For an area of low primary productivity, the Tibetan plateau contains a surprising diversity of large mammal species. Although detailed ecological studies are still lacking, it appears that most large mammals compensate for paucity of resources by occasional migratory movements or by having large home ranges. Thus, long-term persistence of these species requires that human alterations of, or disturbances to, their wild habitat be either restricted geographically or of limited intensity. It is not surprising that all species of Tibetan plateau ungulates are listed on either Appendix I or II of CITES, and under various classifications in the IUCD Red List (Wilson & Reeder 1993, Hilton-Taylor 2000). Nomadic pastoralism has been the predominant land-use of the Tibetan plateau for centuries, and most wild species have long since accommodated to such low-intensity use. With the recent push to ‘exploit’ China's western regions (China Daily 2001, Glantz, Ye & Ge 2001), socio-economic conditions on the Tibetan plateau are undergoing rapid fluxes, and thus the future of this unique fauna is in question.

Although situated in the neighbouring province of Qinghai rather than in the Tibetan Autonomous Prefecture itself, a portion of the Kunlun Mountains colloquially named Yeniugou (literally ‘wild yak valley’) has for years been recognized by Chinese scientists and authorities as a stronghold of Tibetan plateau ungulates (Cai 1997, Cai, Liu & O'Gara 1990), particularly the wild yak Bos grunniens. Although not completely clear, we suspect that the principle reason for the relatively high densities of large mammals in Yeniugou is geographical: relatively high amounts of precipitation have produced glaciers and permanent snow fields along the north-facing slopes of the Kunlun Mountains, whereas much drier and somewhat lower slopes characterize the mountains just a few kilometres to the north. The diversity of habitats caused by the proximity of lush, snow-fed meadows and dry, desert conditions evidently allows for high reproduction to offset the inevitable human-caused mortality. Despite its importance, Yeniugou has never been accorded any official conservation status, and land use continues to be dominated by pastoralism. In 1995, approximately 45,000 km² of lands directly to the south and west of Yeniugou were designated as the Kekexili Nature Reserve, which was elevated to ‘national level’ status in 1997 in an attempt to protect the declining herds of Tibetan antelope Pantholops hodgsoni (also called chiru). However, Yeniugou was not included within Kekexili, and thus wildlife congregating there is protected only by the national wildlife law that bans poaching.

Beginning in 1990, collaborative surveys by the Chinese Academy of Sciences and the University of Montana (UM), USA, have documented the status and conservation prospects of wildlife in Yeniugou (Harris 1993, Harris 1995, Harris & Miller 1995, Harris, Miller, Cai & Pletscher 1996, Miller, Harris & Cai 1994). In 1997, a group from UM returned to Yeniugou to assess trends during the 1990s (Harris, Pletscher, Loggers & Miller 1999). The major finding of that survey was that resident Tibetan antelope, which had been numerically dominant in the early 1990s, had been decimated by poaching. Most other species had evidently remained relatively stable, although we noted troubling declines in both blue sheep Pseudois nayaur and argali Ovis ammon (Harris et al. 1999). Here, we report on an additional follow-up survey we conducted in 2002.

Study area

Yeniugou is an interior valley within the Kunlun Mountains in Haixi Mongolian People's Autonomous Prefecture, situated in Qinghai Province at approximately 35° 50′N, 91–93° E (Fig. 1). Elevations on the valley floor vary roughly within 3,800–4,200 m a.s.l.. The main Kunlun range (elevations up to 5,921 m a.s.l.) lies to the south of the Yeniugou River, and isolated massifs of up to 5,521 m a.s.l. in the north. The valley totals an area of approximately 3,900 km², although studies were conducted only within a smaller, core area of approximately 1,200 km². Yeniugou is characterized by high elevations, low annual precipitation and shallow soils. Vegetation consists predominately of graminoids (notably Stipa, Kobresia and Carex spp.) and forbs. Trees do not grow at these elevations. Height of vegetation ground cover tends to be low (5–25 cm; Zhou 1990, Harris & Miller 1995), and parts of the area at higher elevations are devoid of all vegetation. Further information on vegetation in Yeniugou can be found in Harris (1993) and Miller & Bedunah (1994).

Figure 1.

Location of the Yeniugou study area within the province of Qinghai, China with inset showing the location of Qinghai within China.

Yeniugou is situated in a transition zone between traditionally Mongol and Tibetan ethnic areas. During the 1930s, a group of Kazaks seeking refuge from persecution in Xinjiang also began settling in the area. Tibetans were the original inhabitants of Yeniugou, but they were removed in the early 1950s, and their grazing rights revoked. From the early 1950s through 1983, grazing rights belonged to the Kazaks. In 1984, the area was officially handed over to a group of Mongol pastoralists when the majority of Kazaks moved back to Xinjiang (Qinghai Bianjizu 1985). Thus, three different ethnic groups have occupied Yeniugou in the past 45 years.

Mongol pastoralists initially used Yeniugou only during summer returning to winter pastures near the town of Golmud each year. There appeared to be little systematic allocation of pastures among pastoral groups in the early 1990s, and drainages occupied by an encampment in a given summer would not necessarily be occupied by the same, or indeed any, pastoralist the following summer (Harris 1993). In recent years, government policies have increasingly stressed a clarification of grazing boundaries and fixing of grazing seasons. The number of Mongol pastoralists using the valley has evidently remained approximately stable, but resident pastoralists have adopted a more sedentary style of animal husbandry. These pastoralists now spend all year within Yeniugou, and boundaries of both summer and winter pastures are defined and allocated to specific families. According to Kazak informants, who witnessed both periods, two phenomena characterized the transition from Kazak to Mongol occupancy of Yeniugou: 1) the movement of gold miners, based in eastern Qinghai, through Yeniugou to suspected deposits to the west, and 2) commercial hunting of wildlife by outsiders, primarily of yaks and argali, and primarily in winter. Both of these activities were illegal; in practice, both were virtually uncontrolled. Further information on the cultural history of the area is reported in Harris (1993).

Since the adoption by national authorities of the ‘Great Western Development Strategy’ in 1999 (Crisp & Tang 2000), remote areas such as Haixi Prefecture in western Qinghai (in which Yeniugou lies) have been the focus of considerable economic development activity. The economic centre of Haixi, Golmud, doubled in population between 1984 and 2001 (Wu & Li 2001). However, China's extensive rail transportation network had never penetrated onto the Tibetan plateau (despite failed efforts to do so in the 1950s), and it was in the context of the greater, regional strategy that the central government announced in February 2001 the initiation of work on a rail line connecting Golmud in western Qinghai with Lhasa in the Tibetan Autonomous Regions (Economist 2001). Construction teams were quickly mobilized, and by July 2001 a massive construction project involving an estimated 400,000 workers was underway along the present route of the motorway connecting the two cites (Wu & Li 2001). The railroad under construction passes approximately 50 km from the nearest part of the study area.

Methods

We spent the period 3–22 September 2002 in Yeniugou. We interviewed local pastoralists encountered, as well as county officials in Golmud and Xining, Qinghai's provincial capital. We observed wildlife during foot-surveys along predetermined routes (vehicles were used to transport base camps along the main river valley). To maximize the validity of comparisons, we used the same methods as during our 1997 surveys and surveyed the same subdrainages on approximately the same calendar date (Fig. 2). We documented all travel routes on 1:100,000 topographic maps and later calculated distances travelled. Most observations were aided by binoculars or 20–42x zoom spotting scopes. We focused our attention on Yeniugou's wild ungulate fauna, particularly the three species we believed most susceptible to recent declines at the hands of humans: wild yak, Tibetan antelope and argali. Time constraints did not allow us to conduct distance sampling surveys for Tibetan wild ass Equus kiang, Tibetan gazelle Procapra picticaudata and Tibetan antelope Pantholops hodgsoni as we had during 1991; violation of critical assumptions rendered distance sampling inappropriate for wild yak, blue sheep and argali in any case (Harris 1993,1996). When appropriate, we estimated the probability that multiple sightings in the same general area represented duplicate counts and adjusted our counts accordingly (Harris 1993,1994). We make no claim that the number of animals observed during any of the four survey periods represent fully unbiased population estimates as we can not account for animals that may never have been observed, and only have qualitative means to assess varying detectability. However, by replicating observations under similar conditions (Table 1), we believe that gross population changes were unlikely to be confounded with counting inaccuracies.

Results

Wild yak

Our best estimate of the number of wild yaks observed during September 2002 is 1,697; about 400 more than we observed in the same area during September 1997 (see Table 1). Yak groups often move among subdrain-ages, particularly during the transition from summer to autumn foraging sites (Harris & Miller 1995), which makes assessment of possible duplicate counts difficult. However, in September 2002 most yaks occurred in large, reasonably distinguishable maternal groups (including all sex/age classes) that were relatively sedentary during the survey period, and on a number of occasions we were able to document what appeared to be the same group in approximately the same location during intervals as long as 17 days. Thus, we believe that any duplicate counting of individuals was negligible.

As in earlier surveys, most large yak groups occurred along the north-facing slopes characterized by moist, Kobresia-dominated sedge meadows, typically at 4,500–4,800 m a.s.l. (Fig. 3). Also as in previous years, we noted scattered individual adult males throughout the study area, especially in the drier, south-facing slopes north of the Yeniugou River. However, unlike in previous years, we also noted large maternal groups north of the river. In contrast to the north-facing slopes fed by glaciers and snow-fields, these drier valleys have only small, scattered sedge meadows, but most of these were occupied by large groups of yaks. In 2002, we tallied 379 of the total 1,697 animals counted north of the river, whereas in 1997, only 120 of the estimated 1,297 animals were observed north of the river.

Table 1.

Survey conditions and estimated number of animals of each species observed in the large mammal surveys conducted in Yeniugou, Qinghai Province, China, during 1991–2002. Confidence limits (given in parentheses) apply to number of animals observed, not necessarily to animals present.

Figure 2.

Study area within Yeniugou, showing routes walked (—) and driven (- -) during the September 2002 survey. The main crest of the Kunlun Mountains (elevation markers 5762 and 5921 m) bounded our surveys to the south, the Yeniugou river flows west to east through the middle of the map and the blank area in upper left comer is the Black Lake.

Prior to this survey, we had no reason to suspect a history of hybridization between wild and domestic yaks. However, in 2002 we observed a single white-faced yak. No domestic yaks were present in Yeniugou during 2002, but a herd of about 25 had been kept by a Mongol pastoralist during 1996–1997.

Argali

Our best estimate of the number of argali we observed during September 2002 was 94 (see Table 1) which is substantially below our estimates of 141 animals observed in 1997 (Harris et al. 1999), 191 in 1992 and 245 (95% CI = 238–256) in 1991 (Harris 1993). Of the 94 argali, 18 were adult males and 15 were lambs (we were unable to quantify yearlings because some maternal groups were too distant at the time of observation). Argali are the most difficult ungulate species to find in Yeniugou, but we have little doubt that they were less numerous within the study area than in earlier years. Weather conditions during September 2002 were often cloudy and snowy; thus, it is possible that proportionally more argali than in previous years escaped our detection. However, the observation conditions were even more limiting in 1992 when we observed 191 animals (at least 67 of which were adult males, suggesting that the abundance in 1992 was probably similar to the higher number of 245 observed during 1991; Harris et al. 1999). Additionally, we attempted to compensate for poorer sighting conditions by spending more observation time in likely argali habitat as the survey progressed, and as we realized we were seeing fewer argali than in the past. We saw very few argali tracks, faecal material or skulls, which further confirmed that the cloudy conditions were most likely not the sole cause for our observing so few animals directly by sight.

The spatial distribution of argali also appeared to have changed considerably from similar time periods in previous years, with many fewer seen near mid-elevation Stipa-dominated grasslands, and relatively more seen in drier, rocky slopes, where one would more likely expect to find blue sheep. Of the 94 animals, 37 were observed further west (and further from pastoralists) than we had recorded during any earlier survey. In contrast, we saw only two ewes in the Hongshanbao area and no argali at all in the Ainiwa'er area, both of which (routinely) held large groups in the early 1990s (Fig. 4).

Figure 3.

Approximate locations of yak groups observed during the September 2002 survey in Yeniugou, Qinghai. Large circles demarcate locations of large maternal groups with numbers observed indicated; smaller circles represent smaller groups or scattered individual animals.

Other species

White-lipped deer Cervus albirostris, which are generally associated with more mesic grasslands further east within the Tibetan plateau (Kaji, Ohtaishi, Miura & Wu 1989, Schaller 1998), and which were considered rare prior to our surveys in the 1990s, seem to have continued their gradual increase (see Table 1). We observed 111–127 individuals. In contrast to wild yaks in Yeniugou, white-lipped deer appeared frequently to move among drainages, and we were rarely able to confirm the presence of a group by observing it again on a subsequent day. Our uncertainty arises because of a group of 15 animals that appeared similar in composition to a group of 16 seen nine days earlier at a position 21.5 km away.

Figure 4.

Approximate locations and numbers of argali observed during the September 2002 survey in Yeniugou, Qinghai (), and locations at which argali had been observed during similar time of year in 1991, 1992 and 1997 (). Light, shaded areas bounded by dashed lines represent the approximate extent of regions affected by pastoral activity during September 2002. Dark shaded areas bounded by solid lines represent the approximate extent of regions affected by winter domestic livestock grazing, as indicated by location of encampments and evidence of past grazing.

Our best estimate of the number of blue sheep observed during September 2002 was 701, somewhat fewer than the 839 estimated observed during 1997 (Harris et al. 1999) and the 1,150–1,261 estimated observed in 1991 (see Table 1). Blue sheep groups were found in most, but not all of the previous locations, but, when found, group size was generally similar to those documented in previous surveys. Weather conditions often made observing blue sheep difficult; thus we suspect under-counting was more pronounced for blue sheep than other species.

We estimate that we observed 506 individual Tibetan wild ass during September 2002, which is similar to the total of 418 seen in 1997 (Harris et al. 1999), fewer than our 1991 estimate of 843 wild ass (95% CI = 618–1,052), but more than we observed during 1992 when < 100 were documented (see Table 1). We believe that the wide disparity in counts of wild ass reflects their mobility and variable choice of location for resource availability rather than true changes in abundance.

Inability to conduct distance sampling (line-transect estimates) during 2002 precluded us from quantifying the number of Tibetan gazelle. However, frequency of observation, group size and group distribution suggested that gazelle abundance was roughly similar to the 1,511 (95% CI = 1,037–1,985) estimated in 1991.

Distance sampling was unnecessary to quantify Tibetan antelope, numerically the most abundant ungulate in Yeniugou during the early 1990s, with an estimated 2,076 (95% CI = 927–3,247) present in 1991. Despite extensive searching all calving and wintering areas which we had documented antelope inhabiting during the same time period in the early 1990s, we did not record a single Tibetan antelope in Yeniugou during September 2002.

As during earlier surveys, we made opportunistic observations of carnivores, including wolves Canis lupus, brown bears Ursus arctos, lynx Felis lynx, red and Tibetan foxes Vulpes vulpes and V. ferrilata and Altai weasels Mustela altaica. During our September 2002 survey, we documented the presence of Eurasian badger Meles meles for the first time in Yeniugou. We accounted for at least six individual brown bears during our 2002 survey (five observed directly including a female with two cubs, and one inferred from footprints which could be reliably dated and, from their location, were unlikely to have been made by any of the five bears observed). Although we hesitate to impute much meaning to such a small sample, the relative ease with which we encountered bears struck us as noteworthy (one of our observed bears was encountered independently on three separate occasions). Bears on the Tibetan plateau are extremely wary of people, so our observations suggested to us that human disturbance in Yeniugou had remained low.

Numbers of Himalayan marmots Marmota himalayana, Tibetan hares Lepus oiostolus and Mongolian five-toed jerboa Allactaga sibirica appeared unchanged from the early 1990s. Plateau pikas Ochotona curzoniae, the subject of considerable reduction efforts elsewhere on the Tibetan plateau, were abundant. We noted that brown bears, Tibetan foxes, Saker falcons Falco cherrug and Hume's groundpeckers Pseudopodoces humilis were always associated with habitats containing high pika densities (Smith & Foggin 1999). In one case, a Tibetan fox was seen closely following the movements of a brown bear that was excavating pikas. The fox appeared to be attempting to benefit from capturing pikas disturbed (but not captured) by the bear.

Discussion

As its name suggests, Yeniugou has long been associated with wild yaks. Our interest in the area began in the late 1980s, when reports of substantial poaching of yaks prompted concerns about their status. Commercial poaching of wild yaks appeal's not to have occurred since that time, and subsistence poaching by pastoralists has evidently occurred at a low-enough level to allow for a population increase. The proximity of moist sedge meadows, favoured during summer, with lower elevation Sfipa-dominated grasslands used during winter, evidently provides high quality habitats for wild yaks. Wild yaks are still very vulnerable to human disturbance, but are evidently able to tolerate low density pastoralism in the area. Yeniugou probably contains the densest congregation of wild yaks anywhere in the world, and clearly represents an important component of their entire range (Schaller & Liu 1996).

We considered two hypotheses regarding the continuing, slow decline of argali observed in Yeniugou (from approximately 250 to < 100 within 10 years): 1) poaching and 2) displacement by pastoralists, resulting in either permanent migration, or lowered reproduction incapable of balancing the mortality. We lack evidence that would conclusively reject poaching as a principal factor, but although we do not doubt that some occurred, we are sceptical that it alone caused the observed decline. With the increased awareness of the poaching problem in recent years (largely generated by publicity about declining numbers of Tibetan antelope), we believe that commercial poaching (and associated marketing of meat or other products) would have become known to law enforcement authorities (none was reported). Small-scale subsistence poaching could well have occurred, but pastoralists are no longer armed, do not lack meat and our experience in the neighbouring province of Gansu suggests that pastoralists rarely have easy access to argali because their own activity displaces the animals to more distant slopes (Harris & Pletscher 2001) . We observed no field evidence of argali poaching.

Documenting displacement of wildlife by human activity is always difficult, to say nothing of linking displacement with demographic responses. Thus, we can not be entirely confident that pastoralism is to blame for the argali decline. However, changes in the patterns of pastoralism and of argali distribution (see Fig. 4), together with indications from work in nearby Gansu (Hams & Pletscher 2002) form a pattern consistent with this explanation. Although pastoral encampments remained few in 2002, they had become more sedentary than during earlier decades. Perhaps more importantly, they were located in precisely the areas that argali had preferred during late summer and early autumn of earlier years, as well as in areas we inferred were important for argali during winter. These areas, in which we observed the majority of argali during the 1990s, held no argali in 2002. Instead, the majority of the argali we observed were located in the most remote places in the valley system, where forage production appeared to be lower.

Unlike the related blue sheep, argali appear unable to obtain the quantity and quality of forage they need from sparse, high-elevation sedges and forbs. Argali depend on availability of a substantial quantity of grass in summer (Harris & Miller 1995) supplemented with shrubs in winter (Harris & Bedunah 2001). Among Tibetan plateau ungulates, argali also appear to be the most intolerant to human disturbance, moving to alternate areas when people are present.

In contrast to the situation with argali, widespread poaching of Tibetan antelope during the 1990s was documented by county-level officials in Golmud, and was observed personally by a Kazak pastoralist whom we knew to be a reliable observer. We doubt that domestic livestock were implicated in the disappearance of Tibetan antelope because we have no evidence that domestic herds had used areas formerly frequented by the Tibetan antelope. Because Tibetan antelope are capable of long-distance movements, we can not definitively dismiss the possibility that what appeared to be a dramatic population reduction was, instead, a redistribution or alteration in timing of movements. However, our previous surveys and information from pastoralists present from the 1950s through the early 1980s suggest that our study area encompassed the year-round distribution of Tibetan antelope within Yeniugou (Harris et al. 1999). Additionally, we noted a complete lack of skulls, tracks and droppings from Tibetan antelope in 2002, all of which we commonly encountered prior to our 1997 survey. Thus we interpret the absence of antelope in 2002 as confirming our earlier interpretation that the species is now extirpated from Yeniugou. Its future in this valley system will require recolonisation from nearby herds.

The fact that white-lipped deer and brown bears had persisted in Yeniugou suggests that, with the exception of antelope poachers in the mid-1990s, human disturbance had remained relatively low. White-lipped deer and brown bears (along with wild yaks and argali) are likely to disappear when human presence increases. For example, local respondents and officials were unanimous in reporting that these four species were very rare or had been eliminated from the central and eastern Qilian Mountains in Sunan County, Gansu, whereas blue sheep, Tibetan wild ass and Tibetan gazelle numbers remained locally high (R.B. Harris, unpubl. report to the World Bank Gansu and Xinjiang Pastoral Development Project, 2002). The reduction in traffic and poaching from gold miners no doubt had a beneficial effect on these sensitive species.

However, whether the relatively undisturbed nature of Yeniugou can continue in the face of the rapid development in western China (and lacking specific management plans to prioritize wildlife) is unclear. We conclude that the long-term persistence of Yeniugou's unique wildlife requires a government policy favouring the maintenance of it in a primitive state, a direction that is contrary to the general thrust of development, exploitation and modernization in western China.