77 surface sediment samples and three core samples, A6, A28 and A34, were used in this analysis. The samples were recovered from the NER during the 2019 and 2020 voyages of the South China Sea Bureau of the Ministry of Natural Resources (see Figure 1 for the locations of the sampling points).

Particle size analyses were conducted on the three core samples. The pretreatment process was as follows: an appropriate amount of sediment was extracted every 1 cm downcore, 15 ml of 30% hydrogen peroxide was added, and the samples were allowed to stand for 24 h to remove any organic matter. Once the reaction was complete, approximately 5 ml of 3 M HCl were added to the samples which were then left for 24 h to acidify carbonate material. After the reaction was completed, the samples were centrifuged and then sonicated to mix the remaining sedimentary particles. Grain size was then measured in the School of Marine Science, Sun Yat-sen University by a Beckman Coulter (LS13320) laser diffraction particle size analyser with an analytical range of 0.4-2000 µm. The relative error of repeated measurements was less than 3%.

The Sr-Nd isotopes were measured from all of the 77 surface samples and the three sediment cores. After drying, grinding and sieving, the samples were sent to Nanjing FocuMS Technology Co. Ltd. for measurement of the Sr-Nd isotopic ratios. The organic carbon and carbonate in the sediments were removed by H₂O₂ and acetic acid, and then the samples were digested with concentrated HNO₃ and HF. The sample solution was eluted by a Biorad AG50W-X8 cation exchange column, then the matrix elements and Rb were removed by 2.0mol/L HCl, then the Sr components were received by leaching with 2.5mol/L HCl, and finally the total rare earth components were received by leaching with 6.0mol/L HCl. The SR component of the previous step was further purified by SR special resin. In the purification of Nd, Sm and Nd were separated by LN exchange column. The internal calibration ratios used for mass fractionation during the analysis were, ⁸⁶Sr/⁸⁸Sr = 0.1194 and ¹⁴⁶Nd/¹⁴⁴Nd = 0.7219. NIST SRM 987 and Nd isotope JNDI-1 were used as external calibration standards to account for instrument drift.

Qualitative tests of clay minerals were carried out on the sediments of A6, A28 and A34 at different downcore depths. Pretreatment before testing was done according to the following procedure (Liu et al., 2003; Trentesaux et al., 2003). To separate and extract the clay minerals, first dry a sample of the sediment, add a specific proportion of distilled water for soaking and rinsing, let stand, and remove the supernatant. Next add more distilled water, stir to suspend the sediment, let stand, and get a certain volume of the suspension. Due to the low clay mineral content of the samples, repeat the extraction process as needed. Once the suspension settles and the supernatant is removed, the remaining mud is the clay material needed for the analysis. Transfer the mud into the drying oven at 60 °C for 12 h. Once dry, grind the sample into a powder. Finally, sift the samples through a 2 μM sieve to concentrate the clay minerals. Saturate the samples with ethylene glycol for 24 hours at room temperature. Then heat the samples at 490°C for 2 h. Once the samples had been properly prepared, they were subjected to three X-ray powder diffraction (XRD) runs. XRD patterns were obtained using a d/max 2500 PC diffractometer with CuK radiation at a voltage of 40 kV and a current intensity of 100 Ma. Diffraction patterns (2θ) were scanned from 5° to 30° in steps of 0.02°. The mineral composition was identified by jade software and the PDF2 database.

Accelerator mass spectrometry (AMS) ¹⁴C dating analyses for the three sediment cores were done in the Beta laboratory, and the results were corrected for δ¹³C fractionation to obtain the conventional radiocarbon age. The experimental process was as follows: Appropriate amount of sediment samples were placed in a clean beaker, dried at 50°C, appropriate amount of mill - Q water was added, and the samples were fully soaked to disperse completely. Then transfer the sediment samples to a 63 μm stainless steel sieve and rinse samples under distilled water. Brush lightly with a brush until the sample is completely rinsed. Transfer the remaining sample in the sieve to a small beaker for drying and weighing. The uniform size of the foraminiferan G. menardii shells were selected under a solid microscope, and 3% H₂O₂ was added to remove the organic matter. After ultrasound, the turbidities were removed and the samples were dried. Finally, the G. menardii shell was collected into the plastic pipe and was performed ¹⁴C dating on Ac ⁃ celerator Mass Spectrometry (AMS) in Beta laboratory. The conventional radiocarbon ages were then calibrated with CALIB 7.2 software using a marine calibration curve to account for the marine reservoir effect (Liu et al., 2016a) and to convert the ages into the correct calendar age.

We analysed the lithology of three sediment cores by particle size analyse experiments ( Figure 2 ).

Core A6: The core is about 28 cm long. The colour of the sediment core is black from 1-2 cm, brown from 2 ~ 17 cm and gradually transitioning to black from 17 ~ 28 cm. The particle size measurement results show that the median particle size of the core varies from 6 to 48 μm, and the mean particle size is from 21.2 to 70.6 μm. The sediment composition of the core is clay and silty mudstone with clay minerals (<4 μm) and silty sand (4 ~ 63 μm) were 26.4% and 54.5%, respectively. The proportion of sand is less, the average content with sand size (> 63 μm) is 19.1%. The proportion of clay minerals decreased gradually from the bottom to the top of the core, while the sandy sediment showed an increasing trend. The average and median particle size suddenly increased at a depth of 9 ~ 10 cm ( Figure 2 ).

Core A28: The core is about 45 cm long. The colour of the sediment core is white with a little black mixed in 1 ~ 3 cm, black on top and gradually lightened down to brown in 3 ~ 45 cm. The median particle size of the core is 21.4 ~ 73.6 μm, and the mean particle size is 39.6 ~ 96.1 μm. The sediment composition of the core is clay and silty mudstone with clay minerals (< 4 μm) and silty sand (4 ~ 63 μm) are 11.5% and 50.8%, respectively. The proportion of sand is larger than that of clay, and the average content with sand size (> 63 μm) is about 37.7%. The proportion of clay minerals increases gradually from the bottom to the top of the core, while the proportion of sandy sediments decreases.

Core A34: The core is about 40cm long. The colour of the sediment core is dark grey in 1 ~ 9 cm and brown in 9 ~ 40 cm. The median particle size of the core ranges from 5.1 to 10.3 μm, and the mean particle size varies from 15.1 to 48.1 μm. The sediment composition of the core is clay and silty mudstone with clay minerals (< 4 μm) and silty sand (4 ~ 63 μm) are 36.1% and 51.9%, respectively. The proportion of sand is less than that of clay, the average content with sand size (> 63 μm) is about 20%. The proportion of clay minerals in the core increases gradually from the bottom to the top of the core, while the proportion of sandy sediment decreases. The core contains a sudden increase in average and median particle size at a depth of 28 ~ 29 cm.

The Sr-Nd isotopic ratios for the 77 surface samples from the study area ( Table 1 , the duplicate sites in this table are two groups of parallel data, and the first data is used in mapping) show that the ⁸⁷Sr/⁸⁶Sr value ranges from 0.70919 to 0.71042, with an average value of 0.70951. This is close to the Sr isotope ratio of seawater (0.70924) and higher than the 0.7025 Sr isotope value for MORB (mid-ocean ridge basalt) (Goldstein et al., 1988). ¹⁴⁴Nd/¹⁴³Nd values ranged from 0.512042 to 0.512270, with an average of 0.512165, corresponding to ϵ Nd values. The ϵ Nd value varies from -11.631990 to -7.172703, with an average value of -9.238031, which is similar to that in the Indian Ocean ϵ Nd (- 7 ~ - 9) (Deng et al., 2012). Overall, the changes in isotopic ratios occur within a small range, with a positive (⁸⁷Sr/⁸⁶Sr) value and a negative ϵ Nd value.

The Sr-Nd isotope ratio results of the cores from the three NER stations, A6, A28 and A34, show that the core sediments and ¹⁴⁴Nd/¹⁴³Nd values vary little with depth. The ⁸⁷Sr/⁸⁶Sr ratio values fall between 0.7094321 and 0.7117061, with an average of 0.709937. The ratios of ¹⁴⁴Nd/¹⁴³Nd show minimal variability between 0.5121194 and 0.5122696, with an average of 0.512197, corresponding to the ϵ Nd range of -10.116300 to -7.1863576 ( Table 2 ). The Sr-Nd ratio from the cores is similar to that of the surface samples, but the ⁸⁷Sr/⁸⁶Sr and ¹⁴⁴Nd/¹⁴³Nd ranges are slightly greater for the surface samples.

The analysis of mineral components less than 2 μM in cores A6, A28 and A34 of core samples shows that the main diffraction peaks are obvious in the measurement curves of natural slices of clay minerals, and it can be identified that the components are mainly montmorillonite, illite, kaolinite, chlorite and a small amount of quartz, of which illite is the main component, chlorite and kaolinite are the second, and the diffraction peaks of montmorillonite are not obvious ( Figure 3 ).

¹⁴C enters marine sediments after a series of long-term cycles and decays at a constant rate (half-life: 5730 years). In the sedimentary layer, when the burial depth increases, the ¹⁴C index decreases (Zhang and Zhou, 2012). The maximum age of the three sediment cores in the study is 35000 years before present, and the overall sedimentation rate is low, with an average of 1 cm ka^-1 ( Table 3 ). The deposition rate accelerated at 12 ~ 18 ka Bp. The deposition rate of core A28 peaked at 25 cm ka^-1, but was followed by an extremely rapid decrease ( Figure 4 ). Visual analysis of the core showed that the sediment colour at this interval (19 ~ 20 cm depth) is significantly darker, represented by a dark sediment layer as shown by Figure 2 . This abrupt change in the rate of sediment deposition corresponds to the Younger Dryas (12.9 ~ 11.5 ka BP).

The Sr-Nd isotopes have a long half-life and can maintain relative stability as they undergo various geological processes (Huang et al., 2005). These isotopes have become a good tracer for determining sediment sources. It is generally believed that when the ⁸⁷Sr/⁸⁶Sr ratio in the sediment is > 0.710, the material source is considered to be from the crust; when ⁸⁷Sr/⁸⁶Sr is < 0.705, the material likely originated from the mantle (Fang et al., 2020). ¹⁴³Nd/¹⁴⁴Nd in the continental Erosion Products = 0.51204 ± 0.0002( ϵ Nd = − 11.4 ± 4), and the ¹⁴³Nd/¹⁴⁴Nd ratio of continental crust is close to 0.5119 (DePaolo, 1988; Goldstein et al., 1984). Therefore, sediments flowing through older crustal rocks usually have higher ⁸⁷Sr/⁸⁶Sr ratios and lower εNd values (< -6) (De Paolo and Johnson, 1979), such as the low εNd values (-14 to -15) and the high ⁸⁷Sr/⁸⁶Sr ratio (0.710 to 0.714) of the North China Dabie complex, that indicates the hybridization of the ancient crust (Jahn et al., 1999).

The obvious differences between different rivers around the Bay of Bengal make it a good location to trace sediment sources (Galy and France-Lanord, 1999; Kessarkar et al., 2005; Tripathy et al., 2014; Joussain et al., 2016). By analysing the Sr-Nd isotope ratios, the εNd values previously obtained from the river sediments around the Bay of Bengal, and the corresponding marine sediments, it is found that the distribution of Sr-Nd isotopes in the marine sediments is consistent with the potential terrigenous source of the river system that transport materials into these areas (Tripathy et al., 2011). For example, the ⁸⁷Sr/⁸⁶Sr value of the Eastern Bay of Bengal (Andaman Sea) is 0.712~0.722, which is close to the ⁸⁷Sr/⁸⁶Sr value of the middle Irrawaddy River (Colin et al., 1999), while the ⁸⁷Sr/⁸⁶Sr value of marine sediments in the west of the Bay of Bengal is 0.723~0.732, which is similar to the ⁸⁷Sr/⁸⁶Sr value of sediments in the Godavari River on the central and western coasts (Colin et al., 1999; Ahmad et al., 2005; Ahmad et al., 2009; Chaitanya et al., 2021).

Our Sr-Nd isotope test results ( Tables 1 and 2 ) show that the surface sediments of the NER and the sediments from the three sediment cores all have low εNd values (- 11.631990 to -7.172703) and high ⁸⁷Sr/⁸⁶Sr ratios (0.70919 to 0.71042), indicating that their source rocks contain a large amount of ancient crustal materials.

The average value of ⁸⁷Sr/⁸⁶Sr of the surface sediments obtained from the NER is 0.7095, and the average value of ¹⁴³Nd/¹⁴⁴Nd is 0.5122, indicating that a large number of diagenetic materials in the sediments come from continental crust sources and have been eroded. It can be seen from the correlation distribution diagram of εNd and ⁸⁷Sr/⁸⁶Sr ( Figure 5 ) that the ⁸⁷Sr/⁸⁶Sr values of the surface sediments and the core sediments are very close to the ⁸⁷Sr/⁸⁶Sr range of 0.712~0.719 (Colin et al., 1999), indicative of the Irrawaddy River, whose provenance is the Qinghai Tibet Plateau. These values are also similar to the value of the Eastern Bay of Bengal (Andaman Sea) ( Figure 5 ). The average value of εNd for the surface sediment in the study area is -9.238, which is also within the range of the Irrawaddy River (- 6.9 ~ 10.86) ( Figure 5 ). Only a small amount of surface sediments have εNd and ⁸⁷Sr/⁸⁶Sr values suggestive of an origin in the Ganges Brahmaputra River system close to the northern and southern foothills of the Himalayas.

The research method of Sr-Nd isotopes combined with clay mineral analysis is widely used. The fine-grained, clay dominated sediments are mainly transported by rivers, and there are obvious differences in the clay mineral composition representative of potential sources. Therefore, the study of clay composition within fine-grained sediments can be used as important evidence to determine sedimentary sources and transport pathways (Ahmad et al., 2012; Liu et al., 2016b). On the other hand, the formation and diagenesis of clay minerals are closely related to the environment in which they are located, and the effects of these processes may be reflective of local environmental conditions (Tang et al., 2002).

Various sediment inlets in the Northeast Indian Ocean have different clay mineral assemblages. The clay minerals in the surface sediments of the southeastern Andaman Sea are mainly composed of illite (45%), kaolinite (31%), chlorite (18%), and smectite (6%) (Cao, 2015). The sediments of the Mahanadi River are characterized by high illite content and low montmorillonite content (Li et al., 2018). The clay mineral contents in the sediments of the Ganges-Brahmaputra River system are illite (57%), kaolinite, chlorite and montmorillonite (less than 5%) from high to low (Khan et al., 2019). The sediment in the offshore basin near the Krishna-Godavari estuary has a high content of montmorillonite and a low content of chlorite (Phillips et al., 2014). Although the runoff and sedimentary input of the Ganges-Brahmaputra-Meghna River system is much larger than that of other basins (Milliman et al., 1983), Himalayan materials from the Irrawaddy River and Yarlung Zangbo River are more important and stable sources (Song et al., 2021). Illite is the predominate material in the clay mineral composition of the NER, followed by chlorite and kaolinite. Montmorillonite is the least abundant ( Figure 3 ). The clay mineralogy of the NER is similar to that of the surface sediments of the Andaman Sea, indicating a shared provenanced, though small inputs from the Ganges Brahmaputra River and Mahanadi River affect the sedimentary source.

To summarize the Sr-Nd isotope and clay mineral analysis, the sediments in the study area contain a large amount of older continental crust materials from the Qinghai-Tibet Plateau, which have been subjected to certain erosional processes and may have been imported to the NER via the Irrawaddy River and the Andaman Sea.

A large volume of surface sediments encompassing a wide area were collected in the NER. The Sr-Nd isotope test results can not only be used for provenance tracing but also to study the distribution pattern of terrestrial material after it enters the study area. The distribution of the εNd (0) value and ⁸⁷Sr/⁸⁶Sr isotope ratios of surface sediments are an important indication of the current source input pathways. By analysing the Sr-Nd isotope distribution characteristics of 77 surface samples obtained from the NER, we found that the ¹⁴³Nd/¹⁴⁴Nd isoline map ( Figure 6A ) showed that the values decreased from south to north. The ⁸⁷Sr/⁸⁶Sr contour map ( Figure 6B ) shows a decreasing trend from east to west.

Sedimentation has little influence on Nd and its ratio in sediments. Nd isotopes can effectively identify and trace terrigenous sedimentary materials and indicate their evolutionary history (Shao et al., 2009). The εNd value of the materials imported into the sea from the land is smaller than that of marine biomass (Liu, 2004). According to the distribution pattern that the ¹⁴³Nd/¹⁴⁴Nd ratio in the study area gradually increases from north to south, it is inferred that the material sources are mainly transported from north to south. The larger the ratio of ⁸⁷Sr/⁸⁶Sr is, the older the sediments and the greater the chemical weathering (Meng et al., 2000). As the ⁸⁷Sr/⁸⁶Sr value decreases from east to west, the chemical weathering in the east of the NER is greater than that in the west. This may be because the monsoon weakens after it passed through the ridge, the hydrodynamic effect of the seawater carrying terrestrial materials was weakened, thus reducing its ability to further transport sediments from east to west. Therefore, the migration mode of the sediment transported to the NER is from east to west and from north to south.

The NER is relatively high and located in the open ocean far from the source area. Turbidity currents and isobaric currents have little impact on it. The isotope and clay mineral analysis results show that the winter monsoon and summer monsoon in the Bay of Bengal can transport the terrigenous sediments of the northern mountain system to the northern end of the NER (Wei et al., 2007; Zhang et al., 2007; Weber et al., 2018), but it is difficult to transport them further southward. However, after the sediments exported from the Irrawaddy River flow through the relatively closed Andaman Sea via the summer monsoon, the monsoon-driven circulation in this area can be used to transport more sediments to the northern NER (Schott and McCreary, 2001; Shankar et al., 2002; Joussain et al., 2016). This interpretation is supported by the observed decrease in ¹⁴³Nd/¹⁴⁴Nd and ⁸⁷Sr/⁸⁶Sr values from northeast to south and east to west, as shown in the isotope contour maps ( Figure 6 ).

Therefore, we comprehensively analysed the Sr-Nd isotopes and their spatial distribution and clay mineralogy in the study area and proposed the following transport mode of terrigenous materials into the NER. In the Qinghai-Tibet Plateau, a large volume of continental crust materials was weathered and denuded and then transported into the Irrawaddy River. Terrigenous clastic materials were carried to the Andaman Sea with the flowing waters of the Irrawaddy River. It is transported westward to the north of the NER through the surface monsoon circulation of the Andaman Sea. Under the influence of the monsoon circulation in the study area, the terrigenous materials continue to be carried southward along the NER ( Figure 7 ), and the Sr-Nd ratio remains relatively stable along the entire transport path.

To explore the temporal and spatial changes in this transport mode of sediments to the NER, we performed AMS ¹⁴C dating analysis on three core sediments.

The maximum ages of the A6, A34 and A28 cores are 18ka, 19ka and 35ka respectively. A6 and A34 cores in the northern part of the NER recorded the sedimentary and environmental changes after the last glacial maximum (LGM, 26.5 ka ~ 19 ka BP), including the Younger Dryas (12.6 Ka ~ 11.5 ka BP). The two cores reflect similar sedimentation rates, and there is no significant change during the Younger Dryas (12.6 Ka ~ 11.5 ka BP). After the Younger Dryas (12.6 Ka ~ 11.5 ka BP), the sea level rose and the sedimentation rates began to decrease, suggesting that the northern part of the NER may be mainly affected by sea level changes. The deposition of A28 core in the southern NER records the sedimentary and environmental changes since the late period of the last glacial period (70 ka ~ 11.5ka BP), including the LGM and the Younger Dryas. AMS ¹⁴C dating analysis of the core shows that the sediments in the southern NER were stable from 35ka BP to 15ka BP and were not affected by the LGM, indicating that the southern NER was not affected by provenance and sea level during this period. From 15ka BP to 12.6ka BP (before the Younger Dryas), A sudden increase in the sedimentation rate in the southern NER could be linked to a number of factors. The first factor may be that as the climate warms, the increase of provenance denudation and stream input amount during this period (Kolla and Pierre, 1973; Kolla et al., 1976). And the second reason may be that large amounts of volcanic material from the Indonesian archipelago were deposited, carried by the southeast monsoon or equatorial easterly winds (Dehn et al., 1991; Hovan and Rea, 1992; Zhang et al., 2004). Since the Younger Dryas (12.6ka BP), the sedimentation rate of the southern NER decreased, which may be affected by the decrease of provenance denudation and the stream input amount.

We also found that the mean particle size of the A6, A28 and A34 cores are relatively uniform, but there is a sudden change in sedimentation during the period of rapid climate change (12.9 ~ 11.5 ka BP) in the Younger Dryas (Broecker et al., 2010; Liu, 2012). The deposition processes at the NER may also be affected by this climatic change. The Sr-Nd isotope values of the A28 and A34 core samples had obvious changes after the Younger Dryas, especially the εNd values, which decreased rapidly, confirming the change in source ( Figure 8 ). Core A6, located west of the ridge, differs from the other two cores potentially due to their different locations (east of the ridge). However, A6 also contains evidence of a change at approximately 15 ka BP, after the Younger Dryas. In general, the sediment transport regime in the study area has been relatively stable for the past 35000 years and is essentially the same as the current model. However, during the Younger Dryas (12.9 ~ 11.5 ka BP), the quantity of sedimentary input, particle size and Sr-Nd isotope ratio changed. The relationship between this change and the climate of the Younger Dryas needs to be further studied.