Integration of new zircon U–Pb ages with biostratigraphy to establish a high-precision age model of the Miocene Nakayama Formation on Sado Island in Central Japan

Abstract

The most common age constraint for the diatomaceous sediments is biostratigraphy of siliceous microfossils. Although biostratigraphy is a powerful tool to establish stratigraphy and correlate with sedimentary sequences in other sites, biostratigraphy generally includes uncertainties difficult to evaluate. In this study, we measured zircon U–Pb ages of eight tuff beds intercalated with diatomaceous mudstone of the Nakayama Formation on Sado Island in Central Japan and integrated the U–Pb ages with diatom and radiolarian biostratigraphy, whose ages and errors were re-evaluated by this study, to establish an age model precisely representing the sedimentary age. Two tuff beds in the upper and middle part of the formation offered zircon U–Pb ages of 6.7 ± 0.2 Ma and 10.87 ± 0.07 Ma, which are consistent with biostratigraphy, and provided a good example of effective integration of zircon U–Pb ages with the biostratigraphy. On the other hand, zircon U–Pb ages of the other six tuff beds in the lower part are around 12 Ma and not distinguishable from each other. In addition, older zircon grains in the 6 tuff beds are assembled in the interval from 30 to 20 Ma, which is consistent with the age of the volcanic basement rocks forming most part of Sado Island. Similarities in chemical compositions of glass shards and age distributions of zircon grains indicate that the volcaniclastic components in the tuff beds should originate from single or associated magmatic activities.

1 Introduction

Diatoms, siliceous marine algae, contribute ca. 40% of marine primary production and particulate organic carbon export as part of the biological pump (Tréguer et al. 2018). Diatoms and organic matter transported to the seafloor form into diatomaceous sediments. Therefore, diatomaceous sediments have attracted scientific attention in their role in organic carbon burial, which is one of the important factors controlling the atmospheric CO₂ concentration (e.g., Ragueneau et al. 2000; Renaudie 2016). In fact, the prosperity of diatomaceous sediments in the North Pacific marginal basins including the Japan Sea since the middle Miocene has been associated with global climate change (e.g., Vincent and Berger 1985; Flower and Kennett 1994). However, the vulnerability of the old diatomaceous sediments to silica diagenesis and the scarcity of carbonate microfossils are preventing the paleoceanographic and paleoclimatologic use of diatomaceous sediments.

One of the greatest concerns in the use of diatomaceous sediments for paleoceanographic studies is the precision of their sedimentary age. The scarce content of carbonate makes it difficult to use chemostratigraphy such as the oxygen stable isotope. The most common age constraint for the diatomaceous sediments is diatom and radiolarian biostratigraphy. Biostratigraphy is a powerful tool to establish chronostratigraphy and correlate a sedimentary sequence with those in other sites. International Ocean Discovery Program/Integrated Ocean Drilling Program (IODP), Ocean Drilling Program (ODP), and Deep Sea Drilling Project (DSDP) have significantly improved the Neogene diatom and radiolarian biostratigraphy in the North Pacific region (e.g., Barron et al. 1995; Yanagisawa and Akiba 1998; Kamikuri et al. 2017). However, biostratigraphy generally includes uncertainties difficult to evaluate, for example, geographic or regional differences in biostratigraphic events and uncertainties deriving from age models of the reference sequences, like deep-sea drilling cores, which are used to construct the biostratigraphy.

In case there are tuff beds intercalated with diatomaceous sediments and the tuff beds contain zircon grains, zircon U–Pb dating is available. The recent improvement of in situ U–Pb age measurement and the establishment of new standard zircon preferable to the measurement of young zircon have enabled precise U–Pb dating for Miocene samples (Iwano et al. 2013; Hattori et al. 2017; Obayashi et al. 2017). Whereas a zircon U–Pb age is an excellent indicator of the crystallization age of the zircon, the zircon U–Pb age does not always represent the age of the deposition of the tuff bed, for example, early crystallization of zircon a long time before the eruption and re-sedimentation of old pyroclastic materials containing zircon (e.g., Danhara et al. 2005).

The Miocene Nakayama Formation distributed on Sado Island has a great potential to be used for paleoceanographic studies during the middle to late Miocene. It has orbital-scale oscillations in sedimentary structures such as laminations and bioturbations representing the bottom water oxygenation level (Sakamoto 1992). Well-preserved siliceous microfossils and good exposures of outcrops of the Nakayama Formation have contributed to the establishment of high-resolution diatom biostratigraphy (Yanagisawa and Watanabe 2017a). The Nakayama Formation also has a lot of felsic tuff beds intercalated with diatomaceous sediments (Sakamoto 1992; Yanagisawa et al. 2009; Yanagisawa and Watanabe 2017a), which enables the integration of biostratigraphy and radiometric dating and establishment of more plausible age constraints.

In this study, we measured zircon U–Pb ages of eight tuff beds intercalated with diatomaceous mudstone of the Nakayama Formation and compared their U–Pb ages with diatom biostratigraphy already reported by Yanagisawa and Watanabe (2017a) and radiolarian biostratigraphy newly established by this study in order to confirm that these age constraints indeed represent the sedimentary age. The ages and error ranges of diatom and radiolarian biostratigraphy were re-evaluated by constructing new age models of reference sedimentary sequences for biostratigraphy. In addition to age determination, we described the detailed stratigraphy of the lowest part of the Nakayama Formation (SW Section), where a lot of thick tuff beds have been identified.

1.1 Geological setting

Sado Island (854.8 km²) is located in the Japan Sea northwest of Niigata Prefecture in Central Japan. Sado Island has two highland areas, Osado Mountains and Kosado Hills, with the Kuninaka Plain area between them (Fig. 1). Most parts of the highland areas are composed of volcanic rocks, which are mainly formed during the early Miocene volcanism (e.g., Metal Mining Agency of Japan 1994). Sedimentary rocks on Sado Island started to be formed during the middle Miocene, preceding the marine transgression widely reported from the Japanese Islands (Kobayashi 2000; Yanagisawa and Watanabe 2017b), and are mainly exposed along the southeastern foot of Osado Mountains and in the southern part of Kosado Hills (Fig. 1; Shimazu et al. 1977). The Neogene and Quaternary sedimentary rocks are classified as, from bottom to top, the Orito, Hanyugawa, Nakayama, Nozaka, Kawachi, Kaidate, and Shichiba Formations (Fig. 2; the Compilation Committee of the Geological Map of Niigata Prefecture 2000; Yanagisawa and Watanabe 2017b).

Fig. 1

Location and a simplified geological map of Sado Island. The target formation in this study, the Nakayama Formation, is included within the dark brown shade category, Miocene-Pliocene sedimentary rocks. A black rectangle shows the studied area, whose detailed map is described in Fig. 3. Geological distribution is based on Sumi (1989) and the Compilation Committee of the Geological Map of Niigata Prefecture (2000), and stratigraphy is based on Yanagisawa and Watanabe (2017b)

Fig. 2

Cenozoic stratigraphy and simplified lithologies of Sado Island in Niigata Prefecture, Central Japan. Stratigraphy and lithologies are based on Yanagisawa and Watanabe (2017b). Color bars (dark gray, light gray, dark brown, and light brown) represent the classification categories used in Fig. 1

The first phase of the sedimentary rocks is called the Orito Formation. The Orito Formation is composed of conglomerate, sandstone, and mudstone, which are considered fluvial to inner bay deposits (Yanagisawa and Watanabe 2017b). It is thought that the depositional age of the Orito Formation is within the Neogene North Pacific diatom zone (NPD) 3A₁ (ca. 16.9–16.6 Ma) due to the occurrence of Arcid-Potamid Fauna from the tropical Kadonosawa Fauna (Yanagisawa and Watanabe 2017a).

The Hanyugawa Formation, which conformably covers the Orito Formation, is composed of conglomerate, calcareous sandstone, glauconite sandstone, and gray mudstone (Yanagisawa and Watanabe 2017b). A basal conglomerate of the Hanyugawa Formation is recognized because of the marine transgression (Yanagisawa and Watanabe 2017b). The age of the top of the Hanyugawa Formation was estimated as ca. 12.3 Ma based on extrapolation of a sedimentation rate curve, which was obtained from diatom biostratigraphy established in the upper Nakayama Formation (Yanagisawa and Watanabe 2017a). While the thickness of the Hanyugawa Formation is no more than 15 m, the existence of glauconite beds supports the long duration of the Hanyugawa Formation equivalent to ca. 4 Myr (Yanagisawa and Watanabe 2017b).

The Hanyugawa Formation is conformably covered by the Nakayama Formation, which is composed of diatomaceous mudstone and a few glauconite beds. Diatom biostratigraphy shows that the Nakayama Formation, whose thickness is 230–320 m, corresponds to NPD5B to NPD7A (ca. 12.3–6.5 Ma; Yanagisawa and Watanabe 2017a). In the lowest part of the Nakayama Formation, diatomaceous mudstone has often altered into hard siliceous mudstone due to silica diagenesis, and the hard mudstone used to be called the Tsurushi Formation (Sakamoto 1992). However, the use of “the Tsurushi Formation” as a formation name is not recommended since the horizon of the silica diagenesis boundary is not consistent among sites (Yanagisawa and Watanabe 2017b). Then, we avoid using the name of the Tsurushi Formation in this study.

The Nakayama Formation is conformably overlain by the Nozaka Formation, which is composed of massive slightly diatomaceous mudstone formed during the late Miocene to Pliocene (Yanagisawa and Watanabe 2017b). The boundary between the Nakayama and Nozaka Formations is a tuff bed named WKA-5, which is one of our samples for zircon U–Pb dating. Glauconite is also developed around the boundary.

The Kawachi Formation, which is composed of the Pliocene to Pleistocene sandy mudstone with a lot of seashell fossils, unconformably covers the Nozaka Formation (Yanagisawa and Watanabe 2017b). The Kawachi Formation is unconformably covered by the Kaidate Formation, which is composed of Pleistocene sand and gravel with seashell fossil layers (Yanagisawa and Watanabe 2017b). The Kaidate Formation is conformably overlain by the Shichiba Formation, which is composed of alternation of gravel, sand, and sandy silt with a lot of seashell fossils and foraminifera (Yanagisawa and Watanabe 2017b).

The studied area is in the southern part of the Osado Mountains foot area (Fig. 1). In this area, the Orito, Hanyugawa, Nakayama, and Nozaka Formations are exposed with east dips and N–S to NE–SW strike directions without major unconformities, and the Kawachi Formation unconformably covers the Nozaka Formation in the northern part and the Nakayama Formation in the southern part (Fig. 3; Sakamoto 1992; Yanagisawa and Watanabe 2017a). There are inferred faults in the middle part, and the degree of silica diagenesis is lower in the south of the faults than in the north (Yanagisawa and Watanabe 2017a). Diatom biostratigraphy was established in most parts of the Nakayama and Nozaka Formations (Yanagisawa and Watanabe 2017a). On the other hand, the lowest part of the Nakayama Formation, exposed in SW (Hanyugawa) Section, is poorly constrained, but it has been assigned within the diatom zone NPD5B_3–4 (ca. 12.3–11.5 Ma; Yanagisawa and Watanabe 2017a).

Fig. 3

A detailed geological map of the studied area. The geological distribution and diatom biohorizons in each section are based on Yanagisawa and Watanabe (2017b). The Nakayama Formation is divided into two colors (sky blue and purple) depending on the degree of the silica diagenesis: The sky-blue area represents diatomaceous mudstone in the opal-A zone, and purple area represents hard mudstone in the opal-CT zone. The measured tuff samples were taken from WK Section (WKA-5), TB-II Section (TBA-3), and SW Section (SWT-1aU, SWT-2U, SWT-3U, SWT-3L, SWT-5U, and SWT-5L). Brown zigzag lines represent sampling locations of WKA-5 and TBA-3. Sampling locations in SW Section are shown in Supplementary Fig. 1

2 Methods/experimental

2.1 Sampling strategy

In the studied area, many felsic tuff beds and several mafic tuff beds are intercalated with the muddy sediment of the Nakayama and Nozaka Formations (Fig. 4). Since most of the tuff beds are thinner than 5 cm and finer than medium sand, we took samples for zircon U–Pb dating from coarse-grained felsic tuff beds selectively. Each sample was equally collected from the top to the bottom of each tuff bed.

Fig. 4

Columnar sections in the studied sections and close-up columnar sections around WKA-5 and TBA-3. The sampling locations of WKA-5 and TBA-3 are shown in Fig. 3. A composite columnar section in the north of the assumed fault (WK and SN Sections composite) is based on Yanagisawa and Watanabe (2017a). The sedimentary sequence covers six Neogene North Pacific diatom zones from NPD5C to NPD7Ba and contains 15 diatom biohorizons from D55 to D77 (Yanagisawa and Watanabe 2017a). Abbreviation: Fm. Formation

The two tuff samples (WKA-5 and TBA-3) were taken from the sections with better-constrained diatom biostratigraphy, WK (Wakamiya) and TB-II (Shichibagawa-II) Sections, corresponding to the upper and middle part of the Nakayama Formation (Figs. 3 and 4). WKA-5, which is exposed in WK Section, is a 30-cm-thick coarse felsic glassy tuff with laminated intervals (Fig. 4). WKA-5 is intercalated with glauconite-bearing mudstone, and the top of WKA-5 is defined as the boundary between the Nakayama and Nozaka Formations (Yanagisawa and Watanabe 2017b). TBA-3, which is exposed in TB-II Section, is a 10-cm-thick coarse felsic glassy tuff with pumice fragments (Fig. 4). While its lower boundary with diatomaceous mudstone is sharp, the upper boundary is weakly bioturbated.

From SW Section, six tuff samples (SWT-1aU, SWT-2U, SWT-3U, SWT-3L, SWT-5U, and SWT-5L) were taken for zircon U–Pb dating (Fig. 5; Supplementary Fig. 1). They are coarse felsic glassy tuff, whose thickness ranges from 15 to 60 cm. Detailed descriptions are given in Sect. 3.1. Additionally, 15 diatomaceous mudstone samples were taken from SW Section for radiolarian analysis to further constrain the age model (Fig. 5). Almost all sampling intervals are shorter than 4.2 m (with one exception of 11.9 m).

Fig. 5

A composite columnar section with age-diagnostic radiolarian occurrences in SW Section. Three close-up columnar sections around five sets of thick tuff beds (SWT-1a, SWT-1b, SWT-2, SWT-3, and SWT-5) are also displayed. Tuff names with underlines express the measured tuff samples in this study. Alphabets A to H beside the composite columnar section represent outcrops described in the route map of SW Section (Supplementary Fig. 1). Only one diatom biohorizon (D55) was recognized in SW Section (Yanagisawa and Watanabe 2017a). Abbreviations: FO first occurrence, LO last occurrence, Fm. formation

2.2 Major element analysis of volcanic glass shards

Volcanic glass shard samples for major element analysis were separated as a light-mineral fraction by a panning process. After sieved with a 63-µm mesh, residue grains were mounted in epoxy resin discs and polished until their surfaces were flattened. We measured 10 major element compositions (SiO₂, TiO₂, Al₂O₃, FeO, MnO, MgO, CaO, Na₂O, K₂O, and P₂O₅) of volcanic glass shards by using a JEOL JXA-8900R electron probe microanalyzer (EPMA) equipped with wavelength-dispersive X-ray spectrometers (WDS) at the Atmosphere and Ocean Research Institute, the University of Tokyo (Kashiwa, Japan). EPMA operating conditions were 15 kV acceleration voltage, 7 nA beam current, beam diameter of 10 µm, and counting time of each element was 10–60 s at the peak position and 10–30 s at the background (Matsumoto et al. 2015; Nakanishi et al. 2020).

2.3 Zircon U–Pb measurement

Heavy minerals including zircon grains were separated from tuff samples by elutriation, panning, magnetic separation, and heavy liquid techniques at the National Museum of Nature and Sciences (Tsukuba, Japan). Zircon grains were mounted in epoxy resin discs by handpicking. The zircon samples were polished until the midsection of the grains was flatly exposed.

Zircon samples were analyzed by a Nu Plasma II (Nu Instruments, UK), multi-collector inductively coupled plasma-mass spectrometry (MC-ICP-MS), connected to a Jupiter solid nebulizer system (ST Japan, Japan) at the Geochemical Research Center, the University of Tokyo (Tokyo, Japan). The laser with a wavelength of 257 nm from CARBIDE (Light Convection, Lithuania) ablated a square-shaped area with 20-µm sides. The mass spectrometer simultaneously detected ²⁰²Hg, ²⁰⁴Pb (+²⁰⁴Hg), ²⁰⁶Pb, ²⁰⁷Pb, ²⁰⁸Pb, ²³⁵U, and ²³²Th (Obayashi et al. 2017; Hattori et al. 2017). The ²³⁸U signal intensity was calculated based on the ²³⁵U signal intensity by normalizing the ²³⁸U/²³⁵U ratio at 137.818 (Hiess et al. 2012). We used NIST SRM 612 and GJ-1 zircon (Jackson et al. 2004) as primary standards. Three spots of both NIST SRM 612 and GJ-1 were analyzed to calibrate the fractionation of Pb isotopic ratios and ²³⁸U/²⁰⁶Pb ratios, respectively, before and after 10–20 spot analyses of unknown samples. We also analyzed OD-3 zircon (Iwano et al. 2013) as a secondary standard to evaluate the reliability of the measurements. OD-3 zircon shows a weighted mean ²⁰⁶Pb/²³⁸U age of 32.90 ± 0.12 Ma (n = 29), which agrees with previously reported ages in Iwano et al. (2013) within the 2 σ error ranges. We used Isoplot 3.75 program (Ludwig 2012) to calculate weighted means of ²⁰⁶Pb/²³⁸U ages and ²⁰⁷Pb-corrected ages and to draw Tera-Wasserburg concordia diagrams.

2.4 Construction of radiolarian biostratigraphy

Samples for radiolarian analysis were prepared following the procedures described by Kamikuri et al. (2017). The dried sediment samples were placed in beakers with 15% hydrogen peroxide to remove organic materials for several hours. A 5% solution of hydrochloric acid is generally added to remove the calcareous fraction from the sediments; however, because of the very poor content in calcareous matters of the used sediments, we did not use hydrochloric acid this time. Subsequently, after the samples were washed and sieved through a 45-μm mesh, they were dried again in an oven at 40℃. Then, each sample was scattered randomly on a glass slide. Next, Norland Optical Adhesive #60 (refractive index = 1.56) was used as a mounting medium for a 24 × 36-mm cover glass. Then, all species on a slide were checked to identify key biostratigraphic species useful to constrain sediment ages. Approximately at least 500 specimens were checked for each sample.

2.5 Estimation of age errors of diatom and radiolarian biostratigraphy

There are a lot of error factors difficult to estimate, for example, errors of the age model of the reference sedimentary sequence and inter-regional effects on the timing of the species extinctions and occurrences. The inter-regional effects could be minimized by adopting the biostratigraphy constructed in the neighboring basins. In the case of diatom and radiolarian biostratigraphy of the Japan Sea sediments, the seafloor sediments taken by IODP, ODP, and DSDP in the Japan Sea and the North Pacific are usually used as reference sedimentary sequences.

In this study, we calculated errors of some diatom and radiolarian biohorizons by estimating 2 σ error ranges of the age models of the reference sedimentary sequences by using an open-source age-depth modeling routine, the Undatable software in MATLAB produced by Lougheed and Obrochta (2019). Based on a sediment deposition simulation, the Undatable can estimate the uncertainty resulting from inconstant sedimentation rates. New age models containing the 2 σ error ranges were established from five sedimentary sequences (Supplementary Fig. 2), ODP Sites 887C and 884B (the North Pacific), DSDP Site 438B and ODP Site 1151A (the Northwest Pacific), and IODP Site U1430 (the Japan Sea), in order to obtain the age errors of diatom biohorizons reported in the Nakayama Formation by Yanagisawa and Watanabe (2017a) and radiolarian biohorizons newly identified in SW Section in this study (Table 1). The input age constraints for the Undatable calculations are listed in Supplementary Table 1. The age models of Sites 887C and 884B were constructed based on magnetostratigraphy (Barron et al. 1995). The age model of Site 438 was based on selective diatom biohorizons (Yanagisawa and Akiba 1998). The age model of Site 1151A was based on magnetostratigraphy and diatom biohorizons (Motoyama et al. 2004). The age model of Site U1430 was based on cyclostratigraphy (Kurokawa et al. 2019), which was established on the perfectly continuous spliced sequence (Irino et al. 2018). The geomagnetic polarity timescale used in this study was revised after Geologic Time Scale 2020 (Ogg 2020). The ages and 2 σ error ranges of diatom biohorizons used for Sites 438 and 1151A were derived from the diatom biostratigraphy in Sites 887C and 884B (Barron and Gladenkov 1995; Watanabe and Yanagisawa 2005) and revised after the new age models. The input 2 σ errors of the age constraints were set at 5 kyr for magnetostratigraphy, individual widths of the calculated 2 σ error ranges for diatom biostratigraphy, and 100 kyr for cyclostratigraphy (there are listed in Supplementary Table 1). The calculation on the Undatable was executed under the following conditions: nsim = 10⁵, bootpc = 0, xfactor = 0.1, and combine = No (details of the parameters are given in Lougheed and Obrochta 2019).

Table 1 Newly calculated ages of diatom and radiolarian biohorizons

3 Results

3.1 Stratigraphy of SW section

Outcrops of SW Section are dispersedly located along the Hanyugawa River (Supplementary Fig. 1). In this study, the Nakayama Formation represented by diatomaceous mudstone with tuff and hard siliceous mudstone (Outcrops A to G), and the Orito Formation represented by sandstone and conglomerate (Outcrop H) were recognized, but the lack of continuous exposure prevented us from finding the outcrop containing the Hanyugawa Formation represented by thin glauconite sandstone, which was described in Yanagisawa and Watanabe (2017a). The thickness of the sedimentary sequence of the Nakayama Formation we described in SW Section measures up to ca. 48 m. The upper (Outcrops A–C) and middle parts (Outcrops D–F) of the sequence are composed of laminated and massive diatomaceous mudstone, several thick coarse tuff beds, and thin hard mudstone layers, while the lower (Outcrop G) is composed of massive diatomaceous mudstone, thick coarse tuff beds, and hard mudstone (Fig. 5).

Following the Neogene radiolarian biostratigraphic scheme proposed in the Japan Sea by Kamikuri et al. (2017), we identified four key radiolarian biohorizons from the sedimentary sequence of SW Section: (1) the First Occurrence (FO) of Hexaconthium akitaensis in Outcrop A, (2) The Last Occurrences (LO) of Eucyrtidium inflatum between Outcrops D and F, (3) the LO of Calocyclas motoyamai between Outcrops D and F, (4) the LO of Lithopera reanzae in Outcrop G (Fig. 5). The mean ages of the FO of H. akitaensis and the LOs of E. inflatum, C. motoyamai, and L. renzae were estimated as 11.35 Ma, 11.75 Ma, 11.75 Ma, and 12.07 Ma, respectively, in this study (Table 1). This radiolarian biostratigraphy in SW Section is consistent with the diatom biohorizon D55 (11.55–11.33 Ma) previously reported by Yanagisawa and Watanabe (2017a).

SW Section has several thick and coarse felsic tuff beds. These tuff beds are useful for correlations among outcrops because of their unique sedimentary structures and their appearances as a set with another nearby tuff bed (Fig. 5). In the following paragraphs, we described below the detailed sedimentary structures of five sets of thick felsic tuff beds in SW Section: SWT-1a, SWT-1b, SWT-2, SWT-3, and SWT-5.

SWT-1a consists of the upper 15-cm-thick tuff bed (SWT-1aU) and the lower 10-cm-thick tuff bed (SWT-1aL) with intercalation of 10-cm-thick massive diatomaceous mudstone between the tuff beds. The top boundary of SWT-1aU is gradual, and the bottom boundary is sharp. SWT-1aU shows laminations in the middle part. The top boundary of SWT-1aL is gradual, and the bottom boundary is erosional. The grain size of both tuff beds is coarse to medium. SWT-1a is exposed in Outcrops B and C.

SWT-1b consists of the upper 35-cm-thick coarse tuff bed (SWT-1bU) and the lower 40-cm-thick coarse-to-fine tuff bed (SWT-1bL) with intercalation of 10-cm-thick massive diatomaceous mudstone between the tuff beds. The top boundary of SWT-1bU is gradual, and the bottom boundary is erosional. The uppermost part of SWT-1bU looks like gomashio (dark-colored grains in white matrix grains). The top boundary of SWT-1bL is gradual, and the bottom boundary is sharp. SWT-1bL shows laminations in the upper and lower parts and graded structures in the middle part. SWT-1b is partially exposed in Outcrop B and fully exposed in Outcrop C.

SWT-2 consists of the upper 35-cm-thick coarse tuff bed (SWT-2U) and the lower 5-cm-thick coarse tuff bed (SWT-2L) with intercalation of 15-cm-thick massive diatomaceous mudstone between the tuff beds. The top boundary of SWT-2U is bioturbated, and the bottom boundary is sharp. SWT-2U shows laminations and pumice fragments. The top boundary of SWT-2L is bioturbated, and the bottom boundary is erosional. SWT-2L also shows laminations. SWT-2 is exposed in Outcrops D, E, and F.

SWT-3 consists of the upper 60-cm-thick coarse-to-medium tuff bed (SWT-3U) and the lower 40-cm-thick coarse-to-fine tuff bed (SWT-3L) with intercalation of 10-cm-thick glauconite-bearing massive diatomaceous mudstone between the tuff beds. The top boundary of SWT-3U is bioturbated, and the bottom boundary is erosional. SWT-3U shows laminations and pumice fragments in the upper part and graded bedding in the lower. The top boundary of SWT-3L is gradual, and the bottom boundary is erosional. SWT-3L shows laminations, cross-laminations, frame structure, and pumice fragments. SWT-3 is exposed in Outcrop F.

SWT-5 consists of the upper 60-cm-thick coarse tuff bed (SWT-5U) and the lower 40-cm-thick coarse tuff bed (SWT-5L) with intercalation of 35-cm-thick glauconite-bearing massive diatomaceous mudstone between the tuff beds. The top boundary of SWT-5U is gradual, and the bottom boundary is erosional. SWT-5U is almost massive except for a thin silt band. The top boundary of SWT-5L is gradual, and the bottom boundary is erosional. SWT-5L shows pseudoconglomerate in the upper part, cross-laminations in the middle part, and laminations in the lower part. SWT-5 is exposed in Outcrop G.

3.2 Petrology of volcanic glass shards

The results of the major element analysis of volcanic glass shards are listed in Table 2. The total of 10 major element concentrations ranges from 88.93 to 92.95 wt%. They are relatively lower than those of the Quaternary tephra beds measured by the same machine (Nakanishi et al. 2020). It possibly resulted from the hydration of volcanic glass shards due to the longer duration after its sedimentation.

Table 2 Major element composition and shape of the volcanic glass shard samples

After the normalization of the major element concentrations as the total got to 100 wt%, the measured volcanic glass shard samples similarly show high SiO₂ concentrations ranging from 77.09 to 78.60 wt% and are plotted in the rhyolite zone of the total alkali-silica diagram by Le Bas et al. (1986) (Fig. 6a). On the other hand, the K₂O–TiO₂ diagram provides some variances, 3.45–4.93 wt% of K₂O concentrations and 0.093–0.205 wt% of TiO₂ concentrations (Fig. 6b). WKA-5 displays the lowest K₂O concentration and the highest TiO₂ concentration among the measured samples. In the K₂O–TiO₂ diagram, the samples from SW Section were clearly divided into two groups: U-group (the higher K₂O: 4.66–4.93 wt% and the lower TiO₂: 0.036–0.059 wt%) and L-group (the moderate K₂O: 4.27–4.30 wt% and the moderate TiO₂: 0.087–0.093 wt%). Interestingly, the U-group consists of the upper tuff beds of SWT-1a, SWT-2, SWT-3, and SWT-5 (SWT-1aU, SWT-2U, SWT-3U, and SWT-5U), and the L-group consists of the lower ones of SWT-3 and SWT-5 (SWT-3L and SWT-5L).

Fig. 6

a The total alkali-silica diagram (Le Bas et al. 1986) and b the K₂O-TiO₂ diagram. a All of the volcanic glass shard samples are plotted in the rhyolite zone. Error ellipses are 1 σ. b The volcanic glass shard samples from SW Section were clearly divided into two groups: U-group and L-group in the K₂O-TiO₂ diagram. Error ellipses are 1 σ

In this study, the classification of the shape of volcanic glass shards is mainly based on Yoshikawa (1976): the bubble-wall type (Bw), the pumiceous type (Pm), the intermediate type (Im), and the irregularly shaped type (Irr). Bw-type glass shards show planar shapes like fragments of large bubbles. Pm-type shards have a lot of small bubbles or elongated ones in shards and are usually porous. Im-type shards show intermediate characteristics between Bw- and Pm-type shards. Irr-type shards show different shapes from other types, for example, glass shards by quench fragmentation. The observational characteristic of each tuff sample is described below.

WKA-5 contains abundant high-temperature quartz and glauconite other than volcanic glass shards. The volcanic glass of WKA-5 is composed of abundant Bw-type glass shards with minor Pm-type ones. TBA-3 shows an unusual combination of heavy minerals: zircon, monazite, and barite. The volcanic glass of TBA-3 is composed of abundant Im-type glass shards with minor Pm- and Bw-type ones. SWT-1aU contains a small portion of lithic fragments and sponge spicules. The volcanic glass of SWT-1aU is composed of abundant Pm-type glass shards with minor Im- and Bw-type ones. SWT-2U contains a small amount of glauconite and sponge spicules. The volcanic glass of SWT-2U is composed of abundant Im-type glass shards with minor Pm- and Bw-type ones. SWT-3U contains a small amount of glauconite. The volcanic glass of SWT-3U is composed of abundant Im-type glass shards with minor Pm- and Bw-type ones. SWT-3L contains a small amount of glauconite and sponge spicules. The volcanic glass of SWT-3L is composed of abundant Im-type glass shards with minor Pm- and Bw-type ones. SWT-5U contains a small amount of glauconite and a lot of weathered lithic fragments and sponge spicules. The volcanic glass of SWT-5U is composed of abundant Im-type glass shards with minor Pm- and Bw-type ones. SWT-5L contains weathered lithic fragments and sponge spicules with a small amount of glauconite. The volcanic glass of SWT-5L is composed of abundant Im-type glass shards with minor Pm- and Bw-type ones.

3.3 Zircon U–Pb ages

Obtained analytical data of the zircon grains are listed in Table 3. We measured 18–39 euhedral zircon grains for each sample. The non-radiogenic Pb ratios and the ²⁰⁷Pb-corrected ages were calculated for all grains by estimating the ²⁰⁷Pb/²⁰⁶Pb ratios of non-radiogenic Pb based on the two-stage model of Pb isotopic evolution constructed by Stacey and Kramers (1975). Zircon grains with higher non-radiogenic Pb than 5% were excluded from the following U–Pb age determination protocol. To obtain information about the sedimentary age, we distinguished the youngest group of zircon grains from older grains. The youngest group was defined as grains whose ²⁰⁷Pb-corrected ages overlap with any of the ²⁰⁷Pb-corrected ages overlapping with that of the youngest grain within the 2 σ error range. We reported a weighted mean age of the ²⁰⁷Pb-corrected ages in the youngest group as a zircon U–Pb age of each tuff bed (Fig. 7). The weighted mean ²⁰⁷Pb-corrected age of each tuff bed is consistent with its concordia age (except for SWT-2U, which has only 2 concordant grains and no concordia age).

Table 3 U–Pb measurement data of zircon grains from the sampled tuff beds in the Nakayama Formation

Fig. 7

Weighted mean age plots and Tera-Wasserburg concordia diagrams of the youngest zircon group. Results of a WKA-5, b TBA-3, c SWT-1aU, d SWT-2U, e SWT-3U, f SWT-3L, g SWT-5U, and h SWT-5L are shown. In the weighted mean age plots (left sides), green and purple bars represent accepted and rejected ²⁰⁷Pb-corrected ages of individual grains with 2 σ errors, respectively, and black bars represent weighted mean ages. In the Tera-Wasserburg concordia diagrams (right sides), black solid and dash ellipses represent concordant and discordant data with 2 σ errors, respectively, and red ellipses represent weighted means with 2 σ errors

The zircon U–Pb ages of the upper two tuff beds revealed that they were formed by the late Miocene volcanic eruptions. The youngest group of WKA-5 yielded a weighted mean ²⁰⁷Pb-corrected age of 6.68 ± 0.18 Ma (2 σ, MSWD = 3.3, n = 6) (Fig. 7a). The older grains of WKA-5 are divided into two groups, 7.49–7.12 Ma (n = 3) and 12.03–11.36 Ma (n = 10). The youngest group of TBA-3 yielded a weighted mean ²⁰⁷Pb-corrected age of 10.874 ± 0.065 Ma (2 σ, MSWD = 1.09, n = 10) (Fig. 7b). The older grains of TBA-3 range from 69.10 to 57.00 Ma (n = 5).

As for the zircon U–Pb ages of the youngest group, all of the measured tuff beds in SW Section show the middle Miocene ages, which are undistinguishable within the 2 σ error ranges. In addition, most of the older grains display similar ²⁰⁷Pb-corrected ages, typically ranging from 28 to 21 Ma. The youngest group of SWT-1aU yielded a weighted mean ²⁰⁷Pb-corrected age of 11.972 ± 0.080 Ma (2 σ, MSWD = 0.31, n = 9) (Fig. 7c). The ²⁰⁷Pb-corrected ages of older grains range from 25.21 to 21.84 Ma (n = 10) except for one grain representing 12.568 Ma. The youngest group of SWT-2U yielded a weighted mean ²⁰⁷Pb-corrected age of 11.68 ± 0.51 Ma (2 σ, MSWD = 2.2, n = 3) (Fig. 7d). The ²⁰⁷Pb-corrected ages of older grains range from 27.87 to 21.46 Ma (n = 31) except for one grain representing 176.1 Ma. The youngest group of SWT-3U yielded a weighted mean ²⁰⁷Pb-corrected age of 11.957 ± 0.065 Ma (2 σ, MSWD = 1.07, n = 11) (Fig. 7e). The ²⁰⁷Pb-corrected ages of older grains range from 28.06 to 21.29 Ma (n = 25) except for one grain representing 12.27 Ma. The youngest group of SWT-3L yielded a weighted mean ²⁰⁷Pb-corrected age of 11.943 ± 0.073 Ma (2 σ, MSWD = 0.73, n = 17) (Fig. 7f). The ²⁰⁷Pb-corrected ages of older grains range from 27.52 to 22.76 Ma (n = 9) except for one grain representing 97.64 Ma. The youngest group of SWT-5U yielded a weighted mean ²⁰⁷Pb-corrected age of 11.925 ± 0.088 Ma (2 σ, MSWD = 0.68, n = 7) (Fig. 7g). The ²⁰⁷Pb-corrected ages of older grains range from 27.79 to 21.77 Ma (n = 32). The youngest group of SWT-5L yielded a weighted mean ²⁰⁷Pb-corrected age of 11.927 ± 0.082 Ma (2 σ, MSWD = 1.3, n = 13) (Fig. 7h). The ²⁰⁷Pb-corrected ages of older grains are divided into three groups, 24.86–22.12 Ma (n = 3), around 63 Ma (n = 2), and 1986 Ma (n = 1).

4 Discussion

4.1 Comparison of U–Pb ages with biostratigraphy

4.1.1 The upper and middle part (WKA-5 and TBA-3)

One of the outstanding results of this study is enabling us to treat biostratigraphic age constraints as quantitative ages. Based on our newly calculated ages and errors of diatom biostratigraphy (Table 1) originally reported by Yanagisawa and Watanabe (2017a), zircon U–Pb ages of WKA-5 and TBA-3 are concordant with diatom biostratigraphy within the 2 σ errors (Fig. 8a). Especially, the zircon U–Pb age of TBA-3 completely falls within the range of the diatom biostratigraphic age model.

Fig. 8

Summary of age constraints by zircon U–Pb ages and biostratigraphy in the Nakayama Formation. a The upper and lower part of the Nakayama Formation contains two zircon U–Pb ages (black bars; this study) with diatom biostratigraphy (red rectangles; Yanagisawa and Watanabe 2017a), while b the lower part (SW Section) contains six zircon U–Pb ages (black bars; this study) with diatom (red rectangles; Yanagisawa and Watanabe 2017a) and radiolarian (green rectangles; this study) biostratigraphy. Newly calculated diatom and radiolarian ages are used. All age constraints are described with 2 σ errors

TBA-3 occupies an intermediate horizon between the diatom biohorizons D55.8 (10.35–10.07 Ma) and D55.2 (11.50–10.78 Ma). Therefore, the zircon U–Pb age of TBA-3 (10.87 ± 0.07 Ma) is a significant age constraint compensating the ca. 1.05-Myr interval with no age restriction by diatom biostratigraphy. As a result, the combination of the dual age constraints by our high-precision zircon U–Pb ages and newly evaluated diatom biostratigraphic ages was able to provide a more accurate age model with a smaller error range.

WKA-5 is recognized as the boundary between the Nakayama and Nozaka Formations near the diatom biohorizon D75 (6.58–6.36 Ma). The proximity of WKA-5 to the diatom biohorizon has an advantage for comparison of their ages. Our results revealed that the zircon U–Pb age of WKA-5 (6.7 ± 0.2 Ma) seems slightly older (but not significantly) than the age of the diatom biohorizon D75. The zircon U–Pb age generally represents the age of crystallization of zircon in magma, but not strictly the sedimentation age of volcanic tuff. Thus, some zircon U–Pb ages could be older than the sedimentation ages depending on the magmatic evolution and the eruption–deposition system.

Another explanation comes from the occurrence of glauconite above and beneath WKA-5 (Fig. 4). Glauconite is an authigenic clay mineral, and one of the key factors for its formation is long-time residence near the water–sediment interface (e.g., Fernández-Landero and Fernández-Caliani 2021). Actually, diatom biostratigraphy proved that some Miocene glauconite-rich beds observed in the Japan Sea coastal area of Japan, such as the Noto Peninsula in Central Japan and Sasamori Hills in northern Japan, represented a hiatus or an extremely low sedimentation rate (Kato and Yanagisawa 2021; Watanabe 1990; Yanagisawa 1999). The slow sedimentation inferred from glauconite around WKA-5 could explain the minor misalignment between the diatom and zircon ages.

4.1.2 The lower part (SW section)

While only one diatom biohorizon (D55) was recognized in SW Section (Yanagisawa and Watanabe 2017a), we established radiolarian biostratigraphy and identified four key radiolarian biohorizons from three stratigraphic horizons (Fig. 5). Age constraints by diatom and radiolarian biostratigraphy are stratigraphically consistent, whereas zircon U–Pb ages are not completely consistent (Fig. 8b).

Except for SWT-2U, which has only three youngest-group zircon grains and a long 2 σ error bar, highly accurate U–Pb ages around 12 Ma were obtained from the tuff beds in SW Section, and not distinguishable from each other within the 2 σ errors although the maximum stratigraphic distance is over 20 m (equivalent to ca. 0.2–0.4 myr assuming the sedimentation rate of 6–10 cm/kyr). The similarity of the zircon U–Pb ages of the tuff beds possibly indicates the same origin of the youngest zircon grains in each tuff bed.

Compared with the biostratigraphic age model, the zircon U–Pb age of SWT-1aU is relatively older, and those of SWT-3U and SWT-3L are probably older. On the other hand, zircon U–Pb ages of SWT-5U and SWT-5L are in conformity with biostratigraphy and can be considered as depositional ages of the tuff beds.

The major element compositions of the volcanic glass shards of the measured tuff beds in SW Section also show similar properties. The K₂O-TiO₂ diagram, which is often used to classify Quaternary tephra beds (e.g., Aoki and Machida 2006), reveals that the volcanic glass shards of the tuff beds in SW Section are divided into two groups: U-group and L-group (Fig. 6b). The U-group is characterized by higher K₂O and lower TiO₂ concentrations than the L-group and consists of all upper counterparts (SWT-1aU, SWT-2U, SWT-3U, and SWT-5U). L-group contains the lower counterparts (SWT-3L and SWT-5L). The thickness of diatomaceous mudstone intercalated between the upper and lower tuff beds of each tuff set ranges from 10 to 35 cm, which corresponds to the time interval of 1–6 kyr assuming a sedimentation rate of 6–10 cm/kyr. If each tuff set in SW Section was formed by each series of eruptions, the similarity of the compositional difference or transition between the upper and lower tuff beds of each tuff set may reflect some indigenous features in the magmatic evolution, for example, assimilation and crystallization differentiation.

The thickness and the sedimentary structures of the tuff beds in SW Section indicate that the tuff beds are formed by proximal eruptions or resedimentation of nearby pyroclastic materials (Fig. 5). Considering the similarity in the chemical compositions of the volcanic glass shards and the U–Pb ages of the youngest zircon grains, volcaniclastic components in the thick tuff beds in SW Section could originate from single or associated magmatic activities around Sado Island.

4.2 Possible source of older zircon grains

Although considerable zircon grains have younger U–Pb ages of ca. 12 Ma, many more grains show U–Pb ages from 30 to 20 Ma (Fig. 9a). Especially, SWT-1aU, SWT-2U, SWT-3U, and SWT-5U have large proportions of the zircon grains representing 20–30 Ma: 55%, 89%, 66%, and 82%, respectively. These older zircon grains seem to be inherited zircon grains incorporated during the eruption or the assimilation in the magma chamber.

Fig. 9

a An age histogram of zircon grains and b age distributions of the basement rocks. A yellow shade represents the age range from 20 to 30 Ma. Colors in the histogram represent individual tuff samples taken from SW Section. Reported ages of the basement rocks on Sado Island, the Nyukawa, Aikawa, Masaragawa, and Kimpokusan Formations, are shown. FT ages are based on Ganzawa (1982) and Metal Mining Agency of Japan (1987). K–Ar ages are based on Shibata et al. (1979), Konda and Ueda (1980), Metal Mining Agency of Japan (1987), (1988), (1989), and Shinmura et al. (1995). These ages are listed in Supplementary Table 2. (1) The Aikawa Formation includes the Oonogawa and Sanze Formations. (2) The Masaragawa Formation includes the Toyooka Formation. (3) The Kimpokusan Formation includes the Kyozukasan Formation

The basement rocks forming most part of Sado Island are volcanic rocks formed during the Oligocene to the early Miocene (Figs. 1 and 2). Figure 9b shows the compiled data of fission-track (FT) ages and K–Ar ages of the volcanic basement rocks on Sado Island, the Nyukawa, Aikawa, Masaragawa, and Kimpokusan Formations (Supplementary Table 2; Shibata et al. 1979; Konda and Ueda 1980; Ganzawa 1982; Metal Mining Agency of Japan 1987, 1988, 1989; Shinmura et al. 1995). Some consideration would be needed because all FT ages were reported before the standardization of FT dating calibration by International Union of Geological Sciences subcommission on geochronology (Hurford 1990) and some samples possibly experienced thermal alterations (Metal Mining Agency of Japan 1988). However, the ages of the volcanic basement rocks are possible to explain the age distributions of the older zircon group in the tuff samples from SW Section (Fig. 9). The proximity of the volcanic eruption site inferred by the thickness and the sedimentary structures of the tuff beds in SW Section also supports that the older zircon grains representing 30–20 Ma derive from the volcanic basement rocks around Sado Island.

Thick tuff beds in the lower part of the Nakayama Formation were also reported in the Kosado region. In the Ogi-Donokama area, several tuff beds with dozens of centimeters to several meters in thickness are intercalated with hard siliceous mudstone equivalent to the Nakayama Formation. Ages ranging from 13 to 11.5 Ma are reported as FT ages of the tuff beds though the detailed sampling points are unclear (Kanzo 1989). In the Ogi-Shiroyama area, several thick felsic tuff beds, one of which is almost 10 m thick, were recognized in diatomaceous mudstone corresponding to the diatom zone NPD 5B₃ (ca. 12.3–11.7 Ma) in the Nakayama Formation (Yanagisawa 2012). Comprehensive studies including those tuff beds in the Kosado region are needed to reveal the magmatic system and the volcanic activities that formed thick tuff beds in SW Section during the middle Miocene.

5 Conclusions

In this study, we obtained new zircon U–Pb ages of 8 tuff beds intercalated with the middle to late Miocene diatomaceous mudstone in the Nakayama Formation on Sado Island in Central Japan. In addition, we newly evaluated the ages and error ranges of the diatom and radiolarian biostratigraphy used in the Japan Sea region to compare its ages with our zircon U–Pb ages. As a result, we obtained the following findings,

• 1. The zircon U–Pb ages of WKA-5 (6.7 ± 0.2 Ma) and TBA-3 (10.87 ± 0.07 Ma) are consistent with the diatom biostratigraphy within the 2 σ. The combination of the zircon U–Pb ages with the biostratigraphic age constraints resulted in a more accurate age model with a smaller error range.

• 2. The zircon U–Pb ages of the measured tuff beds in SW Section are around 12 Ma and not distinguishable within the 2 σ. The only 2 zircon U–Pb ages from the lowest set of the tuff beds, SWT-5U and SWT-5L, are consistent with the diatom and radiolarian biostratigraphy and can be used as sedimentary ages.

• 3. The similarities in the U–Pb age distributions of zircon grains and the chemical compositions of volcanic glass shards obtained from the measured tuff beds in SW Section indicate that they should originate from single or associated magmatic activities.

• 4. The U–Pb ages of the older zircon grains make an assemblage from 30 to 20 Ma. These older ages can be explained by the contamination of zircon grains from the volcanic basement rocks around Sado Island.

As the final remark, we emphasize that integration of zircon U–Pb dating with biostratigraphy realizes stronger age determination by making up for the individual weak points.