Progress,problems and prospects of palynology in reconstructing environmental change in inland arid areas of Asia

YongTao Zhao,YunFa Miao,2*,Yan Lei,2,XianYong Cao,MingXing Xiang,2

1. Key Laboratory of Desert and Desertification, Northwest Institute of Eco-Environment and Resources, Chinese Academy of Sciences,Lanzhou,Gansu 730000,China

2.University of Chinese Academy of Sciences,Beijing 100049,China

3. Key Laboratory of Alpine Ecology, Institute of Tibetan Plateau Research, Chinese Academy of Sciences, Beijing 100101,China

ABSTRACT Studying the climatic and environmental changes on different time scales in inland arid regions of Asia can greatly im‐prove our understanding of climatic influences for the Qinghai-Tibet Plateau in the context of global change. Pollen, as a remnant of seed plants,is sensitive to environmental factors including precipitation,temperature and altitude,and is a clas‐sic proxy in environmental reconstruction. In the last two decades, great progress in the application of palynology to in‐land areas of Asia has highlighted the role of palynology in paleoclimatic and paleoenvironmental research. The main progress is as follows. (1) On the tectonic time scale of the late Cenozoic, the palaeoclimatological sequence has been es‐tablished on the basis of pollen percentage, concentration and taxon. Pollen data have revealed a continuous enhancement of drought in the inland arid region of Asia,in contrast to evidence acquired based on other proxies.(2)In the late Quater‐nary, an increase in herbaceous plants further supports the intensification of drought associated with global cooling. In more detail,the palynological record shows a glacial-interglacial pattern consistent with changes in global ice volume.(3)The Holocene pollen record has been established at a high resolution and across a wide range of inland areas.In general,it presents an arid grassland environment in the early Holocene,followed by the development of woody plants in the mid-to late-Holocene climate optimum.This pattern is related to moisture changes in areas dominated by the westerlies.There are also significant regional differences in the pattern and amplitude of vegetation response to the Holocene environment. (4)Modern pollen studies based on vegetation surveys, meteorological data and statistics show that topsoil palynology can better reflect regional vegetation types (e.g., grassland, meadow, desert). Drier climates yield higher pollen contents of drought-tolerant plants such as Chenopodioideae, Ephedra, and Nitriaria, while contents of Artemisia and Poaceae are greater under humid climates. Besides these achievements, problems remain in palynological research: for example, pol‐len extraction,identification,interpretation,and quantitative reconstruction.In the future,we encourage strengthened inter‐disciplinary cooperation to improve experimental methods and innovation. Firstly, we should strengthen palynological classification and improve the skill of identification; secondly, laboratory experiments are needed to better constrain pol‐len transport dynamics in water and air; thirdly, more rigorous mathematical principles will improve the reliability of re‐constructions and deepen the knowledge of plant geography; and finally, new areas and methods in palynology should be explored, for example DNA, UV-B and isotopic analysis. It is expected that palynology will continue to develop, and we hope it will continue to play an important role in the study of past climatic and environmental changes.

Keywords:Palynology;inland arid areas;Late Cenozoic;Quaternary;Holocene;modern environmental processes

1 Introduction

The inland arid areas of Asia are located far from the ocean and are affected by high terrain,thus forming the world's most extensive mid-latitude arid zone. The area as a whole is characterized by low precipitation,sparse vegetation, strong wind and frequent dust storms. At the same time, glaciers have developed in the foothills,and runoff from snowmelt forms relatively flat alluvial fans at the foot of the mountains.These pre‐cious freshwater resources support the oasis landscape in arid areas.The region has played an important role in the spread of modern human populations from Africa during the Quaternary, for the cultural exchange be‐tween East and West during the Holocene, and for the sustainable development of modern society. Under the current global warming background, sustainable devel‐opment and ecological changes in this area reflect the process of human adaptation to rapid climate changes.The rate of change, spatial extent and magnitude of droughts have received widespread attention in this re‐gion; furthermore, a large amount of surface material from the region is carried by the westerly circulation and/or near-surface winds to the ocean (Uematsuet al.,1983; Reaet al., 1985; Zhanget al., 2003; Zheng,2016), which influences the surface radiation balance and energy exchanges with the atmosphere, biosphere,and lithosphere. Such processes have profound impacts on regional and global climate (Arimoto, 2001; Harri‐sonet al., 2001; Jickellset al., 2005). Studies have shown that dust storms occurred in the Eocene (Liet al.,2018),gradually increased in the late Cenozoic(Reaet al., 1985; Anet al., 2001; Yanget al., 2016), and reached a peak in the Quaternary (Dinget al., 1999; Luet al., 2019). Therefore, it is of great significance to study aridification processes at different time scales in inland arid areas of Asia,to better understand global cli‐mate change, human evolution, and modern human re‐sponse to a drought-prone environment. Some specific questions in this respect are as follows.How is aridifica‐tion characterized on tectonic and orbital time scales?How can drought characteristics provide information on human activities at Holocene and modern time scales?

Palynology is a science that studies the morpholo‐gy and classification of spores and pollen in plants and can be applied to many fields (Traverse, 1988;Mooreet al., 1991). Briefly, palynology can be divid‐ed into two aspects: namely, modern palynology and paleopalynology. The focus of paleopalynology is on reconstructing paleovegetation and paleoenvironmen‐tal change.In the study of aridification in inland areas of Asia,pollen is considered to be one of the most use‐ful environmental indicators (Sun and Wang, 2005).As pollen comes directly from plants, it is very sensi‐tive to climatic factors such as precipitation and tem‐perature,as well as terrain.It also has advantages of be‐ing preserved in greater quantities than other plant fos‐sil organs, thus enabling its widespread application in paleoenvironmental reconstruction. This paper summa‐rizes the role of palynology in revealing aridification at different time scales since the late Cenozoic(Figure 1),and discusses its existing problems,to assess the advan‐tages and weaknesses of the subject in future research and to stimulate palynological development.

Figure 1 The geographical location of inland arid regions of Asia(modified from(Wang et al.,2017)).The red cross,black dot,and green dot represent late-Cenozoic,late-Quaternary,and Holocene research sites,respectively(The detailed information of the study sites are attached in supplemental material Table 1)

2 Important research progress

2.1 Late Cenozoic tectonic timescales

At the tectonic time scale, large amounts of detri‐tal material have been deposited at the northern margin of the Qinghai-Tibet Plateau in Inner Asia. Subsequent erosion and cutting under tectonic action have provid‐ed ideal natural outcropping profiles for studies.There‐fore, studies on palynology at the tectonic scale of the late Cenozoic have mainly focused on this area. Re‐cords established in other palynological studies at the periphery of the Qinghai-Tibet Plateau, including cen‐tral Asia (north Iran (Song, 1999; Ballatoet al.,2010)),north and west Kazakhstan(Akhmetyevet al.,2005), Zaisanskaya Basin (Akhmetyevet al., 2005),the Mongolia plateau and surrounding areas (Wang,1990; Alexeevaet al., 2001; Kataokaet al., 2002)have not yielded a continuous time sequence. On the northern edge of the Qinghai-Tibet Plateau, based on the combination of a long-sequence of cores and seis‐mic profiles, studies have achieved greater detail, es‐tablishing the sedimentation sequence and sedimenta‐ry age, yielding abundant materials for paleoclimate research.For example,pollen assemblages were estab‐lished on the north and south sides of the Tianshan Mountains, to reconstruct the climate change on lon‐ger time scales, at sites including the Taxihe profile on the northern slopes (26.5−2.6 Ma, Figure 2a) (Sun and Zhang, 2008), Jingouhe profile on the northern slopes (28 −4.2 Ma, Figure 2b) (Tanget al., 2011),Kuchetawu profile on the southern slopes(13.3−2.6 Ma,Figure 2c) (Zhang and Sun, 2011), Sikouzi profile in the Liupan Mountains(20.13−0.08 Ma,Figure 2d)(Ji‐ang and Ding, 2008), Laojunmiao profile in the Jiu‐quan Basin (<13 Ma, Figure 2e) (Maet al., 2005),and Maogou profile in the Linxia Basin (30.6−5.0 Ma,Figure 2f) (Maet al., 1998) (Figure 1 for locations of late-Cenozoic sites). Then, based on ordination analy‐sis methods (e.g., DCA, PCA), or by using the per‐centage of xerophytic types to construct a drought in‐dex, the trends and general processes of aridification in Inner Asia were discussed.

Figure 2 Significant paleorecords since the Miocene in Inner Asia:(a)Taxihe,the northern slope of the Tian Shan Range(Sun and Zhang,2008);(b)Jingouhe,the northern slope of the Tian Shan Range(Tang et al.,2011);(c)Kuchetawu,the southern slope of the Tian Shan Range(Zhang and Sun,2011);(d)Sikouzi,Liupan Mountains(Jiang and Ding,2008);(e)Laojunmiao,Jiuquan Basin(Ma et al.,2005);(f)Maogou,Linxia Basin(Ma et al.,1998);(g)Tianshui Basin(Hui et al.,2011;Liu et al.,2016);(h)KC-1,xerophytic taxa percentages(Miao et al.,2011a;Cai et al.,2012;Miao et al.,2013a);(i)KC-1&SG-3,xerophytic concentration(Cai et al.,2012;Miao et al.,2016a);(j)ASM(Asian summer monsoon)intensity reconstructed using special pollen of Fupingopollenites(Miao et al.,2016d);(k)Long-term microcharcoal concentration(Miao et al.,2016b).Xerophytic types include Amaranthaceae,Ephedra,and Nitraria

We have conducted a series of palynological stud‐ies in the Qaidam Basin during the past decade. The Qaidam Basin is located in the transition zone be‐tween extreme drought and monsoon areas, with an annual precipitation of about 300 mm in the east and less than 50 mm in the west. It is a typical arid region of Inner Asia (Figure 1). Based on palynological anal‐ysis of the KC-1 core in the western Qaidam Basin it has been found that, since 18−5 Ma in the late Ceno‐zoic, drought-tolerant plants such asEphedra, Cheno‐podiaceae (now transferred into Amaranthaceae) andNitrariahave been widely distributed in the basin and have shown an increasing trend indicating continuous drying. At the same time, the surrounding mountains(the Altyn Taph Mountains to the northwest, Qilian Mountains to the northeast and Kunlun Mountains to the south) are dominated by coniferous species such asPinus,Picea,Abies, and Podocarpaceae, and their decreasing percentages reflect the deterioration of the mountain climate (Figure 2h) (Miaoet al., 2011a).When combined with sporopollen results of other re‐cords in the Qaidam Basin(Caiet al.,2012),a contin‐uous and complete Late Cenozoic record of the basin was established, revealing details of drying processes in the basin (Figure 2h) (Miaoet al., 2013a). Recent‐ly, pollen concentrations have enabled discussion of aridity(Miaoet al.,2016a)(Figure 2i),and it is worth mentioning a special pollen type −Fupingopollenites.This is an extinct palynomorph assigned to the new fossil genus:F. wackersdorfensis(Thiele-Pfeiffer)Liu, andF. minutusLiu (Liu, 1985), reflecting the type locality, Fuping County in Guangxi, South Chi‐na, dated as Miocene. This type is usually found in the basin's profiles and its environmental significance has been fully discussed by Miaoet al.(2016d).Based on a thorough literature search and confirma‐tion of the palynological morphology, the "Coexis‐tence Approach" method was used to determine the annual precipitation in its living environment, reveal‐ing this type requires annual precipitation of more than 1,000 mm.On this basis,pollen data from central and eastern Asia, including the Yangtze River delta(Zhanget al., 2013) were analyzed in detail.The west‐ernmost location of Fuping pollen was in the eastern part of the basin at 18 −14 Ma, and after 14 Ma, its range contracted to the southeast before disappearing during the Quaternary. This change clearly reveals that aridity expanded eastwards to Southeast Asia, consis‐tent with decreasing precipitation in Southeast Asia(Figure 2j) (Miaoet al., 2016d). In addition, the study of long-term microcharcoal records in the Qaidam Ba‐sin has revealed a general trend of increasing wildfires in Inner Asia since 18 Ma (Figure 2k), also indicating gradual drying (Miaoet al., 2016b). These records(Miaoet al., 2011a,b; Miaoet al., 2013a; Miaoet al.,2016a; Miaoet al., 2016d) are consistent with sporo‐pollen records from the Tianshui Basin in the eastern part of the arid area (Figure 2g) (Huiet al., 2011; Liuet al., 2016), non-sporopollen records from the arid ar‐ea and the Loess Plateau, expansion of inland deserts(Lu and Guo, 2014; Luet al., 2019), records from Sea of Japan and the northern Pacific region (Reaet al.,1985),and the latest results of many studies in the cen‐tral part of the North Pacific Ocean(Yanget al.,2016).

Thus, advantages of the palynology proxy are highlighted in the aforementioned studies of aridifica‐tion in the Asian interior (Thilakanayakaet al., 2019),along with corresponding research methods,sporopol‐len identification and naming systems (Voigtet al.,2017).At the same time, pollen records have been re‐garded as vital evidence in the arid region of Inner Asia (Nieet al., 2018), supporting simulations of at‐mospheric dynamics (Zhanget al., 2019), and provid‐ing an important reference for the study of the rela‐tionship between carbon (Caveset al., 2016), oxygen isotopes(Botsyunet al.,2016),vegetation type chang‐es(Shenet al.,2018)and climate changes in the Qing‐hai-Tibet Plateau. In addition, the northern part of the Qinghai-Tibet Plateau is tectonically active, and these palynological studies enable distinction between tec‐tonic and climatic signals in sedimentary records, fur‐ther emphasizing the important role of high-quality paleoclimatic records(Changet al.,2015).

2.2 Late Quaternary orbital time scales

Similar to the late Cenozoic, studies of the late Quaternary have largely focused on the central and eastern part of the arid region. For example, in west‐ern Turkmenia (Figure 1 for the geographical loca‐tion), woody plants were dominant in the early Pleis‐tocene(Malgina,1961),and the presence of forest ele‐phant remains (Dubrovo, 1960) indicates a warm-wet climate. The steppe/desert vegetation dominated byArtemisiaand Chenopodioideae was widely devel‐oped during the Late Quaternary, which reflects a dry climate(Chupina,1974).Likewise,the vegetation pre‐sented a similar pattern in the eastern part of inland ar‐id areas (mainly referring to the Xinjiang region),changing from a warm-moist vegetation community(dominated byPinus,Picea,Tsuga,Podocarpus,Jug‐lans,Ulmus,Celtis,Tilia,Carpinus,Quercus,Acer,Lonicera) to drought-tolerant plants (Wang and Yan,1987). Specifically, the mountain areas were mainly coniferousPinusandPicea,while the basin was dom‐inated by desert and/or steppe with herbaceous plants(e.g., Poaceae and Sedges) and shrubs (Artemisiaand Chenopodioideae)(Yan,1991).By reviewing the pub‐lished literature, we found that drought-tolerant shrubs and herbs have dominated most periods in the inland arid areas since the Late Quaternary. Some re‐cords (XKL(Wuet al., 2020)), SG−1 (Koutsodendriset al., 2019), SG−3 (Caiet al., 2012), Chaona (Wuet al., 2007) (see Figures 1 and S1 for locations of late-Quaternary sites) show substantial increases in their drought-tolerant plant contents (Figure 3), suggesting aridification was more severe. We assume this was closely connected with cooling during this period.

Palynological records of some sites can show characteristics of the glacial-interglacial cycle in the context of aridification. For example, in the central Tianshan mountains, the forest line moved upwards and the forest belt widened during the Quaternary in‐terglacial intervals, while the forest line shifted down‐ward and the forest belt narrowed during glacials(Yan, 1991). Similarly, a record from the Lop Nur Lake (eastern margin of the Tarim Basin) revealed an arid vegetation community dominated by desertshrubs and herbs (e.g.,Artemisia, Chenopodioideae,Ephedra) (Haoet al., 2012); vegetation coverage was relatively high during interglacial periods and low during glacial periods (Yanget al., 2013). New pollen data from the Zoige Basin (Figure 3e) at the eastern margin of the Qinghai-Tibet Plateau reveal that vege‐tation has been driven by changes in ice sheets in the northern hemisphere since the Late Quaternary (Fig‐ure 3f), and has a characteristic about 100,000 years periodicity (Zhaoet al., 2020). However, it should be noted that compared with records from Europe, the extent of glacial-interglacial changes recorded in the palynological records of inland Asia is much weaker and more complex. For example, the Yandonggou(YDG) section in the eastern arid region recorded al‐ternating patterns of dark coniferous forest,sparse for‐est, grassland, shrub grassland and desert since the Late Quaternary (Pan, 1999). In general, precipitation in inland arid areas was low,and although it increased during interglacial periods, it did not reach the mini‐mum water threshold for the growth of irrigation-de‐manding trees. In addition, when compared with Eu‐rope, the inland terrain of Asia is very complex and precipitation gradients in the mountainous areas and basins vary widely. Under these conditions, the gla‐cial-interglacial vegetation pattern is distinct from that in European areas with woody plants in the inter‐glacial period and the development of herbs in the gla‐cial period(Allenet al.,1999;Williset al.,2000;Tze‐dakiset al., 2004; Tzedakis, 2005; Wohlfarthet al.,2008; Allen and Huntley, 2009). The vegetation changes in the late quaternary in this arid region need to be further studied,to reveal in more detail the char‐acteristics of changes on orbital time-scales.

Figure 3 Pollen records from inland areas of Asia during the late Quaternary,and the global ice volume:(a)XKL(Wu et al.,2020);(b)SG-1(Koutsodendris et al.,2019);(c)SG-3(Cai et al.,2012);(d)Chaona(Wu et al.,2007);(e)Zoige Basin(Zhao et al.,2020);(f)LR04 global benthic δ18O stack(Lisiecki and Raymo,2005)

2.3 Holocene sub-orbital time scales

Studies in this period are characterized by their wide spatial coverage (see Figures 1 and S1 for loca‐tions of Holocene sites) and good age control, en‐abling a deeper understanding of vegetation changes across large areas. Early studies on Holocene vegeta‐tion change were mainly based on lake sediments,e.g.,Aibi Lake (Wen and Zheng, 1988), Balikun Lake(Hanet al., 1989) and Manas Lake (Rhodeset al.,1996),and they found a complex pattern between veg‐etation and climate, reflecting different combinations of temperature and humidity, although they did not reach consistent conclusions. Subsequent integration of additional Holocene climate data documented spa‐tiotemporal differences in precipitation/moisture con‐ditions between arid central Asia and monsoonal Asia on different time scales (Chenet al., 2016; 2019b).Palynological results have revealed that herbs and shrubs developed in the early Holocene with low veg‐etation coverage, indicating a relatively dry climate in this region (Herzschuh, 2006; Chenet al.,2008;Chenet al., 2016) (Figures 4a, 4b).Woody plants and over‐all vegetation coverage increased in the middle Holo‐cene (Figure 4f), reflecting a good combination of temperature and moisture conditions. However, a re‐cent publication documented a drought condition dur‐ing the mid-Holocene in arid central Asia (Xuet al.,2019), which was accordant with outputs from cli‐mate models(Jianget al.,2013).This raised an impor‐tant issue: whether the expansion of trees during the mid-Holocene was mainly affected by temperature changes, or was controlled by the moisture/precipita‐tion condition as previous studies suggested (Zhaoet al., 2011).At present, woody plants, including conifer‐ous forest (mainlyLarix sibirica,Picea schrenkiana)and mixed broadleaf-conifer forest(mainlyBetula,Lar‐ix sibirica,Picea schrenkiana),are restricted to high-al‐titude mountain areas (Fang and Yoda, 1990). Körner(2012) pointed out that tree growth and productivity at high elevations were often limited by the brevity of the growing season due to prevailing low temperatures.To better understand this, a synthesized comparison of variations in trees and their climatic implications was necessary, especially in the drylands in Central Asia.During the late Holocene, trees were replaced byAr‐temisa-dominated grassland in most basins and even the low mountain areas which use to be occupied by coniferous or mixed-forest trees.This can be explained by the increased moisture condition resulting from in‐creased jet-stream activity caused by elevated annual latitudinal insolation gradient (Routsonet al., 2019),which was proved by other proxy-based reconstruc‐tions (Chenet al., 2016; 2019b; Zhang and Feng,2018)and simulations(Zhanget al.,2017).Further ex‐pansion of research sites to more areas, including the northern part of Xinjiang, the Altai Mountains and the Mongolian Plateau (Wang and Feng, 2013; Zhang and Feng,2018;Zhanget al.,2020)(Figures 4c,4e)has in‐dicated significant spatial differences in vegetation re‐sponses to climate. For example, increasing precipita‐tion driven by the westerlies in the early Holocene would have first reached the threshold for forest expan‐sion in the high-elevation sub-region, but was signifi‐cantly delayed in the low-elevation sub-region.In addi‐tion, sensitivities of vegetation to Holocene tempera‐ture and precipitation changes varied between species.For example,Pinus sibiricaandPinus sylvestriswere sensitive to the westerlies-associated precipitation changes, whilePiceaandArtemisiawere more sensi‐tive to temperature changes (Zhanget al., 2020).Thus,studies on the proper quantification of different species or vegetation biomes vulnerability to stressing factors such as climate change and other factors should be stressed. For example, satellite-based remote sensing data documented that decrease in precipitation weak‐ened plant growth for temperate biomes in the last de‐cades (Huxmanet al., 2004). Similarly, the growing season Normalized Difference Vegetation Index (ND‐VI), which can be an indicator of plant growth, de‐creased with a reduction in precipitation for temperate grasslands of China (Fanget al., 2005). More studies are needed to apply these short-term observations to the reconstruction of Holocene vegetation dynamics,hence evaluating the impacts of future climate change on terrestrial ecosystems.Also, the influence of human activity on vegetation is another factor that needs to be addressed, even though we still have limited evidence on how human beings directly alter Holocene vegeta‐tion structures in inland arid areas of Asia.Recent stud‐ies documented an obvious increase in archaeological sites in Northwestern China during the late Holocene(Donget al., 2013), indicating an elevated intensity of prehistoric human settlement in these areas.Anthropo‐genic activities have altered the response of regional bi‐ome burning to climate changes (Xueet al., 2018;Zhanget al., 2020). Thus, distinguishing the patterns,processes and mechanisms on how human activities af‐fect the natural vegetation succession at local and re‐gional scales is beneficial to the understanding of the human−environmental interactions(Dong,2018).

Figure 4 The integrated effective moisture conditions in inland areas of Asia during the Holocene.(a)Westerly Central Asia(Chen et al.,2008);(b)Monsoonal Central Asia(Herzschuh,2006);(c)Altai Mountains(Zhang and Feng,2018);(d)Northern Xinjiang(Wang and Feng,2013);(e)North Mongolia Plateau(Wang and Feng,2013);and(f)variations of woody plants in eastern Tibetan Plateau(Zhao et al.,2011).The effective moisture and stacked woody pollen types were both normalized by Z-score value

2.4 Modern processes

Based on pollen percentages, statistical methods,vegetation cover analysis and meteorological data,studies of modern pollen processes enable a better un‐derstanding of pollen production,dispersion and depo‐sition, hence representing an important foundation for reconstructing paleovegetation and paleoclimate (An‐dersonet al., 1989; Guiot, 1990; Horowitz, 1992; Lu‐ly, 1997; Couret al., 1999; Maet al., 2008; Shenet al., 2008; Zhao and Herzschuh, 2009; Lüet al., 2011;Xuet al., 2012; Guoet al., 2020; Liuet al., 2020; Luet al., 2020). Hitherto, research on modern pollen samples in the inland arid and semi-arid regions of Asia has mainly included top soils, lake sediments and moss polsters(Yan and Xu,1989;Li,1991;Wenget al., 1993;Wanget al., 1996; Xuet al., 1996; Liuet al., 1999; Chenet al., 2004; Lüet al., 2004a; Lüet al., 2004b; Zhang, 2004; Zhuet al., 2004; Liet al.,2005a; Liet al., 2005c; Xuet al., 2005; Shenet al.,2006; Herzschuh, 2007; Xuet al., 2007; Zhaoet al.,2007; Maet al., 2008; Shenet al., 2008; Luoet al.,2009; Xuet al., 2009; Zhao and Herzschuh, 2009;Cheng and Chen,2010;Huanget al.,2010;Luoet al.,2010; Zhanget al., 2010; Zhao and Sun, 2010;Weiet al., 2011; Zhaoet al., 2012a; Wuet al., 2013a; Zhao and Li, 2013; Qinet al., 2015; Wei and Zhao, 2016;Yanget al., 2016; Changet al., 2017; Geet al., 2017;Huanget al., 2018; Chenet al., 2019a; Cuiet al.,2019; Guoet al., 2020; Liuet al., 2020; Wanget al.,2020), while studies of airborne pollen are limited(Couret al., 1999; Xuet al., 2006; Liet al., 2008; Panet al., 2013; Liet al., 2015b; Xuet al., 2015; Liet al.,2019b; Lüet al., 2020).Therefore, we will focus here on reviewing surface pollen research(Figure 5).

Figure 5 Locations of topsoil samples in inland arid areas of Asia(Pollen data(black cross)are from the Eastern Asia Surface Pollen Dataset(Cao et al.,2014)and our filed collection(green cross),meteorological data are from China Meteorological Data Network)

Early studies on surface sediments in inland re‐gions of Asia were concentrated in the central and eastern regions (Yan and Xu, 1989; Li, 1991; Pan,1993; Wenget al., 1993; Yan, 1993; Wanget al.,1996; Xuet al., 1996; Yanet al., 1996; Liuet al.,1999), including the Altai, Tianshan (Nanshan sec‐tion, Chaiwopu Basin, Tianchi, north slope), West Kunlun, Taklimakan desert, and southeastern Inner Mongolia Plateau. Pollen assemblages were dominat‐ed by xerophytic pollen types such as Chenopodioide‐ae,ArtemisiaandEphedra, which generally reflect the respective vegetation types, thus indirectly reveal‐ing the influence of different drought intensities on re‐gional vegetation types. By applying theRmodel(R=p/v,whereprepresents the percentage of the spe‐cies'pollen,andvrepresents the coverage of the corre‐sponding species (Davis, 1963)), Xuet al.(1996)found varyingRvalues of plants in different families and genera. For example, Chenopodioideae,Artemis‐iaandEphedrawere over-represented in arid regions(Yan and Xu, 1989; Wenget al., 1993; Wanget al.,1996; Xuet al., 1996;Yanet al., 1996), while Poace‐ae, Leguminosae and Compositae were under-repre‐sented (Liuet al., 1999). Based on these differences in pollen representation, corrections should be applied to the interpretation/reconstruction of regional vegeta‐tion types when using sedimentary achieves (Pan,1993).

Methods of pollen analysis in earlier studies(Wenget al., 1993; Wanget al., 1996; Liuet al.,1999) can be divided into the following three catego‐ries: (1) pollen percentage; (2) pollen ratios of specif‐ic types, including theArtemisiato Chenopodioideae ratio (A/C),ArtemisiatoBetularatio (A/B), and arbo‐real pollen to non-arboreal pollen ratio (AP/NAP);and(3)statistical methods,including principal compo‐nents analysis(PCA),cluster analysis,regression anal‐ysis, and theR-value. Although these early studies were conducted in limited regions, and were largely qualitative or semi-quantitative, they have improved our understanding of the arid environment and provid‐ed a solid foundation for further investigation of the relationships between vegetation zones in different en‐vironments.

Further studies of palynological processes expand‐ed the spatial coverage to include a wide range of arid and semi-arid regions, such as Xinjiang, Qinghai,Gansu, Ningxia, Inner Mongolia, and Mongolia(Chenet al., 2004; Lüet al., 2004a; Lüet al., 2004b;Zhuet al., 2004; Liet al., 2005a; Liet al., 2005c; Xuet al., 2005; Shenet al., 2006; Herzschuh, 2007; Xuet al., 2007; Zhaoet al., 2007; Maet al., 2008; Shenet al., 2008; Luoet al., 2009; Xuet al., 2009; Zhao and Herzschuh, 2009; Cheng and Chen, 2010; Huanget al., 2010; Luoet al., 2010; Zhanget al., 2010;Zhao and Sun, 2010; Weiet al., 2011; Zhaoet al.,2012a; Wuet al., 2013a; Zhao and Li, 2013; Qinet al., 2015; Wei and Zhao, 2016; Yanget al., 2016;Changet al., 2017; Geet al., 2017; Huanget al.,2018; Chenet al., 2019a; Cuiet al., 2019; Guoet al.,2020; Liet al., 2020b; Liuet al., 2020; Luet al.,2020; Wanget al., 2020). The results of those studies revealed good correspondence between modern pol‐len spectra and vegetation types in different regions,consistent with previous research (Yan and Xu, 1989;Li, 1991; Pan, 1993; Wenget al., 1993; Yan, 1993;Wanget al., 1996; Xuet al., 1996; Yanet al., 1996;Liuet al., 1999). However, the strength of correlation between the pollen records and vegetation showed re‐gional variability. A representative study of a wider family of pollen genera highlighted the significance of the relationship between modern pollen assemblag‐es and environmental evolution. In addition, as meth‐ods of pollen data analysis developed beyond those mentioned above, the pollen ratios method (Her‐zschuhet al., 2003; Chenet al., 2004; Lüet al.,2004b; Liet al., 2005a; Liet al., 2005c; Xuet al.,2005; Liuet al., 2006; Herzschuh, 2007; Luoet al.,2007; Xuet al., 2007; Shanget al., 2009; Zhao and Herzschuh, 2009; Cheng and Chen, 2010; Weiet al.,2010; Weiet al., 2011; Zhao and Li, 2013; Wanget al., 2020) expanded to include, for example, theArte‐misiato Cyperaceae ratio(A/Cy),Poaceae toArtemis‐iaratio (P/A), and the sorting method (Chenet al.,2004; Lüet al., 2004b; Lüet al., 2004c; Liet al.,2005a; Liet al., 2005b; Liet al., 2005c; Xuet al.,2005; Liuet al., 2006; Shenet al., 2006; Xuet al.,2007; Zhao and Herzschuh, 2009; Weiet al., 2011;Yanget al.,2011;Zhanget al.,2017).Quantitative re‐construction methods were developed, such as Biomi‐sation (Herzschuhet al., 2003; Herzschuhet al.,2004; Zhenget al., 2008b; Fenget al., 2011), calcula‐tion of representation (Herzschuhet al., 2003; Liet al., 2005a; Liet al., 2005b; Luoet al., 2008; Weiet al., 2009; Zhao and Herzschuh, 2009; Weiet al.,2010; Wuet al., 2013b,c), the modern analogue tech‐nique (Zhenget al., 2009), and transfer functions(Zhang, 2004; Shenet al., 2006; Maet al., 2008;Wanget al.,2009;Lüet al.,2011).Meanwhile,the es‐tablishment of the Eastern Asian Modern Pollen Data‐base(EAMPD)and its application in the northwestern region(Zhenget al.,2008a;Luoet al.,2010),provided the foundation for the study of surface sporopollen on a large spatial scale in arid and semi-arid regions, as well as improving the accuracy of the functional rela‐tionship between precipitation and pollen assemblages.Subsequently, on the basis of the large-spatial-scale and large-data-volume studies on the Qinghai-Tibetan Plateau (Lüet al., 2011) and in the northwestern re‐gions (Zhaoet al., 2012a), EAMPD was updated and improved (Zhenget al., 2014), becoming an impor‐tant palynological resource in climate reconstruction methods in East Asia (Caoet al., 2014). These stud‐ies serve as references for understanding ecological evolution and its controls (e.g., temperature; precipi‐tation) in desert environments (Luet al., 2020), in‐cluding changes over geological time scales. For ex‐ample, the relationship between pollen of woody plants in the topsoil and the intensity and direction of the monsoon suggested that modern pollen of woody plants might be a good indicator of a monsoon cli‐mate (Liet al., 2020b). In short, modern palynologi‐cal research applied to environmental change has con‐tinued to develop, the number of samples has in‐creased, and focus has shifted to comparative studies across wide areas. Research methods have developed from qualitative to semi-quantitative/quantitative,pro‐viding references for paleoclimatic and environmental reconstructions.

In summary, palynology has achieved great prog‐ress over different time scales in arid regions of in‐land Asia, yielding valuable plant ecological data that forms the basis for discussing the driving mechanisms of environmental changes, as well as providing new insights into other scientific objectives.

3 Problems

Based on the research progress described above and on our experiences in palynology, it can be con‐cluded that the application of palynology in arid re‐gions still has shortcomings which need addressing,in two key aspects. The first relates to the extraction and identification of pollen and spores. The tradition‐al extraction method has been retained with few changes, but without systematic and professional un‐derstanding of the experimental process or any at‐tempt at experimental fine-tuning to increase the suc‐cess rate of extraction. The difficulty of identification has increased because of the complexity and similari‐ty of pollen and spore morphologies and/or the strong deformation of pollen and spores in the sedimentary profile. The second problem concerns the interpreta‐tion and quantitative reconstruction of paleoclimates.Differences between the different pollen and spore taxon in terms of their production, dispersal, deposi‐tion and preservation, as well as the influence of the vertical zonality associated with the uplift of the pla‐teau in the late Cenozoic, all remain poorly con‐strained.Therefore, the interpretation of palynological results is complicated, and limits quantitative recon‐structions.These problems present a barrier to palyno‐logical research and innovation.

3.1 Problems in extraction and identification

The methods of pollen and spore extraction are fundamental tasks in palynology. In China, the most common methods used for extracting pollen and spores are mainly referred to as the method (Figure 6)developed by Liet al.(1995), while in other coun‐tries, the methods are mainly based on references by Doher(1980),Traverse(2007)and Brown(2008).Ad‐ditionally,Lycopodiummarkers are added to calculate pollen concentration as proposed by Stockmarr(1971). Subsequently, Kitaba and Nakagawa (2017)attempted to employ black ceramic spheres as a new kind of maker for calculation of pollen concentra‐tions. Unfortunately, this method is rarely adopted. In China, many studies have been carried out to explore targeted methods for various kinds of sediments. For example,Liet al.(1999)introduced the sieving-analy‐sis method to analyze pollen and spores in loess sedi‐ments because this method has advantages of needing a smaller sample, short processing time, and low lev‐els of pollen and spore losses. Afterward, Li and He(2004) suggested that loess sediments should be treat‐ed with HCl and HF repeatedly during the extraction,considering the high content of SiO2, Al2O3and Ca‐CO3. Moreover, a comprehensive study on various types of Quaternary sediments (Liet al., 2007) con‐cluded that loess-paleosol samples should be analyzed using HF treatment and sieving; lake sediments in North China should be analyzed by adding a heavy-liq‐uid processing stage after the traditional HF treatment method; aeolian sand samples should be analyzed with a combination of the HF treatment method and heavy-liquid method; peat and bog sediments should be analyzed using the heavy liquid method; some im‐pure samples require the sieving-analysis method, ar‐chaeological sample analysis should employ the heavy liquid method and then HF treatment; finally, surface samples require the heavy-liquid method without acidalkali(Liet al.,2007).In addition,the use of a disper‐sant and nitric acid, and the appropriate order of acids and alkalis,has been discussed(Liet al.,2007).

Figure 6 The extraction procedure for pollen and spores

A few improvements in the laboratory extraction process for Cenozoic stratigraphic samples have been made,as these are different from the loose Quaternary sediments. Potential problems such as extraction effi‐ciency, effectiveness and safety during the traditional pollen and spore extraction process have been ad‐dressed, achieving improvements in the extraction procedure.Among these, a simple and safe automatic device was developed to replace the manual water ex‐change (two washings to neutralize), which shortened the processing time by 70% and decreased the loss of pollen and spores(Miaoet al.,2013b,c).Furthermore,an automatic device suitable for pollen and spores was specially designed for the artificial stirring step,guaranteeing the automatic mixing and stirring at a uniform and slow rate. This device shortened the acid holding time by 80%, and improved the reaction effi‐ciency by 50%(Miaoet al.,2014).Our improvements increased the extraction efficiency,effectiveness,safe‐ty, and time cost, enabling the device to processes more samples while achieving the quantity-for-quality strategy. This approach guarantees high-quality paly‐nological records.

During identification, many factors could influ‐ence accuracy: for example, similarities in the mor‐phologies of different pollen genera;damage to pollen and spores during the sedimentary deposition and lab‐oratory extraction processes. Therefore, to improve the accuracy of identification,we suggest:1)strength‐ening the study of standard slides of modern pollen,especially the comparison of those with similar mor‐phology; 2) evaluating the influence of different treat‐ment steps on the outer wall of pollen in the laborato‐ry processing, to minimize damage to the external wall; and 3) increasing the use of scanning electron microscopy, as an auxiliary identification method which can enhance reliability of the results.

3.2 Challenges in interpretation and quantitative reconstruction

Sporopollen grains are very small, and after leav‐ing their host plant, will have undergone processes in‐cluding transportation, sorting, sedimentation, preser‐vation, and disturbance. These are often neglected,hindering the application of sporopollen in explaining climate changes. The northwest region of China is mostly composed of basins and mountains,with rapid‐ly changing precipitation gradients and unclear zonal patterns, in great contrast to the simple spatial varia‐tion of European vegetation zones.Therefore, it is not appropriate to simply apply the European model;meanwhile, the vertical temperature gradient is signif‐icant, yet the humidity pattern is also complex, mak‐ing it hard to find a clear relationship between pollen assemblage and climate. Many areas in this region are high-energy, glacial-fluvial or tail-end lake systems,where rapid changes in sedimentary phase also impact the palynological interpretation. For instance, in the Shiyang River Basin in eastern arid areas, different sections show different Holocene sporopollen assem‐blages,even in two adjacent records:one indicates co‐niferous forest during the Holocene, while the other suggests dominance of xerophytic herbs and shrubs,showing that sporopollen assemblages vary under dif‐ferent sedimentary environments and sedimentary fa‐cies (transport media) (Zhuet al., 2001). In addition,the extremely arid climate biases the modern palyno‐logical reconstruction, leading to an overestimation of precipitation (Caoet al., 2017). The relationship be‐tween surface sporopollen and environmental factors is fundamental to the quantitative reconstruction of paleoclimate, and is very difficult to quantitatively re‐construct paleoclimate on the basis of such complex sedimentary facies and diverse pollen sources. There‐fore, problems remain in the study of surface sporo‐pollen, as follows. (1) The temporal (Liet al., 2005c;Sugita, 2007) and spatial (Zhao and Herzschuh, 2009;Geet al., 2017) representativeness of surface sporo‐pollen analysis are unclear, with surface sporopollen patterns depending on the local environment, pollen yield, dispersal, preservation, and other factors: there is a large deviation between palynological assemblage and land cover type.(2)The sampling site characteris‐tics are not completely matched to those of the meteo‐rological station. The spatial distribution of surface pollen samples is heterogeneous, and studies are based on meteorological data from weather stations;however, in arid and semi-arid areas, the spatial het‐erogeneity of climate (particularly east-west), and the strong locality of atmospheric circulation, causes in‐terpolated meteorological station data to poorly repre‐sent conditions at the site of the surface pollen sample(Liet al., 2003; Song and Zhang, 2003). This impairs the accuracy of inferred relationships between paly‐nology and climate. (3). The collected samples may not fully represent the natural vegetation,due to artifi‐cial disturbance. On the one hand, because the meteo‐rological stations are mostly distributed in densely populated areas, the palynological results obtained by direct sampling near the meteorological stations may not represent natural conditions. On the other hand,the sampling points are mainly distributed in areas with strong human activities, or near busy roads; usu‐ally, the surface vegetation in these areas has been se‐verely altered, leading to very noisy palynology-cli‐mate relationships.

Caution is certainly needed when simply applying modern results to paleoenvironmental reconstructions.For example,ArtemisiaandChenopodioideaeare widely distributed in deserts and steppes, and their ra‐tio (A/C) has been proposed as an effective proxy based on their different water requirements in the veg‐etation growing season. It is generally believed that a low A/C represents an arid environment, while a high A/C represents a semi-arid environment (El-Moslima‐ny, 1990). This index has become widely used as a proxy of paleo humidity/rainfall in arid and semi-arid regions, over different time scales including the Holo‐cene (Zhaoet al., 2007; Zhaoet al., 2008; Wanget al., 2009; Zhanget al., 2010; Zhaoet al., 2010b;Chenet al., 2013), Pleistocene (Herbet al., 2013;Koutsodendriset al., 2018), Pliocene (Wanget al.,2006; Caiet al., 2012; Herbet al., 2015), and late Miocene (Haoet al., 2012; Koutsodendriset al.,2019).However,some studies have pointed out funda‐mental problems in the A/C ratio,such its spatial foot‐print and minimum content requirement (Wenget al.,1993; Sunet al., 1994; Zhaoet al., 2008; Zhaoet al.,2012b; Wanget al., 2020).Artemisia L. originated in arid or sub-arid areas of temperate Asia in the mid-Ce‐nozoic (Wang, 2004), consistent with evidence based on molecular biology (Tkachet al., 2008). Whether the A/C ratio can be used to indicate rainfall/humidi‐ty during this early period is an open question (Miaoet al., 2011c). For instance, in two adjacent sediment cores SG-1 (about 7.5 Ma) and SG-3 (about 3 Ma) in the Qaidam Basin, the respective A/C ratios showed diametrically opposite trends since 3 Ma (Figure 7)(Caiet al., 2012; Koutsodendriset al., 2019), and it is unclear what other factors control the A/C ration besides moisture conditions. The upper age limit on the reliability of the A/C ratio in pollen records as a proxy for moisture availability in Central Asia re‐mains uncertain.

3.3 Innovation problems: tradition limitation,slow updates and limited innovation

Due to the problem of poor pollen preservation in arid areas, research into microcharcoals needs to be strengthened. Microcharcoals are normally considered as micron-sized carbon chips with similar sizes to spo‐ropollen grains, and obtained by the same palynologi‐cal extraction method.On long-time scales,wildfire is considered as one of the most important factors driv‐ing the evolution of the earth's ecosystems and is closely related to other environmental factors. There‐fore, long time-scale wildfire records are of great sig‐nificance to the understanding of past environmental and ecological changes (community compositing and abundance).As the residual matter of plants is burned by fire, the resulting microcharcoal called the "fossils of fire" provides a unique material by which to study fire history, such as the scale, intensity, fuel burning type (herb or wood) (Umbanhowar Jr and Mcgrath,1998; Enache and Cumming, 2006) distance from the study site by using charcoal concentration (Patterson IIIet al., 1987; Clark, 1988; Whitlock and Anderson,2003), and ratio of charcoal length to width (Enache and Cumming, 2006; Daniauet al., 2013; Crawford and Belcher,2014).Also,fire remains offer the oppor‐tunity to assess climate condition and human activi‐ties and their driving factors(Thevenon and Anselmet‐ti, 2007; Bowmanet al., 2009). Many studies have been carried out on Holocene records in inland Asia,exploring the relationships between climatic factors and human activities (Huanget al., 2006; Zhanget al., 2008; Zhaoet al., 2010a; Zhouet al., 2011; Jianget al., 2012; Zhouet al., 2012; Wanget al., 2013; Liet al., 2015a; Miaoet al., 2016c; Zhouet al., 2016;Donget al., 2017; Miaoet al., 2017b; Xueet al.,2018; Liet al., 2019a; Miaoet al., 2020). However,over long time-scales, the only microcharcoal records obtained to-date are coarse-resolution records from the North Atlantic reported by the Ocean Drilling Pro‐gram (Herring, 1985). Through nearly 10 years of work in the hinterland of Asia, such as the Qaidam Basin, we obtained charcoal sequences for different periods and at different resolutions, which have re‐vealed a general trend of increasing wildfires in the Asian interior during the past 18 Ma (Figure 2) (Miaoet al.,2016b),this supports the interpretation of the re‐gion's increasing aridity as revealed by palynological records. Furthermore, based on the morphology of the charcoal grains, wildfires were considered to have mainly occurred in the forest-steppe ecotone, charac‐terized by a shrinking area and increasing fire frequen‐cy (Miaoet al.,2019).Under the definition of general palynology, we suggest here that charcoal should be included as a sub-field of general palynology, as this would enrich palynological research.

Figure 7 (a)Four major stages in the development of Artemisia.Stage I,Presence(Wang,2004;Tkach et al.,2008);II,expansion but at low levels(Wang,2004;Webb,2013);III,Artemisia becomes the most abundant component(Wang,2004),IV,expansion and diversification(Wang,2004).(b)A/C in SG-3(Cai et al.,2012).(c)A/C in SG-1(Koutsodendris et al.,2019)

Untraditional palynological methods are of great significance to paleoclimate studies. Concentrations or fluxes of microparticles such as fungus spores(e.g.,AcritarchsandCladocera) (Miaoet al., 2017a;Miaoet al.,2018a;Miaoet al.,2018b)and palynolog‐ical facies (Powellet al., 1990; Li and Batten, 2005;El-Soughieret al., 2010) should be considered as valuable data. When pollen grains are scarce, bryo‐phytes should be included in palynological statistics to obtain a more reliable palynological assemblage,in‐stead of being discarded because the samples cannot meet the palynological statistical standards (Miaoet al.,2015).

Few studies have addressed far-infrared and isoto‐pic analysis methods; however, in recent years, new indicators have been explored. First is the analysis of sporopollen carbon isotopes based on SWiM-IRMS(Nelsonet al., 2007; Urbanet al., 2013; Zhaoet al.,2017)and laser ablation nano combustion gas chroma‐tography/isotope ratio mass spectrometry (Roijet al.,2017), which enables the use of nanogram-level solid C (single sporopollen particles) to explore the rela‐tionship between a plant's water use efficiency and the environment (Nelson, 2012), as well as indicating changes of C3/C4 in the past (Nelsonet al., 2006;Nelsonet al., 2008; Urbanet al., 2010; Urbanet al.,2015; Nelsonet al., 2016). Secondly, microspectros‐copy research based on Fourier Transform Infrared(FTIR) spectroscopy has been employed to analyze the contents of absorbing pigments (aromatic rings)in pollen, which are sensitive to UV-B radiation; the UV-B intensity in the surrounding environment can then be evaluated for that geological period to recon‐struct the paleoenvironment (Rozemaet al., 2001;Blokkeret al., 2005; Watsonet al., 2007; Lomaxet al., 2008; Fraseret al., 2011), ozone layer changes in polar regions (Lomaxet al., 2008), altitude (Watsonet al., 2007), and extinction events (Bencaet al.,2018). In this regard, due to the limitation of singlepollen particle extraction technology and the require‐ment for expensive instruments in low-noise surround‐ings, no research using these methods has been car‐ried out in Central Asia. This contrasts with Europe and North America, where such studies have achieved initial successes; therefore, far-infrared and isotopic research urgently needs promotion.

4 Summary and prospects

To summarize the progress and existing problems of palynological studies described above in inland ar‐id regions of Asia, we are encouraged to continue to improve the extraction method and strengthen innova‐tion, and to reinforce the palynological morphology classification analysis. Advances in Microscopy tech‐nology(e.g.,Scanning Electron Microscope,SEM)of‐fers a wide variety of capabilities suitable for imaging and analysis of pollen morphology and chemical com‐ponents on nanometer scales. Thus, the morphologi‐cal differences between genus and species can be dis‐tinguished using SEM technology, especially for the types being ecological significant. For example, for‐mer studies assumed that the Chenopodioideae were characteristic of highly continental climates with cold winters and dry summers, their abundance elevated with increasing aridity (El-Moslimany, 1990). But our fieldwork in north-western China found that halophyt‐ic tribes of Chenopodioideae, such as Salicornieae,Suaedeae, mainly grew around near-shore lakebed ar‐eas, and they were adapted to the halophytic environ‐ment which thus could not be used as a moisture indi‐cator. The palynological morphology of halophytic types cannot be distinguished from xerophytic types(e.g., Chenopodieae, Atripliceae, Salsoleae, Campho‐rosmeae, Hablitzieae, Corispermeae, Beteae) under light microscopy. Thus, further comparisons between these two types and other similar pollen groups will be conducted under SEM in our future work.After the comparisons, we can develop new climatic proxies or re-evaluate previous work without considering the ecological differences between different plant groups.

Laboratory experiments of flume and wind tunnel experiments can better constrain pollen transport dy‐namics in water and air. Flumes can be used to obtain quantitative information about dispersal and deposi‐tional parameters when they are designed in a way that pollen move under shear stress conditions that would be reasonable in natural environments. Recent flume studies have shown that mud suspensions can travel in bedload and form ripples that accrete into beds by means of forming flocculation, which means pollen incorporated into the fluvial sediments also can be deposited from currents rather than by setting from slow moving or still water (Schieber, 2011). This is different from the traditional proposition that pollen may be deposited at a low flow rate (Brush and Brush 1972; Brownet al., 2007). Thus, the flume experi‐ment can potentially provide a quantitative basis for waterborne pollen sedimentology in the future. In ad‐dition, wind tunnels can provide information about the characteristics of particles (including pollen)movement by wind. Former studies have found that pollen influx coming from dust weather contribute more to the total pollen influx than that coming from non-dust weather (Liet al., 2015), and wind speed can be the primary factor determining airborne pollen assemblages, in addition to temperature and precipita‐tion. A laboratory wind tunnel equipped with particle image velocimetry (Donget al., 2009) can be em‐ployed to provide an air stream for the study of mov‐ing pollen under various conditions.

We will also conduct studies of modern sporopol‐len production, dispersal, and deposition in the basin.For example, we have established approximately 20 airborne traps in various vegetation biomes in arid ar‐eas of northwestern China in the last two years, and plan to put some waterborne pollen traps in the lake volume in the near future. By studying the monthly and/or seasonal changes in the pollen types and abun‐dance, we can obtain a better knowledge of the rela‐tionship between modern sporopollen assemblages and vegetation source.Besides,we are encouraging to explore new fields and new methods of palynological analysis (e.g., using DNA, far-infrared and isotopes).The ultimate aim is to promote palynology and in‐crease its contribution to research arid environmental evolution in inland Asia.

Acknowledgments:

The project is supported by the NSFC (41772181,41807440 and 41888101); the Strategic Priority Re‐search Program of CAS (No. XDA20070200); Young Top Talents Project of the "Ten Thousand Youth Pro‐gram" of the Organization Department of the Central Committee of the CPC; Youth Innovation Promotion Association, CAS (2014383); "Light of West China"Program, CAS; and the NSF of Gansu Province(18JR3RA395). We thank Miao's group members and anonymous referees for discussions and their con‐structive suggestions.