Late Pleistocene-Holocene vegetation history and anthropogenic activities deduced from pollen spectra and archaeological data at Guxu Lake, eastern China

This study presents high-resolution pollen and charcoal records from Guxu Lake in the Taihu Lake Basin, eastern China, spanning the last 23,000 years. The sedimentary sequences revealed dynamic terrestrial and lacustrine environments during 23.0-11.7 cal ka BP, the climate was relatively cold and dry, and the vegetation was dominated by evergreen-deciduous broadleaf and coniferous mixed forest. During 11.7-4.4 cal ka BP, the Quercus- and Castanopsis-dominated evergreen-deciduous broadleaf mixed forest expanded, while the Poaceae and Artemisia were still the major terrestrial herbs under warmer and more humid conditions. After this period, the climate became relatively cool and dry again, and the vegetation landscape was comparatively stable, as it remains today. Wild rice likely grew before Neolithic humans occupied this area. The variations in Oryza-type Poaceae pollen spectra and distributions of Neolithic archaeological sites indicate rice agriculture may have first appeared and developed with human occupation in ca. 7.0-4.4 ka BP. During the historical period, beginning approximately 4 ka BP, a clear signal of intensified anthropogenic disturbance is evident from the clearing of forests, high charcoal concentrations and the presence of rice pollen in large quantities. These results suggest more intensified rice farming was widespread, with increasing human impact on the environment.

Reconstructing the history of vegetation helps shed light on environmental shifts and human adaptations in response to climatic changes, especially during the Last Glacial Maximum and the Holocene. Such reconstruction can also elucidate the response of humans to global changes and suggest ways to coordinate the relationship between man and nature in the future 1 . To date, millennial and orbital-scale events 2,3 , autoecological processes and/or high regional topographical variability 4 have been considered as the main factors controlling the vegetation changes through hemispherical or semi-hemispherical records from ice cores and speleothems 5,6 , regional and/or local palynological data 7,8 and microcharcoal 9,10 and diatom analyses 11,12 . Organic materials (fossil pollen in particular) are often sufficiently well preserved in waterlogged deposits 13 . Lake sediments in particular are considered the primary archives 14 of proxy data of past environmental change within and around a lake 15 . In addition, these materials are often the richest source of information on the subsistence strategies of prehistoric inhabitants 16,17 .
The Taihu Lake Basin, located in the lower reaches of the Yangtze River, East China, serves as an example. Many studies on paleoclimatic and paleoenvironmental changes of the Taihu Lake Basin have been carried out through the analysis of sedimentary cores from the eastern, southern and western parts of the area (Fig. 1). In these studies, multi-proxy records of pollen, diatom, carbonate, TOC, TN and δ 13 C of organic matter, grain size Pollen and charcoal records. Pollen data description. Results of fossil sample analysis (101 samples, 67,165 total grains) are shown in the pollen spectra in Figs. 3 and 4. According to the changes in pollen percentages and concentrations, we divided the pollen diagram of the core from Guxu Lake into seven pollen zones, main characteristics of the single pollen zones are presented in Table 2.
The average (n = 101) total pollen concentration was 76,684 grains per gram dry sample, among which the arboreal pollen concentration was the highest (45,580 grains per gram dry sample), that of terrestrial herbs and shrubs was 21,232 grains per gram dry sample, and that of aquatic herbs was as low as 9872 grains per gram dry sample. Moreover, the total concentration of fern spores was relatively high at 16,556 grains per gram dry sample on average. Generally, pollen assemblages were dominated by arboreal pollen (approximately 61%), whereas the average percentage of terrestrial herb and shrub pollen was close to 39% of the pollen assemblages. Moreover, the average percentage of aquatic pollen, calculated by the value of aquatic pollen to total pollen, was as much as 10%. The concentration of charcoal ranged from 0.003 to 6.757 cm 2 per gram dry sample (1.109 cm 2 per gram dry sample on average).
PCA of pollen. The PCA results of pollen taxa and 101 samples of Guxu Lake GDP Core 1 are represented in the pollen diagram and reveal which taxa react similarly to environmental changes (Fig. 5). Together, the first two principal components explained 40.73% of the variation in the pollen data (axis 1: 23.08%; axis 2: 17.65%). Terrestrial herbs and shrubs, as well as coniferous forest, were distributed on the positive end of the first axis while the broadleaved forest mostly scattered on the negative end, which indicates that positive and negative scores based on the Axis 1 (PCA Factor 1) represent dry-cold and humid-warm conditions, respectively. However, no apparent environmental characteristics were revealed by the second axis. Besides, Oryza-type Poaceae and Artemisia reached the highest scores on the negative and positive ends of Axis 2 (PCA Factor 2) respectively, which probably suggest Man-made and non-human environment, correspondingly.
Presence of rice pollen. Before 11.7 cal ka BP, prior to the Holocene. The percentages of Oryza-type Poaceae were rather high, ranging from 0% to 24.9% (approximately 8.6% on average), which is approximately two-fifths that of wild grass Poaceae. In addition, the concentrations of Oryza-type Poaceae were notably as low as 1,049 grains per gram dry sample on average (varying from 0 to 20,070 grains per gram dry sample). www.nature.com/scientificreports www.nature.com/scientificreports/ 11.7-7.9 cal ka BP, early Holocene. The percentages of Oryza-type Poaceae were relatively low, varying from 1.5% to 12.3% (approximately 6.7% on average), whereas that of wild grass Poaceae ranged from 5.0% to 13.2% (approximately 8.7% on average). The concentrations of Oryza-type Poaceae were notably high, at an average value of 14,543 grains per gram dry sample (varying from 383 to 43,980 grains per gram dry sample), whereas that of wild grass was rather high as well, at an average value of 15,417 grains per gram dry sample (varying from 1,804 to 38,405 grains per gram dry sample).
7.9-4.4 cal ka BP, mid-Holocene. The percentages of Oryza-type and wild grass Poaceae decreased to an average value of approximately 4.3% (varying from 2.2% to 12.8%) and approximately 5.3% (ranging from 2.6% to 16.8%), respectively, and their concentrations also decreased to 5,068 grains per gram dry sample on average (varying from 307 to 11,488 grains per gram dry sample) and 6,089 grains per gram dry sample on average (varying from 282 to 12,265 grains per gram dry sample), respectively.

Discussion
Vegetation and environmental history in the Guxu Lake area. The palynological data from GDP Core 1 have a broad similarity to the δ 18 O records of a Greenland ice core-Greenland Ice Sheet Project 2 (GISP2) 52 and stalagmites from Donge Cave 6,53 , Hulu Cave 5 and Sanbao Cave 3,54 in eastern and central China; Qunf Cave in southern Oman 55,56 ; and Moomi Cave on Socotra Island, Yemen 57 (Fig. 6). Based on variations in the pollen record and in the charcoal concentration, in conjunction with changes in sedimentary lithology, the evolution of vegetation and paleoclimate changes around Guxu Lake since the Late Pleistocene can be classified into four main stages as follows: Figure 2. Age-depth model for sedimentary Core 1 of the Guxu Lake obtained by BACON program 48 . Upper panels show the MCMC random walks (left panel, greyscale shading), the prior (green curves) and posterior (grey histograms) distributions for the accumulation rate (middle panel) and memory (right panel). Bottom panel depicts the calibrated 14 C dates (1σ age probability distributions represented in transparent blue) and the age-depth model (darker greys indicate more likely calendar ages; dotted grey lines show 95% confidence intervals; dotted red lines show single 'best' model based on the mean age for each depth). The parameter settings are shown at the upper panels (red font).
www.nature.com/scientificreports www.nature.com/scientificreports/ 23.0-18 cal ka BP, the Late LGM period. The evergreen-deciduous broadleaf and coniferous mixed forest represented by Quercus, Castanopsis, and Pinus was distributed near the mountains and/or hills, whereas the herbs and shrubs, dominated by Poaceae, grew on the plain and the aquatic herb Typha stood at the lakefront. Considering these factors, together with the rather low pollen concentration, the climate in this period was likely relatively cool and dry. This climate corresponds to the Last Glacial Maximum (LGM) (24-18 ka BP), when the global ice sheets reached their maximum integrated volume [58][59][60][61] . However, the pollen record of Guxu Lake (e.g., the lower PCA Factor 1 scores, the higher ratios of AP to NAP) exhibited a warming trend approximately 22 cal ka BP, which was anomalous within the regional context and was likely the result of local processes. Approximately 20 to 19 cal ka BP, a new warming period started and was coincident with continuous global warming at the end of the LGM 62-64 , which is also recorded in the Hulu Cave stalagmite 5 and in the Greenland ice core 52 .  www.nature.com/scientificreports www.nature.com/scientificreports/ 18-11.7 cal ka BP, the Last Deglaciation. The Last Deglaciation (approximately 18-11.5 cal ka BP) is known for various high-frequency oscillations, such as the Bølling/Allerød warm periods 65,66 , the Heinrich event 1 (H1) 59,64,67,68 , the Oldest Dryas (OD) 65,69 and the Younger Dryas (YD) 70-72 cold phases, which are also reflected in the pollen assemblage from Guxu Lake. From 17.2 to 15.6 cal ka BP, the mixed evergreen-deciduous broadleaf and coniferous forest continued to shrink. However, herbaceous plants dominated by Artemisia and Poaceae expanded to a larger area and aquatic plants (mainly Typha) decreased sharply, suggesting that the climate in this period remained cold. The higher PCA Factor 1 scores, together with the lower ratios of AP/NAP associated with cooling and drying, are in strong agreement with the δ 18 O record of the stalagmites from the North Hemisphere 3,5,57 and a Greenland ice core 52 , corresponding to the OD and/or H1 events, as is also recorded in the Greenland ice core and stalagmites 3,5,52,57 . During the period of 15.6-12.6 cal ka BP, the hills were covered by a mixed evergreen-deciduous broad leaved forest, and shrub and herbaceous plants were distributed in the plain and aquatic herbs dominated by Typha, together with fern spores, increased considerably, suggesting a relatively warmer and more humid climate, which is likely a reflection of the Bølling/Allerød warming oscillation. In addition, a notably cold and dry oscillation occurred during approximately 12.6-11.7 cal ka BP, as recorded in GDP Core 1. This oscillation is indicated by the reduction in the evergreen-deciduous broadleaf mixed forest, while herbaceous and shrub plants increased in conjunction with sharply decreased aquatic herbaceous and fern spores and decreased ratios of AP/NAP. This pattern likely correlates with the Younger Dryas event, which is also observed in the Chinese and Yemen stalagmites 3,5,53,57 , as well as in the Greenland ice core 52 Table 2. Pollen zone number, depth, age, characteristic pollen assemblages from GDP Core 1. www.nature.com/scientificreports www.nature.com/scientificreports/ number of fern spores, represented by Polypodiaceae and Ceratopteris, grew under the forest. The abrupt decrease in the PCA Factor 1 scores, the increase in the ratios of AP/NAP, and the high total terrestrial and hydrophyte pollen concentrations during 11.7-7.9 cal ka BP suggest that the start of the Holocene was associated with warm and humid conditions. This hypothesis is consistent with the δ 18 O records from stalagmites in China and Oman 53-56 , as well as in the Greenland ice core 52,65,73 . Meanwhile, the charcoal record of the Guxu Lake presents some noteworthy issues. For example, the overall charcoal concentration in the Holocene is relatively high, especially that reached the peak in the Early Holocene, but there is no regional record of corresponding human activity. This is probably the result of the Holocene climate warming and vegetation development.
Then, 7.9-4.4 cal ka BP, which is known as the Holocene Optimum in the lower Yangtze River according to previous climatic and environmental studies of the middle Holocene e.g.46,74-85 , a warmer and wetter climate was present, as indicated by pollen record. Both pollen and charcoal records indicate an increase in summer precipitation for the Taihu Lake Basin coincident with the Holocene summer insolation maxima 86 . However, the stalagmites in China recorded the wettest phase occurred during ca. 10.2-5.7 ka BP, which seems to indicate inconsistencies with the pollen records from the Guxu Lake (Fig. 6). This is consistent with the time-transgressive Holocene optimum mode 75 though the duration of Holocene optimum has not been beyond debate e.g. 87 .
Multiple-plant remains analysis 88 has been conducted on a Neolithic Yangjia site (ca. 6.3-5.9 cal ka BP) close to northeast of the Guxu Lake (Fig. 1, site 1). Pollen and phytolith records reveal the regional landscape at or around the site was a mosaic of Quercus-Ulmus-Castanopsis-Poaceae vegetation assemblage, and a general warm and humid conditions was present. This is consistent with that inferred from the nearby Guxu Lake. Besides, pollen analysis on other Neolithic cultural deposits in the Taihu Lake region reveals a relatively consistent vegetation landscape and environmental feature of the Majiabang Culture (7.0-5.8 ka BP).
However, a general weakening trend in the Asian monsoon during the Holocene caused by the change in insolation was punctuated by Asian Monsoon (AM) cooling events 6 , among which the 8.2 ka BP, 6.3 ka BP, 5.5 ka BP and 4.4-4.0 ka BP AM events were present in the pollen record of GDP Core 1. Among these events, the GDP Core 1 recorded suddenly reduced arboreal pollen percentage and the ratio of AP/NAP at around 8.2 ka BP (Fig. 6, Global chill), which is correlated with the "Global chill" cold event centered near 8.2 ka BP 89,90 . Cultural interruption during ca. 8.3-8.0 ka BP was witnessed at the Neolithic Kuahuqiao site 91 (Fig. 1, site 10), which is www.nature.com/scientificreports www.nature.com/scientificreports/ likely to be affected by the cooling event. The end of this cooling event marked the beginning of the Holocene optimal in many areas.
The decrease of arboreal and total terrestrial pollen concentrations and the charcoal concentration at approximately 6.3 ka BP (Fig. 6, GDP 1) is also reflected in the δ 18 O record from the Dongge Cave stalagmite in eastern China 6 , which is correlated with an ice-rafted debris event 92 and likely also attributed to preliminary anthropogenic perturbation from the Neolithic Majiabang Culture (7.0-5.8 ka BP) that occupied the (North) Taihu Lake Basin at approximately 6.5 ka BP. This is also present in pollen records of the profile of the Yangjia site 88 (Fig. 1,  site 1).
The cooling event occurred at approximately 5.5 ka BP (Fig. 6, Bond 4) and is correlated with the "Bond event 4" in the North Atlantic 92 , which is also documented in the δ 18 O record of stalagmites in eastern China and Oman 6,55,56 . This event presumably accelerated cultural development in Neolithic China, especially the Liangzhu Culture (5.5-4.3 ka BP), and even indirectly caused the emergence of civilizations in Egypt and Mesopotamia 93 . In addition, the suddenly changing pollen record and abrupt reduced charcoal concentration at approximately 5.0 ka BP (Fig. 6, GDP 2) was probably attributed to expanding and intensified human activities during the Liangzhu Culture according to archaeological records.  61,95 . Climatic change at 4.4-4.0 ka BP, known as "Holocene Event 3" or "Bond events 3" (Fig. 6, Bond 3) 92,96,97 , is presumed to have played a very important role in the shift of Neolithic cultures in China 98,99 . In particular, the collapse of the Liangzhu Culture in the Taihu Lake Basin is considered to have resulted directly from this cooling event 100,101 . This hypothesis is consistent with the pollen record from Guxu Lake and the δ 18 O records of stalagmites from Dongge Cave and Oman 6,55,56 .
In addition, during the late phase (1.9-0 cal ka BP), the forest was dominated by broadleaved evergreen Castanopsis and Quercus growing on the surrounding hills, with Poaceae and Artemisia on the plain and a small population of aquatic herbaceous Typha on the lake shore. Nevertheless, the notable change in the pollen assemblage and concentration reflects a more intensive human influence on local vegetation and the environment, though it might be correlate with the Little Ice Age 102,103 during its early period.
Occurrence and development of rice agriculture. Though pollen of Poaceae comprises a small portion overall, it is can still be used as a reference when it is the dominant terrestrial herb, especially combined with so-called Oryza-type and wild grass. In the Lower Yangtze River, Poaceae pollen with a grain diameter is greater than 40 μm and an aperture diameter wider than 4.0 μm is generally identified as Oryza-type Poaceae 32,104-106 , which will benefit our discussion on rice collection, utilization and even cultivation. In general, proxy data of rice pollen extracted from the Guxu Lake sediments present changes over time (Fig. 6G,H). Prior to the Holocene (before 11.7 cal ka BP), the presence of rice pollen in GDP Core 1 together with a certain amount of rice pollen and wild rice phytoliths present in the parent soil of the nearby Yangjia site 88 , may suggest that wild rice was once distributed in the surroundings of Guxu Lake before human occupation. In the early Holocene (11.7-7.9 cal ka BP), rice pollen was distributed with relatively lower percentage and notably higher concentration, which was most likely due to the warm and humid climate that accelerated the natural distribution and growth of wild rice. During the mid-Holocene (7.9-4.4 cal ka BP), the decreased rice pollen percentage and concentration may correlate with human occupation and rice agriculture that occurred and developed during this period. Moreover, a relatively high concentration of charcoal, approximately 1.868 cm 2 per gram dry sample on average, might also reflect perturbation by humans. Correspondingly, rice remains unearthed from the nearby Yangjia site 88 , including charred rice grains, domesticated rice spikelets, rice phytoliths from rice husks and leaves, as well as rice pollen, illustrated a scenery of rice farming. In addition, a large number of previously excavated Neolithic archaeological sites and previous studies on rice agriculture in this area e.g. 30,31,107-113 were dated to have begun and/or developed in the early and middle part of the Holocene (ca. 7.0-4.0 ka BP), which is in agreement with our results. In the late Holocene (after 4.4 cal ka BP), the significantly increased rice pollen percentage and concentration together with high concentration of charcoal is a clear signal of intensified anthropogenic disturbance that corresponds to widespread intensified farming activities and the control of local fire events in historical periods as well as relatively stable environmental conditions.
The presence and level of rice agriculture in the Taihu Lake area, as an example of typical human activities, could be reflected to some extent by the number and size of Neolithic human occupations. Based on Neolithic archaeological sites in this area [114][115][116][117] and the latest archaeological excavation reports, plots were drawn to demonstrate changes in the quantity and distribution density of 331 known archaeological sites in different periods. Generally, Neolithic archaeological cultures in this area can be divided into three main periods, namely, the Majiabang Culture (7.0-5.8 ka BP), Songze Culture (5.8-5.3 ka BP), and Liangzhu Culture (5.3-4.3 ka BP). The number and density of sites increased significantly from the Majiabang Culture (40) to the Liangzhu Culture (243), then decreased sharply to 8 for the Guangfulin-Qianshanyang Culture (4.3-3.9 ka BP) (Fig. 7), illustrating the intensified human activities in the Taihu Lake area during the most of the Neolithic period (ca. 7.0-4.2 ka BP), which is in good agreement with the Holocene Optimum (Fig. 6).
In this case, human settled locally when the Guxu Lake was relatively stable (ca. 8.0-6.5 cal ka BP) as reflected by the hydrophyte concentration (Fig. 6E). Meanwhile, the first occurrence of human activities since the Majiabang Culture was recorded in the vegetation landscape, such as changing rice pollen and charcoal concentration illustrated in the GDP Core 1 (Fig. 6, GDP 1). In the second stage, the Songze Culture developed slowly, which might be related to the "Bond 4" cold event around 5.5 ka BP. Then, the rapid development of Liangzhu Culture witnessed a fine social structure system consisted of dam, city, noble cemetery, exquisite jade, Scientific RepoRtS | (2020) 10:9306 | https://doi.org/10.1038/s41598-020-65834-z www.nature.com/scientificreports www.nature.com/scientificreports/ and developed paddy field agriculture in generally stable warm and humid conditions during 5.3-4.3 ka BP as recorded in the Guxu Lake sediments. Especially the abrupt reduction of pollen and charcoal concentration around 5.0 ka BP, is probably a reflection of expanding and intensified human interference with natural vegetation. However, a considerable shift occurred at the end of the Neolithic, i.e., the collapse of the Liangzhu Culture. This is coincident with the abrupt cooling event at 4.4-4.0 ka BP, which might exacerbate change within prehistoric society in the Taihu Lake region.
Generally, 'slash-and-burn' cultivation appeared beginning ca. 7.0 ka BP (early Majiabang Culture) and rice agricultural activities expanded in the Taihu Lake area during the Neolithic. Expanding and intensified anthropogenic activities affected the distribution of arboreal plants, and probably resulted in decreased charcoal concentration. Wild rice was supposed to be distributed in this area before Neolithic humans occupied the area, and it may have been collected as supplementary food together with cultivated rice as the main plant food resources, especially during the Liangzhu Culture period.
conclusion Changes in vegetation on the East Asian continent were considered to controlled primarily by the East Asian monsoon, including the East Asian summer and winter monsoons 118 . In the Northern Hemisphere, orbital forcing of insolation increased in the summer and decreased in the winter during the early-to mid-Holocene, causing generally warmer summers and colder winters compared with the present 119  Our main inferences on vegetation evolution and human dynamics are as follows:

Methods
Sample collection and description. According to preliminary surveys and drilling data, we selected the relatively thick sedimentary location (31°30′47.8″N, 120°07′13.4″E, 14.6 m a.s.l.) of Guxu Lake to carry out the Guxu Drilling Project (GDP) in March 2013 and obtained an 18.40-m-long sediment core. Specific lithological features are described as in Table 3.
Generally, the sediments between 17 m and 2 m of the core at Guxu Lake are characterized by light or dark gray silt intercalated with bluish gray silt and gray brown silt interbeds. We divided the core into samples for pollen, phytolith and other analyses per 2 cm segments and chose the upper 10 m with intervals of 10 cm for pollen analysis (101 samples in total).
Dating. Organic sedimentary samples were dated with an accelerator mass spectrometer (AMS) using 14 C in the Beta Analytic Radiocarbon Dating Laboratory (Miami, Florida, USA) and Radiocarbon Dating Laboratory at Peking University (Beijing, China) respectively, then calibrated using IntCal13 120 and OxCal v4.3.2 121 to convert the radiocarbon ages to calendar years.
Pollen and charcoal analysis. A palynological analysis was carried out on the basis of procedures suggested or applied by Moore, et al. 122 and Horrocks 123 . Samples of 5 g dried powder were processed with HCl (37%, 50 ml), KOH (10%, 50 ml), KI/IH (2.0 g/ml in density, 5 ml), acetic acid (50 ml) and an acetolysis mixture (1 ml concentrated H 2 SO 4 , 9 ml acetic anhydride). A tablet of Lycopodium marker (20,848 grains) was added to each sample. The pollen samples were spread uniformly on glass slides, and at least 500 pollen grains, excluding aquatic pollen and spores, were identified and counted using a Nikon Eclipse LV100POL microscope. Identification was aided by the use of reference materials collected by the Key Laboratory of Vertebrate Evolution and Human Origins of Chinese Academy of Sciences, Institute of Vertebrate Paleontology and Paleoanthropology, Chinese Academy of Sciences and published keys 124 . Charcoal analyses were carried out as part of the routine pollen analysis 125 . Charcoal content was estimated using the point count method described by Clark 126 by counting the proportion of 5,500 points of charcoal that were touching an 11-point ocular scale. To reduce potential error from small dark particles that are difficult to assign as charcoal, only points falling on particles greater than 5 μm were counted. Diagrams were constructed using Tilia 127,128 and zoned according to variations in pollen percentages and concentrations using CONISS 129 . Multivariate analysis. To obtain information regarding similar pollen composition among different pollen zones, as well as to determine which pollen taxa exhibited similar reactions in the diagram of pollen percentage, a detrended correspondence analysis (DCA) and principal component analysis (PCA) were conducted based on the square-root-transformed pollen percentage data (except aquatics) using CANOCO 5 software 130 for ordination and plotting a "species-sample bi-plot. " The DCA showed that the pollen percentage data set has a mainly linear structure, resulting in the use of a linear-based PCA 131 . In the PCA, the data were "species-centered. "