1 INTRODUCTION

Submarine volcanoes include extinct or active volcanic landforms widely distributed in the ocean (Kim and Wessel, 2011), and they are usually classified according to subduction zone volcanoes, oceanic ridge volcanoes, and intraplate volcanoes (Fan et al., 2017; Zhao et al., 2020b). Previous studies have revealed that submarine volcanoes play important roles by, serving as sedimentary barriers (Chen et al., 2014; Yin et al., 2021), as potential sites for mineral resources (Grigg et al., 1987; Du et al., 2017), and as the cause of serious natural disasters (Omira et al., 2016). Since partial melting materials migrate from deep to the surface via submarine volcanoes, which themselves were controlled by plate tectonic evolution and crust-mantle interaction, direct clues to understand geological evolution can be explored (Yan et al., 2015b). Submarine volcanoes' morphology, scale, composition, and distribution provide evidence of regional tectonic evolution, insight into magmatic dynamics (Yan et al., 2006; Sun et al., 2020a; Zhang et al., 2021), and information for resource development (Sun et al., 2020b, 2021a).

The South China Sea (SCS) is one of the largest and deepest marginal seas in the western Pacific Ocean, with a unique geomorphology and numerous intraplate volcanoes (Kim and Wessel, 2011; Sun et al., 2019a, 2020a, b). Extensive studies in the SCS northern margin have provided large amounts of geophysical data and petrological samples, including invaluable insights into the geochemical characteristics (Yan et al., 2014, 2015a), temporal and spatial distribution (Zhang et al., 2016; Song et al., 2017; Zhu et al., 2021), volume and magmatic activity scale (Fan et al., 2017; Zhao et al., 2020b), hydrothermal fluid (Sun et al., 2012; Wang et al., 2018b; Zhao et al., 2021b), and deep conduits (Xia et al., 2018; Fan et al., 2019; Wang et al., 2019) of submarine volcanoes there. Previous studies found that magma intrusions at the continental margin of the SCS mostly occurred in the strongly thinned continental slope and Continent-Ocean Transition (COT) (Fan et al., 2017; Song et al., 2017; Xia et al., 2018; Sun et al., 2019c). In the northwestern margin, volcanic eruptions are broader and younger, and the post-spreading volcanism is stronger than in the northeastern margin (Zhang et al., 2016; Song et al., 2017; Wang et al., 2019; Sun et al., 2021b). These indicate that volcanism in the northern margin of the SCS may be related to a variety of factors, such as continental margin rifting mode (Zhao et al., 2016), lithospheric cooling and subsidence (Song et al., 2017), Neotectonic movements (Zhang et al., 2016), and mantle plume (Fan et al., 2017).

Submarine volcanoes have been found in scattered multi-channel seismic (MCS) data in the southern margin of the SCS (Li et al., 2013; Franke et al., 2014; Chang et al., 2017a; Peng et al., 2019; Luo et al., 2021), and volcanic rocks were obtained by partial petrological sampling, as well (Kudrass et al., 1986; Zhou et al., 2005; Yan et al., 2008). Previous studies have discussed the genetic relationship between volcanism and high-velocity bodies (HVBs) in the lower crust or hyperextended continental crust in the southern margin on the basis of the MCS profile, and gravity and magnetic data (Chang et al., 2017a). However, submarine volcanoes' characteristics, spatial and temporal distribution, and genesis in the southern margin of the SCS remain obscure. Due to the lack of comparison between the shallow structure obtained by MCS and the deep crustal structure found by refraction seismic profile, the effects of Nansha Block's drifting process and deep materials on submarine volcanic eruption are not well understood. Therefore, it is necessary to combine various research data to study the characteristics and tectonic evolution of submarine volcanoes in the southern margin of the SCS.

In addition to the newly acquired reflection seismic data, we collected previous MCS profiles, gravity and magnetic data, drilling and dredging samples in the southern margin. We identified 43 submarine volcanoes in the study area. Our results assessed the approximate range of this Pliocene-Pleistocene volcanism, and revealed the temporal and spatial distribution characteristics of the submarine volcanoes in the southern margin of the SCS. Combined with the crustal structure from adjacent ocean bottom seismometer (OBS) wideangle seismic profiles, the influence of stretching faults, crustal thinning and magma underplating of mantle upwelling on submarine volcanism are further discussed.

2 GEOLOGICAL SETTING

The SCS, situated at the junction of the IndoAustralian, Pacific and Eurasian plates, has long been a key area for earth science research (Sibuet et al., 2016). The north-south continental margins of the SCS are unique areas for studying the rifting process during the transition from the extension of the continental lithosphere to the formation of oceanic crust (Ding et al., 2013), including several micro blocks, such as the Xisha-Zhongsha Blocks, Nansha Block, and Reed Bank (Ding and Li, 2011). The southern margin consists of the Nansha Block and Reed Bank, which are characterized by thinning of the continental crust and by hundreds of islands, reefs, and shoals (Yan and Liu, 2004; Hutchison and Vijayan, 2010). The north-south passive continental margins are generally considered to be conjugated, and the Nansha Block and the Reed Bank in the southern margin drifted southward with seafloor spreading during the Oligocene-Early Miocene (Sibuet et al., 2016).

2.1 History of tectonic evolution

In topography of SCS, the abyssal basins can be divided into the Northwest Sub-basin, Eastern Subbasin, and Southwest Sub-basin. Located between the Xisha-Zhongsha Blocks and the Nansha Block, the Southwest Sub-basin has a water depth slightly greater than 4 km, and a "V" shape that opens to the northeast (Fig. 1). According to the most accepted magnetic anomaly modeling (Briais et al., 1993; Li et al., 2014) and the drilling results from the International Ocean Discovery Program (IODP) (Li et al., 2015b), the seafloor spreading of SCS established at ~33 Ma and terminated at ~15.5 Ma. Recent studies have shown that the spreading of the SCS underwent intermittent southward ridge jumps (Ding et al., 2018), with the direction of spreading changed several times from Early Oligocene to Miocene (Sibuet et al., 2016; Sun et al., 2019c). In the Miocene, the Nansha Block collided with the Northwest Borneo, and almost simultaneously the spreading of the SCS stopped (Clift et al., 2008; Cullen et al., 2010; Hutchison, 2010).

The seafloor spreading of SCS gradually weakened from east to west, and its southwest margin gradually transformed to a thinned crust that had not completely broke up (Luo et al., 2021). The southern margin represents a unique transition from Cenozoic continental rift to seafloor spreading and subsequent collision, preserving crucial information on the evolutionary process and tectonic characteristics of the SCS. Geophysical data and partially dredged samples indicate that the Nansha Block is a hyperthinned continental crust (Schlüter et al., 1996; Zhou et al., 2005; Qiu et al., 2008; Hutchison and Vijayan, 2010; Yan et al., 2010). Though the Paleozoic-Mesozoic basement of the Nansha Block is highly homologous to the neighboring regions of Southeast Asia (Kudrass et al., 1986; Yan and Liu, 2004; Zhou et al., 2005), recent drilling results of Well NK-1 in Nansha Block suggest that the basement in the southern margin was similar to that of the South China Block (Miao et al., 2021).

Average crustal thickness of the southern margin is 15–20 km based on previous reflective, refractive seismic data and gravitational inversion results (Qiu et al., 2011; Ding et al., 2013; Pichot et al., 2014; Gozzard et al., 2019). The average thickness of the oceanic crust in the SW Sub-basin is 5.3 km (Yu et al., 2017), with a relatively flat Moho surface (10–12 km). The sedimentary deposits of the Reed Bank in the eastern part of the southern margin are thin (1–2 km thick), with a crustal thickness of 21–23 km (Niu et al., 2014). The central part of the southern margin, having a crustal thickness of 15–20 km, is characterized by a thinned continental crust (Qiu et al., 2011). In the southern margin's western part, a hyper-stretched continental crust with a thickness of 10–18 km was observed, and depth-dependent extension of the crust occurred (Wei et al., 2020).

The numerous studies of petrology and seismic stratigraphy along the southern margin show the characteristics of the sedimentary layer. The Ocean Drilling Program (ODP) Site 1143 revealed the lithological composition of semi-deep marine calcareous mudstones predominantly from the Late Miocene in the Nansha Block (Wang et al., 2000). Tectonic events in the SCS during the Cenozoic are recorded in sedimentary sequences, forming extensive unconformities (Zhang et al., 2020b). The rift-onset unconformity, breakup unconformity, and collision unconformity, some significant unconformities in the southern margin (Hutchison and Vijayan, 2010; Sun et al., 2011; Franke et al., 2014), temporally corresponds to continental rifting, seafloor spreading, and continental collision events, respectively (Sun et al., 2011).

2.2 Submarine volcanism

In the SCS, Cenozoic magmatism extensive, particularly on continental slopes and abyssal basins (Yan et al., 2015a; Ding et al., 2018). Pre-spreading magmatism occurred mainly on the northern continental margin and on coastal areas of South China (Xu et al., 2012). Syn-spreading magmatism well developed in the northern continental margin, such as the Zhujiang (Pearl) River Mouth Basin and the Dongsha slope (Sun et al., 2014; Zhao et al., 2014, 2016; Fan et al., 2017; Deng et al., 2019; Zhang et al., 2021). Shortly after seafloor spreading ceased in the SCS, intraplate magmatism affected large areas around the SCS, such as the Zhujiang River Mouth Basin, the Leiqiong Peninsula, the Beibu Gulf, the Indochina Block, the Reed Bank, the Nansha Block, and the SCS Basin (Ho et al., 2000; Yan et al., 2014; Hui et al., 2016; An et al., 2017; Zhao et al., 2018; Sun et al., 2020a).

In the northern continental margin, Fan et al. (2017) calculated the volume of intrusive magma above the Moho for some identified volcanoes and suggested that the volume of magmatism was comparable to the standard of large igneous province (LIP) (Wignall, 2001). In the abyssal basin, there are multiple seamount chains along the fossil spreading ridge and adjacent areas, as well as many isolated seamounts (Yan, 2008; Yan et al., 2014). Dredged seamount samples range in ages from ~16 Ma to < 1 Ma (Zhao et al., 2020b), revealing that most of them formed during post-spreading and generally are younger near the fossil spreading ridge (Sun et al., 2019b). IODP Expedition 349 also found interbedding between basalt volcanoclastic breccia and clastic sediment layers, indicating multiple volcanic eruptions after seafloor spreading (Li et al., 2015b). From the topographic map (Fig. 1), the SW Sub-basin's seafloor was found to be flattened, marked by isolated seamounts and linear structures.

Most intraplate volcanism within the SCS occurred millions of years after the cessation of seafloor spreading (Yan et al., 2015a & b), mainly manifested as early tholeiites and late alkali-basalts, with the characteristics of oceanic island basalt-type (OIB) (Yan, 2008; Xu et al., 2012; Li et al., 2015a). The SCS began to erupt predominantly alkali magma after ~8 Ma (Yan et al., 2014), which is consistent with the age of alkali-basalts in southern Vietnam (An et al., 2017). The compositional evolution of post-spreading magmatism in the SCS is similar to that of the Indochina block and the Leiqiong Peninsula (Yan et al., 2014). Geochemical analysis showed that most volcanism exhibited features of enrichment components, which previously were assumed to be related to deep mantle materials and subduction plate circulating materials (Yu et al., 2018; Zhang et al., 2018).

Cenozoic basalts found in the northern margin, the Leiqiong area, the SCS Basin and the Indochina Block range from 28.5 Ma to < 0.1 Ma and share the same petrologic and geochemical characteristics (Hoang and Flower, 1998; Ho et al., 2000; Yan et al., 2018), which constitute the vast southeastern Asian basalt province (Zhao et al., 2021a). Since volcanism in the southern margin is obscure and has not yet been included in the southeastern Asian basalt province, the southern margin is an important missing piece for constructing the integrated characteristics of SCS volcanism. Comprehensive geophysical and petrological results can help explore fundamental problems about the structure and tectonic evolution of submarine volcanoes in the SCS's southern margin.

3 DATA AND METHOD

This study relies on the 2D reflection seismic data (XD01) newly acquired by the R/V Shiyan 2 of the South China Sea Institute of Oceanology, Chinese Academy of Sciences in 2019. The profile crosses the SW Sub-basin and Nansha Block. The total length of the survey line is about 310-km, and the reflection seismic data was acquired through a 24-channel cable. The air-gun array, with a total volume of ~0.098 m³, consisted of four large capacity Bolt-guns as artificial sources. An isochronous shooting method was adopted, with shot intervals of 90 s, a ship speed of approximately 5 knots, and the shooting pressure of 12.5–13.5 MPa. A total of 1 601 shots were fired.

The 2D seismic data were processed using routine procedures such as format conversion, gain recovery and filtering to obtain the reflection seismic profile along the line. Volcanism in the SW Sub-basin and Nansha Block are revealed in combination with gravity and magnetic data. To obtain a more comprehensive and accurate understanding of the regional submarine volcanism, we collected previous data published on the southern margin, including 45 MCS profiles, 4 OBS seismic profiles, drilling and dredging samples (Fig. 2). The characteristics, tectonic evolution, and dynamic mechanism of submarine volcanoes in the southern margin of the SCS are discussed.

Since composition within submarine volcanoes is generally igneous rocks, their geophysical characteristics significantly differ from seafloor sediments or mud volcanoes (Yin et al., 2021). In a reflection seismic profile, igneous rocks have higher densities and faster seismic velocities, show a strong impedance contrast to sedimentary rocks (Zhang et al., 2016; Song et al., 2017), and when located below sedimentary deposits can usually be identified by positive high amplitude reflections from seismic data (Zhao et al., 2014, 2016; Zhu et al., 2021). Usually intrusive igneous rocks would cause local deformation to surrounding rocks, often forming forced folds, whereas extrusive rocks are onlapped by the overlying strata at their apical interface (Trude et al., 2003; Zhang et al., 2016). Additionally, igneous rocks usually have a high shielding effect that produces blank or chaotic reflections inside it (Sun et al., 2020b; Zhu et al., 2021). Since volcanoes are directly exposed on the seafloor and can reach closer to sea level, they often have strong gravity and magnetic signals (Song et al., 2017).

In summary, submarine volcanoes in the study area are mainly identified by the following criteria based on whether they have: (1) distinctly high topography on the seafloor, (2) special morphology of a volcano-like or mounded cone, (3) positive high amplitude external seismic reflections, (4) low amplitude and chaotic internal seismic reflections, (5) obvious uplift of the adjacent strata at the flanks, (6) higher positive gravity anomaly, and (7) higher (positive or negative) magnetic anomaly.

4 RESULT

The newly acquired seismic reflection profile (Fig. 3a) crosses NE to SW through the SW Subbasin (0–80-km offset), continental slope (80–130-km offset), fault-block belts (130–200-km offset) and graben-horst zones (200–310-km offset). Due to water depth in the Yongshu Island area, the shipborne seismic source could not be shot normally, leaving a data blank window of approximately 16.5 km. The relatively flat terrain of the SW Subbasin is approximately 4 km deep. Its fossil spreading ridge is a central rift bounded by faults. The IODP Sites U1433 and U1434 were drilled for the first time in the Southwest Sub-basin to meet the basalt base (Fig. 2), revealing the oceanic crust's lithological characteristics and overlying sedimentary deposits (Li et al., 2014).

We referred to the drilling data of IODP Site U1433 in the SW Sub-basin and ODP Site 1143 in the Nansha Block, and the sequence stratigraphic division (Yan and Liu, 2004; Ding et al., 2013; Zhang et al., 2020a). The main reflection interfaces of the shallow sediments (T1, T2, T3, T4) could be identified based on the reflection characteristics of this seismic profile (Fig. 3b). With strong amplitudes and continuous high-frequency reflections, T1, T2, and T3 have nearly horizontal characteristic in the oceanic basin. In the southern margin, T2 and T3 are approximately parallel to the seafloor surface while T1 is not pronounced (Zhang, 2020). According to previous studies, the age corresponds to 2.58 Ma (T1), 5.3 Ma (T2), and 11.6 Ma (T3) (Wang et al., 2000; Li et al., 2015b). T2 is the interface between Miocene and Pliocene, corresponding to the end of the collision and uplift event of the Nansha Block (Ding et al., 2013). T3 is the interface between the Middle Miocene and Late Miocene (Zhang, 2020). T4 (~15.5 Ma) formed at the end of the spreading, with continuous moderate amplitude reflection (Zhang et al., 2020b). On newly acquired seismic profile, we first identified 5 submarine volcanoes, using S1–S5 as their identification codes (Fig. 3b). S1 is located in the SW Sub-basin, whereas S2–S5 developed on the thinned continental crust of the Nansha Block.

4.1 Qualitative identification of submarine volcanoes

S1 in the SW Sub-basin has a height of more than 2 km above the basement (Figs. 3 & 4a), a width at the basal of approximately 13 km, and a maximum flank dip of approximately 14°. Not a common cone-shaped tip but rather a caldera that slopes to the northeast, S1 is completely buried in thick sedimentary deposits, proving that it has been active for a long time. Sedimentary reflections on either side of S1 differ. Northwest of S1, sedimentary reflection characteristics are continuous and horizontal with weaker reflections, proving that properties of sediments do not relatively change. The reflections of deposits in the southeast of S1 show horizontal stratification and strong amplitude. Sedimentary barrier effect of volcanoes (Chen et al., 2014; Yin et al., 2021) left the two sides of S1 with distinct characteristics, suggesting that S1 formed before the deposition of T4 (~15.5 Ma). Significantly different from the other four seamounts (S2–S5) with negative magnetic anomalies, S1 has a positive free-air gravity anomaly and a strong positive magnetic anomaly (Fig. 3c). The differences may reflect the different geomagnetic directions at the different periods. The adjacent MCS profile DZ02 located the S1 near the Continent-Ocean Transition (COT) (Zhang et al., 2020a). Approximately 20 km southeast of S1, a NW-dipping normal fault (Fig. 4a) forms a steep terrain. The thicker sediments above there have a strong forced fold, judged to be an intrusion.

S2, located at the uplift of the continental slope (Fig. 4b), has a height of more than 2.5 km, and a width of more than 20 km at the basal with a maximum flank dip of approximately 15°. The sediments in the northwest are relatively thin, while those in the southeast slowly transition to the normal sedimentary sequence of the Nansha Block. Although S2 has a distinct positive gravity anomaly, and corresponds to a negative magnetic anomaly, it is not in the center of the magnetic anomaly (Fig. 3c), perhaps because of the geomagnetic polarity reversal (Fan et al., 2017), the anomaly of the intrusive rocks or magmatic conduit. Vesicular olive basalts (~3.8 Ma) were collected from the S08-69-1 dredged site on S2 with alkali-basalts petrological characteristics (Yan, 2008), indicating S2 erupted during the Pliocene. Isotopic results indicate two mixed terminal elements in the source region of these alkali-basalts: asthenosphere or lithospheric mantle (DMM) and deep mantle upwelling (EM2) (Yan et al., 2008). OIB alkali-basalts indicate that S2's magmatic source is probably the deep mantle material.

S3 is located to the southeast of the Yongshu Basin with a height of approximately 800 m (Fig. 4b). The width at the basal is approximately 5 km with the maximum flank dip of about 17.8°. Volcanic rocks are directly exposed out of the seabed. S3 corresponds to the most prominent magnetic anomaly in the entire profile (Fig. 3c), with the maximum negative anomaly even exceeding -96 nT. Such a high magnetic anomaly may be related to the abundant magnetic minerals in volcanic rocks. Within 20 km northwest of S3, the intrusion caused the loss of all sedimentary layers below T2, strong uplift in surrounding strata and nearly vertical faults above T2, so it is judged that S3 erupted after T2 (< 5.5 Ma). To the southeast of S3, the Yongshu Island corresponds to the positive free-air gravity anomaly with the highest value (> 95 mGal), it was previously thought that its carbonate reef mass may be accreted on a volcano (Zhang, 2020), but lacking data in this profile, we cannot discuss the basement properties of the Yongshu Island.

Between S2 and S3, Yongshu Basin, with sediments more than 2 km thick, was formed in the depressions of tilted fault blocks controlled by several NW-dipping normal faults. These tilted fault blocks are the main tectonic feature of the basement in the Nansha Block (Yan et al., 2006; Hutchison and Vijayan, 2010; Ding et al., 2013). There are complete sedimentary sequences in the Yongshu Basin, from bottom to top, which are pre-rift basement, syn-rift sedimentary deposits from Late Oligocene to Early Miocene, and post-rift draping strata from Middle Miocene to present (Hutchison and Vijayan, 2010).

ODP Site 1143 is only approximately 9 km away from the profile (Fig. 5). The depth of the drilled core is more than 512 m (Wang et al., 2000). The shallow strata of this study can be compared with drilling data and seismic section across Site 1143. The basement in the graben-horst zones is sharply undulating, with multiple submarine volcanoes and faults. To the southeast of Yongshu Island, a NW-dipping synsedimentary fault (Fig. 5a) cuts directly through the uppermost sediments, and an obvious fault scarp can be seen on the seabed topography. Maximum sedimentary thickness of the syn-fault sequence is more than 1.5 km. The synsedimentary fault is characterized by strata tilting toward the main fault, which is very active during the sedimentary process. As seen on the magnetic anomaly map (Fig. 3c), the negative magnetic anomalies on both sides of this fault suddenly change into very sharp positive magnetic anomalies. In the southern margin, the fault that remains active today was probably reactivated by the pre-existing syn-rift extensional fault, evidence that some extensional faults are currently active.

The maximum depth of the submarine canyon is approximately 3 km (Fig. 5a), showing a distinctly negative terrain. Beneath this submarine canyon, a sedimentary basin was formed by the sinking of the basement due to strong faulting. S4 is located to the northwest of the submarine canyon (Fig. 5a), with the volcanic rock directly exposed out of the seabed. S4, with a height of approximately 1 km, a width of approximately 5 km at the basal, and a maximum flank dip of approximately 22°, has a positive free-air gravity anomaly and a negative magnetic anomaly with a high value (Fig. 3c). The ODP1143 drill near S4 found dacitic-rhyolitic tuff, volcanic ash, and volcanic glass with an age of < 2 Ma (Wang et al., 2000). The volcanic glass has higher density and generally comes from a nearer eruption source, indicating these pyroclastic materials are most likely related to the eruption of S4. Dacitic-rhyolitic rocks may represent the transition from mafic magma in the deep to intermediate-acid magma in the shallow by fractional crystallization of the upward transport.

To the southeast of the submarine canyon (Fig. 5a), the sedimentary basement has undergone significant uplift, which may be attributed to intrusions based on its reflective characteristics. The width of these intrusions near the basal is more than 20 km. The pull-up disturbance of the surrounding sedimentary strata can be observed around these intrusions. Intrusions penetrated T2, with homologous faults of magmatic intrusion formed in the strata above T2, leading to a conclusion that intrusions may have been active during Pliocene-Pleistocene. Because these intrusions and S4 are located on both sides of the submarine canyon with similar active eras, it is possible that their conduits are related to the extensional faults of the submarine canyon.

S5 has a height of approximately 1.5 km above the basal (Fig. 5a), and a width of approximately 8 km at the basal with the maximum flank dip of approximately 20°. On both flanks of S5, the pull-up disturbance of the T2 and T3 is obvious. The top of S5 was superimposed by thinner surface sediments. To the southeast of S5, an intrusion that looks like a dike can be recognized. The dike caused some forcing folds and contemporaneous faults in all strata above T3, so the intrusive age may be Quaternary. In the vicinity of S5 and the dike, there are SE-dipping normal faults, and a continuous sedimentary basin controlled by these normal faults on the southeast side of the dike.

4.2 Distribution characteristics of submarine volcanoes

To further ascertain the distribution characteristics of submarine volcanoes at the southern margin of the SCS, we collected 45 previous MCS profiles covering the area. Instead of reinterpreting these MCS profiles, we directly counted locations of submarine volcanoes discovered by previous studies, marking all volcanoes in Fig. 6 along with submarine volcanoes identified in profile XD01. Excluding submarine volcanoes in the abyssal basin, we found a total of 43 submarine volcanoes in the southern margin. We also collected the results of igneous rocks found by available dredging and drilling at the southern margin (Table 1). Except for the older Mesozoic igneous basement, post-spreading volcanic rocks were sampled at 8 volcanoes (Kudrass et al., 1986; Wang et al., 2000; Zhou et al., 2005; Yan, 2008).

By combining submarine volcanoes identified in this study and post-spreading volcanic rocks revealed by drilling and dredging, we circled the approximate range of volcanism around the outer edge of these volcanic eruptions (Fig. 6). Since MCS data are not evenly covered, the range of submarine volcanism we delineated is not fully representative of the actual situation, but the trend of submarine volcanism is relatively obvious. Our research found that submarine volcanism in the southern margin developed continuously from the Wan'an Basin to the east of Reed Bank. An interesting phenomenon is that a narrow NE-trending volcanic belt along the southwest rift and the southern Continent-Ocean Transition zone, was obviously affected by the crustal extension. Our results show that there are more volcanoes in the north than in the south of the Nansha Block.

The delineated range of submarine volcanism in this study, combined with post-spreading basalt eruptions in southern Vietnam and the southern Xisha Block from previous studies (Hoang and Flower, 1998; Hoàng et al., 2013; Yan et al., 2014) and the results of the Global Seamount Survey (Kim and Wessel, 2011), reveal the large-scale postspreading volcanism in the southern margin (Fig. 6). Since volcanism at the southern margin obviously did not occur in isolation, the spatially adjacent and contemporaneous intraplate volcanism throughout the extensive region seems to indicate a common mantle-scale tectonic cause.

5 DISCUSSION 5.1 Temporal and spatial distribution of volcanism

The igneous rocks in Table 1 are clearly divided into two categories in terms of chronology. The earlier category includes Mesozoic granite, basalt, rhyolite, etc., which are generally considered to be the Mesozoic basement. The latter category includes young volcanic rocks that erupted during the Pliocene-Pleistocene, with the dating ages between 4.3 Ma and 0.4 Ma. The lithology is dominated by alkali-basalt. By analyzing the relationship between the submarine volcanoes (S2–S5) identified in this study and the adjacent strata, it was also indicated that S2–S5 erupted after T2 (~5.5 Ma). Additionally, some volcanoes in the western part of the Nansha Block were confirmed to have erupted during Pliocene (Li et al., 2013). Limitations in obtaining rock samples from submarine volcanoes prevent us from proving that all seamounts at the southern margin erupted in this period, but it can be proved that the Pliocene-Pleistocene was an important active period of submarine volcanism in there.

Previous studies have divided Cenozoic magmatism in the northern margin of the SCS into three periods: pre-spreading (Paleocene and Eocene), syn-spreading (Early Oligocene-Middle Miocene), and post-spreading (Middle Miocenepresent) (Yan et al., 2006; Hui et al., 2016; Zhang et al., 2016; Sun et al., 2020a). During the Pliocene-Pleistocene, massive magmatic intrusions and volcanic eruptions occurred in the northern SCS, Xisha Uplift and continental shelf of central Vietnam (Tan et al., 2014; Fan et al., 2017; Wang et al., 2018a; Sun et al., 2019a). The most abundant magmatism since 5.5 Ma in the northern margin of the SCS is considered to be a significant regional event (Zhang et al., 2016), coinciding with the ages of the submarine volcanoes we found at the southern margin. This indicates that during the Pliocene-Pleistocene, the northern and southern margins of the SCS were in an active period of volcanism.

Submarine volcanoes are mainly distributed in fault-block belts and graben-horst zones where the crust is strongly stretched and weakened. Volcanoes identified in the northern SCS are mainly distributed along faults (Fan et al., 2017). Coincidentally, most submarine volcanoes in the southern margin developed in these areas with dense faults (Fig. 7). In the narrow belt of the northern Nansha Block, submarine volcanoes are widely developed, and extensional faults are dense in the northern Nansha Block with dominant NE-trending (Clift et al., 2008; Hutchison and Vijayan, 2010), indicating intense crustal fragmentation and cutting. The magmatic materials in the crust tend to migrate upward along pre-existing stretching faults to form submarine volcano (Fan et al., 2017; Wang et al., 2019; Zhao et al., 2020a). The Reed Bank is a rigid block with minimal Cenozoic stretching deformation (Ding and Li, 2011), and there is no submarine volcano. In such an area, it may be more difficult for magma to break through the crust.

This study shows that pre-existing stretching faults were still active in the later period, and a clear correlation exists between submarine volcanoes and stretching faults. Pre-existing Faults may provide vertical weak belts, which can help magma rise through the thinned continental crust (Zhao et al., 2014; Fan et al., 2017; Wang et al., 2019; Wen et al., 2021). The extensional faults of fault-block belts and graben-horst zones in the southern margin have obviously spatial correspondence with post-spreading volcanism, from which we infer that these fault systems provide conduits for submarine volcanism.

5.2 Magmatism and tectonic evolution of submarine volcanoes

For the deep structure and tectonic evolution of volcanoes, we consulted the results of OBS seismic data. Profile OBS-PR and profile OBS973-1 are relatively close to profile XD01 (Fig. 2). The crustal structure of profile XD01, shown in Fig. 8a, provides key information for discussing tectonic evolution of submarine volcanism. The thickness of the crust below S2 is only approximately 10 km. The extensional faults as magmatic conduits of S3, extended aslant to the necking zone with thinner crust. S4 and S5 are located in graben-horst zones with crustal thicknesses less than 15 km, indicating extreme stretching and thinning. Thinner crust has lower overburden pressure and shorter ascending conduits, which may be more conducive to magmatism (Fan et al., 2017; Xia et al., 2018). Therefore, we infer that the submarine volcanoes in the southern margin are controlled by the thickness of the lithosphere after rifting and stretching.

The crustal structure of refraction seismic profile OBS-PR across the Zhenghe Massif is shown in Fig. 8b. There are HVBs in the lower crust with velocities ranging from 7.0–7.7 km/s (Pichot et al., 2014). After reconstructing profile OBS-PR by adding shallow constraints of MCS data, the results further confirmed the existence of the HVBs under the Nansha Block (Liang et al., 2019). There are also high-velocity bodies (HVBs) in the lower crust with velocities greater than 7.0 km/s in the western portion of the southern margin (Wei et al., 2020). It is believed that the HVBs under the Nansha Block has a spatial connection with volcanoes and intrusions based on seismic profile and gravity data (Chang et al., 2017a).

To form such a large area of submarine volcanism in the southern margin, there must be a large scale of conduits and magmatic sources. Recently, increasing studies have found that the HVBs under the continental margins of the SCS has a genetic relationship with magma underplating and submarine volcanism (Lester et al., 2014; Fan et al., 2017, 2019; Xia et al., 2018; Cheng et al., 2021; Wen et al., 2021). The HVBs in the continental margin of the SCS is generally not associated with the Cenozoic stretching crust (Lü et al., 2017). Considering the crustal thickness of the Nansha Block, the possibility of serpentinization is also ruled out (Pichot et al., 2014). Therefore, HVBs are more likely to be formed by the magma underplating of mantle material. In the process of mantle upwelling, molten magma intruded into the crust. In the Nansha Block, crustal thickness of the faultblock belts and graben-horst zones is only 10–20 km (Qiu et al., 2011; Pichot et al., 2014). In these areas with strong stretching and thinning, the fault systems provide conduits for the upward migration of the magma in the deep crust. We infer that the submarine volcanoes in the southern margin may have a causal connection to the HVBs below them.

Many studies suggested that magma underplating driven by mantle upwelling is the dominant mechanism of post-spreading seamounts in the SCS (Xu et al., 2012; Yan et al., 2014; Xia et al., 2016; Fan et al., 2017; Zhang et al., 2018). Long-standing mantle upwelling leads to massive magmatism and hydrothermal leakage at the northern margin of the SCS (Zhao et al., 2021b; Lin et al., 2022). Trace elements and Sr-Nd-Pb-Hf isotope analysis indicated that oceanic crusts of the SCS have deep mantle sources (Zhang et al., 2018). Recent seismic tomography studies have found multiple lowvelocity anomalies in the deep mantle of the SCS (Huang et al., 2015; Zhao et al., 2021a; Hua et al., 2022; Wang et al., 2022), suggesting that mantle upwelling occurs beneath the broader SCS region. Surrounding by a curved subduction system, the SCS has shown increasing geochemical evidence of subduction-induced mantle upwelling (Li et al., 2021). With evidence of alkali-basalt (OIB-type) eruptions, with isotopic results showing the mantle source, with the HVBs in the lower crust and mantle upwelling with low-velocity anomalies, we infer that the mantle upwelling in the SCS should make a direct or indirect contribution to the post-spreading volcanism of the southern margin. Magma underplating caused by mantle upwelling with low-velocity anomalies in the southern SCS may have provided a clue of genesis for the submarine volcanism.

We propose the formation mechanism of postspreading submarine volcanoes in the southern margin of the SCS (Fig. 9). Mantle upwelling with low-velocity anomalies penetrating the lithospheric mantle may lead to magma underplating at the bottom of the crust, thus forming the HVBs in the lower crust with a thickness of 2–3 km in the southern margin. Under appropriate conditions, magma can break through the lithosphere weakened zones along pre-existing stretching faults. Finally, submarine volcanoes preferentially erupted in the region with strong lithosphere stretching and thinning. Therefore, we infer that the intensity of postspreading volcanism in the southern margin might be affected by the stretching faults, crustal thinning, and magma underplating caused by mantle upwelling, and have caused the special spatial distribution of submarine volcanoes.

6 CONCLUSION

There are a large number of Pliocene-Pleistocene submarine volcanoes (dating from 5.5–0.4 Ma) in the southern margin of the SCS, indicating an extensive distribution of post-spreading volcanism. These volcanoes are concentrated in the fault-block belts and graben-horst zones with strong stretching and thinning in the crust, and enveloped to form the volcanic belts, indicating that the submarine volcanoes are more likely to occur in areas with weakened crust. The stretching faults have clear spatial correspondence with submarine volcanoes, which may provide magmatic conduits. We propose that submarine volcanoes identified in this research are related to the HVBs in the lower crust and magma underplating, giving clues for further understanding the evolution of the post-spreading submarine volcanism in the southern margin.

7 DATA AVAILABILITY STATEMENT

The data that support the findings of this study are available from the corresponding author upon reasonable request.

8 ACKNOWLEDGMENT

We thank the crew of R/V Shiyan 2 and the scientists who participated in the science voyage. Genggeng WEN, Jiangnan LIN, Tao GOU, and Changrong ZHANG from South China Sea Institute of Oceanology, Chinese Academy of Sciences, are greatly appreciated for their useful discussions. We are also grateful to editors and anonymous reviewers for their comments and suggestions, which greatly improved the manuscript.