The AMS shows that the minimum susceptibility axes (κ ₃ ) of most samples are perpendicular to the bedding plane, while the maximum (κ ₁ ) and medium (κ ₂ ) susceptibility axes lie in the bedding plane (Figure 5A). Plots of shape parameter (T) versus anisotropy degree (P), and lineation (L) versus foliation (F) show a dominance of the oblate form (Figures 5B,C). These results indicate that most samples have undisturbed sedimentary fabrics.

Magnetic components were calculated by principal component analysis (PCA) (Kirschvink, 1980) using the software PaleoMag (Jones, 2002). A component with the steep positive inclination can be removed before 30 mT or 250°C (Figure 6), which is most likely an overprint induced by the coring process that was identified in many paleomagnetic studies on the ODP/IODP cores (Acton et al., 2002; Tauxe et al., 2012; Zhao et al., 2013). The characteristic remanent magnetization (ChRM) was identified through the stepwise AF demagnetization with a maximum field of 80–120 mT or thermal demagnetization (TD) up to 600°C.

For samples from IODP Site U1501, polarity ambiguity arises because the samples are azimuthally unoriented and the expected inclination ∼25.5° is relatively shallow. We applied the following methods to establish the magnetozones: (a) positive/negative inclination (±37.3 ± 6.6°) corresponding to a paleolatitude of ±20.8 ± 3.3°N is interpreted as normal/reversed polarity, respectively. (b) polarity reversal is identified by near 180° shift in declinations on long coherent core sections. In order to obtain reliable magnetic polarity zones, the samples with the following scenarios are discarded from the magnetostratigraphic interpretation: (a) greigite as the dominated magnetic carrier, (b) inclination of κ ₃ axes of the AMS less than 60°, and (c) maximum angular deviation (MAD) larger than 15°. A succession of 12 normal and 11 reversed events are recognized, while some suspected reversed events with only one or two samples are also shown (Figure 2H).

The sedimentary succession of Hole 1501C is characterized by a hiatus at 103.63 m (section U1501C11H6W, 135 cm), marked by significant discontinuity and downhole color change based on the core section image (Figure 2I) and spanning approximately 10.5–9.0 Ma (Wang and Jiang, 2020). Assuming there are no other significant hiatuses, an initial magnetostratigraphic scale (Table 1; Figure 2) was constructed by correlating observed polarity sequences with a reference GPTS in GTS 2020 (Ogg, 2020), constrained by shipboard and post-expedition biostratigraphy (Larsen et al., 2018; Wang and Jiang, 2020) and δ¹⁸O isotopic record (Zhang et al., 2020).

The base of magnetozone N1 correlates with the base of Chron C1n (Brunhes) (Figure 2H), according to shipboard paleomagnetic results (Larsen et al., 2018). The top of magnetozone N5 has a date of 1.36–1.88 Ma based on the shipboard biostratigraphy (last appearance of Calcidiscus macintyrei (1.6 Ma) and Globigerinoides fistulosus (1.88 Ma) at 38.08–46.95 m; Larsen et al., 2018) and δ¹⁸O isotopic record (42 m, 1.36 Ma; Zhang et al., 2020), thus correlates with the top of C2n (Olduvai). Uncertainty polarities below the N5 make us tend to connect the base of N5 and C2n (Olduvai), which is slightly younger than the shipboard magnetic age located at the base of C2r.1n (Feni). According to shipboard paleomagnetic results, the base of polarity zone R5 correlates with the base of C2r.2r (Matuyama). A series of biological events including the last appearance of Triquetrorhabdulus/Orthorhabdus rugosus (81.74 m, 5.28 Ma), Discoaster quinqueramus (83.09 m, 5.59 Ma) and Nicklithus amplificus (87.65 m, 5.94 Ma), and the first appearance of Ceratolithus acutus (84.77 m, 5.35 Ma) and Ceratolithus atlanticus (84.77 m, 5.39 Ma), display the age of sediment from 82–88 m is 5.28–5.94 Ma (Wang and Jiang, 2020), supporting the correlation of magnetozones N6–N8 to Chrons C2An.1n (Gauss)–C3n.2n (Nunivak). Finally, polarity zone R8 correlates with Chron C3r (Gilbert). Below 85.04 m, dominant normal polarities hamper the determination of magnetic polarities.

The NGR series in the 0–66 m interval has been converted to the time domain (Figure 7A) based on the initial paleomagnetic results including the base of N1 (Chron C1n) and N5 (Chron C2n) and the top of N6 (the top of Chron C2An.1n) and oxygen isotopic records (Figure 2H). Zhang et al. (2020) obtained the planktonic foraminiferal oxygen records (Globigerinoides ruber) of the upper 46 m of Site U1501 (Figure 8A). However, the establishment of their age model was based on the shipboard biostratigraphy model, which has a discrepancy with paleomagnetic results here. Therefore, this study adopts the age-depth points until 18 m, where the feature of the δ¹⁸O curve is largely compatible with the LR04 stack (Lisiecki and Raymo, 2005). Then age points including MIS 22, 36, 37, and 54 were determined based on oxygen isotope features around the boundary of polarity zones of C1n, C1r.2n, and C2n (Figure 8A).

The time-domain NGR spectrum (Figure 7B) has prominent peaks at ∼400, 125, ∼100 and 40 kyr above 90% confidence level, which are close to the value of Milankovitch periodicities of 405-kyr long eccentricity, ∼124, 99-kyr short eccentricity and 40-kyr obliquity, respectively, (Laskar et al., 2004; Laskar et al., 2011). The evolutionary spectrum shows significant short eccentricity and obliquity cycles since 0.4 Ma (Figure 7C). Robust long eccentricity cycles and short eccentricity cycles are displayed during 0.4–0.8 Ma. Short eccentricity cycles are dominant before 0.8 Ma.

Dominant short eccentricity components were extracted using the Taner bandpass filtering and tuned to the La2010 astronomical solution (Figure 8A; Laskar et al., 2011). The high NGR values reflect organic matter and clay-rich sediments, which are sensitive to seasonal precipitation, fluvial input, and current delivery. These East Asian hydrological cycles are modulated by precession and eccentricity (Wang et al., 2019; Wang, 2021). Therefore, we make a peak-to-peak correlation between the La2010 orbital eccentricity cycles and the NGR cycles, assuming there is no time lag between these two series. During the tuning process, the correlations of the δ¹⁸O records of Site U1501 and the LR04 stack (Lisiecki and Raymo, 2005) provided extra constraints during the 0–0.4 Ma and 1.3–1.7 Ma intervals, causing the slight offset <50 kyr between the NGR cycles and La2010 orbital eccentricity cycles.

The sedimentation rate in the 66–103.63 m interval is relatively low (∼0.6 cm/kyr) based on the paleomagnetic results and the biostratigraphic framework (Figure 2H). The depth-domain NGR was transformed to a time-domain (Figure 7G) using the initial age constraints of the magnetozone R8 (Chron C3r) and post-expedition biostratigraphic datum including the first appearance of Discoaster berggrenii (96.25 m, 8.29 Ma) and the last appearance of Discoaster bollii (102.24 m, 9.21 Ma). These anchor points yield a stable sedimentary rate. The 2π MTM spectra and evolutionary FFT power spectrum analysis on the initial age constraint NGR series shows ∼1.3 Myr, 250, 160, 125 and 100 kyr cycles. The ∼1.3 Myr periods are consistent with ∼1.2 Myr obliquity amplitude cycles that originate from the precession of nodes of Earth and Mars (Laskar et al., 2004). Despite the fact that the 250 kyr cycle was identified by Hinnov (2000) without a specific origin, here it may be linked to the interaction between the eccentricity and obliquity signal. The weak signal of short eccentricity in the NGR profile is attributed to the resolution of time-domain NGR data being too low to recover ∼100-kyr scale behavior adequately.

The orbitally tuned age model for 66–103.63 m was constructed in three steps. First, the 1.3-Myr-filtered NGR profiles were roughly tuned to 1.2-Myr obliquity amplitude cycles (Supplementary Figure S2A). The resulting NGR profiles (Supplementary Figure S2B) display a stronger 405-kyr long eccentricity signal in the 2π MTM spectra and evolutionary FFT power spectrum analysis (Supplementary Figures S2C,D). Second, using the 1.2-Myr-tuned age model, the 400-kyr-filtered NGR data was tuned to the long eccentricity of the La2010 solution (Supplementary Figure S2E; Laskar et al., 2011) to adjust the 1.2-Myr-tuned age model. Finally, the tuning was done by correlating the extracted short eccentricity signal of the resulting NGR data to the orbital eccentricity of the La2010 solution (Figure 8B).

The biostratigraphic framework (Figure 2H; Wang and Jiang, 2020) displays a steady accumulation rate of ∼1.25 cm/kyr in the 103–160 m interval. The post-cruise biostratigraphic datum provides the following anchor points: the first appearance of Catinaster coalitus (105.75 m, 10.89 Ma), the base of the common occurrence of Discoaster kugleri (117.05 m, 11.90 Ma), the top of the common occurrence of Discoaster kugleri (113.61 m, 11.58 Ma), Calcidiscus premacintyrei (123.55 m, 12.38 Ma) and Cyclicargolithus floridanus (134.55 m, 13.28 Ma), and the last appearance of Coronocyclus nitescens (120.35 m, 12.12 Ma) and Sphenolithus heteromorphus (138.15 m, 13.53 Ma). The 2π MTM spectra and evolutionary FFT power spectrum analysis (Figures 7N,O) on the time-domain NGR series (Figures 7M) reveal ∼400, 125 and 90 kyr cycles. The short eccentricity periods have been extracted from the time-calibrated NGR series and tuned to the orbital eccentricity target of the La2010 solution (Laskar et al., 2011; Figure 8C). The 2π MTM spectra and evolutionary FFT power spectrum analysis on the resulting NGR series reveal a strong long eccentricity signal (Figures 7P,R).

Our tuning results provide a new timescale spanning 15.54 Myr. During 10.9–15.54 Ma, the sedimentary rate is stable at ∼1.25 cm/kyr. The duration of the hiatus located 103.63 m is estimated as ∼1.2 Myr. There is a relatively lower sedimentary rate between 2.84 and 9.8 Ma, followed by relatively higher and oscillating sedimentary rates in the 0–0.23 Ma and 1–2.84 Ma intervals (Figure 2J). The variation of sedimentation rate of Site U1501 suggests complicated sedimentary processes in the northern SCS.