Introduction

Reservoir quality of sandstones is critically impacted with the clastic composition controlled by provenance distribution patterns. Therefore, potential sources and sedimentary transporting routes of the northern South China Sea (SCS) are of remarkable significance on hydrocarbon exploration as well as on geological information archive, such as paleo-environmental changes and tectonic evolution (Fig. 1a) (Morton et al. 2001; Rossi et al. 2002). Nevertheless, it virtually remains further investigation due to the complexity of geological framework, limited borehole coverage and poorly dated stratigraphy in this region, particularly northwestern SCS area. Submarine canyons, which strongly dissect continental margins, are commonly regarded as significant conduits for sediment transfer from continental shelf into deep-sea basins (Popescu et al. 2004; Antobreh and Krastel 2006; Harris and Whiteway 2011). As important hydrocarbon reservoirs, these canyons or channels possess the potential of accumulating huge amount of sandy deposition, which have always attracted widespread interests all over the world (Mayall et al. 2006; Gong et al. 2011). Globally, sediments transported in canyon systems not only gathered voluminous oil and natural gas resources, but also kept high-quality record of regional structural evolution, environmental and climatic change information (Hale et al. 2014; Bayliss and Pickering 2015; Flecker et al. 2015; Reimchen et al. 2016). Enhanced understanding of these geological structures on morphology, composition, evolution as well as depositional dynamics has been greatly generalized resulting from high-resolution seismic data and sequence stratigraphy (Arzola et al. 2008; Mountjoy et al. 2009; Jobe et al. 2011).

Fig. 1
figure 1

a Topographic map of the Yinggehai and Qiongdongnan Basins relative to Southeast Asia and the South China Sea (SCS) and the bottom zoomed plot. b The special tectonic units, studied borehole samples and published data. Crosses and lines represent the analyzed samples of rare earth element (REE) and detrital zircon U–Pb dating, respectively

Full size image

The present “S-shaped” Central Canyon is a large axial submarine system with an approximate length of 570 km and width of ~ 9–30 km, which lies paralleled to the shelf break and stretches with a NE–NEE orientation in the northwestern margin of the SCS (Fig. 1b). Due to its unique internal architecture and interplay of erosion and aggradation processes, forming mechanisms and sedimentary provenances of the Central Canyon area have long been in debate over the past 10 years (He et al. 2013). It has been implied that sufficient sediments for the initial growth of Central Canyon were provided by rapid and broad Tibetan Plateau uplift around 5.5 Ma (Gong et al. 2011). Meanwhile, the Red River Fault movement also induced high-energy gravity flows providing voluminous material (Yuan et al. 2009). By contrast, Li et al. (2017a) favored that eastern Vietnam has already delivered sufficient sediments with large-scale submarine fans stepwise developing in Central Canyon since 10.5 Ma. Based on further integration and compilation of 2D/3D seismic and well log profiles, provenance evolution was reconstructed for different segments in the Central Canyon source-to-sink system (Su et al. 2013; Li et al. 2017b). After describing morphological characteristics and dissecting internal architectures of the Central Canyon system, Su et al. (2013) developed multi-phases including initial, eroded infilling, tranquil infilling and rejuvenation period in order to reconstruct provenance evolution. On one hand, northern continental slope and major potential sources including Red River, eastern Vietnam or Hainan Island were dominant suppliers to the northwestern SCS, on the other hand, Qiongdongnan tectonic transformation was likely to constrain the evolution of Central Canyon eastern segment around 11.6 Ma (Xie et al. 2006, Su et al. 2014). However, due to lack of convincible evidence requested by source-to-sink system analysis, temporal and spatial variations of sediment composition patterns in Central Canyon area since late Miocene are still perplexing issues from the prospective of origination and infilling mechanisms, which seem even harder to be deeply interpreted by the complicated potential sources adjacent to Yinggehai–Qiongdongnan Basin illustrated in detail as follows. Although details on provenance of Red River remain controversial, its modern drainage system has been revealed as maximum discharge delivering high volumes of sediments into the northwestern SCS during the Neogene, likewise, the paleo–Red River was also presumed to be a significant source (Clift et al. 2006; Milliman and Farnsworth 2011; Zhao et al. 2015). Seismic profiles initially indicated that progradational slope clinoforms were not largely generated in Hainan Island until the late Cenozoic (Van Hoang et al. 2010; Xie et al. 2008). Verified by U–Pb age dating, Hainan Island has already exerted its impact on sediment supply to its adjacent basins during Oligocene–Miocene (Shao et al. 2016; Yan et al. 2011). Additionally, eastward progradation from the Vietnamese margin into the western Qiongdongnan Basin and low degree of weathering conditions from geochemistry analysis together imply a Central Vietnam provenance since the Miocene, which is contrast to the speculation of little sediment delivery due to absorption of the Mekong River drainage basin (Ma et al. 2016). A series of intensive local mafic volcanic activities was also indicated by mineral assemblage of augite and olivine and positive Eu anomalies in the Qiongdongnan Basin during late Cenozoic (Fyhn et al. 2009; Zhao et al. 2015). Similarly, how exactly this suite of provenances influenced the Central Canyon’s sedimentary evolution since its origination is scarcely systematically discussed and remains rather elusive.

Provenance studies of sedimentary sequences are critical to basin history analysis and natural resource exploration. It is ineffectively to interpret the source-to-sink system solely based on traditional petrography or heavy minerals particularly in the complicated source areas. Zircon is uniquely characterized by its mineral stability which can withstand mixed effects of weathering, erosion as well as thermal alteration processes (Kosler 2003). Due to its stable isotopic system under different pressure, temperature and fluid composition conditions (Moecher and Samson 2006), detrital zircon U–Pb age spectra dated by Laser Ablation Induction Coupled Plasma Mass Spectroscopy have been increasingly considered as a valuable tool in wide range of geological researches (Cao et al. 2018; Horton et al. 2008; Condie et al. 2009). In addition to zircon U–Pb geochronology, immobile trace elements, especially rare earth elements (REE) are also robust indicators of source rock geochemistry as these elements suffer minor modification from sorting, fractionation and diagenesis processes during transportation (Shao et al. 2017; Mclennan 1989). Although detrital single-grain analysis has gained much popularity among provenance studying field, bulk rock geochemistry is still able to provide quantitative insights into siliciclastic sediments.

In this study, we aim to address the following problems by focusing on new results of detrital zircon U–Pb ages and geochemical analysis from Yinggehai Basin and Qiongdongnan Basin, in comparison with recalculation and integration on published or unpublished data from Central Vietnam, modern Red River, Hainan Island and SCS basement: (1) what exactly are the sediment contributors surrounding the SCS for the Central Canyon and what types of clastic materials did they provide? And (2) what is the sedimentary infilling history for the entire Central Canyon during late Miocene–Pliocene?

Geological setting

Basically, Yinggehai Basin is located between Hainan Island and Indochina block with a NNW–SSE orientation. Perpendicular to the trend of Yinggehai Basin, Qiongdongnan Basin is surrounded by Hainan Island, Xisha Uplift and Shenhu Uplift (Fig. 1b). Both tectonic evolution and stratigraphic structures of Yinggehai and Qiongdongnan Basin have been investigated for decades. Mechanisms including the strike-slip events of the Red River Shear Zone, the Indochina Block extrusion model and the SCS seafloor spreading have been debated but remain disputed for the interpretation of these two oceanic basins’ evolutionary histories (Zhu et al. 2009; Zhang et al. 2013; Zhu and Lei 2013). Nevertheless, typical passive continental margins of Yinggehai–Qiongdongnan Basin have still been identified: a large-scale rifting stage prior to the late Oligocene and a post-rifting stage with accelerated thermal subsidence since early Miocene (Taylor and Hayes 1983). In particular, intensive tectonic activities of the Red River Fault Zone took place during the Neogene post-rifting phase when the previous strike-slip reversal processes had been switched into dextral movement in the northern Vietnam (Rangin et al. 1995). This could somehow modify or change the characteristics of the parent rocks exposed in this area. Meanwhile as additional complementary information, Neogene magmatism proved to be relatively strong around the SCS continental margins. Widely distributed basalts along the central and southern Indochina block and basaltic extrusions on Hainan Island were also explicitly recorded (Yan et al. 2006). Under controlled by this complexity of the tectonic backgrounds, littoral to neritic sedimentary environment has gradually been changed into bathyal to abyssal facies since late Miocene in most regions of the Qiongdongnan Basin where submarine fans or deltas were abundantly generated. On the other hand, shelf–slope systems were partly distributed in the Yinggehai Basin facilitated by high rate of sediment supply (Cai et al. 2013; Zhang et al. 2015; Li et al. 2016). In this deep-water sedimentary setting, the main drainages of Central Canyon transporting sediment material through Qiongdongnan abyssal plains were initially developed (Fig. 2).

Fig. 2
figure 2

Schematic stratigraphic column of the major northern SCS basins (Mi and Zhang 2011) with relative and global sea-level variations (Haq 1988), sedimentary environment, SCS evolution history (Li et al. 2016) and tectonic backgrounds (Taylor and Hayes 1983; Leloup et al. 1995; Yan et al. 2006)

Full size image

Analytical methods

Detrital zircon U–Pb dating implies the time of zircon growth, which is generally the rock crystallization age. The zircon U–Pb system is particularly characterized by a high closure temperature and in most cases unaffected by high-grade metamorphism to temperatures at least as high as 750 C (Cherniak and Watson 2001; Carter and Bristow 2015). Extensive oil and gas drilling programs have been conducted in the northern SCS area by China National Offshore Oil Corporation (CNOOC). Specific sample locations in our research are displayed in Fig. 1b, with detailed sampling information provided in Table 1. Age models of all these boreholes are derived from biostratigraphical and unpublished seismic data owned by CNOOC. Zircons in our samples were extracted from the bulk sediments by conventional magnetic and heavy liquid separation techniques in laboratory of the Institute of Regional Geology and Mineral Resources, Hebei Province, China. Prior to analysis, zircon grains have been randomly selected followed by cathodoluminescence (CL) imaging on the polished epoxy disks to locate proper analytical spots in oscillatory zoning (Supplementary Fig. S4 for CL images). The U–Pb isotopic dating was processed by laser ablation–inductively coupled plasma–mass spectroscopy (LA–ICP–MS) at the State Key Laboratory of Marine Geology, Tongji University, China. The instrumentation involves a Thermo Elemental X-Series ICP–MS coupled to a New Wave 213 nm laser ablation system, of which the laser beam has been set to a 10 Hz repetition rate with a spot size of 30 µm. The ablated material was then transported into the ICP–MS with a He–Ar gas mixture. 20 s of gas blank was incorporated for each analysis before 50 s duration of data acquisition. Our method employs an external standard zircon 91500 (1065.4 ± 0.3 Ma) analyzed twice every seven analyses (Wiedenbeck et al. 1995) and Plešovice (337.1 ± 0.4 Ma) for accuracy calibration (Sláma et al. 2008). The U–Pb ages were finally reported with 1σ level of uncertainty calculated by ICPMSDATACAL combined with the common Pb correction (Supplementary Table S1, 2; Supplementary Fig. S5) (Andersen 2002; Liu et al. 2010), while visualization for age distribution patterns were achieved via histograms and kernel density estimation plots (Vermeesch 2012). Generally, 206Pb/238U ages are more precise for younger ages (< 1000 Ma), whereas 207Pb/206Pb ages are applied for older ages (> 1000 Ma).

Table 1 Stratigraphic position of analyzed samples in this study and published research (Wang et al. 2014; Cao et al. 2015)
Full size table

Sediments aimed for REE analysis were first washed with deionized distilled water at the State Key Laboratory of Marine Geology, Tongji University, China. After being dried at 50 °C for 48 h and crushed in an agate mortar, about 3 g of dry powdered samples were heated to 600 °C for 2 h and then digested with 0.1 M HCl in order to remove organic matter and interlayer water as well as small part of marine carbonates, separately. Dissolved in a HF–HNO3 mixture, trace elements were measured using Thermo Elemental X-Series (ICP–MS). Certified standards including GSR-5, GSR-6 and GSD-9 provided by the Institute of Geophysical and Geochemical Exploration, China, have been repeatedly analyzed as unknown samples for the precision and accuracy assessment. The external precision (1σ) was generally better than 5% and our reported concentrations turned out to be in consistent with the recommended data (Supplementary Table S3). The measured REE abundances have been normalized to the chondrite average (Boynton 1984).

Results

In our study, Yinggehai–Qiongdongnan samples display typical temporal and spatial variations from east to west in upper Miocene–Pliocene (Fig. 3; data of Th/U against ages for Supplementary Fig. S6). Generally, representative tectonic periods are divided into Himalayan (~ present-66 Ma), Yanshanian (~ 135–205 Ma), Indosinian (~ 205–257 Ma), Hercynian (~ Veriscan) (257–386 Ma), Caledonian (~ 400–570 Ma), Jinningian (~ 1000–1400 Ma) and Lvliangian (~ 1800–2500 Ma) movements. In the westernmost studied areas, both DF-1 and DF-2 are similarly observed with rather broad age spectra covering Yanshanian, Indosinian, Caledonian, Jinningian even Lvliangian peaks. Compared to DF-1, Pliocene sample of DF-2 displays Indosinian cluster (~ 235 Ma) of larger proportion within its overall pattern. As for another two Qiongdongnan boreholes, age combination differences among separate samples are more clearly identified. For well QW1 further proximal to Central Vietnam, although showing multiple peaks from upper Miocene to Pliocene, QW1-1 is characterized by its Indosinian (240 Ma) and Caledonian (410 Ma) peaks of almost same proportion, while QW1-2 is apparently dominated by its Indosinian (220 Ma) population with the Caledonian (420 Ma) sharply decreased and Jinningian (680 Ma) largely increased. Comparatively, borehole QE in the easternmost is also identified for weaker Caledonian signal (445 Ma) within younger strata sediments of QE-2. Moreover, both QE-1 and QE-2 are commonly featured by their higher Yanshanian peaks than the other analyzed wells. It should also be noted that the 155 and 240 Ma age peaks of QE-2 have already become the dominant clusters while the rest populations of 445 and 800 Ma have shrunken into secondary proportion.

Fig. 3
figure 3

Histograms and probability density distribution plots for detrital zircon U–Pb ages of the Yinggehai–Qiongdongnan Basin samples from upper Miocene to Pliocene with typical age peaks numbered (Ma). The youngest zircon U–Pb age (Ma) is marked with * in the upper left corner. n represents the number of concordant analyses relative to all grains. Published data in Dongfang gas field of Yinggehai Basin from Wang et al. (2014) and data in Qiongdongnan Basin from Cao et al. (2015) have been renamed and displayed on plot for comparison

Full size image

Figure 4 has provided abundant geochemical information on samples along Central Canyon drainage area. With all studied wells roughly considered, light rare earth elements (LREE) enrichment is evidently identified in all borehole samples. Especially among upper Miocene–Pliocene wells including DF, QW3, QW1 and QE, a decreasing trend in both values and numerals of positive Eu anomalies is discerned eastwards. In detail, well DF in northwestern Yinggehai Basin displays pronounced positive Eu anomalies within almost all the layers with only one slight positive value detected in samples (1506–2030 m) of DF-2. To the further east, the slight negative Eu anomaly is first observed in Pliocene samples (1500–1865 m) of QW3-2 in spite of obvious positive Eu anomalies for its other layers. Difference between upper Miocene and Pliocene sample for QW1 is more apparent. No negative anomalies are identified in QW1-1 while upper-layer samples in QW1-2 are featured by moderate-to-obvious negative values. As for the easternmost well, nearly all the QE samples are exclusively characterized by negative Eu anomalies except for the only slight positive data within 3070–3168 m. Contrast to resemblance with PAAS for other boreholes, QW2 located closer to Hainan Island exhibits larger disparity with PAAS pattern. Moreover, vertically, QW2-2 shares relatively higher-level of similarity to PAAS with larger overall REE concentration than QW2-1. However, most samples of QW2 are exhibiting intensive Eu depletion with the only slight positive data within 960–1100 m. Additionally, as complementary evidence, middle Miocene samples including MW, MM and ME also display remarkable spatial shifting in Eu anomalies. MW lying in the westernmost is featured by highest positive value while ME to the east shows strong negative signal.

Fig. 4
figure 4

Chondrite normalized REE patterns of upper Miocene–Pliocene samples in the study area; Eu anomaly values are highlighted on the plot

Full size image

Discussion

Provenance hypotheses

Geographical location of the Central Canyon indicates primary potential provenances including the modern Red River drainages, Kontum Massif of Central Vietnam areas, Hainan Island, SCS basement and other possible source suppliers (Fig. 5). Abundant zircon U–Pb dating and REE geochemical data on bedrocks and sediments of Indochina, South China blocks, surrounding fluvial networks and northern SCS basement has made great contributions to constraining the age distribution patterns of those potential source areas. Specifically, Eu anomalies are ubiquitously utilized in provenance discrimination studies as mafic or ultramafic rocks are typically featured by positive values while acidic or sedimentary rocks usually generate the opposite results. Therefore, by comparing zircon ages and REE patterns in the source-to-sink system, the Central Canyon sedimentary infilling evolution could be better interpreted. However, it should still be considered cautiously due to possible sedimentary mixing effect from various end members.

Fig. 5
figure 5

a Simplified geological map of the potential source areas with tectonic units including Yinggehai Basin, Qiongdongnan Basin, Red River Fault Zone, Song Ma Fault Zone, Song Ca Fault Zone, Kontum Massif and Khorat Plateau (based on the geological map of Asia 1975, Chinese Academy of Geological Sciences, 1975). Major drainage systems include Red River, Mekong River, Changhua River and Ningyuan River (Zeng and Zeng 1989; Li et al. 2012). b Compilation of published and unpublished detrital zircon U–Pb ages are displayed for different potential source areas (Van Hoang et al. (2009) for modern Red River; Burrett et al. (2014) for Central Vietnam; Xu et al. (2014), Cao et al. (2015) and Wang et al. (2015) for Hainan Island). S number of samples, N number of concordant analyses. c REE distribution patterns for potential sources are also shown (Zhao et al. (2015) for modern Red River; Shao et al. (2010) for Hainan Island)

Full size image

Modern Red River extends along the Ailao Shan–Red River Fault Zone, whose upstream bedrocks are mainly comprised of Mesozoic sedimentary rocks while downstream is ubiquitously featured by schists and gneisses (Fig. 5) (Leloup et al. 1995). Particularly, age signatures of the Red River might be imprudently analyzed, as its paleo-drainage courses may have suffered drastic changes from the East Asia tectonic activities, such as Tibetan uplift. Possible upstream capturing over the Cenozoic was likely to take place to change the sediment supply of the paleo–Red River (Clark et al. 2004). Hence, U–Pb chronology data cited in this paper only represents its modern topographic framework. Due to wide drainage areas, great lithological complexity and severe geotectonic modification, a broad age range (~ 2500–30 Ma) with apparent Indosinian and Jinningian peaks are identified (Van Hoang et al. 2009). To be addressed, there is a unique Himalayan cluster (~ 30 Ma) in its multimodal U–Pb combination compared to age patterns of other sources.

Central Vietnam of the Indochina block is widely occupied by Paleozoic–Mesozoic sedimentary rocks and mafic-to-ultramafic bedrocks (Fig. 5). Obviously, its Indosinian–Variscan (260 Ma) and Caledonian (430 Ma) peaks are almost of same proportion combined with a relatively smaller Jinningian cluster (945 Ma) (Burrett et al. 2014). Large-scale submarine fans were confirmed by seismic profiles and heavy mineral analysis on the deepwater sediments of Dongfang gas field, Yinggehai Basin, which evidently indicate a critical contribution from Kontum Massif of Central Vietnam (Zhong et al. 2013). According to previous studies, the Indochina Block experienced drastic extruding and deforming processes during late Triassic, resulting in high volumes of Paleozoic even Proterozoic zircon grains in the Central Vietnam sediments and thick successions in SCS basins. With continual Tethyan subduction beneath Eurasian Plate, extensive magmatic activities were especially taking place during Indosinian movement leading to a relatively high peak of 260 Ma. In the following Cenozoic period, this potential source might be partially modified under the influence from the activity of Red River, Song Ma and Song Ca Fault Zones (Kesselmans et al. 2001; Usuki et al. 2009). Consequently, the age signatures with three well-defined U–Pb peaks for this potential mafic-to-ultramafic sediment contributor kept a good record of the regional tectonic evolution history. As previously mentioned, albeit exhibiting similar wide Indosinian and Jinningian peaks, modern Red River is discriminative from Central Vietnam for its Himalayan cluster (~ 30 Ma) and lack of obvious Caledonian populations (~ 430 Ma). Since Cenozoic tectonic movement had a limited influence on Central Vietnam area, this contributor is devoid of zircons with rather young ages.

According to historical data report, Qiongdongnan and Yinggehai Basins seem to be primary adjacent areas receiving input from Hainan Island due to its relatively steep hillsides and huge-volume runoff (Zeng and Zeng 1989). Changhua and Ningyuan River are two important rivers generated from the central mountains down to surrounding plains (Fig. 5). Parent rocks of Hainan Island are mainly exposed of intermediate–acidic intrusions with limited modern-age basalts in its northern areas. With regard to the U–Pb signatures, Hainan Island displays well-shaped bimodal pattern clustering at Yanshanian (100 Ma) and Indosinian (235 Ma) periods, which is totally different from modern Red River and Central Vietnam.

Compared to multimodal or bimodal U–Pb combinations of peripheral suppliers, the wide northwestern SCS basement shows relatively simple age signatures. B1 is located in Beibu Gulf Basin while B2 and B3 lie in the middle Qiongdongnan Basin. These three samples are all apparently dominated by the confined Yanshanian peak (~ 140–145 Ma). As facilitated evidence, Zhu et al. (2017) confirmed the late Jurassic metamorphic basement around Xisha Uplift, which was then intruded by early Cretaceous granitic magmatic bodies. Slightly different from B1 and B2, B3 near to Shenhu Uplift exhibits an additional smaller Caledonian peak (~ 415 Ma) on its age spectrum.

As REE pattern proves to be a vigorous indicator for provenance studies, geochemical analyses of the potential contributors are also systematically conducted, accordingly. Interestingly, modern Red River, Hainan Island and SCS basement share high-level similarity bearing with negative Eu anomalies and LREE enrichment (Fig. 5). Their overall REE patterns are comparable to PAAS representing acidic or sedimentary rock types. By contrast, borehole DF on the submarine delta is uniquely characterized by strong Eu enrichment since middle Miocene. Since this borehole lies rather close to the continental margin and also confined by seismic profiles, its REE patterns might be reflective of abundant basic terrigenous materials from the contiguous block of Central Vietnam.

Provenance evolution of Central Canyon

As potential source areas assessed in the former section, previous studies indicated that Mesozoic suturing along Indochina–Yangtze boundary turned to be one of the most significant events, which had been defined as the Indosinian orogeny. This crucial movement not only includes multiple-stage subduction and collision processes but widespread igneous activities and subsequent metamorphism and deformation. Specifically, huge serpentinite bodies in Kontum Massif of Central Vietnam representing a remnant of paleo-oceanic between Indochina and South China were exposed in abundance during this period (Trung et al. 2006; Liu et al. 2015), which was then proved to be a substantial mafic-to-ultramafic supplier to the western SCS basins. Accordingly, geochemical results of middle Miocene samples of MW, MM and ME tracing the Central Canyon course strongly imply the unbalanced distribution of basic sediments in Qiongdongnan Basin prior to late Miocene. MW nearest to the Vietnamese margin received much larger volume of basic material than the other two wells close to Hainan Island which provided acidic type of sediments. Naturally, highest positive Eu anomalies were observed in its REE pattern. Basically, implication could be made that influence of mafic-to-ultramafic sediments from Kontum Massif on Central Canyon might be weakened or diluted when transported eastward (Fig. 4).

Based on compilation of previous geochronologic analysis, detrital zircon U–Pb ages have revealed a stable provenance of the upper Miocene to Pliocene sediments west of the Central Canyon drainage. Three broad age groups corresponding to Indosinian, Caledonian and Jinningian events in well DF show high compatibility with U–Pb signatures of Central Vietnam (Fig. 3). Meanwhile, the remarkable positive Eu anomalies for both DF samples also suggest clastic materials in northern Yinggehai Basin principally and continually derived from Kontum Massif basic rock regions (Fig. 6). In addition, although being detected in an extremely small scale, both Himalayan (~ 30 Ma) and Lvliangian peaks (~ 2500–1800 Ma) become larger in proportion for Pliocene sample of DF-2, which resemble with modern Red River characteristics to a certain degree. Despite a dominant contribution from the erosion off Kontum Massif highland, this slight change in U–Pb pattern might suggest stronger influence of other secondary sources controlled by Cenozoic tectonic modification processes, such as Red River Shear Zone activities. As confirmed by Chen et al. (2015), appearance of 30 Ma age peak is consistent with the timing of tectonothermal events around Red River Fault Zone.

Fig. 6
figure 6

Schematic source-to-sink evolution model for the late Miocene–Pliocene Yinggehai–Qiongdongnan Basin with Central Canyon highlighted. Arrows stand for sediment transportation routes from different potential source areas to the studied locations. Cyan—areas tend to be more influenced from basic rock sources; orange—areas tend to be more influenced from acidic or sedimentary rock sources

Full size image

QW1 and QE confined by geochemical analysis and zircon U–Pb dating are located in the western and eastern part of Central Canyon, separately (Figs. 3, 4). Spatially, both QW1-1 and QE-1 are characterized by obvious similarity to the age signatures of Central Vietnam, displaying high broad Indosinian, Caledonian and Jinningian peaks. Distinct from DF samples discussed above, Yanshanian clusters are also clearly identified indicating non-neglectable clastic materials from other important provenances. Since Hainan Island is featured by unique bimodal pattern with large Yanshanian and Indosinian peaks, sediments might be eroded from its central highs down into the north of Central Canyon channel. Although northwestern SCS basement is featured by single or major confined Yanshanian peak as represented by studied samples (Fig. 5), it turns out that the basement provided extremely limited sediments into our study area. Beibu Gulf Basin lies relatively isolated and farther to the channel while middle-south Qiongdongnan basement has been deeply submerged, even been distributed with voluminous carbonate platforms around Xisha Uplift since early Miocene. Likewise, Shenhu Uplift to the northeast of Qiongdongnan Basin was no more sediment supplier under the transgressional sedimentary environment during this time.

Nevertheless, QW1-2 and QE-2 within Pliocene strata show a larger difference than their upper Miocene samples on U–Pb combinations, indicating both temporal and spatial provenance shiftings. Yanshanian and Indosinian populations become dominant implying increased input from Hainan Island. By contrast, Caledonian clusters have been weakened into a relatively smaller scale for Pliocene samples, suggestive of decreasing trend in Central Vietnam supplement after late Miocene. Overall, age spectra become wider for both samples with higher abundance in Lvliangian clusters. Implication could be made that potential sources with more old Proterozoic zircons were also supplying sediments into Central Canyon. From west to east, QE-2 dominated by major peaks representing Yanshanian and Indosinian events is totally distinctive from QW1-2. This further confirms greater significance of Hainan Island than Central Vietnam when sediments were being transported eastwards along the entire course.

REE patterns tracing mafic-to-ultramafic sediment signals also display an evident spatial variation within the whole Central Canyon (Fig. 4). Since QW3-2 at the original point lies rather close to the Vietnamese margin, continual and abundant basic supplements were being dumped in this area resulting in intensive Eu enrichment from late Miocene to Pliocene. However, for the farther east well QW1, Hainan Island casted its influence on sample QW1-2 with more acidic sediments, hence modest-to-slight positive Eu anomalies even negative values were initially generated. As for the easternmost borehole of QE, basic material signal might have already been diluted to such a large extent that only one average data (3070–3168 m) of slight positive Eu anomaly was observed accompanied by other apparent negative values during late Miocene. Since Pliocene, Hainan Island exerted a dominant influence on QE-2 with large abundance of acidic transport that its REE patterns have all been replaced by obvious negative Eu anomalies. To make it more explicit, being situated at the western margin of Qiongdongnan Basin, QW2 might have been initially influenced by Hainan Island since late Miocene due to its special location leading to the dominance of negative Eu anomalies. Based on this unbalanced distribution pattern of positive Eu anomalies, we could further confirm that provenances surrounding the Central Canyon–Central Vietnam or Hainan Island as major potential sources—have been providing mafic-to-ultramafic or acidic sediments with different abundance largely at a temporal as well as a spatial scale (Fig. 6).

Conclusion

Constraints from detrital zircon U–Pb chronological analysis and REE geochemical results together generated a well-defined relationship between potential provenances and upper Miocene–Pliocene sediments in the Central Canyon. The collision orogenic belt between Indochina block and South China continent located in the Kontum Massif of the Central Vietnam has exerted significant influence on the sedimentary infilling processes of the Central Canyon in the northern SCS. The typical wide U–Pb spectra are identified by broad Indosinian, Caledonian and Jinningian clusters in the western studied samples. Correspondingly, the ophiolite suites generated during the late Triassic period has supplied mafic-to-ultramafic sediments into the Central Canyon course resulting in positive Eu anomaly signals. On the other hand, as another crucial potential contributor, Hainan Island featured by bimodal U–Pb combination also supply eastern samples with abundant terrigenous input, resulting in dominant Yanshanian and Indosinian peaks over other older-age zircon groups. In addition, different from basic sediment supplier of Central Vietnam, Hainan Island has dumped large abundance of acidic materials into its surrounding areas as well as the further western part of Qiongdongnan Basin. As facilitated evidence, REE geochemical results tracing the entire channel provide a clearer explanation of the spatial changes in basic sediment signal distribution patterns. Conclusively, in the western region of Central Canyon, its sedimentary infilling processes have been largely under control by the provenance of Central Vietnam. By contrast, the eastern Central Canyon has gradually tended to be affected from Hainan Island after late Miocene. Overall at a time scale, for the sedimentary evolution of Central Canyon even the entire northwestern SCS basins, it is also implied that influences of Central Vietnam have been remarkably weakened and stepwise replaced by Hainan Island to the north.