The Bayesian age-depth model was applied to the OSL ages above 9.6 m (U2-U3) using the R package “rbacon” (Blaauw and Christen, 2011). During the calculation, samples were divided into 0.25 m thick sections. Given the evident change in sedimentation rate at the boundary between U2 and U3 (located at 8.4 m), the accumulation rates were set at 0.35 ka/m for U2 and 3.5 ka/m for U3, respectively ( Figure 2b ).

A total of 14 samples (Z01–Z15, except Z13) were selected from core DZK3, next to the OSL samples, with a width of 5 cm for particle-size analysis. During the experiment, 0.3 – 0.5 g of each sample was pretreated with 30% H₂O₂ and 10% HCl to remove organic matter and carbonates, respectively. To ensure proper dispersion of the sediment size, the samples were added with an appropriate amount of (NaPO₃)₆ solution before measurement (Lu and An, 1997; Wang et al., 2024a). The particle size distribution was then analyzed using the Malvern Master sizer 3000 laser diffraction instrument, which has a measurement range of 0.24 – 4000 μm, at Shantou University.

Preheat plateau tests and dose recovery tests (with a given dose of 24.8 Gy, test dose 4.96 Gy) were performed on sample DZK3-G05 to determine the optimal preheat temperature and evaluate the accuracy of the SAR procedure (Murray and Wintle, 2003). Preheat temperatures of 200, 220, 240, 260, 280, and 300 °C were tested, with the corresponding preheat temperatures for test doses being 40 °C lower, and four aliquots tested for each temperature. A stable preheat plateau was observed between 200 °C and 280 °C ( Figure 3a ), as indicated by relatively stable D_e values, recovery ratio, and recuperation within the range of 0.9 – 1.1 ( Figures 3b–d ). Finally, a pre-heat temperature of 260 °C was applied for 10 s to measure both the natural and regenerative dose signals, while 220 °C for 10 s was used to measure the test dose signal (Wang et al., 2024b).

For samples using the SAR-SGC procedure, SGC curves were clustered, and squeezed together with little deviation (the relative standard deviation for each sample at each regeneration dose <10%) ( Figures 4a, b ). A comparison of SAR D_es and SGC D_es shows good agreement, with errors within 10% for each sample ( Figure 5 ). The representative decay and dose-response curves for samples DZK3-G02 and DZK3-G05 are presented in Figures 4a, b . The OSL signals rapidly decayed to background levels within the first 2 s of stimulation, indicating a dominance of the fast component. The dose–response curves of all samples were well-fitted using an exponential-plus-linear function ( Equation 1 ) to establish SGC. The details of the OSL dating results are presented in Table 1 .

Moreover, the relative contributions of the fast, medium, and slow OSL components for sample DZK3-G05 were analyzed using the “numOSL” R package (Peng et al., 2013; Peng and Li, 2017). Figure 6 illustrates component proportions obtained from this analysis. The fast component dominates the initial signal decay (0 – 2 s), exhibiting rapid depletion characteristic of well-bleached quartz. Specifically, the fast component contributes approximately 64% of the initial signal (integrated within the first 0.8 s), while the medium component accounts for about 30%. Although the medium and slow components persist beyond 5 s, their relative contributions to the initial signal intensity remain minor compared to the fast component.

For each sample, the over-dispersion (OD) value was calculated to describe the D_e distribution. The OD values of all samples were between 8 and 15% ( Table 1 ). The Abanico plots (Dietze et al., 2016) of the (D_e) distributions for DZK3-G01 to G06 are illustrated in Figures 4c–h . The relatively low OD values represent a completely bleached history before deposition, so the central age model (CAM) (Galbraith et al., 1999) was applied to all samples in D_e calculation.

Samples of DZK3-G07 and below are saturated. The characteristic saturation dose (D₀) of DZK3-G07 was calculated. For sample DZK5-G07, the single exponential function is [Standardized OSL (I) = 17.5378 (1- exp (- X/113.26)) + 0.18]. D₀ is 113.26 Gy. SAR De is 253 Gy, exceeding 2 D₀ (Wintle and Murray, 2006; Lai and Ou, 2013; Mahan et al., 2022). L_x/T_x of SGC aliquots from samples DZK3-G08 to G12 are inserted to the DZK3-G07 SGC curve and SGC D_e are all over 250 Gy. Consequently, OSL ages from DZK3-G07 to DZK3-G12 are considered as minimum ages.

A chronological framework for core DZK3 was established using OSL samples and a Bacon Bayesian age–depth model based on the Bacon model (Blaauw and Christen, 2011) for the upper 9.6 m ( Figure 2b , see Method 3.2). OSL ages range from >100 ka below 10.2 m to 6.2 ± 0.4 ka at 2.60 – 2.70 m. The dating results were generally consistent with the stratigraphic order ( Figure 2 ). The OSL ages in U4 are considered as minimum ages. Sample DZK3-G07 at 10.25 m yielded an age >100 ka, while the D_e values of DZK3-G08 and deeper samples exceeded 250 Gy, indicating ages of at least 100 ka. Based on lithology (see Section 2), grain size results, and the chronological framework, U4 is interpreted as a transgressive environment, likely developed during MIS 5 or earlier. A long hiatus exists between U4 and U3 with an age gap between >100 ka and 19.8 ka. In U3 (8.40 – 9.60 m), OSL dating of samples DZK3-G06 and DZK3-G05 yielded ages of 19.8 ± 1.2 ka and 10.4 ± 0.6 ka, respectively. The lithology of U3 consists mainly of dark-gray organic-rich clay, indicative of a lacustrine environment. Limnetic layers, which were widely developed in the Laizhou Bay during the early Holocene, are regarded as marker layers at the lower boundary of the Holocene (Yuan et al., 2024; Tian et al., 2016; Xu et al., 2018). The chronology of DZK3 shows that the lake developed earlier than the limnetic layers observed in other cores around the area (Tian et al., 2016; Xu et al., 2018), which is closely related to the regional geomorphology (see Section 5.3). OSL ages from U2 (8.4 – 1.6 m) span from 6.2 to 10.5 ka, indicating that transgressive deposition began at approximately 10 ka. The upper 1.6 m consists of fluvial deposits, but OSL samples were not collected for this section.

The sedimentation rates for U3 and U2 were estimated using the Bayesian age–depth model. During the deglaciation period (U3), the sedimentation rate was approximately 0.12 m/ka. However, this rate increased significantly (approximately 1.5 m/ka) during the Holocene transgression. The differential sedimentation rates since deglaciation are attributed to differences in geomorphology and climatic influence.

The hiatus spanning >100 ka to 19.8 ka in core DZK3 likely resulted from weakened sedimentation caused by strong weathering and enhanced downcutting erosion of rivers under dry and cold climatic conditions. Such hiatuses generally exist in coastal plains or delta areas. Previous studies have identified five sedimentary hiatuses along the coastal plain of the western Bohai Sea since the Late Quaternary, all related to cold climates. These hiatus ages are estimated as 4 – 5.5 ka, 10 – 25 ka, 25 – 29 ka, 29 – 68.5 ka, and 80 – 140 ka, respectively (Li et al., 2004). Two ages of 35.1 ± 5.3 ka (OSL, depth at 16.9 m) and 8.5 ± 0.1 ka (¹⁴C, depth at 16.7 m), obtained for core BT113 on the west coast of Bohai Bay, suggest a long depositional hiatus caused by surface exposure and desertification during MIS 3 to LGM period, when sea levels continuously decreased (Chen et al., 2012).

Based on the study of Li et al. (2004), the depositional discontinuity observed in this study (>100 to 19.8 ka) corresponds to the second, third, and fourth sedimentary hiatuses, dating to MIS 2 – 4. Paleontological studies further indicate that this hiatus occurred in a cold and dry climate, during which sea levels dropped to -120 to -80 m, causing the base level of erosion to decline and limiting sedimentation (Li et al., 2004). The hiatus may also be due to the enhancement of river downcutting caused by sea level decline. Deep river valleys commonly formed in the lower reaches of rivers and estuaries in low sea-level stages, especially during the LGM (Xu et al., 2019; Lai et al., 2025). Similar depositional discontinuities have been observed in other cores along the southern coast of the Bohai Sea. The astronomical chronology of core Lz908 indicates two slow deposits at a rate of approximately 0.02 – 0.05 m/ka, which are regarded as possible discontinuities at 16.6 m (34 ka) and 28.2 m (100 ka), corresponding to the end and beginning of T2 (Yi et al., 2012) and belonging to MIS 3 and MIS 5, respectively. A comparable depositional discontinuity, dating to 37 – 71 ka, was observed in core DZK01 between 11.85 m and 12.65 m (Yuan et al., 2024).

Similar cases of depositional discontinuities during MIS 2 – 4 have been widely reported in delta areas (Xu et al., 2022; Yim et al., 2008). For example, recent studies have found that little or no deposition occurred during ca. 38 – 14 ka in the Yangtze Delta (Gao et al., 2022; Liu et al., 2022). In the Pearl River Delta, a depositional discontinuity of 70–8 ka was identified in core P5 – 4 at 26.35 – 26.75 m (Lin et al., 2023). Additionally, core SXG06 shows discontinuous deposition between 10.55 m and 11.5 m, with an age range of117–35 ka, and the first continental layer (12.5 – 10.5 m) consists of a thin 2 m deposit, indicating weak deposition (Lin et al., 2024).

Coastal limnetic and lagoon sediments of the Heituhu Formation are widely developed along the southern coast of Bohai Sea, with ages ranging from 3210 ± 190 to 11785 ± 160 a BP (Wang et al., 1999). These sediments were underlain by floodplain sediments of the LGM and overlain by mid-late Holocene transgression deposits (Xu et al., 2018). Marker layers of the Heituhu Formation, found at the bottom of the Holocene layer, are commonly observed in cores ( Figure 9 ). For instance, The dark-gray organic-rich layer at 10.2 – 10.8 m in core LZ908 yielded a ¹⁴C age of 8.2 ± 0.05 cal ka BP (10.2 m) and an OSL age of 9.7 ± 0.8 ka (10.7 m) (Yi et al., 2012). The same layer in core BH1 has ages of 9.7 ± 0.8 ka (OSL) and 5.5 ± 0.07 cal ka BP (¹⁴C), and in core BH2, the ages are 5.2 ± 0.08 cal ka BP (¹⁴C) and 9.7 ± 0.8 ka (OSL) (Yi et al., 2013). These results indicate that the limnetic facies in the coastal region were early Holocene sediments. Regarding the evolution of lakes along the southern coast of the Bohai Sea, previous studies have focused on lakes in nearby regions that formed during the historical period (Guo, 1990; Zhang et al., 2004). Many of these lakes, such as Judian Lake and Biehua Lake, evolved from lagoons after transgression at 6 ka (Guo, 1990; Han et al., 2002; Zhang et al., 2003).

The lacustrine layer (U3, 8.4 – 9.6 m) in core DZK3 was dated to 19.8 – 10.5 ka, which is significantly older than the limnetic and lagoon layers in other cores. Additionally, there is a hiatus spanning >100–19.8 ka between the lacustrine position and the underlying layer, where there are no floodplain sediments from the LGM. This indicates that the development of this lacustrine facies layer differs from that reported in previous studies (Xu et al., 2018). Additionally, according to a 10,000-year scale climate and geomorphology model, lake development generally occurred in interfluvial depressions after the river valleys were filled (Xu et al., 2019). This time was estimated to be approximately 12 ka by geomorphological studies of the lower Yellow River plain (Wang et al., 1999; Zhao et al., 2023). Specifically, the lacustrine facies layer developed on top of the paleochannel during the Late Pleistocene to Early Holocene in north of the Luzhong Mountains (Tian et al., 2025). The ¹⁴C ages of these lacustrine layers are 9165 ± 120 a BP, 9515 ± 125 a BP, and 9650 ± 190 a BP, which are consistent with the formation age of the Heituhu Formation (Zhang, 1990; Zhang et al., 2004). However, the lacustrine facies of DZK3 in this study are not consistent with the two cases mentioned above. Therefore, their formation may be attributed to a small water storage depression and shaped by the evolution of local microtopography.

As shown in Figure 9 , the Holocene stratum around the southern coast of the Bohai Sea exhibit three distinct characteristics: (1) The base of the Holocene strata is located at approximately -10 m, and the depth increases from land to sea (e.g., GK95, - 8.9 m; DZK3, - 8.4 m; DZK01, - 11.8 m; BH2, - 13 m; GK111, - 12.5 m; BH1, - 16 m); (2) The onset of Holocene deposition generally dated to approximately 9 – 10 ka (e.g., LZ908, 9.7 ± 0.8 ka; DZK3, 10.5 ± 0.8 ka; DZK01, 9.5 ± 0.6 ka; BH2, 8.9 ± 0.7 ka; BH1, 9.7 ± 0.8 ka); (3) A dark-gray, organic-rich layer (limnetic facies, Heituhu Formation) can be used as a marker for the base of the Holocene (e.g., LZ908, BH1, BH2, and DZK01) (Xu et al., 2018; Yi et al., 2013; Yuan et al., 2024).

Since the Late Pleistocene, three large-scale transgressions (from top to bottom: T1, T2, and T3) have occurred along the coast of the Bohai Sea due to sea-level fluctuations. The extents of these transgressions are shown in Figure 1b (Wang et al., 1986; Yi et al., 2013; Liu, 2017; Li et al., 2019; Yuan et al., 2024). On the southern coast of the Bohai Sea, T2 has the largest range, followed by T1, and T3 has the smallest range (Yuan et al., 2024).

The earlier dating framework for the transgression layers along the coast of the Bohai Sea was primarily based on ¹⁴C method, with the three transgressions (T1, T2, and T3) occurring in MIS 1, 3, and 5, respectively (Wang et al., 1986; Zhao, 1986). The age of T1 is well established. However, because of the impact of the dating cap (Lai and Fan, 2014; Song et al., 2015) and contamination by younger carbon (Yim et al., 1990), the ¹⁴C age of T2 should be treated with caution (Zhang, 2021; Long et al., 2022). Although the theoretical upper limit can reach 50 ka, ¹⁴C ages beyond 30 – 40 ka often lead to inaccurate results (Lai and Fan, 2014; Song et al., 2015; Long et al., 2022). Additionally, ¹⁴C ages for pre-Holocene sediments may have been underestimated due to the pollution of new carbon (Yim et al., 1990). As shown in Figure 9 , for core BH1, the OSL ages of T2 are 92 ± 9 ka and 123 ± 15 ka, whereas the adjacent ¹⁴C ages are 46 ± 4 cal ka BP and >46 cal ka BP. In addition, ¹⁴C-based age models suggest that the distribution extent of the MIS 3 transgression is wider than that of the MIS 5 transgression along the Bohai Sea Coast, which is not consistent with the global sea–level change pattern (Yi et al., 2013; Long et al., 2022). Therefore, the formation time of T2 in the Bohai Sea likely corresponds to MIS 5 or earlier than MIS 3, as previously reported (Long et al., 2022).

On the southern coast of the Bohai Sea, studies using OSL dating also suggest that T2 formed earlier than MIS 3. As shown in Figure 9 , Yi et al. (2013) proposed that T2 on the southern coast of the Bohai Sea developed during MIS 3 – 5 based on ¹⁴C and OSL dating (e.g., cores LZ908, BH1, and BH2). Meanwhile, Liu (2017) suggested T2 corresponded to MIS 5 (core GK111, >99 ± 14 ka). For core DZK3, the OSL ages of samples DZK3-G06 and DZK3-G07 are 19.8 ± 1.2 ka and >100 ka, respectively, and there is a sedimentary hiatus between U3 and U4. Therefore, we suggest that no transgression sediments were deposited in the study area during MIS 3, which is similar to the dating results of core DZK01 (Yuan et al., 2024).

Based on the OSL dating results of core DZK01 and DZK3, T2 occurred at least in MIS 5e (115 – 130 ka), and there were no marine sediments in MIS 5a-d (71 – 115 ka). There are two reasons for this phenomenon: One possibility is that marine sediments have been eroded, but there is still a fluvial unit of MIS 4 – 2 in core DZK01, so it is unlikely; The second reason is that T2 in the southern coastal Bohai Sea only occurred in MIS 5e. However, owing to the saturation of quartz OSL ages and a lack of foraminifera identification, especially in the strata above the reddish-brown silt layer of U4, further investigation is required.