Pollen stratigraphy, vegetation and environment of the last glacial and Holocene—A record from Toushe Basin, central Taiwan

(1)

Quaternary International 147 (2006) 16–33

Pollen stratigraphy, vegetation and environment of the last glacial and

Holocene—A record from Toushe Basin, central Taiwan

Ping-Mei Liew

a,

, Shu-Yue Huang

b

, Chao-Ming Kuo

c

a_{Department of Geosciences, National Taiwan University, 1, Sec. 4, Roosevelt Rd. Taipei 106, Taiwan, ROC} b_{Department of Life Science, National Taiwan University, 1, Sec. 4, Roosevelt Rd. Taipei 106, Taiwan, ROC}

c

Chinese Petroleum Corporation, 3, Sung Ren Rd, Taipei 11010, Taiwan, ROC Available online 9 November 2005

Abstract

The pollen record from the Toushe Basin (231490_{N; 120153}0_{E; 650 m above sea level), a peat bog of central Taiwan, displays a}

continuous vegetation history of the past 96,000 yr BP of monsoon Asia. Instead of today’s closed subtropical evergreen broadleaved forest dominated by Machilus–Castanopsis surrounding the basin, temperate deciduous forest predominated during most of the last glacial. In early MIS 4, Alnus reaches the highest value of the whole sequence (60–70%) representing temperate deciduous forest and relatively cold and arid conditions. Following this stadial, Alnus and herbs (mainly Cyperaceae) dominated alternately, with a minor increase of Castanopsis. Peaks of monolete spores between cal. 42.2 and 37.0 kyr BP (kyr BP represent calibrated years) indicate episodic wet conditions. The later glacial, especially between 23.2 and 18.7 kyr BP, shows a high percentage of Gramineae, indicating dry and possibly sometimes cold conditions. The late glacial shows a remarkable increase of warm-temperate to temperate forest elements, such as Ilex, Cyclobalanopsis and Symplocos. At about 15.1 kyr BP a peak of monolete spores indicates wet–warm conditions. A subsequent sharp increase of Salix and then Gramineae between 13.0 and 11.6 kyr BP corresponds to the Younger Dryas. A warming event at 11.5 kyr BP is also evident. The Holocene is characterized by warm–wet conditions of the overwhelmingly abundant monolete spores since 10.7 kyr BP and the prominent increase of Castanopsis.

1. Introduction

Toushe Basin (231490_{N; 120153}0_{E; 650 m above sea level),}

a 1.75 km2 desiccated peat bog in the low hills of central Taiwan, is one of a series of north–south trending middle Pleistocene tectonic basins along the island’s backbone

(Fig. 1). To the north lies Sun-Moon Lake (or Jih-Yueh

Tan, 750 m above sea level) where a pollen sequence covering time since the last glacial has been reported by

Tsukada (1967). Mountains surrounding these basins range

in altitude from 700 m to higher than 1000 m eastward. The ﬂuvial and lake deposits in Toushe Basin are about 80 m thick. The bog became desiccated at about 1.8 kyr BP.

In the context of global climate, changes in low-latitude areas have the same importance as those at high latitudes. Marine records of the last glacial from offshore Taiwan show that low-latitude sea-surface temperature is not as

warm (Huang et al., 1997) as previously estimated by

CLIMAP (1981). This is also shown for other low-latitude

terrestrial areas (Flenley, 1979; Hooghiemstra, 1989;

Farrera et al., 1999). On the other hand, a chronological

discrepancy of climatic events between the two poles is recognized (Sowers and Bender, 1995). Thus, the latitu-dinal variations of regional climates should be better understood before attempting to interpret the global features of climate change, even though the climate conditions of higher latitude since the last glacial have been well documented. The tropical–subtropical record is crucial in understanding the driving force of climate change from a global point of view (Stock, 1999). Recent studies from the stalagmites of Hulu Cave (China) show that the timing of changes in the monsoon generally agrees with the timing of temperature changes from the Greenland ice core GISP2 (Wang et al., 2001). This indicates that the East Asian monsoon is integral to millennial-scale changes in atmosphere/oceanic circulation patterns and is affected by orbitally induced insolation variations. Remarkable

Corresponding author. Tel.: +886 2 33662932; fax: +886 2 23636095. E-mail address: liewpm@ntu.edu.tw (P.-M. Liew).

(2)

vegetation and climate changes during glacial time in the subtropical island of Taiwan were found in a previous study of Sun-Moon Lake (Tsukada, 1967). According to his study, by excluding Alnus pollen, an early stadial at some time subsequent to 60,000–50,000 BP contains pre-dominantly boreal conifers and pine and a low percentage of temperate elements, indicating a temperature decrease of 8–11 1C; the period from about 50,000 to 10,000 BP is dominated by cool-temperate species. However, there are only four radiocarbon dates in the core and a detailed record of climate change is not available. It is important to obtain a high-resolution pollen record with good age control from the last glacial in Taiwan so as to reveal the synchronism (or not) of global climate events. Thus, the Toushe Basin was chosen for a study of its palynostrati-graphy. We use biomization of fossil pollen assemblages as proposed byPrentice et al. (1992, 1996) and used for the reconstruction of vegetation in Europe, Africa and Asia to discuss the climatic conditions of the stadials and interstadials of the last glacial and Holocene based on the record of surface pollen assemblages of the natural forests nearby in the Salixian area (Jolly et al., 1998;

Tarasov et al., 1998; Yu et al., 1998; Allen et al., 2000;

Takahara et al., 2000; Gotanda et al., 2002). We aim to

describe the vegetational changes quantitatively so as to interpret the magnitude of possible environmental changes. The simpliﬁed pollen diagram of the upper 17 m of this site has been published previously (Liew et al., 1998;

Kuo and Liew, 2000), but data to 39.5 m depth will be

described here.

2. Modern vegetation and climate of the study area Taiwan is a subtropical mountain island whose climate is dominated by the East Asian monsoon. Warm–wet summers and cool-relatively dry winters prevail and the

whole island is generally humid. According to the meteorological data near the study site (Fig. 1), the Sun-Moon Lake Station (altitude 1014 m) immediately north of the Toushe peat bog, mean annual rainfall is 2341 mm, annual evaporation is 1098 mm and mean annual tempera-ture is 19.2 1C. The coldest month has an average temperature of 13.9 1C, whereas the warmest month is 23.6 1C, with an average of 155.6 rainy days. The estimated mean annual temperature of Toushe is 21.2 1C, the lapse rate being 0.54 1C/100 m. The present vegetation surround-ing the study area belongs to the subtropical evergreen Lauro-Fagaceae forest. This forest consists mainly of Machilus kasanoi, M. zuihoensis, Beilschmiedia erythrofolia, Phoebe formosana, Sapium discolor, Michelia formosana, Cyclobalanopsis flauca, Pasania uraiana, P. konishii, P. ternaticupula, P. brebicaudata, Ardisia sieboldii, Zelkova fomosana, Engelhardtia roxburghiana, Glochidion hongkon-gensis, Trema orientalis, Liquidambar formosana, Rhus succedanea, Schefflera octophylla, Castanopsis hystrix, Quercus variabilis, Fraxinus formosana, Lagerstroemia subcostata, Symplocos theophrastaefolia and Sapindus mukorosii among others (Lin et al., 1968).

Tsukada (1967) described the vegetation of mountain

forests above the nearby subtropical forest as follows:

Warm-temperate forest (ca. 500–1800 m): Dominated by Castanopsis, Lithocarpus, Cyclobalanopsis and Cin-namomum with other broadleaved species and with conifers such as Keteleeria and Podocarpus species. The undergrowth is crowded with ferns and mosses.

Cool-temperate forest (ca. 1800–2400 m): Composed of deciduous hardwood species of Cyclobalanopsis, Ulmus, Zelkova, Juglans, Carpinus and others, mixed with conifers including Chamaecyparis. Cyclobalanopsis and the Chamaecyparis species form two separate associa-tions but both groups occupy the misty climate belt.

(3)

Subalpine coniferous (or boreal) forest (ca. 2400–3500 m): Composed of Tsuga chinensis, Abies kawakami and Picea morrisonicola mixed with Pinus armandii. Above about 3300 m, alpine shrubs and herbs occur.

Su (1984) studied the vegetation in the mountains of

central Taiwan and identiﬁed the following altitudinal zones (Fig. 2) with annual temperature range:

(1) Ficus–Machilus Zone (below altitude 500 m; 23–26 1C; tropical): Lowland evergreen broadleaved forest in-cluding species of Ficus and Machilus.

(2) Machilus–Castanopsis Zone (500–1500 m; 17–23 1C; subtropical), submontane evergreen broadleaved forest with two major types: 1. Castanopsis type: mainly composed of Castanopsis hystrix, C. Kawakamii, Schima superiba, Engelhardtia, Lithocarpus and 2. Machilus type: mainly Machilus japonica, M. kusanoi, Ficus, Lagerstroemia and tree fern species.

(3) Lower Quercus Zone (1500–2000 m; 14–17 1C; warm temperate), with major components Cyclobalanopsis longinus, C. gilva, Lithocarpus and Litsea species. (4) Upper Quercus Zone (2000–2500 m; 11–14 1C;

tempe-rate), with major components Cyclobalanopsis morii,

C. stenophylloides, Trochodendron and Castanopsis carlesii.

The Upper Quercus Zone is often mixed with montane mixed coniferous forest including Chamaecyparis, Pinaceae and Taxodiaceae, although they may separate into associations. When local conditions are less humid, montane deciduous broadleaved forest appears in Quercus Zone including species of Acer, Juglans, Ulmus, Carpinus, Platycarya and Quercus. When aridity increases Alnus formosana prevails in this zone although this tree may appear at altitudes between 900 and 2600 m. However, Alnus with Salix, Carpinus and Acer frequently appear near 2000 m, whereas Alnus associated with Urticaceae usually occurs below this altitude. In addition, Pinus exists in the still drier conditions of this zone.

(5) Tsuga–Picea Zone (2500–3100 m; 8–11 1C; cool tempe-rate), with major components Tsuga chinensis, Picea morrisonicola and Pinus armandii mastersiana.

(6) Abies Zone (3100–3600 m; 5–8 1C; cold temperate), mainly Abies kawakami. Beyond 3300 m, alpine shrubs and herbs are scattered.

(7) Alpine vegetation (43600 m; o5 1C; cold), mainly Gramineae with some Juniperus.

(4)

In an effort to elucidate the fossil sequence of Toushe, a preliminary study was made of the surface pollen assemblages of the natural forest in Salixian (231280_–231330_{N; 120153}0_–1201590_{E), 30 km southeast of}

the Toushe site (Liew and Chung, 2001). This area includes the Machilus–Castanopsis Zone, Lower and Upper Quercus Zones and Picea–Tsuga Zone (Fig. 2) ofSu (1984)between altitudes 700 and 2800 m (Chung, 1994). Pollen assem-blages above 2500 m, the Picea–Tsuga Zone, are dominated by Pinus, Tsuga and Picea. Below 2500 m, in the Upper and Lower Quercus Zones, pollen assemblages are more complex. Between 2500 and 2250 m, an Alnus-dominant pollen assemblage appears while below 2250 m Castanop-sis-dominant pollen assemblages are common. Cyclobala-nopsis pollen is expected to prevail in the Quercus Zone if sampling in different parts of this zone is wide enough to cover more associations of this zone. Nevertheless, the Upper Quercus Zone may sometimes also be represented by a Castanopsis-dominant pollen assemblage associated with Cyclobalanopsis and Alnus, even though Cyclobala-nopsis is common in this zone. This is possibly because some of the prevailing genera in the zone belong to the Castanopsis pollen types, such as Castanopsis carlesii. Castanopsis-type pollen includes Castanopsis, Lithocarpus and Pasania, totaling 26 species. However, Castanopsis-dominant pollen assemblages in the Quercus Zone (above 1500 m) can still be differentiated from those of the Castanopsis–Machilus Zone (below 1500 m) if there exist signiﬁcant associated elements in the pollen assemblages. The latter are characterized by Castanopsis associated with subtropical to warm-temperate elements such as Elaeocar-pus, Diospyros and Euphorbiaceae, and abundant spores, which are different from those of the Castanopsis-dominant assemblages at higher altitudes.

3. Methods 3.1. Chronology

The 39.5-m-long core was taken in the center of this small, rectangular basin. The sediment from 39.5 to 32.8 m depth is clay. The upper 32.8 m is mainly peaty sediments intercalated with thin layers of clay or gyttja except for backﬁll in the top 0.3 m. Each sample for analysis was about 1 cm3, with an average sampling interval of 9 cm. The total number of samples is 395, yielding 114 pollen taxa. Thirty-one radiometric analyses were performed, including three AMS and 28 radiometric14C dates. Each sample for conventional14C dating was 6–10 cm in length. The sedimentation rate is relatively constant, as shown in

Fig. 3. Ages of sediments beyond the limit of 14C

radiometric dating are extrapolated by tentatively correlat-ing the Alnus-dominant interval with MIS 4 (MIS 3/4, 23.3 m and MIS 4/5a, 26.5 m) (Martinson et al., 1987). This high Alnus-interval, representing cold/dry conditions, is most likely to correspond to MIS 4. It contrasts with the layer beneath where less cold and less dry elements, i.e.

Castanopsis or Cyperaceae, are still signiﬁcant. The bottom of this core at 39.5 m is estimated as 95.2 kyr BP by the least-squares equation shown in the ﬁgure, and the ages of depths between 39.5 and 26.5 m are then extrapolated. 3.2. Vegetation reconstruction

Yu et al. (2002) transferred the surface pollen

assem-blages of Salixian (Liew and Chung, 2001) into modern vegetational types by the biomization technique (Prentice

et al., 1996) by which vegetation can be described by

biomes and major vegetation categories deﬁned on the basis of their dominant plant functional types (PFTs). However, the ecological reference is not as complete at that time as it is at present and the results show overlaps between the upper boundary of subtropical/warm tempe-rate and also between warm-tempetempe-rate/tempetempe-rate vegeta-tion types. In this study, we slightly modiﬁed (Tables 2 and 3) the previous work of Yu et al. (2002)with reference to the altitudinal vegetational zones ofSu (1984)(Fig. 2) and also to newly available ecological data (Editorial

Commit-tee of the Flora of Taiwan, 1993, 1994, 1996, 1998, 2000).

For example, PFT sut are those broadleaved evergreen plants that appear in subtropical evergreen forest (Machi-lus–Castanopsis Zone of Su; 500–1500 m in altitude central Taiwan). We attempted to name biomes by following the aforementioned altitudinal zones proposed by Su (1984). Thus, the tropical evergreen forest corresponds to the Ficus–Machilus Zone (o500 m; 423 1C) and subtropical evergreen forest to the Machilus–Castanopsis Zone (500–1500 m; 17–23 1C); warm-temperate evergreen forest to the Lower Quercus Zone (1500–2000 m; 14–17 1C) and temperate broadleaved and conifer mixed forests (com-prised of the montane evergreen forest, montane mixed coniferous forest; montane deciduous forest) to the Upper Quercus Zone (2000–2500 m; 11–14 1C); cool-temperate coniferous forest to the Tsuga–Picea Zone (2500–3100 m; 8–11 1C). We also included in the biome list the tropical rain forest occurring in southern but not in central Taiwan, and for the biomes absent in Taiwan we followedYu et al.

(2002), and Editorial Committee of the Vegetation of

China (1980)as a reference to forest steppe.

A test of modern forest reconstruction is carried out by using the same set of surface pollen assemblages in Salixian that was used byYu et al. (2000). We apply two strategies when assigning Castanopsis-type pollen in PFTs: (1) include Castanopsis in PFTs sut and wte2, i.e. in subtropical and warm-temperate forests since Castanopsis is the main forest element of subtropical and warm-temperate forests; (2) include Castanopsis in PFTs sub, wte2 and wte1, i.e. in subtropical, warm temperate and also temperate broadleaved and conifer mixed forests, since four of 26 species of Castanopsis-type pollen grow in temperate broadleaved and conifer mixed forests. Among the 16 sites between altitudes 700 and 2800 m in the Salixian area, the result of strategy (1) shows two sites incorrectly assigned (altitudes 2000 and 2150 m: temperate

(5)

broadleaved and conifer mixed forests are incorrectly assigned to warm temperate evergreen forest). The result of strategy (2) shows three sites incorrectly assigned (altitudes 1500 m: warm-temperate forest incorrectly

as-signed to temperate broadleaved and conifer mixed forest; 1780 m: warm-temperate forest incorrectly assigned to subtropical forest; 2150 m: temperate broadleaved and conifer mixed forests incorrectly assigned to subtropical

Fig. 3. Ages and information used to establish age–depth relations at Toushe bog. Radiocarbon yr were converted to calendar yr (kyr) by use of refsBard et al. (1998)andStuiver and Reimer (1993).

(6)

forest). In Japan, Gotanda et al. (2002), following

Takahara et al. (2000), studied the reconstruction of forests

by surface pollen samples. They improved the reconstruc-tion of present forests by assigning Fagus to temperate deciduous forest instead of several forest types and excluding Gramineae. For the biomization of the Toushe fossil pollen sequence, we use strategy (1) based on the better results of the modern forest reconstruction in Salixian, by assigning Castanopsis in subtropical and warm-temperate forests (Tables 2 and 3).

4. Palynostratigraphy of the Toushe bog 4.1. Palynological zones

The pollen zones are described as follows and also in

Table 1. The percentage of each pollen genus is based on

the total pollen sum, whereas that of spores is based on the sum of pollen and spores (Figs. 4.1 and 4.2). In the following discussion, years (e.g. 2.9 kyr BP) represent calibrated years, and14C years are expressed as 2900 yr BP. Zone 20 (39.5–39.0 m, tentatively estimated as 95.2– 94.4 kyr BP): arboreal pollen (AP) is about 70% in which Castanopsis is dominant (460%). Other associated ele-ments are Cyclobalanopsis, Quercus and Symplocos.

Zone 19 (39.0–37.2 m, tentatively estimated as 94.4– 91.4 kyr BP): AP still prevails but Cyperaceae and Grami-neae continuously increase. Among the AP, Castanopsis and Salix are the most important. Associated elements are Quercus and Cyclobalanopsis. Compared with the previous zone, Salix increases to more than 20% at the expense of Castanopsis. Liquidambar and Ilex values also increase.

Zone 18 (37.2–35.6 m, tentatively estimated as 91.4– 88.8 kyr BP): is characterized by an abrupt increase of Pteridophytes (460%). AP is dominant, Symplocos (450%) and Myrica (25%) being the most common ones. Zone 17 (35.6–30.5 m, tentatively estimated as 88.8– 80.5 kyr BP): AP reaches 60%, slightly less than in the previous zones in which the major element Castanopsis (20–40%) again reaches its dominant situation. Myrica reaches 15% in the lower part. Cyperaceae increase upward to 30%.

Zone 16 (30.5–29 m, tentatively estimated as 80.5– 78.0 kyr BP): is characterized by a sharp decrease of Castanopsis, from 40% to o15%, and herbs outnumber arboreal pollen for the ﬁrst time. Cyperaceae increase to 40% or more, whereas Salix increases remarkably in the upper part at the expense of Cyperaceae.

Zone 15 (29–28.2 m, tentatively estimated as 78.0– 76.7 kyr BP): exhibits remarkable changes in both AP and NAP (non-arboreal pollen). AP dominates in this zone (80%) but instead of Castanopsis, the major tree pollen are Ilex (40%) and Alnus (up to 70%). Alnus reaches its climax for the ﬁrst time in this sequence. Castanopsis is at about 5%, even less than in Zone 16, and is lower than

Cyclobalanopsis. Cyperaceae decrease sharply, while the Artemisia content rises slightly.

Zone 14 (28.2–26.5 m, tentatively estimated as 76.7– 73.9 kyr BP): Cyperaceae dominate in the lower part. Alnus sharply decreases while Cyclobalanopsis increases. In the upper part Alnus reaches 40%. Cyclobalanopsis (about 15%) is higher than Castanopsis and maintains this trend during the glacial. Artemisia is relatively higher in Zones 14 and 15.

Zone 13 (26.5–23.3 m, tentatively assigned as 73.9– 59.0 kyr BP): AP dominates (480%) with Alnus as the major component (generally 60% or more). Salix increases while Cyclobalanopsis decreases in comparison with the previous zone. Cyperaceae decrease as well. Pinus decreases from Zone 15 upward.

Zone 12 (23.3–23 m), tentatively estimated as 59.0– 57.6 kyr BP: is characterized by a decrease of Alnus (20%) and Salix and the increase of Ilex (from o5% to 425%).

Zone 11 (23.0–22.4 m, tentatively estimated as 57.6– 54.0 kyr BP): Alnus predominates again (60%). Compared with Zone 13, Castanopsis and Symplocos increase at the expense of Salix. Symplocos occasionally increases in Zones 11 and 12.

Zone 10 (22.4–21.0 m, tentatively estimated as 54.0–48.5 kyr BP): Alnus decreases remarkably to about 40% or even less, although it is still the main component of AP. Cyclobalanopsis and Castanopsis are relatively sig-niﬁcant, similar to the last two zones 12 and 11. Ilex and Symplocos decrease. Cyperaceae dramatically increase again. Castanopsis is relatively higher from Zones 12 to 10, after its decrease to a very low value in Zone 13.

Zone 9 (21.0–18.4 m, tentatively estimated as 48.5– 36.9 kyr BP): Alnus has the same amount as in Zone 10 but peaks at the middle. Cyperaceae frequently ﬂuctuate at the expense of Alnus. The most notable feature is the increase of Pteridophytes, especially monolete spores. Salix increases episodically in this zone. Castanopsis decreases slightly and maintains the trend from this zone upward until the beginning of the Holocene. Ilex is common from Zone 12 upward.

Zone 8 (18.4–16 m, cal. 36.9–31.2 kyr BP; 14C years, 41,100–31,600 yr BP): Pteridophytes and Cyperaceae de-crease sharply. Compared with the previous zone Alnus and Ilex increase while Cyclobalanopsis decreases. Grami-neae increase slightly.

Zone 7 (16–14.6 m, 31.2–27.9 kyr BP; 31,600–23,900 yr BP): Cyclobalanopsis slightly increases. Alnus is still dominant with a level similar to the previous zone. Ilex and Symplocos decrease. Castanopsis increases slightly. Alnus is relatively low but with remarkable ﬂuctuations from Zones 10 to 7.

Zone 6 (14.6–13.8 m, 27.9–27.4 kyr BP; 23,900–23,400 yr BP): compared with Zone 7, Alnus increases clearly. Cyclobalanopsis, Castanopsis, Quercus, Ligustrum and Salix decrease. Artemisia decreases slightly. Zones 6–12 are assumed to belong to MIS 3.

(7)

Table 1 Tous he bog po llen zone s with infe rred paleo vegetati on and paleo climat e Zone depth and age OIS and age C haracte ristics Pa leovegetat ion Paleo environm ent 1b (1.4–0 .3 m) 3.8–1 .8 kyr BP 1 H erbace ous pollen rise (70–80%). Inc rease in Liq uidambar ; decre ase in C yclobala nops is , Ilex . Cyperac eae and Gra minea e fr equent. Increa se in pter idophyt es Gra ssland Subtro pical to w arm tem perate, wet 1a (2.5–1 .4 m) 5.7–3 .8 kyr BP 0–12 .1k yr He rbace ous pollen increa se. Increa se in Salix , Symplo cos , Ilex ; decrease in M allotus . Increa se in Cype raceae Su btropical forest bu t flu ctuated Fluctu ated subtro pical to tem perate 2c (3.1–2 .5 m) 6.3–5 .7 kyr BP Inc rease in Ilex , Lig ustrum ; decre ase in Pin us. Decrease in Cyperac eae and pterido phytes Su btropical to tropica l fore st Subtro pical, less wet 2b (3.9–3 .1 m) 7.0–6 .3 kyr BP Inc rease in Gramin eae and C yperacea e, Salix peak s a t first and then de crease . Increa se in Pin us . Increa se in pter idophyte s (to 50% ) Su btropical forest mai nly Subtro pical, w et 2a (4.7–3 .9 m) 8.0–7 .0 kyr BP W oody taxa sum increase slightly. Increa se in Ilex and Mallotus , Salix pe ak at early par t. Sh arp decre ase in pter idophyte s and Cyperac eae Su btropical forest bu t flu ctuated Fluctu ated subtro pical to tem perate less wet 3 (7.10– 4.7 m) 11.0– 8.0 kyr BP W oody taxa sum decre ases (to 40%) . Increa se in Cast anopsis , Tre ma , Pin us ; decre ase in Cyclobala nops is and Symplo cos. Salix varies; peak of Ilex at middle. Increa se in Cyperac eae and Gramin eae, sharp increa se in pter idophyt es Su btropical to warm-tem perate fore st Subtro pical to w arm tem perate, wet 4d (7.6–7 .10 m) 11.9– 11.0 kyr BP Inc rease in Cyclo balanops is , Sy mplocos and Sal ix ; inc rease in Gra minea e (to 40% ) Te mperat e to subtro pical fore st Temp erate, less dry 4c (9.2–7 .6 m) 14.5– 11.9 kyr BP W oody taxa A 85%. Inc rease in Salix , Symplo cos . Decrease in C yperacea e and A rtemisia . Gra mineae is low but pe ak at end. De crease in pter idophyte s Te mperat e forest Temp erate, less dry 4b (9.8–9 .2 m) 15.7– 14.5 kyr BP 2 W oody taxa sum varie s, increa se in A lnus . Increase in Cyperac eae. Sh arp inc rease in pterido phytes W arm-temperat e fore st Warm tem perate, moist to wet 4a (10.3– 9.8 m) 16.7– 15.7 kyr BP 12.1– 24.1 kyr W oody taxa rise (to 70% ), decre ase in Alnus ; inc rease in Ilex , Sy mplo cos , Cyclo balan opsis . Decrease in Cyperac eae and Gra minea e Te mperat e forest Temp erate less dry 5 (13.8– 10.3 m) 27.4– 16.7 kyr BP De crease of woody ta xa pe rcentag es, he rbs rise (to 75% ) o f pollen sum , Gra minea e rise (to 50%) . Decre ase in Alnus (to 20%); increase in C yclobala nops is , Ilex, Sy mplocos . Increa se in A rtemis ia (to 10% ), C yperacea e Te mperat e forest to forest st eppe Temp erate, dry 6 (14.6– 13.8 m) 27.9– 27.4 kyr BP W oody taxa A 80%; inc rease in Alnus ; decre ase in Cyclo balan opsis , C astano psis , Que rcus , Lig ustrum and Sa lix . Decre ase in Cyperacea e and A rtemis ia Te mperat e deciduou s fore st Temp erate 7 (16–1 4.6 m) 31.2– 27.9 kyr BP W oody taxa sum varie s 60–90% , A lnus fluctuat ed at the compen se of C yperacea e. Increa se in C yclobala nops is and Castano psis . Increa se in Sa lix at the early par t; decre ase in Ilex and Symplo cos , peak in Gra minea e a t end Te mperat e deciduou s fore st Temp erate 8 (18.1– 16 m) 36.9– 31.2 kyr BP 3 W oody taxa sum increa se A 80%, A lnus varies and pe ak at middle; increase in Ilex ; Cyclobala nops is sligh tly decre ase; de crease in C yperacea e and increa se in Gra mineae. pteridop hyte s decre ase sharply Te mperat e deciduou s fore st Temp erate, gettin g drier 9 (21.0– 18.4 m) 48.5– 36.9 kyr BP 27.6– 58.9 kyr W oody taxa sum fluctu ates 40–8 5%, de crease in Castano psis . Flu ctuation of Alnus at the comp ense of Cyperac eae. Increa se in Sy mplo cos ; two peaks of Sa lix at middle and so does Ilex ; increa se in pter idophyt es (to 60%) Te mperat e deciduou s fore st Temp erate, mo ist to wet 10 (22.4– 21.0 m) 54.0– 48.5 kyr BP W oody taxa sum decre ases and then increa ses. Increase in C yclobala nops is and Cast anopsis . D ecrease in A lnus (to 40% ), Ilex , Sy mplo cos ; Increa se in Cyperac eae Te mperat e deciduou s fore st Temp erate, less dry 11 (23.0– 22.4 m) 57.6– 54.0 kyr BP W oody taxa A 80% , inc rease in Al nus (to 60% ); Castano psis the sam e a s the previo us zone. De crease in Ilex and Sal ix . Inc rease in Gramin eae Te mperat e deciduou s fore st Temp erate, dry

(8)

12 (23.4– 23.0 m) 59.0– 57.6 kyr BP 4 Inc rease in Ilex (to 25% ), C astano psis , Symplo cos . D ecrease in A lnus (to 20% ) and Salix Te mperat e deciduou s fore st Temp erate 13 (26.6– 23.4 m) assig ned as 73.9– 59.0 kyr BP 58.9– 73.9 kyr W oody taxa A 80% , mai nly Alnus (to 60%) , increa se in Salix ; 1 peak in Sy mplo cos ; d ecrease in C yclobala nops is . Decrease in Cyperac eae Te mperat e deciduou s fore st Temp erate, dry 14 (28.2– 26.6 m) 76.7– 73.9 kyr BP W oody taxa sum varie s 30–70% , A lnus decline but rises at in the later par t (to 40%), Cyclobala nops is (15% ) highe r than C astano psis ; d ecrease in Sal ix , inc rease in Cyperac eae (to 50% ), Artemisia similar to the pre vious zone Te mperat e deciduou s fore st Temp erate, less dry 15 (29–2 8.2 m) 78.0– 76.7 kyr BP 5a? W oody taxa A 80% ; d ecline in C astano psis (to 5%) and Salix . Increa se in Ilex (40% ) and then Alnus (70% ), decre ase in Pin us , increa se in A rtemis ia , decrease in Cyperac eae Te mperat e deciduou s fore st Temp erate, gettin g dry 16 (30.5– 29 m) 80.5– 78.0 kyr BP 73.9– 85.1 kyr W oody taxa sum varies A 40% , declin e in C astano psis (15% ). Increa se in Sal ix (to 40%) . Increa se in Cype raceae (to 40%) and Gra minea e Te mperat e deciduou s fore st Temp erate, mo ist 17 (35.6– 30.5 m) 88.8– 80.5 kyr BP W oody taxa A 60% bu t u p to 80% at lower, Cast anopsis varie s (2 0–40%). Myric a decre ases upward . Inc rease in C yperacea e (to 30%) , de crease in pteridop hytes (10%) Su btropical to warm-tem perate fore st, less closed Subtro pical to w arm tem perate, less wet 18 (37.2– 35.6 m) 91.4– 88.8 kyr BP 5b? W oody taxa sum varie s 50–90% , d ecline in Castano psis (o 15%) , increase in Symplo cos (to 50% ) and M yrica (25% ). Increase in Po lygon um and pterido phytes (to 80% ) Tro pical to subtro pical fore st Subtro pical, w et 19 (39.0– 37.2 m) 94.4– 91.4 kyr BP 85.1– 93.6 kyr W oody taxa sum 60–8 5%, decre ase in Castano psis , increa se in Salix (to 20% ) and also Ilex , Liquida mbar and Ligustrum . Inc rease in Gramineae and Cyperac eae Te mperat e forest Temp erate, mo ist 20 (39.0– 39.5 m) 95.2– 94.4 kyr BP 5c? 4 93.8 kyr W oody taxa A 70%; do minant taxa is Cast anopsis (to 60%). C yclobala nops is , Quercu s, Symplo cos and Pin us also sign iﬁcan t; C yperacea e abou t 20% , pterido phytes are low W arm-temperat e fore st Warm -temperate, moist

(9)

(10)

0 100 200 300 400 500 600 700 800 900 1000 1100 1200 1300 1400 1500 1600 1700 1800 1900 2000 2100 2200 2300 2400 2500 2600 2700 2800 2900 3000 3100 3200 3300 3400 3500 3600 3700 3800 3900 4000 Depth (cm) 1840 ± 50 2230 ± 50 4230 ± 50 4410 ± 40 4800 ± 50 5640 ± 60 6480 ± 60 7370 ± 60 8780 ± 60 9600 ± 130 10059 ± 87 10309 ± 89 12100 ± 90 12350 ± 90 13700 ± 100 13900 ± 100 15200 ± 110 18150 ± 120 20500 ± 100 23400 ± 190 23900 ± 150 28000 ± 250 29300 ± 300 31100 ± 400 31600 ± 350 32800 ± 500 37200 ± 600 39200 ± 600 8270 ± 70 > 43000 > 41000 > 43000 > 40000 > 45000 > 41000 Age (C-14 yr BP) 20 Artem isia 20 Poly gonum Um belliferae 20 40 60 80 Cyp eraceae 20 40 60 80 Gram ineae 20 40 20 40 60 80 20 40 60 80 100 Zone 1b 1a 2c 2b 2a 3 4d 4c 4b 4a 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9

Subtropical evergreen forest Warm-temperate forest

Tropical evergreen forest

Temperature deciduous and conifer mixed forest Monolete spore other Trilete spore Potam ogeto naceae DCAaxis-1 Biomes Lygodium (%)

Fig. 4.2. Percentage of each spore genus is based on sum of pollen and spore. Reconstructed biomes plotted stratigraphically in relation to sample scores on the 1st axis of a detrended correspondence analysis of the pollen data.

(11)

Zone 5 (13.8–10.3 m, 27.4–16.7 kyr BP; 23,400–13,800 yr BP): NAP prevails in the middle and upper parts of this zone in which Gramineae (up to 40% once in the upper part) as well as Artemisia (at about 10%) reach their highest values. Alnus decreases remarkably to less than 20% and Cyclobalanopsis increases slightly. Symplocos is higher than in the previous zone but decreases gradually upward in the upper part of this zone. This zone belongs to MIS 2.

Zone 4 (10.3–7.10 m, 16.7–11.0 kyr BP; 13,800–9700 yr BP): differs from the previous zone remarkably. The forest elements rise again and NAP falls. AP increases to 80% in the upper part. Cyclobalanopsis together with Ilex, Symplocos and Salix are the main woody elements. Gramineae decrease. Due to the abrupt change of assemblages, the zone is divided into three subzones. In subzone 4a (10.3–9.8 m) Cyperaceae sharply decrease. Ilex, Cyclobalanopsis and Symplocos increase. NAP is clearly lower. In subzone 4b (9.8–9.2 m) spores increase abruptly and Cyperaceae also increase. Alnus is higher and Ilex is lower than in the previous subzone. Cyclobalanopsis maintains the same level as previously. In subzone 4c (9.2–7.6 m) spores decrease remarkably. Symplocos drama-tically increases in the lower part and Salix in the upper part. Ilex is common. Herbs are relatively low. In subzone 4d (7.6–7.1 m) Gramineae rise to 40%. Salix reaches its highest value in the upper part.

Zone 3 (7.10–4.70 m, 11.0–8.0 kyr BP; 9700–7200 yr BP): Pteridophytes are high. Castanopsis is dominant among AP. Trema increases. Salix, Symplocos and Ilex are reduced. NAP increases again. It is probably hydrophyllus because of the accompanying large amount of fern-spores. AP is higher than NAP.

Zone 2 (4.70–2.50 m, 8.0–5.7 kyr BP; 7200–4900 yr BP): is characterized by the increase of Ilex and Mallotus. Pteridophytes reach their maximum value in the middle part. Three subzones are distinguished here: In sub-zone 2a, Salix, Ligustrum (12–15%) and Ilex (25%) increase again. Cyclobalanopsis decreases. Cyperaceae are reduced. Pteridophytes decrease remarkably. Mallotus appears from the middle part of this zone upward. In subzone 2b Pteridophytes increase to maximum (ave-rage in 50%). Ilex decreases while Pinus increases. In subzone 2c, Pinus is lower while Symplocos and Ilex rise again. Herbs increase again. Ligustrum and Symplocos also increase.

Zone 1 (2.50–0.30 m, 5.7–1.8 kyr BP; 4900–1800 yr BP): is characterized by the increase of Cyperaceae. Salix increases but Ilex decreases. Herbs reach their maxi-mum values. Cyperaceae and monolete spores become important. This zone is further divided into two subzones due to the changing amounts of NAP. In subzone 1a (2.5–1.4 m) Pinus is lower while Symplocos and Ilex rise again. Salix increases but monolete spores decrease. Cyperaceae also increase upward. In subzone 1b (1.4 m upward) Cyperaceae reach their maximum. Monolete spores increase as well.

4.2. Paleovegetation: a quantitative reconstruction

The characteristics of vegetation of pollen spectra from Toushe Basin appear in Fig. 4.2 according to PFTs and biomes shown inTables 2 and 3with the biomes plotted on the 1st axis of detrended correspondence analysis (DCA by Tilia; Grimm, 1997). The results show that subtropical evergreen forest dominated during the Holocene, in contrast to temperate broadleaved and coniferous mixed forests, speciﬁcally the temperate deciduous forest within them, during the last glacial.

Vegetation reconstruction shows that the Castanopsis-dominant pollen assemblages of the early glacial are mainly subtropical to warm-temperate evergreen forests except Zone 19. The warmest conditions are in Zone 18 (tropical and subtropical forests) and the lower part of Zone 17 (subtropical to warm-temperate forests). Thus, Zone 19, where temperate broadleaved and conifer mixed forests prevailed, might correspond to MIS 5b. From Zones 16 to 4, the temperate broadleaved and conifer mixed forests, especially the deciduous forest, dominated except for some intervals of the last glacial at cal. 41.6, 38.0 and 37.3 kyr BP and at 22.3 and 18.9 kyr BP as well as 15.1 kyr BP where warm-temperate evergreen forest or subtropical forest appeared. It shows a distinct interstadial at about 42–37 kyr BP. The warm conditions occurred at 22.3 and 18.9 kyr BP just before and within the dry phase (Grami-neae prevail) of the late stadial. However, due to high amounts of Gramineae in these samples, this result needs to be further discussed. Nevertheless, an abrupt warm phase around 22 ka is reported in the Siple Coast of Antarctica (Taylor et al., 2004) and a warm phase very near Last Glacial Maximum (LGM) is also reported in New Zealand (Hormes et al., 1999). Post-Bølling subtropical conditions are found at 11.5 kyr BP. Thus, the subtropical conditions of 15.1 kyr BP (depth 9.59 m) and 11.5 kyr BP (depth 7.28 m) of this study correspond to the spikes of Atlantic melting events (Bond et al., 1992). After 10.7 kyr BP, subtropical conditions continue. Tropical forest appeared at 6.9 and 6.1–5.9 kyr BP. Temperate broadleaved and conifer mixed forest appeared at about 11.2–11.0, 7.5, 7.2 and 7.1, 5.2, 5.0 and 4.9 kyr BP. The cool interval at 3.7–2.0 kyr BP which was found in alpine lakes of Taiwan (Liew and Huang, 1994) appeared here only as a change from subtropical to warm-temperate forests. It indicates that it is a less prominent cooling than those in the ﬁrst half of the Holocene. Subtropical and tropical forests are frequent from 8 to 5 kyr BP. Warm-temperate evergreen forest is common in early Holocene before 8 kyr BP.

However, the result of biomization shows that samples with equal score of warm-temperate evergreen forest and subtropical evergreen forest are common although they should be assigned as warm-temperate forest according to the rule of less number of PFTs in biomization (among the samples assigned as warm-temperate forest are almost with the same score as subtropical forest except at depths 31.1,

(12)

Table 2 Assig nment of pollen ta xa to plant fun ctional types (PFT) PFT co de Pollen taxa te Ficus , Litho carpus , Bischo ffia , Hom alanth us , Pa lmae , Pipera ceae , Trema , Albizz ia , A cacia , C eltis , Sche fflera , Ardisia , Myrica , Elaeoc arpus , Sloane a , M allotus , Oma lanthu s, Macarang a , Melan olepis , Gl ochidian , Daph niphyllum , C itrus , Zantho xylu m , Chlora nthus , Sapium , Miche lia , Melia , Algaia , Syzy gium , Helicia , Meliosma , Sty rax , W endland ia , Myrsin e, Viburn um , Symplo cos , Lig ustrum , Saurau ia , Ilex , Kandelia , R hizopho ra , Lumnit zera , Dyso xyllum , A ralia , Fraxinus , Osma nthus , C allicarpa , Cyca s, La gerstroemia , Maes a , Diosp yros , Rutace ae, Combr etaceae , Mora ceae, Legumin osae, Prote aceae , Melas tomaceae , Ulm aceae tr Diospyros , Dyso xylu m , Aglaia , Albizz ia , Termin alia , Ficus , Syzy gium , Bombac aceae , Bombax , Combr etaceae , Euphor biaceae , Legum inosae , Melan olep is , Micro tropis , M yrsin aceae, Prote aceae, Sapotace ae, Ulmaceae, Ulmu s, Wend landia sut Cyclobala nops is , Cast anopsis , Litho carpus , Pasania , Cryptoc arya , Schef flera , Ardisia , Myric a , Ela eocarp us , Sloanea , Pla tycar ya , Mallotus , Daph niphy llum , Michelia , M eliosma , Ilex , Diospyros , En gelhardti a , Myrsin e, Schima , M aesa , Fr axinus , Lig ustrum , Osma nthus , Prunus , Callicarpa , Actinid ia , Saurau ia , Capap aris , Anac ardia ceae , Araliac eae , Citrus , Zelkov a , Chlora nthus , Came llia , Helicia , Symplo cos , Syzy gium , Celtis , Zantho xylu m , Sapium , Viburn um , Cyca s, M ora ceae, Rhamnac eae, Rub iaceae, Rutace ae, Eb enaceae , M eliaceae, Apoc ynaceae , Eupho rbia ceae, Lagers troemia wte2 Cyclobala nops is , Castano psis , Litho carpus , Quercu s, Viburn um , Daph niphyl lum , Schim a , A ralia , M yrsine , Ligustrum , Osma nthus , Ela eocarp us , Tro choden dron , Actinidia , Cleyera , Eurya , Gordo nia , Platyca rya , Dendrop anax , Illicium , Ilex , Meliosma , schfﬂe ra Magnol iaceae , Aalia ceae, Hydra ngeac eae, Sy mplocos wte1 Ilex , Lig ustrum , Cyclo balanops is , Que rcus , Sh ortia , Viburn um , Trocho dendron , Symplo cos , Prunus , A ctinid ia ts1 Alnus , Ribes , Ulmac eae, Sa lix ts2 Carya , Alnus , Ela eognus , Fagus , Jugla ns , Liquida mbar , Pte rocarya , Que rcus , Sa lix , C astanea , C arpinus , Z elkova , Ulmus , Helwingia , Ulm aceae ts3 R hus , Z elkova , La gerst roemia , Sapind us , Platyca rya , Sapium , Taxillus , Celtis , Ulmu s, Liq uidam bar , Tre ma , K oelreute ria , Te rminali a , A lbizia , Helwingia , Ebenac eae. Diosp yros , Fr axinus ts Acanth opanax , Ace r, Viburn um , Alnus , Ulmus Ruta ceae, Rosace ae, Rha mnac eae ec Pinus , Junip erus wtc Keteleeria , Po docarpu s, Caloc edrus , Ceph alotax is tc Chama cyparis , Crypt omeria , Cunnin gham ia , Taiwa nia , Pinus , T a xodiac eae ctc Picea , Tsuga bec Abies tef Impati ens , M icrolepia , Cusc uta , Zant hoxylum , Rumax , Chenopo dium , Amara nthus , C lematis , Ranuncu lus , Cocc ulus , Nymphae a , Hydran gia , Chlora nthus , Hyp ericum , Dros era , Arabis , Kalancho e, Sedu m , Deutzia , Pittosp orum , Rosa , Rubus , Sangu isorb a , R hamnus , C issus , Ela eognus , Begon ia , Actino stemm a , Melas toma , Bredia , Epilobium , Lu dwigia , Trapa , Myriop hyllu m , Helwingia , Sche fflera , Tetrapa nax , Sh ortia , M aesa , Ardisia , Myrsin e, Ly simachia , Styrax , Symplo cos , Gentia na , Galium , C allicarpa , Callitriche , So lanum , Justic ia , Stro bilan thus , Plantago , Viburn um , Sagittaria , Po tamog eton , Musa , Bryoph yllum , Ty pha , Stautonia , Pip er , Actinid ia , Loranthu s, Mi crotropis , W irkstroem ia , A ca nthaceae , G esne riaceae. Gramineae tf Actino stema , Gentia na , Cuscut a , Galium , Polygon um , Sta utonia , Hypericum , A rabis , Sedum , Deutzia , Hyd rangia , Pittosp orum , Rosa , R ubus , Impa tiens , Eu onymus , Rhamn us , Ela eognus , Epilobium , Short ia , Lysimac hia , Primula , Symplo cos , Swert ia , Plant ago , Lo nicera , Clema tis , Lo ranthu s, Micro tropis , Cuc ubitaceae , Gra minea e, La biata e, Ran unculac eae, Ruta ce ae Scroph ulariaceae, So lanaceae , Umbe liferae af Polygon um , Rhodo dendron , Th alictrum , Salix , Actinidia , Juniperu s, Chen opod ium , Clemat is , Ranun culus , Dam nac anthu s, Eurya , Galium , R osa , Lonicera , Sedum , C rucifera e, Ericaceae, Composit ae . B erberis , R ibes sf Artemisia , Justic ia , Rubus g Gramin eae, Compo sitae h Ericac eae s Cyperac eae wod Symplo cos , Que rcus , Ilex , Cyclo balan opsis Fern s (x) Psilotum , Ly copodiu m , Selag inella , Equise tum , Ophi oglos sum , Angiop teris , Archangio pteris , Maratt ia , Osmu nda , Ly godiu m , Schizae a , Dicranopteris , Hemenoph yllum , Cibotium , Cyathe a , Dennstae dtia , M icrolepia , Pteridium , Dav allia , Vitta ria , Woodwa rdia , Dryo pteris , Dipte ris , Pol ypodium , Azolla , Ctenit is , Dipla ziun , Pla giogy ria , Microsorium , Selag inella trx Cyath eaceae Key to PFT code: te, tropica l evergre en; tr, tr opica l rain green; sut, sub tropica l evergre en; wte2, w arm-temperate evergreen ; wte1, tempe rate ever gre en; ts1, co ol-temp erate summ ergreen; ts2, inte rmediate-tem perate summ ergree n; ts3, sout hern warm-tempe rate summ ergreen; ts , tempe rate summ ergree n; ec, euryt hermic conifer; wtc, sout h warm-temperat e conifer; tc, tem perate conife r; ctc, cool-t emperate evergre en conifer; bec, boreal evergre en conifer; tef, tropica l and subt ropica l evergre en forbs/sh rubs; tf, tem perate forb s/shrub s; af, arctic -alpine forb s/shr ubs; sf , st eppe forb/shr ub; g, grass ; h , Ericac eae; s, sedge ; wod , woodla nd; x, fern s; trx, Cyat heaceae .

(13)

35.4, 39.1 and 39.2 m where warm-temperate forest has the higher score). It possibly indicates that further discrimina-tion of the altitudinal zones within the Lauro-Fagaceae forest using genus level data is not very easy. This is partly due to the fact that the main pollen types of forests lower than the Tsuga–Picea Zone are the Fagaceae—Castanopsis and Cyclobalanopsis—whose altitudinal spread of species is wide.

The other problem in biomization here is the difﬁculty when Gramineae-dominant assemblages are encountered, such as during LGM, due to the wide occurrence of Gramineae in these tropical–subtropical areas. Conse-quently, even though the Gramineae-dominated assem-blages of LGM possibly represent forest steppe or steppe conditions, the biome of temperate broadleaved and conifer mixed forests occurred. We hope more work in the future will further reﬁne the rule set of the biomes.

5. Interpretation of vegetational history and climatic conditions

In the pollen diagram, warm elements, such as Castanopsis and Mallotus, alternated with cold elements, such as Alnus and Salix. Higher spore values in the diagram are regarded as a proxy of higher precipitation conditions or water-transport processes. Precipitation is usually an index of an intensifying summer monsoon, which in turn represents more frequent tropical cyclones. Relatively dry conditions are represented by high values of Gramineae (i.e. Gramineae without accompanying large amounts of monolete spores), Alnus and Salix. According to modern vegetation assem-blages (Fig. 2) Alnus and Salix grow in less humid parts of the Quercus Zone even though they are well known as temperate, boreal or arctic-alpine trees and shrubs indicating wet conditions in other places (Tarasov et al., 1998).

5.1. Zones 20–17, tentatively estimated as 95.2–80.5 kyr BP, probably MIS 5c–a

The early glacial (Zones 20–17) is characterized by assemblages with a high value of Castanopsis-type pollen (20–40%) except Zone 18. Castanopsis-dominant pollen

assemblages usually represent assemblages of subtropical to warm-temperate forest—Machilus–Castanopsis Zone (500–1500 m) in present Taiwan, as shown in pollen records studied earlier (Liew, 1977). However, Castanopsis-domi-nant assemblages of the early glacial have much lower spore and much higher Cyperaceae content if compared with the present Machilus–Castanopsis Zone. They may indicate a climatic condition less humid than that of today’s Machilus–Castanopsis forest. Alternatively, if those Castanopsis-type pollen during early glacial are the few species growing higher than 2000 m, they may represent a forest within today’s Upper Quercus Zone. The increase of Salix in Zone 19 indicates a relatively cold trend although still within the temperate broadleaved and conifer mixed forests (the Upper Quercus Zone), which is tentatively assigned to MIS 5b. After this, a remarkable ﬂuctuation from dry to wet is shown in Zone 18 in which Symplocos and Myrica replace the important role of Castanopsis then followed by high peaks of monolete spores and Polygonum (91.4–88.8 kyr BP estimated by the age model here). Myrica presently grows in the low altitude (below 1500 m) area of Taiwan. Whether this wet/warm phase corresponds to the wet event at 88 ka in the loess record of China (Rousseau

et al., 2000) needs further study. Castanopsis returns to its

dominant role after this warm–wet phase. Another small wet/dry ﬂuctuation in the lower part of Zone 17 is indicated by various amounts of Cyperaceae. Higher Cyperaceae contents might indicate lower lake levels

(Maley and Brenac, 1998). The drought trend in the upper

part of Zone 17 is also witnessed by changes of lithological facies from lake clay to peat. Zones 18 and 17 are assumed to be within MIS 5a. The following zones 16–14 represent the transition from warm to cold conditions.

5.2. Zones 16–14, tentatively estimated as 80.5–73.9 kyr BP The interval begins with the remarkable decrease of Castanopsis and increase of Salix, indicating less humid and less warm conditions than before. Fluctuations of cold/dry to warm/wet conditions from Zones 15 to 14 are shown by the successive dominance of Ilex, Alnus, the low value of Cyperaceae in Zone 15 and the increase of Cyclobalanopsis and Cyperaceae in Zone 14, although still within the temperate broadleaved and conifer mixed forest. 5.3. Zone 13, tentatively assigned as 73.9–59.0 kyr BP, possibly corresponds to or within MIS 4

Alnus rises to more than 60% and Cyperaceae decrease sharply suggesting spread of Alnus in the area. A pollen assemblage with such high value of Alnus is similar to surface pollen assemblages between altitudes 2250 and 2500 m of the present natural forest, central Taiwan. Sharp decrease of Cyperaceae indicates still drier conditions than those of the previous zone. This assemblage represents the temperate deciduous forest within the present Upper Quercus Zone (or temperate broadleaved and conifer

Table 3

Assignments of plant functional types to biomes

Biome PFTs

Tropical evergreen forest te, tef, trx, x Tropical rainforest tr, tef, trx, x, ts3 Subtropical evergreen forest sut, tef, ts3, wtc, trx, x Warm-temperate evergreen forest wte2, tf, ts3, x Temperate broadleaved and

conifer mixed forest

wte1, ec, tc, ts1, ts2, ts, tf Cool-temperate conifer forest ctc, ec, g, h

Cold-temperate conifer forest bec, ec, g, h Alpine conifer forest ec, af, g Forest steppe wod, g, sf Alpine shrub land af, sf, g

(14)

mixed forests). Most of this zone is probably colder than the late stadial (Zone 5), considering the associated woody taxa of Zone 5.

5.4. Zones 12–6 (59.0–27.9 kyr BP): represent interstadial conditions of MIS 3

Alnus and Cyperaceae alternately dominated, within the temperate deciduous forest; Cyperaceae indicate increasing humidity. Ilex and Cyclobalanopsis increase. Castanopsis increases from Zones 12 to 10, indicating a warmer trend between the early and late stadials. There is a remarkable wet episode from cal. 42.2 to 37.0 kyr BP (estimated 36,100–32,500 yr BP) represented by spore-dominant assem-blages (Zone 9). Warm-temperate forest conditions at 41.6, 38.0 and 37.3 kyr BP appeared based on forest reconstruction also. A prominent wet episode (35–25 kyr BP) during MIS 3 has been documented by An (2000). Compared with this study, its age appeared to be younger.

The upper part of this interstadial (from Zone 9 upward) appeared relatively drier than the lower part as revealed by the amounts of Gramineae. Castanopsis has its lowest value between Zones 9 and 6 although showing a slight increase in Zone 7.

5.5. Zone 5, 27.4–16.7 kyr BP, late stadial, belongs to MIS 2 Vegetation in Zone 5, especially the upper part, is characterized by dominance in NAP, representing a temperate forest or possibly forest steppe. Between 23.2 and 18.7 kyr BP Gramineae and Artemisia reach their highest values, indicating relatively dry and sometimes cold conditions. Cyperaceae replace Gramineae at about 18.6 kyr BP and warm elements including Symplocos, Ilex and Cyclobalanopsis increase at about 16.7 kyr BP. How-ever, among the woody elements, Cyclobalanopsis rather than Alnus or Pinaceae appear. This is the interval corresponding to the LGM. According to the surface pollen study of lowland northeast China, 40% NAP marks the existence of forest-steppe zone, and this boundary almost overlapped the 700 mm/yr annual precipitation

(Ren, 1998). Thus, Zone 5 represents relatively dry

conditions but probably not as cold as those of Zone 13 (MIS 4). The less cold late stadial is shown in the result of biomization and also indicated by the study of Tsukada

(1967) at Sun-Moon Lake. In comparison to present-day

temperatures, the estimated temperature was 4–5 1C lower than today during the late stadial and 8–10 1C lower in the early stadial. This study conﬁrms this estimation. Warm-temperate to subtropical conditions appeared at 22.3 and 18.9 kyr BP before or within the Gramineae-prevailing drier phase as mentioned previously.

5.6. Zone 4, 27.4–11.0 kyr BP, the late glacial

The late glacial is characterized by a rise of AP. Among the woody taxa, Cyclobalanopsis, Ilex, Symplocos and Salix

are of changing importance, mainly representing temperate broadleaved and conifer mixed forests. The climatic conditions are drier and cooler than today, but warmer and wetter than before. A strong peak of monolete spores and warm-temperate forest at 15.1 kyr BP (12,800 yr BP) indicates the warm–wet Bølling interval. At about 13.0–12.5 kyr BP (10,900–10,450 yr BP) the increase of Salix marked the beginning of cold conditions. Then Ilex peaked at 12.1 kyr BP and Gramineae (435%) at 11.8–11.6 kyr BP (10,200–10,100 yr BP). They indicate a trend from cold to less cold and then dry-cold conditions of Younger Dryas time similar to the main trend found in the arid–semiarid transition zone of northern China (Zhou et al., 2001). According to a detailed study in the Netherlands

(Hoek, 1997), the late Pleniglacial ended about radiocarbon

age 12,900 yr BP (uncalibrated). From 12,900 to 12,450 yr BP is the Oldest Dryas, 12,450 to 12,100 yr BP is Bølling, 12,100 to 11,900 yr BP is Older Dryas, 11,900 to 10,950 yr BP is Allerød, 10,950 to 10,150 yr BP is Younger Dryas. In the Toushe record, the warm interval with a peak of monolete spores at 15.1 kyr BP marked the warm–wet episode corresponding to Bølling, while peaks of Salix and Gramineae may correspond to Younger Dryas (13.0– 11.6 kyr BP or 10,900–10,100 yr BP). Warm-temperate to subtropical elements increase at 11.5 kyr BP, but soon return to temperate conditions at 11.2–11.0 kyr BP (7.18 m). After 11.0 kyr BP warm-temperate to subtropical forests appeared almost continuously.

5.7. Zones 3–1 (11.0–1.8 kyr BP)

A Salix peak appeared at 11.0 kyr BP (7.07 m) which just preceded the wet phase indicated by the increase of monolete spores at 10.7 kyr BP (6.97 m). Climatic condi-tions become subtropical/warm–wet again as indicated by large amounts of monolete spores, hydrophyllus herbs and an increase of Castanopsis at the expense of Ilex, Ligustrum, Symplocos and Salix. But the rise of Salix, Ilex and Symplocos again at about 9.6–9.4 kyr BP (8600–8400 yr BP) indicates a cold episode. The subtropi-cal-warm elements increased in the middle Holocene, i.e. Mallotus began to increase between 7.3 and 6.8 kyr BP (6500 and 6000 yr BP) and Glochidion increased at 6.2–5.8 kyr BP (5450–090 yr BP). The higher value of Pinus—which grows far from the study site—runs parallel to the trend of the warm element Castanopsis, which might indicate an intensiﬁed monsoon at about 6.9–6.8 kyr BP (6150–6050 yr BP). Salix peaks indicated the prevailing deciduous forest and less warm conditions at about 11.0, 9.6–9.4, 9, 7.9, 7.5, 7.2 and 7.1, 5.2 and 5.0, 4.0 and 3.7 kyr BP.

The most remarkable feature in the Holocene record is a conspicuous dominance of monolete spores which indicate wet conditions. An et al. (2000) mentioned that the Holocene optimum may be represented by a wet interval which is asynchronous in East Asia. There are several intervals during the Holocene with such features especially

(15)

before 6.8 kyr BP in present study, including 10.6–10.3, 10.0–9.7, 9.5–8.5, 6.9–6.8 and 2.9–1.8 kyr BP (9300–9100, 8900–8600, 8500–7700, 6000–6100 and the last 3000 yr BP). However, the record of the last 5000 yr in the alpine lakes at almost the same latitude has been discussed by the authors elsewhere (Liew and Huang, 1994). The higher value of NAP in the early Holocene is considered to be a result of the enlargement of ﬂooding areas due to early Holocene wet conditions (Bush, 2002), while that of the recent several millennia is attributed to human disturbance.

The average score of each zone presented in the ﬁrst and second axis of DCA (Grimm, 1997) is shown inFig. 5. Pollen taxa involved in DCA analysis are those with percentage 42%, including Pinus, Taxodiaceae, Ericaceae, Ilex, Alnus, Quercus, Cyclobalanopsis, Catanopsis, Liquidambar, Symplo-cos, Trema, Ulmus, Ligustrum, Salix, Glochidion, Homa-lanthus, Mallotus, Hydrangea, Tristellateia, Gymnosporia, Aizoaceae, Artemisia, Compositae, Polygonum, Ranunculus, Umbeliferae, Typha, Cyperaceae, Gramineae and Liliaceae. The similarity of Zones 6–13 may indicate cold conditions before the late stadial drought; similarities among Zones 7, 9, 10 and Zones 8, 11, 12 may indicate cyclic changes in humidity during the interstadial. Large shifts occurred between Zones 19 and 18, 18 and 17, 16 and 15, 15 and 14, and 4 and 3 indicating the abrupt change during the early glacial and between the late glacial and Holocene.

6. Discussion

We used a biomization technique to reconstruct past altitudinal vegetation of the Toushe pollen sequence. This allows interpretation of the magnitude of possible environ-mental changes if the site is sensitive to past ecotone changes. For this site, especially in the Holocene, changes in prevailing biomes appear frequently (Fig. 4.2). Tempe-rate deciduous forest dominated by Alnus prevailed during the last glacial. The cold or cold/dry phase in the last glacial is possibly indicated by the times when peaks of

Gramineae appeared although they are hardly recognized in the biome reconstruction of this study. Tsuga–Picea forest does not appear at this site during the last glacial although it appears during the early stadial in Sun-Moon Lake located 100 m higher. Thus, we believe that the boundary between Tsuga–Picea forest and Alnus forest was located near the Sun-Moon Lake in the early stadial (Liew

and Chung, 2001).

The large altitudinal range of Castanopsis pollen type is another concern in the reconstruction of past biomes. In the present study, we include Castanopsis in subtropical and warm-temperate forests, which means excluding it from the temperate broadleaved and conifer mixed forest. If we consider the high-growing species and include Castanopsis-type pollen in wte1 also, with Castanopsis also included in temperate forests, we will ﬁnd only a small change in Holocene biome reconstruction but a relatively clear change in that of the early glacial (six samples among 83 samples in Holocene, i.e. ages at about 4.8, 4.85, 4.9, 7.6, 7.7 and 7.9 kyr BP will change to temperate broadleaved and conifer mixed forests; while most of Zone 17 will become temperate broadleaved and conifer mixed forests). Thus, the Castanopsis-dominant assemblage in early glacial is possibly like that of present subtropical/warm-temperate forests but less humid, or alternatively like parts of temperate broadleaved and conifer mixed forests.

6.1. Maximum cooling of the last glacial

Long-term vegetational records often show the response of local ﬂora to climatic changes. During the last glacial, the winter monsoon intensiﬁed (Huang et al., 1997), and the climate was relatively drier and colder than that of today. As mentioned above, the stadials of the last glacial are represented by Zone 13 (possibly MIS 4) and Zone 5 (MIS 2) in which the early stadial (Zone 13) has the highest value of Alnus (460%). Surface pollen assemblages of Salisien area show that Alnus-dominant (50%) conditions could be found at about altitude 2250 m of today’s forests. Thus, a vertical migration of 1500 m could have occurred between the early stadial of the last glacial and the Holocene. Furthermore, in contrast to the Alnus-dominant assemblage of Zone 13 in Toushe, a contemporaneous Pinus/Tsuga-dominant assemblage is found at the neigh-boring Sun-Moon Lake (Tsukada, 1967). Thus, the level of Sun-Moon Lake (750 m) should represent the lower boundary of Alnus/coniferous forest of the early glacial. Presently the boundary lies at about 2400 m above sea level. Though Taiwan’s uplift rate is generally high, this area is fairly stable (Chen, 2003), and vertical displacement due to tectonism is negligible. Thus, the maximum ecotone migration is estimated to be at least 41000 m and possibly about 1500 m; therefore, the temperature difference be-tween MIS 4 and present is 8–10 1C.

The smooth sedimentation rate in this peat bog (Fig. 3) excludes the conspicuous diastem during the LGM. The severity of cold conditions of the early and late stadials of

Fig. 5. Detrended correspondence analysis. Mean scores for the samples of each zone (from 20 to 1) plotted for the ﬁrst two DCA axes.

(16)

the last glacial here (Zone 5) appeared different from those of many other regions. As described above, Gramineae sometimes reached more than 40% and NAP prevailed during the late stadial, which indicates drought conditions with precipitation less than half the present level probably. However, if we compare the woody elements with the early stadial (Zone 13, MIS 4), the late stadial shows a lower percentage of Alnus but higher percentage of Cyclobala-nopsis. Thus, most of MIS 4 may have been colder than MIS 2. The biomization results show a similar trend. Conditions of relatively higher sea level and lower insolation during MIS 4 than that of MIS 2 may be plausible reasons. The southward shifting ITCZ and trade wind belt as well as the lowering of sea level during the late stadial (MIS 2) possibly resulted in the drought conditions in this area, while the effect of the maximum ice sheet of MIS 2 on subtropical temperature needs to be evaluated.

Many discussions have emphasized the wet interval during the later part (35–25 kyr BP) of MIS 3 and the East Asia monsoon (An, 2000;Shi et al., 2001). In Europe, there is a mild phase during MIS 3 at about 39,000–36,000 yr ago

(Andel and Tzedakis, 1996), similar to the warm–wet

interval of 37–42 kyr BP characterized by remarkable spore peaks in the present study. If they are related, then the warm phase may have occurred not only in Asia.

6.2. Younger Dryas

The Younger Dryas was indicated by the remarkable increase of Salix at about 13.0–12.5 kyr BP which differs from previous Cyclobalanopsis–Symplocos prevailing as-semblages, and culminated in the later part of this cold event by an increase of Gramineae (435%) at 11.8–11.6 kyr BP. They indicate a trend from cold to less cold and then dry-cold conditions of Younger Dryas time, similar to the main trend found in the arid–semiarid transition zone of northern China (Zhou et al., 2001). It is almost synchronous with climate records from Europe

(Peteet, 1995) and from Africa (Bonneﬁlle et al., 1995). In

equatorial and northern Africa, Gasse (2000) also found two drastic arid-humid transitions occurring around 15–14.5 and 11.5–11 ka which correspond to the post-Bølling interstadial and post-Younger Dryas warming event in this study. However, at the present study site, the change of vegetation in 13.0–11.6 kyr BP is still within the range of temperate broadleaved and mixed conifer forest based on biomization. Thus, this cold event is not recognizable from the results of forest change only.

Nakagawa et al. (2003) described Younger Dryas in the

pollen sequence of Lake Suigetsu, Japan, with about a 400 yr lag compared with the Greenland ice core (Alley,

1993; Mayewski et al., 1993). According to the pollen

diagram of Lake Suigetsu, an increase of Fagus and a decrease of Carpinus occurred about 12,600 yr BP. Thus, the onset of Younger Dryas remains equivocal. One possible explanation of this age discrepancy is the

deﬁnition of magnitude of coldness for the beginning of the Younger Dryas.

7. Conclusion

The pollen stratigraphy from the low mountain area of central Taiwan shows vegetational change across the last 96,000 yr. During MIS 5 (probably 5c), the Castanopsis-dominant assemblage possibly represents a subtropical evergreen broadleaved forest, less humid than at present, assuming that these Castanopsis are not high-growing species. A Salix-dominant interval follows, which is tentatively correlated with MIS 5b. After that, Symplocos and Myrica dominate; in conjunction with a high spore level, this indicates a subtropical forest with a wet episode. Castanopsis then prevails again. Thus at least one wet interval existed in the early glacial (MIS 5a). The Alnus-dominant forest replaced the preceding Castanopsis forest during most of the last glacial. In the early stadial (MIS 4), the Alnus percentage is higher than that of the late stadial (MIS 2). A 8–10 1C lower annual temperature than at present and an about 1000–1500 m lower boundary of the Tsuga–Picea Zone/Upper Quercus Zone than at present characterizes the early stadial (MIS 4). However, in MIS 2, the high value of NAP might represent a forest steppe condition; a fall in precipitation to half the present level might be possible. Alnus is lower but Cyclobalanopsis is higher during the late stadial (MIS 2) than in the early stadial (MIS 4) indicating less cold conditions than in most of the early stadial. The generally drier conditions of the glacial stage are interrupted by an episodic warm/wet interval around 42.2–37.0 kyr BP. Abrupt changes during the late glacial and Holocene are also clear. The peak of monolete spores at 15.1 kyr BP (12,800 yr BP) and a warm-temperate forest mark the warm–wet episode correspond-ing to the Bøllcorrespond-ing. At about 13.0–12.5 kyr BP an increase of Salix possibly marks the onset of cold conditions. Then after a peak of Ilex at 12.1 kyr BP, Gramineae (435%) rise at 11.8–11.6 kyr BP which may correspond to the cold to less cold and then cold-dry phases of the Younger Dryas. A subtropical warm period also appeared at 11.5 kyr BP as indicated by biomization. Subtropical conditions continue to prevail from 10.7 kyr BP onward. An increase of warm elements, such as Castanopsis, appeared around 11.2 kyr BP and wet conditions indicated by spore peaks occurred at about 10.7 kyr BP. Possible abrupt cold episodes are indicated by Salix peaks at about 11.0, 9.6–9.4, 7.9, 7.1, 5.2 and 5.0, 4.0 and 3.7 kyr BP. Subtropical to tropical elements such as Mallotus increased in the mid-Holocene at about 7.3–6.8 kyr BP (except 7.2 kyr BP), while Glochidion was higher between 6.2 and 5.8 kyr BP, probably indicating a warmer interval when the summer monsoon intensiﬁed. The warmest climate oc-curred at 6.9 and 6.1–5.9 kyr BP based on forest recon-struction. High early Holocene spore peaks occurred in the intervals 10.6–10.3, 10.0–9.7, 9.5–8.8, 6.9–6.8 and about 2.9–1.8 kyr BP and are considered as a proxy of higher

(17)

precipitation induced by an intensifying East Asian monsoon.

Acknowledgements

The authors are grateful to the reviewers Dr. Pavel Tarasov and Dr. John Dodson for their constructive suggestions in improving this paper. We also wish to thank Prof. C.F. Hsieh in the Department of Lifesciences of National Taiwan University and Prof. Ge Yu and Dr. B. Xue from Nanjing Institute of Geography and Limnology for their information about their biomization study, Dr. M.L. Hsieh and Mr. T.S. Shih for their help in the ﬁeld and Mr. C.Y. Lee and B.C. Chen for their computer work during this study. Thanks are also due to Dr. Grimm for DCA analysis using Tilia Program. This study was funded by the National Science Council for the PAGES project.

References

Allen, J.R.M., Watts, W.A., Huntley, B., 2000. Weichselian palynostrati-graphy, palaeovegetation and palaeoenvironment: the record from Lago Grande di Monticchio, southern Italy. Quaternary International 73/74, 91–110.

Alley, R.B., 1993. Abrupt increase in Greenland snow accumulation at the end of the Younger Dryas event. Nature 362, 527–529.

An, Z.S., 2000. The history and variability of the East Asian paleomonsoon climate. Quaternary Science Reviews 19, 743–762. An, Z.S., Porter, S.C., Kutzbach, J.E., Wu, X.H., Wang, S.M., Liu, X.D.,

Li, X.Q., Zhou, W.J., 2000. Asynchronous Holocene optimum of the East Asian monsoon. Quaternary Science Reviews 19, 171–187. van Andel, T.H., Tzedakis, P.C., 1996. Palaeolithic landscapes of Europe

and environs, 150,000–25,000 years ago. Quaternary Science Reviews 15, 481–500.

Bard, E. Arnold, M., Hamelin, B., Tisnerat-Laborde, N., Cabioch, G., 1998. Radiocarbon calibration by means of mass spectrometric 230Th/ 234U and 14C ages of corals: an updated database including samples from Barbados, Mururoa and Tahiti. Radiocarbon 40 (3), 2, 1085. Bond, G., Heinrich, W., Broecker, L., Labeyrie, J., McManus, J.,

Andrews, S., Huon, R., Jantschik, R., Clasen, S., Simet, K., Tedesco, Klas, M., Bonani, G., Ivy, S., 1992. Evidence for massive discharges of icebergs into the North Atlantic during the last glacial period. Nature 360, 245–249.

Bonneﬁlle, R., Riollet, G., Buchet, G., Icole, M., Lafont, R., Arnold, M., Jolly, D., 1995. Glacial/interglacial record from intertropical Africa, high resolution pollen and carbon data at Rusaka, Burund. Quaternary Science Reviews 14 (9), 917–936.

Bush, M.B., 2002. On the interpretation of fossil Poaceae pollen in the lowland humid neotropics. Palaeogeography, Palaeoclimatology, Palaeoecology 177, 5–17.

Chen, B.C., 2003. Sedimentary and neotectonic studies of the Puli and Yuchi Basins, central Taiwan. Masters Thesis, National Taiwan University, Taipei.

Chung, N.J., 1994. Studies on the vegetation ecology and the conservation characteristics of Salisen Area, central Taiwan. Ph.D. Thesis, National Taiwan University, Taipei (in Chinese).

CLIMAP Project Members, 1981. Seasonal reconstruction of the earth’s surface at the last glacial maximum. Geological Society of America, Map and Chart Series MC-36.

Editorial Committee of the Flora of Taiwan, 1993. Flora of Taiwan, vol. 3, second ed. Department of Botany, National Taiwan University, Taipei (1084pp).

Editorial Committee of the Vegetation of China, 1980. Vegetation of China. Science Publishing Company, Beijing (1375pp).

Farrera, I., Harrison, S.P., Prentice, I.C., Ramstein, G., Guiot, J., Bartlein, P.J., Bonneﬁlle, R., Bush, M., Cramer, W., von Grafenstein, U., Holmgren, K., Hooghiemstra, H., Hope, G., Jolly, D., Lauritzen, S.-E., Ono, Y., Pinot, S., Stute, M., Yu, G., 1999. Tropical climates at the last glacial maximum: a new synthesis of terrestrial palaeoclimate data. I. Vegetation, lake-levels and geochemistry. Climate Dynamics 15 (11), 823–856.

Flenley, J., 1979. The Equatorial Rain Forest. A Geological History. Butterworths, London (162pp).

Gasse, F., 2000. Hydrological changes in the African tropics since the last glacial maximum. Quaternary Science Reviews 19, 189–211.

Gotanda, K., Nakagawa, T., Tarasov, P., Kitagawa, J., Inoue, Y., Yasuda, Y., 2002. Biome classiﬁcation from Japanese pollen data: application to modern-day and late Quaternary samples. Quaternary Science Reviews 21, 647–657.

Grimm, E.C., 1997. Tilia, Version 2.0 b5. Illinois State Museum, Springﬁeld.

Hoek, W., 1997. Palaeogeography of lateglacial vegetations: Aspects of lateglacial and early Holocene vegetation, abiotic landscape and climate in the Netherlands. Netherlands Geographical Studies 230 (147pp).

Hooghiemstra, H., 1989. Quaternary and upper Pliocene glaciation and forest development in the tropical Andes: evidence from a long high resolution pollen record from the sedimentary basin of Bogota, Columbia. Palaeogeography, Palaeoclimatology, Palaeoecology 72, 11–26.

Hormes, A., Shotyk, W., Schlu¨chter, C., Reese, S., Stocker, T.F., 1999. A prominent interstadial between 19.9 and 18.3 kyr BP during the last glacial maximum in New Zealand, South Island-evidence from 14C dating on different organic components of Dy and Gyttja samples. In: Program with Abstracts, XV INQUA Congress, p. 84.

Huang, C.Y., Liew, P.M., Zhao, M., Chang, T.C., Kuo, C.M., Chen, M.T., Wang, C.H., Zheng, L.F., 1997. Deep sea and lake records of the southeast Asian paleomonsoons for the last 25 thousand years. Earth and Planetary Science Letters 146, 59–72.

Jolly, D., Prentice, I.C., Bonneﬁlle, R., Ballouche, A., Bengo, M., Brenac, P., Buchert, G., Burney, D., Cazet, J.-P., Cheddadi, R., Edorh, T., Elenga, H., Elmoutaki, S., Guiot, J., Laarif, F., Lamb, H., Lezine, A.-M., Maley, J., Mbenza, A.-M., Peyron, O., Reille, A.-M., Ssemmanda, I., Straka, H., Umer, M., Van Compo, E., Vilimumbalo, S., Vicens, A., Waller, M., 1998. Biome reconstructed from pollen and plant macrofossil data for Africa and the Arabian peninsula at 0 and 6 ka. Journal of Biogeography 24, 1007–1027.

Kuo, C.M., Liew, P.M., 2000. Vegetation history and climatic ﬂuctuations based on pollen analysis of the Toushe peat bog, central Taiwan since the last glacial maximum. Journal of the Geological Society of China 43 (3), 372–392.

Liew, P.M., 1977. Pollen analysis of Pleistocene sediments at Waichia-taokeng—central Taiwan. Acta Geologica Taiwanica 19, 103–109. Liew, P.M., Chung, N.J., 2001. Vertical migration of forests during the

last glacial period in subtropical Taiwan. Western Paciﬁc. Earth Sciences 1 (4), 405–414.

Liew, P.M., Huang, S.Y., 1994. A 5000-year pollen record from Chitsai Lake, central Taiwan. Terrestrial, Atmospheric and Oceanic Sciences 5 (3), 411–419.