Secular Variations in Provenance of Sedimentary Components in the Western North Pacific Ocean Constrained by Sr Isotopic Features of Deep‐Sea Sediments

Deep‐sea sediments around Minamitorishima Island in the western North Pacific Ocean record the depositional environment in the central to western North Pacific since the late Cretaceous. Previous studies on the bulk chemical composition of deep‐sea sediments in this area have revealed that the sediment column can be divided into five chemostratigraphic units with at least three intercalated layers enriched in rare‐earth elements. However, the end‐member components and their changes in abundance, which indicate compositional variation, as well as the environmental factors affecting them, remain unclear. Here, we report the bulk chemistry, including newly analyzed 87Sr/86Sr ratios of the sediments, seafloor Mn nodules, and fish teeth collected from the deep‐sea basin around Minamitorishima Island, to identify the end‐member components of the sediments. Our results suggest that the pelagic sediments in the study area are mainly composed of terrigenous components, hydrogenous Mn oxides, biogenic calcium phosphate, and volcanic materials. The combination of Sr isotope modeling and bulk chemistry revealed that the input of volcanic materials has increased twice, which might be Izu‐Bonin‐Mariana tephras of the latest Eocene to Oligocene explosive volcanism at the later event but is unclear at the earlier event. The provenance of terrigenous materials has changed from North America to East Asia. The bulk sediment chemostratigraphy around Minamitorishima Island could have resulted not only from the local environmental changes in the pelagic realm, but also from volcanism around the Pacific Ocean and the northwestward movement of the area due to the Pacific plate motion.

Murray, Kato et al., 2011;Plank & Langmuir, 1998;Yasukawa et al., 2016;Zhou & Kyte, 1992). It is worth reconstructing the paleoceanographic/paleoenvironmental conditions of the western North Pacific because this area has moved from the central to the western Pacific since the late Cretaceous (Shipboard Scientific Party, 1990) and thus records the long-term evolution of the pelagic depositional environments in the Pacific Ocean.
In addition, rare earth elements and yttrium (REY)-enriched pelagic clay, which is called "REY-rich mud," has been discovered and reported from the seafloor around Minamitorishima (Marcus) Island in the western North Pacific (Figure 1; Fujinaga et al., 2016;Iijima et al., 2016). It can be seen as a new REY resource owing to its high REY concentrations and immense resource potential (Takaya et al., 2018). To investigate the distribution of the REY-rich mud around Minamitorishima Island, over 70 piston core samples have been collected from this area in recent years Iijima et al., 2016;Takaya et al., 2018;Tanaka, Nakamura, Yasukawa, Mimura, Fujinaga, Iijima et al., 2020, Tanaka, Nakamura, Yasukawa, Mimura, Fujinaga, Ohta et al., 2020. Based on the bulk chemical compositions, Tanaka, Nakamura, Yasukawa, Mimura, Fujinaga, Iijima et al. (2020) concluded that deep-sea sediments around Minamitorishima Island can be systematically classified into five groups. In the studied cores, the sediment groups occurred in a certain sequence from the seafloor to depth, with three intercalated REY-enriched layers (REY peaks, defined as sediment with ΣREY >2,000 ppm; Tanaka, Nakamura, Yasukawa, Mimura, Fujinaga, Iijima et al., 2020, Tanaka, Nakamura, Yasukawa, Mimura, Fujinaga, Ohta et al., 2020 that form a "multi-elemental chemostratigraphy" in the sediment column. The chemostratigraphic units and REY peaks classified by Tanaka, Nakamura, Yasukawa, Mimura, Fujinaga, Iijima et al. (2020) are described as follows, from the seafloor surface to depth: Units I and II, the 1st REY peak, Unit III, the 2nd REY peak, and Units IV and V intercalated by the 3rd REY peak. The characteristic chemical compositions of each unit and the REY peak could be attributed to variations in the proportions of the sediment source end-member components among the units (Tanaka, Nakamura, Yasukawa, Mimura, Fujinaga, Iijima et al., 2020). However, the chemical composition, provenance, and proportions of the sediment end-member components that exert control on the chemostratigraphy of the deep-sea sediments around Minamitorishima Island are not entirely clear because they hitherto have only been discussed with a focus on the bulk chemical and mineralogical compositions.
Radiogenic isotopes (e.g., Nd, Hf, Sr, and Pb isotopes) are well known as powerful tools to identify sediment source end-member components and their provenance. Each end-member component shows characteristic ratios of the isotopes that are independent of their chemical and mineralogical compositions (Godfrey, 2002;Miyazaki et al., 2016;van de Flierdt et al., 2004;Vervoort et al., 2011;Yasukawa et al., 2019). For example, the Sr isotopic ratios ( 87 Sr/ 86 Sr) are greater than 0.712 in the terrigenous component, whereas they are less than 0.705 in the volcanic components and ∼0.7092 in modern seawater (Capo et al., 1998;Frank, 2002). Considering that deepsea sediments are mixtures of such components, it may be suggested that the radiogenic isotopic ratios of the bulk sediment reflect the proportions of these single source components (i.e., end-member components), and thus the proportions can be determined quantitatively. Therefore, the mineralogical and chemical compositions and the radiogenic isotopic ratios of the sediments together can provide higher-resolution information on the origin, and thus the depositional history, of the sediments. Previous research applied radiogenic isotopes to the sediments in the western North Pacific, where the sediments similar to those in deep-sea basins around Minamitorishima exist, to discuss binary mixing between Chinese Loess and volcanic ash (Asahara, 1999;Asahara et al., 1999;Jones et al., 1994Jones et al., , 2000Nakai et al., 1993). However, pelagic clay in the area around Minamitorishima Island also contains biogenic calcium phosphate (BCP) and hydrogenous Mn-oxides (Takaya et al., 2018;Yasukawa et al., 2020Yasukawa et al., , 2021. Thus, for a comprehensive understanding of the depositional history of pelagic clay, various components should be discussed, which will place the pelagic clay research in a broader context than before. Furthermore, high-resolution datasets of the geochemical characteristics of pelagic clay, including the chemical compositions and isotope ratios, could provide detailed constraints not only on the Earth's surface geochemical cycles, but also on the subducting materials, and may thus describe the processes and nature involving them in the Earth's interior (Mimura et al., 2019;Plank et al., 2007;Plank & Langmuir, 1998).
Here, we present new Sr isotope analyses of bulk sediments, Mn nodules, and fish teeth collected from the deepsea basin around Minamitorishima Island. Using geochemical data, including Sr isotope ratios and bulk chemical compositions, and microscopic observations, we estimated the contributions of the possible end-member  (NOAA National Geophysical Data Center, 2009; https://www.ngdc.noaa.gov/mgg/global/global.html) and were obtained by multi-narrow-beam echo sounder (Machida, Sato, et al., 2021), respectively.

KR13-02
This study Yasukawa  components and identified their origin and provenance using a quasi-quantitative method to analyze the transition and provenance of the sedimentary constituents and the environmental changes recorded in the deep-sea sediments in the western North Pacific Ocean.

Geological and Sedimentological Background
In the North Pacific pelagic realm, continental eolian dust is considered a major source of deep-sea sediments (Asahara et al., 1995;Kyte et al., 1993;Rea, 1994). Previous work demonstrated that the transport of eolian dust in the North Pacific has been affected by changes in atmospheric circulation patterns, such as the location of the Intertropical Convergence Zone (ITCZ; Pettke et al., 2000Pettke et al., , 2002Stancin et al., 2006;Hyeong et al., 2011Hyeong et al., , 2016Zhang et al., 2016). Dust in the North Pacific has been supplied mainly by Asia, called Chinese Loess, in the past 25 million years following the aridification of the Asian interior (Zhang et al., 2016). Notably, Hyeong et al. (2016) reported that the dust source in the eastern North Pacific changed at the Eocene-Oligocene boundary; they attributed this shift to the potential migration of the ITCZ. In contrast, although the Asian continent has been considered the dust source of the sediments in the Minamitorishima area during the Quaternary and Pliocene periods (Asahara, 1999;Asahara et al., 1999), possible changes in the dust source throughout the Cenozoic remain poorly understood.
It is noteworthy that abnormal concentrations of BCP were observed in the deep-sea sediments around Minamitorishima Island (Iijima et al., 2016;Takaya et al., 2018). BCP is the main host of REY in deep-sea sediments and accounts for more than 20,000 ppm of ΣREY (Arrhenius et al., 1957;Bernat, 1975;Kashiwabara et al., 2014Kashiwabara et al., , 2018Takaya et al., 2018;Toyoda et al., 1990). The high concentration of BCP results in the formation of REY peaks, including those in deep-sea sediments with ΣREY >5,000 ppm, which are called extremely REY-rich mud (Iijima et al., 2016;Takaya et al., 2018).
In addition, the deep-sea basins in the western North Pacific are surrounded by a number of seamounts and knolls, which could supply volcanic materials. Most of the seamounts in the western North Pacific were formed as hotspots in the area called the South Pacific Isotopic and Thermal Anomaly (5°S to 40°S, 105°W to 180°W) from 140 to 70 million years ago (Ma; Christie et al., 1995;Koppers et al., 2003). Today, the area with many Cretaceous seamounts in the western North Pacific is known as the West Pacific Seamount Province (WPSP; Koppers et al., 2003) and ranges from approximately 0°N to 35°N and 140°E to 175°E. In contrast, very young small knolls in the western North Pacific are called petit-spot volcanos (Hirano, 2011;Hirano et al., 2008Hirano et al., , 2019. Those near Minamitorishima Island have an age of <3 Ma (Hirano et al., 2019). Moreover, the Izu-Bonin-Mariana (IBM) volcanic arc in the western North Pacific was activated by the subduction of the Pacific plate beneath the Philippine Sea plate from 52 to 48 Ma Ishizuka, Tani, et al., 2011;Taylor & Goodliffe, 2004). In the early stage, the IBM arc was aligned east-west near the equator; however, it gradually rotated to a north-south alignment and migrated northward (Hall, 2002;Taylor & Goodliffe, 2004). Recent studies revealed that the volcanism in the early stage was characterized by forearc basalt at 52.5 Ma, and boninite for the infant arc edifice building at 51.3-46 Ma Brandl et al., 2017;Brounce et al., 2021;Ishizuka, Tani, et al., 2011, Ishizuka et al., 2018Robertson et al., 2018), followed by the mature arc edifice building (40-28 Ma;Brandl et al., 2017;Johnson et al., 2021). In particular, the volcanism in the latest Eocene to early Oligocene (∼37-28 Ma) was dominated by subaerial and subaqueous explosive volcanism producing the vitric ash, accompanied by the growth of the initial Izu-Bonin-Mariana arc, including the section called the Kyushu-Palau Ridge (KPR;Shipboard Scientific Party, 2002;Johnson et al., 2021). The KPR volcanism decreased between ∼25 and 22 Ma and stopped at ∼22 Ma, at the same time as the formation of Shikoku and Parece Vela backarc basins (Ishizuka, Taylor, et al., 2011;Taylor, 1992). Notably, volcanic activity continued in the Mariana arc even after the temporary cessation of volcanism in the Izu-Bonin arc (Straub et al., 2015). After vigorous activity resumed in the Izu-Bonin arc ∼17 Ma (Bryant et al., 2003), the IBM arc has been active, and explosive volcanism has continued (Ishizuka, Taylor, et al., 2011;Kutterolf et al., 2018;Robertson et al., 2018;Schindlbeck et al., 2018;Straub et al., 2010Straub et al., , 2015.

Cruise and Sample Information
In this study, bulk sediment samples obtained from five piston cores (KR13-02 PC04, PC05, and PC06; MR13-E02 Leg 2 PC01; and MR15-E01 Leg 2 PC02) and six pilot cores (KR13-02 PCPL04 and MR14-E02 PCPL04, PCPL05, PCPL08, PCPL09, and PCPL11) were analyzed, and all samples were collected within the Japanese exclusive economic zone (EEZ) around Minamitorishima Island during cruises KR13-02 of R/V Kairei and MR13-E02 Leg 2, MR14-E02, and MR15-E01 Leg 2 of R/V Mirai Iijima et al., 2016;Takaya et al., 2018;Tanaka, Nakamura, Yasukawa, Mimura, Fujinaga, Iijima et al., 2020). The sampling locations are shown in Figure 1 and are listed in Table S1 in Supporting Information S1. A piston corer consists of a 15-m long duralumin outerpipe and polycarbonate innertube, whereas a pilot corer is 2-m long exhibiting the same feature, and reaches the seafloor preceding the piston corer. For each site, the sediment lithology and physical properties of the pilot core and the topmost section of the piston core were generally identical. However, the sediment samples collected by pilot cores are more likely to retain detailed information around the topmost part of the sediment column, which is typically disturbed during piston coring.
We analyzed the 87 Sr/ 86 Sr of 38 bulk sediment samples from the five piston and six pilot cores noted above ( Figure 2, Table S2 in Supporting Information S2). In addition, fish tooth samples were collected from 22 horizons in KR13-02 PC05, and their Sr isotopic ratios were analyzed. The sampling points of all bulk sediments and fish teeth are shown in Figure 2, with the lithologies and REY contents of the cores reported in previous studies Iijima et al., 2016;Takaya et al., 2018;Tanaka, Nakamura, Yasukawa, Mimura, Fujinaga, Iijima et al., 2020).
This study newly analyzed bulk chemical compositions of 13 samples from three piston cores (KR13-02 PC05, PC06, and MR15-E01 PC02) and six pilot cores (KR13-02 PCPL04 and MR14-E02 PCPL04, PCPL05, PCPL08, PCPL09, and PCPL11; Figure 2, Table S2 in Supporting Information S2) to obtain a higher sampling resolution than that described in previous studies to complement the bulk sediment data set in the target horizons for Sr isotopic analysis, and to obtain the geochemical information of the uppermost sediment layer in the studied area. The bulk chemical compositions of the other 25 samples were reported by Fujinaga et al. (2016), Iijima et al. (2016), Takaya et al. (2018), and Tanaka, Nakamura, Yasukawa, Mimura, Fujinaga, Iijima et al. (2020), and are listed in Table S2 in Supporting Information S2.
All of the cores in this study were collected below the carbonate compensation depth in the Pacific (∼4.5 km; Van Andel, 1975;Pälike et al., 2012). The sediment samples are dark brown to blackish-dark brown and are categorized as pelagic clay or silty clay. Fujinaga et al. (2016), Iijima et al. (2016), Ohta et al. (2016Ohta et al. ( , 2020, and Tanaka, Nakamura, Yasukawa, Mimura, Fujinaga, Iijima et al. (2020), Tanaka, Nakamura, Yasukawa, Mimura, Fujinaga, Ohta et al. (2020) reported the details of the lithology based on the microscopic observations ( Figure 2). The main component of all cores is a clay-sized fraction of <4 μm. It is notable that the total proportions of silt-sized and sand-sized grains, which are mainly phillipsite and BCP, increase to >10% in the layers from 2.43 to 8.44 and 12.11-12.73 m below the seafloor (mbsf) in KR13-02 PC04, 0.50-3.67, and 6.82-11.45 mbsf in KR13-02 PC05, 0.50-3.00 mbsf, and 6.00-11.68 mbsf in KR13-02 PC06 Iijima et al., 2016;Ohta et al., 2016), and 0.00-1.96 mbsf in MR15-E01 Leg 2 PC02, respectively. The observed minor components are Fe-Mn micronodules, quartz, feldspar, reddish-brown semi-opaque oxides, colored unidentifiable minerals, radiolaria, diatoms, and siliceous sponge spicules. This study utilized the proportion of phillipsite based on the microscopic observations during/after the research cruises in Section 5.3.2, in addion to the reported data by Ohta et al. (2016Ohta et al. ( , 2020. All protocols followed those described by the International Ocean Discovery Program (IODP)  Tanaka, Nakamura, Yasukawa, Mimura, Fujinaga, Iijima et al. (2020), and this study. Orange, blue, and light blue triangles indicate the sampling points for bulk sediments (major and trace elements, and Sr isotopes), bulk sediments (Sr isotopes), and fish teeth (Sr isotopes), respectively. The chemostratigraphic units next to the REY content of the piston cores are from Tanaka, Nakamura, Yasukawa, Mimura, Fujinaga, Iijima et al. (2020). In addition, the chemostratigraphic units next to the REY content of the pilot cores are those of top 2 m of KR13-02 PC04 and MR14-E02 PC04, PC05, PC08, PC09, and PC11 from Tanaka by Mazzullo and Graham (1988) in the same manner as Ohta et al. (2016), and we estimated the proportions of all constituents visually and normalized the values to 100% in total.
The depositional ages of Units I and II, and the 1st REY peak are constrained by the Os isotope and ichtyolith stratigraphies of the two piston cores in this area (KR13-02 PC05 and MR14-E02 PC11; Nozaki et al., 2019;Ohta et al., 2020). Based on them, it is suggested that the bottom of Unit I corresponds to ∼11 Ma at KR13-02 PC05 and 12.0 Ma at MR14-E02 PC11; Unit II was deposited between 20.0 and 34.08 Ma at KR13-02 PC05 and between 15.0 and 34.0 Ma at MR14-E02 PC11; the 1st REY peak was deposited between 34.16 and 34.6 Ma at KR13-02 PC05 and between ∼34.0 and ∼35.0 Ma at MR14-E02 at MR14-E02 PC11. Although there are uncertainties owing to possible erosion(s) (Tanaka, Nakamura, Yasukawa, Mimura, Fujinaga, Iijima et al., 2020;Tanaka, Nakamura, Yasukawa, Mimura, Fujinaga, Ohta et al., 2020), it is suggested that Unit I was deposited after 12 Ma, Unit II was deposited between at least ∼15 and ∼34 Ma, and the 1st REY peak was deposited just before the Eocene-Oligocene boundary. On the other hand, Usui and Yamazaki (2021) reported paleomagnetic constraints for the initiation of the 1st REY peak deposition, from 38.615 Ma to 34.999 Ma during the latest Eocene, indicating that further investigations would be required to determine the precise age and sedimentation rate of the REY peak.

Methods
The bulk sediment samples were dried at 110°C prior to the analysis. In contrast to fish tooth samples (see below), they were not acid-leached before digestion. They were prepared by the glass bead method using Li 2 B 4 O 7 as a flux and LiBr as a releasing agent for major-element analyses (Kato et al., 1998(Kato et al., , 2002Yasukawa et al., 2014), by mix-acid (HF-HNO 3 -HClO 4, and HNO 3 -HCl-HF) digestions for trace-element analyses (Kato et al., 2005;Yasukawa et al., 2014Yasukawa et al., , 2019, and by acid digestion and purification using HF, HNO 3 , HClO 4 , and HCl for Sr isotope analyses (Miyazaki et al., 2018). Mn nodule samples were leached using 2.5 M HCl following the methods for Mn crusts described by Ling et al. (2005) and Amakawa et al. (2017) to extract elements associated with hydrogenous Mn oxides. Hand-picked fish tooth samples were precleaned by (a) reductive, (b) oxidative, and (c) acetic acid cleaning to remove (a) oxyhydroxides, (b) organic matters and opal, and (c) carbonate on the sample surface, following Gleason et al. (2002), and completely digested using HNO 3 . The leachate and digested samples of Mn oxides and fish teeth were prepared using the same method as that for the bulk sediment samples.
Major-and trace-element analyses were conducted at the Department of Systems Innovation, the University of Tokyo. The major-element content was determined using X-ray fluorescence (XRF) spectrometry (Rigaku ZSX Primus II) following the methods described by Kato et al. (1998Kato et al. ( , 2002 and Yasukawa et al. (2014). The trace-element content was determined using inductively coupled plasma quadrupole mass spectrometry (ICP-QMS; Thermo Fisher Scientific i-CAP Q) following Kato et al. (2005) and Yasukawa et al. (2014Yasukawa et al. ( , 2019. The analytical results were within 3% for major elements and generally within 5% for trace elements (relative percent difference) of the accepted values reported for the Geological Survey of Japan (GSJ) reference materials JB-1b and JB-2, respectively.
Digestion and purification for Sr isotopic analysis were conducted in a Level 100 clean room at Japan Agency for Marine-Earth Science and Technology (JAMSTEC) mainly following Miyazaki et al. (2018). Sr separation was performed using the AG50W-X8 cation ion-exchange resin (Bio-Rad, California, USA) and the Sr-spec resin (Eichrom Technologies, Inc., Illinois, USA). Column separation was performed using a fully automated open-column chemical separation system called COLUMNSPIDER, which was developed by JAMSTEC and HOYUTEC Co., Ltd. (Kawagoe, Japan; Miyazaki et al., 2012). Sr isotopes were measured via thermal ionization mass spectrometry using the TRITON TI ® instrument (Thermo-Finnigan, Bremen, Germany) at JAMSTEC following Takahashi et al. (2009). The mean 87 Sr/ 86 Sr values of the standard reference materials from the US National Institute of Standards and Technology SRM 987, and GSJ JB-2, and JMn-1 were 0.710241 ± 0.000024 (2 S.E., n = 34), 0.703674 ± 0.000025 (2 S.E., n = 12), and 0.709178 ± 0.000020 (2 S.E., n = 1), respectively. The total procedural blank concentration was below 35.9 pg. The analytical details are presented in Text S1 in Supporting Information S1.

Chemical Compositions of Bulk Sediments
The major-and trace-element contents of 13 bulk sediment samples analyzed in this study are in the range of previously reported data for the study area (Table S2 in Supporting Information S2; Fujinaga et al., 2016;Iijima et al., 2016;Takaya et al., 2018;Tanaka, Nakamura, Yasukawa, Mimura, Fujinaga, Iijima et al., 2020). On the basis of the chemostratigraphic unit classification proposed by Tanaka, Nakamura, Yasukawa, Mimura, Fujinaga, Iijima et al. (2020), a sample from KR13-02 PC05, two samples from KR13-02 PC06, and three samples from MR15-E01 PC02, whose major-and trace-element contents were measured in this study (Figure 2), could be categorized as Unit II, the 2nd REY peak, and Unit V, respectively (Table S2 in Supporting Information S2). In contrast, all seven samples from the top of the pilot cores ( Figure 2) are identified as the topmost unit, that is, Unit I (Table S2 in Supporting Information S2). The post-Archean average Australian shale-normalized (PAAS; Pourmand et al., 2012) REY patterns of the six samples from KR13-02 PC05, PC06, and MR15-E01 PC02 (i.e., Unit II, the 2nd REY peak, and Unit V, respectively; Figure S2 in Supporting Information S1) show negative Ce anomalies with relative enrichment of heavy rare earth elements; however, three Unit V samples from MR15-E01 PC02 have smaller negative Ce anomalies than those reported in the previous studies ( Figure S2 in Supporting Information S1). In addition, the seven samples from the pilot cores (i.e., Unit I) do not show a clear Ce anomaly ( Figure S2 in Supporting Information S1).

Sr Isotopic Ratios of Bulk Sediments, Mn Nodules, and Fish Teeth
The Sr isotopic ratios of the 38 bulk sediment samples are listed in Table S2 in Supporting Information S2. The maximum value was 0.716845 at 12.22 mbsf in MR15-E01 PC02, whereas the minimum was 0.708715 at 3.17 mbsf in KR13-02 PC05. In terms of the chemostratigraphy (Figure 3, Table S2 Table S4b in Supporting Information S2). In the depth profile of KR13-02 PC05 (Figure 4c), the 87 Sr/ 86 Sr values of the fish teeth decrease from 0.709123 at 0.13 mbsf to 0.708402 at 1.50 mbsf, although they are less variable below 2.12 mbsf (less than approximately 0.7084).

Geochemical Characteristics of the Fish Teeth and Mn Nodule Samples
First, we discuss the geochemical characteristics, in particular the Sr isotopic ratios, of the fish teeth and Mn nodule samples around Minamitorishima Island. The range of Sr isotopic ratios of the fish teeth in this study ( 87 Sr/ 86 Sr = 0.707854-0.709123) are similar to those in the seawater 87 Sr/ 86 Sr records during the Cenozoic reported by Veizer et al. (1999), Frank (2002), and McArthur et al. (2012. Strontium isotope ratios in pelagic seawater were homogenous at each age, regardless of the location, because the residence time of Sr (>10 6 years) is sufficiently longer than the timescale of ocean circulations (several hundred to 1,500 years; Broecker & Peng, 1982;Frank, 2002;Ravizza & Zachos, 2014). However, considering the depositional ages of Units I and II, and the 1st REY peak (Middle Miocene to Pliocene, Oligocene to Early Miocene, and the latest Eocene to the Eocene-Oligocene boundary, respectively; Nozaki et al., 2019;Ohta et al., 2020;Usui & Yamazaki, 2021), the 87 Sr/ 86 Sr values of the fish teeth could be higher (0.709123 in Unit I, 0.708015-0.708837 in Unit II, and 0.708022-0.708189 in the 1st REY peak) than those in seawater at each plausible time interval of deposition (0.708250-0.709050 during the Middle Miocene to Pliocene, 0.707850 to 0.708250 during the Oligocene to Early Miocene, and 0.707800 to 0.707900 during the latest Eocene to the Eocene-Oligocene boundary; Veizer et al., 1999;McArthur et al., 2012McArthur et al., , 2020. Therefore, it is likely that the fish teeth did not perfectly preserve the marine Sr isotopic values at the time of deposition. A possible reason for the shift to higher Sr isotopic ratios is the effects of diagenetic alteration by pore water (Ingram, 1995;Martin & Scher, 2004). Although the degree of alteration of fish teeth in pelagic clay by pore water in this area has not been assessed, we have accepted the measured values as the limits of the range of Sr isotopic ratios maintained by fish teeth in the sediments in this study.
In contrast, all of the Mn nodule leachate samples show very similar Sr isotopic ratios ( 87 Sr/ 86 Sr = 0.709096-0.709181; Table S4 in Supporting Information S2), as noted in Section 4.2. These values are close to those of modern seawater (Frank, 2002;McArthur et al., 2012;Veizer et al., 1999) and a Mn nodule on the sediment surface reported in previous research (Ito et al., 1998). This similarity could be attributed to the continuous exchange of Sr between the leachable Mn oxide phase and the surrounding seawater, regardless of the origin of the Mn oxide (i.e., whether it is hydrogenous, diagenetic, or hydrothermal; Clauer et al., 1984;Ito et al., 1998). Therefore, the Mn nodules in this study could also reflect Sr isotopes in modern seawater.

Continental Components
The geochemical characteristics of the bulk sediment samples are described on the basis of 87 Sr/ 86 Sr ratios and the PAAS-normalized REY patterns. The bulk sediment samples show significant variations in 87 Sr/ 86 Sr, which ranges from 0.708715 (at 3.17 mbsf in KR13-02 PC05) to 0.716845 (at 12.22 mbsf in MR15-E01 PC02; Figure 3, Table S2 in Supporting Information S2). This variation can be attributed to variable mixtures of the end-members. The 87 Sr/ 86 Sr of samples from the uppermost and lowermost sediment layers (Units I and V; Tanaka, Nakamura, Yasukawa, Mimura, Fujinaga, Iijima et al., 2020) are relatively high (>0.712; Figure 3, Table S2 in Supporting Information S2). In contrast, the sediments in Units II to IV have lower 87 Sr/ 86 Sr values than those in Units I and V ( Figure 3, Table S2 in Supporting Information S2), suggesting an increase in the proportion of components with low 87 Sr/ 86 Sr ratios or a decrease in the proportion of components with high 87 Sr/ 86 Sr ratios in Units II to IV.
Possible components with high Sr isotopic ratios are continental rocks, which generally have high 87 Sr > 0.712;Chen et al., 2007;Gallet et al., 1996;Jahn et al., 2001;Kanayama et al., 2005;McCulloch & Wasserburg, 1978;Pearce et al., 2015;Sun, 2005;Yokoo et al., 2004); the values in Units I and V are consistent with this range. The PAAS-normalized REY patterns of the Unit I samples are almost flat ( Figure S2 in Supporting Information S1). In addition, previous studies have reported the predominance of terrigenous material in Unit I, specifically from the Chinese Loess Plateau or Taklimakan desert (Tanaka, Nakamura, Yasukawa, Mimura, Fujinaga, Iijima et al., 2020;Yasukawa et al., 2019). Therefore, the high 87 Sr/ 86 Sr values of the bulk sediments in Unit I are mainly attributed to terrigenous material from East Asia. In contrast, the bulk REY concentrations in Unit V are higher than those in Unit I ( Figure S2, Table S2 in Supporting Information S2) and are generally in the same range as those in Units II to IV. This result suggests that the 87 Sr/ 86 Sr values of Unit V could result from the mixing of terrigenous materials with higher 87 Sr/ 86 Sr values than those in Unit I and other materials with relatively low 87 Sr/ 86 Sr values.

Proportions of BCP and Mn Oxides in Bulk Sediment
In this section, we discuss the mixing proportions of the terrigenous components and other possible constituents on the basis of the characteristic element compositions. Here, we assume that the eolian dust of Chinese Loess (Table S5 in Supporting Information S2; Chen et al., 2007;Gallet et al., 1996;Jahn et al., 2001;Sun & Zhu, 2010;Taylor et al., 1983;Yokoo et al., 2004) is a modern, plausible terrigenous end-member of the pelagic sediments in the study area. This is because the Unit I samples are dominated by the terrigenous component, as suggested in Section 5.2.1, which seems to reflect the final deposition of materials as a result of wind-driven dust transport from Asia (Asahara, 1999;Asahara et al., 1999;Yasukawa et al., 2019;Zhang et al., 2016).
For other possible end-member constituents, we next focus on BCP as the main host of REY in the deep-sea sediments, as well as hydrogenous Mn oxides, owing to their high concentrations of transition metals (Arrhenius et al., 1957;Bernat, 1975;Kashiwabara et al., 2014Kashiwabara et al., , 2018Machida et al., 2016;Takaya et al., 2018;Toyoda et al., 1990;Yasukawa et al., 2016), both of which were commonly observed in the REY-rich mud in microscopic observations .

Proportions of BCP
The samples with relatively low Sr isotopic ratios-Units II, III, and IV and the REY peaks-may have been more affected than Unit I by components other than terrigenous materials. Their PAAS-normalized REY patterns show a clear negative Ce anomaly and enrichment in middle to heavy rare-earth elements ( Figure S2 in Supporting Information S1; Fujinaga et al., 2016;Iijima et al., 2016;Tanaka, Nakamura, Yasukawa, Mimura, Fujinaga, Iijima et al., 2020). These features resemble those of BCP represented by fish teeth and bone fragments in deep-sea sediments (Takaya et al., 2018;Toyoda et al., 1990). Moreover, these samples are categorized as REY-rich mud, that is, sediments with ΣREY >400 ppm (Kato et al., 2011;Tanaka, Nakamura, Yasukawa, Mimura, Fujinaga, Iijima et al., 2020), and contain several percent to 30% BCP according to microscopic observations . Therefore, the addition of fish teeth (BCP) with relatively low Sr isotopic ratios ( 87 Sr/ 86 Sr = 0.707854-0.709123; Section 4.2) can explain the lower 87 Sr/ 86 Sr values in bulk REY-rich mud compared to those of Unit I.
The proportion of BCP in the bulk sediment samples was estimated by considering the mixing between the terrigenous component and BCP. For the estimation, we used the P 2 O 5 content as an indicator of BCP. The BCP in deep-sea sediments in the study area contains 29.8 wt.% P 2 O 5 on average (Table S5 in Supporting Information S2; Takaya et al., 2018). In contrast, the terrigenous component (e.g., Chinese Loess) and Mn nodules of mainly hydrogenous origin that were collected ∼300 km east of Minamitorishima Island (hereafter referred to as Minamitorishima Mn nodules) contain ∼0.15 wt.% and ∼0.99 wt.% P 2 O 5 , respectively (Table S5 in Supporting Information S2; Gallet et al., 1996;Fujinaga et al., 2016;Machida et al., 2016). The effects of organic-derived phosphorous could be considered to be negligible because organic matter was not confirmed by smear slide observations . Indeed, sediment samples collected in the same area generally contain less than 0.1 wt.% total organic carbon (i.e., the core at ∼14 km northwest of KR13-02 PC05; Y. . Therefore, almost all of the phosphorous content in the samples can be considered to originate from BCP. The estimation method is described in Text S2 in Supporting Information S1. The estimated proportions of BCP for each sample are shown in Table 1. The BCP proportion in Unit I is very low (<1%), whereas those of Units II to V and the REY peaks are greater than 1%. Notably, the REY peak samples have the largest proportions of BCP, ranging between 12% and 33%. The proportions of BCP in Units II, III, and V range between 1.5% and 5%, except for two samples, whereas those in Unit IV are relatively high (5%-12%).
Notably, the bulk chemical compositions of Units II to V reflect the BCP input. In addition to phosphorous, CaO, Th, U, and REY are enriched in BCP (Arrhenius et al., 1957;Toyoda et al., 1990;Trueman & Tuross, 2002 this study) are almost identical to those of P 2 O 5 content. In addition, they resemble the depth profile of the estimated proportions of BCP in Cores KR13-02 PC04 and PC05 (Table 1), which supports the validity of our estimation. This result also implies that REY, Th, and U contents, which are widely used in the discussion of petrological characteristics, could be less useful for the sediments rich in BCP.

Proportions of Hydrogenous Mn Oxides
Another material that may lead to a decrease in the 87 Sr/ 86 Sr ratio of bulk sediment is hydrogenous Mn oxides. Dubinin et al. (2017) reported that the 87 Sr/ 86 Sr of Mn micronodules in sediments from the Brazil basin in the Atlantic Ocean is similar to that of modern seawater; thus, they proposed that Sr exchange has occurred between the micronodules and ambient pore water/seawater. Yasukawa et al. (2020) estimated that the major portion (50%-90%) of Fe-Mn (oxyhydr)oxides in the REY-rich mud around Minamitorishima Island is in the form of hydrogenous Mn microparticles, which were originally reported in the South Pacific pelagic clay (Uramoto et al., 2019). Considering that the leachate of macroscopic Mn nodules of hydrogenous origin showed modern seawater-like values ( 87 Sr/ 86 Sr = 0.709096-0.709181) regardless of the sampling position (i.e., in both the surface and inner layers of the nodules), the exchange of Sr with seawater could be a common phenomenon in both macroscopic Mn nodules on the seafloor and Mn microparticles in the sediment. Therefore, the hydrogenous Mn oxides in the REY-rich mud could affect the Sr isotopic composition of the bulk sediments.
The proportion of hydrogenous Mn oxides in each bulk sediment sample was estimated by considering the mixing between the terrigenous component and hydrogenous Mn oxides. For the estimation, we used the Co content as an indicator of hydrogenous Mn oxides. Zhou and Kyte (1992) suggested that Co precipitates slowly at a constant  Machida et al., 2016). Therefore, most of the Co content in the bulk sediment could have been derived from the hydrogenous Mn oxides by slow precipitation.
The proportion of hydrogenous Mn oxides in Unit I is less than 2% (Table 1), as determined by the estimation method described in Text S2 in Supporting Information S1. In contrast, most of the samples from Units II to V and the REY peaks are enriched in Co and hydrogenous Mn oxides, with contents of 2%-5.5% (Table 1). It is notable that three samples in MR15-E01 PC02 show relatively low REY contents (ΣREY = 410-660 ppm; Table S2 in Supporting Information S2), suggesting that they contain less BCP compared to the other samples. Therefore, the negative Ce anomaly of BCP could be partly offset by hydrogenous Mn-oxides, resulting in the smaller negative Ce anomaly in the PAAS-normalized REY patterns of bulk sediments (see Section 4.1, Figure  S2 in Supporting Information S1). Figure 6 shows the depth profiles of Co, MnO, and Ni contents of Cores KR13-02 PC04 and PC05 (Iijima et al., 2016;this study), and those of the proportions of Mn-oxides (Table 1). The depth profiles of the MnO and Ni contents in Cores KR13-02 PC04 and PC05 resemble those of the Co contents and estimated Mn-oxide proportions; that is, the increasing trends from Unit I to the 1st REY peak and from the bottom of Unit III to 2nd REY peak, the maximum value in the 1st REY peak, the decreasing trend in Unit IV, and the minimum value from the bottom of Unit IV to Unit V. MnO and Ni are also incorporated into the Mn-oxide phase regardless of its size; that is, macro-nodules, Fe-Mn micronodules, and microparticles (Hein & Koschinsky, 2014;Uramoto et al., 2019;Yasukawa et al., 2020). These consistent features in the depth profiles imply that the proportions of Mn-oxides were estimated properly.

Model 1: Terrigenous Component + BCP + Mn Oxides
A model calculation is introduced to reproduce the observed depth profiles of the bulk Sr isotopic ratios, as shown in Figure 4. Considering the observed geochemical features discussed in Section 5.2, we first considered a model of three-component mixing of the terrigenous component, BCP, and hydrogenous Mn oxides (Table 2), which were labeled as Model 1. This ternary mixing model is consistent with the result of independent component analysis by Yasukawa et al. (2019), although the data set statistically analyzed (n = 567) was merely a part of the entire data set in the study area (totally n = 1,685, including the above 567 samples; Tanaka, Nakamura, Yasukawa, Mimura, Fujinaga, Iijima et al., 2020;Tanaka, Nakamura, Yasukawa, Mimura, Fujinaga, Ohta et al., 2020). The calculation methods are based on those reported by Langmuir et al. (1978), Holland (1984), and Palmer and Elderfield (1985), details of which are provided in Text S3 in Supporting Information S1. Table S5 in Supporting Information S2 shows representative compositions of the potential end-member components used in the model calculation.
Model 1 can reproduce a major portion of the measured 87 Sr/ 86 Sr ratio (Figure 7b). The exceptions are the intervals at Unit II and the 1st REY peak, and Units III and IV. In these intervals, the calculated 87 Sr/ 86 Sr values are higher than the observed values. This discrepancy suggests that other end-member(s) with low 87 Sr/ 86 Sr are present in these intervals and that a three-component mixing model is insufficient to explain the 87 Sr/ 86 Sr ratios of the bulk sediments.

Volcanic Components as Additional End-Member Components
Model 1 requires the addition of components with low 87 Sr/ 86 Sr ratios between Unit II and the 1st REY peak and in Unit IV. One possible component with low 87 Sr/ 86 Sr ratios is a volcanic material.  (Gale et al., 2013;Koppers et al., 2003;Machida et al., 2009;Miyazaki et al., 2015;Savov et al., 2006;Straub et al., 2010Straub et al., , 2015Straub et al., , 2017White & Klein, 2014). Previous studies have not discussed any notable enrichment of volcanic components in the study area on the basis of the chemical compositions of the bulk sediments Iijima et al., 2016;Yasukawa et al., 2019). Regarding the statistical analysis reported by Yasukawa et al. (2019), this is probably because only a limited number of elements were used (i.e., major elements, REY, Co, Ni, and Cu). However, Asahara (1999) and Asahara et al. (1999) reported that a volcanic component was supplied to the sediment in the area around Minamitorishima Island. In addition, Ohta et al. (2016) reported zeolite-rich intervals in KR13-02 PC05: 0.5 to 2.7 mbsf and 7.9 mbsf to the bottom of the core (Unit II to the 1st REY peak, and Units IV to V). These intervals are generally classified as zeolitic clay or clay with zeolite, owing to the presence of 10%-30% phillipsite according to smear slide observations (cf. Section 3). Phillipsite has been considered to form during early diagenesis of volcanic glass under conditions with a low sedimentation rate (Bernat & Goldberg, 1968;Bernat et al., 1970;Czyscinski, 1973;Stonecipher, 1976). It is notable that phillipsite shows low Sr concentrations (<35 ppm; Takaya et al., 2018), suggesting an elution of  (Iijima et al., 2016; this study) with chemostratigraphic units (Tanaka, Nakamura, Yasukawa, Mimura, Fujinaga, Iijima et al., 2020) Bryant et al., 2003;Christie et al., 1995;Gale et al., 2013;Koppers et al., 2003;Machida et al., 2009;Miyazaki et al., 2015;Savov et al., 2006;Straub et al., 2010Straub et al., , 2015Straub et al., , 2017White & Klein, 2014). The observed modal composition of phillipsite in each sample, obtained by smear slide observations, was applied as the abundance of volcanic components ; Table S7 in Supporting Information S2). For the Mariana and Izu-Bonin arcs, only tephra records between 40 and 23 Ma were applied because they formed mature arcs before the cessation of the Izu-Bonin volcanism in the Oligocene (Brandl et al., 2017;Johnson et al., 2021;Shipboard Scientific Party, 2002).

Model 2: Terrigenous Component + BCP + Mn Oxides + MORB
Model 2 is based on the assumption that MORB (Gale et al., 2013;White & Klein, 2014) is a volcanic end-member (Table 2). The calculated 87 Sr/ 86 Sr values using Model 2, as shown in Figure 7c, are similar to the measured values at ∼8 mbsf, or Unit IV. As a result, the addition of MORB results in better fitting of the 2nd REY peak to Unit IV compared to Model 1, although the calculated values remain higher than the measured values of Unit II.

Model 3: Terrigenous Component + BCP + Mn Oxides + Lavas From WPSP
The WPSP was active during the Cretaceous, from approximately 140 to 70 Ma (Christie et al., 1995;Koppers et al., 2003). In addition, the boundary of the pelagic clay and chert around Minamitorishima Island is reportedly late Campanian (Shipboard Scientific Party, 1990). The depositional age of the pelagic clay around Minamitorishima Island suggests that the WPSP may affect the lowermost part of the piston core. Based on the calculation, the measured 87 Sr/ 86 Sr values are in the range of the 87 Sr/ 86 Sr values of Model 3 throughout the core (Figure 7d). Thus, the WPSP might be a possible source of volcanic materials on the basis of the Sr isotope modeling.

Model 4: Terrigenous Component + BCP + Mn Oxides + Lavas From Petit-Spot Volcanos
Petit-spot volcanos are also among the closest sources of volcanic material in the area around Minamitorishima Island (e.g., Hirano et al., 2019), although the reported ages of petit-spot volcanos around Minamitorishima Island are markedly younger (<3 Ma; Hirano et al., 2019) than the onset of Unit II deposition (Nozaki et al., 2019;Ohta et al., 2020;Usui & Yamazaki, 2021). Model 4 considers the mixing of the terrigenous component, BCP, hydrogenous Mn oxides, and petit-spot basalt (Table 2). It yields similar Sr isotopic values to the sample data in Units II and IV, as in the case of Model 3 ( Figure 7e); thus, petit-spot volcanos could be another possible source of the volcanic material in the sediments.

Constraints of the Volcanic End-Member Sources Based on the Bulk Chemical and Mineralogical Compositions With Sr Isotope Ratios
The model calculations in Section 5.3.2 support the possibility that the input of volcanic materials, such as MORB, lavas from WPSP and petit-spot volcanos, and tephra from the Mariana and Izu-Bonin arcs (40-23 Ma), could lower the bulk 87 Sr/ 86 Sr isotope values at Unit II, the 1st REY peak, and Unit IV. In addition, they suggest that phillipsite could be useful for estimating the proportions of volcanic materials, even though strontium could have been eluted during phillipsite crystallization or alteration of the original volcanic materials. To strongly test the Sr modeling results, this section discusses the source of the terrigenous and volcanic materials based on the bulk chemical and mineralogical compositions, in addition to the 87 Sr/ 86 Sr values. The discussion in this section includes the 87 Sr/ 86 Sr ratios not only reported in this study (Table S2 in Supporting Information S2), but also reported by Yasukawa et al. (2019), that is, samples from KR13-02 PC04 and PC05, MR13-E02 PC03, and MR14-E02 PC05 and PC08 (Table 1). Furthermore, to expand the data set and to discuss the variation among the units, the scatter diagrams in this section include all chemical compositional data obtained from all eight piston and six pilot cores ( Tanaka, Nakamura, Yasukawa, Mimura, Fujinaga, Iijima et al. (2020), and this study.
The proportional changes in the terrigenous and volcanic components can be observed in the depth profiles of Rb, Cs, and Ba contents in KR13-02 PC04 and PC05 (Figure 8). Previous studies reported that Rb, Cs, and Ba content values are high in continental materials, but are low in volcanic materials Plank & Langmuir, 1998). Figure 8 shows that Unit II, the 1st REY peak, and Unit IV contain less than 100 ppm of Rb, 7.5 ppm of Cs, and 400 ppm of Ba. This strongly suggests that the proportions of the volcanic materials in Unit II, the 1st REY peak, and Unit IV are greater than those in continental materials. In other words, Units I and III, which show high Rb and Cs contents, are dominated by the input of continental materials. This is supported by the observation that the estimated proportion of terrigenous materials exhibits a depth profile similar to that of the Rb and Cs content values in KR13-02 PC04 and PC05 (Figure 8). The characteristics of Unit V are different from those of Units I and III; Unit V shows high Rb (approximately 100 ppm) but low Cs (less than 7.5 ppm), which will be discussed in Section 5.4.3.
The La/Yb ratio and Th content of bulk sediments provide another constraint for the provenance of the volcanic materials in Unit II (Figures 9a and S3 in Supporting Information S1). Since, Th is incorporated into BCP (cf. Section 5.2.2.1), the 1st and 2nd REY peaks exhibit Th enrichment around La/Yb = ∼6-7. On the other hand, Unit I sediments, which are mainly derived from terrigenous components, show a trend toward Chinese Loess in Figure 9a. Although Units III and V are located in the mixing trend between BCP and Chinese Loess (or between the REY peaks and Unit I), Unit II deviates from the mixing trend due to lower La/Yb ratios than those of BCP and Chinese Loess and shifts toward the Izu-Bonin and Mariana tephras or MORB, rather than the lavas from WPSP or petit-spot volcanos.
In   (Iijima et al., 2016;Tanaka, Nakamura, Yasukawa, Mimura, Fujinaga, Iijima et al., 2020; this study) with chemostratigraphic units (Tanaka, Nakamura, Yasukawa, Mimura, Fujinaga, Iijima et al., 2020)  Some of the Unit IV samples exhibit low Th and La/Yb values (Th = ∼10 ppm, La/Yb = ∼7-7.5) relative to those of Units I, III, V, and the REY peaks in Figure 9a, and exist near the Unit II samples in Figure 9b. These results suggest that the Unit IV samples may be affected by volcanic components, like Unit II. However, in Figure 9c, the Unit IV samples have lower Th/La values than those of Unit II. In addition, they show slightly higher Nb/La values compared to those of BCP and REY peaks and exist in the simple Chinese Loess-BCP mixing trend. Thus, the Unit IV values in Figures 9a-9c imply that the volcanic materials in Unit IV may be different from those in Unit II, which has similar 1/Sr and 87 Sr/ 86 Sr values to those of the tephra from the IBM arc (Figure 9b), but higher La/Yb ratios than those of the tephra from the IBM arc ( Figure 9a).
The Unit I samples on the scatter diagram between 1/Sr and 87 Sr/ 86 Sr show lower Sr isotope ratios than those of Chinese Loess, implying that Unit I could contain a certain amount of volcanic materials. Although it is not mentioned in previous studies, the shipboard core description recognized the presence of volcanic glass in the Unit I sediments in certain cases (e.g., MR13-E02 cores). Indeed, another two-component mixing model based on Sr isotope ratios suggested that the western North Pacific pelagic clay deposited during the last 3 Myr contains volcanic materials constituting up to 20%-25% (Asahara, 1999).

Change in the Sedimentary End-Members Around Minamitorishima Island
Our model calculation suggested a fundamental change in the sedimentary end-members around Minamitorishima Island during the depositional history of the area. As discussed above, the contributions of BCP and Mn oxides to the bulk chemical composition of the sediments are 0%-33% and 0%-5.5%, respectively (cf. Section 5.2, Table 1). These values are almost the same as the results based on microscopic observations reported by Ohta et al. (2016Ohta et al. ( , 2020 Figures 7 and 9). The depth profiles of Sr isotope modeling illustrate almost the same trends as the analyzed Sr isotope ratios (Figure 7), which suggests (a) the effectiveness of the visual estimation by microscopic observations and (b) the validity of the assumption that eluted volcanic Sr has been almost maintained in the surrounding sediments. Significant volcanic inputs occurred at the following two intervals: from Unit II to the 1st REY peak and below the 2nd REY peak (i.e., Units IV and V), based on the smear slide observations. The maximum estimated contribution of the volcanic components is over 30% (Table S7 in Supporting Information S2). The bulk chemical compositions of Unit II and the 1st REY peak suggested that they could contain tephra from the IBM arc (cf. Section 5.3.3, Figure 9). This indicates that a substantial amount of volcanic material began to be supplied to the seafloor around Minamitorishima Island from the Izu-Bonin and Mariana arc during the deposition of Unit II and the 1st REY peak. The input of volcanic material from the Izu-Bonin and Mariana arc is consistent with the silicate fraction observed in the western North Pacific pelagic clay by Asahara (1999). In contrast, the bulk 87 Sr/ 86 Sr and chemical compositions revealed that the provenance of volcanic materials in Unit IV could be different from that of Unit II and the 1st REY peak (cf. Section 5.3.3, Figures 7 and 9), and those of Unit V are discussed in Section 5.4.3.

Relationship Between the Input of Volcanic Material and Geographical Setting
Here, the additional input of volcanic material to the study area is considered in terms of spatiotemporal changes in the geographical setting. The 1st REY peak and Unit II were deposited between the latest Eocene and the Eocene-Oligocene boundary, and between ∼34 and ∼15 Ma, respectively (Nozaki et al., 2019;Ohta et al., 2020;  Tanaka, Nakamura, Yasukawa, Mimura, Fujinaga, Iijima et al. (2020), and this study. In the figures, the representative endmembers are also shown. Chinese Loess are from Taylor et al. (1983), Gallet et al. (1996), Jahn et al. (2001), Yokoo et al. (2004), and Chen et al. (2007); biogenic calcium phosphate (BCP) is from Takaya et al. (2018), and this study; North American Shale Composite is from Haskin et al. (1966), Gromet et al. (1984), and Pearce et al. (2015); MORB are from Class & Lehnert 2012; yellow area with the global average (yellow star; Gale et al., 2013;White & Klein, 2014); data on lavas from West Pacific Seamount Province are from the data set of Anewetak Seamount Trail, Gilbert Ridge Seamounts, Magellan Seamount Trail, Ratak Seamount Group, and Ujlan Seamounts retrieved from GeoROC, 2021a, 2021b, 2021c, 2021d, 2021e; orange area with the average value (orange star; Christie et al., 1995;Koppers et al., 2003); data on lavas from petit-spot volcanos are from Machida et al. (2009) and Miyazaki et al. (2015); tephra from the Mariana arc (40-23 Ma) is from Savov et al. (2006) and Straub et al. (2015Straub et al. ( , 2017; tephra from the Izu-Bonin arc (40-23 Ma) is from Bryant et al. (2003) and Straub et al. (2010). A blue dashed circle in (b) implies the Units II samples with low 87 Sr/ 86 Sr values, which are not located between the Unit I and the rare earth elements and yttrium (REY) peaks. As for (a) and (c), the mixing curves between Chinese Loess and BCP, and between Chinese Loess and the tephra from the Izu-Bonin arc with their reference ratios are shown in Figure S3 in Supporting Information S1. Usui & Yamazaki, 2021). During these intervals, the input of volcanic materials from the Izu-Bonin-Mariana arc increased. Previous studies suggested that the mature IBM arc was formed in the late Eocene to early Oligocene (40-28 Ma) with explosive volcanism (37-28 Ma; Johnson et al., 2021;Shipboard Scientific Party, 2002). Therefore, the results of this study, as well as the estimated age of the volcanic activity of the IBM arc (i.e., the latest Eocene to Oligocene), suggest that the introduction of volcanic materials to the area around Minamitorishima Island could be caused by the explosive volcanism forming the mature IBM arc.
According to the paleogeographic reconstructions reported by Müller et al. (2019), Minamitorishima Island has remained in the east of the IBM arc at almost the same latitude since at least the middle Eocene (∼45 Ma; Figure 10). Therefore, in addition to the regional tectonic activity, the northwestward movement of Minamitorishima Island toward the IBM arc due to Pacific Plate motion could also be responsible for the onset of volcanic input to the area around Minamitorishima Island. Although Hyeong et al. (2016) suggested that the ITCZ migrated across the equator from south to north during the global cooling around the Eocene-Oligocene boundary (∼34.3-34.0 Ma), the Minamitorishima area was most likely under the influence of the westerlies around the Eocene-Oligocene transition, because of the increased input of the Mariana arc volcanic component to this area during the 1st REY peak and Unit II deposition. This implies that the volcanic materials from the IBM arc could be transported to the area located more than 3,000 km away at that time.
Unit IV could have also experienced an increase in volcanic components that could be different from those in Unit II (i.e., the tephra from the Izu-Bonin-Mariana arc), as discussed in Section 5.3.3. The depositional age of Unit IV has not been determined, but the onset of pelagic clay deposition around Minamitorishima Island was estimated to be ∼75 Ma (Shipboard Scientific Party, 1990). Considering the paleo-location of the study area from 75 to ∼34.4 Ma (Figure 10), the longitude of Minamitorishima Island was approximately 170°W-170°E in the central Pacific Ocean, which should have been several thousand kilometers from both the East Pacific Rise (EPR) and the IBM arc. This result implies that neither MORB from the EPR nor volcanic tephra from the IBM arc could have affected Unit IV. Therefore, further research on the volcanic activity that influenced Unit IV is necessary to explain the chemical compositional data (e.g., slightly high Nb/La in Figure 9c).
Th content values of Unit V are similar to those of Chinese Loess and Unit I; however, their La/Yb ratios are lower than those of Chinese Loess (Figure 9a). This tendency of Unit V is similar to that of the North American shale composite (NASC; Gromet et al., 1984;Haskin et al., 1966). In addition, in the scatter diagram between Rb and Cs (Figure 9d), the Unit V samples clearly deviate from the other units and REY peaks, shifting toward the NASC composition. These chemical compositional data support that North America could be considered the provenance of the terrigenous component in Unit V.
Focusing on the eastern Pacific Ocean, there was a substantial amount of volcanism in the Central/South America, such as the Caribbean Large Igneous province (CLIP; Hauff et al., 2000;Sinton et al., 1998) in the Late Cretaceous, and multiple explosive volcanic activities forming the ignimbrite and tephra layers in the Caribbean area during the Eocene and Miocene (Central American Tertiary Ignimbrite Province; Jordan et al., 2006Jordan et al., , 2007. Although Unit V contains 1%-40% of phillipsite, which implies the input from substantial volcanic materials, the Caribbean volcanic activities do not account for the Sr isotope ratios or formation ages of the sediments: Unit V with higher Sr isotope ratios than the other units deposited between ∼75 Ma and the latest Eocene,   (Hauff et al., 2000;Jordan et al., 2007;Sigurdsson et al., 2000). Although a part of the eruption age of the Central American Tertiary Ignimbrite Province is within the range of ∼75 Ma to the latest Eocene, Carr et al. (2007) concluded that the prevailing wind direction around the province during the eruption was from west to east. Thus, it is unlikely that this volcanism was a source of materials for Unit V.
Meanwhile, the chemical and isotopic compositional features imply that the provenances of the terrigenous components of Units I and III may be different. The Unit III samples exhibit high Cs and Rb contents (Figures 8  and 9d) and are scattered on the mixing trend between Chinese Loess-BCP (Figure 9c), which implies that they are less affected by volcanic materials compared to the other units or REY peaks. On the other hand, the observed Rb content is higher than that calculated in Unit III ( Figure 8) and may be attributed to a change in the terrigenous dust source. Indeed, the observed 87 Sr/ 86 Sr values of KR13-02 PC05 are lower than the calculated 87 Sr/ 86 Sr between 5.97 and 3.97 mbsf, Unit III, in all models (Figure 7). This tendency might be explained by a brief input of terrigenous components with low 87 Sr/ 86 Sr values. Based on the discussion on the dust source of Unit V mentioned above, it can be inferred that one possibility is a substantial input of Central/South American dust with low 87 Sr/ 86 Sr ratios, which could be eroded and transported by the trade winds, potentially affecting Unit III. Another possibility is the terrigenous components called "eolian 2," which was deduced from ODP Site 1149 by Scudder et al. (2014), located east of the Izu-Bonin arc. They suggested the presence of a continental crustal material, other than the Chinese Loess, that had been supplied to the western North Pacific until the Paleogene when Unit III was deposited. In addition, the 87 Sr/ 86 Sr ratios of Chinese Loess used in this study were mainly obtained from the Pliocene section; therefore, the actual 87 Sr/ 86 Sr ratios of dust from Asia during the early Paleogene are unclear, and have the potential to lower the 87 Sr/ 86 Sr ratios of the deep-sea sediments. Therefore, based on the currently available data, we cannot constrain the provenance of Unit III. More comprehensive data for Unit III as well as Unit V, such as other isotope ratios (e.g., Nd, Pb, etc.), are required for future investigations.

Conclusions
This study measured the 87 Sr/ 86 Sr of deep-sea sediments, fish teeth (BCP), and Mn nodules collected from the seafloor around Minamitorishima Island in the western North Pacific Ocean. We demonstrated that the BCP could be affected by a diagenetic process, and thus its 87 Sr/ 86 Sr ratio seems to be different from that of seawater at the time of deposition. Mn oxides also show modern seawater-like 87 Sr/ 86 Sr values in all of the inner layers of the nodules, suggesting a continuous exchange of Sr between the Mn oxides and contemporary seawater.
According to the PAAS-normalized REY patterns, the sediment samples from the top of the cores (chemostratigraphic Unit I) are composed mainly of terrigenous materials, whereas the samples from the deeper part of the cores (chemostratigraphic Units II to V) are affected by the addition of other components with lower 87 Sr/ 86 Sr values compared to the continental crust, such as BCP and hydrogenous Mn oxides. The P 2 O 5 content indicates that the proportion of BCP reached 12%-33% during the deposition of the 1st and 2nd REY peaks. In addition, the Co content suggests that the proportion of hydrogenous Mn oxides in Units II to V is 2%-5.5%.
Calculations of simple Sr mixing models and bulk chemical compositions of the samples revealed that four end-member components have been affected by the deep-sea sediments around Minamitorishima Island: a terrigenous component, BCP, hydrogenous Mn oxides, and a volcanic component. The input of volcanic materials increased in Unit II and the 1st REY peak, which originated from the Izu-Bonin-Mariana arc in the western North Pacific deposited since the latest Eocene explosive volcanism. In contrast, volcanic materials from a different source might have been introduced at Unit IV, although the depositional age and origin of the volcanic material are yet to be determined. We further recognized the differences in terrigenous components among Units I, III, and V. The provenance of the eolian dust could have shifted from North America in Unit V to eastern Asia in Unit I through the depositional history around the Minamitorishima area, although that in Unit III is unknown.