The Bohai Sea covers ~77,000 km² and is characterized as an inland, shallow gulf (Liu et al., 2016a; He et al., 2019). The three major bays within the Bohai Sea (Liaodong, Bohai, and Laizhou Bays) have a water depth of less than 20 m, except for the central Bohai Sea, where depths can reach up to 40 m (Qin et al., 1990). It is connected to the North Yellow Sea through a 104.3 km-wide passage, known as Bohai Strait, where the water depth can reach ~60 m ( Figure 1A ).

The Huanghe originates from the Tibetan Plateau, and flows across the Chinese Loess Plateau and the North China Plain. It is renowned for its substantial sediment load, and annually carries approximately 1.1×10⁹ tons of sediment into the Bohai Sea over the past 2000 years (Milliman et al., 1987; Wang et al., 2007). The total sediment discharge from other rivers around the Bohai Sea is minimal, constituting only 3.6% of the Huanghe’s sediment discharge (Xue et al., 2018).

Tidal patterns in the Bohai Sea are semidiurnal, with average and spring tidal ranges of 1.5-2 m and 3-4 m, respectively, in the western Bohai Sea (Xiong, 2012). Flood tidal currents can reach velocities of up to 5 knots (~2.5 m/s) around the north Bohai Strait (Liu et al., 2016a). The coastal current and oceanographic circulation within the Bohai Sea is predominantly influenced by high-salinity water from the Yellow Sea Warm Current, a branch of the Kuroshio Current. This circulation is counterclockwise in Liaodong Bay and clockwise in Bohai Bay (Liu et al., 2016a). The winter coastal current system, crucial for long-term sediment transport and prodelta evolution, is dominant because the Yellow Sea Warm Current is less intrusive in the summer; as a result, the circulation system predominantly transport river-borne sediments in winter, while storing them in summer (Su and Yuan, 2005; Xue et al., 2018). A bifurcation of the current occurs off the Luanhe River delta in the northwestern Bohai Sea ( Figure 1A ) (Su and Yuan, 2005; Liu et al., 2016a; 2016b).

The Bohai Sea experienced transgression since the deglaciation period, and as an isolated gulf, seawater has entered Bohai Sea from the Bohai Strait since ~12 kyr BP. During the maximum flooding of the transgression ~7 kyr BP, the shoreline extended 50-90 km west of its present position (Xue, 1993; Xue et al., 2018). From the mid-Holocene onward, the shoreline advanced along the northern, western, and southern coasts of Bohai Bay, associated with the deltaic/alluvial progradation of the ancient Huanghe and other smaller river (e.g., Luanhe and Haihe Rivers) (Tian, 2010; Liu et al., 2019) ( Figure 1A ). These rivers, particularly the Huanghe, contribute the most sediment loads to the western and northwestern Bohai Sea (He et al., 2019).

The substantial sediment discharge originating from the Huanghe has played a crucial role in shaping the Huanghe delta both onshore (Xue, 1993; Saito et al., 2000; He et al., 2019) and offshore (Liu et al., 2009; Liu et al., 2016b, 2019) in the western Bohai Sea. The Huanghe River has experienced frequent avulsion and lobe-switching events (Cheng and Xue, 1997; Wang et al., 2007), leading to the formation of a deltaic plain comprised of distinct deltaic lobe (referred to as superlobes). Tranditionally, the delta plain was divided into 10 superlobes (Xue and Cheng, 1989; Xue, 2009), but a recent analysis of borehole cores along the western Bohai coast (He et al., 2019) has suggested a revised division into eight superlobes, accompanied by refined dating spans ( Supplementary Figure 1A ). Notably, except for the period from 1128 to 1855 AD when the Huanghe altered its course to the South Yellow Sea, forming a separate superlobe, all other superlobes are located along the western Bohai coast ( Supplementary Figure 1A ). North of the Huanghe delta plain, the northernmost regions of the western Bohai coast and the northwestern Bohai coast are dominated by the Haihe and Luanhe River delta plains, respectively ( Figure 1 ; Supplementary Figure 1A ). Similar to the Huanghe delta plain, these areas have undergone evolution since the mid-Holocene, driven by the progradation of the two deltaic systems (Tian, 2010; Liu et al., 2019).

Offshore, there is a pervasive Huanghe-dominated mud region across the western, southern, and northwestern Bohai Sea ( Figure 1B ). The subaqueous delta clinoform of the Huanghe, as the subaqueous extension of the Huanghe delta plain, is situated predominantly in the western Bohai Sea with a thickness of up to 20 m ( Figure 1B ) (Liu et al., 2019). Dominated by eastward/southeastward and northeastward coastal currents, the Huanghe prodelta experiences deflection along these two directions, resulting in the formation of a unique dual-mud belt system ( Figure 1B ) (Liu et al., 2022). The eastward/southeastward deflected prodelta (mud belt) eventually reaches east of the Bohai Strait, entering the Yellow Sea and forming a mud wedge (up to ~40 m thick) offshore of the Shandong Peninsula (Yang and Liu, 2007). The northeastward deflected prodelta extends to the offshore regions of the Luanhe delta and even farther northeast, forming a depocenter (~10 m thick) near the furthest end of this mud belt ( Figure 2C ). This mud belt is separated from the adjacent Luanhe River prodelta (attached to the Luanhe River delta plain) by a shore-parallel erosion-dominated sandy region, which is ~10-20 km in width ( Figure 1B ) (Liu et al., 2022).

The seismic stratigraphy in the western and northwestern Bohai Sea has been extensively examined recently (Tian et al., 2017; Liu et al., 2019; 2020; 2022), providing a critical foundation for this research ( Figure 2 ). Postglacial stratigraphy in these regions comprises three sets of stratigraphic units (Liu et al., 2020): (1) terrestrial sediments formed during Late Pleistocene when the study area was subaerially exposed and devoid of seawater influence, primarily representing the Lowstand Systems Tract; (2) transgressive lags formed during the transgression through winnowing processes, especially wave action and ocean currents, often exhibiting as sand sheets and corresponding to the Transgressive Systems Tract; and (3) a unit of prodelta sediments formed since the mid-Holocene due to the progradation of neighboring river deltas, representing the Highstand Systems Tract (HST). These units are bounded by a transgressive (wave) ravinement surface (wRs) and a Maximum Flooding Surface (MFS).

The focus of this study is on the HST, and its internal configuration reveals notable disparities between the western and northwestern Bohai Sea. In the former, the unit exhibits a lobate configuration, consisting of four subunits associated with the Huanghe-dominated prodelta (Liu et al., 2019; corresponding to seismic units 4-6 and 8 in that study) ( Figure 2B ). Therein, the most recent subunit represents the accumulation of the Modern Huanghe Delta (MHD) formed since 1855 CE. Its basal boundary truncates internal dipping reflections of the underlying subunits on their tops ( Figure 2B ), representing a regional erosional unconformity; this aligns with the shift of the Huanghe course to the South Yellow Sea between 1128 and 1855 CE, resulting in a lack of sediment supply and degradation of the prodelta in our study area (Liu et al., 2013; 2019). In the northwestern Bohai Sea, the HST is internally uniform, characterized by stratified reflections, with no lobate structure observed ( Figures 2C, D ). Due to energetic longshore transport regimes, transgressive sand sheets have transformed into a series of alongshore-arranged erosional ridges, a phenomenon particularly pronounced in the erosion-dominated region ( Figures 2C, D ). Consequently, HST sediments have been partitioned by this erosional region along the shore-normal direction, with (1) a wedge attached to the Luanhe River delta plain corresponding to the Luanhe River-dominated prodelta, (2) isolated mud patches filling the troughs among the ridges in the erosion-dominated area, and (3) the Huanghe-dominated prodelta, extending from coast to offshore ( Figures 2C, D ) (Liu et al., 2022).

The Bohai Sea and its surrounding area are characterized by a warm and wet climate during summers and a cold and dry climate in winters, influenced by the East Asian Monsoon, with southerly and northerly winds prevailing in the summer and winter, respectively (Zang, 1996). The lowest temperatures (-1 to 4°C) typically occur in January, while the highest temperatures (24 to 27°C) are recorded in July and August. Precipitation is particularly concentrated in the summer, with an annual average of ~613.6 mm (IOCAS, 1985; Wang, 2013).

The vegetation surrounding the Bohai Sea ( Supplementary Figure 1B ) is predominantly characterized by deciduous broadleaved forests associated with warm temperate climates and shrub grasslands (Wang, 1993). Natural vegetation is predominantly observed in mountainous areas due to intensive agricultural and human activities on the plains (Meng and Wang, 1987). In the hills, deciduous broadleaved forests are dominated by Quercus, co-dominated with Pinus. Some deciduous broadleaved trees are present in the plain area, while others, e.g., Betula, Tilia, and Carpinus, are mainly found in the hills and lowlands. Coastal salt marshes are characterized by herb dominance, particularly Chenopodiaceae and Artemisia (Wang, 1993). Within the Huanghe delta plain, the vegetation is predominantly composed of herbs and shrubs, with trees dominated by Pinus densiflora, Pinus tabuliformis, and some other deciduous broadleaved forests. In the Luanhe delta plain, dominance is observed in P. tabuliformis, Quercus liaotungensis, and Quercus dentate ( Supplementary Figure 1B ).

In the Bohai Sea, the distribution of pollens and spores in seafloor sediments has been investigated by Yang et al. (2019), suggesting that Huanghe serves as the primary source. They also proposed that (1) pollen concentration is related to grain size, and (2) water depth is proportional to the percentages of Pinus pollen and Pteridophyte spores, while inversely related to other arboreal (excluding Pinus) and herbaceous pollen.

Previous palynological studies on sediment cores in this region primarily focused on paleoclimatic reconstruction (Yi et al., 2003; Chen et al., 2012; Lu et al., 2023), a scope does not align with the objective of this study. Nevertheless, these studies offer a valuable dataset for us to reevaluate the distribution of pollen and spore records in relation to the mid- and late-Holocene prodelta evolution.

As mentioned earlier, the seismic stratigraphy in the western and northwestern Bohai Sea has been well characterized through recent investigations. To place our pollen records in the context of stratigraphic evolution via a seismic-to-core correlation, we employed three seismic profiles from these studies: one from the western Bohai Sea, extending east-west north of MHD (Liu et al., 2019), and two from the northwestern Bohai Sea along shore-parallel and shore-normal directions (Liu et al., 2022) ( Figures 1B ; 2B-D ). The acquisition of these profiles utilized the Applied Acoustic Engineering CSP2200 subbottom system (United Kingdom). Additionally, we presented a new seismic profile covering approximately two-thirds of the offshore region around the MHD ( Figures 1B ; 2A ). The eastern quarter of this profile employed the same subbottom system and firing parameters as the boomer profiles reported by Liu et al. (2022), while the western three-quarters utilized chirp data collected with an Edgetech 512i towfish featuring a 0.5-7.2 kHz, 30 ms pulse.

Seismic interpretation in the western and northwestern Bohai Sea followed Liu et al. (2019) and (2022), respectively. The interpretation of early Holocene sand and the overlying MFS at the location east of MHD was not addressed in Liu et al. (2019), as this sand body did not develop within their study area. These interpretations, instead, is based on the findings of Li et al. (2023) (detailed in section 4.1). Depth estimation used two-way travel time (TWTT) with an average acoustic velocity of 1650 m/s, following Liu et al. (2016a); Liu et al. (2019).

Five new boreholes (BHB1, YRD1, H1, H3, and H4) were drilled in the western and northwestern Bohai Sea along with the seismic profiles ( Figure 1B ). The recovery rates for these cores exceed 90% (considering total length), except for the top 1 m of BHB1 and top 1.2 m of YRD1, which were unrecovered. Core positions were determined using DGPS. Given the primary focus on pollen records within prodeltaic sediments in this study, specific segments of these cores were selected for analysis: the top ~6 m of core BHB1 and the top 5 m of cores YRD1, H1, H3-H4. The selection of these segments was based on their correlation with the seismic stratigraphy. In the laboratory, these segments were longitudinally split, described, photographed, and subsampled for subsequent analysis. Additionally, five sediment cores (H9601, H9602, BH264, L1, H2) were sourced from existing literature, selected based on established chronological constraints and palynological data availability.

Cores BHB1, YRD1, H9601, H9602, and BH264 are situated in the western Bohai Sea, both onshore and offshore of the Huanghe delta plain. The remaining five boreholes (L1 and H1−4) are located in the northwestern Bohai Sea, both onshore and offshore of the Luanhe River Delta ( Figure 1B ). Basic information for these cores is summarized in Table 1 .

Ten samples (including bulk sediments, peat, plant and benthic foraminifera) from the newly acquired cores were selected for ¹⁴C dating at Beta Analytic Inc. (Miami, United States) and Pilot National Laboratory for Marine science and Technology (Qingdao, China) using accelerator mass spectrometry (AMS). Past research in and around the Bohai Sea has suggested that conventional ages of marine samples should be calibrated using the Marine curve when their δ¹³C values exceed -10‰, while the IntCal curve is more appropriate for values typically around -25‰ (e.g., He et al., 2018; Liu et al., 2023). Given this, our bulk sediment, peat, and plant samples, characterized by δ¹³C values nearing −25‰, were calibrated using the latest IntCal 20 curve, while shell and benthic foraminifera samples were calibrated using the most recent Marine 20 model (Reimer et al., 2020) with the ΔR=-334 ± 50. All calibrations were conducted using Calib Rev. 8.2 software with a two standard deviation (2σ) of uncertainty (Stuiver et al., 2022).

Thirty-nine AMS ¹⁴C ages have been documented along with the other five cores (from literature sources), but they were not calibrated with the same standards (i.e., the Marine20 and IntCal20 models were not used). In this regard, a recalibration was conducted using these latest models. The dating (and calibration/recalibration) results newly reported in this study and those from earlier research are summarized in Tables 2 , 3 , respectively.

A total of 178 samples were collected at 5-10 cm intervals (for cores BHB1 and YRD1) and 15-20 cm (for cores H1, H3 and H4) intervals for grain size analysis. The grain size data were determined using a Malvern Mastersizer-2000 laser particle size analyzer (United Kingdom) after pretreating the samples with 10% H₂O₂ and 0.1 N HCl to remove organic matter and biogenic carbonate. The grain size parameters (e.g., mean diameter) were calculated following the procedure of Folk and Ward (1957). The classification of sediment types was determined using the classification scheme established by Shepard (1954).

Fifty-two samples were taken at 5-30 cm (for cores YRD1 and BHB1) or 50-100 cm (for cores H1, H3 and H4) intervals for pollen and spore analysis. Each sample, weighing 5.0 ± 0.3 g (cores YRD1 and BHB1) or 10.0 ± 0.3 g (cores H1, H3 and H4) in dry weight, were processed following the presentative palynological treatment procedures (Moore et al., 1991). This involved the use of 10% HCl to dissolve calcareous minerals, 10% KOH treatment for the removal of organic matter, and 45% HF to eliminate siliceous materials. The residues were sieved over a 7-mm mesh screen in an ultrasonic water bath to remove tiny impurities and facilitate pollen identification. To facilitate identification, exotic Lycopodium marker spores (10,315 ± 281 grains/tablet) were added before the initial sample treatment. The prepared specimens were subsequently mounted in glycerin jelly for examination and identification.

Pollen identification and counting were conducted using a Nikon Eclipse Ni stereomicroscope at 400× or 1000× magnification. Palynomorphs were identified in accordance with the Quaternary Pollen Atlas (Wang et al., 1995; Tang et al., 2020). For each sample, a minimum of 200 pollen grains (excluding spores) were counted. A pollen diagram was created using TILIA 2.0.29 (Grimm, 1991 and the subsequent updated version). The pollen and spore percentages, indicating relative concentrations, were calculated based on the terrigenous pollens and spores.

All ten sediment cores examined in this study have been integrated into the pre-existing seismic stratigraphic framework (section 2.2), utilizing seismic-to-core correlation ( Figures 2A-D ) and incorporating AMS ¹⁴C dating results ( Tables 2 , 3 ). Stratigraphic transects of the core columns, integrated with the seismic stratigraphic framework, from the western and northwestern Bohai Sea are depicted in Figures 3 , 4 , respectively. The variations in mean grain size for the five newly acquired cores are illustrated in Figures 5 – 9 .

In the five newly acquired boreholes, over 80 sporopollen types were identified, predominantly originating from deciduous trees and herbs ( Figures 5 – 9 ). Therein, the arboreal pollens (AP) were dominated by gymnosperms, particularly Pinus and Picea, while angiosperm pollens included mainly deciduous Quercus, Betula, Ulmus, Ephedra, Morus, and Carpinus. Non-arboreal pollen (NAP) grains were mainly composed of Artemisia, Gramineae, Chenopodiaceae, Typha, and Cyperaceae. Fern spores included Selaginella, Triletes, and unidentified Monoletes. Pollen concentrations ranged from 1 to 8723 grains/g, with varying percentages and concentrations of AP and NAP. Pinus was notably abundant in several sections ( Figures 5 - 9 ).

In the proximal core transect, core BHB1 shows the highest percentages of SP and AP for SL2, while core YRD1 exhibits the highest percentages for SL3 ( Figure 10 ). Considering the alignment of SLs 2 and 3 with superlobes I and II on the Huanghe delta plain, along with the locations of these superlobes and contemporaneous paleo-shorelines, core YRD1 is the farthest core site from the contemporaneous Huanghe mouth along the direction of the offshore sediment dispersal regime during the evolution of SL3/superlobe II, and core BHB1, although not the farthest, is situated far away from the Huanghe mouth and within the upcurrent location of the sediment dispersal regime during the formation of SL2/superlobe I ( Figure 1A ; Supplementary Figure 1A ). From a source-to-sink perspective, the upcurrent location can also be considered a considerably far distance for sediment transport, as it is not the typical destination for sediment deposition. In this context, our observation suggests that SP and AP can disperse to more distant locations in the proximal region, aligning with previous studies indicating that pollen with air sacs can travel greater distances compared to NSP (Wodehouse, 1935; Schwendemann et al., 2007; Leslie, 2010). The consistent dispersal pattern between SP and AP is likely due to a significant portion of AP consisting of airborne pollen species (Nilsson et al., 1977; Rahman et al., 2020). These findings also extend to individual pollen species; Pinus, a representative species of SP due to its lightweight nature allowing it to float in both water and air for an extended period with two air sacs (Schwendemann et al., 2007), shows a consistent increase in percentage within both SLs 2 and 3 as the transport distance increases ( Figure 10 ).

Contrastingly, core BHB1 exhibits the lowest percentages for NSP and NAP in SL2, while core YRD1 shows the lowest percentages for SL3 ( Figure 10 ). This suggests that the dispersal of NSP and NAP differs from that of SP and AP and is not dominant in long-distance transport. Despite NAP grains being generally smaller than AP and presumably dominated by water transport over longer distances, both NAP and Artemisia (a representative NAP species) show decreasing percentages with dispersal distance in SL2 and SL3 ( Figure 10 ). This implies that NAP tends to accumulate earlier in the dispersal process, reflecting short-distance transport, which is consistent with previous studies indicating NAP has concentrated in the nearshore areas of the Bohai Sea (Xu and Sun, 1988; Yang et al., 2019). In summary, in the proximal accumulations of the Huanghe prodelta, the percentage of SP and AP correlates positively with offshore dispersal, while that of NSP and NAP shows a negative correlation. Notably, this process is also influenced by the direction and strength of dispersal and the hydrodynamic regime, discussed further in Section 5.4.

The HST sediments in core H3 represent the Huanghe-dominated distal accumulation ( Figure 1 ). Although its formation age is unclear, it provides an opportunity to compare pollen records between proximal and distal accumulations. The five proximal cores cover most of the evolutionary stages of the Huanghe delta, potentially including the timing for distal accumulation. Therefore, differences in pollen assemblages between core H3 and these cores can be attributed to disparities between proximal and distal accumulations. In core H3, the percentage of AP is slightly lower than NAP, and Pinus percentage is insignificant ( Figure 11 ). This indicates that, while AP, particularly Pinus pollen, can travel longer distances than NAP, both AP and NAP show low concentrations after extended long-distance transport, and the difference in their percentage observed in proximal sediments is no longer significant. Previous studies on seafloor sediments in the Bohai Sea suggest that wind-driven dispersal (in air) becomes the primary driver of pollen dispersal when water depth exceeds 20 m (Yang et al., 2019). However, this is not observed in core H3 (core top over 20 m water depth; Table 1 ), with lower SP and an insignificant percentage of Pinus ( Figure 8 ). This could be due to complex pollen provenance and preservation processes in this distal mud, as further discussed in Section 5.3.

Pollen records in deltaic-prodeltaic sediments were often used for paleoclimatic reconstruction (Hao et al., 2021; Ye et al., 2024). In the Huanghe delta, Yi et al. (2003) used variations in Pinus and NAP percentages in cores H9601 and H9602 to indicate climatic warmth and cooling. They associated a decrease in Pinus in Pollen Zone I/1 and Zone IIb/2 with a warm climate, while an increase in NAP and Pinus in Pollen Zone II/1 and Zone III/2 suggested cooling.

Our stratigraphic analysis ( Figure 4 ) reveals that in core H9601, Zones I/1 and II/1 roughly correspond to SL2 and SL3, respectively, while in core H9602, Zone III/2 spans SL2, SL3, and superlobe V. In contrast to cores H9601 and H9602, where Pinus decreases in SL2 and increases in SL3, our core BHB1 shows the highest Pinus percentage in SL2 but a sharp decline in SL3, accompanied by an increase in NAP. Core YRD1 also does not exhibit an increase in NAP and Pinus in SL3; instead, both decrease slightly ( Figure 10 ). This comparison highlights a noticeable variation in the vertical (downcore) distribution of pollen assemblages in prodeltaic sediments across different locations within the Huanghe delta. Given that significant changes occurred among different delta lobes, this variation is likely attributed to lobe switching, which involves alterations in offshore distance, sediment/pollen transport distance, and different time spans of accumulation (even for a single lobe, it may accumulate diachronously) across various sites.

In the northwestern Bohai Sea, core site L1 underwent a transition from a prodelta to a delta-plain environment due to recent lobe switching of the Luanhe River, which resulted in resulting in the shoreline protrusion and creation of deltaic land at this location (Yu, 2019; Liu et al., 2022) ( Figures 1 , 3 ). In prodeltaic sediments, the pollen assemblages are dominated by AP, particularly Pinus and Quercus, with AP significantly higher than NAP in percentages and concentrations ( Figure 11 , Supplementary Figure 4 ). In delta-plain sediments, NAP becomes the dominant taxa, and the percentage of Selaginella is significantly higher than in prodeltaic sediments ( Figure 11 ). Although the increased NAP and Selaginella percentages in delta-plain sediments most likely associated with the transition in sedimentary facies related to lobe-switching, these changes might be alternatively interpreted as indicators of increased precipitation and more frequent river floods in pollen-based paleoclimatic studies. In earlier studies, these elements are typically used for as climatic indictors because Selaginella spores, being large in grains, settle and sediment primarily through river flooding (Xu et al., 1995), and NAP content is generally higher than AP during flooding periods (Xu et al., 2004). In addition, the decline in AP from prodeltaic to delta-plain sediments is also probably influenced more by lobe-switching-related changes in the sedimentary environment than by climate. Previous studies (e.g., Xu et al., 1995) have indicated a gradual decline in AP content with greater transport distance on the Luanhe River alluvial plain-delta plain. This pattern likely plays a significant role in the decline of AP during the prodelta to delta plain transition at core site L1, associated with increased onshore transport distance for this core site.

The northwestern Bohai Sea, receiving sediment from both the Huanghe and Luanhe River (Liu et al., 2022), is expected to show varied pollen assemblages in their prodeltas due to diverse vegetation in their drainage basins ( Supplementary Figure S1B ) (Yang et al., 2019). Along the core transect ( Figure 3 ), HST sediments in core L1, representing Luanhe River input, displays a Pinus-Artemisia-Quercus-dominated pollen assemblage ( Supplementary Figure S4 ), with the dominance of AP, which is consistent with offshore seafloor sediments (Xu and Sun, 1988). However, subsurface sediments within this offshore region, identified as the Huanghe-dominated mud belt distinct from the Luanhe River prodelta (Liu et al., 2022), exhibit variations from surface sediments and among themselves, showing Artemisia-Ulmus-Chenopodiaceae-Pinus ( Figures 8 , 9 ) with NAP dominance in cores H3 and H4 ( Figure 11 ). Such a discrepancy in pollen assemblages between surface sediments and subsurface records, aligning otherwise with those of the separated Luanhe River delta, likely results from the abandonment of the Huanghe-dominated mud belt. We hypothesize that the mud belt has not received Huanghe supply for at least over 1 kyr; historical evidence indicates that the Huanghe flowed into the southern Yellow Sea between 1128 and 1855 CE (Liu et al., 2013), and there is no evidence of sediment supply from the Huanghe to the mud belt since 1855 CE (instead prone to disperse southeastward from MHD) when it re-entered the Bohai Sea (Liu et al., 2023). Thus, while subsurface pollen records may reflect the influence of the Huanghe, surface sediments appear to receive pollen sedimentation sourced from the surrounding region. This discovery underscores the complex nature of pollen provenance in seafloor sediments, especially during periods of sediment hiatus.

In cores H3 and H4, Ulmus dominates the AP, constituting ~41.2% of the total pollen assemblage ( Figure 11 ). Notably, no such high content of Ulmus has been observed in sediment cores from the Luanhe River and Huanghe drainage basins, delta plains, or the Bohai Sea seafloor (Yi et al., 2003; Chen et al., 2012; Yang et al., 2019), where the representatives of AP are Pinus, Quercus, Tilia, and Corylus (Xu et al., 2007). Instead, Ulmus and other broad-leaved trees were dominant in the forests of the eastern Northeast China region during the early to mid-Holocene (Ren, 1999). Particularly, in the pollen assemblage of core DYD in the eastern Liaodong Peninsula, Ulmus pollen accounted for as much as 37.65% in Holocene sediments (Liu, 2022). As Ulmus is an airborne pollen primarily flowering from April to June, coinciding with prevailing southerly winds in the Bohai Sea area (Yang et al., 2019), we hypothesize that the Ulmus pollen in cores H3 and H4 originated from airborne and/or seawater dispersion from the Northeast China region rather than being transported with sediment from the Huanghe or Luanhe River. This hypothesis can be supported by the water depths of cores H3 and H4, both ~20 m ( Table 1 ), and the Bohai Sea circulation system ( Figure 1A ); both facts favor westward transport from Northeast China, facilitating dispersion with water flow (cf., Yang et al., 2019). The abundant Ulmus pollen in cores H3 and H4 provides insights into how prodeltaic sediments preserve pollen from other regions through airborne and seawater dispersion during prodeltaic sedimentation, adding greater uncertainty and complexity to the pollen assemblages of prodeltaic sediments.

In the northwestern Bohai Sea, HST sediments in core H2 exhibit higher pollen concentrations (excluding Ulmus) than adjacent offshore cores H1, H3 and H4 ( Figure 11 and Supplementary Figure 5 ), even surpassing delta-plain sediments in the onshore core L1 ( Supplementary Figure 4 ), indicating that H2 is a pollen sink. Our stratigraphic analysis places core H2 in the erosion-dominated area with energetic longshore tidal currents (Xue et al., 2009), strong enough to remold the transgressive sand sheets into erosional sand ridges, but the core site is on an isolated mud patch filling a trough between the ridges, representing an area of accumulation/erosional relict (Liu et al., 2022) ( Figures 2 , 3 ). Despite the prevailing belief that quiescent conditions favor pollen deposition (Chmura and Eisma, 1995), the concentration of pollen in core H2 suggests that accumulations in energetic hydrodynamic conditions might provide an ideal setting for pollen deposition, potentially more so than in prodeltaic sediments (compared to cores H1, H3, and H4).

Fern spores, predominantly Selaginella, also exhibit the highest percentage and concentration in core H2 along the transect of the five cores ( Figure 11 ), which can further support the afore mentioned hypothesis. Specifically, Selaginella grains, as previously noted, being large and heavy, rely on strong hydrodynamic regimes for transport (Xu et al., 1995). Additionally, the high percentage of fern spores aligns with their buoyant nature, which allows them to be transported over longer distances (Dai and Weng, 2011; Dai et al., 2014; Yang et al., 2019). This nature may contribute to their high percentage in regions with strong transport regimes, particularly in seafloor sediments with increasing offshore distance and abundant precipitation, where fern spores far exceed pollen in percentage (Wang et al., 1987; Sun et al., 1999; van der Kaars and de Deckker, 2003).

Energetic longshore currents in the erosion-dominated area likely also influence pollen deposition in neighboring prodeltas, especially at their marginal parts. Cores H1 and H4, located at the boundary between the prodelta and the erosion-dominated area ( Figures 1B ; 3 ), represent such marginal sediments and exhibit lower pollen content compared to non-marginal core H3 ( Figure 11 ). Additionally, we hypothesize that these currents control the sedimentation of Ulmus pollen along different zones. As Ulmus is likely transported from the east, possibly via seawater dispersal, strong longshore currents may act as a barrier, hindering Ulmus pollen from reaching more landward (northwestward) locations. This hypothesis explains the observed high Ulmus content in core sites (H3 and H4) southeast of the erosion-dominated area, while the content is much lower in the remaining cores ( Figure 11 ).

Regionally, transport regimes can significantly impact pollen records, creating pollen sinks, due to their flow directions. This phenomenon is primarily observable in the western Bohai Sea, providing three lines of evidence. Firstly, an anomalous pattern in SL2 shows higher percentages of fern spores and Quercus near the river mouth in core H9601 but higher concentrations in core BHB1 farther away ( Figure 10 ; Supplementary Figure 6 ). Despite typical nearshore concentration for Quercus (showing high percentages) in the western Bohai Sea (Yang et al., 2019) and an expected increase in fern percentage after long-distance transport, opposite trends for both elements were observed, potentially influenced by the local coastal current direction promoting nearshore pollen accumulation. Secondly, ferns exhibit nearshore concentration with percentages increasing offshore, a normal trend. However, the highest Quercus percentage and concentration appear in core H9602, not in core BHB1 closer to the contemporaneous river mouth ( Figure 10 ; Supplementary Figure 1A ), possibly due to core BHB1 being more offshore with deeper water depth, and the southeastward longshore transport regime at that time influencing a shallower region than core BHB1. The third evidence comes from total pollen concentration in core BHB1, which is the highest in both SL2 and SL3 ( Figure 5 ). Given the core site BHB1 is neither the farthest nor closest to the contemporaneous river mouth for both SLs, the highest pollen concentration is most likely associated with a pollen sink likely induced by the local transport regime. Similar pollen sinks associated with the transport regime direction are observable in Bohai Sea surface sediments, particularly in Laizhou Bay (displaying the highest pollen concentration; Yang et al., 2019), a down-current region for the coastal current, indicating a sink ( Figure 1A ).

Considering that clinoform-scale construction of prodelta typically leads to specific arrangements of sediment grain size distribution (Patruno and Helland-Hansen, 2018), it is crucial to examine how this distribution affects the pollen records. We observed a positive correlation between grain size/mud content and pollen concentrations in various core segments ( Figure 12 ). Specifically, in core BHB1, SL3 and SL2 exhibit the highest pollen concentrations (>4000 n/g; Figure 5 ) with the smallest grain size (clay or silt clay; Figure 4 ), while SL2 in core H9602 shows the lowest concentration (<50;Yi et al., 2003) with a significant increase in grain size (fine sand; Figure 4 ). In core L1, lower pollen concentrations are associated with coarser delta-plain sediments compared to the higher concentrations in prodeltaic sediments ( Figures 3 , 11 ).

This positive correlation, unsurprisingly, is also evident in seafloor sediments across various continental shelf regions, e.g., the northern South China Sea (Tong et al., 2012) and the Bohai Sea (Yang et al., 2019), likely attributed to the susceptibility of both pollen and fine-grained sediments to suspension-related dispersion in underwater conditions (Heusser, 1988). Nevertheless, in Huanghe prodelta sediments, this positive correlation is valid only when the mud content exceeds ~8-12%, and lower mud content may impede pollen accumulation ( Figure 12 ).