In: G.Miehe & Zhang Yili (eds.) 2000: Environmental Change in High Asia, Marburger Geographische Schriften 135, 201-227. ISBN 3-88353-062-X
Patterns of Forest Distribution and the Impact of Fire and Pastoralism in the Forest Region of the Tibetan Plateau
The present pattern of forest distribution in the forest region of the Tibetan Plateau is characterised by forest fragmentation, especially the phenomenon of forest-free south-facing slopes. Thus far it was widely assumed that the absence of forests on south slopes in an otherwise forested region reflects a primary forest distribution pattern caused by climatic conditions. Yet juniper forests are for example well adapted to the climatic site conditions of south slopes. It will be shown that the impact of pastoralism is responsible for continued absence of forests on south-facing slopes. Initial forest destruction is most likely caused by the impact of fire. Presently fire is a very rare act by nature, but a common event after human ignition, be it an accident or intentional forest or pasture burning by pastoralists. In this context all available references on forest fires of the region are presented. Furthermore, natural forest regeneration is prohibited by continued grazing and additional periodic burning, since south-slope pastures are of utmost importance for winter livestock grazing. A case study from Zitsa Degu / Jiuzhaigou (Aba Tibetan AP, W-Sichuan Province, 33°N/104°E) in connection with preliminary observations from Riwoqê (Qamdo Prefecture, Tibet AR, 31°13'N/96°36'E) show the impact of traditional land-use strategies on forest distribution. In light of new research it is evident that vast areas of the Tibetan Plateau need to be regarded as cultural landscape rather than wilderness.
So far it was believed that the forest distribution patterns in the eastern Tibetan Plateau reflected to a large degree the natural state of potential forest distribution. Not only most explorers of the first half of this century believed to have encountered a landscape only slightly altered by human activities, but still in 1985 Zheng (1986: 33) stated: "historically human activities have been insignificant". Furthermore a new review on China's biodiversity perpetuates a similar notion. Although for each biounit's ecosystem there is a graph totaling its status (protected, destroyed, etc.), the status of 'cold coniferous forests' reflects available Chinese figures on the devastating impact of modern logging, but does not indicate historical deforestation (Mackinnon et al. 1996: 275, 311, 329). However, recent field research (Frenzel, 1994, 1995; Miehe et al. 2000, Thelaus, 1992; Winkler, 1994, 1997, 1998b) indicates clearly that the impact of humans and their grazing animals are being greatly underestimated. In the forest region of Tibet, this is exemplified especially by the phenomenon of forest-free south slopes or forest-clad north facing slopes, besides disjunct relict forests in otherwise at present completely forest free areas. Moreover, the impact of fire has been mentioned briefly in many accounts, but no attempt has been made in trying to understand its role in shaping forest distribution patterns. Similarly, the far-reaching impact of pastoralism has not been fully taken into account. Besides direct impact from trampling and grazing the most powerful agent of the herder is fire. Fire is set to replace forest and shrublands with desirable grazing grounds. This analysis of forest distribution patterns is based on field work carried out in Zitsa Degu (Jiuzhaigou, Aba TAP, NNW-Sichuan) in 1991, and preliminary studies in Riwoqê (Leiwuqi, Qamdo Prefecture, TAR) in 1997 as well as observations in Central Tibetan sites and along the northern route connecting Lhasa via Nagqu (Naqu, Nagchu), Qamdo (Changdu, Chamdo), Garzê (Ganzi, Kandze), and Kangding (Dartsedo) with Chengdu [locations within Tibet Autonomous Region (TAR) and Tibetan Autonomous Prefectures (TAP) are referred to by their official Sino-Tibetan name. If the Mandarin or Tibetan name (according to Dorje 1996) differs greatly, it is given in brackets]. While it took millennia to develop eastern Tibet's cultural landscape, modern forest destruction inflicted by state controlled commercial logging seems to accomplish a similar degree of forest reduction in some areas within decades (see Winkler, 1998a/b).
2 Forest-free south slopes
While all exposures of the humid outer reaches of the southern and south-eastern Tibetan plateau are typically forested, areas of inner eastern Tibet and the northern part of south-eastern Tibet show a very different pattern of forest distribution. Here in this more continentally influenced climate most south-facing slopes are presently grasslands, while the north-facing aspects are forest-clad, predominately by spruce-fir forests (fig. 1).
Fig. 1: The typical forest distribution pattern of forest-free south slopes and forest-clad north slopes encountered on a W-slope between Lhakang (Tagong) and Xinduqiao (Dzongzhab), near Qianning ca. 40 km W of Kangding (Dartsedo). The distribution of solitary trees indicates that forest growth on the at present deforested slopes is possible.
Photo: (c) D. Winkler July 11, 1997 (3.600 m a.s.l., towards E), near Qianning, Kangding County, Ganzi TAP, Sichuan
Fig. 2: Map of Zitsa Degu / Jiuzhaigou (top left) and the south-eastern Tibetan Plateau and adjacent regions: In the shaded area (based on Wissmann 1961: 25) forests are mainly distributed on the north-facing slopes; south-facing slopes are mostly forest free. To the north-west of the shaded area forests are at present almost absent). Map: Daniel Winkler 1998.
Wissmann (1961: 25) mapped out the areas of High Asia, where forest-clad north slopes predominate (fig. 2). Although this mapping is outdated, due to the availability of Chinese vegetation maps and recent fieldwork, it gives a general idea of the vastness of the area concerned. This phenomenon of forest distribution is not limited to eastern Tibet. It is also present in the Himalayas, although to a lesser extend. For example, south slopes in some locations of the Nepal Himalaya are cleared by fire to create pastures (Schmidt-Vogt, 1990; Miehe, 1990). Furthermore in the adjoining Karakorumsimilar patterns are to be found, yet their recent development is clearly documented (Schickhoff, 1995). In addition detailed phyto-sociological studies carried out in Mongolia (Hilbig, 1987) show the devastating impact of pastoral land-use strategies on Mongolia's forests. Hilbig attributes the present forest distribution pattern ('forest-steppe mosaic') to human impact and concludes the potential forest area has been reduced significantly in the last millennium. In several sites herbaceous species composition and shrubs under forests or trees and out in the open ('steppe') do not differ significantly.
2.1 Previous explanations
The striking pattern of eastern Tibet's forest distribution has been reported by many scientists and western travellers in the first half of the 20th century (Tafel, 1914; Rock, 1930, 1956; Weigold, 1935; Schäfer, 1938; Hanson-Lowe, 1940; Ku & Cheo, 1941; Teng, 1947a; Duncan, 1952). Only Teng, a pioneering Chinese forester and Tafel, a trained German medical doctor turned explorer, have fathomed the far reaching human impact in Eastern Tibet. Otherwise it was believed that the sunny slopes' climatic conditions were too extreme for forest growth. Rock (1930:161) attributed the absence of forests on the south-facing slopes "to earlier melting of snow". Weigold (1935: 224) assumed that the south facing slopes where drying out too quickly because of the intense insolation. Similarly Ku & Cheo (1941:127) stated that "this difference in insolation [S-slope - N- slope] causes a considerable difference in the rate of evaporation and hence, in turn in the nature of vegetation". Hanson-Lowe (1940: 366) pointed out "the greatest contrasts in insolation, exposure to winds, duration of snow cover, amount of precipitation, rapidity of run-off, etc." between the northern and southern exposures. Schäfer (1938: 56/7) wrote (all quotes of German sources are translations by the author), "scientists tried to explain the very regular distribution of vegetation in many ways. Most commonly it was attributed to the prevailing wind direction, especially in connection with the summer's precipitation conditions". According to Schäfer's own observations, the main factors are evaporation and distribution of snow. Wissmann (1960: 268), who regarded the absence of forests on south slopes as a direct result of exposure to insolation, summarized Schäfer's explanation: "The highly irradiated south-facing slopes are free of snow cover nearly throughout the whole dry winter. High evaporation in connection with high diurnal changes of temperature, which causes thawing in the day and frosts usually as low as -15°C to 20°C at night, can only be survived by grass cover, not by trees or shrubs". This view was also perpetuated by Schweinfurth (1972: 282). All of these hypotheses recognize relevant aspects of the climatic conditions of the south-facing slopes. However, if Wissmann's hypothesis is fully accurate, any growth of woody vegetation on sunny slopes should be impossible, which is a claim not made by anyone who traveled the area.
2.2 Juniper forests
Junipers are especially well adapted to south slope's high insolation and extreme diurnal ranges of temperature. They are very competitive in areas with precipitation between 400 and 600 mm (Li, 1993: 60). Miehe et al. (2000) recorded a juniper stand in Nagarzê near Yamzho Yumco (Yamdrok Tso, Central TAR) where present mean annual precipitation is only 336 mm. The minimal annual average temperature tolerated is around 0°C (Li, 1993: 60). Juniper's ability to endure a cold and dry climate makes it the most widespread and competitive forest community in the transition area between spruce-fir forest communities and shrub and grasslands devoid of tree growth. Yet juniper forests can also be found in eastern Tibet in areas which seemingly have a much higher precipitation. For example in Zitsa Degu precipitation is recorded at 744 mm/a, but this data is collected at 2,400 m a.s.l. on the lower range of the condensation level. In the cloud forest belt precipitation seems to be at least 1,000 mm/a (Winkler, 1997:145). Yet above the upper limit of the condensation level the climatic water balance changes, precipitation decreases and evaporation increases, creating dry-cold sites favoured by junipers.
Juniper forests might consist of one or a combination of the following species: Juniperus tibetica, J. convallium, J. komarovii, J. saltuaria, J. wallichiana (syn. J. indica), J. przewalskii, and J. chinensis.However the taxonomy of junipers is in need of revision. Some of these species might be varieties or sub-species. Adding to the confusion Chinese taxonomists differentiate junipers into the generaJuniperus and Sabina. Besides junipers eastern Tibet's high altitude south-slopes might typically be clad by evergreen oak (Quercus aquifolioides), pine (i.e. Pinus densata), as well as spruce (Picea balfouriana, P. purpurea), and larch (Larix potaninii, L.p. var. macrocarpa L.griffithiana [see Wang & Zhong, 1995]). Often these species intermix according to site conditions. Although junipers are heliophilous, juniper forests are distributed in some areas on either slope, i.e. Juniperus przewalskii in the north-east of the A'nyêmaqên range (Huang, 1987: 476).
Junipers grow slowly, but can grow very old. Chow (1947: 202) sampled height growth of seedlings; based on an average of 40 to 80 seedlings found in SW-Gansu's Towho (Tao He) watershed. While it took Juniperus tibetica 11.2 years to reach 15 cm and 34 years to reach 105 cm, J. saltuaria reached 15 cm in 22.6 and 105 cm in 79 years. These results do not compare well to Li (1993: 60), who states that the average age of juniper forests varies between 100 to 200 years with an average height of 8 to 10 m. However, both studies together reflect the amplitude of juniper growth rates under different conditions. Junipers can reach 1,500 years (Frenzel et al., 1995: 492), one sample from stands close to the Zhopel La (pass) between Qamdo and Riwoqê analyzed by Bräuning (1994: 88) supplies dendrochronological data reaching back to the 7th century. Due to its slow growth rate juniper is banned officially from timber trade in TAR, yet their timber is very durable and has been widely used traditionally, though when available other timber (spruce, pine etc.) is often preferred for construction, since it is much easier to work with. The average growing stock of junipers is only 20 to 100 m³/ha (Li, 1993: 60). Furthermore juniper is in high demand as firewood and for ritual purposes; its branches are used commonly as incense. The practice of lopping off tops of young junipers, and not only branches, was observed above the village of Dzamthog (Jomda [Jiangda] County, Qamdo Prefecture). In many locations juniper growth is in shrub form due to lopping, although juniper trees are present also. However, they are only directly around homesteads or on inaccessible sites. The impact of using juniper (Tibetan: "shukpa") for incense burning is not to be underestimated. Every household traditionally performs 'sang', the offering ritual of burning incense, every morning.
Natural regeneration is according to Li (1993: 60) "usually satisfactory in mature forests. There are numerous young seedlings under the sparse canopy of the forest (4,000 to 5,000 per ha), however, only about 10% of them can survive until the next phase." Ku & Cheo (1941: 91) report, that "seedlings, in general, are present in small quantities. Juniper seeds are dispersed by birds, thus junipers can regenerate far from the location of the seed source. Slow growth exposes seedlings much longer to possible damage through animal trampling; though the seedlings themselves are usually not browsed due to the foliage's high aromatic oil content. But the stands open character and its herbaceous ground cover are welcome features for grazing. Intruding herds seriously disturb the forest's ability of self-regeneration. Moreover, junipers are very vulnerable to fire. Arend (1950: 129) reports about the North American Eastern redcedar (Juniperus virginiana) that "fire is the worst natural enemy"; it "can not maintain itself in areas that burned over frequently. The bark is so thin - rarely more than 0.3 inch [= 0.75 cm] thick, that heat from one surface fire usually kills even mature trees". J. virginiana is disliked by American cattle ranchers for invading grasslands after human fire suppression. Similarly, Tibetan juniper species are very susceptible to fire damage, because of thin bark, high content in resin and the inflammable nature of their foliage (see Teng, l947b: 227; Schmidt-Vogt, 1990: 198). Taking into account that many juniper forests are already the specialists in an environment on the edge of sustaining a forest vegetation due to low water availability and a short growing season, any impact deteriorating the harsh growing condition further will reduce forests' vitality and regeneration.
An interesting question, which has not been raised for Tibet yet, regarding juniper forests is, to which extent does their occurrence depend on natural site conditions and to which extend is it a result of degradation by human impact. This question pertains to the occurrence of more or less pure juniper stands in altitudes, aspects and areas where Pinaceae, such as Picea, Abies and Larix, are also present. In Riwoqê County, 50 km W of Qamdo town (precipitation 536 mm/a), juniper forest are confined to south slopes, which are to a high degree free of forests. Around the potential treeline of 4,600 m to 4,700 m juniper forests are presently a rare occurrence. One small stand was observed near the Dzekri-La (Pass) at around 4,600 m a.s.l. Juniper forests (Juniperus tibetica and J. wallichianaaccording to the local forest department) are presently more widespread in lower elevations between 3,800 m to 4,300 m a.s.l., often in the proximity to settlements. However, several juniper forests (i.e. W of Riwoqê, N of Sibta, around Yiri, fig. 3) contain well developed Picea balfouriana trees or stands (ca. 15-20 m, dbh of 0,7 m, 150 to 300 years) towering above the junipers (8-12 m, dbh of 0,5 m). A grazed site in Yiri showed no evidence of differing site conditions, but also no stumps indicating recent selective felling of spruce. However, stumps are a very transient indicator. Once they are dried out, they disappear quickly, since the procurement and transport of dried wood is more convenient (pers. comm. Lhakpa Sherpa). The presence of spruce trees and stands within the juniper forests indicates that Picea balfouriana is able to endure the south slope's site conditions, forming an open spruce forest interspersed with junipers. Thus Riwoqê's present juniper forest predominance on south-slopes below 4,300 m a.s.l. suggests that pure juniper forests are a result of degradation rather than natural vegetation. Also pollen-analysis from Zoigê (Ru`ergai) indicates that in the process of past deforestation spruce disappears at first. Juniper remains for a transition period to finally give way to grasslands (Thelaus, 1992, see also below). Similar patterns of Juniperus recurva stands within a lower altitudinal belt otherwise dominated by Abies spectabilis have been documented by Miehe (1990: 181) in Nepal's Langtang Himal. It seems that the development of Tibet's pure juniper forests several hundred altitudinal meters below treeline is caused by the practice of selective cutting of Pinaceae, the preferred wood for construction and burning. In Riwoqê, the removal of the taller and stronger shadowing spruces, causes a pronounced change of climatic water balance, exposing the ground to strong insolation. Furthermore, both grazing and firewood collection are degrading the shrub and herb layer, which aggravates site conditions even further. These changes of the microclimate seem to inhibit spruce regeneration under present land-use practices.
Fig. 3: Forest successions under human impact in East Tibet
photo will be provided later
In the left foreground the south slope is covered by a degraded juniper forest, which is grazed. It still contains a few well developed spruces (Picea balfouriana). In the center rear forests of a west facing slope have been fragmented for creating pastures. Here traces of forest fire were evident. Photo: D. Winkler 4.7.1997. Above Yiri Hot Springs (4,100 m a.s.l., Riwoqê County, towards NE)
2.2.1 Relict juniper forests
Juniper forests are of special interest regarding the question of present forest distribution versus potential forest distribution. The present distribution pattern seems not to demonstrate the forests' full original range. Several disjunct juniper forests have been discovered in the last years by science. Relict forests in Central Tibet, such as near Damxung, Reting, Nagarzê (see S. Miehe et al. 2000) indicate clearly a much wider former range of distribution. It is not fully clear if these relict forests are testifying to a past climate more conducive to forest growth as assumed by Li B. (1993: 521), or if these relic forests indicate potential of present forest growth, were it not for ongoing intensive human impact. It is evident that these forests were spared from human destruction, since most relict forests are present around monasteries and regarded as sacred. Monastery forests are also common in areas where recent deforestation is not disputed, for instance behind Gambel Gompa between Songpan and Kungari Pass (Gonggang Ling, Sonpan County, Aba TAP), 40 km SWS of Zitsa Degu (Winkler, 1994: 60), and in Manigango and Zoqen (Dzogchen, both Dêgê County, Garzê TAP). Unquestionably human protection is an important factor for their survival. Some forest stands, previously undocumented, have been observed by the author in the Zhorong Tsangpo Valley between Drigung Til (Zhigung Si) and Trulda (3,980 m a.s.l., Maizhokungar County, Central TAR, see fig. 2), where precipitation is recorded at 490 mm/a (S. Miehe et al. 2000). Here on an otherwise completely deforested south-slope, a last forest of junipers survived and a few solitary trees around a homestead. Further up the valley around Terdrom Monastery solitary juniper trees survived on cliffs in inaccessible sites. Interestingly in the Zhorong Valley there are also some north slope birch forests, mostly higher up on the slope.
3 The impact of fire
Fire has been man's most powerful tool in shaping subtropical and temperate ecosystems since time immemorial (i.e. Bird, 1995). However, fire has not really been considered as an active agent in shaping Tibet's forest distribution pattern and environment, although many reports contain a variety of references to forest fires. First accounts are to be found in Rockhill (1891: 231). He witnessed only traces of a small fire and was told that fires are of rare occurrence. However, Tafel (1914: 173) reported "Traces of forest fires are a common sight". He observed a fire raging for many days completely destroying several square kilometres of "beautiful primary forests". Most fires, Tafel realized, occur in the dry winter season. Heim (1933: 145) attributed oscillations of the treeline to insolation, wind and the impact of fire. Teng (1940: 389) laments the conditions of the coniferous forests in NW-Yunnan and SE-Sikang [presently SW-Sichuan]: "Sheer neglect and wanton destruction have reduced the forest in this region to a deplorable state". "Forest fires, intentional or unintentional, have wrought startling disasters". "Had it not been for the heavy precipitation which prevents the spread of fire, these forests would have vanished decades or perhaps centuries ago". Also Ku & Cheo (1941: 131) report that "forest fire [..] has swept over enormous areas". Wang (1961: 34) understood for China's montane-boreal coniferous forest, that its "replacement [..] by lesser vegetation can be attributed to destructive agents, especially fire. Forest fire is evident in the most secluded and sparsely inhabited areas". Chen (1987: 223, 269) regards pine (Pinus tabulaeformis and P. densata) and larch (Larix potaninii) forests as extremely endangered by fire in the dry winter season, a fact regarding pine forests (Pinus tabulaeformis, P. densata, P. yunnanensis) also stated by Li (1993: 69, 73, 75). Yang reports (1987: 76) "fires do enormous damage to forests" in West-Sichuan. The area inflicted comprises at least 755,000 ha (which is 6.3% of the whole forest area) with an annual average of 1,240 fires. He points out that pine forests are most prone to fire; 79% of the impacted area (ca. 600,000 ha) is located in the south-east of W-Sichuan, which is rich in pine forests. Still over 10% of the area destroyed (ca. 80,000 ha) consists of subalpine spruce-fire forests. Sichuan Senlin (1992: 1252) states that in general fires in the spruce-fir forest belt are less frequent, but when occurring they tend to be very intense and destroy big areas. Due to the steep terrain it is very difficult to fight fires once they do occur (Yang 1987: 76), not to mention the lack of equipment in these remote areas.
As Tafel (1914) already observed most of the fires do occur in winter. Higher summer temperatures usually very favourable for ignition elsewhere are not able to offset the summer's increased precipitation brought in by the monsoon. More precisely, fires are most prevalent in late winter, once the moisture brought in by the summer monsoon has dissipated by run-off and evaporation, until the next onset of summer monsoon, a fact also encountered in the high altitude forests of the Nepal Himalaya (Schmidt-Vogt, 1990). Fire prevention research by MA et al. (1991) on Pinus yunnanensisforests in SW-Sichuan and by LIU (1994) on Pinus densata-forests in Nyingchi (Linzhi) Prefecture (SE-TAR) state, that over 90%, respectively 99% in Nyingchi, of forest fires occur between December and May, while only 2% occur in SW-Sichuan and are absent in Nyingchi during the summer monsoon. Sichuan Senlin (1992: 1252) reports that the months with the highest impact of fire are January to May, but especially February, March and April; Liu also reports from the area around the Yarlung Zhangbo (Tsangpo) bend, that most fires occur in Pinus densata forests in exposed south-slopes below 3,300 m. Fire in fir-spruce forests above 3,300 m a.s.l. are less frequent, yet after extended dry periods in connection with 'favourable winds' fire might ascend into the fir-spruce forests, which are very rich in organic matter, in one case fire continued as a subsurface fire. Ogilvie (1996) reports on successful fire prevention measures in Deqen TAP (Diqing, NW-Yunnan). Fire occurrence has been decreased from over 100 per year down to 5 to 10. The success is attributed to stricter regulations banning burning, extension programs and remote sensing monitoring from Kunming. Nearly all recorded fires (155) in Nyingchi, which occurred between 1981 and 1991, were caused by man. In case of known origination (121 fires, 78%) most prevalent was accidental fire setting. Lightning as a cause was the rare exception (1 fire, <1%; Liu 1994). Yang (1987: 76) attributes 40% of forest fires to intentional ignition, 40% to accidents and a surprisingly high 20% to lightening. This must be an error, since Sichuan Senlin (1992: 1253), of which Yang was a chief editor, attributes only 0.5% of fires to lightning.
3.1 Pasture burning and forest clearing
Using fire to transform forests into pastures and to repeatedly burn established pastures to improve grazing is a common and widespread practice in Tibet. Tafel (1914: 254) concluded from his observations: "As every pastoralist the Tibetan nomad is an ardent enemy of all forest [..]. He is recklessly burning down forests". Ward (1947: 73) recognized that "in forested areas, woody vegetation is cut and burnt to extend grazing, and considerable change results". In the River Gorge country, herds must descend in winter to the forest belt where there is no grazing at all except in artificial clearings. Thelaus (1992: 336) reports from the Zoigê Basin (Aba TAP) that "south aspects are repeatedly burnt and consequently dominated by grasses". About 200 km to the SSW of Zitsa Degu in the Wolong Panda Reserve (Wenchuan County, Aba TAP), which has been settled traditionally by Chiang (Qiang) people, Campbell (1984: 32) and Schaller (1993: 149) also observed slope burning for improving pasture. Johnson (1944: 737) observed around Dawu (Dawu County, Garzê TAP): "Winter is the dry season when travellers and new settlers often burn off the countryside for heat, to create ash for fertilizer or to clear land for farming. South slopes exposed to the sunshine and wind become dry. The fire destroys the conifer-rhododendron combination. North slopes remain snow covered and are moist. Here nature protects the conifers until man's ax moves in". Yang (1987: 76) sees "pasture burning" as one of the causes for repeated forest fires. Recently pasture burning has been banned in many counties. Yet a successful ban of burning resulted in widespread regrowth of shrubs, especially Rhododendron (pers. comm. Wu Ning). This policy is not very popular with pastoralists, since they are losing quality grazing grounds. Just to the north of Zitsa Degu in SW-Gansu Teng (1947a: 196) observed that the south slope's juniper and pine forests were destroyed for timber extraction or pasture extension and reduced to shrublands. "Repeated fires have been responsible for converting the scrub areas into grasslands. This condition has led certain observers to uphold the erroneous view that southern slopes are natural grassland". He suggests for the Qilian Shan, the north-eastern most mountains of the Plateau: "When the early tribesmen first settled in the region [..] they needed pasture on sunny aspects for their animals. Quite naturally they cut the junipers for various construction purposes and even burned up the woodland for pasturage. Owing to the inflammability of the bark of junipers and the xeric nature of the southerly slopes, the woodland was readily destroyed by fire and once destroyed, was difficult to restore. Thus it was turned into grassland and has been kept rather permanently so by continuous grazing. The savannah-like juniper stand is also an outcome of fire and grazing" (Teng 1947a: 189). Besides pasture extension by burning Heim (1933: 166) observed forest clearing by removing bark rings from firs in the Dawu area. Continuous grazing and periodical burning also prevents the occurrence and especially the spread of fire. Uncontrollable grassland fires which are common in other less intensely grazed grassland regions are absent. Furthermore the risk of unwanted fires and their spread is also reduced.
4 The direct impact of grazing
In addition to fire as a factor in forest fragmentation the direct impact of grazing must also be considered. The development and extension of pastoralism surely reduced wild ungulate populations, which might have also impacted vegetation, especially in favoured habitats such as along rivers or south slopes. So far research available on the impact of wildlife grazing in Tibet focuses on the grassland and steppe of north-west Tibet's Changtang (i.e. Schaller & Gu, 1994). However, the extent of deforestation in Tibet's forest region suggests that wild herbivores were not only replaced by domesticated herbivores, but that livestock now greatly outnumbers former wild ungulate populations. In Chamoling (Changmaoling) Reserve (Riwoqê County) for example a few hundred wild red deer (Cervus elaphus macneilli) share grazing grounds with thousands of livestock grazing historically deforested south slopes. South slopes are the preferred grazing grounds in winter, because snow thaws usually within a day or two. Although air temperature is low, the irradiation is still powerful due to the regions location in subtropic latitudes, the thin pure air in high elevation and the nearly vertical angle in which the insolation hits the steep slopes in winter. On the contrary northern slopes usually carry a snow cover throughout the winter, because air temperatures usually do not reach above freezing without direct insolation. The availability of fodder in winter is the crucial factor in determining herd size. Development and maintenance of winter grazing grounds is the prerequisite for successful livestock keeping. The south slope's winter pastures secure a reliable fodder source, sparing the labor intensive practice of fodder collection, which is necessary in regions with higher snow fall. However, in exceptional winters with harsh snowstorms, as in February 1996 in Yushu (Jeykundo) Prefecture (SW-Qinghai) and in the winter of 1997/98 in Nagqu Prefecture (TAR), all grazing grounds are covered with deep snow. In both cases hundreds of thousands of Yak starved and froze to death.
Rock (1956: 100) encountered the negative results of pasturing herds in Picea asperata forests growing along an upper Yellow River (Huang He / Ma Chu) tributary in south Amdo (S-Qinghai): "The nomads with their yak and sheep have ruined this region. From all appearances the forests covered once all the valley slopes, at least those facing north, but they are in a dying condition. There undergrowth [..] is a species of Mnium moss, and this has here almost entirely disappeared, and only where there are small groups of healthy trees to be found, this moss is also present, often over a foot in thickness covering the ground completely. Thousands of dead trees are witness to the evil work of the yak and sheep". Continuous forest grazing impedes, and in the long run, prohibits forest regeneration. Grazing animals are reducing undergrowth, which protects the soil from evaporation, or as described in the case of moss can retain a crucial amount of moisture. Animals trample and destroy seedlings with their sharp hooves. Furthermore they compact the topsoil. Compacted topsoil decreases the soil's capacity to retain moisture and reduces water infiltration. Moreover, forest regeneration is compromised by animals biting off shoots and branches of seedlings and young trees. While visiting a small patch of freshly planted spruces and inquiring about ripped out seedlings, members of the Riwoqê forest department attributed them to grazing yaks. In any case, selective grazing of livestock favours less browsed species, like Juniperus, over more palatable species, likeAbies. Due to reduced water availability on south slopes juniper, larch and pine forests are in general more open and thus supporting a ground cover rich in grass, sedges and forbes. These forests are predestined for forest grazing, assuming that their present open state is not merely a product of intentional burning and grazing as Teng (see above) assumed. Forest grazing also impacts possible fire patterns. Continuous grazing reduces the possibility that recurrent ground fires prevent the build up of inflammable debris, which might turn a relative harmless ground fire into a devastating crown fire.
5 Case study from Zitsa Degu / Jiuzhaigou (Aba TAP, Sichuan)
In Zitsa Degu (Jiuzhaigou) the differences in forest cover of south and north facing slopes were also clearly present (fig. 4, 5) Jiuzhaigou Nature Reserve (NNW-Sichuan, 103°46'-l04°05'E / 32°55'-33°20'N) is located in the eastern most part of the area mapped by Wissmann (see map, fig. 2).
Fig. 4: Altitudinal Belts of Zitsa Degu / Jiuzhaigou Forest Types and Fire Impact
Repeated burning keeps the slope free from shrubs and trees. Salix spp. initiates forest succession in the humid lower part of the slope. In the far left and right some forest stands remain on rock outcrops. On the right 'horizon' note the dead conifers killed from fire while burning the slope. The pasture is reserved for winter grazing. Photo: D. Winkler 2.8.1991 (3,100 m a.s.l., towards N)
Here the absence of forests on south-slopes is not as pronounced as in inner eastern Tibet or in the upper Min Jiang (Zhung Chu) catchment around Kungari Pass (Gonggang Ling), which is merely 20 km straight to the SW. In Zitsa Degu the absence of forests is most common on fully south-exposed slopes, usually above 3,000 m a.s.l. This more limited range of forest absence is explained by two factors: First precipitation in Zitsa Degu (744 mm/a in 2,400 m) is at the higher end of eastern Tibetan averages, where precipitation in valley grounds is between 500 mm to 800 mm/a. However climatic data often reflects only the extrazonal climate of dry valley grounds that receive considerably less precipitation than surrounding slopes. The higher precipitation gives the forests better conditions to reclaim former sites after human impact has been discontinued or reduced. Second and of greater importance, is the factor that human impact is not as pronounced as in other areas which have a higher concentration of population and livestock and probably a longer history of human settlement. Thus Zitsa Degu offers a very interesting opportunity for research. On one hand there is still enough pristine vegetation to study the environment in its original state, on the other hand areas already impacted by human management offer insights in to the stages of development of a cultural landscape.
Most forest-free slopes in Zitsa Degu are located on south aspects. And just as many eastern Tibetan slopes, they are not at all free of shrubs or trees. In the upper Ritse (Rize) Valley (fig. 6) a forest-free south-slope (3,000-3,400 m, see Winkler, 1994: 48/49) contains a variety of shrubs such as Lonicera hispida, Spiraea mongolica, Berberis dasystachya and Juniperus in shrub form. These species represent some of the most common genera of the Tibetan south-slope shrub flora, be it under forest cover or out in the open. On the lower half of the slope, close to the neighboring north slope spruce-fir forest, Betula albo-sinensis and Salix spp. are moving into the grassland initiating forest succession. Moreover, a single small spruce tree is growing on the upper slope, which has to be regarded as an indicator for potential forest growth. As Ellenberg (1988: 392) states discussing the treeline in European mountains, "where one tree grows then others can grow near it if man and animals permit, provided the soil is deep enough". Thus even one small tree or little grove can testify to the general possibility of forest growth and probably the existence of former forests.
Fig. 6: Forest distribution pattern created by fire in Juizhaigou / Zitsa Degu above "Panda Lake"
The upper reaches of the south slopes are especially heavily impacted by fire, which has destroyed most forests. Forest fragmentation is more common in the cloud forest belt (2,700 m-3,500 m). Here forest stands remain most commonly along creeks and on rocky outcrops. The E-W running ridges are lined by dark forests reaching over from the densely forested north slope. Photo: D. Winkler 31.7.1991 (2,650 m a.s.l. Towards N)
On the upper Ritse Valley's south slope soil test sites were investigated and the soil profiles showed deeply developed Cambisols on the south slope, fully capable of supporting forest growth. While both profiles (3,090 m; 3,440 m) showed pedogenetic processes active until at least 0.9 m, the upper site was more similar to the soil types typical for the more humid north slope's soils. Its plastic loamy B-horizon was more pronounced, yet not as brightly brownish developed as typical for the north slope's Chromic Cambisol (Terra fusca). Traces of solifluidal processes or solifluction, which could prevent tree growth by periodically injuring the root system were absent. Furthermore these processes are not to be expected several hundred meters below the upper treeline. Heim (1936: 449) observed solifluidal processes in the Minyak Gonkar (Gongga Shan) region around and above the treeline. The same applies for possible climatic changes as a hypothesis to explain the present widespread absence of forests on south-slopes. Although it is not clear up to which altitude forests could reclaim south slopes under present climatic and pedological conditions, the answer is not as central as could be assumed. If the phenomenon were only to be encountered some hundred meters below the present treeline, recent climatic changes must be taken into account, as Miehe (1996) exemplifies analyzing the distribution patterns of alpine Cyperaceae turfs in S-Tibet. Yet recent climatic change cannot explain the absence of south-slope forests sometimes up to 1,000 altitudinal metres below the north slope's present treeline. Moreover, not only single trees or stands are left on some of Zitsa Degu's south aspects, but in many locations true forests. There is at least one fully south exposed juniper forest at the Yala Pass in 3,600 m a.s.l. (Winkler, 1994: 81), but probably more stands are to be found in inaccessible sites. The junipers (here Juniperus convallium and J. saltuaria) start to build pure forests onwards from around the upper limit of condensation level, which is to be found at 3,500 m (see fig. 4). In the upper part of the condensation level (3,200-3,500 m a.s.l.) junipers are found as admixture in the sub-canopy. The southern slopes top-canopy consists of Picea purpurea, Abies faxoniana and in sites with initial soil development or very shallow soil Larix potaninii. Passing through Tibet's forest region, lone trees, small stands or forest patches on south-facing slopes are not an exception, rather the rule. Among others Cheng (1939: 24, 187), Ku & Cheo (1941: 91), Teng (l947a: 190; 1947b: 227), Duncan (1952: 186), Wang (1961: 55), and Li (1993: 60) report juniper forests on south-facing slopes. Even Schäfer's (1938: 56) photos of the forest-free south slopes depict some conifers within the grasslands.
5.1 The impact of fire
In Zitsa Degu traces of fire impact were abundant in many locations (Winkler, 1994: 31-34, 48, 73, 85, 86). In general the impact of fire was most evident on south-facing slopes. North slopes were only inflicted from valley ground in 2,000 m a.s.l., which is influenced by a dry valley climate, up to the lower limit of the cloud forest belt at around 2,700 m a.s.l. The higher precipitation (ca. 1,000 mm/a) in the condensation level (2,700-3,500 m a.s.l.) and the lower evaporation of the north slope prevents the occurrence of fires due to humid conditions. Wide areas between 2,000 m to max. 2,700 m a.s.l. were impacted by devastating fires in the 1950s. In many places this fire apparently destroyed all former forests. Around the village of Shukcho (Shuzheng) only in a few places some old growth survived in the form of single pines (Pinus tabulaeformis) or rare stands of spruce (Picea wilsonii). In addition, one small stand of deciduous forest remained, containing oaks (Quercus aliena, syn. Quercus liaotungensis see Chen 1987: 213) and Carpinus turczaninowii. Otherwise presently this area is mostly under pine-oak forests of around 40 years of age. Pinus tabulaeformis dominates the top canopy; the sub-canopy consists mostly of oak (Quercus aliena). In the undergrowth there are a few seedlings or young trees of Picea wilsonii. A mostly north exposed area had been spared by the 'great fire' and is presently used as a source of timber. Locals were removing mature conifers through selective cutting. Here in 2,450 m a.s.l. a WSW-exposed stand of Pinus tabulaeformis-Quercus aliena forest showed a more recent impact of a limited ground fire, probably caused by man. In addition the bases of mature pines showed multiple impact of fire. No spruce was to be found in this stand (Winkler, 1994: 36). In many more sites in the pine-oak belt multiple impact of fire was detected on old growth trees.
The present dominance and the wide distribution of Pinus tabulaeformis and Quercus aliena below the cloud forest belt in Zitsa Degu clearly indicates that their dominance is closely related to the impact of fire. The thick fire protective bark enables pines to coexist with ground fires, which are frequent events in many pine forests. Furthermore pine seedlings are very competitive on freshly burned over and highly exposed slopes. Quercus is favoured by its sprouting ability. Even if the above ground part is destroyed by fire, its rootstock is able to sprout again, an ability also employed by locals in coppice systems. Already Wang (1961: 54) concluded that the distribution of Pinus densata is favoured by the impact of fire, since its stands are most common around settlements. Teng (1940: 368) regards NW-Yunnan's Pinus yunnanensis forests below 3,350 m a.s.l. as a subclimax community "kept more or less permanent as a consequence of repeated burning and clear cutting". Moreover, he observed a connection between fire and the abundance of oak (Quercus semecarpifolia); this evergreen oak is "frequently occurring in pure stands as a successional stage after fire" (Teng: 381). The presence of spruce seedlings around Shukcho indicates that the present prevalence of pure pine-oak forests is a transitional stage towards a climax of spruce-mixed forests, richer in biodiversity.
Fig. 7: A recent stand-replacing crown-fire completely destroyed this SE-slope forest in Zitsa Degu's upper Ritse Valley
Photo: D. Winkler 2.8.1991 (3,450 m a.s.l., Towards E)
In the fir-spruce dominated cloud forest belt (2,700 m-3,500 m a.s.l.) and the subalpine belt (3,500 m-3,800 m a.s.l.) no traces of ground fires were observed. Past fires manifested as stand replacing crown fires (fig. 7). In recent decades in the lower Ritse Valley above Panda Lake crown fires burned several south-facing slopes from base to crest (fig. 5). However, untouched were several stands along creeks and open forests on rocky cliffs, where the lack of ground cover and the wide spacing of trees prohibited the spread of fire. The encountered patterns of fire impact reflect general differences in the content of humidity in soil and litter. This fire caused pattern of forest fragmentation is observable in many otherwise forest-free south slopes of the region, and was already reported by Tafel (1914: 81) and is to be seen in Schäfer's photographs (1938: 56/57). The adjoining north-facing slopes were untouched by fire, since most fires occur in winter or early spring, when north-slopes are snow covered or at least saturated with water from thaw. This creates a striking impression along E-W running ridges. Their south slopes are mostly bare of forests, and the top of the ridges are lined by dark forests reaching over from the north slope, a pattern also common in the area of forest- free south slopes elsewhere in Tibet and described by Hilbig (1987: 2) from Mongolia. In several transition areas from forest-free south to forested north slopes a few burned trunks (above Swan Lake, W of Ritse) or small burned stands (above Danzhou) remained fragmented along the forest edge, indicating that the fires, which caused limited havoc, reached these trees from the adjoining forest-free south slope. While researching a forest-free south slope (3,000-3,400 m a.s.l.) above Swan Lake local Tibetan guides confirmed that this slope is burned repeatedly for better grazing. In early August there were no traces of grazing. Upon inquiring it was pointed out that this south slope is reserved for winter grazing.
Also in Riwoqê County several sites showed recent fire impact, which destroyed spruce and spruce-juniper forests (i.e. between Charmed and Sibta, near Takzham). Around Riwoqê's remote Yiri village many slopes had been exposed to fire (fig. 3) in recent decades, evidently to create new pastures in an otherwise well wooded area. The transition from wilderness to cultural landscape is clearly underway.
6 Evidence of prehistoric and historic human impact in Tibet
Having presented the herders' rationale in developing south slope forests into pastures and the negative impact on forests of an expanding pastoralism, we clearly have to dismiss the hypothesis that the widespread absence of forests on south-facing slopes in otherwise forested areas is caused simply by natural conditions, such as lower water availability, high diurnal temperature ranges, intense insolation and differing pedological conditions or the combination of all of these factors. Off course, all these factors impede reforestation after forest destruction. Furthermore there are locations where some of these conditions are actually the cause of present non-forestation, i.e. climatic conditions within the dry valleys.
Pollen analyses suggest that human impact on forests in the Zoigê Basin (Ruo`ergai / Hongyuan Counties, NW-Sichuan), which is about 125 km to the east of Zitsa Degu, dates back to Neolithic times. According to Frenzel (1994) conifer pollen in peat sediments start to decline 5,100 yr B.P. From 2,000 yr B.P. tree pollen reduction clearly accelerated, being replaced by grass pollen. At both times there is no climatic evidence, which could explain such a change in vegetation. Thelaus reports that at 4,000 yr B.P. a reduction of tree pollen, especially Picea, occurred, while Juniperus pollen sedimentation still continued, but all tree pollen declined strongly from 2,500 yr B.P., once again without evident climatic causation. This suggests that already back then human impact had started to change the environment. However, according to Schlütz (1995) pollen analyses from the Nyenpo Yurtse Range (Nianbaoyeze, Qinghai-Sichuan border) do not reflect a more extensive forest cover there in the past.
Traces of increased human presence in this region are found from around 7,500 to 3,000 yr B.P. (Huang, 1994:214). Furthermore Frenzel et al. (1995: 493) found in many locations in eastern Tibet forest soils under present steppe-soils. Traces of agricultural activities can be dated back to 5,000 yr B.P. in Kharu (Guro) just north of Qamdo town in the Mekong valley (Za Chu; Lancang Jiang). Presently this stretch of the Mekong valley is completely deforested and covered only by low, open shrub vegetation.
Any attempt to quantify the full extent of historical forest reduction in Tibet's forest region remains a guess. Frenzel (1994: 160) gives over 70 localized estimates on present forestation versus potential forestation along two expedition routes. But there will be no clear quantification possible until remote sensing techniques backed up by precise localized studies have been applied. So far these technologies have not produced any satisfying results, partially due to technical difficulties presented by north-slope forest detection, partially because no major study has been undertaken or at least been published (see Winkler, 1998a). Still the nature of the phenomenon allows an estimate. Besides comparatively rare alluvial areas or plateaus below treeline (which are presently mostly forest-free in the core areas of Tibetan civilization anyway), the landscape is evenly divided into south slopes and north slopes. Thus, the area of historical deforestation of south slopes might comprise anywhere around 40% of the region's potential forest area.
From the above stated it is clear that Tibet's forest area has clearly diminished within the last millennia, occurring even before the onslaught of commercial timber extraction through Chinese modernization in the 1950s. Beyond doubt the development and expansion of pastoralism has dramatically reduced the forest area. Such an extensive impact was only possible through utilizing fire as an agent to clear primary forests. Continuous grazing and repeated burning prevents forest regrowth. These land-use strategies influence strongly the present forest distribution patterns. Eastern Tibet's forest-free south slopes bear witness to herders' efforts to replace forests with pastures as exemplified by the case study of Zitsa Degu / Jiuzhaigou. Thus wide areas of the potentially forested region of Tibet have to be regarded as cultural landscape rather than a wilderness presenting natural vegetation.
The need for further fieldwork to clarify the issues raised above is evident. Finding the answers is not only of academic interest, but also of importance for the future of Tibet. Traditional land-use techniques have been altered with the recent imposition of a state-planned economy, which is presently in transition towards a market economy. Traditional subsistence economy is being replaced at a fast paste. The resources of the plateau's fragile high altitude ecosystems are now exposed to high pressure from the outside, this is especially true for the forests and grasslands, the forestry and animal husbandry sector being the economic backbone of Tibet. Both traditional indigenous knowledge and modern scientific insight are needed to ensure that present changes will favour the implementation of truly sustainable ecosystem management strategies for the benefit of Tibet's population, its wildlife and the whole ecosystem, as much as for the nearly two billion people living along and depending on the rivers originating from the Tibetan Plateau.
Arend, J.L. (1950): Influence of Fire and Soil on Distribution of Eastern Redcedar in the Ozarks. Journal Forestry 48(2): 129-130.
Bird, M.I. (1995): Fire, Prehistoric Humanity, and the Environment. Interdisciplinary Science Revue 20(2): 140-154.
Bräuning, A. (1994): Dendrochronology for the last 1.400 years in Eastern Tibet. GeoJournal 34(1): 75-95.
Campbell, J.J.N. (1984): Natural History and Conservation Problems of Forests, Bamboos and Giant Pandas in the Sichuan Region - Ecological Background, WWF Giant Panda Project (unpubl.): 95 pp.
Chen, C.G. (1987): Standörtliche, vegetationskundliche und waldbauliche Analyse chinesischer Gebirgsnadelwälder und Anwendung alpiner Gebirgswaldbau-Methoden im chinesischen fichtenreichen Gebirgsnadelwald. Diss. Univ. Bodenkultur Wien 30: 316 pp.
Cheng, W.C. (1939): Les forets du Se-tchouan et du Sikang oriental. Trav. Lab. For. Toulouse, Sect.1, Vol.1, Art.2: 231 pp.
Chow, C.K. (1947): Height Growth of Seedlings of Important Timber Trees in the Upper Tohwo Valley Growth, Kansu. Botanical Bulletin Academia Sinica 1: 201-202.
Dorje, G. (1996): Tibet Handbook with Bhutan. Bath: 768 pp.
Duncan, M.H. (1952): The Yangtze and the Yak. Alexandria, Va: 353 pp.
Ellenberg, H. (1988): Vegetation Ecology of Central Europe. 4th Ed., Cambridge.
Frenzel, B. (1994): Über Probleme der holozänen Vegetationsgeschichte Osttibets. Göttinger Geographische Abhandlungen 95: 143-166.
Frenzel, B., J. Li & S.J. Liu (1995): On the Upper Quaternary Palaeoecology of Eastern Tibet - Preliminary Results of an Expedition to the Eastern Tibetan Plateau, 1992. Science in China (series B) 38(4): 485-494.
Hanson-Lowe, J. (1940): A Journey along the Chinese-Tibetan Border. The Geographical Journal 95: 357-367.
Heim, A. (1933): Minya Gonkar, Forschungsreise ins Hochgebirge von Chinesisch Tibet. Bern, Berlin: 237 pp.
Heim, A. (1936): The Glaciation and Solifluction of Minya Gonkar. The Geographical Journal 87: 444-454.
Hilbig, W. (1987): Zur Problematik der ursprünglichen Waldverbreitung in der Mongolischen Volksrepublik. Flora 179: 1-15.
Huang, R.F. (1987): Vegetation of the Periglacial Zone in the north-eastern Part of the Qinghai-Xizang (Tibet) Plateau. In: Hövermann, J. & Wang Weying (eds.): Reports on the north-eastern Part of the Qinghai-Xizang (Tibet) Plateau by the Sino-W-German Scientific Expedition, Beijing: 438-495.
Huang, W.W. (1994): The Prehistoric Human Occupation of the Xizang-Qinghai Plateau. Göttinger Geographische Abhandlungen 95: 201-219.
Johnson, R.G. (1944): Exploring a Grass Wonderland of Wild West China. National Geographic Magazine 85(6): 713-742.
Ku, C.C. & Y.C. Cheo (1941): A Preliminary Survey of the Forests in W-China. Sinensia 12: 81-113.
Li, B.S. (1993): The Alpine Timberline of Tibet. In: Alden, J., J.L. Mastrantonio & S. Odum (eds.): Forest Development in Cold Climates, NYC.: 511-527.
Li, W.H. (1993): Forests of the Himalayan Hengduan Mountains of China and a Strategy for their Sustainable Development. ICIMOD, Kathmandu: 175 pp.
Liu, D.Y. (1994): Preliminary Investigation of Forest Fire Regularities in south-eastern Xizang. Forest Fire Prevention 3: 25-27 (in Chinese).
Ma, Z.G., J.X. Wang & K.H. Mu (1991): Forest Fires in Sichuan's Pinus yunnanensis-Forests and Classification into Fire Danger Zones. Resource Development and Conservation 7(4): 266-268 (in Chinese).
Mackinnon, J., M. Sha, C. Cheung, G. Carey, X. Zhu & D. Melville (1996): A Biodiversity Review of China. WWF International, Hongkong: 530 pp.
Miehe, G. (1990): Langtang Himal - Flora und Vegetation als Klimazeiger und -zeugen im Himalaya. A Prodromus of the Vegetation Ecology of the Himalayas. Mit einer kommentierten Flechtenliste von Josef Poelt. Dissertationes Botanicae 158, Stuttgart: 529 pp.
Miehe, G. (1996): On the Connexion of Vegetation Dynamics with Climatic Changes in High Asia. Palaeogeography, Palaeoclimatology, Palaeoecology 120: 5-24.
Miehe, S.; Miehe, G. Huang, Jian; Otsu Tsewang; Tuntsu Tseren; Tu, Yanli (2000): Sacred Forests of South Central Xizang and Their Importance for the Restauration of Forest Resources. In: Environmental Change in High Asia edited by G.Miehe & Zhang Yili, Marburger Geogr. Schriften 135, 228-249.
Ogilvie, J. (1996): Forestry in Diqin Prefecture, NW-Yunnan, S-China. Commonwealth Forest Review 75.4: 290-295.
Rock, J.F. (1930): Seeking the Mountains of Mystery - An Expedition on the China-Tibet Frontier to the Unexplored Amnye Machen Range. National Geographic Magazine 57: 131-163.
Rock, J.F. (1956): The Amnye Ma-chen Range and Adjacent Regions. Ser. Orientale 12: 1-194.
Rockhill, W.W. (1891): The Land of the Lamas. London (reprint 1991, Delhi):
Schäfer, E. (1938): Ornithologische Ergebnisse zweier Forschungsreisen nach Tibet. Journal für Ornithologie 86, Sonderheft, Berlin: 350 pp.
Schaller, G.B. (1993): The Last Panda. Chicago: 299 pp.
Schaller, G.B. & B.Y. Gu (1994): Ungulates in NW Tibet. National Geographic Research 10.3: 266-293.
Schickhoff, U. (1995): Himalayan Forest-Cover Changes in Historical Perspective: A Case Study in the Kaghan Valley, northern Pakistan. Mountain Research Development 15.1: 3-18.
Schlütz, F. (1995): Zur holozänen Vegetationsgeschichte im Gebiet des Nianbaoyeze an der Grenze der Provinzen Qinghai und Sichuan. Göttinger Geographische Abhandlungen 102: 185-200.
Schmidt-Vogt, D. (1990): Fire in high altitude forests of the Nepal Himalaya. In: Fire in Ecosystem Dynamics, J.G. Goldammer & M.J. Jenkins (eds.), 3rd Intern. Symp. Fire Ecology: 191-199.
Schweinfurth, U. (1972): The Eastern Marches of High Asia and the River Gorge Country. In: Erdwissenschaftliche Forschung 5, C. Troll (ed.), Wiesbaden: 276-287.
Sichuan Senlin. Editorial Board of "Forests in Sichuan" (1992): 1535 pp. (Chin. / Engl. abstract).
Tafel, A. (1914): Meine Tibetreise, 2 vol., Stuttgart: 694 pp.
Teng, S.C. (1940): Studies of Chinese Timber Trees in Reference to Forest Management I. Sinensia 11: 363-395.
Teng, S.C. (1947a): The Forest Regions of Kansu and their Ecological Aspect. Botanical Bulletin Academia Sinica 1.3: 187-200.
Teng, S.C. (1947b): Silviculture of Kansu Trees. Botanical Bulletin Academia Sinica 1: 221-241.
Thelaus, M. (1992): Some Characteristics of the Mire Development in Hongyuan County, Eastern Tibetan Plateau. Proceed. 9th Int. Peat Congress June 1992, Vol.1, Uppsala: 334-351.
Wang, C.W. (1961): The forests of China with a survey of grassland and desert vegetation, M.M. Cabot Found. 5, Cambridge: 313 pp.
Wang, S.M. & S.X. Zhong (1995): Ecological and Geographical Distribution of Larix and Cultivation of its Major Species in SW China. In: W.C. Schmidt & K.J. Mcdonald (eds.) Ecology and Management of Larix Forests, Ogden: 38-40.
Ward, F.K. (1947). Tibet as a grazing land. The Geographical Journal 110: 60-76.
Weigold, H. (1935). Südost-Tibet als Lebensraum. Jahrbuch Geographische Gesellschaft Hannover: 203-247.
Winkler, D. (1994): Die Waldvegetation in der Ostabdachung des Tibetischen Hochlandes - Dreidimensionale Vegetationszonierung mit Bodenuntersuchungen am Beispiel des Jiuzhai-Tals in NNW-Sichuan. Berliner Geographische Abhandlungen - Beiheft 2, Berlin: 128 pp.
Winkler, D. (1997): Waldvegetation in der Ostabdachung des Tibetischen Hochlands und die historische und gegenwärtige Entwaldung - Beispiel Jiuzhaigou (Zitsa Degu; NNW-Sichuan). Erdkunde 51.2: 143- 163.
Winkler, D. (1998a): Deforestation in Eastern Tibet: Human Impact - Past and Present: In: G.E. Clarke (ed.) Development, Society and Environment in Tibet, Proceed. 7th Seminar Intern. Assoc. Tibetan Studies (lATS) 1995 - Seggau, Austria; Vienna: 79-96.
Winkler, D. (1998b): The Forests of the Eastern Part of the Tibetan Plateau - A Case Study from Jiuzhaigou (Zitsa Degu; NNW Sichuan). Plant Research and Development 47/48, Tübingen: 184-210.
Wissmann, H. v. (1960;61). Stufen und Gürtel der Vegetation und des Klimas in Hochasien und seinen Randgebieten. A: Hygrische Raumgliederung und Exposition; B: Thermische Raumgliederung und Frostböden. Erdkunde 14: 249-272; 15: 19-44.
Yang, Yupo (1987): Alpine forests in Western Sichuan, China and the effects of forest management. In: T. Fujimori & M. Kimura (eds.) Human Impact and Management of Mountain Forests. Proceed. Intern. Workshop (lUFRO), Ibaraki, Japan: 67-79.
Zheng, D. (1986): A comparative study of geoecological and environmental problems between the Himalayas and the Hengduan Mountains. Proceed. Intern. Workshop "Watershed Management", Chengdu (ICIMOD & CAS), Kathmandu: 25-38.