Clockwise rotation of SW Japan and timing of Izanagi–Pacific ridge subduction revealed by arc migration

Abstract

Igneous rocks associated with the Cretaceous to Paleogene volcanic arc in SW Japan show ages that young from west to east in a direction parallel to the Median Tectonic Line suggesting corresponding translation of a heat source traditionally interpreted in terms of oblique subduction of a spreading ridge. However, recent oceanic plate reconstructions suggest ridge subduction may be younger than the main arc activity. Age compilations of 1227 points of felsic to intermediate Cretaceous and Cenozoic igneous rocks from the Japan arc show arc magmatism that can be separated into an early active period 130–60 Ma (stage 1), a subsequent period of quiescence 60–46 Ma (stage 2), which is followed by a resumption of igneous activity from 46 Ma onward (stage 3). In southwest Japan, the orientations of the magmatic arcs of stages 1 and 3 show and angular discordance of about 20°. The lack of active arc magmatism and the occurrence patterns of adakitic and high-Mg andesitic magmas indicate that ridge subduction occurred during stage 2. The arc age distribution pattern of stage 1 is explained by the slab shallowing related to a younging of the subducting slab as the ridge approaches. Furthermore, the obliquity of the arcs formed at stages 1 and 3 is explained by a 20° clockwise rotation of the inner zone of southwest Japan during the ridge-subduction phase. Oceanic plate reconstructions show counterclockwise rotation in the subduction direction after the ridge subduction phase, and coupling of the subducting oceanic plate with the upper plate would support microplate rotation in the inner zone. The new proposed tectonic reconstructions provide a framework to related Paleogene subduction of an active spreading ridge along the east Asia margin not only to the distribution of granitic bodies but also to rift-related basin formation on the eastern margin of the Eurasian continent and to rotation of crustal blocks indicated by paleomagnetic data of Cretaceous terranes.

1 Introduction

The approach and subduction of active spreading ridges is a rare but inherent part of the plate tectonic cycle. Ridge subduction is associated with anomalously warm subduction zones and has been called on to account for a broad range of events in the geological record, such as changes in magmatism in the overriding plate, changes in plate movement vectors, and the onset of phases of orogenesis including exhumation of high-pressure metamorphic rocks (e.g., Sisson et al. 2003a). On the modern Earth, ridge-trench-trench (RTT) triple junctions have been identified around Chile and the Woodlark Basin, and these are two valuable field areas of study that provide real-time data on the characteristic of igneous activity and crustal deformation in domains of active ridge subduction (Taylor and Exon 1987; Stern 2004; Ramos 2005). The reaction of the hanging wall to ridge approach, collision, and subduction over geological time scales, can be examined by studying the geologic records of such convergent margins (Bradley et al. 2003).

The Japanese Islands record evidence of active subduction throughout much of their geological history, and subduction of a spreading ridge has been highlighted as a potentially important process since the early days of plate tectonics (Uyeda and Miyashiro 1974). The large number of geological data sets and the lack of any major continental collision make the Japanese islands well-suited to studying possible interactions between spreading ridges and the eastern margin of the Eurasian continent since the Cretaceous (Fig. 1; Taira et al. 2016). Cretaceous terranes comprise a major part of the basement rocks of the Japanese Islands and consist of a set of granitoids with pairs of high-P/T and high-T/P type metamorphic rocks (Figs. 2, 3), and it has long been suggested that their formation is closely related to the approach and subduction of a spreading ridge (Uyeda and Miyashiro 1974; Takahashi 1983; Maruyama and Seno 1986; Kinoshita and Ito 1986, 1988; Brown 1998; Iwamori 2000; Uehara and Aoya 2005; Aoya et al. 2003, 2009; Isozaki et al. 2010). These ideas are in agreement with many of the earliest oceanic plate reconstructions for the Pacific realm, which proposed a spreading ridge between the Pacific and Izanagi or Kula plates was subducted at a high angle to the trench in the upper Cretaceous (e.g., Woods and Davies 1982; Engerbreston et al. 1985; Maruyama et al. 1997). However, the uncertainties in these reconstructions are large and recent oceanic plate reconstruction models that incorporate both newly analyzed ocean floor magnetic anomaly data and tomographic data constraining the amount of subducted plate—information that was not available in the 1980s—indicate that the spreading ridge (Izanagi–Pacific ridge) was oriented roughly parallel to the trench of the East Asian continental margin and subducted underneath the paleo-Japan arc in Paleogene (Fig. 1; Whittaker et al. 2007; Seton et al. 2015; Wu et al. 2022a, b). However, only a limited number of the available onshore geological data have been used to examine the predictions of this model and examine its implications for the geological evolution of the Japanese Islands.

Fig. 1

The topography map of the eastern margin of Eurasia continent (data from Amante and Eakins 2009). The approximate paleo-location of the Japanese islands before the formation of the Sea of Japan in Miocene is shown in gray. The paleo-magmatic arcs (> 60 Ma) of NE and SW Japan are adjusted to be parallel when the southwest Japan is rotated 20 degrees counterclockwise. Several models have been proposed for estimating paleo-positional relationships (e.g., Otofuji et al. 1985; Jolivet et al. 1995; Yamakita and Otoh 2000). The selected basins record rifting during the early Cenozoic period of interest (Yi et al. 2003; Imaoka et al. 2011; Gu et al. 2017; Song et al. 2018; Zhu et al. 2020; Liu et al. 2022). The hatched area represents the estimated Izanagi–Pacific Slab gap based on P-wave tomographic observations and is recognized at an approximate depth of around 1000 km (Li et al. 2008; Wu et al. 2022a, b). White stars indicate the presence of a geologic record that indicates a clockwise rotation of approximately 20 degrees with respect to Eurasia continent in Late Cretaceous to Early Eocene (Otofuji et al. 1985; Uchimura et al. 1996; Fukuma et al. 2003; Lin et al. 2003; Park et al. 2005; Wang et al. 2011)

Fig. 2

Distribution of geological terranes in NE and SW Japan (data from Geological Survey of Japan, AIST 2022). NE and SW Japan are divided by the Tanakura Tectonic Line (TTL). In this study, we interpret the Median Tectonic Line (MTL) as a major fault that defines the northern limit of occurrence of geologic terranes in the outer zone (Sanbagawa, Chichibu, and Shimanto belts) and is essentially identical to the Usuki–Yatsushiro Tectonic Line (UYTL)

Fig. 3

Distribution of Cretaceous–Paleogene igneous rocks in NE and SW Japan (data from Geological Survey of Japan, AIST 2022)

In this contribution, we first present a detailed compilation of age data and a review of available geochemical data for contemporaneous magmatic rocks to examine proposed ideas for the timing and orientation of ridge subduction in the paleo-Japan Arc. The dataset we have compiled includes information only available in papers written in Japanese (Additional files 1, 2: Table S1), and we hope this contribution will also serve as a gateway to make these valuable data more widely available.

2 Geological setting

The Cretaceous–Paleogene geological terranes of the Japan can be divided into NE Japan and SW Japan, separated by the Tanakura Tectonic Line (TTL) (Fig. 2). Throughout this area, plutons are dominantly felsic to intermediate and mafic magmatism is minor (Fig. 3).

In central part of NE Japan, Cretaceous magmatic intrusions into pre-Cretaceous basement rocks are recognized in the Kitakami and Abukuma areas. Igneous activity along the inner arc part of Japan adjacent to the Japan Sea coast is broadly similar to that in the Kitakami–Abukuma areas, although knowledge of the basement rock distribution is limited by the presence of Neogene and later cover. On the Pacific coast side, information on the Cretaceous forearc is lacking due to progressive tectonic erosion by the Pacific plate subduction (von Huene and Scholl 1991). Cretaceous to Paleogene volcanic rocks are also present in parts of the Kitakami area (Fig. 3).

SW Japan is divided into an inner zone north of the Median Tectonic Line (MTL) (inner part of the arc) and an outer zone to the south (Fig. 2) (outer part of the arc). In Kyushu, the Usuki–Yatsushiro Tectonic Line (UYTL) is recognized as the western extension of the MTL. In the inner zone, Cretaceous-Paleogene felsic magma bodies intrude pre-Cretaceous basement rocks. Associated volcanism is recognized over a widespread area in SW Japan. The age range of high-T/P metamorphism associated with synchronous igneous activity is ~ 120–110 Ma for the southern Abukuma and Higo metamorphic rocks (Hiroi et al. 1998; Sakashima et al. 2003), and ~ 100–80 Ma for the Ryoke metamorphic rocks (e.g., Suzuki et al. 1994a, b; Takatsuka et al. 2017; Kawakami et al. 2022).

3 Post-Cretaceous spreading ridge subduction around the Japanese Islands depicted in previous studies: an overview

3.1 Highly oblique subduction of the spreading ridge at Cretaceous

The possibility of a correlation between active magmatism and high-T/P type metamorphism in the Cretaceous Japan arc and spreading ridge subduction was first addressed by Uyeda and Miyashiro (1974), and subsequently, numerical simulations of the thermal structure of subduction zones have examined possible relationships between metamorphic P–T conditions of the Ryoke metamorphic rocks in SW Japan and associated melting of the lower crust due to a large heat source related to subduction of a spreading ridge and very young oceanic lithosphere (e.g., Iwamori 2000; Okudaira and Yoshitake 2004; Aoya et al. 2009; Okudaira and Suda 2011). Thermal modelling of the high-P/T Cretaceous metamorphic rocks of SW Japan (Sanbagawa metamorphic rocks) showed the recorded metamorphic P–T conditions are a good match for the thermal models shortly before ridge subduction (Aoya et al. 2003; Uehara and Aoya 2005). However, it is important to note that approach and subduction of a spreading ridge subduction is a sufficient but not necessary conditions to explain the moderately warm to warm environments in the geologic record. A more recent study reveals that consideration of the effect of shear heat generated at subduction boundaries is a viable alternative explanation for the recorded P–T conditions of the Sanbagawa metamorphic rocks that does not require ridge subduction (Ishii and Wallis 2020).

Regional spatiotemporal information such as the occurrence, spatial movement with time, and cessation of igneous activities with a particular chemical composition may offer stronger constraints on tectonic reconstructions (Bradley et al. 1993, 2003) than is possible from thermal models alone. In SW Japan, it is recognized that there is a systematic E–W trend of younger-aged I-type granitoids parallel to the trend of the Median Tectonic Line (MTL) from ~ 115 Ma in the west to ~ 65 Ma in the east, and it has been argued that this trend in geologic features can be directly related to the subduction of a spreading ridge (e.g., Kinoshita and Ito 1986; Nakajima et al. 1990). This model assumes that a spreading ridge, which subducts at a high angle, i.e., highly oblique to the volcanic arc, serves as a heat source for magma generation. The granitoids of SW Japan also show a younging trend in a direction subperpendicular to the MTL with younger-aged units present in areas further north from the MTL than the older aged units. In NE Japan, where granitoids are present with slightly older ages than SW Japan (~ 130–95 Ma), a similar origin to SW Japan has been sought for the origin of the magmatism calling on subduction of a young slab and subsequent approach and subduction of an active spreading ridge. The main line of evidence is the adakitic chemical composition of magma (e.g., Osozawa et al. 2019). However, other tectonic settings have also been highlighted as compatible with the characteristics of the magmatism in NE Japan: slab rollback or slab rupture (Tsuchiya et al. 2015).

Oblique subduction ridge models for the Cretaceous-Paleogene history of Japan face two significant problems when considering the regional distribution of igneous activity and the associated ages. (i) Migration of a RTT triple junction from west to east along the length of the Japanese archipelago requires large-scale left lateral movements with a displacement that is equal to or greater than the extent of SW Japan after the magmatism to explain the presence of older granite bodies in NE Japan (Sakashima et al. 2003; Osozawa et al. 2019). Detrital zircon geochronology does not support the idea that the geologic terranes of NE Japan and the outer zone of SW Japan migrated from positions far to the south of their present locations because zircon age spectra show a strong connection to the North China Craton, implying that this was the provenance region and the two areas were not geographically isolated (e.g., Tokiwa et al. 2019; Li and Takeuchi 2021). (ii) The proposed ridge subduction models directly relate increased magmatism with the passage of a subducting spreading ridge. However, this view contradicts the observed cessation of magmatic activity around modern-day ridge subduction domains. The most likely explanation for this reduction in magmatism despite the expected higher temperatures is the lack of water supplied by an actively subducting slab due to the formation of a slab window (Thorkelson 1996). Aoya et al. (2009) suggest that dehydration of serpentinized mantle wedges at elevated temperatures may provide the water necessary to cause widespread melting, but this would imply Cretaceous magmatism in the forearc region.

Differential exhumation has been proposed as alternative model to explain the east–west trends in granite ages seen in a direction measured parallel to the MTL in SW Japan based on the monazite chronometry of granite bodies in the southern margin of the inner zone (Suzuki and Adachi 1998). However, both volcanic and plutonic rocks show the same overall younging to the east (e.g., Nakajima et al. 1990; Sato et al. 2016). The preservation of volcanic rocks that were formed at the earth’s surface shows that the effects of erosion are limited and suggests that changes in the locus of magmatic activity are more important than any differential exhumation.

Intrusion of mid-ocean ridge basalt (MORB) magma into forearc regions has commonly been highlighted as evidence for interaction between an active spreading ridge and an oceanic trench (Hibbard and Karig 1990; Kaeding et al. 1990; Sisson et al. 2003b). In SW Japan, Kiminami et al. (1994) report a set of in situ MORB intrusions in the Shimanto accretionary complex with ages that young from west to east that can be chronologically compared to trends in the inner zone igneous rocks. However, the results of subsequent field work suggest the lithological boundaries presented in this paper may be structural rather than intrusive (Onishi and Kimura 1995).

3.2 Subparallel subduction of the spreading ridge at early Paleogene

Recently proposed oceanic plate motion models make use of both an expanded data set of magnetic anomalies in modern day ocean domains and the information on the volume of subducted slabs present within the mantle from seismic tomography. The resulting revised plate reconstructions show roughly parallel subduction of a spreading ridge under the East Asia continental margin during the Paleogene (Whittaker et al. 2007; Müller et al. 2008; Facccenna et al. 2012; Seton et al. 2015; Matthews et al. 2016; Wu et al. 2022a, b). The two types of reconstruction suggest contrasting times for the ridge–trench interaction and geometrical relationships with the arc-trench system, and the Cretaceous–Paleogene onshore geological record of the Japanese Islands has great potential to contribute to testing the viability of the two contrasting types of tectonic model.

Spreading ridge subduction potentially causes extensive geomorphic changes and thermal anomalies in the forearc region associated with the approach and subduction of a warm and buoyant domain, and the corresponding geologic features recorded in accretionary complexes from Sakhalin to southwest Japan may be interpreted within the framework of this model (Wu and Wu 2019; Kimura et al. 2019). The cessation of continental arc magmatism recognized over a wide region that occurred at around 50 Ma, and the abrupt changes in Sr and Nd isotopic compositions suggesting a depleted mantle influx recognized immediately after the period of magmatic quiescence can be interpreted as signatures of the existence of the slab window associated with the spreading ridge subduction (Wu and Wu 2019; Liu et al. 2020). These proposed models are consistent with observations of modern examples of spreading ridge subduction, including the cessation of arc igneous activity. However, by itself this model does not account for the trend in the ages of igneous activity in SW Japan seen over a distance of ~ 1000 km E–W. In the following section, we present a compilation of ages of the igneous activity that sheds more light on the timing of ridge subduction since the Cretaceous and use it to propose a tectonic framework for the basement units of the Japanese Islands.

4 Method: compilation of ages of igneous activity

4.1 Radiometric ages

We collected 1227 data from the available literature that report the timing and location of Cretaceous to Paleogene igneous activity in the Japanese islands (Table 1). The age range analyzed is from 130 Ma, when Cretaceous igneous activity begins (Tsuchiya et al. 2015; Osozawa et al. 2019; Harada et al. 2023), to 30 Ma, shortly before the opening of the Sea of Japan event in the Miocene (Otofuji et al. 1985). This age range covers the time period relevant to the proposed ridge subduction models described above. For the compilation, we focused on intermediate to felsic intrusive and extrusive rocks (Tables 1, Additional files 1, 2: Table S1). Mafic rocks were not included. The ages in this study are the averaged ages of multiple analyses from a single rock sample rather than individual measurements. Reliable crystallization ages for igneous rocks are provided by zircon U–Pb ages from a number of publications. This method uses zircon, which is resistant to hydrothermal alteration and weathering, has a high closure temperature and is therefore resistant to resetting by later thermal events. In addition, two separate radiometric decay sequences can be used for the same analysis to verify the age. We also include U–Th–total Pb ages for uraninite (UO2) and thorite (ThSiO₄). The associated closure temperatures for both minerals may be somewhat lower than zircon (Yokoyama et al. 2010, 2016), but the results are compatible with the zircon U–Pb ages and we consider they can be regarded as crystallization ages and make a useful contribution to the tectonic discussion of this study (Fig. 4A). Rb–Sr whole rock or mineral ages and monazite U–Th–total Pb ages have been given prominent treatment in former studies. However, these are commonly in marked disagreement with the age values obtained by the above methods (Shibata and Ishihara 1979; Skrzypek et al. 2018, 2020) and were excluded from the compilation to avoid complicating the discussion. We also compiled K–Ar and Ar–Ar age values (whole rock and mineral ages). These ages are interpreted as cooling ages rather than crystallization ages. It is possible that some of the results were modified by later thermal events. However, when K–Ar ages are compared to zircon U–Pb ages from the same samples or from samples taken in very close proximity, there is a good agreement in the observed trends with the K–Ar ages generally showing ages ~ 10–15 Myr younger than the zircon ages (Fig. 4B). We conclude that the K–Ar age data set is a useful addition to the geochronological record used in this study. Due to the limited number of results, it was more difficult to compare Ar–Ar ages to the U–Pb ages. However, we also included these ages because they show good consistency with neighboring K–Ar ages. In our compilation of cooling ages, we exclude data from samples where significant alteration of the measured minerals are described or where the original research considers them to have undergone secondary alteration. Filtering of data as described above leaves a data set consisting of 697 crystallization ages and 530 cooling ages (Fig. 5, Additional files 1, 2: Table S1).

Table 1 List of compiled chronological data counts by method

Fig. 4

Comparison of uraninite and thorite U–Th–total Pb ages (Yokoyama et al. 2016). A and K–Ar age values, B against zircon U–Pb ages based on the database of Table 1. Each point is a comparison between samples from the identical or proximal area (within a few kilometers) and expected to have the similar rock-forming age. Error bars are shown as 1σ

Fig. 5

Spatial distribution of chronological data and associated dating method

4.2 Adakitic magma and high-Mg andesitic magma

In our literature review, we also recorded reports of samples with geochemical data showing adakite and high-Mg andesitic magma (HMA) compositions that occur within the spatiotemporal domain covered by our compilation (Additional files 3, 4: Table S2). Such magmas can form as a result of the subduction of an oceanic plate into relatively warm mantle regions. Adakites mainly exhibit high-Sr/Y and high-La/Yb signatures which are most likely controlled by high-pressure melting in the presence of the garnet or amphibole and the absence of plagioclase (Castillo 2012). These features of adakites are generally interpreted to indicate a slab melting origin (e.g., Defant and Drummond 1990). However, the possibility of an origin controlled by intracrustal differentiation processes continues to be debated (e.g., Yada and Owada 2003; Kamei et al. 2004; Takahashi et al. 2005; Takahashi 2008; Tsuchiya 2008; Chiaradia 2015; Chapman et al. 2015). HMA magmatism is thought to be associated with partial melting of mantle under relatively low-pressure conditions with hydrated conditions (Tatsumi 1982, 1995). Occurrences of both adakitic and HMA magma compositions are rare in normal arc magmatism and are generally interpreted as features characteristic of the subduction of young slabs and magmatism during the onset of subduction.

5 Compilation results

5.1 Radiometric ages distribution

5.1.1 NE Japan

In NE Japan, there is a concentration of data on the eastern side corresponding to the Kitakami–Abukuma domain. Basement rocks of the western part of NE Japan are only locally exposed beneath extensive post-Miocene to Quaternary overlying strata on the Sea of Japan side (Fig. 6). The oldest activity is ~ 130 Ma east of the Kitakami area, with ages showing a good correlation with longitude, and an overall westward-younging trend up to ~ 100 Ma and an average migration rate for NE Japan can be calculated as ~ 130 km per 20 Myr (~ 6.5 km/Myr) (Fig. 7). Slightly younger ages down to ~ 95 Ma are also observed in the Abukuma area, but the overall trend is consistent with the ages of the Kitakami area. Younger igneous activity is also recorded at Jodogahama, but has limited distribution (e.g., Tsuchiya et al. 2005).

Fig. 6

Spatial distribution of compiled age values. Specific colors are assigned according to age value. Data sources are shown in Additional files 1, 2: Table S1

Fig. 7

Spatiotemporal variation of arc igneous activity in NE Japan, with longitude on the horizontal axis because the largest age gradient in NE Japan is in the east–west direction (e.g., Tsuchiya et al. 2015; Yokoyama et al. 2016). Colored circles indicate crystallization ages, and colors correspond to latitude. White circles indicate cooling ages. Note that the east–west direction on the horizontal axis is reversed compared to the map to facilitate a direct comparison with SW Japan in Fig. 14

5.1.2 SW Japan

West of the ISTL, there is a notable difference in the spatial configurations of igneous activity before ~ 60 Ma and after ~ 46 Ma. Major igneous activity prior to 60 Ma is 115–100 Ma in Kyushu, 100–90 Ma in the Shikoku, and 80–70 Ma in the Chubu area, and shows an overall trend of younging to the east when locations with similar distances measured perpendicular to the MTL are compared. There is also gradual younging trend to the north with increasing distance from the MTL (Figs. 6, 8A). These age distributions indicate that temporal variations in the age of igneous activity cannot be adequately expressed solely as a function of distance from the MTL (Fig. 8A). A small number of igneous rocks showing ages of 110–90 Ma are also recognized in the Chubu and Kinki areas, independent of the overall trend of younger age at greater distances from the MTL (Fig. 8A). More careful tracing of the migration of the locus of igneous activity with time shows that the isochronous lines are oriented roughly parallel to the modern lines of latitude in a direction ~ 15–20 degrees oblique to the MTL (Figs. 6, 8B, Additional files 1, 2: Table S1). When plotted in a space with horizontal axes of latitude and longitude and a vertical axis of age, the geochronological data define a clear planer surface (Figs. 8B, Additional files 1, 2: Table S1). This surface shows a horizontal migration of 300 km migration of the arc associated with a change in age from 110 to 60 Ma. The average migration rate of the arc calculated from this observation is ~ 6–7 km/Myr, which is in good agreement with the rates of migration recorded in NE Japan. On the continental side of the arc, there is vigorous igneous activity 115–90 Ma, which is followed by a quieter period 90–70 Ma and a phase of moderate activity 70–60 Ma. Thus, the region north of ~ 35°N records two peaks of igneous activity at 110–90 Ma and 70–60 Ma. The spatiotemporal trends of the > 60 Ma igneous activity of SW Japan do not extend southward beyond the MTL. In the area between the TTL and ISTL (identified as part of SW Japan in Fig. 2), a less distinct spatial change is recognized, although the area does show a similar bimodal age distribution with activity peaks at 110–90 Ma and 75–60 Ma (Figs. 6, 7C) similar to the region north of ~ 35°N in SW Japan. Most of the areas around the TTL on the NE Japan side show ages in the range 105–95 Ma, and thus, the TTL forms the border between the age distribution patterns. As shown in Fig. 6, the two major tectonic lines, the MTL and TTL, crosscut the trends of the > 60 Ma igneous activity.

Fig. 8

Spatiotemporal variation of arc igneous activity in SW Japan west of ISTL, with distance from the MTL (A) and latitude (B) on the horizontal axis. Colored circles indicate crystallization ages. The colors correspond to longitude. White circles in (B) indicate cooling ages. The age distribution in the region between the ISTL and TTL is shown in (C)

Magmatic crystallization ages between 60 and 46 Ma are almost completely absent from the rock record. The only exception is the Daito granodiorite, which shows a zircon U–Pb age of 57 Ma (Ishihara and Tani 2013). Other ages from igneous bodies that fall in this age range are all cooling ages (Fig. 8). After 46 Ma, the site of igneous activity changes drastically and becomes concentrated along the Japan Sea coast. The < 46 Ma arc front, which also takes into account the distribution of igneous rocks from 30–20 Ma, is aligned parallel to the MTL and outer zone zonal structures, and no subsequent large-scale migration is recognized. This arc front is roughly consistent with the southern boundary of the San-in Belt in the traditional classification of granitic zones (e.g., Murakami 1974).

5.2 Adakitic magma and high-Mg andesitic magma

The compiled information on adakite and HMA localities shows a variety of ages and spatial distributions (Figs. 9, 10). In many areas, adakite and HMA show a similar spatial distribution. Based on Sr and Y contents, the ages of both magma types can be separated into three distinct stages (Fig. 10). Data for La and Yb contents in the 100–30 Ma range were not available in sufficient numbers for our discussion, but there is a good correlation between high La/Yb ratios and high Sr/Y ratios for the samples with ages in the 130–100 Ma range (Additional file 3: Table S3). The first adakite- and HMA-related magmatism is observed from NE Japan to Kyushu in the period 130–100 Ma. The second is observed in the period 70–60 Ma in the Chubu region and adjacent to the TTL in SW Japan. The third is observed in the period 45–30 Ma in the Chugoku region, the Noto peninsula, the Oga peninsula, and the Jodogahama area. A comparison of these results with the overall age distribution shows the first stage corresponds to extensive igneous activity that extended over a wide part of the continental margin, whereas the second and third stages correspond to periods immediately before and after the period of magmatic quiescence.

Fig. 9

Spatial distribution of adakite and high-Mg andesitic magmas. Both types of magma are represented by different symbols for each active stage. Data sources are shown in Additional files 3, 4: Table S2

Fig. 10

A Adakite discrimination diagram using Sr and Y content in felsic magma during the period 130–30 Ma (Defant and Drummond 1990). Rocks showing SiO₂ < 56 wt%, Al₂O₃ < 14.5 wt% and Na₂O < 3 wt% are not included in the plot to find Sr and Y features in the range of major elements typical of adakite (e.g., Castillo 2012). ADR = andesite–dacite–rhyolite. B Re-plotted data of (A) with age as the horizontal axis. High-Sr/Y ratios signify adakite activity. The data sources are shown in Additional file 3: Table S3

6 Discussion: possible tectonic scenarios

In this section, we use our compilation of data related to the timing, location and type of igneous activity to propose a tectonic history for the Japan arc. We divide this history into four distinct stages: (i) subduction initiation of the Izanagi plate, (ii) subducting slab shallowing, (iii) ridge subduction and clockwise rotation of SW Japan, (iv) subduction of the Pacific plate (Fig. 11).

Fig. 11

Conceptual tectonic model for SW Japan that can explain the spatiotemporal evolution of arc magmatism. The onset of subduction of the Izanagi plate (IZA) beneath the eastern margin of Eurasia (A), the migration of the arc front to the continental side accompanied by slab shallowing due to the approach of the spreading ridge (B–D), cessation of the arc magmatism accompanied by ridge subduction and formation of the slab window (E), and the onset of subduction of the Pacific plate (PAC) after spreading ridge subduction (F) are shown. The remnants of the arc formed during (A–D) are oriented at an angle of about 20 degrees to the present arc orientation shown in (F). This rotation suggests microplate rotation associated with the IZA–PAC ridge subduction

6.1 Onset of extensive magmatism: subduction initiation of the Izanagi plate

The earliest igneous activity of the Cretaceous is recognized in NE Japan at ~ 130 Ma, and this magmatism is accompanied by significant adakite and calc-alkaline magma production (Tsuchiya et al. 2015; Osozawa et al. 2019) (Fig. 12). Compilations of detrital zircon U–Pb ages from the Shimanto and Sanbagawa belts show a lack of igneous zircons in the 160–130 Ma range, suggesting cessation of igneous activity in the paleo-Japan realm (Fig. 13). The presence of this group of igneous bodies and the age patterns of igneous detrital zircons indicate that igneous activity in Cretaceous Japan initiated with an intense phase at 130 Ma marking a major shift from a lull during 160–130 Ma. The significance of the period in the geological development of east Asia is shown by the similar magmatic histories observed in the Korean Peninsula (e.g., Sagong et al. 2005) and the Sikhote-Alin area (e.g., Wu et al. 2017).

Fig. 12

Schematic illustration of magma arc configuration changes in SW and NE Japan for the periods 130–60 Ma (A) and 46–30 Ma (B) and the present (C) in map view. The colored bars indicate the general locations of arc fronts. Subduction of the Izanagi-Pacific Ridge between 60 and 46 Ma produced a magmatic hiatus, and the change in convergence vector associated with the change in the subducting plate induced a 20° clockwise rotation (CW) of the inner zone of SW Japan and possibly the Korean peninsula. B The blue dotted line indicates that the eastern part of the paleo-magmatic arc has been eroded to a greater extent in the east in association with MTL activity during the period 60–46 Ma. For simplicity, the TTL–ISTL regions where the IBM arc collided with the other Japanese arcs and the location of the magmatic arc after 30 Ma are not shown in (C)

Fig. 13

Probability density plot based on a complication of U–Pb ages from detrital zircons in the Sanbagawa and Northern Shimanto belts in representative literatures. The source data do not include U–Pb ages for zircon rims grown by metamorphism, all of which can be interpreted as igneous ages. The bars show the U–Pb age for each analysis. Sanbagawa: Okamoto et al. (2004), Aoki et al. (2007, 2012, 2009), Tsutsumi et al. (2009, 2012), Knittel et al. (2014), Endo et al. (2018), Jia and Takeuchi (2020), Shimanto: Aoki et al. (2012, 2015), Tokiwa et al. (2019, 2017, 2021), Shimura et al. (2017, 2020, 2021) and Jia and Takeuchi (2020)

We interpret the onset of this igneous activity as marking the onset of active subduction of a former plate generally identified as the Izanagi Plate (Woods and Davies 1982). Initiation of subduction of the Izanagi plate has been linked to both the termination of a phase of flat slab subduction of a marginal oceanic plate, and subduction at a former transform boundary due to a change in relative plate motion (Khanchuk et al. 2016; Grebennikov et al. 2016; Wu et al. 2022a, b).

The onset of subduction is expected to be associated with the formation of a hot and juvenile mantle upwelling due to the presence of a slab gap and/or the development of a return flow in the mantle as the slab is subducted (Wu et al. 2022a; b). The formation of adakites and HMAs in the period 130–95 Ma can be interpreted as magma generated as a result of the interaction of the subducting slab with this hot mantle (Fig. 11A, B). Furthermore, such high-temperature mantle influx can induce long-term and widespread mantle-tectonic thermal anomalies (Fig. 11B). Thus, it can explain the extended igneous activity on the continental side of SW Japan that continues up to 90 Ma. Evidence for widespread igneous activity from 120 to 90 Ma can be traced to the Korean Peninsula (e.g., Kim et al. 2012; Zhang et al. 2021).

Further evidence for subduction initiation at this time is also given by studies of metamorphic rocks. A counterclockwise P–T path showing isobaric cooling documented in higher-grade units of the subduction-type Sanbagawa metamorphic belt associated with a peak metamorphic age of 126–116 Ma has been interpreted by Endo (2010) and Endo et al. (2012) in terms of subduction initiation. The presence of highly depleted ultramafic rocks of mantle wedge origin in the same unit showing high degrees of influx partial melting is compatible with high-T conditions and active mantle flow associated with the early stages of the Sanbagawa metamorphism (Mizukami and Wallis 2005; Hattori et al. 2010).

The common occurrence of thick pelagic charts accompanying units showing accretion ages around the Albian–Aptian in the Shimanto belt (e.g., Ota et al. 2019) and the relationship between monotonous landward arc migration and younger plate ages as shown in the next section, suggest that the subducting oceanic plate subducting beneath Japan during the mid-Cretaceous was not young.

6.2 Arc migration: subducting slab shallowing 130–60 Ma

In NE and SW Japan, there is evidence of arc migration at similar rates of 6–7 km/Myr. The younging trends of currently observed paleo-magma arcs formed in the period > 60 Ma show a nearly orthogonal relationship between NE and SW Japan. However, reversing the NE Japan and SW Japan rotation vectors related to the Miocene Japan Sea opening, as estimated on the basis of paleomagnetic studies (e.g., Otofuji et al. 1985; Hoshi 2018), shows the younging trends in both NE Japan and SW Japan are oriented roughly parallel to the former continental margin (Fig. 12). These similarities allow us to interpret the present NE Japan and SW Japan age distributions as different parts of the same former magmatic arc (Fig. 12A). The absence of a rock record of the 130–120 Ma arc in the western part of SW Japan can be explained by excision due to movements on the MTL and erosion. These ideas are compatible with proposals for the former existence of a ‘paleo-Ryoke’ domain, which is originally located to the south of the present Ryoke belt prior to MTL activity (Fig. 12B, Ichikawa 1990; Ono 2002; Takagi and Shibata 2000; Takagi and Arai 2003). In this study, we estimate that the magmatic arc migrated away from the trench over a distance of ~ 400 km (Fig. 14). Such large-scale continentward migration of the arc is best explained by shallowing of the dip of the subducting slab (Fig. 11B, C; Gianni and Pérez Luján 2021). As we show in the following section, such shallowing is consistent with increased buoyancy of the subducting slab as its age decreases until the ridge reaches the subduction zone between 60 and 46 Ma.

Fig. 14

Summary of arc magmatism 130–30 Ma. The white circles in the left show the crystallization ages in NE Japan as in Fig. 7. The black circles show the crystallization ages in SW Japan as in Fig. 8B, C. The age ranges of occurrences of adakite and high-Mg andesitic (HMA) magma are shown on the right side. Metamorphic ages based on zircon U–Pb chronology for the high-T/P type metamorphism are shown in purple. 1: Abukuma metamorphism (Hiroi et al. 1998). 2: Higo metamorphism (Sakashima et al. 2003; Maki et al. 2014; Tsutsumi 2020; Kawaguchi et al. 2020). 3–6: Ryoke metamorphism (Skrzypek et al. 2018; Takatsuka et al. 2018; Kawakami et al. 2019, 2022; Miyazaki et al. 2022)

The approach or subduction of an active spreading ridge has commonly been called on to explain the P–T conditions of the Ryoke metamorphic rocks (e.g., Brown 1998; Iwamori 2000). However, the spatiotemporal configuration of high-T metamorphism including that in the Abukuma, and Higo regions is in good agreement with the major younging trends we found in the distribution of paleo-magma arcs, indicating that high-T metamorphism occurred in association with major arc magmatism for at least 120–80 Ma (Fig. 14). The temporal and spatial proximity of the high-T metamorphism to arc magmatism indicates that the high-T metamorphism has moved continentward with time together with the main magma arc trend and that these P–T conditions are best explained by the subduction of young oceanic lithosphere at some distance from the spreading ridge. Since the age and subduction angle of the subducting slab may have changed with time, the slightly older Abukuma–Higo metamorphism (~ 120–110 Ma) is likely associated with older slab ages and deeper subduction angles, and more recent Ryoke metamorphism (~ 110–80 Ma) with younger and shallower subduction.

6.3 Magmatic hiatus: ridge subduction at 60–46 Ma

The 60–46 Ma magmatic hiatus in southwest Japan recognized in this study can be correlated with that of Wu and Wu (2019). However, since igneous activity several million years younger than 60 Ma is reported not only from the Daito granodiorite but also the Korean Peninsula and detrital zircon chronology (e.g., Tokiwa et al. 2016; Cheong and Jo 2017; Nakano et al. 2021), it is possible that more data on igneous ages in SW Japan will show the start of this magmatic hiatus is somewhat younger.

As discussed by Wu and Wu (2019), the 60–46 Ma hiatus in igneous activity is best explained by the arrival of the Izanagi–Pacific spreading ridge at a low angle to the trench. This hypothesis implies that the periods just before and after the magmatic hiatus should be associated with young slab subduction. Adakite and HMA igneous activity is commonly interpreted as associated with subduction of young oceanic lithosphere and the presence of these types of magmatism immediately before and after the igneous hiatus supports the idea of ridge subduction during the 60–46 Ma igneous hiatus (Figs. 9, 12).

In more general terms, the 70–60 Ma period represents an igneous flare up stage with widespread and voluminous igneous activity accompanied by very large ignimbrite forming events typified by the Nohi-rhyolite deposits (Koido 1991; Yamada and Koido 2005; Sonehara and Harayama 2007). Likewise, igneous activity after 46 Ma is also characterized by high activity with numerous caldera formations (e.g., Imaoka et al. 2011). These stages associated with common caldera-formation both before and after ridge subduction may be related to high crustal temperatures due to subduction of younger slabs and the presence of tensile stress fields.

6.4 Clockwise rotation of SW Japan during the ridge subduction time

The > 1000-point igneous age compilation allows the recognition of two linear non-parallel arrays of former arcs in the SW Japan basement (Fig. 6). The linear nature of the arrays shows there has been minimal deformation within the separate basement domains and the main effect of crustal movements has been to cause rotations about vertical axes. The ages of the arc arrays and their obliquity allow us to estimate the timing and of this rotation of the different basement domains or blocks. The locations of the arc arrays can also be used as marker horizons to estimate displacement of the blocks.

We interpret the observed ~ 20° obliquity between the 46–30 Ma arc and the 130–60 Ma arc as representing a clockwise rotation of the inner zone of SW Japan between 60 and 46 Ma (Fig. 11). Since the overall age trend of igneous activity before 60 Ma in the Chubu region and ISTL–TTL regions is very similar, we consider the ISTL–TTL region acted as a continuous microplate to the west of ISTL region during the period 130–60 Ma. However, the TTL and MTL cross-cut the continuous 130–60 Ma arc sequence seen in SW Japan, indicating that both these tectonic lines are major boundaries between two distinct microplates and their cross-cutting relationships developed after 60 Ma. The age gap along the TTL is recognized between the ~ 105 Ma arc in NE Japan and magmatic activity on continent side of the 65–60 Ma arc in SW Japan. This gap indicates that there is at least a 200 km arc-normal component of sinistral displacement along the TTL relative to NE Japan. Hence, the trend noted in previous studies that the igneous age of the Cretaceous arc front near the MTL is younger to the east can be explained by clockwise rotation of the inner zone of SW Japan relative to the outer zone and NE Japan, and the loss of the eastern part of the arc due to movement along the MTL (Figs. 11, 12). This crustal loss is discussed in Sect. 6.5.

Paleomagnetism is an ideal method to test for the rotation about a vertical axis proposed in this study. Two such studies by Fukuma et al. (2003) and Uno et al. (2021) report a ~ 20 degrees clockwise rotation of the SW Japan basement area between 70 and 20 Ma that occurred prior to rotation associated with the Miocene Japan Sea opening event (Otofuji et al. 1985). The good continuity and parallel nature within each of the 70–60 Ma and 46–20 Ma arcs revealed in our data set imply rotation did not occur during these periods. This is in agreement with the results of Fukuma et al. (2003) and Uno et al. (2021) but potentially narrows the timing of rotation to the period 60–46 Ma.

A similar rotational history has been confirmed at multiple locations around the Korean Peninsula (Fig. 1; Lin et al. 2003; Huang et al. 2007), suggesting that clockwise rotation relative to Eurasia continent may have occurred within a 1000 km-scale microplate that includes SW Japan (Fig. 1; Uno et al. 2021).

A change in the direction of oceanic plate subduction may help explain the trigger for the rotational motion. Recent oceanic plate reconstructions and analysis of deformational structures as kinematic indicator in the Shimanto accretionary complexes suggest there was a significant rapid shift from pre-ridge subduction of the Izanagi plate with a sinistral lateral convergence to a dextral lateral convergence of the Pacific plate after ridge subduction (Whittaker et al. 2007; Byrne and DiTullio 1992; Onishi and Kimura 1995; Raimbourg et al. 2014; Müller et al. 2022). This counterclockwise change in convergence vector changes the overlying plate at the eastern margin of the Asian continent into a dextral transpression zone, which in turn allows clockwise rotation of the inner zone of southwest Japan (Fig. 12).

Ridge subduction and the resulting plate rearrangement have the potential to have a major influence on regional geology. A series of fault basins around the Japan Sea and the East China Sea have been recognized that developed or reactivated during the period we propose corresponds to the phase of the Izanagi–Pacific ridge subduction (Fig. 1). The development of these basin-forming faults may have been influenced by mantle flow driven by subduction of the Izanagi plate or Izanagi–Pacific ridge and the changes in subduction vectors described above (Cao et al. 2018). A multistage faulting system in the Bohai Bay Basin including movements before and after the proposed time of ridge subduction records a clockwise shift of the main extensional trend of basin-forming lifting compatible with our proposed rotation history (Zhu et al. 2020; Yuan et al. 2022) and suggests that the switching of subducting oceanic plates with different convergent vectors can result in widespread deformation of the continental margin.

6.5 Crustal erosion: relationship with the Median Tectonic Line

Reconstructions of Cretaceous–Paleogene arc migration in SW Japan suggest the presence of a stage of large-scale crustal removal along the MTL. The recognition of an arc deficit and the ~ 20 degrees obliquity of the SW Japan arc front to the MTL indicates that shallower crustal erosion occurred associated with block rotation of SW Japan during 60–46 Ma (Fig. 12).

The locations of the paleo-arcs can be used to estimate the amount of material lost perpendicular to the arc. If the parallelism of the paleo-arcs is preserved, the crustal loss can be estimated as the distance to the envisaged 130 Ma arc front. For example, since arc fronts of the 110–100 Ma and 70–60 Ma periods can be observed in the Kyushu and Chubu regions, respectively; the horizontal distances perpendicular to the 130 Ma arc estimated based on arc migration rates (6–7 km/Myr) are 100–200 km and 400–500 km. Note that these estimated amounts of crustal shortening are assumed to be all accounted for by the CW block rotation, but they are minimum estimates because they do not consider the possibility of erosion of the forearc region between the 130 Ma arc and the younger accretionary complexes.

The age of igneous activity constrains the timing of crustal removal associated with block rotation to have occurred 60–46 Ma. The crustal removal during this stage is inferred to have been surface erosion, since the MTL during the relevant period shows large-scale normal fault movement with the foot wall on the outer zone side uplifting (Kobayashi 1995; Fukunari and Wallis 2007; Kubota and Takeshita 2008; Kubota et al. 2020). Large-scale normal fault movement at this time is also supported by the exposure of the Sanbagawa metamorphic belt (Narita et al. 1999; Kusuhashi et al. 2022).

7 Conclusions

A reassessment of the spatiotemporal evolution of felsic to intermediate arc magmatism in NE and SW Japan in the period 130–30 Ma based on a compilation of a 1227-entry database, combined with geochemical considerations and contemporaneous the geological events in Japan and surrounding regions leads to the following conclusions.

• 1. Arc magmatism occurred throughout the period 130–60 Ma in both northeastern and southwestern Japan and during this period underwent a ~ 400 km migration toward the continental side. This migration is best accounted for by a shallowing of the slab subducting angle associated with a slab progressively younging following the onset of subduction of old lithosphere of the Izanagi plate at around 130–120 Ma.

• 2. Cessation of arc magmatism is recognized in the period 60–46 Ma throughout the Japanese arc and is explained by the subduction of the Izanagi–Pacific spreading ridge subparallel to the trench.

• 3. A new volcanic arc formed in the period 46–30 Ma in SW Japan and remained in a similar location throughout this time with no significant arc migration.

• 4. The ~ 20° obliquity between the 46–30 Ma arc and the 130–60 Ma arc suggests a clockwise rotation of the inner zone of SW Japan during the ridge subduction stage. Such a rotation is supported by paleomagnetic data. The proposed rotation is accompanied by dextral transpression tectonics immediately after the Izanagi–Pacific ridge subduction.

• 5. Magmatic arcs developed during the period 130–30 Ma in Japan provide a robust reference frame for analyzing microplate rotational motion and recognizing localized crustal excision within this convergent plate margin domain.