Isotopic Characteristics of Neogene-Quaternary Tephra From IODP Site U1438: A Record of Explosive Volcanic Activity in the Kyushu-Ryukyu Arc
Abstract
International Ocean Discovery program (IODP) Site U1438 is located within the Amami-Sankaku Basin, ~50 km SW of the Kyushu-Palau Ridge and ~500 km SE of the present-day volcanic front of the Kyushu-Ryukyu arc. Thirty-eight rhyolitic tephra layers (1–15 cm thick) deposited over the last 4 Myr were recovered at this site. Representative ash layers dominated by unaltered, colorless glass shards reaching 100–300 μm in size, in association with rare pyroxene, amphibole, biotite, and zircon, were sampled for isotopic analysis. The isotopic compositions of the tephra have a narrow range in 87Sr/86Sr (0.704 to 0.706), 206Pb/204Pb (18.32 to 18.46), 207Pb/204Pb (15.57 to 15.62), and 208Pb/204Pb (38.46 to 38.75) and are more variable in εNd (−3.5 to +4.8) and εHf (+2.0 to +13.3). This record indicates the tephra were not sourced from the Izu-Bonin-Mariana arc or from Central Japan but derived from felsic volcanism from the Kyushu-Ryukyu arc. The 4.8-km-deep Amami-Sankaku Basin was located up to 600–900 km from the Kyushu-Ryukyu arc over the last 4 Ma, which is significantly more distal than other drill sites from which extensive tephra layers sourced from SW Japan have been recovered. Site U1438 tephra are thus likely related to widely distributed (M > 6) rhyolitic eruptions. We propose the unique tephra record and high-precision isotope analysis of recovered tephra from Site U1438 can be used as an important marker to identify submerged calderas within the Ryukyu arc and/or constrain the history of activity (>1 Ma) of major calderas-forming eruptions within the Kyushu-Ryukyu arc.
Key Points
- IODP Site U1438 records a unique 4-Ma history of rhyolite eruptions from the Kyushu-Ryukyu arc
- Explosive (>6 M) eruptions at 4 Ma, 2–2.5 Ma and <1 Ma in southwest Japan recorded in deep ocean basin (Amami-Sankaku Basin)
- Radiogenic isotopes of tephra might be key to distinguish the source of eruptive products
1 Introduction
Large caldera-forming eruptions are regional and global hazards and may act as an important control on short- and long-term climate change (e.g., Self, 2006). This is particularly the case in Japan where the historical and geological records shows periods of intense volcanism associated with caldera-forming, explosive rhyolitic eruptions (e.g., Machida, 1999, 2002; Yamamoto et al., 2018). The historical tephra record is thus of fundamental importance in any effort to build a comprehensive representation of the explosive volcanic activity through time of particular arc segments in order to better mitigate potential volcanic hazards (e.g., Self, 2006) and constrain the long- and short-term tectonic, mantle and subduction control on timescales of magma production and volcanism (e.g., DeCelles et al., 2009; Ducea et al., 2015; Kimura et al., 2014).
Tephra recovered from ocean sediments are an important resource aiding the completion of tephra records developed through on-land studies (e.g., Kutterolf et al., 2018; Schindlbeck et al., 2018, and references therein). Subaerial tephra records are typically more prone to erosion, burial, and alteration, particularly for older sequences, and so are often less complete (e.g., Mahony et al., 2016). Moreover, subaerial tephra deposits might also reflect the effects of more local eruptions whose effects are negligible at the larger (e.g., Japan) scale (e.g., Torii et al., 2000). Correlations between individual tephra layers over several hundreds of kilometers in both marine and subaerial settings are also hampered by uncertainties in chronology, postdepositional erosion, bioturbation and localized versus regional tephra fallout (e.g., Mahony et al., 2016). Though individual tephra layers might be correlated through depositional ages, the mineralogy and composition of glass shards is crucial in determining their sources (e.g., Kotaki et al., 2011; Schindlbeck et al., 2018; Tsuji et al., 2017). Nevertheless, constraining the source and extent of ash deposits is an ongoing challenge, as tephra deposits may also derive from currently unknown volcanic sources, such as submerged calderas (e.g., Tatsumi et al., 2018), volcanic centers covered by products of younger eruptions, or might be related to calderas with unconstrained ages, or whose relations with particular tephra markers are disputed (e.g., Moriwaki et al., 2008; Nishizawa & Suzuki, 2016).
The Quaternary tephrostatigraphy of Japan, based on the detailed study of sedimentary basins and Ocean Drilling Program (ODP)/Deep Sea Drilling Project/International Ocean Discovery program (IODP) stratigraphy and samples, is well established. Consequently, the timing of recent silicic, caldera-forming eruptions, which cover parts of Japan in thick deposits, is relatively well known (e.g., Kimura et al., 2015; Machida, 1999, 2002). Though source characterization and correlations of large-scale rhyolitic eruptions in the last 5 Ma have been recently attempted using Pb isotope composition of glass shards (Kimura et al., 2015), the tephrostratigraphy and locations of caldera-forming eruptions older than 1 Ma in Japan remains significantly less well constrained (e.g., Mahony et al., 2016).
In order to determine more precisely the extent and volumes of tephra fallout related to large-scale eruptions >1 Ma, it is necessary to target tephra layers from ocean basins hundreds of kilometers from any arc-front or (known) volcanic source. This paper reports new isotopic data (Sr-Nd-Hf-Pb) for 17 rhyolitic tephra layers younger than <4 Ma and one older tephra layer (ca. 14 Ma) sampled during IODP Expedition 351 from a deep ocean basin SE of Kyushu. On the basis of comparisons with existing data for tephra of known provenance, we conclude that rhyolitic tephra from the Amami-Sankaku Basin (ASB) were derived from previously unrecognized and major eruptive sources in the Kyushu-Ryukyu arc. The tephra record of the ASB might therefore be used as markers to extend the record of major caldera-forming eruptions in southwest Japan beyond 1 Ma.
2 Background
In 2014, IODP Expedition 351 targeted Site U1438 within the ASB (4.8 km below sea level) in the north central portion of the Philippine Sea plate (Figure 1). The basin is located ~50 km SW of the Kyushu-Palau Ridge (KPR), a remnant of subduction at the Izu-Bonin-Mariana (IBM) Arc that was created between 50 and 25 Ma (e.g., Arculus et al., 2015; Ishizuka et al., 2011). The ASB is located ~500 km SE of the present-day volcanic front of the Kyushu-Ryukyu arc, ~850 km south of the Southwest Japan arc (e.g., Kamata & Kodama, 1999), and ~640 km west of the currently active Izu-Bonin arc (Figure 1). The formation of the Southwest Japan and Kyushu-Ryukyu arcs results from the subduction of the Philippine Sea plate beneath southwestern Japan. Variations in magmatism over the last 15 Ma reflect changes in subduction dynamics, including changes in orientation of subduction in Pleistocene-Pliocene times, subduction of the KPR and Shikoku Basin (e.g., Kamata & Kodama, 1999; Kimura et al., 2014; Mahony et al., 2016; Pownall et al., 2017; Shinjo et al., 2000).
One of the primary aims of IODP Expedition 351 was to target the sedimentary and volcaniclastic record at Site U1438 to constrain the evolution of the arc following inception at ~50 Ma (Arculus et al., 2015). Sediments and sedimentary rocks recovered at Site U1438 up to 1461 m below seafloor (mbsf) were separated into four distinct lithostratigraphic units (Units I to IV; Arculus et al., 2015) and overly a unit of igneous basement (Unit 1: 1461–1611 mbsf). The deeper portions of the volcaniclastic and sedimentary record in the core (Unit III: ~310–1361 mbsf, and IV: ~1361–1461 mbsf) are primarily related to the Paleogene Izu-Bonin arc developed at the KPR (50 to 25 Ma; Brandl et al., 2017). These 1,300 m of volcaniclastic gravity flow deposits (tuffaceous mudstones, tuffaceous sandstones, and tuffaceous breccia-conglomerates) overlie an igneous basement of low-Ti, low-K tholeiitic basaltic lava flows of 49.3- to 46.8-Ma age (Ishizuka et al., 2018). The uppermost 160.3 m (Unit I) of the ASB sediments, however, comprise tuffaceous mud and mud with ash containing occasional centimeter-thick ash layers (Figures 2a and 2b). The pelagic sediments of Unit I were deposited at low rates consequent to the cessation of volcanism and magmatic activity of the KPR at circa 25 Ma (Arculus et al., 2015).
3 Samples and Methods
Tephra layers recovered in Unit I were identified during shipboard core description on IODP Expedition 351 using standard description methods, including smear-slides, changes in sediment color, magnetic susceptibility, and grain size. The presence and thickness of individual layers was then double-checked onshore using high-resolution, color-scanned core images. Tephra layers were sampled onboard from IODP Holes U1438A and U1438B in Unit I (Arculus et al., 2015). The ashes form discrete layers (Figures 2a and 2b), often varicolored, 1–15 cm thick with most layers ranging from 1–4 cm in thickness and deposited between 0 and 4 Ma (Figures 3 and 4). The tephra usually show sharp to bioturbated basal contacts and gradational upper contacts into the mud, though both upper and lower contacts are often slightly to strongly bioturbated. One particular ash layer (Table 1) was found associated with a foraminiferal ooze, suggesting that the thickness might occasionally not be well constrained. Colorless vitric glass shards and pumices dominate the ash layers, with maximum grain sizes usually from 100–300 μm (Figures 2c–2f). The shards are angular and have well-preserved bubble walls (Figure 2c). Signs of alteration are rare and limited to the presence of spotted, whitish (less transparent) glass in certain samples (Table 1). Some tephra layers contain <1% pyroxene, amphibole, biotite ± zircon (e.g., Figure 2e). Representative unaltered tephra samples distributed throughout the last 4 Ma were sampled for Sr-Pb-Nd-Hf isotopes (Figure 4). Age-depth relationships for holes U1438A-B is based on the correlation between magnetozones and the geomagnetic polarity timescale (Arculus et al., 2015). The shipboard correlation of 87 geomagnetic reversals to the geomagnetic polarity timescale (Cande & Kent, 1995; Gradstein et al., 2012) at Site U1438 allows for high-resolution magnetostratigraphy up to Chron C16n.1r (36.051 Ma). For further details regarding the methods and correlations, the reader is referred to Arculus et al. (2015).
Site/hole | U1438A | U1438A | U1438A | U1438B | U1438B | U1438B | U1438B | U1438B | U1438B | U1438B | U1438B | U1438B | U1438B | U1438B | U1438B | U1438B | U1438B | U1438B |
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
Core/Section | 1H2W | 1H4W | 3H1W | 1H2W | 4H1W | 4H6W | 5H2W | 5H3W | 5H4W | 6H4W | 6H5W | 7H1W | 7H3W | 7H3W | 7H5W | 8H1W | 8H2W | 14H4W |
Interval sampled (cm) | 94–96 | 128–130 | 90–92 | 42–44 | 135–138 | 46–48 | 113–115 | 79–82 | 104–106 | 115–117 | 73–75 | 27.32 | 74–85 | 136–138 | 115–116 | 18–20 | 73–75 | 138–140 |
Thickness (cm) | 2 | 10 | 2 | 2 | 9 | 3 | 4 | 3 | 1 | 2 | 15 | 10 | 8 | 4 | 2 | 6 | 1 | 3 |
Estimated Age (Ma) | 0.12 | 0.28 | 0.91 | 0.04 | 1.45 | 1.84 | 2.10 | 2.17 | 2.28 | 2.91 | 2.98 | 3.19 | 3.43 | 3.47 | 3.67 | 3.86 | 4.00 | 14.67 |
minerals associated | — | — | biotite, amphibole | — | — | — | — | — | amphibole | amphibole | — | — | — | — | amphibole | plagioclase | — | biotite, amphibole, zircon |
Max shard size (μm) | 300 | 350 | 300 | 200 | 200 | 100 | 300 | 200 | 200 | 200 | 100 | 200 | ||||||
glass alteration | unaltered | unaltered | unaltered | unaltered | unaltered, Found with Foraminifera ooze | slight devitrification of ash | unaltered | unaltered | slight devitrification of ash | unaltered | unaltered | unaltered | slight devitrification of ash | unaltered | slight devitrification of ash | unaltered | unaltered | unaltered |
Sm ppm | 3.4 | 2.4 | 2.8 | 3.4 | 3.0 | 3.9 | 6.1 | 4.3 | 3.6 | 3.0 | 4.3 | 2.9 | 3.1 | 6.7 | 2.9 | 4.3 | 6.1 | 2.4 |
Nd ppm | 15 | 15 | 16 | 18 | 15 | 20 | 26 | 22 | 14 | 16 | 24 | 14 | 15 | 31 | 14 | 22 | 26 | 9 |
143Nd/144Nd | 0.512756 | 0.512582 | 0.512600 | 0.512461 | 0.512684 | 0.512793 | 0.512718 | 0.512770 | 0.512617 | 0.512721 | 0.512742 | 0.512681 | 0.512880 | 0.512677 | 0.512682 | 0.512649 | 0.512654 | 0.512555 |
1σ | 2 | 3 | 2 | 3 | 3 | 3 | 3 | 2 | 2 | 3 | 3 | 3 | 3 | 2 | 2 | 2 | 3 | 3 |
eNdm | 2.30 | −1.09 | −0.75 | −3.45 | 0.89 | 3.03 | 1.55 | 2.57 | −0.41 | 1.62 | 2.03 | 0.84 | 4.73 | 0.76 | 0.86 | 0.21 | 0.31 | −1.62 |
Rb ppm | 87 | 97 | 109 | 124 | 106 | 106 | 76 | 88 | 119 | 102 | 99 | 103 | 51 | 102 | 99 | 124 | 102 | 189 |
Sr ppm | 51 | 74 | 65 | 67 | 65 | 82 | 154 | 113 | 50 | 73 | 86 | 53 | 86 | 90 | 62 | 47 | 74 | 25 |
87Sr/86Sr | 0.70471 | 0.70554 | 0.70505 | 0.70593 | 0.70533 | 0.70426 | 0.70449 | 0.70438 | 0.70578 | 0.70455 | 0.70453 | 0.70535 | 0.70434 | 0.70497 | 0.70542 | 0.70586 | 0.70592 | 0.70735 |
1σ | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 |
176Hf/177Hf | 0.283063 | 0.282927 | 0.282916 | 0.282830 | 0.283013 | 0.283116 | 0.283024 | 0.283094 | 0.282934 | 0.282967 | 0.282979 | 0.282973 | 0.283149 | 0.282998 | 0.282989 | 0.282942 | 0.282958 | 0.282865 |
1σ | 3 | 3 | 3 | 3 | 4 | 4 | 4 | 4 | 5 | 3 | 4 | 4 | 4 | 4 | 4 | 4 | 4 | 5 |
eHfm | 10.3 | 5.5 | 5.1 | 2.1 | 8.5 | 12.2 | 8.9 | 11.4 | 5.7 | 6.9 | 7.3 | 7.1 | 13.3 | 8.0 | 7.7 | 6.0 | 6.6 | 3.3 |
U ppm | 2.3 | 1.8 | 2.5 | 2.5 | 2.9 | 2.6 | 1.6 | 1.9 | 2.3 | 2.4 | 2.1 | 2.1 | 1.2 | 2.2 | 2.1 | 2.2 | 1.9 | 4.1 |
Th ppm | 8.2 | 11.2 | 9.8 | 12.1 | 11.4 | 11.5 | 12.5 | 12.0 | 10.1 | 11.7 | 14.7 | 11.2 | 5.5 | 12.1 | 9.2 | 14.4 | 10.5 | 7.4 |
Pb ppm | 19.9 | 13.2 | 18.8 | 20.4 | 15.4 | 14.7 | 11.1 | 11.1 | 16.1 | 15.5 | 15.2 | 20.2 | 9.8 | 16.4 | 15.3 | 15.4 | 14.6 | 18.4 |
206Pb/204Pb | 18.3877 | 18.4196 | 18.3588 | 18.3946 | 18.4056 | 18.4402 | 18.4397 | 18.4162 | 18.4630 | 18.3303 | 18.3223 | 18.4735 | 18.3233 | 18.4276 | 18.4676 | 18.4483 | 18.4321 | 18.4354 |
1σ | 3 | 2 | 3 | 2 | 2 | 3 | 2 | 2 | 3 | 3 | 2 | 3 | 2 | 2 | 3 | 2 | 3 | 2 |
207Pb/204Pb | 15.569 | 15.600 | 15.585 | 15.598 | 15.598 | 15.594 | 15.599 | 15.592 | 15.620 | 15.575 | 15.571 | 15.615 | 15.577 | 15.601 | 15.618 | 15.617 | 15.612 | 15.598 |
1σ | 3 | 2 | 3 | 2 | 2 | 3 | 2 | 2 | 3 | 3 | 2 | 3 | 2 | 2 | 3 | 2 | 3 | 2 |
208Pb/204Pb | 38.459 | 38.619 | 38.555 | 38.631 | 38.647 | 38.646 | 38.660 | 38.623 | 38.747 | 38.485 | 38.465 | 38.712 | 38.477 | 38.632 | 38.740 | 38.725 | 38.706 | 38.650 |
1σ | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 |
Correlated with | Kikai Volcano (Kyushu) | Ryukyu Arc-Kyushu, ODP Leg 126 Hole 795 2H-2, 84-86 | Ryukyu Arc-Kyushu: ODP Leg 126 Hole 811 8H-CC, 2-4 | Aira-Tanzawa Volcano (Kyushu) | Ryukyu-Kyushu Arc | Ryukyu-Kyushu Arc | Ryukyu-Kyushu Arc | Ryukyu-Kyushu Arc | Ryukyu-Kyushu Arc | Ryukyu-Kyushu Arc | Ryukyu-Kyushu Arc | Ryukyu-Kyushu Arc | Ryukyu-Kyushu Arc | Ryukyu-Kyushu Arc | Ryukyu-Kyushu Arc | Ryukyu-Kyushu Arc | Ryukyu-Kyushu Arc | Ryukyu-Kyushu Arc |
Samples were processed and analyzed at the School of Earth, Ocean, and Environment at the University of South Carolina (USA). Samples were wet-sieved between 180 and 90 μm using single-use Nylon sieves. In order to remove the clay fraction, 5 g/L of sodium hexametaphosphate Na-(PO3)6 was used to deflocculate the clay in an ultrasonic bath. This procedure was repeated until the water was clear. Rare carbonate was removed by letting samples rest in 15% HCl for up to 15 min. Samples were then repeatedly cleaned in 18 MΩ H2O in order to remove any possible influence of Pb contamination from the Na-(PO3)6 (measured Pb contents of the Na-(PO3)6 were <0.45 ppb). Density separation by panning in order to remove the heavier mineral fraction was followed by hand picking underneath a microscope in order to produce 100–150 mg of pure glass shards (Figures 2d and 2e). Most samples are crystal poor or crystal free and dominated by colorless, transparent glass shards. Samples were then further rinsed three times in 18 MΩ H2O prior to being digested for 24 hr in a 3:1 HF:HNO3 mixture in capped Teflon vials. Digested samples were subsequently uncapped and dried down on a hot plate. Fluorides were removed by adding a 3–4 ml of concentrated HNO3 followed by sealing the Teflon capsules at ~90 °C before being dried down again. This step was repeated three times.
After the final digestion steps, separate aliquots of the digest were processed for Nd, Sr, Pb, and Hf column chemistry. Rare earth elements were first isolated using the TRU-spec resin (Eichrom) in HNO3. Neodymium was then separated from this cut using LN-spec resin in HCl (e.g., Pin & Zalduegui, 1997). Hafnium was separated using a method modified after Münker et al. (2001). Pb was extracted in HNO3-HBr extraction techniques in anion resin (e.g., Abouchami et al., 1999). Procedural blanks where always more than 1,000 lower than the amount of sample processed.
Isotope ratios were measured on the Thermo Finnigan Neptune Multi collector inductively coupled plasma source mass spectrometry (ICP-MS) at the Center for Elemental Mass Spectrometry at the University of South Carolina. Solutions were introduced with a 100-ml self-aspirating Teflon nebulizer (ESI, USA) coupled to an ESI APEX-Q system using a Jet sample cone and X-skimmer cone. Pb isotopes were measured using the Tl addition technique (White et al., 2000). Isotope ratios were corrected for fractionation using 203Tl/205Tl = 0.418911 and the exponential law. Average ratios measured on NBS 981 standard solutions were 206Pb/204Pb = 16.9358 ± 0.00097, 207Pb/204Pb = 15.4893 ± 0.00096, and 208Pb/204Pb = 36.6927 ± 0.0027 (2 standard errors, n = 13). Sample compositions were corrected for instrumental bias using reference values for NBS 981 reported by Todt et al. (1996).
For Hf and Nd, the samples were diluted to achieve a signal of 2–5 V on 143Nd and 176Hf, and 30–45 analyses were run on each sample, depending on sample volume. Samples were bracketed every three to six solutions by the JNdi-1 standard for Nd, and the JMC-457 standard for Hf. Repeat analysis of U.S. Geological Survey rock standard AGV-1 (n = 2) run as an unknown agrees with published reference values of Weis et al. (2006, 2007) to within 0–0.4 epsilon units. The Hf and Nd isotopic composition of samples are reported relative to the accepted values for these standards (176Hf/177Hf = 0.282160; Vervoort & Blichert-Toft, 1999; 143Nd/144Nd=0.512115, Tanaka et al., 2000).
Analyses of glass shards for major and minor elements were performed at Oregon State University with a CAMECA SX-100 electron microprobe using 15-keV accelerating voltage and 10-nA sample current. Counting times ranged from 10 to 60 s depending on the element and desired detection limit. In order to reduce the effects of alkali migration, the alkalis were measured first and a 0-time intercept function was applied to the data. Data reduction was performed online using a stoichiometric PAP correction model (Pouchou & Pichoir, 1984). All analyses are normalized to 100% to eliminate the effects of variable post depositional hydration. Analyses with total oxides less than 90 wt % were excluded from the data set. Laser ablation inductively coupled plasma mass spectrometer (LA-ICP-MS) analyses of trace elements in the same ash shards as microprobe analyses were performed using a Photon Machines Analyte G2 excimer laser system (ArF excimer 193 nm) coupled with a Thermo XSeries II quadrupole ICP-MS at the WM Keck Collaboratory for Plasma Spectrometry at Oregon State University. Shard analyses were conducted in spot mode using a masked spot rectangle of 27 μm × 75 μm, oriented along the long axis of the ash shard, with a laser pulse rate of 5 Hz, analysis time of ~45 s per spot, and laser energy fluence of 4.84 mJ/cm2. Reference glass standard GSE-1G (analyzed after ~10 unknowns) was used for calibration, natural glass standards ATHO and BCR-2G (Jochum et al., 2005) as secondary standards to monitor accuracy and precision, and 29Si was used as an internal standard in conjunction with SiO2 contents measured by electron microprobe.
4 Results
Results are shown in Table 1 and plotted against age (depth) in Figure 6. The U1438 ashes show a narrow range in 87Sr/86Sr (0.704 to 0.706), 206Pb/204Pb (18.32 to 18.46), 207Pb/204Pb (15.57 to 15.62), and 208Pb/204Pb (38.46 to 38.75) with greater variation in εNd (−3.5 to +4.8) and εHf (+2.0 to +13.3). There is an inverse relationship between 87Sr/86Sr and 143Nd/144Nd, with 87Sr/86Sr becoming gradually more radiogenic as 143Nd/144Nd becomes less radiogenic (Figure 7a). Major and trace element compositions indicate that all samples are rhyolitic, with >70 wt % SiO2 and Na2O+K2O of 4–9 wt % (Figure 5 and the supporting information). These compositions are similar to those of known rhyolitic caldera-forming eruptions over the last 5 Ma from throughout Japan (Kimura et al., 2015; Figure 5).
5 Discussion
5.1 The Source of Neogene and Quaternary Tephras From Site U1438
A depositional environment wherein fine-grained air-fall tephra are rapidly transported to the seafloor as gravity currents in the form of turbid plumes is likely (speeds >2 cm/s; e.g., Carey, 1997; Lowe, 2011). This implies that the tephra could have reached the floor of the ASB, currently at 4.8-km water depth, in ≤65 hr. Descending tephra plumes were therefore most likely not significantly affected by remobilization, redistribution by strong oceanic currents before final deposition at depth within the ASB. We conclude that the tephra are most likely to be distal air-fall deposits from explosive eruptions that occurred a few days prior to deposition. A rapid depositional environment and lack of remobilization of tephra layers is also consistent with the preserved angular shape of ash shards and the absence of identifiable seawater contamination (Figures 2 and 6b). Indeed, had seawater alteration affected the tephra compositions significantly, Sr isotopes would increase at constant εNd due to the amount of radiogenic Sr in ocean water, as is shown in the basement basalts from Site U1438 (e.g., Yogodzinski et al., 2018). Moreover, 87Sr/86Sr is not correlated with shard alteration (as indicated by patchy white, colorless glass). Thus, Sr isotopes appear to have been unaffected by interaction with seawater. In only one case (Table 1 and Figure 6b, tephra sample at age ca. 14 Ma) is it possible that 87Sr/86Sr might be affected by seawater alteration as neither Pb-Nd-Hf show any distinct composition shifts compared to other samples. Pb isotopes as well show no effect of alteration, as shown by their near homogenous 208Pb/204Pb, 206Pb/204Pb and 207Pb/204Pb (Figure 6b), consistent with the lack of radiogenic Sr seawater alteration and the general elevated abundances of Pb in these tephra (10–20 ppm; Table 1 and the supporting information). Thus, Sr and Pb, reflect the magmatic isotopic composition of the tephra layers, as is the case for Hf and Nd isotopic ratios.
As can be seen in Figures 6-8, the isotopic and trace element ratios of the tephra from the ASB are distinct from IBM arc tephra. Figures 6a–6c and 7a and 7b show that the isotopic compositions of tephra and volcanic rocks from the IBM arc are more homogenous and clearly derived from more depleted sources than most volcanic samples from Northeast or Southwest Japan, or from the Kyushu-Ryukyu arc. This distinction has prevailed throughout the 50-Ma history of IBM Arc, which has produced volcanic rocks with εNd of 5–10, 87Sr/86Sr <0.7045, and εHf generally >14 (e.g.,Straub et al., 2015 ; Yogodzinski et al., 2018). In contrast, U1438 tephra have relatively unradiogenic Nd (εNd +5 to −5) and Hf (εHf +14 to +2) and relatively radiogenic Sr (87Sr/86Sr >0.704). The fact that U1438 tephra are distinct from IBM volcanic rocks is further supported by the variation in εNd with Sm/Nd (Figure 8). IBM arc rocks and tephra show a subhorizontal trend of decreasing Sm/Nd with nearly constant εNd, which is controlled primarily by sedimentary inputs from the downgoing slab (Straub et al., 2015; Yogodzinski et al., 2018). In contrast rhyolitic tephra from the Kyushu-Ryukyu arc and silicic volcanism from the Honshu quaternary arc have lower Sm/Nd (<0.3) and εNd (<5; Figures 6 and 8). In terms of Pb isotopes, U1438 tephra are homogenous and distinct from the array of volcanic rocks from the IBM arc (Figure 7), which have higher 208Pb/204Pb and 207Pb/204Pb for a given 206Pb/204Pb (Figures 6 and 7). The U1438 tephra are similar in this respect to rhyolitic tephra sampled from throughout Japan (Figure 7), though they have a significantly more restricted compositional variability. Thus, from first-order observations, it is clear that tephra younger than 4 Ma recovered from Site U1438 show no genetic relationship with the IBM Arc, but clear overlap isotopically with rhyolitic tephra from Honshu and Kyushu.
Kimura et al. (2015) assert that most rhyolitic tephra from the main Japanese Islands are formed through partial melting of the crust. This hypothesis is supported by the fact that the extended Pb isotopic range of the tephra mimics that of the crustal basement (Kimura et al., 2015). Alternatively, given the high abundances of Pb in crustal rocks compared to most mantle-derived rocks (e.g., Miller et al., 1994), it is possible that the Pb isotopic ratios of rhyolitic melts derived by fractional crystallization of more mafic parental magmas may be controlled by assimilation of a crustal component. The U1438 tephra only occupy a restricted portion of the Pb isotopic space defined by tephra from Japan (Figure 7). It is thus likely that the U1438 tephra originated from the islands of Honshu or Kyushu or from nearby islands that share a similar geologic history. Nevertheless, it is not possible to unambiguously distinguish sources in the Kyushu-Ryukyu, Southwestern, and Central Japan arcs using Pb isotopes alone (Kimura et al., 2015).
In terms of 87Sr/86Sr and εNd, U1438 tephra overlap with isotopic compositions of known silicic eruptions from Kyushu, which overall show a limited range in 87Sr/86Sr (e.g., Ata, Aso, Aira, and Sendai ± Kirishima calderas, Arakawa et al., 1998; Figures 6g and 6h). Due to their generally more radiogenic Nd and less radiogenic Sr, U1438 tephra define a compositional array similar to silicic rocks from the Kyushu-Ryukyu arc, including samples from the Aso and Aira calderas (Figures 6g and 6h, 7a, and 8). On the same basis, the Site U1438 tephra appear unlike rhyolites and subvolcanic felsic plutons from Central Japan (Takedani Pluton, Hartung et al., 2017, and felsic volcanic eruptions, Kersting et al., 1996), which trend toward less radiogenic Nd and more radiogenic Sr (Figure 7). This is consistent with rhyolite melts having acquired part of their isotopic signature through interaction with Honshu basement crust, in part comprising isotopically distinct granites and gneisses that have highly radiogenic Sr, with 87Sr/86Sr ~0.7200, and unradiogenic Nd, with εNd ~−15 (e.g., Arakawa & Shinmura, 1995). Moreover, the isotopic compositions of U1438 tephra show slightly higher 208Pb/204Pb and 207Pb/204Pb at a given 87Sr/86Sr than silicic Central Japan magmatism and overlap with silicic magmatism from the Ryukyu-Kyushu arc (Figures 7e and 7f). Thus, it is difficult to unambiguously distinguish specific possible source(s) for the U1438 tephra within Japan. Nonetheless, the homogenous Pb values and combined Sr-Nd isotopic data patterns are consistent with the hypothesis that these deposits were sourced from eruptions in a single region, most likely derived from within the Kyushu-Ryukyu arc.
In general, the distribution of tephra at Site U1438 (Figures 4 and 9) indicates a decrease in volcanic activity between 1 and 2 Ma, and more moderate volcanic activity between 2.8 and 3.5 Ma, with peak activity between 2 and 2.5 Ma, around 4 and circa <1 Ma. Based on the tephra record, volcanic activity observed in and around Japan over the last several Ma also shows periods of large-scale volcanism clustered around 4–6, 2–2.5, and <1 Ma (Mahony et al., 2016). Thus, the distribution in time of tephra at Site U1438 is broadly consistent with the known recent evolution of volcanism on the main islands of Japan. However, the Southwest Japan arc is an unlikely contributor to these rhyolite tephra, because it shows both a volcanic hiatus between 5 and 3 Ma and a lack of caldera-forming eruptions from 3–0 Ma (Kimura et al., 2005, 2014; Figure 9). Volcanism in Central Japan (Norikura segment) shows an increase in activity between 1.8 and 2.5 Ma (Kimura et al., 2015; Mahony et al., 2016) and could thus be contributing to the U1438 tephra at 2–2.5 Ma (Figure 9), but the generally less radiogenic 87Sr/86Sr and more radiogenic εNd of Site U1438 rhyolites makes this unlikely. On the other hand, there are a number of observations consistent with an origin for U1438 within the Kyushu-Ryukyu arc. These are (1) the observed increase in magmatism ranging from basalt to rhyolite in the Central Ryukyu arc at circa 4–6 Ma (Shinjo et al., 1999 and references therein); (2) the occurrence of contemporaneous and large caldera-forming eruptions in Central Kyushu (e.g., the 1.3- to 0.6-Ma ignimbrites) and in Southern Kyushu (Kagoshima volcanotectonic graben) including the Kikai, Ata, and Aira calderas (Shinjo et al., 2000; Figure 7); and (3) the overlap of isotopic signatures of U1438 and Kyushu-Ryukyu rhyolites, combined with a lack of significant isotopic heterogeneity in U1438 rhyolites over the last 4 Ma (Figures 6 and 7).
In Japan, most late Quaternary tephra are distributed to the east of their source volcanoes, because the plumes are transported by the prevailing westerly winds (Machida & Arai, 2003). It is thus unsurprising that the isotopic record of rhyolitic tephra at Site U1438 is consistent with sources in the Kyushu-Ryukyu arc. As the demise of volcanic activity at the KPR at ~25 Ma (Ishizuka et al., 2011) was subsequently followed by subduction initiation along the Kyushu-Ryukyu arc at 15 Ma (Wu et al., 2016), it is likely that the approach of Site U1438 to the Ryukyu-Kyushu arc led to an increase in the frequency of ash deposition at the site.
5.2 Correlating Individual Tephra Layers Within the Kyushu-Ryukyu Arc
Our ability to correlate individual ash layers with known volcanic sources is hampered by the lack of detailed isotopic data for tephra throughout Japan and by the overall lack of well-constrained rhyolitic tephra deposited in SW Japan since ~1 Ma (Kimura et al., 2015). Machida (1999, 2002) and Kimura et al. (2015) correlated 15 tephra layers from the Kyushu-Ryukyu arc in the last 1 Ma with well-known tephra markers throughout Japan (Figure 9), but major tephra markers in SW Japan in the 1- to 4-Ma time window remain scarce. There have been several caldera-forming eruptions Quaternary in the Kyushu-Ryukyu Arc, including those at Aso, Aira, Ata, Kikai, Kakuto-Kobayashi, and Shishimita (e.g., Arakawa et al., 1998; Machida & Arai, 2003), but most tephras from these eruptions do not appear to be well preserved at Site U1438. The great distance from the source could be the main reason for the patchy record of certain caldera-forming silicic eruptions from Kyushu, as Site U1438 is close to the distal limits of known tephra fallout of certain caldera-forming eruptions (Machida, 1999; Machida & Arai, 2003; Tsuji et al., 2017). It is also possible that the water depth at Site U1438 might induce localized heterogeneous distribution of tephra on the seafloor during deposition by density currents. As shown in Allan et al. (2008), tephra preservation can be patchy at a given drill site, with individual cores only sampling a limited proportion of tephra layers deposited in across a basin. Thus, the lack of evidence of certain large-scale Quaternary eruptions is unsurprising. Nevertheless, based on Sr-Nd isotopic composition, depositional age, and overlapping major element compositions of ash shards compared to tephra samples of known rhyolitic eruptions measured by Tsuji et al. (2017) and Kimura et al. (2015), we are confident that the caldera-forming eruption of Aira (Aira-Tanzawa Tephra, AT), the most prominent tephra marker in the late Pleistocene at circa 29 ka (Machida et al., 2002), and Kikai Volcano (eruption at 77–101 ka, K-Tz; Machida & Arai, 1992, Tatsumi et al., 2018) are well preserved within the tephra record of U1438 (Table 1). The 2-cm-thick ash falls from the Aira and Kikai eruptions are also consistent with the estimated thickness of tephra deposits at this location according to tephra isopach maps of Machida (1999).
Figures 6g and 6h and 9 show that the depositional age and isotopic composition of several tephra layers from the IBM forearc and sampled during ODP Leg 792 (Egeberg et al., 1992) are similar to those at Site U1438. This supports the interpretation proposed by Egeberg et al. (1992) and more recently by Kutterolf et al. (2018) and Schindlbeck et al. (2018) that occasional tephra markers within the IBM arc (Figure 1 for location of ODP Site 792, IODP Exp. 350 Sites U1436-U1437 and IODP Exp. 352 Sites U1439-U1442) are derived from the Kyushu-Ryukyu arc. Occurrence of AT tephra and Kikai-Akahoya (K-Ah) tephra on some Izu-Bonin islands (e.g., Ishizuka & Geshi, 2018) provides further support for this interpretation. Additional correlations for tephra at 2–4 Ma might be possible on the basis of the 2- to 4-Ma tuff layers (rhyolites) in the Miyazaki Group from the Miyazaki basin in the southeastern part of Kyushu (Torii et al., 2000; Figure 9), but the lack of isotopic data and imprecise age estimates for the subaerial deposits means that such correlations would be tenuous.
We find that the predominant source(s) of the U1438 rhyolitic tephra lie predominantly within the Kyushu-Ryukyu arc, but the exact volcanic source(s) of most of the layers remain unknown. The abundance of caldera-forming eruptions from the Kyushu-Ryukyu arc, which dominate the sedimentary deposits of ocean basins to the east of their location, as shown by the abundance of rhyolitic tephra from Kyushu-Ryukyu arc in the ASB and not from the IBM arc, is also consistent with recent studies showing that tephra from the Kyushu-Ryukyu arc and Central Japan arc contribute significantly to the sedimentary record within the IBM system (Egeberg et al., 1992; Kutterolf et al., 2018; Schindlbeck et al., 2018) and broader Pacific region (e.g., Scudder et al., 2016).
5.3 Implications for Large Rhyolitic Eruptions in Southwest Japan at 0–4 Ma
Though Site U1438 is located at the closest, ~50 km from the KPR, the isotopic composition of the tephra imply a source within the Kyushu-Ryukyu arc. At present, Site U1438 is located ~600 km from Central Ryukyu (Io-Torishima Volcano), 590 km from South Kyushu (Aira Volcano) and 680 km from Central Kyushu (Aso Volcano). Site U1438 is significantly farther away from Southwest Japan than other drill sites with important tephra records on the Philippine Sea plate, such as Deep Sea Drilling Project sites 296 and 297 (e.g., Ingle et al., 1975), both of which are 300–400 km from Central Kyushu calderas. As Site U1438 is located on the Philippine Sea plate, the location of the drill site became closer to Southwest Japan with time. Assuming a subduction rate at the Kyushu-Ryukyu Trench of ~ 6 cm/year, the site could have been ~240 km further to the southeast from its current location at 4 Ma. The tephra layers would therefore have been derived from eruptive centers located at distances of greater than 800 km. Though preservation of thick tephra deposits in ocean sediments is not unusual, as shown in the IBM forearc and in the Shikoku and Ryukyu sedimentary basins (e.g., Egeberg et al., 1992; Reagan et al., 2015; Straub et al., 2015; Tsuji et al., 2017), preservation of thick tephra layers from eruptive centers situated at such distances (>600 km) are significantly less common (e.g., Schindlbeck et al., 2018). Mahony et al. (2016) modeled the tephra thickness with distance to the volcanic source as a function of eruption magnitude, demonstrating that tephra deposited several hundreds of kilometers from the trench are most likely to be related to large (>M5) caldera-forming eruptions. Moreover, Schindlbeck et al. (2018) estimated the magnitude of volcanic eruptions based on tephra thickness recovered at IODP Site U1437 within the IBM forearc, and distance from a known source in Kyushu, as M6.3–7.3. We conclude the thick ash layers at Site U1438, allowing for the age-distance relations from the Kyushu-Ryukyu arc, record >M6 eruptions. Though such large-scale eruptions should be identifiable by the presence of widespread tephra deposits in and around Japan and the Philippine Sea, there is, as of now, no clear silicic tephra record of volcanic eruption from the Kyushu-Ryukyu arc that can be clearly linked to thicker tephra fallout at circa 1.5, 3.0–3.2, and 4.1 Ma (Kimura et al., 2015; Machida, 2002; Figure 9). We suggest that the location of the deep (4.8-km water depth) ASB located >600km from the Kyushu-Ryukyu arc makes it an ideal candidate to identify currently unknown large-scale eruptions that occurred in the Neogene and Quaternary. High-precision dating and isotope analysis of such calderas and sedimentary basins containing an array of tephra markers (e.g., Miyazaki Basin) could help constrain the history of explosive eruptions in southwestern Japan, with Site U1438 in the ASB representing a crucial marker of large, caldera-forming rhyolitic eruptions in the last 4 Ma within the Kyushu-Ryukyu arc. We suggest that unidentified, submerged large-scale calderas within the Kyushu-Ryukyu arc, similar to submerged calderas already known to exist within the northern IBM arc (e.g., Sumisu caldera, Tani et al., 2008) and Northern Ryukyu (Shishimuta Caldera, Kamata, 1989) might in part be responsible from these tephra deposits, as well as unidentified older historical eruptions from known calderas within Kyushu.
6 Conclusions
Isotopic analysis of ash layers from IODP Expedition 351 indicates the tephra layers at Site U1438 from the ASB preserve a record of rhyolitic eruptions derived primarily from the Kyushu-Ryukyu arc, with no clear evidence of any significant eruptions derived from Central Honshu, southwestern Japan,and the IBM arc. The tephra record at Site U1438 can in part be correlated with Quaternary eruptions of Aira and Kikai on Kyushu Island, though other crucial markers seem to be missing. The <4-Ma tephra record from IODP Site U1438 records significant tephra accumulations (9–15 cm thick) from rhyolitic eruptions for which volcanic source(s) are as yet unconstrained but were at considerable distance from Site U1438 (>600 km). The implication is that the eruptions must have been M > 6. The unique rhyolitic tephra record of Site U1438 might therefore be used as a marker for caldera-forming eruptions within the Kyushu-Ryukyu arc over the last 4 Ma.
Acknowledgments
A. M. acknowledges support from the Swiss National Science Foundation (grant 200020/135511 to Othmar Müntener and grant P2LAP2_171819 to A. M.) and Swiss IODP. This work was also supported by grants to G. M. Y. and M. B. from NSF (OCE-1537135) and IODP. The authors thank W. Buckley at the Center for Elemental Mass Spectrometry for their assistance with data collection at the University of South Carolina. The work and assistance of the captain and crew of the R/V JOIDES Resolution are gratefully acknowledged. We are grateful for reviews by J. Schindlbeck and S. Kutterolf as well as editorial handling by M. Edmonds. The full description of Site U1438 cores can be accessed publicly from http://web.iodp.tamu.edu/LORE/ website. Geochemical analyses of glass shards are accessible in the supporting information data set.