasuhara, Hunt, Cronin, Hokanishi, Kawahata, Tsujimoto and Ishitake2012), who examined the correlation between the BOAR values and TOC content in the Quaternary sediments of the Shatsky Rise in the northern Pacific, concluded that the TOC content in the sediment displayed a hump-shaped relationship with the BOAR. On the other hand, Stepanova and Lyle (Reference Stepanova and Lyle2014) recognized high BOAR values during the glacial periods with high TOC contents at an eastern equatorial Pacific site. They considered that a high food supply increased the abundance of ostracods. Our results suggest that the BOAR values exhibited a weak correlation with the export productivity, as Yasuhara et al. (Reference Yasuhara, Hunt, Cronin, Hokanishi, Kawahata, Tsujimoto and Ishitake2012) pointed out.
Changes in water mass structure
In the relative abundance of the radiolarian subtropical water taxa, Motoyama et al. (Reference Motoyama, Yamada, Hoshiba and Itaki2016) indicated that the subtropical water taxa account for 21–30% of radiolarian assemblages in the Holocene sediments near the Oki Ridge (samples 34, 46, 49, and 50). At Site 1426, the subtropical water taxa frequently exceed 10% above ~58 m CCSF-D patched Ver1 (Itaki, T., Sagawa, T., Kubota, Y., unpublished data). The Pleistocene subtropical water taxa indicate 0.3–9.7%. Their relative abundance suggests a smaller influx of the TWC than the modern influx. As Kitamura and Kimoto (Reference Kitamura and Kimoto2006) pointed out, the TWC influx has increased since the Pleistocene. Here we consider that relative abundance of more than 1% indicates the TWC flow over the Oki Ridge.
Though the middle bathyal zone, the ostracod TIW taxa often appear. The TIW taxa remained for a shorter period than those in the Pliocene Kuwae Formation (Table 2). The TIW taxa persisted for 1–8 ka at Site U1426. If Krithe spp. are included in the TIW taxa, these taxa persisted for 1–10 ka. The Pliocene TIW taxa remained for 2.0–42.5 ka (Irizuki et al., Reference Irizuki, Kusumoto, Ishida and Tanaka2007). Even the occurrence of G. inflata was not consistent with that of the TIW taxa on Oki Ridge sites. The differences in duration and combination in the water mass between Site U1426 and the Kuwae Formation were caused by the bathymetric setting. The Kuwae Formation contains shallower deposits than Site U1426. The Kuwae sediments were in the lower sublittoral to upper bathyal zones shallower than ~800 m depth. The appearances of the TIW taxa indicate that the TIW possibly has developed to a greater extent than in a shallower setting.
We recognize four modes in the pattern of the radiolarian subtropical water taxa, G. ruber, and the ostracod faunas and infer paleoceanography (Fig. 9). Mode 1 is the intervals without the subtropical water taxa, G. ruber, and ostracods. The TWC did not flow over the Oki Ridge. The seafloor was exposed on the intermediate water with low oxygen content. Mode 1 fell in Marine Oxygen Isotope Stage (MIS) 40, early MIS 47–late MIS 48, late MIS 61, and MIS 63. Mode 2 is defined as the interval with the combination of ostracod JSIPW taxa and the TWC indicator, the radiolarian subtropical water taxa or G. ruber. The ostracod JSIPW taxa often associate with the TIW taxa. The mode occurred during late MIS 47, early MIS 49, MIS 51, MIS 54, late MIS 55, MIS 57, late MIS 58, and MIS 69. In the mode, the TWC flowed over the Oki Ridge and ventilated in the north of the Sea of Japan to form the JSIPW. During MIS 54 and late MIS 58, the TIW developed and extended to the middle bathyal zone. The temperature gradient of the water column was reduced temporally. Irizuki et al. (Reference Irizuki, Kusumoto, Ishida and Tanaka2007) consider that the strong East Asian winter monsoon would force the temperature gradient steep and drive the TIW to be shrunken. The temporal development of the TIW was possibly caused by occasional weakening of the winter monsoon. During the intervals with the TIW taxa, temperature gradients became gentle in the water column and the TIW expanded. Mode 3 is the interval of the subtropical water taxa without ostracod faunas. It fell in MIS 50, early MIS 55, MIS 66, and MIS 68. The subtropical water taxa indicate that the TWC influx would ventilate in the north of the Sea of Japan to form intermediate water with high oxygen content. The intermediate water could not covey oxygen into deep water. The ventilation of the surface water mass was possibly shallower than mode 2. Mode 4 is the interval of the ostracod fauna without the radiolarian subtropical taxa during early MIS 51 and MIS 61. The sea surface over the Oki Ridge was not influenced by the TWC. Ostracods indicate oxic bottom water. During the early MIS 51, temperature gradients were reduced in the water column and the TIW expanded.
Schematic representation of vertical circulation in the Japan Sea. The topography is at the present. JSIPW, Japan Sea Intermediate–Proper Water; MIS, Marine Oxygen Isotope Stage; TIW, Temperate Intermediate Water.
The power spectra indicate changes in TIW values in cycles of ~15 ka (Fig. 6, Table 4). It is not a typical periodicity of the orbital precession such as 19 and 23 ka in the East Asian monsoon (Sun et al., Reference Sun, Clemens, An and Yu2006; Ao et al., Reference Ao, Dekkers, Qin and Xiao2011) or the early Pleistocene dominant periodicity, 41 ka, which represents the orbital eccentricity. The periodicity of 41 ka is observed in changes in shallow-marine faunas (e.g., Kitamura et al. Reference Kitamura, Kondo, Sakai and Horii1994; Takata, Reference Takata2000; Ozawa and Kamiya, Reference Ozawa and Kamiya2001). Irizuki et al. (Reference Irizuki, Kusumoto, Ishida and Tanaka2007) hypothesize that a weak East Asian winter monsoon would force temperature gradients to become gentler and cause the development of the TIW. The seafloor environments were affected by food supply and oxygen content that were conveyed from the sea surface. Even the Holocene dark layers have variable fabric, bioturbation, and geochemical and paleontological compositions, indicating variable seafloor environments (e.g., Watanabe et al., Reference Watanabe, Tada, Ikehara, Fujine and Kido2007; Usami et al., Reference Usami, Ohi, Hasegawa and Ikehara2013). The TIW periodicity resulted from the confluence of food supply and oxygen content, as well as a temperature gradient in the water column.
Results of the Steel-Dwass test with the Monte Carlo simulation. Single and double asterisks indicate significance at levels of P<0.05 and P<0.01, respectively.
Results of applying the multitaper method to the time series data of the Temperate Intermediate Water values. ID is shown in Figure 6. AR1, autoregressive-1.
Deep-sea ostracod faunas through 1.7 Ma
At 1.7 Ma, the taxonomic composition did not undergo a clear change. Throughout the time interval of 1.0–2.0 Ma, the faunas consisted of the TIW and JSIPW taxa. The formation and structure of the intermediate water did not change at ~1.7 Ma. On average, the BOAR values exhibited a decline. The LOWESS-filtered BOAR values exhibited an obvious decrease from 1.7 to 1.6 Ma (Fig. 8). The BOAR values are linked with the food supply, as was discussed previously. The decrease in BOAR values suggests an alteration in the food supply from the sea surface. According to the model of Itaki (Reference Itaki2016), the inflow of the TWC with a high nutrient content was enhanced after 1.7 Ma. The decrease in BOAR values may have resulted from a high nutrient content that was transported by the enhanced inflow of the TWC.
Migration of ostracods in response to changes in climate
The Japan Sea Intermediate Water conveys oxygen into the deep sea. This oxygen is easily available for deep-sea animals. In general, the deep-sea faunas are adapted to cold conditions and are scanty eurybathic species as a result of the repeated environmental changes in the Plio-Pleistocene (Kojima et al., Reference Kojima, Adachi and Kodama2007; Malyutina and Brandt, Reference Malyutina and Brandt2013). During the glacial periods, the whole of the Sea of Japan was isolated and conditions became anoxic in deep water, which might have eliminated most of the benthic faunas. Shallow-marine ostracods exhibit clear changes in taxonomic composition in response to the interglacial and glacial environmental changes (e.g., Ozawa and Kamiya, Reference Ozawa and Kamiya2001; Yamada et al., Reference Yamada, Irizuki and Tanaka2002). Irizuki et al. (Reference Irizuki, Kusumoto, Ishida and Tanaka2007) hypothesized that the faunal changes could have been caused by bathymetric migration of ostracod taxa. To test this hypothesis, it is necessary to compare the taxonomic compositions of shallow-marine and deep-sea faunas.
Our data indicate the disappearance and appearance of ostracod faunas in response to deep-sea environmental changes (Fig. 4). In the interval with the radiolarian subtropical water taxa, we found Acanthocythereis dunelmensis, Argilloecia toyamaensis, Robertsonites hanaii, and R. tabukii, which are diagnostic for the JSIPW (Fig. 8). The occurrence of the ostracod species with the subtropical water taxa suggests that these species appeared in the deep sea during the interglacial periods. They appeared during MIS 73, MIS 67, MIS 63, MIS 53, MIS 51, and MIS 49. The four species have already been reported in the shallow-marine Omma and Hamada formations exposed on the coasts of the Sea of Japan (Cronin and Ikeya, Reference Cronin and Ikeya1987; Ozawa, Reference Ozawa1996; Ozawa and Kamiya, Reference Ozawa and Kamiya2001; Ozawa and Domitsu, Reference Ozawa and Domitsu2010). These outcrops consist of coastal and shelf sediments of which the ages range from 2.0 to 1.2 Ma (e.g., Ozawa and Kamiya, Reference Ozawa and Kamiya2005). The species occur in the intervals of both the interglacial and glacial periods. The occurrence of the four species at Site U1426 indicates an expansion of their habitats into deep-sea floors during the transitions from the glacial to the interglacial periods. During the glacial periods, dark-colored sediments were formed under low-oxygen conditions with low export productivity. The ostracods could not survive on the deep-sea floors. The taxa limited their habitats to shelves and coasts.
The model of migration is consistent with the hypothesis of Irizuki et al. (Reference Irizuki, Kusumoto, Ishida and Tanaka2007) on the migration of the taxa in response to the glacial–interglacial environmental changes. Irizuki et al. (Reference Irizuki, Kusumoto, Ishida and Tanaka2007) recognized decreases in the relative abundance of A. dunelmensis and R. tabukii during the interglacial periods and inferred the bathymetric migration of the taxa in response to the Pliocene changes in sea level and water mass. According to their work, the decrease in the relative abundance of A. dunelmensis and R. tabukii was attributable to migration to deeper seafloors during the developments in the TIW from the glacial to the interglacial periods. However, Irizuki et al. (Reference Irizuki, Kusumoto, Ishida and Tanaka2007) neither documented faunal changes in both deep-sea and shallow-marine settings nor described the species composition in the glacial deep-sea sediments. The correlation of ostracod faunas between our deep-sea data and the reported shallow-marine data enable us to corroborate the hypothesis of Irizuki et al. (Reference Irizuki, Kusumoto, Ishida and Tanaka2007).
Today, A. dunelmensis, A. toyamaensis, R. hanaii, and R. tabukii are not living in shelf and coastal environments in the Japan Sea (Ozawa, Reference Ozawa2003). These species flourish in colder conditions than the TWC. The TWC inflow has increased since 1.7 Ma (Kitamura and Kimoto, Reference Kitamura and Kimoto2006; Gallager et al., Reference Gallagher, Kitamura, Iryu, Itaki, Koizumi and Hoiles2015; Itaki, Reference Itaki2016). It has been accompanied by increases in sea surface temperature. We consider that the increases in the temperature dismissed them from the shelf and coastal environments. Ozawa and Kamiya (Reference Ozawa and Kamiya2005) recognized the disappearance of 24 shallow-marine taxa in the Japan Sea after 1.5 Ma. They called the extinct taxa the “now-extinct species” and inferred that increase in amplitude of the eustatic change gave a stress to shallow-marine taxa and could have caused the extinction of the now-extinct species. A. dunelmensis, A. toyamaensis, R. hanaii, and R. tabukii are not included in the now-extinct species. The high amplification of sea level changes possibly also caused their disappearance in shallow-marine environments.
CONCLUSIONS
We revealed changes in the fauna of deep-sea ostracods that were correlated with the glacial–interglacial environmental changes.
The occurrence of the deep-sea ostracods was linked with redox conditions on the seafloors and export productivity because more ostracods were found at higher L* values and in biogenic sediments. The benthic foraminifera and redox-sensitive elements indicate that the ostracods were found in intervals with oxic conditions but were absent in intervals with dysoxic conditions.
Under a small influx of the TWC to the Sea of Japan, the JSIPW was formed. The ostracod faunas indicate expansion of the TIW to the middle bathyal zone during MIS 43, MIS 51, MIS 54, MIS 58, MIS 61, MIS 63, MIS 72, and MIS 73. The TIW suggests a gentler temperature gradient in the water column than the modern gradient. By analogy with the relationship between the modern JSIPW and the winter monsoon, the occasional expansion of the TIW was possibly caused by a weakened winter monsoon. The temporal change in the TIW taxa shows the ~15 ka periodicities that resulted from a confluence of temperature gradient in the water column, oxygen content, and food supply.
From 1.7 to 1.6 Ma, the taxonomic composition did not indicate changes, but the BOAR values decreased markedly, which suggests an alteration in export productivity.
The repeated appearance and disappearance of the JSIPW taxa indicates changes in their habitat ranges in response to the glacial–interglacial environmental changes. During the interglacial periods, the taxa expanded their habitats to deep-sea floors. During the glacial periods, they were eradicated from deep-sea floors, and their habitats became restricted to coastal and shelf environments.
Acknowledgments
This research used samples provided by the IODP. This work involves KK’s undergraduate thesis at Shinshu University. We express our thanks to Profs. Koichi Hoyanagi and Kohki Yoshida (Shinshu University, Japan) for their discussion throughout this study; Dr. Ryoichi Nakada (Japan Agency for Marine-Earth Science and Technology, Kochi) for suggestions on geochemical proxies for the redox conditions; Profs. Ryuji Tada (University of Tokyo, Japan), Takahiro Kamiya (Kanazawa University, Japan), and Dr. Tomohisa Irino (Hokkaido University, Japan) for their suggestion on stratigraphy and physical properties at the site, as well as ostracods; and Dr. Hirokazu Ozawa (Nihon University, Japan) and an anonymous reviewer for their critical reviews and constructive comments. Also, we would like to thank Enago (http://www.enago.jp) for the English-language review.
SUPPLEMENTARY MATERIAL
To view supplementary material for this article, please visit https://doi.org/10.1017/qua.2017.68
