Mount Taibai, with a maximum altitude of 3,767 m a.s.l., is the highest part of the Qinling Mountain range and the highest mountain in central and eastern China. The lithology is granite formed by the large-scale incursion of acidic magma during the Yanshan Movement in the Cretaceous (Tian and Huang, 1990). There are numerous Quaternary relict glacial landforms in the Mount Taibai area, with periglacial landforms generally developed in areas below the snow line of ∼2,600 m a.s.l. (Pang, 1998). However, the snow line has risen above Baxiantai, which is the highest peak of Mount Taibai (3,767 m a.s.l.).

There are pronounced differences in precipitation, temperature, topography, and vegetation distribution on the southern and northern sides of the Mount Taibai. Moreover, diverse temperate and subtropical plant species, as well as numerous medicinal herbs, survive in the region (Pang, 1998). The combination of thermal and hydrological conditions results in a vertical distribution of vegetation belts in the area. The area below 780 m a.s.l. is extensively cultivated and the natural vegetation has been substantially altered by human activities. Above this cultivated zone, the following vegetation zones on the northern slope of Mount Taibai can be defined, with increasing altitude: Quercus spp. (Quercus wutaishanica and Quercus aliena var. acuteserrata) forest (780–2,200 m a.s.l.), Betula spp. forest dominated by Betula albosinensis (2,200–2,730 m a.s.l.), coniferous forest divided into Abies fargesii-forest and Larix chinensis-forest (2,730–3,400 m a.s.l.), and alpine shrubs and meadow (>3,480 m a.s.l.).

Daye Lake (33°57’25.90″ N, 107°45’38.39″ E; 3,615 m a.s.l.; Figure 1) is a small, closed glacier lake with a surface area of 0.5 ha and maximum water depth of 18 m, formed after the retreat of the Quaternary glacier activity, located above the tree line on the northern slope of Mount Taibai. It is mainly supplied by precipitation which is concentrated in summer. The regional climate is influenced by the East Asian summer monsoon, and the annual mean air temperature and precipitation are 8.1°C and 720 mm, respectively. The coldest month (January) and warmest month (July) have respective average temperatures of −4 and 19.3°C, based on a 30–year meteorological dataset (CE 1981–2010) from Taibai county meteorological station (1,545 m a.s.l) (see Figure 2 in Zhang et al., 2021). Based on the lapse rates for the northern slopes of the Qinling Mountains proposed by Ma et al. (2017), the mean annual, January and July temperatures are 2.8, −7.3, and 13.6°C, respectively. Daye Lake is surrounded by alpine shrubland and meadow. The shrub taxa are dominated by Salix cupularis, Rhododendron capitatum, and Spiraea, and the local herb taxa are mainly Ranunculaceae, Brassicaceae, and Poaceae (Zhang et al., 2021).

A 2.35 m–long sediment core (DYH16A) was retrieved from Daye Lake in April 2016 using a UWITEC piston corer. The undisturbed core was transported to the laboratory and subsampled at 1 cm–intervals. Seventeen samples of bulk organic sediments were analyzed by accelerator mass spectrometry (AMS) at Arizona University, United States, and Peking University, China (Table 1). The ages of bulk sediment samples may be influenced by old carbon due to the complexity of carbon sources. Therefore, we used the intercept method after fitting a 2nd-order polynomial curve to get the carbon reservoir age which is comparable to the age difference between pollen concentrates and bulk organic from the surface sample (see the Supplementary Material). Finally, 1540 years was selected as the reservoir age of Daye Lake. A chronological model (Figure 2) was developed using the R package rbacon, version 2.5.0 (Blaauw and Christen, 2011) based on calendar ages before present (BP = 1950 CE) and calibrated with the IntCal20 dataset (Northern Hemisphere) (Reimer et al., 2020). Note that only the top 1.26 m of the core were discussed in this study.

The top 1.26 m of core DYH16A at 1 cm–intervals were analyzed in this study. A known number of Lycopodium spores were added to 0.5 g of dried sediment to calculate pollen concentrations (grains g^–1). Samples were then treated sequentially with HCl, KOH, and HF, following the procedures described in Fægri and Iversen (1989) and Moore et al. (1991). Pollen identification was carried out under a light microscope at 400× magnification with reference keys (e.g., Wang et al., 1995; Tang et al., 2016). At least 500 terrestrial pollen grains were counted each sample. Pollen percentages were calculated based on the sum of terrestrial taxa, excluding aquatic taxa and spores. Pollen diagrams were draw and constrained incremental sum of squares (CONISS) analysis was conducted using Tilia software of version 2.0 (Grimm, 1987).

A lower-resolution pollen record from the top 1.03 m of the core DYH16A has been published by Zhang et al. (2021) to reconstruct the regional plant abundances based on the REVEALS model. This record spans the past ∼7000 years which the sediment accumulation rate was relatively uniform. The higher-resolution pollen record presented here spans the entire Holocene and extends to the base of the core where the sediment accumulation rate was lower. But the entire core is continuously deposited which can reflect the Holocene changes of vegetation and climate in the region.

The pollen record from Daye Lake was used for biomes reconstruction in order to explore the vegetation response to Holocene climate change. The procedures used for biomization analysis follow those described in Prentice et al. (1996). First, pollen types (taxa) were assigned to one or more plant functional types (PFTs) (Table 2), and the PFTs were then classified into biomes (Table 3). Finally, the affiliated scores for each biome were calculated based on the pollen percentages and the PFT sets and were then used to represent the abundance of a particular biome. Assignment of the identified pollen taxa from Daye Lake to PFTs and biomes follows Yu et al. (2000) and Ni et al. (2014) for China, together with reference to the modern regional vegetation distribution. Biome definition depends on the influence of the dominant PFTs with the highest scores. Biome affinity scores provide an objective evaluation of the vegetation composition and have been applied to several surface and fossil pollen datasets (Tarasov et al., 1998; Yu et al., 2000; Rudaya et al., 2008; Ni et al., 2014). Biome modeling was conducted using PPPBase software (Guiot and Goeury, 1996), based on three matrixes (PFT × taxon, biome × PFT and biome × taxon matrix) which comprised 51 pollen types, 13 PFTs and 5 biomes in this study (Tables 2, 3).

The reconstructions of palynological diversity can be used to directly interpret the composition and structure of the vegetation.

In order to reconstruct the palynological diversity history over the Holocene in Qingling Mountains region from different aspects, Hill’s indices (Hill, 1973) and rarefaction analysis (Hurlbert, 1971) are used to estimate the diversity richness including palynological richness, Shannon diversity, and Simpson diversity. Palynological richness is the simple count of the number of taxa which represents the total number of taxa in a community to describe the vegetation community; the number of common taxa is known as the Shannon diversity index; and the number of dominant taxa is a measure of the Simpson diversity (Rudaya et al., 2020; Kulkarni et al., 2021). We also calculated evenness index to reflect the community structure and to assess palynological diversity at the regional scale and variations in vegetation composition (Pielou, 1966). The value of evenness is between 0 (only one species in the pollen assemblage) and 1 (each species with the same abundance in the pollen assemblage), the low value indicates the greater difference in the abundance of the plant types in pollen assemblages (Li and Zhao, 2018). All of these indices were calculated using the “vegan” package, version 2.5–7, in R (Oksanen et al., 2020).

The pollen assemblages from Daye Lake (Figure 3) were conducted with CONISS analysis and divided into five significant zones, i.e., three pollen zones and two subzones, suggested by broken stick model. The zones are described below.

Zone I (11,700–10,700 cal yr BP). This zone is characterized by the highest percentages of coniferous trees (average of 42.18%). Among them, Pinus, Abies, and Picea have high percentages, with averages of 27.60, 9.18, and 3.23%, respectively, while Tsuga has the lowest (average of 2.17%). Deciduous broadleaf trees are relatively poorly represented (30. 65%). Among the shrub taxa, Ephedra is well represented (average of 1.26%), while Rosaceae, Fabaceae, and Euphorbiaceae are poorly represented. Herbs are dominated by Artemisia which has an average of 21.45%. The pollen concentration is the lowest during the entire Holocene.

Zone II (10,700–3,800 cal yr BP). This zone is characterized by the lowest proportion of coniferous trees (average of 22.66%) among the three zones and the significant increase of deciduous broadleaf trees (from 30.65 to 48.62%). The obvious increase of Corylus/Carpinus (with average values of 18.51%) and evergreen oak (Quercus_evergreen) are at the expense of the decrease of Pinus (from 27.60 to 9.88%). However, the herb pollen proportion decrease, which is largely due to the decreases in Artemisia (from 9.31 to 5.30%) and Amaranthaceae (from 2.18 to 1.32%), whereas shrub taxa such as Rosaceae (from 2.16 to 5.66%) and Fabaceae (from 0.20 to 1.33%) increase. The pollen concentration is relatively high in this zone. Owing to the different trends of Pinus, Abies, Corylus/Carpinus, Fagus, and Artemisia within this zone, two subzones, i.e., Zone IIa (10,700–6,000 cal yr BP) and Zone IIb (6,000–3,800 cal yr BP) is further divided. Zone IIa shows the lowest abundance of coniferous trees (20.49%) and the highest value of deciduous broadleaf trees (51.48%) in the Holocene. In Zone IIb, Artemisia reach the lowest percentages in the entire record (averages of 5.04%).

Zone III (3,800–0 cal yr BP). There is a minor increase in the percentages of coniferous trees and herbs (average of 26.87 and 23.51%, respectively). The abundance of deciduous broadleaf trees and shrubs shows a decreasing trend (average of 40.80 and 6.10%, respectively). This zone can also be further divided into Zone IIIa (3,800–1,200 cal yr BP) and Zone IIIb (1,200–0 cal yr BP). Zone IIIa is characterized by continuous decrease in deciduous broadleaf trees, increase in coniferous trees and herbaceous plant (which is mainly due to increases in Artemisia and Amaranthaceae). However, Zone IIIb is characterized by the increase of coniferous trees, decrease of deciduous broadleaf trees, and a substantial increase in herbs, especially in Artemisia (from 7.99 to 13.62%, the maximum value over the Holocene) and Poaceae (from 2.04 to 3.09%). The pollen concentration in Zone IIIa is high whereas the one in Zone IIIb is low.

The biome reconstruction from Daye Lake (Figure 4) shows that the coniferous forests (CO) vegetation type displays the highest affinity scores during the first stage of the early Holocene (11,700–10,700 cal yr BP). This vegetation reconstruction suggests drought- and cold-tolerant needles communities played a major role in the regional vegetation cover. However, the temperate deciduous broadleaf forests (TEDE) become dominated during the Holocene interval after which suggests a warm and moist environment during the Holocene. There is a pronounced inverse relationship between TEDE and CO throughout the Holocene.

The Holocene period experiences drastic changes in the biome composition, characterized by a marked increase in TEDE, herbaceous vegetation (STEP) and warm mixed forest (WAMF) scores, which is accompanied by a decrease in the CO scores. The drastic decrease of CO (which is corresponding to the apparent reduce of Pinus, see Figure 3), and the visible increase of TEDE suggests a warm and moist environment during the period of∼10,700 and 6,000 cal yr BP. The period of 6,000–3,800 cal yr BP is characterized by the minimum scores of CO, decreased scores of TEDE, and the lowest scores of STEP, but a visible increased scores of WAMF. CO increase slightly compared with the previous stage from 3,800 to 1,200 cal yr BP. The increasing trends are observed in herbs and WAMF. During the past ∼1200 years, WAMF decrease and STEP increase from low to high values. There is an obvious low value of TEDE between 1000 and 600 years and then the scores increase significantly, whereas conifers show the opposite trend.

Diversity analysis based on pollen records from lake sediments can accurately reflect the richness and composition of the regional vegetation (Stirling and Wilsey, 2001; Birks et al., 2016a). The Shannon and Simpson diversity indexes show similar temporal trends and are positively correlated with evenness (Figure 5). The palynological richness shows an increasing trend, with pronounced fluctuations, throughout the Holocene. Therefore, Shannon diversity and evenness index are employed in the following discussion. During 11,700–10,700 cal yr BP, the palynological diversity increases substantially, corresponding to a trend of increasing evenness. Palynological richness, diversity and evenness index are in increasing trends until 6,000 cal yr BP; however, there is an interval of pronounced low palynological diversity and evenness index during 9,000–8,200 and 7,800–7,000 cal yr BP based on the raw data (gray line in Figure 5). Palynological diversity increases substantially from 6,000 to 3,800 cal yr BP, and then decreases slightly between 3,800 and 1,200 cal yr BP. The decreasing trend is observed from ∼1,200 cal yr BP, which is followed by another increase during the past ∼600 years.

Climate change is a major factor influencing the distribution and abundance of vegetation (Bartlein et al., 1986; Ma et al., 2013); however, the response of vegetation and plant diversity to climate change shows strong regional variations. For example, temperature is the major influence on diversity in high-altitude mountains and high-latitude regions (Hou, 1983; Zhang et al., 2020), while precipitation is more important in arid and semi-arid regions such as the Loess Plateau of China (Shang and Li, 2010; Sun et al., 2013). Elsewhere, such as in southwest China, temperature and precipitation both have major influences on the vegetation (Wang et al., 2018; Trivedi et al., 2020). In the case of east-central China, the Qingling Mountains region, the relationship between climate change and vegetation over the Holocene need to be investigated.

The Qinling Mountain range is a typical region influenced by the East Asian summer monsoon, where plant growth is influenced predominantly by temperature and precipitation (Gansert, 2004; Zhang et al., 2006). The response of the vegetation at Daye Lake to climate change is manifested by changes in community composition and in the relative abundance of the dominant taxa (Figures 3, 4). Coniferous trees such as Abies and Picea can tolerate low temperatures and drought, and generally grow at high altitudes (Achuff and Roi, 1977; Liu Z. L. et al., 2002). By contrast, deciduous broadleaf forest (e.g., Quercus_deciduous, Carpinus_Corylus, and Ulmaceae) is favored by a warm and moist environment (Wang et al., 2018; Xu et al., 2019; Trivedi et al., 2020). Evergreen trees such as Quercus_evergreen and Cyclobalanopsis are widely distributed in subtropical and tropical regions, especially in locations with high temperatures and precipitation (Srivastava et al., 2021). Herbaceous plants such as Artemisia and Amaranthaceae are drought–tolerant (Kolb and Sperry, 1999; Uotila et al., 2021).

In order to determine the vegetation response at Daye Lake to Holocene climate change, we compared the pollen assemblages, biome scores, Shannon diversity and evenness index with paleoclimatic records from other sites in the region. The cited records in the study comprise the following: monsoon precipitation and temperature reconstructions based on pollen records from Gonghai and Daihai Lakes, both to the north of the Qinling Mountains (Xu et al., 2010; Chen F. H. et al., 2015); temperature and precipitation reconstructions based on records of branched fatty alcohol ratio BNA₁₅ and pollen assemblages from the Dajiuhu wetland to the south of the Qinling Mountains, respectively (Huang et al., 2013; Sun et al., 2019); alkenone-based temperature and pollen-based precipitation records from Qinghai Lake, to the northwest of the Qinling Mountains (Herzschuh et al., 2010; Hou et al., 2016; Figure 6).

In the first stage of the early Holocene (11,700–10,700 cal yr BP), coniferous forests with Pinus, Abies and Picea were abundant in the study area, as well as steppe habitats dominated by Artemisia. Deciduous broadleaf taxa were poorly represented in the vegetation (Figures 3, 4). The dominance of coniferous forest indicates that the climate during this period was cold and dry, which is confirmed by the paleoclimatic record from sites elsewhere in the region (Figure 6).

During the early Holocene (10,700–7,000 cal yr BP), there were significant decreases in Pinus and Artemisia, and evergreen trees such as Quercus_evergreen began to appear (Figure 3), and there was a substantial increase in deciduous broadleaf and warm mixed forests. The pollen percentages and biome results indicate that the region was occupied by deciduous broadleaf forests, suggesting that the climate in the early Holocene was warm and humid. At the same time, temperature and precipitation commenced an increasing trend and reached their highest levels in the region of the southern and northern China, and the northeastern Qinghai-Tibet Plateau (Figure 6). However, the climate change at this stage is not stable, and there are fluctuations on a centennial time scale. For example, the abundance of deciduous broadleaf taxa decreased at 9,000–8,200 cal yr BP and 7,800–7,000 cal yr BP in Daye Lake, revealing deteriorating climate, while it corresponded to the declining of temperature and precipitation in Qinghai Lake and Dajiuhu wetland, respectively. And the Gonghai Lake in the northern China also has a low precipitation.

In the middle Holocene (7,000–3,000 cal yr BP), coniferous forest expanded slightly, as indicated by increases in Pinus and Picea pollen percentages, while percentages of thermophilous deciduous broadleaf taxa gradually decreased. The representation of Artemisia reached the lowest level in the profile. The hydrothermal combination at this time were probably worser than during the previous stage; nevertheless, the climate was still conducive to the growth of warm-temperate mixed forest (Figure 4). The substantial climatic differences existed between the northern, western, and southern parts of China at this time. At Qinghai Lake in the northeastern Tibetan Plateau, to the northwest of the Qinling Mountains, there was strong decrease in temperature which reached its lowest level during the Holocene, probably because of intensification of the Westerlies (Hou et al., 2016); while precipitation decreased substantially. However, temperatures at sites to the north of the Qinling Mountains were close to those sites to the south. At Daihai Lake, temperatures reached their highest level during the Holocene, while in the Dajiuhu wetland the temperature remained at a high level. This dipolar pattern of precipitation is a feature of the climate of China on various timescales in north and south of China and is discussed in detail in Chen J. H. et al. (2015). Overall, our results of pollen percentages and biome reconstruction may indicate that the changes of precipitation pattern at Daye Lake during the Holocene were between northern and southern China.

During the late Holocene, especially after 3,000 cal yr BP, Pinus and Artemisia commenced a pronounced increasing trend and there was the increased representation of herbaceous vegetation (Figure 3). The biome reconstruction shows an increase in coniferous forests and herb-dominated vegetation, accompanied by decreases in deciduous broadleaf and warm mixed taxa (Figure 4), indicating a climatic deterioration. It is noteworthy that decreases in temperature and precipitation occurred at other sites adjacent to Qingling Mountains region (Figure 6).

Pollen is an important indicator for the reconstruction of past plant diversity. There have been many studies on plant diversity based on the palynological diversity (Xiao et al., 2008; Liang et al., 2019; Rudaya et al., 2020; Mottl et al., 2021). Even though there are several uncertainties involved in reconstructing plant diversity from fossil pollen records. For example, palynological diversity cannot represent the full range of plant diversity due to the limitations of pollen identification, sample size, and the complexity of depositional processes (Odgaard, 2001; Weng et al., 2006; Meltsov et al., 2011; Birks et al., 2016a; Liang et al., 2019). Several taxa can only be identified to family level, and poorly represented entomophilous taxa can lead to the underestimation of plant diversity. However, the vegetation of the Qinling Mountains is dominated by woody plants which are identified mainly at the genus level (4 families + 16 genera of trees and shrubs in Figure 3). And at least 500 pollen grains were counted for each sample in this study to ensure a statistically adequate and representative sample of observations. Moreover, Zhang et al. (2021) suggested that the pollen loading in Daye Lake is predominantly wind–transported, indicating that the palynological diversity reconstructed in this study is representative of the vegetation of the Qinling Mountains region. Nevertheless, several studies have shown that palynological diversity is positively correlated with plant diversity (Odgaard, 1999; Meltsov et al., 2011; Birks et al., 2016a; Felde et al., 2016). Overall, despite these potential limitations of palynological diversity, we assume that the pollen records from Daye Lake can provide a reliable record of changes in plant diversity in the Qingling Mountains region during the Holocene.

The record of palynological diversity at Daye Lake is in good agreement with records of regional climate change (Figure 6), indicating that palynological diversity was highly dependent on the evolution of the regional climate. The low values of palynological diversity and evenness index at the beginning of the early Holocene are consistent with the relatively low temperature and precipitation indicated by the paleoclimate reconstructions. The increased palynological diversity and evenness index during the early Holocene is also consistent with the inferred warm and humid climate, and numerous studies have demonstrated that intervals of warm and humid climate are characterized by high vegetation cover and high palynological diversity in northwest, southwest and Loess Plateau regions of China (Fang et al., 2003; Zhang et al., 2005; Xiao et al., 2008; Zhao and Ding, 2014). Notably, palynological diversity and evenness index at Daye Lake continued to increase during the middle Holocene (7,000–3,000 cal yr BP), even though the precipitation decreased; temperatures remained at a high level (Figures 6A, F), the environmental conditions in the Qingling Mountains continued to favor a high degree of palynological diversity, indicating that the vegetation in the Qinling Mountains was more sensitive to temperature than to precipitation. The palynological diversity also commenced a decreasing trend after ∼3,000 cal yr BP, and it’s clear that the environmental conditions in the area became less favorable for certain plant species, leading to their decreased representation or disappearance. Cheng et al. (2017) also suggested that the alpine timberline on Mount Taibai moved upward to a higher altitude during ∼4,200–2,000 cal yr BP and started to move downward from ∼2,000 cal yr BP. The visible decrease in the palynological diversity of the Daye Lake between 1,200 and 600 cal yr BP may also be resulted from the cold and dry environment which was consist with the sudden decrease in temperature and precipitation in the northern and southern of the Qinling Mountains within the recognizable range of the age error. At the same time, archeological evidence indicated that this period corresponded to the cultural and economic prosperity of the Song Dynasty (Archaeological research institute of Sui-Tang Dynasties of Shanxi Academy of Archaeology, 2018) which revealed the possible influence of human activities on the vegetation. Interestingly, during the past few centuries, palynological diversity increased, accompanied by an increase in temperature and a decrease in precipitation (Figure 6). This seems to support evidence that temperature had been a major factor affedting vegetation evolution in Qinling Mountains region, and that the rising temperatures resulting from global warming had a positive effect on palynological diversity in the region. However, the possible roles played by human activities should also be considered. For example, during the past several hundred years, the increases in Artemisia, Amaranthaceae, and Poaceae pollen percentages in the sediments can be attributed to the development of agriculture in low-altitude areas and open land may be one of the reasons for the increase in palynological diversity (Woodbridge et al., 2021).

Overall, the results of this study indicate that plant diversity in the Qinling Mountains during the Holocene was driven primarily by climate change. The relatively warm and humid environment to the south of the Qinling Mountains has resulted in a higher plant diversity in southern China than in northern China (Sun et al., 2009; Shang and Xing, 2016). Sun et al. (2013) concluded during the warm period of the Holocene, subtropical vegetation in the south expanded and migrated northward into the Qinling Mountains region. The impact of climate on vegetation is also exerted via extreme climatic events such as droughts, floods, and abrupt cooling, which lead to species extinctions and the reduction of biodiversity (Srivastava et al., 2021). Park et al. (2019) showed that in coastal East Asia successive droughts during the ∼8.2 ka caused forest damage and a major reduction of plant diversity. Additionally, Tao et al. (2013) demonstrated that in northwest China the cold and dry climatic event at 4.2 ka resulted in the conversion of grassland to desert, with a greatly reduced plant diversity and vegetation structure. However, the pollen record from Daye Lake seems not to contain clear evidence for the influence of abrupt climatic events on palynological diversity (Figure 6). Notably, climate change will affect the vegetation phenology which will subsequently impact the plant diversity. For example, changes in seasonal temperatures will alter the flowering time of plants, with increases in annual average temperature overall resulting in an earlier flowering time (Hart et al., 2014; Yu et al., 2016).

There is a long history of human impacts on the composition and structure of vegetation, associated with the development of agriculture and the use of fire (Ruddiman, 2003; Rudaya et al., 2020; Kulkarni et al., 2021). In southern China significant agricultural activity began ∼6000 years ago (Cheng et al., 2018), while the large-scale alteration of the vegetation composition and ecosystems has occurred during the past few millennia (Lewis and Maslin, 2015; McMichael, 2021). In the Qinling Mountains, agricultural activity began as early as ∼8000 cal yr BP, associated with cultures such as the Peiligang and Yangshao (Zhang et al., 2018). However, even though a long history of human activities was documented in this area, there is no ambiguous evidence for human disturbance of the vegetation at Daye Lake, possibly due to its high altitude (3,615 m a.s.l.), location above the forest line (Zhang et al., 2021). Liu H. et al. (2002) suggested that the manifest influence of human activities on vegetation is concentrated on the low latitudes in Mount Taibai.

In summary, the temperature of the Qinling Mountains has increased in recent decades due to global warming, which will impact the regional vegetation composition and palynological diversity (Li et al., 2018; Huang et al., 2020). Additionally, the palynological diversity of the Qinling Mountains region may increase with the ongoing climate warming in the future when the direct human interference is absent.