Introduction

Vegetation-based palaeoclimatic reconstruction can be used as a valuable input for modelling future global climate change. However, human interventions and influences on both vegetation and climate in the late Holocene can result in uncertainties in such modelling efforts. Owing to increasing interest in this field, several recent studies have focused on the vegetation and climate changes that occurred during the late Holocene, and particularly the influence of human activities on vegetation changes (Liu and Qiu 1994; Faust et al. 2004; van der Linden et al. 2008; Xu and Zhang 2013; Li et al. 2014; Turner et al. 2014; Cichon and Niedzielski 2015; Tian et al. 2015). Additionally, some authors have reported significant climate and environmental changes during the medieval warm period (MWP) and the little ice age (LIA) (Chu et al. 2002; Cronin et al. 2003; Driese et al. 2004; Tan et al. 2009; Zeng et al. 2012; Quamar and Chauhan 2014; Sadori et al. 2015). Mountain peat sediments in particular are capable of preserving evidence of vegetation and climate change history, including the events such as the MWP and LIA (Viau et al. 2012; Zhang et al. 2014; Chen et al. 2015). Thus, studies into the dynamic vegetation landscape and contemporary climate conditions based on multiple proxy data from mountain peat sediments can be considered as critically important for understanding the late Holocene palaeoenvironment.

Southeastern subtropical China lies directly in the path of the summer monsoon circulation, and is a key region for the study of monsoon variability and vegetation reconstruction during the late Quaternary. Daiyun Mountain, in Fujian Province, is located in this subtropical region of southeastern China, and is significantly affected by monsoon changes. As such, the changing East Asian Monsoon climate may be recorded as vegetation changes upon Daiyun Mountain, and therefore reconstruction of the palaeovegetation of this region is of great importance. The peat sediments of Daiyun Mountain are predominantly located within a nature reserve and are surrounded by low hills supplied with abundant atmospheric precipitation. Thus, with typical swamp-like conditions, the Daiyun Mountain peat is extremely sensitive to climatic changes, and offers an ideal site for studying regional climatic variability under a changing global climate. To date, the majority of Holocene pollen studies in Fujian Province have focused on coastal areas, flood plains and river deltas, while mountain areas have been somewhat neglected (Lan et al. 1986; Zheng 1991; Jin and Zheng 1993; Wu 1993; Wang et al. 1995; Chen et al. 1998a, b; Zheng et al. 2002; Zhi et al. 2003). Many studies have reconstructed vegetation and environmental changes (Liu and Qiu 1994; Qiu 2006; Xu et al. 2013; Huang et al. 2014; Yue et al. 2012, 2014) as well as the impact of human activities (Liu and Qiu 1994; Qiu 2006; Xu et al. 2013) in the monsoonal regions of southeastern China. Despite this, the climatic conditions, vegetation history and the impacts of human activities on Daiyun Mountain during the late Holocene remain poorly understood (Liu and Qiu 1994; Qiu 2006). Earlier studies in this region have relied on low-resolution sequences with poor chronological control (Liu and Qiu 1994; Qiu 2006). In contrast, the high-resolution multi-proxy record presented in this study offers a chronology based on precise accelerator mass spectrometry (AMS) radiocarbon (14C) dating of a peat profile from Daiyun Mountain. Using this technique, we reconstruct the late Holocene vegetation and climate changes on a regional scale and analyze human impact on the vegetation and environment over the last millennium. Our study provides additional data for research into global change, both in terms of tracing past conditions and modelling future scenarios.

Study area

The peatland selected for this study is located in a nature reserve on Daiyun Mountain, Fujian Province in southeastern China. The Daiyun Mountain nature reserve (25°38′07″−25°43′40″N, 118°05′22″−118°20′15″E) is located 40 km north of Dehua (Fig. 1). The Daiyun Mountain range occupies the central part of this province, and comprises several peaks, including Daiyun (1,856 m a.s.l.), Jiuxian (1,655 m a.s.l.), Shiniu (1,781 m a.s.l.) and Zhiyundong (1,629 m a.s.l.). The mean annual temperature in the region is 15.6–19.5 °C and the mean annual precipitation is 1,700–1,900 mm. Daiyun Mountain is a subtropical oceanic climate under the strong influence of the southeast monsoon. Around 80 % of the annual precipitation falls during the rainy season between April and September (Lin 2002).

Fig. 1
figure 1

Maps showing a, the geographical location of the study site and the modern vegetation of the study area: 1 Wanxiang cave (Zhang et al. 2008), 2 Dongge cave (Wang et al. 2005), 3 Pingnan peat (Yue et al. 2012), 4 Lianhuachi site (Qiu 2006), 5 Daiyun Mountain peat (DYS2), 6 Lake Huguangyan Marr (Yancheva et al. 2007; Wang et al. 2007), b modern vegetation of the DYS2 profile

Full size image

More than 90 % of the area is covered by subtropical evergreen broadleaf forests, mainly comprising evergreen trees such as Quercus and Castanopsis (Fig. 1). The vegetation community varies with altitude (Fig. 2a): below 500 m, the evergreen broadleaved forest is dominated by Castanopsis tibetana, C. eyrei and C. fabri, among others; from 500 to 1,000 m on the southern slopes of the mountain, broad-leaved evergreen forests, coniferous forests, and coniferous and broad-leaved mixed forests are dominated by species that include Cunninghamia lanceolata, Pinus massoniana, Castanopsis fargesii, C. eyrei, Cyclobalanopsis myrsinifolia, Taxus chinensis var. mairei and Cryptomeria fortunei; from 500 to 800 m on the northern slopes of the mountain, forests with some deciduous features are often interspersed with evergreen broadleaved forests, conifer forests, and conifer and broadleaved mixed forests. The species within the deciduous forests include Quercus variabilis and Liquidambar formosana; from 1,000 to 1,600 m on the southern slopes of the mountain and from 800 to 1,600 m on its northern slopes, vegetation is dominated by coniferous trees and wetland plants, which mainly include Cryptomeria fortunei, Tsuga longibracteata, Juncus effusus, Isachne globosa and Eleocharis tetraquetra; above 1,600 m, the mountain summits are covered by broad-leaved evergreen shrubs dominated by Rhododendron simsii, R. mariesii and Eurya saxicola.

Fig. 2
figure 2

a Vegetation composition with altitude. b View of the peatland studied. c The vegetation landscape surrounding the DYS2 profile

Full size image

The peatland investigated in this study developed extensively during the Holocene. The total peat-covered area on Daiyun Mountain is approximately 100 m by 200 m, and the marshy environment allowed the accumulation of peat layers ranging from 1 to 3.2 m in thickness. Due to the area’s poor drainage and the absence of any significant water bodies, the peatland has a stable sedimentary setting, with well-preserved continuous autochthonous sediments.

Materials and methods

Field sampling and dating

A 100 cm deep profile was excavated from the peatland on Daiyun Mountain in 2010 (25°40.040′N, 118°12.381′E; altitude, 1,579 m a.s.l.). Terrestrial plant macrofossils, peat (60–180 μm) and pollen samples were collected from four locations and dated using AMS (Table 1). Three samples were dated at the AMS Dating Laboratory of Beijing University, China, and one was dated by Beta Analytic, USA. Pollen extracts contained both terrestrial and aquatic pollen, in addition to other organic matter. While pollen dating can be problematic when aquatic plants are involved, we consider the use of pollen extraction for dating to be effective in reducing the influence of old carbon to give more reliable ages, compared with bulk sample dating techniques. Many previous studies have indicated that pollen extraction from sediments is able to give more reliable stratigraphic ages (Zhou et al. 1997, 2005). All dates were calibrated to calendar years before present (0 = ad 1950) using CALIB Rev. 6.0.1 with the IntCal 09 calibration data set (Stuiver and Peimer 1993; Stuiver et al. 1998; Reimer et al. 2009). The age-depth model was established by fitting spline functions to the age control points using the Bacon model (Blaauw 2010); this task was carried out using the statistical software package R.

Table 1 AMS 14C dates from DYS2 (this study) and the Lianhuachi site (Qiu 2006) from Fujian province, China
Full size table

Pollen analysis

Forty-five samples were taken at 2 cm intervals for pollen analysis. Each sample contained 1–2 cm3 of peat material. Sample preparation followed the standard procedure, using 10 % HCl to remove carbonate and 10 % KOH to remove organic matter. Finally, pollen and charcoal were extracted using a heavy liquid (2.1 g ml−1) composed of HBr, Zn and KI (500 ml: 120 g: 510 g) and sieved on a 7 μm mesh screen to obtain a higher concentration of pollen grains (Nakagawa et al. 1998). The concentrate was mounted in glycerol gel and Lycopodium tablets (27,637 grains/tablet) were added to each sample to enable calculation of the pollen concentration (Maher 1981). Each pollen sample was counted under a light microscope at ×400 magnification and 1,000× was used for critical identifications. Pollen counts were usually >500 terrestrial pollen grains. Pollen identification was carried out using published pollen atlases by Wang et al. (1995) and Xi and Ning (1994), as well as the modern pollen collections of the laboratory. Pollen percentages of trees, shrubs and dry land herbs were calculated based on their pollen sums. The relative pollen percentages of wetland herbs and fern spores were then established from the sums of all counted palynomorphs. The pollen diagram was plotted using Tilia 1.7.16 (Grimm 2011) and pollen assemblage zones were provided by stratigraphically constrained cluster analysis (CONISS) (Grimm 2011).

The charcoal record was prepared from pollen-slide charcoal counting in widely spaced profile samples. Micro-charcoals larger than 50 μm were counted in order to calculate their accumulation rates, as they may be used to reconstruct fire history, to indicate palaeovegetational and palaeoenvironmental changes and to mark any significant fire event in the study region (Patterson et al. 1987; Lv et al. 2002).

Geochemical analysis

The degree of peat humification, as indicated by fluctuations in absorbance, was determined following the alkaline extraction method described by Blackford and Chambers (1993). The analyses were performed using a personal computer-based 2102 UV–Visible spectrophotometer in the Soil Laboratory of the School of Geographic and Oceanographic Sciences, Nanjing University, China. The sediment core was sampled at 1 cm intervals and a total of 85 samples, each weighing 0.1 g, were used for the humification analysis.

Eighty-four subsamples weighing 2 g each were taken from depths of 3–86 cm in the sediment core for loss-on-ignition (LOI) analysis. Sequential combustion at 550 °C was used to estimate the organic matter content (Dean 1974).

Data analysis

Numerical analyses were conducted using only those taxa that occurred in at least three samples with a percentage of >1 %. A total of 30 pollen taxa were used and the analysis was carried out using CANOCO v.4.52 (Ter Braak and Smilauer 2003). Detrended correspondence analysis (DCA) was used to determine whether linear or unimodal based techniques should be employed in the subsequent ordination analysis. The length of the gradient was 1.363 standard deviation (SD) units, which was less than 3. Therefore, a principal component analysis (PCA) (Birks and Gordon 1985; Birks 1998) was performed to analyze the pollen assemblages using inter-species correlations and square-root transformations of pollen percentages.

Results

Lithology and radiocarbon dates

Four stratigraphic units are identified, and are presented in Fig. 3. Unit 1 (uppermost 32 cm) consists of yellow–brown peat, with an organic matter content of 24.28 %. Unit 2 (32–61 cm) is distinguished by its black colour and high organic matter content of around 40 %; this banded peat section is dark with highly humified peat layers (Fig. 3). Unit 3 (61–90 cm) consists of brown peat with an organic matter content of 22.92 %. Finally, Unit 4 (90–100 cm) consists of sand and gravel with the lowest amount of organic matter.

Fig. 3
figure 3

Age/depth model based on four calibrated AMS 14C dates (see Table 1 for details). The chronology for the section 0–7 cm is difficult to read in the chronology curve, but consisted of a linear function for this short section

Full size image

The four calculated dates (in calendar ages) were used to construct the age-depth model, which was established by fitting spline functions to the age control points using the Bacon model (Blaauw 2010) (Fig. 3). This chronology indicates that the peat profile spans the last 4,460 years. Therefore, the temporal sampling resolution for the lithological analysis is 45 years for each contiguous 1 cm interval, while each centimetre represents 90 years of the fossil pollen record. Ages based on this model are used in the discussion section below. The lower unit (77–100 cm) was found to have a sedimentation rate of 3.71 mm a−1; the upper units (0–7, 7–57 cm) had similar rates of 0.363 mm a−1 and 0.449 mm a−1, respectively. However, the sedimentation rate of the middle unit was considerably lower, at 0.080 mm a−1. Comparing the lithologies and peat accumulation rates at our study site with those at the nearby Lianhuachi site (25°40′15″N, 118°11′12″E; 1,550 m a.s.l.) reported by Qiu (2006), we can see that the peat depths in both areas are the same (approximately 90–100 cm), and that the accumulation rates in the upper units (approximately 0–60 cm) of the two sites are high, while those of the lower units are low (Table 1). The correspondence of our results with those from pollen stratigraphy at Lianhuachi confirms the reliability of the chronology constructed in this study.

Fossil pollen data

A total of 96 pollen and spore types were identified in the 45 samples, including 57 trees and shrubs, 33 herbs, six ferns and three types of moss and algal palynomorphs. The identified pollen taxa spanned coniferous, tropical, subtropical and temperate vegetation types. The coniferous taxa mainly included Pinus, Picea, Tsuga, Podocarpaceae and Taxodiaceae. The tropical and subtropical broad-leaved taxa mainly include Castanopsis (including Lithocarpus), Ilex and evergreen Quercus (including Cyclobalanopsis). The temperate broad-leaved taxa include Alnus, Carpinus, Betula, Pterocarya, Fagus, deciduous Quercus, Ericaceae, Corylus, Carya, Juglans, Acer, Tilia and Rosaceae. The main dry land herbaceous taxa are Poaceae, Asteraceae, Artemisia, Lamiaceae, Humulus, Brassicaceae, Fabaceae, Apiaceae, Thalictrum and Campanulaceae. The main wetland herbs are Cyperaceae, Liliaceae, Polygonum, Ranunculaceae, Potamogetonaceae, Myriophyllum and Typha. The identified spores mainly consist of Polypodiaceae, Hymenophyllaceae, Pteris, Selaginella, Parkeriaceae and trilete spores. Two pollen zones, each with two subzones, were identified (Figs. 4, 5).

Fig. 4
figure 4

Pollen percentage diagram of the DYS2 profile with the 28 main taxa and the four main pollen zones obtained from a constrained cluster analysis. The pollen sum used is “trees and shrubs (AP) + dry land herbs (NAP)”, except for wetland herbs and ferns, which use the pollen sum “trees and shrubs + dry land herbs + wetland herbs + ferns”; shadow is ×5 magnification

Full size image
Fig. 5
figure 5

PARs from the DYS2 profile with the 28 main taxa and the four main pollen zones obtained from a constrained cluster analysis; shadow is ×5 magnification

Full size image

Zone 1 (91–48 cm depth; 4,350–1,000 cal bp): pollen percentages are dominated by evergreen Quercus (ranging from 9.7 to 26.9 %, mean of 19.7 %) and Castanopsis (mean of 2.3 %), with some Castanea, Liquidambar, Ilex and Ericaceae. The pollen percentage of dry land herbs fluctuates between 37.9 and 66.2 % with Poaceae as the major taxon (mean 38.8 %). In addition, there is a relatively high percentage of wetland herbs (mean 7.3 %) and a low percentage of fern spores (mean 8.6 %). The pollen accumulation rate (PAR) fluctuates between 632 and 138,420 grains cm−2 a−1. This zone is characterized by the highest abundance of evergreen Quercus but the lowest abundance of herbaceous pollen. The zone can be divided into two subzones: Subzone 1-1 (91–62 cm depth; 4,350–1,550 cal bp) contains the highest percentage of evergreen Quercus and the lowest Poaceae percentage within the entire profile. In addition, the percentages of Castanopsis, Castanea, Liquidambar, Ilex and Cyperaceae are relatively high, whereas the percentages of Asteraceae and Artemisia are low. The PAR is low, at 35,093 grains cm−2 a−1. Subzone 1-2 (62–48 cm depth, 1,550–1,000 cal bp) is marked by a sharp increase in Poaceae and a slight decrease in evergreen Quercus. However, pollen assemblages are still dominated by evergreen Quercus. The PAR is 37,770 grains cm−2 a−1.

Zone 2 (48–0 cm depth; 1,000–0 cal bp): This zone is characterized by the highest percentage of dry land herbs (mean 78.78 %); however, it has the lowest percentage and PAR of trees and shrubs (mean 21.22 %, 4,837 grains cm−2 a−1). In addition, there is a relatively low percentage of wetland herbs (mean 4.4 %) and a high percentage of fern spores (mean 22 %). Poaceae, Pinus and Asteraceae increase markedly, while evergreen Quercus shows a notable decrease. This zone can also be divided into two subzones. Subzone 2-1 (48–27 cm depth; 1,000–550 cal bp) is marked by the highest percentage of dry land herbs and the lowest percentage and PAR of trees and shrubs. Pinus, Poaceae and Asteraceae increase markedly, while evergreen Quercus, Castanopsis, Castanea, Liquidambar and Ilex decrease considerably. There is a relatively low content of wetland herbs and a high content of trilete spores. Subzone 2-2 (27–0 cm depth; 550–0 cal bp) also contains a high percentage and high PAR of dry land herbs and a low percentage and PAR of trees and shrubs. The percentages and PARs of Poaceae, Pinus and Asteraceae are high, but slightly less than in Subzone 2-1, whereas the percentages and PARs of evergreen Quercus, Castanopsis, Castanea, Liquidambar and Ilex are low. There is a relatively low content of wetland herbs and a high content of trilete spores.

PCA results

PCA results based on pollen samples and pollen taxa reflect the characteristics of the pollen assemblages and effectively summarize the main vegetation changes (Fig. 6). The first two principal components (Axis 1 and Axis 2) capture 83.8 and 6.5 % of the total variance in the data set. PCA Axis 1 (PCA-1) mainly reflects relative changes between warm temperature taxa such as evergreen Quercus, Ilex, Fagus and Tsuga, as shown on the right side of Fig. 6a, and cold-tolerant taxa such as Poaceae and Asteraceae, as shown on the left side of the figure. PCA Axis 2 (PCA-2) separates the mesic taxa such as Ranunculaceae from the drought-resistant ones such as Artemisia. Based on the ecological affinities of individual taxa, it is evident that Axis 1 therefore represents changes in summer temperature, because pollen assemblages typically reflect vegetation conditions during the flowering seasons. In contrast, Axis 2 represents changes in effective moisture. Four clusters can be identified on the basis of the PCA scores along the first two axes, as shown in Fig. 6b; these correspond to the pollen assemblage zones defined by CONISS (Grimm 2011) (Figs. 4, 5).

Fig. 6
figure 6

PCA results for fossil pollen data from DYS2 (Campa: Campanulaceae; Miche: Michelia; Poace: Poaceae; Ligus: Ligustrum; Myrio: Myriophyllum; Apiac: Apiaceae; Arter: Artemisia; Potam: Potamogetonaceae; Aster: Asteraceae; Ranun: Ranunculaceae; Carpi: Carpinus; Brass: Brassicaceae; Reere: Reevesia; Taxod: Taxodiaceae; Liqui: Liquidambar; E-Quer: evergreen Quercus; Casis: Castanopsis; Casta: Castanea; Erica: Ericaceae; Lamia: Lamiaceae; Sympl: Symplocos; Rosac: Rosaceae; Cyper: Cyperaceae)

Full size image

Humification and LOI

Figure 7 shows that both the mean degree of humification (17.52 %) and LOI (26.45 %) are low during the period 4,350–1,000 cal bp. Organic matter varies between 25 and 41.18 % during 1,000–550 cal bp. After 550 cal bp, the mean LOI is relatively lower (22.72 %). Additionally, the degree of humification shows a decreasing trend from 1,000 cal bp to 550 cal bp, with the lowest mean value (10.2 %) found between 550 cal bp and the present.

Fig. 7
figure 7

Comparison of multiproxy data in the DYS2 profile and pollen percentages (trees + shrubs) from Lake Huguangyan Maar (Wang et al. 2007), of evergreen Quercus from the Pingnan peat (Yue et al. 2012) and magnetic susceptibility from Lake Huguangyan Maar (Yancheva et al. 2007)

Full size image

Discussion

Vegetation history and climate since 4,350 cal bp

Knowledge of the size of the pollen catchment is essential for interpreting pollen assemblages in terms of pollen source areas. Typically, the boundary for local pollen ranges between 20 and 30 m within the sedimentary basin. The presence of local pollen is therefore a reflection of the surrounding vegetation and has important implications for the reconstruction and succession of local vegetation (Bradshaw 1981; Jacobson and Bradshaw 1981; Prentice 1985; Xu and Zhang 2013). In contrast, regional pollen refers to that pollen which covers a range of 100–2,000 m within the sedimentary basin and is important for studies into vegetation reconstruction and climate change (Nielsen and Sugita 2005; Wang and Herzschuh 2011; Xu and Zhang 2013). The peat bog investigated in this study is approximately 100 m wide and 200 m long, and has accumulated a 1.0 m thick peat layer. As this peatland is enclosed without the presence of a river, we can consider the sources of pollen to be both local and regional. This allows a high resolution reconstruction of vegetation history and climate changes in the Daiyun Mountain region during the late Holocene.

4,350 to 1,550 cal bp

This pollen zone was characterized by high pollen percentages of wetland herbs, trees and shrubs, as well as a small pollen percentage of dry land herbs. Taxa typical of the warm subtropics, such as evergreen Quercus, Castanea, Castanopsis, Ilex and Liquidambar, dominate this zone. Poaceae and Lamiaceae are the major contributors to the dry land herbs, while Cyperaceae and Ranunculaceae are the major wetland herbs. The vegetation mainly comprised evergreen forests dominated by broad-leaved trees, indicating a warm and wet climate at this time. Previous studies have also found high pollen percentages of Castanopsis and evergreen Quercus in the Fuzhou basin (Yue et al. 2014) and the Pingnan peat (Yue et al. 2012) during 4,000–1,000 cal bp, supporting the interpretation of a warm climate during this period. Additionally, results from Lake Huguangyan Maar (Fig. 7; Wang et al. 2007; Yancheva et al. 2007; Wu et al. 2012), indicating low magnetic susceptibility (also see Fig. 7) and high total organic carbon, support the interpretation of a warm climate during 4,350–1,550 cal bp.

1,550 to 1,000 cal bp

The pollen diagram shows a decrease in the pollen percentage from trees and shrubs, but increased pollen percentage and PAR from dry land herbs. Evergreen Quercus, Castanopsis, Castanea, Ilex and Liquidambar decline significantly, while Poaceae and Asteraceae increase. Additionally, wetland herbs, predominantly Cyperaceae and Ranunculaceae, show a significant decline. The vegetation at this time was therefore dominated by evergreen forests, suggesting a warm and wet climate, similar to the previous stage. Wang et al. (2007) also reported high relative pollen abundance of trees and shrubs from Lake Huguangyan Maar in southern subtropical China, while Yancheva et al. (2007) reported low magnetic susceptibility in the same lake. Both of these results support our interpretation of a warm and wet climate during this period.

1,000 to 550 cal bp

This zone is characterized by a high pollen percentage and PAR of dry land herbs, and low pollen percentages and PARs of trees, shrubs and wetland herbs. The vegetation mainly comprised degraded forests at this time, with a sharp decline seen in trees and shrubs, corresponding to an abundance of well-developed herbs, indicating ecological succession due to disturbance of the vegetation. A significant change in the vegetation occurred after 1,000 cal bp, with a sharp reduction of evergreen trees including Quercus, Castanopsis, Castanea, Ilex and Liquidambar, and an increase in Poaceae and Asteraceae. LOI, absorbance and the ratio of evergreen to deciduous groups are high, whereas the wetland herbs are scarce during this period, suggesting that a cool and dry climate prevailed in the Daiyun Mountain region at this time (Fig. 7). Most of the late Holocene pollen records from southeastern China consistently show a gradual decline in evergreen forests, such as those dominated by evergreen Quercus, after 1,000 cal bp (Liu and Qiu 1994; Qiu 2006; Yue et al. 2014). It is believed that the climate in southeastern China became cooler during the late Holocene, probably associated with a weakening of the summer monsoon.

550 cal bp to present

The pollen percentage and PAR of trees and shrubs are low, whereas dry land herbs, largely composed of Poaceae and Asteraceae, are high at this time. The vegetation had become increasingly degraded, as indicated by the sharply decreased pollen percentage and PAR from trees and shrubs and plentiful pollen from herbs. LOI, absorbance and the ratio of evergreen to deciduous groups are low whereas wetland herbs are abundant (Fig. 7), suggesting a wet and cold climate since 550 cal bp. A previous study by Wang et al. (2007) reported a similar low abundance of tree and shrub pollen and Yancheva et al. (2007) documented low magnetic susceptibility from Lake Huguangyan Maar in south China. Additionally, a number of pollen records also show a gradual decline in trees, especially evergreen forests, since 550 cal bp in Fujian Province (Liu and Qiu 1994; Qiu 2006; Yue et al. 2014). All of these studies suggest a similar climate to that inferred from our results.

Overall, vegetation was dominated by evergreen forests mainly composed of broadleaf trees during 4,350–1,000 cal bp, suggesting a warm climate at this time. The pollen percentage and PAR of trees and shrubs sharply decrease after 1,000 cal bp, indicating a cooling climate. Pollen records from the SZY core in Fujian also show an abrupt decrease in evergreen broadleaved tree pollen around 2,600 cal bp, indicating a cooling climate, driven by the weakening East Asian monsoon (Yue et al. 2012). We find that most of the late Holocene pollen records consistently show a gradual decline in evergreen forests, such as those dominated by evergreen Quercus, after 2,000 cal bp (Liu and Qiu 1994; Qiu 2006; Yue et al. 2014). It is therefore believed that the climate in southern China became cooler during the late Holocene, probably associated with a weakening summer monsoon. Oxygen isotopes measured in Dongge Cave exhibit negative values, indicating a weak Asian monsoon during the late Holocene (Wang et al. 2005). Similar patterns in oxygen isotope values have also been reported from Shanbao Cave further north (Shao et al. 2006; Wang et al. 2008). Together, these results support our interpretations of the varying climate conditions in our study area during the late Holocene. In addition to this, Xu et al. (2013) suggested that human activities have influenced the vegetation changes in the mountainous regions of Fujian Province of southeastern China during the last 1,300 years. In the section below, we further investigate human impacts on vegetation and the environment since 1,000 cal bp.

Climate events: the warm period from ad 670 to 960, the MWP and the LIA

In our study, we suggest that the chronology may not be sufficiently robust for discussion of climate events, as they have not been accurately dated. However, by comparing the lithologies and peat accumulation rates from our study site with those from the nearby Lianhuachi site (25°40′15″N, 118°11′12″E; 1,550 m a.s.l.) from Qiu (2006), we can be more certain of the reliability of the chronology constructed in this study. Thus, we can identify the timings of climate events with some degree of uncertainty through the reconstruction of late Holocene vegetation and climate change, and therefore consider that the timings of climate events described here are reliable.

Both LOI and humification are good proxies for past climate and environment conditions (Aaby 1976; Zhang et al. 1998; Borgmark and Schoning 2006; Daley and Barber 2012). High LOI and humification suggest a dry climate, while low LOI and humification indicate a wet climate (Aaby 1976; Zhang et al. 1998; Zhou et al. 2004; Borgmark 2005; Borgmark and Schoning 2006; Ma et al. 2009). In this study, we analyzed the peat humification, LOI, pollen accumulation rates and the ratio of evergreen groups to deciduous groups in our study area, and compared the results with the winter half-year mean temperature change over the last 2,000 years in eastern China (Ge et al. 2002), in addition to the δ18O record from Wanxiang Cave (Fig. 8; Zhang et al. 2008). While the differences between summer temperatures in the north and south of China are small, a significant difference can be seen in winter temperatures. As a result, temperature changes caused by varying solar radiation are more obvious during the winter under the background of climate change. Thus, we consider it reasonable to compare our data to the temperature changes in the winter half-years. Overall, the changes recorded in the six curves are largely consistent, since two warm periods and a cold period can be identified. For ad 670–960, the pollen accumulation rates peak, suggesting a warm period, which can be confirmed by the winter half-year mean temperature record in eastern China. Additional evidence in the form of negative δ18O at Wanxiang Cave and low LOI and absorbance further suggest that the climate was warm and wet from ad 670 to 960. This stage corresponds to the third warm period (ad 600–1000) in the climate evolution of eastern China over the past 5,000 years as suggested by Zhu (1973). However, the Daiyun Mountain peat records reveal that this warm period began later and ended earlier than the times established by Zhu (1973) and that the event was shorter and cooler than the medieval warm period (MWP). We suggest that the area was possibly cooled by the winter monsoon.

Fig. 8
figure 8

Overview chart illustrating a peat humification in the DYS2 profile, b the loss on ignition in the DYS2 profile, c PARs of the DYS2 profile, d the ratio of evergreen groups to deciduous groups, e temperature changes of winter half-years in eastern China (Ge et al. 2002), f δ18O record from Wanxiang Cave (Zhang et al. 2008)

Full size image

Many studies have attempted to reconstruct the climate changes during the MWP, however the beginning and end of this event remain controversial (Adhikari and Kumon 2001; Mann and Jones 2003; Yang et al. 2003; Driese et al. 2004; Pla and Catalan 2005; Matul et al. 2007; Yang et al. 2007). Nevertheless, the period from ad 800 to 1400 is generally considered to cover the MWP time span (Zuo and Jin 2009; Wang 2010). Thus, the climate fluctuations during the MWP can be recognized in our peat record. Based on our results, the MWP occurred around ad 1050–1520, with relatively a dry and warm climate compared with the warm period from ad 670 to 960. The ratio of evergreen to deciduous pollen was also higher during the MWP, suggesting a warmer climate, while the higher absorbance and LOI indicate dry conditions at this time. The evergreen to deciduous pollen ratio was high during ad 1060–1280 and 1430–1520, indicating that the climate was relatively warm during these two stages; these correspond to the two temperature peaks, with a 150 year long relatively cool period between. This pattern is considered broadly consistent with the main features of the climate fluctuations during the MWP.

From ad 1580 to 1850, both the evergreen to deciduous pollen ratio and pollen accumulation rates are low, corresponding to the low temperatures of the reconstructed winter half-year temperature record in eastern China. This suggests that the climate during this period, which corresponds to the LIA, was cold. In addition, low LOI and absorbance and negative δ18O values from Wanxiang Cave suggest a wet climate at this time. As such, the LIA may have corresponded to the cold periods of the Ming and Qing dynasties (Zhu 1973).

Human impact on vegetation and environment since 1,000 cal bp

Since 1,000 cal bp, the abundance of evergreen trees such as Castanopsis, Ilex and evergreen Quercus, and deciduous broad-leaf trees and shrubs such as Rosaceae and Ericaceae, decreased in our pollen data. Both human activities and climate change could have been the cause of such substantial changes in trees and shrubs. Human activities tend to result in the decline of selective woody plant taxa in small patches in the same region (Li et al. 2008). In contrast, a reduction of climate-sensitive trees and shrubs over large areas tends to be driven by climatic change (Li et al. 2008). In our study area, therefore, the declining numbers of evergreen trees and deciduous broadleaved trees and shrubs may have been caused by the use of these as building materials or fuel. In addition, our data suggest that secondary pine forests and grasses increased significantly at the same time, suggesting the beginning of farming in the region. According to Liu and Qiu (1994) and Qiu (2006), who have also reconstructed vegetation change in Fujian Province, the collapse of primeval vegetation communities occurred around 1,200 cal bp. This was represented by an abrupt decline in Castanopsis and evergreen Quercus pollen and a noticeable increase in Pinus and Poaceae pollen percentages. Additionally, the pollen record from Pingnan peat (Xu et al. 2013) also indicated a gradual decrease in Castanopsis and evergreen Quercus pollen, accompanied by a continuous increase in Poaceae pollen since approximately 800 cal bp; this pattern may be interpreted as reflecting the spread of rice cultivation. A large number of pollen records from lake sediments and soil profiles are also characterized by a reduction in arboreal pollen, such as Pinus, evergreen Quercus and Cyclobalanopsis, in parallel with a rise in herbaceous pollen, such as Artemisia, Poaceae and Chenopodiaceae, as a result of human influence during the late Holocene in east-central, southern, and northeastern China (Tarasov et al. 2006; Chen et al. 2009; Cao et al. 2010; Li et al. 2014).

A high concentration of charcoal usually indicates abundant fire activity (Chen 1990) and previous studies have suggested that the concentration of charcoal (>50 μm) is a convincing indicator of regional or local fire activity (Minckley et al. 2007; Sadori and Giardini 2007; Egli et al. 2012; Zhao et al. 2012). Combined with pollen data, charcoal can be used to reconstruct fire history, palaeovegetation and palaeoenvironmental changes (Patterson et al. 1987; Lv et al. 2002). In this context, charcoal can not only indicate the long-term fire regime and climate changes, but also the dynamic intensity of human activities (Jiang et al. 2008; Marquer et al. 2008). The pollen percentages of dry land herbs, trees and shrubs, and Poaceae were compared with charcoal (>50 μm) accumulation rates and δ18O from the Wanxiang cave (Fig. 9; Zhang et al. 2008). From this, we found that the charcoal accumulation rates can precisely indicate the frequency and intensity of fires, which are closely linked with human activities in our study area. Figure 8 shows positive δ18O, suggesting a weaker monsoon since 1,000 cal bp, corresponding to a low pollen percentage of trees and shrubs. Moreover, Poaceae pollen and the charcoal accumulation rates (>50 μm) increase and reach their peaks during this period, which suggests increased human activities since 1,000 cal bp.

Fig. 9
figure 9

Map showing changes in dry land herbs, trees and shrubs, Poaceae pollen, charcoal concentration (>50 μm), and the δ18O record from Dongge Cave (Wang et al. 2005) during the last 4,350 years (red shading highlights the influence of human activities)

Full size image

The population of Quanzhou City, which falls within our broad study area, increased rapidly during the Sui and Tang dynasties (Chen and Chen 2011). According to historical documents, many people moved to Quanzhou in the early Tang dynasty, resulting in a great expansion of farming in the surrounding regions (Chen and Chen 2011). The population doubled during the Song and Yuan dynasties (Chen and Chen 2011) and the increasing population had major impacts on the vegetation and the environment.

The natural vegetation in our study area is mostly coniferous and broadleaved mixed forest and broadleaved forest with some pine forest. It is known that the forest of Dehua County became an important source of fuel for the development of the ceramics industry from the Song dynasty onwards (Chen and Chen 2011). Increasing charcoal accumulation rate may therefore be linked to the clearance of trees for fuel. Liu and Qiu (1994) reported similar vegetation changes and speculated that human activities were the cause of vegetation degradation. Additionally, several environmental archaeology studies into the Fujian shell mound have suggested that people abandoned aquatic resources in the region after 2,000 cal bp, due to the expansion of paddy rice culture (Cai 1998; Qu and Chen 2010). Therefore, the sharp decrease of woody plants identified here may conceivably be linked to human activities.

Conclusions

In this study, a 4,350 year history of vegetation and climate changes was reconstructed based on pollen, LOI and humification records in peat sediments. We conclude:

  1. 1.

    Vegetation was dominated by evergreen forests, mainly composed of broadleaved trees during the period 4,350–1,000 cal bp, indicating a warm and wet climate. From 1,000 to 550 cal bp, a cool but dry climate can be inferred from the decreased pollen percentages and PARs of trees, shrubs and wetland herbs. The pollen percentage and PAR of dry land herbs, as well as LOI, increased between 1,000 and 550 cal bp. Finally, a higher pollen abundance of dry land and wetland herbs, along with low pollen percentage and PAR of trees and shrubs, low absorbance and LOI, suggest a cooler but wetter climate after 550 cal bp. In the future, further analysis of modern pollen assemblages is necessary, to permit a quantitative reconstruction of palaeoclimate based on the pollen database of East Asia.

  2. 2.

    Climate events of the warm period from ad 670 to 960, the MWP (ad 1050–1520) and the LIA (ad 1580–1850) can be identified from our records, with the climate of each period being characteristically warm and wet, warm and dry, and cold and wet, respectively. In this study, the chronology of these climate events may not be adequately robust, as there are as yet no accurate dates for them. However, we can identify the period in which the climate events took place from reconstructions of late Holocene vegetation and climate change.

  3. 3.

    Human activities may have had a very significant impact on the natural vegetation during the last 1 ka. Pollen and charcoal signals indicate strong human influence since 1,000 cal bp, which may be linked to the development of the porcelain industry in Dehua County and the rapidly increasing population within Quanzhou City in our study region.