Influence of plant coverage and environmental variables on pollen productivities: evidence from northern China

Abstract

Pollen productivity is a critical parameter in the interpretation of pollen-vegetation relationships, and in the quantitative reconstructions of past vegetation from fossil pollen records. One-year monitoring records were collected for 143 pollen traps in various parts of northern China, together with modern vegetation data. Absolute Pollen Productivity Estimates (APPE) were calculated for 11 taxa using the ratio of pollen influx to plant coverage at each applicable sampling site, in which the plants of the target taxon were present. Relative Pollen Productivity Estimates (RPPE) were calculated for the 11 taxa (taking Poaceae as the reference taxon) at those sites in which each taxon occurred together with Poaceae. Artemisia and Chenopodiaceae were found to have the highest RPPEs and the largest RPPEs ranges, while Pinus and Quercus also had higher RPPEs than Poaceae; Abies, Betula, Larix, Picea and Cyperaceae had relatively low RPPEs. Variations in RPPE between different areas may be explained by variations in climatic conditions, plant coverage and land use practices which might influence plant growing situation. Marked effect that variations in pollen productivity can have on vegetation reconstructions was demonstrated by applying these distinct RPPEs to reconstructions of Holocene vegetation in the Lake Daihai area (northern China), such as a large range of RPPE produces a large range of plant coverage. Variations in RPPEs within a single taxon, related to vegetation coverage and climatic conditions, therefore need to be considered in future vegetation reconstructions.

Keywords

pollen influx / pollen productivity / vegetation reconstruction / REVEALS model / northern China

Introduction

The quantitative reconstruction of vegetation compositions is a primary objective in Quaternary palaeoecology and palynology, and such information on local, regional and global scales is a prerequisite for vegetation model validation and environmental archeology. Pollen-based quantitative vegetation compositions can be modeled on the basis of taxa-specific pollen transport characteristics, lake size and Relative Pollen Productivity Estimates (RPPE; ^{Sugita, 2007a} and ^2007b). The reliability of these approaches has been demonstrated in North America (^{Sugita et al., 2010a}), Denmark (^{Nielsen and Odgaard, 2010}), Sweden (^{Hellman et al., 2008a} and ^2008b), the Swiss Plateau (^{Soepboer et al., 2010}), the Tibetan Plateau (^{Wang and Herzschuh, 2011}), northern Asia (^{Cao et al., 2019}) as well as temperate China (^{Li et al., 2020}). The results of these quantitative vegetation reconstructions rely heavily on RPPE for the plant taxa involved (^{Sugita, 2007a} and ^2007b; ^{Broström et al., 2008}; ^{Trondman et al., 2015}; ^{Cao et al., 2019}; ^{Li et al., 2020}). Pollen productivities have consequently been estimated for many taxa (^{Sugita et al., 1999}; ^{Broström et al., 2004}; ^{Bunting et al., 2005}; ^{Soepboer et al., 2007}; ^{Räsänen et al., 2007}; ^{Mazier et al., 2008}; ^{Sjögren et al., 2008}; ^{von Stedingk et al., 2008}; ^{Grindean et al., 2019}), but most of these investigations have been restricted to Europe. In eastern Asia, studies on pollen productivities have increased sharply in recent years, particularly in temperate regions of China (e.g., ^{Wang and Herzschuh, 2011}; ^{Xu et al., 2014}; ^{Li et al., 2015}; ^{Han et al., 2017}; ^{Li et al., 2017a}; ^{Li et al., 2017b}), and these RPPEs have already been incorporated quantitatively into land-cover reconstruction (^{Li et al., 2020}). However, the reliability of RPPE need to be further confirmed because, for example, one pollen taxon might have quite different RPPEs among different regions.

It is usually assumed that the RPPEs which the vegetation is based on have remained constant throughout the entire period covered by the vegetation reconstruction. However, the RPPEs derived for major plant types are only broadly consistent between different regions, and it has been suggested that pollen productivity for a particular plant taxon can vary regionally as a result of environmental factors such as climate, edaphic parameter and vegetation composition (^{Calcote, 1995}; ^{Hicks, 2001}; ^{Pardoe, 2001}; ^{Schofield et al., 2007}; ^{Hättestrand et al., 2008}; ^{Bunting et al., 2013}; ^{Chen et al., 2019}). For example, Cyperaceae and Empetrum plants in northern Finland are characterized by vegetative reproduction rather than prolific flowering because of the severe climatic conditions. Furthermore, the lower RPPE obtained for Fagus on the Swiss Plateau than in the Swiss Jura, southern Sweden and Denmark may be simply due to a reduced representation of this taxon in the vicinity of lakes (^{Broström et al., 2008}). However, differences in RPPEs may also arise as a result of different methodologies in the collection of pollen samples and/or vegetation data (^{Broström et al., 2008}; ^{Bunting and Hjelle, 2010}; ^{Ge et al., 2015}; ^{Niemeyer et al., 2015}). For example, the RPPE obtained for Filipendula based on lake sediment surface samples from Estonia are markedly higher than those based on moss polster samples from southern Sweden and Norway, and non-distance-weighted vegetation data from Norway resulted in lower herbaceous RPPEs than for southern Sweden (^{Broström et al., 2008}). Larix and Cyperaceae obtained higher representation in the lacustrine than in terrestrial moss samples (^{Niemeyer et al., 2015}). In addition, the poor taxonomic resolution of pollen identification renders it impossible to estimate the productivity of a single plant species. Hence, productivity estimates at higher taxonomic levels can be biased if pollen productivities vary within a single taxon (^{Watrin et al., 2007}).

Hitherto, pollen productivity estimation is mainly restricted to the ratio between productivity of the target taxon to that of the reference taxon, named by the RPPEs, and all corresponding models and software are prepared for the RPPEs and land-cover reconstruction based on RPPEs (e.g., ^{Sugita, 2007a} and ^2007b). Besides the RPPEs, the APPE is quite important in pollen-based vegetation reconstruction. Since absolute measures of pollen deposition in sediments are more effective than pollen percentages for distinguishing plant communities near the limits of trees as well as in the areas of sparsely occurring trees (^{Hicks, 2001}; ^{Sugita et al., 2010b}). On the Tibetan Plateau, pollen samples from the last glacial period harbours were with quite low pollen concentration but abundant arbor pollen grains, which is generally considered to be transported by wind from a long distance, while there is no direct evidence (e.g., ^{Sun et al., 1993}; ^{Zhu et al., 2015}); absolute pollen deposition could be a potential solution to separate the local or regional pollen grains. ^{Sugita et al. (2010a)} employed the slope of the linear relationship between pollen loading and distance-weighted plant abundance to represent APPE, which supports us a new approach to investigate the pollen-plant relationship.

The main objective of this study was to derive APPEs for six major tree types (Abies, Betula, Larix, Picea, Pinus, Quercus), one shrub type (Ostryopsis) and four types of herbs (Artemisia, Chenopodiaceae, Cyperaceae, Poaceae) in northern China, on the basis of pollen influx data from one-year pollen traps, in order to address the following questions: 1) What are the ranges and median values of the APPEs and RPPEs for these taxa in northern China? 2) How are APPEs related to plant coverage and climate parameters? 3) How do the different RPPEs based on Holocene pollen records from Lake Daihai in northern China affect vegetation reconstructions?

Study area

To investigate variations in pollen productivity with different environmental settings we collected modern pollen traps and vegetation data from a large area (Fig.1)(33.97°N–43.98°N, 104.54°E–117.58°E) covering steep environmental gradients, particularly in terms of mean annual temperature (ranging from −4.4°C to 9.2°C) and mean annual precipitation (ranging from 150 to 800 mm). This region encompasses major forest, steppe and desert vegetation zones. Forest trap sites were mainly located in mountainous areas which had a mean annual precipitation between 400 and 800 mm and were dominated by Larix, Pinus, Picea, Abies, Quercus and Betula plant communities. Steppe trap sites were located in areas with a mean annual precipitation ranging from 200 to 450 mm, where the dominant plant species included Stipa, Aneurolepidium chinense, Festuca ovina, Artemisia frigida, Cleistogenes squarrosa and Thymus mongolicus. The key plants in desert steppe areas, which have annual mean precipitation of 150–200 mm, include herbaceous plants such as Stipa gobica, Stipa breviflora, Stipa glareosa and Stipa klemenzii, and shrubby species such as Ajania, Hippolytia and Artemisia salsoloides. Cyperaceae are dominant in alpine meadow vegetation. Shrub communities are characterized by Ostryopsis, Caragana and Nitraria (^{Wu, 1980}; The Inner Mongolia and Ningxia Investigation Group of the Chinese Academy of Sciences, 1985; ^{Gao and Dai, 1988}). In the eastern part of the study area, the original vegetation had been substantially modified by human activities, and for this reason, the less impacted mountainous areas were chosen for sampling. The western and northern parts of the study area mainly belong to the steppe vegetation zone used only as pastures, and consequently have been less impacted by human activities than the eastern area. The pollen traps were nevertheless placed in sparsely populated areas to reduce human impact as much as possible.

Lake Daihai is situated on the eastern edge of the Ordos Plateau in Inner Mongolia (112°33′E–112°46′E, 40°29′N–40°37′N, 1221 m a.s.l.). It has a surface area of 133.5 km², a maximum water depth of about 14 m (in the year of 1999), a catchment area of 2289 km², and several rivers providing seasonal inflow. The lake is located in the transition zone between semi-humid and semi-arid areas in the middle temperate zone of China, and is surrounded by temperate steppe.

Materials and methods

Some 143 pollen traps were deployed between July 2004 and July 2005, including 56 traps in forested areas, 16 in shrublands, 28 in steppes, 21 in alpine meadows and 22 in desert areas, thus covering the major vegetation types in northern China. The traps consisted of a bucket 10 cm in diameter and 30 cm deep, with a lid that had a 5.2 cm diameter aperture surrounded by a sloping collar. The traps were sunk into the ground at each site so that the opening was 5–10 cm above the ground surface to minimize the risk of surrounding surface soil getting into the trap. The original pollen data were published by ^{Li et al. (2008)} and ^{Xu et al. (2009)} yet the representation of different pollen taxa has not been discussed then. The vegetation components in a 50 m radius surrounding forest traps and a 5 m radius surrounding steppe traps were recorded as percentages of plant coverage. The species compositions, the total coverage of each plant community and the coverage percentages of different plant species were recorded for each of these sites. Unfortunately, the vegetation surveying data are infeasible to calculate the plant abundance at different distances around the sampling sites, which is essential for the Extended R-Value (ERV) model; thus, we had to select another reliable approach to estimate pollen productivity.

Daily weather records from 1961 to 1990 measured at 243 meteorological observatories in northern China were obtained from the China Meteorological Statistical Annals (supplied by the China National Weather Service), and averages were calculated. The mean annual precipitation (P_ann), mean annual temperature (T_ann), mean temperature of the warmest month (Mt_wa) and mean temperature of the coldest month (Mt_co) for each trap site were calculated by parabolic interpolation using the Polation 1.0 program (Nakagawa, The Polation 1.0 program), with the inverse square distance as the weighting factor.

Pollen influx was reflected in the pollen sedimentation patterns and recorded as grains per cm² per annum. We calculated the APPEs using the ratio of pollen influx (grains cm⁻² a⁻¹) to plant coverage (%) for individual taxa (Abies, Betula, Larix, Picea, Pinus, Quercus, Ostryopsis, Artemisia, Chenopodiaceae, Cyperaceae and Poaceae) based on those applicable sampling sites in which the target taxa were present in the vegetation communities; then each individual taxon obtains series of APPE values based on individual sampling sites. The ranges of APPEs for each pollen taxon are presented in box-plots produced using SPSS 11.5 software (Fig. 2). To explore which climatic factors (P_ann, T_ann, Mt_co or Mt_wa) affect the pollen productivity of individual taxa, we fitted univariate tree models using the Brodgar software package (Highland Statistics Ltd., available at Brodgar website). The determination of tree size was guided by cross-validation; the smallest tree was selected using the one standard deviation rule (^{Breiman et al., 2015}). A LOESS smoother was then applied to highlight pollen productivity correlations with climate, using a default smoother span of 0.67 in the CanoDraw program. The thresholds obtained from univariate tree analysis are presented in the LOESS plots (Fig. 3).

We calculated the RPPEs for the 11 taxa by comparing the influxes per unit of plant coverage for the target taxa with those for the inference taxon. Poaceae was chosen as the reference taxon because it is common in the present-day vegetation of northern China as well as in both modern and fossil pollen records, and it has well defined pollen morphology, an intermediate pollen productivity, and has been used in previous studies (e.g., ^{Herzschuh et al., 2003}; ^{Broström et al., 2008}; ^{Mazier et al., 2008}; ^{Zhao and Herzschuh, 2009}; ^{Wang and Herzschuh, 2011}; ^{Li et al., 2017a}). It is, however, not an ideal taxon since it includes more than one species and different areas are likely to contain different species.

To demonstrate the impact of differences of pollen productivitiesy on vegetation modeling, we set up a series of 100 REVEALS (version 3.9) model runs using Sugita’s lake model (^{Sugita, 1993} and ^2007a) involving 100 randomly selected RPPE subsets and fossil pollen data from the DH99a Lake Daihai sediment core (^{Xiao et al., 2004}). The eight taxa (Betula, Pinus, Quercus, Ostryopsis, Artemisia, Chenopodiaceae, Cyperaceae and Poaceae) that were used in the model, make up between 72.1% and 98.1% (median: 93.4%) of the pollen assemblages in the fossil record. To establish the 100 RPPE data sets for model running, RPPE values for each taxon were selected randomly for 100 times respectively using the sample function in REVEALS. The fall speed used for the calculation draws on information from ^{Sugita et al. (1999)}, ^{Broström et al. (2004)} and ^{Tian (2010)}, and is shown in Table 1. Approximate error ranges (due to variations in the present-day RPPEs) could thus be obtained for each taxon and each sample. To be able to directly compare the original pollen percentages with the model-based reconstructed vegetation percentages, the pollen percentages for the eight taxa included in the pollen totals were recalculated.

Results

Calculation of APPEs and RPPEs

The PPE ranges and median values for the 11 taxa investigated are shown in Fig. 2 and Table 2, which also includes the main climatic factors affecting these values. Artemisia and Chenopodiaceae had the highest APPEs and the largest ranges. Artemisia recorded one APPE (grains·cm⁻²·a⁻¹, unit for APPE omitted below) value higher than 10000 and 21 values below 1020, with the other 13 values falling between 1400 and 5600; six of the sites for which RPPEs were available had values between 152 and 690, while all other sites had values less than 62. The maximum APPE for Chenopodiaceae was 9122, with three sites ranging between 1000 and 1300 and 10 sites ranging between 140 and 970; its two highest RPPEs were 830 and 517, while other values were all less than 50. The highest APPE for Poaceae was 1658, with 42 sites below 100 and 10 sites between 108 and 427. Pinus and Quercus had higher PPEs than Poaceae, and Pinus had the smallest range of PPEs out of all the 11 taxa. Abies, Betula, Larix, Picea and Cyperaceae all had relatively low PPEs. The fact that Ostryopsis has a low APPE median but a relatively high RPPE median value is a methodological artifact: there were eight traps applicable for APPEs, but only three of these could be used for Ostryopsis RPPEs as they were the only ones in which Ostryopsis and Poaceae occurred together in the surrounding vegetation. No traps were applicable for calculating RPPEs for the Abies, Pinus and Quercus taxa, but since they are common taxa in fossil pollen records as well as in the modern vegetation of mountain areas in northern China we used the ratios of their APPEs to the APPE median for Poaceae to calculate their individual RPPEs.

The relationship of APPEs to plant coverage and key climatic factors

Variation in plant coverage has a systematic effect on the APPEs for Larix, Picea, Ostryopsis, Artemisia, Chenopodiaceae, Cyperaceae and Poaceae. The plant coverages for Betula and Quercus have a positive correlation with their respective APPEs (Fig. 3).

Of all the environmental factors the APPEs for Betula, Larix are revealed by univariate tree modeling to be the most strongly related to P_ann, while the APPEs for Quercus and Cyperaceae are related to Mt_co, those for Ostryopsis and Poaceae are related to T_ann, and those for Artemisia and Chenopodiaceae are related to Mt_wa. The APPEs for Betula, Quercus, Artemisia and Cyperaceae have a apositive correlation with their key climatic factors, but no obvious trends have been found for Larix and Picea. The APPE for Poaceae remains consistent with increasing T_ann, while a reduction in the APPE for Chenopodiaceae is mainly due to the effect of a single outlier (Fig. 3).

Holocene vegetation reconstructions for the Lake Daihai area

The vegetation composition reconstructed on the basis of a series of 100 runs of the REVEALS model on randomly selected RPPE schemes is markedly different from that based on the original pollen percentages. For example, extremely high proportions of Artemisia (up to 77%) and Chenopodiaceae (up to 48%) were found in the pollen spectra throughout the entire Holocene, but the proportions appear much lower (less than 40% and 16%, respectively) in the reconstructed vegetation (Fig. 4). In contrast, the abundance of Cyperaceae (0%–2.5%, median: 0.3%) in pollen assemblages is extremely low compared to the abundance of Cyperaceae plants (between 0%–55%, median: 9%) in the reconstructed vegetation because of its low RPPE. Pinus, Quercus and Ostryopsis are present in similar proportions in both the pollen and vegetation data. As indicated by the large error ranges, particularly for Betula and Cyperaceae (Fig. 4), the individual reconstruction results vary greatly depending on the RPPEs involved.

Fossil pollen records indicate a steppe vegetation during the early Holocene (before 8 cal ka BP) dominated by herbs, with forest patches probably growing around the lake. However, according to the quantitative vegetation reconstruction, the regional vegetation around Lake Daihai was, at this stage, dominated by open birch forests rather than by steppe vegetation. Between 8 and 2.4 cal ka BP, the vegetation reconstruction indicates that the presence of pines and oaks showed a distinct and synchronous increase at the expense of birches, which is in accordance with the inferences based on pollen percentages; however, Cyperaceae is one of the main herbaceous taxa in the vegetation reconstruction while Artemisia and Chenopodiaceae are the dominant herbs in the original pollen percentage data. After 2.4 cal ka BP, pollen data indicates a scrub-steppe environment in which the forests almost disappeared, and that since that time the mountain slopes were composed mainly of Artemisia and Poaceae, with Chenopodiaceae growing in the valleys and depressions. However, the reconstruction indicates that birches began to recover during this period, although Artemisia, Chenopodiaceae, Cyperaceae and Poaceae reached their highest coverage while the Pinus and Quercus coverage began to decrease. The most significant difference between pollen data and the REVEALS-based vegetation reconstruction is that Betula and Cyperaceae were dominant taxa in the vegetation but contributed very little to the total pollen.

The sample-specific relative vegetation coverage reconstruction ranges for Betula and Cyperaceae are large, while those for Artemisia and Chenopodiaceae are much smaller. Poaceae only has a large range in reconstructed vegetation coverage during the late Holocene.

Discussion

PPE variations between different areas

The highest productivity estimates were for Artemisia and Chenopodiaceae (median APPEs of 963 and 514, and median RPPEs of 25.38 and 25.7, respectively), while lower estimates were obtained for Cyperaceae and Poaceae (median APPEs of 25 and 38, and median RPPEs of 0.13 and 1, respectively), which is consistent with the over-representation of Artemisia and Chenopodiaceae relative to Poaceae and Cyperaceae reported by ^{Li et al. (2011)}. However, in contrast to our results, ^{Li et al. (2011)} obtained markedly higher productivity for Artemisia than for Chenopodiaceae. In our study, Cyperaceae shows consistently low PPEs, which is consistent with studies from the Alashan Plateau (^{Herzschuh et al., 2003} and ^2006a), the Tibetan Plateau (^{Herzschuh et al., 2006b}; ^{Zhao and Herzschuh, 2009}; ^{Wang and Herzschuh, 2011}; ^{Wei et al., 2011}), and southern Sweden (^{Sugita et al., 1999}; ^{Broström et al., 2004}), but contrary to a study from Estonia (^{Poska et al., 2011}). In addition, Pinus (median APPE/RPPE: 145/3.78) and Quercus (median APPE/RPPE: 194/5.05) have generally higher PPEs than Poaceae, which is in agreement with results from Europe (^{Sugita et al., 1999}; ^{Räsänen et al., 2007}; ^{Soepboer et al., 2007}; ^{von Stedingk et al., 2008}; ^{Poska et al., 2011}).

Although the RPPEs for some taxa are more or less consistent between regions, some differences do exist. The largest differences are recognized in Abies and Picea, which have distinctly lower RPPEs in northern China than those obtained from several European areas. The RPPEs for Artemisia and Chenopodiaceae in northern China are also markedly higher than on the Tibetan Plateau (alpine meadow), where the cold climate might be not optimum for Artemisia and Chenopodiaceae; and Betula has a lower RPPE in northern China than in other regions (Table 3). The differences in RPPEs between our study and other studies could be caused by (at least to some extent) the different pollen source areas between the small (5.2 cm diameter) opening-area of traps and the surface-soil or lake-surface samples.

Some older studies assumed that Larix had a pollen productivity as high as other boreal conifers and concluded that its under-representation in pollen spectra was mainly due to the poor preservation and dispersal abilities of its pollen (^{Gunin et al., 1999}). However, our own results for Larix (median APPE/RPPE: 20/0.12) support the conclusions of, for example, ^{Gunin et al. (1999)} and ^{Ma et al. (2008)}, that Larix has a low pollen productivity as well as poor pollen dispersal. The PPE values for Larix obtained in our study should be reliable since all four of the Larix traps were located in pure Larix forest and since glycerin was added to each trap at the time of installation to preserve the pollen grains.

The dependence of PPEs on taxa coverage and climate

Critical assessment of the data and method

The correlations between APPEs and the environmental variables were found to be complex for most of the pollen taxa. The following methodological limitations might have enhanced the natural ambiguity.

1) Low number and limited distributions of sites.

Although 120 traps were included in this study, only very few traps were useable for some of the pollen taxa and their distribution was very limited. For example, only four traps were applicable for the Pinus APPEs, which were located on the Manhan and Li Mountains, but none of these traps could be used to calculate RPPEs for Pinus since Poaceae was not present in the vegetation at these sites. Due to the small number of available traps, the obtained PPE-range may have been underestimated.

2) Taxonomic resolution

In most cases, a single pollen taxon corresponds to several plant species or genera, with variable reproduction strategies, life forms, habitats and climatic requirements, which can have a marked effect on the PPEs and the associated vegetation reconstructions (^{Watrin et al., 2007}). Within our study area Artemisia and Chenopodiaceae have many more species than the arboreal taxa Abies, Betula, Larix and Picea; investigated (^{How, 1998}), which, together with the larger sampling area for Artemisia and Chenopodiaceae than for arboreal taxa, may explain their larger PPE ranges. Clearly taxonomic resolution presents a challenge to quantitative vegetation reconstruction and needs further investigation.

3) Pollen source area and vegetation data

The source area of pollen has important implications for the interpretation of pollen-vegetation relationships and palaeoecological records (^{Bunting et al., 2004}; ^{Niemeyer et al., 2015}). However, ^{Li et al. (2008)} concluded that the pollen source radius is less than 50–100 m for forest sites and other studies have also concluded that the plants in the immediate vicinity of a trap supply most of the pollen in the trap (^{Jackson and Kearsley, 1998}; ^{Sugita et al., 1999}; ^{Pérez et al., 2009}; ^{Xu et al., 2009}; ^{Sugita et al., 2010b}). Therefore, we surveyed vegetation components within a radius of 50 m for the forest traps and 5 m for steppe traps, aiming to account for the different dispersal mechanisms of arborous and herbaceous taxa. The background pollen was not included in the calculations but may have contributed a substantial proportion of the pollen load in the traps, and the proportion of this load may vary between the sites which is even more problematic.

4) Temporal variations in pollen productivity

The inter-annual variability in PPEs for the same taxon is well documented (^{Andersen, 1974}; ^{Saito et al., 1984}; ^{Hicks et al., 1994}; ^{Hicks, 2001}; ^{Sugita et al., 2010b}; ^{McLauchlan et al., 2011}) and hence monitoring records of pollen influxes over 5 to 10 years is required to obtain average pollen productivities (^{Hicks, 2001}). The observed spatial differences in productivity estimates may, to some extent, be due to the same factors that cause differences between the years.

Possible reasons for PPE variances

Differences in vegetation structure and vegetation cover density, both between and within species, have been found to influence pollen productivity (^{Hicks, 2001}; ^{Broström et al., 2008}; ^{von Stedingk et al., 2008}; ^{Sugita et al., 2010b}; ^{Baker et al., 2016}). In this study, the APPEs for Betula and Quercus clearly increase when the forest coverage exceeds 40%, which may be a result of optimal growth conditions. In contrast, higher APPEs for non-arboreal taxa (Ostryopsis, Artemisia, Chenopodiaceae and Cyperaceae) occur in sites with lower plant coverage while lower APPEs appear in sites with higher plant coverage. It is possible that these species expend less energy on sexual reproduction (but more on vegetative reproduction) at sites with high coverages (^{Cresswell et al., 2011}). The response of PPEs to plant coverage clearly differs between arboreal and non-arboreal species.

It has been suggested in previous studies that differences in vegetation structure, vegetation cover, reproduction strategy, age class, and growth behavior, both between and within species, might influence pollen productivity (^{Hicks, 2001}; ^{Broström et al., 2008}; ^{von Stedingk et al., 2008}; ^{Sugita et al., 2010b}; ^{Han et al., 2017}). The higher APPEs for Betula and Quercus in sites with higher coverages might have resulted from the optimal growing conditions, while the lower PPEs for Artemisia in forest zones (median APPE/RPPE: 299/4) than in non-forest zones (median APPE/RPPE: 962/34) may have been due to lower plant coverage and less suitable growing conditions in forest areas.

The impact of climate conditions on pollen productivity is complicated and is likely to vary from different pollen taxa (^{Autio and Hicks, 2004}; ^{Broström et al., 2008}; ^{van der Knaap et al., 2010}). And vegetation evolution is sensitive to temperature (^{Wu et al., 2016}). In our study, the PPEs for Larix are related to the mean annual precipitation, while those for Quercus and Cyperaceae may be affected by the mean temperature of the coldest month. The productivity of Ostryopsis and Poaceae may be related to the mean annual temperature and those of Artemisia and Chenopodiaceae to the mean temperature of the warmest month. The major climatic factors affecting the PPEs for an individual taxon also vary from one region to another. For example, in northern China, the PPEs for Betula and Picea have been shown to be related to the mean annual precipitation, while in the Swiss Jura Mountains the PPE for Betula responds positively to both winter and spring temperature (^{Sjögren et al., 2006}), and in northern Finland and the Caucasus the PPE for Picea is affected by the summer temperature of the previous year (^{Autio and Hicks, 2004}; ^{van der Knaap et al., 2010}). A possible explanation for these differences could be that moisture is the climatic factor controlling vegetation distribution in northern China, particularly in the relative dry steppe and desert areas, while temperature is the controlling factor in western and northern Europe. ^{Sjögren et al. (2008)} suggested that the pollen productivity of trees growing close to the tree line in mountain areas might be climatically limited. Our RPPE for Betula (range: 0.13–2.14, median: 0.29) is far lower than those for European areas (Table 3), possibly because our traps were located in the transition zone between forest and alpine meadow, close to the upper tree limit.

Although we tried to avoid human impact as much as possible when positioning the traps, the selected areas would have still been impacted by grazing. The pollen productivity of herb species in pasture-lands are likely to be lower than in non-pasture areas as a result of grazing disturbing the flowers and thus reducing the pollen productivities (^{Groenman-van Waateringe, 1993}; ^{Dahlström, 2008}). Since Poaceae is one of the favorite forages for livestock species, the lower APPE for Poaceae in non-forested areas (median: 31) than in forested areas (median: 112) may be related to human impacts, which would unfortunately also affect the RPPEs for other taxa since Poaceae was used as the reference taxon in our study.

Implications of differences in RPPEs for vegetation reconstruction

The basic assumption of the REVEALS model (^{Sugita, 2007a}) is that wind is the dominant agent of pollen transport. Lake Daihai has several inflowing rivers that originate from the surrounding mountains; and previous studies on sediment-surface pollen distribution for Lake Daihai revealed that the pollen assemblages represent regional vegetation surrounding Lake Daihai well and the influence of background pollen transported by wind is quite weak (^{Xu et al., 2005}; ^{Tian et al., 2008}). As taphonomy did not change substantially, we assume that the vegetation reconstruction should represent the regional vegetation coverage well. In addition, the drainage catchment area may be overrepresented in the Daihai pollen record compared to areas that only receive contributions from air-borne pollen (^{Xu et al., 2012}). This may explain the high contributions of arboreal taxa to the early and middle Holocene fossil pollen spectra despite the fact that the lake was surrounded by steppe vegetation.

In the present-day vegetation, Betula and Pinus are the main tree taxa in forest communities on the northern slopes of the mountains in the Lake Daihai area, accompanied by Ostryopsis in the forest under-storey. However, the vegetation reconstructions indicated by REVEALS modeling suggest that birches were the main component in forests throughout the entire Holocene. Previous studies found that both Betula and Pinus are over-represented in the pollen taxa (^{Xu et al., 2007}). In our reconstruction the proportion of Pinus in the vegetation is slightly lower than it is in the pollen, which agrees with the previous conclusion that it is over-represented (^{Xu et al., 2007}). The contribution of Betula in the vegetation is markedly higher than in the pollen, which disagrees with the previously interpreted over-representation, hence we that argue the birch coverage could be over-estimated, because Betula RPPE in this study is lower than previous RPPE studies in north China.

Artemisia is widely distributed within the Lake Daihai area, occurring in both steppes and alpine meadows, while the Chenopodiaceae species are only common on hilly steppes (^{Xiao et al., 2004}). ^{Liu et al. (1999)} found that Artemisia and Chenopodiaceae are dominant components in modern pollen samples from temperate-steppe areas in northern China, even in sites with low plant coverage. In our reconstructions their vegetation coverage percentages are substantially lower than their pollen percentages due to their high pollen productivities. In contrast, Cyperaceae and Poaceae plants are major components of halophilic meadows and patches of boggy meadow that are predominant along the lake shores and on the leading edges of diluvial fans (^{Wang et al., 1990}; ^{Xiao et al., 2004}). Their high proportions in the reconstructed vegetation are consistent with those in present-day vegetation.

Large ranges in reconstructed vegetation estimates were found for the individual samples, for almost all species, indicating that the vegetation reconstructions are strongly affected by the large ranges of the pollen productivity estimates. Taxa with large RPPE ranges such as Artemisia, Chenopodiaceae, Cyperaceae and Quercus are reconstructed to behave in large ranges in plant coverage; while taxa with narrow RPPE ranges such as Pinus are estimated to have small coverage-ranges (Table 2; Fig. 4). Our experiment reveals the importance of RPPE in the vegetation reconstruction. We found that the pollen productivity of individual species varied with different climatic conditions. If only one PPE-set for the investigated species is applied throughout the entire period covered by the vegetation reconstruction (as is commonly the case), the effect of climatic variations on the vegetation reconstruction would be obscured, especially for a series of strong climatic fluctuations. Further investigations into pollen productivities under various climatic conditions are required to reveal the relationships between pollen productivity and environmental factors; the application of different PPE-sets for different climatic periods might then be expected to produce more accurate reconstructions of past vegetation.

Conclusions

1) Pollen influxes per unit plant coverage can be used as an index to scale pollen productivity for different taxa. In this study, Artemisia and Chenopodiaceae had the highest pollen productivity estimates and the largest ranges, while Pinus and Quercus were moderate pollen producers and Abies, Betula, Larix, Picea, Ostryopsis, Cyperaceae and Poaceae were all low pollen producers.

2) We found that environmental factors, climatic conditions and plant coverage could all produce variations in the PPEs for individual taxa.

3) Using REVEALS modeling to reconstruct Holocene vegetation in the Lake Daihai area of northern China, we found that the large range of taxa-specific PPEs produced a large variability in the vegetation reconstructions, which could even prevent the recognition of major changes in vegetation. Further investigations into the relationship between pollen productivity and climatic conditions are required to improve the reconstructions of Holocene vegetation changes.

References

[1] Andersen S T (1974). Wind conditions and pollen deposition in a mixed deciduous forest II Seasonal and annual pollen deposition 1967–1972. Grana, 14(2–3): 64–77

[2] Autio J, Hicks S (2004). Annual variations in pollen deposition and meteorological conditions on the fell Aakenustunturi in northern Finland: potential for using fossil pollen as a climate proxy. Grana, 43(1): 31–47

[3] Baker A G, Zimny M, Keczyński A, Bhagwat S A, Willis K J, Latalowa M (2016). Pollen productivity estimates from old-growth forest strongly differ from those obtained in cultural landscapes: evidence from the Białowieża National Park Poland. Holocene, 26(1): 80–92

[4] Breiman L, Friedman J H, Olshen R A, Stone C G (2015). Classification and regression trees. Belmont, California: Wadsworth International Group. Ency Ecology, 57(1): 582–588

[5] Broström A, Sugita S, Gaillard M J (2004). Pollen productivity estimates for the reconstruction of past vegetation cover in the cultural landscape of southern Sweden. Holocene, 14(3): 368–381

[6] Broström A, Nielsen A B, Gaillard M J, Hjelle K, Mazier F, Binney H, Bunting J, Fyfe R, Meltsov V, Poska A, Räsänen S, Soepboer W, von Stedingk H, Suutari H, Sugita S (2008). Pollen productivity estimates of key European plant taxa for quantitative reconstruction of past vegetation: a review. Veg Hist Archaeobot, 17(5): 461–478

[7] Bunting M J, Gaillard M J, Sugita S, Middleton R, Broström A (2004). Vegetation structure and pollen source area. Holocene, 14(5): 651–660

[8] Bunting M J, Armitage R, Binney H A, Waller M (2005). Estimates of ‘relative pollen productivity’ and ‘relevant source area of pollen’ for major tree taxa in two Norfolk (UK) woodlands. Holocene, 15(3): 459–465

[9] Bunting M J, Hjelle K L (2010). Effect of vegetation data collection strategies on estimates of relevant source area of pollen (RSAP) and relative pollen productivity estimates (relative PPE) for non-arboreal taxa. Veg Hist Archaeobot, 19(4): 365–374

[10] Bunting M J, Schofield J E, Edwards K J (2013). Estimates of relative pollen productivity (RPP) for selected taxa from southern Greenland: a pragmatic solution. Rev Palaeobot Palynol, 190: 66–74

[11] Calcote R (1995). Pollen source area and pollen productivity: evidence from forest hollows. J Ecol, 83(4): 591–602

[12] Cao X Y, Tian F, Li F, Gaillard M J, Rudaya N, Herzschuh U (2019). Pollen-based quantitative land-cover reconstruction for northern Asia during the last 40 ka cal BP. Clim Past, 15: 1503–1536

[13] Chen H, Xu Q H, Zhang S R, Sun Y H, Wang M, Zhou Z Z (2019). Relative pollen productivity estimates of subtropical evergreen and deciduous broadleaved mixed forest in Ta-pieh Mountains. Quat Sci, 39(2): 469–482

[14] Cresswell J E, Hagen C, Woolnough J M (2011). Attributes of individual flowers of Brassica napus L are affected by defoliation but not by intraspecific competition. Ann Bot, 88: 111–117

[15] Dahlström A (2008). Grazing dynamics at different spatial and temporal scales: examples from the Swedish historical record A. D. 1620–1850. Veg Hist Archaeobot, 17(5): 563–572

[16] Gao Z Z, Dai F H (1988). Vegetation of Ningxia. Yinchuan: Ningxia Renmin Press, 52–61 (in Chinese)

[17] Ge Y W, Li Y C, Li Y, Yang X L, Zhang R C, Xu Q H (2015). Relevant source area of pollen and relative pollen productivity estimates in Bashang steppe. Quat Sci, 35(4): 934–945

[18] Grindean R, Nielsen A B, Tanţău I, Feurdean A (2019). Relative pollen productivity estimates in the forest steppe landscape of southeastern Romania. Rev Palaeobot Palynol, 264: 54–63

[19] Groenman-van Waateringe W (1993). The effects of grazing on the pollen production of grasses. Veg Hist Archaeobot, 2(3): 157–162

[20] Gunin P D (1999). Vegetation dynamics of Mongolia. Geobotany, 26

[21] Han Y, Liu H Y, Hao Q, Liu X, Guo W C, Shangguan H (2017). More reliable pollen productivity estimates and relative source area of pollen in a forest-steppe ecotone with improved vegetation survey. Holocene, 27(10): 1567–1577

[22] Hättestrand M, Jensen C, Hallsdóttir M, Vorren K D (2008). Modern pollen accumulation rates at the north-western fringe of the European boreal forest. Rev Palaeobot Palynol, 151(3–4): 90–109

[23] Hellman S, Gaillard M J, Broström A, Sugita S (2008a). The REVEALS model a new tool to estimate past regional plant abundance from pollen data in large lakes: validation in southern Sweden. J Quat Sci, 23(1): 21–42

[24] Hellman S, Gaillard M J, Broström A, Sugita S (2008b). Effects of the sampling design and selection of parameter values on pollen-based quantitative reconstructions of regional vegetation: a case study in southern Sweden using the REVEALS model. Veg Hist Archaeobot, 17(5): 445–459

[25] Herzschuh U, Kürschner H, Ma Y Z (2003). The surface pollen and relative pollen production of the desert vegetation of the Alashan Plateau western Inner Mongolia. Chin Sci Bull, 48(14): 1488–1493

[26] Herzschuh U, Kürschner H, Battarbee R, Holmes J (2006a). Desert plant pollen production and a 160-year record of vegetation and climate change on the Alashan Plateau NW China. Veg Hist Archaeobot, 15(3): 181–190

[27] Herzschuh U, Winter K, Wünnemann B, Li S J (2006b). A general cooling trend on the central Tibetan Plateau throughout the Holocene recorded by the Lake Zigetang pollen spectra. Quat Int, 154–155: 113–121

[28] Hicks S, Helander M, Heino S (1994). Birch pollen production transport and deposition for the period 1984–1993 at Kevo northernmost Finland. Aerobiologia, 10(2-3): 183–191

[29] Hicks S (2001). The use of annual arboreal pollen deposition values for delimiting tree-lines in the landscape and exploring models of pollen dispersal. Rev Palaeobot Palynol, 117(1–3): 1–29

[30] How F C (1998). A dictionary of the families and genera of Chinese seed plants. 2nd ed. Beijing: Science Press (in Chinese)

[31] Inner Mongolia and Ningxia Research Group of Chinese Academy of Sciences (1985). Vegetation of Inner Mongolia Beijing: Science Press, 427–641 (in Chinese)

[32] Jackson S T, Kearsley J B (1998). Quantitative representation of local forest composition in forest-floor pollen assemblages. J Ecol, 86(3): 474–490

[33] Li F R, Gaillard M J, Sugita S, Mazier F, Xu Q H, Zhou Z Z, Zhang Y Y, Li Y C, Laffly D (2017a). Relative pollen productivity estimates for major plant taxa of cultural landscapes in central eastern China. Veg Hist Archaeobot, 26(6): 587–605

[34] Li F R, Gaillard M J, Cao X Y, Herzschuh U, Sugita S, Tarasov P E, Wagner M K, Xu Q H, Ni J, Wang W M, Zhao Y, An C B, Beusen A H W, Chen F H, Feng Z D, Klein Goldewijk C G M, Huang X Z, Li Y C, Jia X (2020). Towards quantification of Holocene anthropogenic land-cover change in temperate China: A review in the light of pollen-based REVEALS reconstructions of regional plant cover. Earth Sci Rev, 203: 103–119

[35] Li Y C, Xu Q H, Wang X L, Cao X Y (2008). Modern pollen assemblages of the forest communities and their relationships with vegetation and climate in northern China. Acta Geogr Sin, 63(9): 945–957 (in Chinese)

[36] Li Y C, Bunting M J, Xu Q H, Jiang S X, Ding W, Hun L Y (2011). Pollen-vegetation-climate relationships in some desert and desert-steppe communities in northern China. Holocene, 21(6): 997–1010

[37] Li Y, Ge Y, Bunting M J, Zhang Z, Li J, Wang C, Li B, Li C (2017b). Relative pollen productivities and relevant source area of pollen in the forest-steppe ecotone of northern China. Rev Palaeobot Palynol, 224: 1–12

[38] Li Y, Nielsen A B, Zhao X, Shan L, Wang S, Wu J, Zhou L (2015). Pollen production estimates (PPEs) and fall speeds for major tree taxa and relevant source areas of pollen (RSAP) in Changbai Mountain northeastern China. Rev Palaeobot Palynol, 216: 92–100

[39] Liu H Y, Cui H T, Pott R, Speier M (1999). The surface pollen of the woodland-steppe ecotone in southeastern Inner Mongolia, China. Rev Palaeobot Palynol, 105(3-4): 237–250

[40] Ma Y Z, Liu K B, Feng Z D, Sang Y L, Wang W, Sun A Z (2008). A survey of modern pollen and vegetation along a south-north transect in Mongolia. J Biogeogr, 35(8): 1512–1532

[41] Mazier F, Brostöm A, Gaillard M J, Sugita S, Vittoz P, Buttler A (2008). Pollen productivity estimates and relevant source area of pollen for selected plant taxa in a pasture woodland landscape of the Jura Mountains (Switzerland). Veg Hist Archaeobot, 17: 479–495

[42] McLauchlan K K, Barnes C S, Craine J M (2011). Interannual variability of pollen productivity and transport in mid-north America from 1997 to 2009. Aerobiologia, 27(3): 181–189

[43] Nielsen A B, Odgaard B V (2010). Quantitative landscape dynamics in Denmark through the last three millennia based on the Landscape Reconstruction Algorithm approach. Veg Hist Archaeobot, 19(4): 375–387

[44] Niemeyer B, Klemm J, Pestryakova L A, Herzschuh U (2015). Relative pollen productivity estimates for common taxa of the northern Siberian Arctic. Rev Palaeobot Palynol, 221: 71–82

[45] Pardoe H S (2001). The representation of taxa in surface pollen spectra on alpine and sub-alpine glacier forelands in southern Norway. Rev Palaeobot Palynol, 117(1–3): 63–78

[46] Pérez C F, Latorre F, Stutz S, Pastorino S (2009). A two-year report of pollen influx into Tauber traps in MarChiquita coastal lagoon Buenos Aires Argentina. Aerobiologia, 25(3): 167–181

[47] Poska A, Meltsov V, Sugita S, Vassiljev J (2011). Relative pollen productivity estimates of major anemophilous taxa and relevant source area of pollen in a cultural landscape of the hemi-boreal forest zone (Estonia). Rev Palaeobot Palynol, 167(1–2): 30–39

[48] Räsänen S, Suutari H, Nielsen A B (2007). A step further towards quantitative reconstruction of past vegetation in Fennoscandian boreal forests: pollen productivity estimates for six dominant taxa. Rev Palaeobot Palynol, 146(1–4): 208–220

[49] Saito H, Mishima Y, Nogawa S, Takeoka M (1984). Pollen production rates in 75-year-old Japanese pine forest. Scientific Report of Kyoto Prefectural University. Agriculture, 36: 9–18

[50] Schofield J E, Edwards K J, McMullen J A (2007). Modern pollen-vegetation relationships in subarctic southern Greenland and the interpretation of fossil pollen data from the Norse landnam. J Biogeogr, 34(3): 473–488

[51] Sjögren P, van Leeuwen J F N, van der Knaap W O, van der Borg K (2006). The effect of climate variability on pollen productivity AD 1975–2000 recorded in a Sphagnum peat hummock. Holocene, 16(2): 277–286

[52] Sjögren P, van der Knaap W O, Huusko A, van Leeuwen J F N (2008). Pollen productivity dispersal and correction factors for major tree taxa in the Swiss Alps based on pollen-trap results. Rev Palaeobot Palynol, 152(3–4): 200–210

[53] Soepboer W, Sugita S, Lotter A F, van Leeuwen J F N, van der Knaap W O (2007). Pollen productivity estimates for quantitative reconstruction of vegetation cover on the Swiss Plateau. Holocene, 17(1): 65–77

[54] Soepboer W, Sugita S, Lotter A F (2010). Regional vegetation-cover changes on the Swiss Plateau during the past two millennia: a pollen-based reconstruction using the REVEALS model. Quat Sci Rev, 29(3–4): 472–483

[55] Sugita S (1993). A model of pollen source area for an entire lake. Quat Res, 39(2): 239–244

[56] Sugita S, Gaillard M J, Broström A (1999). Landscape openness and pollen records: a simulation approach. Holocene, 9(4): 409–421

[57] Sugita S (2007a). Theory of quantitative reconstruction of vegetation I: pollen from large sites REVEALS regional vegetation. Holocene, 17(2): 229–241

[58] Sugita S (2007b). Theory of quantitative reconstruction of vegetation II: all you need is LOVE. Holocene, 17(2): 243–257

[59] Sugita S, Parshall T, Calcote R, Walker K (2010a). Testing the landscape reconstruction algorithm for spatially explicit reconstruction of vegetation in northern Michigan and Wisconsin. Quat Res, 74(2): 289–300

[60] Sugita S, Hicks S, Sormunen H (2010b). Absolute pollen productivity and pollen-vegetation relationships in northern Finland. J Quat Sci, 25(5): 724–736

[61] Sun X J, Du N Q, Chen Y S, Gu Z Y, Liu J Q, Yuan B Y (1993). Holocene palynological records in Lake Selincuo, Northern Xizang. Journal of Integrative Plant Biology, 35: 943–950 (in Chinese)

[62] Tian F, Xu Q H, Li Y C, Cao X Y, Wang X L, Zhang L Y (2008). Pollen assemblage characteristics of lakes in the monsoon fringe area of China. Chin Sci Bull, 53(21): 3354–3363

[63] Tian F (2010). Pollen sedimentary processes of different lakes in Northern China. Dissertation for Master’s Degree. Shijiazhuang: Hebei Normal University China (in Chinese)

[64] Trondman A K, Gaillard M J, Mazier F, Sugita S, Fyfe R, Nielsen A B, Twiddle C, Barratt P, Birks H J B, Bjune A E, Björkman L, Broström A, Caseldine C, David R, Dodson J, Dörfler W, Fischer E, van Geel B, Giesecke T, Hultberg T, Kalnina L, Kangur M, van der Knaap P, Koff T, Kuneš P, Lagerås P, Latałowa M, Lechterbeck J, Leroyer C, Leydet M, Lindbladh M, Marquer L, Mitchell F J G, Odgaard B V, Peglar S M, Persson T, Poska A, Rösch M, Seppä H, Veski S, Wick L (2015). Pollen-based quantitative reconstructions of Holocene regional vegetation cover (plant-functional types and land-cover types) in Europe suitable for climate modelling. Glob Chang Biol, 21(2): 676–697

[65] van der Knaap W O, van Leeuwen J F N, Svitavská-Svobodová H, Pidek I A, Kvavadze E, Chichinadze M, Giesecke T, Kaszewski B W, Oberli F, Kalnina L, Pardoe H S, Tinner W, Ammann B (2010). Annual pollen traps reveal the complexity of climatic control on pollen productivity in Europe and the Caucasus. Veg Hist Archaeobot, 19(4): 285–307

[66] von Stedingk H, Fyfe R, Allard A (2008). Pollen productivity estimates from the forest—tundra ecotone in west-central Sweden: implications for vegetation reconstruction at the limits of the boreal forest. Holocene, 18: 323–332

[67] Wang Y B, Herzschuh U (2011). Reassessment of Holocene vegetation change on the upper Tibetan Plateau using the pollen-based REVEALS model. Rev Palaeobot Palynol, 168(1): 31–40

[68] Wang S M, Yu Y S, Wu R J, Feng M (1990). The Daihai Lake: Environment Evolution and Climate Change. Hefei: University of Science and Technology of China Press (in Chinese)

[69] Watrin J, Lézine A M, Gajewski K, Vincens A (2007). Pollen-plant-climate relationships in sub-Saharan Africa. J Biogeogr, 34(3): 489–499

[70] Wei H C, Ma H Z, Zheng Z, Pan A D, Huang K Y (2011). Modern pollen assemblages of surface samples and their relationships to vegetation and climate in the northeastern Qinghai-Tibetan Plateau China. Rev Palaeobot Palynol, 163(3-4): 237–246

[71] Wu J, Liu Q, Chu G Q, Wang L, Liu J Q (2016). Vegetation history and climate change recorded by stomata evidence during the late glacial in the Great Khingan Mountain Region Northeastern China. Chin Sci Bull, 61(36): 3940–3945 (in Chinese)

[72] Wu Z (1980). Vegetation of China. Beijing: Science Press (in Chinese)

[73] Xiao J L, Xu Q H, Nakamura T, Yang X L, Liang W D, Inouchi Y (2004). Holocene vegetation variation in the Daihai Lake region of north-central China: a direct indication of the Asian monsoon climatic history. Quat Sci Rev, 23(14–15): 1669–1679

[74] Xu Q H, Li Y C, Yang X L, Xiao J L, Liang W D, Peng Y J (2005). Source and distribution of pollen in the surface sediment of Daihai Lake inner Mongolia. Quat Int, 136(1): 33–45

[75] Xu Q H, Li Y C, Zhou L P, Li Y Y, Zhang Z Q, Lin F Y (2007). Pollen flux and vertical dispersion in coniferous and deciduous broadleaved mixed forest in the Changbai Mountains. Chin Sci Bull, 52(11): 1540–1544

[76] Xu Q H, Li Y C, Tian F, Cao X Y, Yang X L (2009). Pollen assemblages of tauber traps and surface soil samples in steppe areas of China and their relationships with vegetation and climate. Rev Palaeobot Palynol, 153(1–2): 86–101

[77] Xu Q H, Tian F, Bunting M J, Li Y C, Ding W, Cao X Y, He Z (2012). Pollen source areas of lakes with inflowing rivers: modern pollen influx data from Lake Baiyangdian China. Quat Sci Rev, 37: 81–91

[78] Xu Q H, Cao X Y, Tian F, Zhang S R, Li Y C, Li M Y, Li J, Liu Y, Liang J (2014). Relative pollen productivities of typical steppe species in northern China and their potential in past vegetation reconstruction. Sci China Earth Sci, 57(6): 1254–1266

[79] Zhao Y, Herzschuh U (2009). Modern pollen representation of source vegetation in the Qaidam Basin and surrounding mountains north-eastern Tibetan Plateau. Veg Hist Archaeobot, 18(3): 245–260

[80] Zhu L, Lü X, Wang J, Peng P, Kasper T, Daut G, Haberzettl T, Frenzel P, Li Q, Yang R, Schwalb A, Mäusbacher R (2015). Climate change on the Tibetan Plateau in response to shifting atmospheric circulation since the LGM. Sci Rep, 5(1): 3940–3945

RIGHTS & PERMISSIONS

Higher Education Press