The Longwangmiao (LWM) site, one 9-m representative sedimentary profile, is located on the site of a disused brick factory, 10 km southeast of Daming County (Figure 1B). The LWM profile has been sensitive to environmental change since the Holocene, as fluvial-lacustrine sediments were interbedded with sandy soils. Based on our prior studies in the profile (Yang et al., 2023a; b), the Holocene paleoflood periods have been reconstructed using analysis of sedimentary characteristics, grain size and chemical composition. Our stratigraphy and paleoflood results are consistent with historical documents and regional archaeological work.

Due to the scarcity of plant fossils, only one humus soil sample at the top of lacustrine deposits was submitted to the Xi’an Accelerator Mass Spectrometry Centre, China, for radiocarbon dating. The result is reported as radiocarbon ages as years before present with an analytical precision <0.5%. Calibration of age was carried out using the IntCal20 calibration curve (Reimer et al., 2020) and OxCal 4.4 (https://c14.arch.ox.ac.uk), which allows direct comparison with other ages.

Seven optically stimulated luminescence (OSL) samples, including six samples of flooding deposits and one sandy soil sample, were dated at the Key Laboratory of Quaternary Chronology and Hydro-Environmental Evolution, China Geological Survey. All of the samples were pretreated under the red light. The central parts of samples were treated with HCl (10%) and H₂O₂ (30%) to remove organic matter and carbonates. Coarse grains (63∼90 μm and 90∼125 μm) or fine grains (4∼11 μm) were extracted based on particle size distribution of different samples (Table1). Then, the coarse grains were etched by HF (40%) for about 1 h, while the fine grains were treated with 30% fluorosilicic acid for about 5 days. The equivalent doses of the samples were measured by Lexsyg smart OSL/TL dating device. Single aliquot regenerative dose (SAR) protocol (Murray and Wintle, 2000; Wintle and Murray, 2006) was used. The U, Th and K concentrations were measured by Inductively Coupled Plasma Mass Spectrometry (ICP-MS) method. The water contents were estimated based on the values measured at laboratory and the sedimentary environments of the samples. The total dose rate was calculated by using the DRAC (Durcan et al., 2015). Based on all the radiocarbon dating and OSL results, the Bayesian age-depth model was established in R v.4.3.1, using package bacon v.2.3.9.1 (Blaauw et al., 2018).

Bulk samples for pollen and grain size analysis were obtained every 5 cm. 180 samples were conducted in the Key Laboratory of Quaternary Chronology and Hydro-Environmental Evolution, China Geological Survey. Pollen analysis followed using the conventional method (Faegri and Iversen, 1964). A tracer of modern Lycopodium spores with 27,637 ± 563 grains was added to each sample before chemical treatment. Pollen was counted under an Olympus BX51 microscope at ×400 magnification. Pollen taxonomy mainly followed Pollen Flora of China (Wang et al., 1995). The samples with a minimum of 100 grains were included in statistical analysis in accordance with the guide to the standardization of global palaeoecological data (Flantua et al., 2023) and related pollen research in nearby sites (Liu et al., 2013; Ren et al., 2019). In total, 97 out of 180 samples were counted for further analysis. Over 400 pollen grains and spores per sample were counted, with the maximum count reaching 798 grains (in one sample). Taxa percentage was calculated based on the total pollen and spore sum. CONISS was used for cluster analysis of pollen taxa. Pollen concentration and percentage diagrams were constructed using the rioja R package (Juggins, 2015).

To gain additional insights into the depositional environment and potential post-depositional processes, deteriorated pollen grains were analyzed separately (Hall, 1981; Chmura and Liu, 1990). Deteriorated pollen was differentiated in morphology as the grains are dark in colour with a damaged or corroded appearance. Several types of deteriorated pollen grains have been summarized, including corroded, degraded, mechanically damaged and obscured pollen grains (Lowe, 1982; Fernandes Pinto, 2012). These degraded grains are common in fluvial deposited pollen assemblages and are probably sourced during flood stages. Therefore, they are assumed to be an indicator of fluvial transport (Chmura and Liu, 1990; Wagstaff et al., 2013).

Grain size distributions of samples were determined using a Malvern Mastersizer 2000 with (NaPO₃)₆ as a dispersing agent after pre-treatment with HCl (10%) and H₂O₂ (30%) to remove secondary carbonates and organic matter. The measurement range of Mastersizer 2000 was 0.02–2,000 μm in grain diameter with a relative error of <1%.

According to our previous research (Yang et al., 2023a; b), flood deposits in the profile have different sedimentary characteristics from lake deposits and sandy soils. In general, major flood deposits commonly contain an inverse-grading couple of fine-grained slackwater deposits (SWDs) and coarse-grained overbank deposits (ODs). The SWDs are red-brown clay with massive structures, and the ODs are a mixture of grey-yellow silt and sand with parallel laminations and ripples. The stratigraphy sequence from bottom to top was summarized as follows (Figure 2):

1) Unit I (900–825 cm) consists of coarse-grained sands with mud gravel at the bottom.

Seven OSL dating results, ranging from 7.2 ± 0.4 ka to 0.86 ± 0.07 ka, follow the stratigraphic sequence (Table 1). At the bottom of Unit I, the OSL result is 7.2 ± 0.4 ka, and the radiocarbon dating result of lacustrine deposits in Unit II is 4,218–3,901 cal a BP. Five OSL results of flood deposits in Unit III are distributed around 2.7–2.5 ka, and one OSL result of the paleosol in Unit IV is 0.86± 0.07 ka. Considering the episodic deposition and hiatus in alluvial sediments, the Bayesian accumulation age-depth model was updated based on the lithological boundaries (Blaauw et al., 2018; Lombardi and Davis, 2022) (Figure 2). As a result, five sedimentary units were identified: before 7.3 ka (Unit I, 900–825 cm), 6.8–3.9 ka (Unit II, 825–755 cm); 3.2–2.2 ka (Unit III, 755–358 cm), 2.1–0.8 ka (Unit IV, 358–190 cm), and after 0.6 ka (Unit V, 190–0 cm). Both the sedimentary records and chronological results are mainly consistent with previous research sites (Yang et al., 2023b).

A total of 66 pollen types were identified in 97 samples, including 20 arboreal pollen, 32 herb pollen, and 11 fern spores, with an average count of 414 grains per sample and an average concentration of 95 grains/g. Only taxa with a representation of >1% were used in Figure 3: Pinus, Quercus, Cereal, Amaranthaceae, Artemisia, Polygonum, Malvaceae, Amethystea, Typha, Selaginella, Selaginella sinensis, and Hymenophyllaceae. The LWM profile was divided into five pollen and three pollen-poor zones according to changes in the relative abundances and the CONISS cluster analysis result (Figure 3).

Pollen zone 1 (900–855 cm, before 7.3 ka): The average pollen concentration is 31 grains/g in six samples. The percentage of trees in this zone is the highest average in the profile (mean 70.4%), ranging from 58.8% to 82.4%. Pinus, the most dominant pollen type found throughout the LWM profile, also reaches its maximum (mean 65.7%). The herb pollen representation ranges from 8.2% to 19.0% (mean 14.8%), dominated by Amaranthaceae (mean 10.2%). The average percentage of ferns is 14.9% and is dominated by S. sinensis (mean 11.1%).

Pollen-poor zone 1 (855–765 cm, 7.4–4.3 ka): No analyzable samples were obtained.

Pollen zone 2 (765–720 cm, 4.3–3.1 ka): This zone consists of 9 samples. The mean pollen concentration increases to 77 grains/g, ranging from 25 to 146 grains/g. Arboreal pollen decreases to 55.8% in this zone. Pinus has a representation of 42.6%, while Quercus increases to 6.4% in the layer of slackwater deposits with a maximum of 17.0%. Total herbaceous pollen percentages increase to 27.2%, mainly Cereal (0.4%–22.4%, mean value 4.1%), Amaranthaceae (2.6%–23.4%, mean value 9.8%), and Typha (0.5%–35.8%, mean value 7.6%). Ferns have similar characteristics to zone 1.

Pollen-poor zone 2 (720–610 cm, 3.1–2.9 ka): Only two samples have more than 100 grains per sample. These samples are characterized by a high percentage of ferns, dominated by Selaginella sinensis. Selaginella sinensis reaches its maximum value (78.0%) in this profile.

Pollen zone 3 (610–435 cm, 2.9–2.4 ka): This zone consists of 22 samples. The percentage of ferns (4.6%–77.9%, mean value 22.2%) decreases, especially Selaginella sinensis (3.6%–71.8%, mean value 23.0%) decrease rapidly. The herbaceous pollen increases from 5.3% to 62.0%, in which Amaranthaceae (mean value 10.9%), Artemisia (mean value 3.0%) and Typha (mean value 3.0%) increase significantly.

Pollen-poor zone 3 (435–365 cm, 2.4–2.2 ka): No samples meet the statistical data requirements.

Pollen zone 4 (365–110 cm, 2.2–0.5 ka): In this zone, 48 samples have high total grains (109–734 grains, mean value 438 grains). The pollen assemblages resemble those in zone 3, except that herbaceous pollens of Polygonum (mean value 3.8%) and Malvaceae (mean value 7.2%) start to appear. Arboreal pollen taxa dominated by Pinus (32.1%–77.8%, mean value 49.8%) increased to the mean value of 52.3%, while ferns dominated by Selaginella sinensis (1.8%–52.5%, mean value 15.1%) decreased to 20.9%.

Pollen zone 5 (110–0 cm, after 0.5 ka): This zone consists of 10 analyzable samples. The pollen grains decrease slightly to an average of 378 grains per sample (114–618 grains). This zone is distinguished by a substantial increase of herbaceous pollen, averaging 30.7%, ranging from 13.1% to 57.2%. Cereal (mean value 9.9%), Amaranthaceae (mean value 11.9%), and Artemisia (mean value 5.1%) all increase to the peak in the profile, while Malvaceae, Amethystea and Typha almost disappear from the record.

It is widely accepted that pollen and spore distribution are closely related to the sedimentary environment (Brown, 1985; Xu et al., 1996; Zhu et al., 2002; Xu et al., 2016). In accordance with previous studies (Yang et al., 2023a; b), Unit II (825–755 cm) of lacustrine deposits and Unit IV (358–190 cm) dominated by sandy soils, represent relatively stable environments during flood-poor periods, while the other alluvial sediments get deposited in flood-prone periods. This interpretation is also supported by the regional stratigraphic comparison (Kidder et al., 2012; Storozum et al., 2018a; Yu et al., 2020). Pollen zones are consistent with the paleoenvironmental reconstruction results. For instance, Typha, an aquatic taxon, is mainly derived from lakes and wetlands in the NCP (Li et al., 2019). Typha appears in pollen zone 2, corresponding to the lacustrine deposit in Unit II. The percentage of herbs in pollen zone 4 increases to its highest value, corresponding to pedogenesis processes in Unit IV.

The pollen concentration is inversely correlated with grain size, suggesting hydrodynamics likely affect the distribution of types of sediment and pollen (Figure 4A). As Figure 4A shows, pollen concentration has the lowest value in coarse-grained deposits (mean size: 34.5 μm), but relatively high values in fine-grained slackwater deposits (mean size: 18.1 μm), and sandy soils (mean size: 19.1 μm). Fine-grained slackwater deposits were often accumulated in a relatively low-energy environments where the flood velocity was close to zero, and where pollen and spores were more likely to be enriched (Brush and Brush, 1972; Brown, 1985). For sandy soils, Pang et al. (2011) inferred that pollen assemblages from surface soil in the NCP have not only locally sourced plant types but also pollen components from regional mountains. Thus, local herbaceous taxa, including Polygonum, Malvaceae, and Amethystea, were mixed, and the total pollen content increased in paleosols (Liu et al., 2013; Liu et al., 2020). Unexpectedly, the low pollen concentration of lacustrine samples in the LWM profile (average 39.1 grains/g) was distinct from the results of regular lake deposits in the NCP, such as Baiyangdian and Ningjinpo (>100 grains/g) (Li et al., 2019). The result should be interpreted with caution, though, as our sample size was relatively small, based on only two valid samples.

Differences in pollen and spores existed between flood deposits and inter-flood deposits (Figure 4B). Previous research in the NCP argued that the percentage of arboreal pollen in flood deposits was higher than that in neighbouring paleosols (Liu et al., 2013;Liu et al., 2020). Nevertheless, the difference is not remarkable, and the relationship is even opposite in some cases (Liu et al., 2016). Moreover, the difference in pollen assemblages in multiphase flood deposits has not been discussed, as different types of paleoflood sediments are the most common scenario in the NCP (Kidder et al., 2012; Storozum et al., 2018a; b; Yu et al., 2020). In the LWM profile, the arboreal pollen percentage, dominated by Pinus, remains relatively high in different sediments. Pinus is regarded as an over-representative indicator because its pollen can be transported long-distances by airflows and rivers (Yang et al., 2021). In contrast, flood deposits have more ferns and fewer herbs than inter-flood deposits. Ferns dominated by Selaginella sinensis in overbank deposits showed higher average percentages (28.6%) than ferns in other deposits, despite its herbaceous pollen percentage being the lowest value (14.8%) overall. Selaginella sinensis mainly grows in alkaline soils in mountainous regions and was deposited with floods on plains, especially after regional forests were destroyed (Zhang and Kong, 1999; Tang et al., 2013). On the other hand, herbaceous pollen grains are more likely to accumulate in inter-flood deposits, because these pollen grains do not disperse over distance from their nearby parent plants (Yang et al., 2021).

The complex relationship between pollen results and flood events was analyzed by comparison of pollen zones with paleoflood results. Six flooding periods have been reconstructed by sedimentary characteristics and confirmed by the grain size and chemical composition of sediments, including F1 of Unit I; F2, F3 and F4 of Unit III; F5 of Unit V; and F6 of Unit 5 (Yang et al., 2023b) (Figure 5). Both flooding Unit I (900–825 cm) and Unit III (755–360 cm) consist of pollen-poor zones at the top of flood periods. F6 of Unit V shows a similar trend with significantly low pollen concentration in pollen zone 5. However, our results suggest that pollen zones/subzones in the NCP cannot indicate flood events directly, which is inconsistent with previous research (Liu et al., 2020). The behavior of pollen grains and spores transported by river water is similar to that of sedimentary particles and they experience sorting effects (Brown, 1985; Moss et al., 2005). The protracted legacy of human activity, exacerbated by susceptible sedimentary environments in the NCP, can significantly complicate the interpretation of pollen and spores. Moreover, different sedimentary facies are closely interconnected in composition and formation, because inter-flooding sediments on floodplains are generally formed on pre-deposited alluvial deposits and are probably inundated by following overbank floods (Yang et al., 2023a; b). Thus, using pollen and spores as indicators for dividing sedimentary rhythms should be circumspect in an inconstant hydrodynamics environment.

It is noteworthy that deteriorated pollen counts in the LWM profile are concentrated in coarse deposits (Figure 5). The mean 39 grains per sample of deteriorated pollen were detected in 46 samples of the LWM profile, with the maximum count reaching 311 grains at 85 cm depth (F6 in Figure 6). Furthermore, deteriorated pollen grains are consistent with the other flooding sedimentary units, including floods in Unit III (F2, F3 and F4 in Figure 6) and Unit IV (F5 in Figure 6). Deterioration of pollen, influenced by several factors, including mechanical damage, microbial activity, and geological processes (Lowe, 1982; Chmura and Eisma, 1995; Flantua et al., 2023). In alluvial environments, deteriorated and indeterminable pollen fossils are often related to overflows and floods, indicating an unstable depositional environment (e.g., Chmura and Liu, 1990). Hall (1981) indicated that higher frequencies of deteriorated grains corresponded with lower pollen concentration in a single stratigraphic column. Our findings produce similar conclusions to these studies. This strong consistency between deteriorated pollen and flood periods probably confirms its indicative of strong hydrological conditions at the flood-peak period.

Representative non-arboreal pollen taxa in the NCP suggests increased anthropogenic activity since ca. 4.0 ka. No obvious palynological evidence of anthropogenic impacts is detected in pollen zone 1. However, herbaceous pollen percentages increase rapidly and cereal pollen starts to appear at the onset of pollen zone 2. These characteristics are closely related to human activities (Li et al., 2008; Cao et al., 2010), which probably reveal that anthropogenic forest disturbance and cereal cultivation have occurred since ca. 4.3 ka. Furthermore, Selaginella sinensis spores aggregated at 640–610 cm and increased to the highest percentage of 77.98% (ca. 2.9 ka) (Figures 3, 6A). Selaginella sinensis is an indicator of deforestation by intense human activities in uplands, which is an adopted practice in the NCP (Zhang et al., 2007; Cao et al., 2010; Xu et al., 2010). Based on the above pollen analysis, it can thus be concluded that human activity has been amplified around 4.0–3.0 ka.

Local agriculture in the Daming area is intensified in pollen zone 4, which is inferred from the significant increase of Malvaceae. A strong correlation between cotton and Malvaceae has been observed in the field of botanical research (Wang and Li, 1994). Malvaceae increased to its peak value between 285–215 cm (Figures 3, 6B), suggesting that widespread cotton cultivation probably appeared since ca. 1.6–1.0 ka. In addition, agricultural history in the Hebei province also supports the initial period of cotton planting from the Tang Dynasty (ca. 1.4–1.1 ka) (Cui et al., 2019). Therefore, it is safe to infer that this stage represents intensified agricultural activity.

Human impacts inferred from the pollen record are consistent with regional research, which further confirms that large-scale human activities in the study area have gradually accelerated since the late Holocene. The average sedimentation rate within the NCP is characterized by rapid increases around 5.0 ka and 2.3 ka (Figure 6C), which are theorized to be due to growing human influence (Xu, 1998; Shi et al., 2002). This sediment pattern is supported by the concentrated percentages of Selaginella sinensis (Figure 6A) and Malvaceae (Figure 6B), which are indicative of deforestation and agricultural expansion, respectively. Furthermore, pollen assemblages from archaeological sites in the piedmont region of the Taihang Mountains suggested that the growing human footprint and impact of urban settlement was accelerated, especially large-scale deforestation occurring after ca. 3.4 ka (Cao et al., 2010; Ren et al., 2019). Nearby geoarchaeological research in the study area also uncovered that humans began to influence alluvial deposits around 4.0 ka, and agricultural production intensified during the Tang-Song Dynasty (ca. 1.4–0.7 ka) (Storozum et al., 2018a). In conclusion, the palynological findings align with regional research, providing compelling evidence for the progressive intensification of human activities within the NCP.

In this paper, there may be two major limitations that could be addressed in future research. Firstly, the valid samples are relatively limited, and therefore, the results of the study may not be representative of the entire region or period. Future studies should focus on increasing pollen data in fluvial-lacustrine sediments to improve the statistical power of the results (Flantua et al., 2023). Secondly, the possibility of redeposited pollen grains should be considered. The deteriorated pollen in the profile, dominated by Pinus and Hymenophyllaceae, cannot be confirmed to be reworked pollen at present because the separation of reworked Holocene pollen from recent pollen is next to impossible without other evidence (Stanley, 1965).