Habitat Variability and Ethnic Diversity in Northern Tibetan Plateau

There are 56 officially-recognized ethnic groups in China. However, the distinct geographic patterns of various ethnic groups in relation to the physical environment in China have rarely been investigated. Based on the geo-referenced physical environmental parameters of 455 Han, Tu, Hui, Salar, Mongolian, and Tibetan communities in Qinghai, we found that the communities could be statistically demarcated by temperature and aridity threshold according to their ethnicity, implying that the geographic distribution of each ethnic group is mediated by the physical environment. We also observed that the habitat of each ethnic group is ecologically compatible with current subsistence strategies. Tibetans settle in cold and humid high-altitude regions owing to the cultivation of highland barley and the breeding of yak, dzo, Tibetan sheep and Tibetan goat. Mongolians survive by animal husbandry in cold and dry grassland areas. Han and Tu people settle in the Huangshui River Valley, which offers relatively humid climate and flat land for agriculture. Hui and Salar people occupy the Yellow River Valley with its relatively arid environment and grassland vegetation suitable for animal breeding. Our findings offer a new perspective in explaining the geographic patterns and the varieties of ethnic groups in China and elsewhere.

As per China's Sixth Census in 2010, there are 55 ethnic groups distributed unevenly in Qinghai. In the present study, we based on 455 communities of the six most heavily represented ethnic groups in Qinghai, including the Han, Tibetan, Hui, Tu, Salar, and Mongolian groups, to investigate the possible connection between ethnic groups' distribution and physical environmental factors. These are also the only ethnic groups in Qinghai that have population over 10,000. Our classification of ethnic groups was based on their dominant population stated in chronographs.

Results
The density maps of the communities of each of the major ethnic groups reveal some notable patterns about their geographic concentration (Fig. 2). Tibetan communities are mostly concentrated in Huangnan, as well as Hainan and Golog Tibetan Autonomous Prefectures in the Yellow River Basin, Yushu Tibetan Autonomous Prefecture in the Yangtze headwater region, and the eastern margin of Haixi Mongolian-Tibetan Autonomous Prefecture in eastern Qaidam Basin (Fig. 2a). The distribution of Mongolian communities is dispersed, and small clusters are found in Henan Mongol Autonomous County, southern Huangnan Tibetan Autonomous Prefecture, and the headwater of the Yellow River (Fig. 2b). Clusters of Hui communities are found in the Yellow River (Hualong Hui and Minhe Hui-Tu Autonomous Counties) and the Huangshui River (Datong Hui-Tu Autonomous County) Basins (Fig. 2c). Salar communities are distributed only in Xunhua Salar Autonomous County in the Yellow River Basin (Fig. 2d). Han communities concentrate around Xining City in the Huangshui River Basin (Fig. 2e). Tu communities locate in the Guanting Basin (within the Yellow River Basin), Datong Hui-Tu and Huzhu Tu Autonomous Counties (within the Huangshui River Basin) (Fig. 2f). The altitude of each community is represented by the elevation of its administrative center (Fig. 3). We found that 53.2% of the communities are set at 2,000-3,000 m a.s.l., while 30.8% of the communities found are at 3,000-4,000 m a.s.l. Different ethnic groups locate in different altitudes. Han, Tu, Hui, and Salar communities dominate the region below 3,000 m a.s.l., while Mongolian and Tibetan communities dominate the region above 3,000 m a.s.l. The cold resistance of barley and sheep helped humans settle above 2,000 m a.s.l. around 3,600 cal. yr B.P. 15 . Given that elevation is closely associated with local climate and biomass, we proceeded to compare the communities' locational environmental characteristics among various ethnic groups.
We found that the communities' mean annual temperature and mean aridity threshold among various ethnic groups are significantly different (p < 0.05) ( Table 1). The mean annual temperature associated with the Tibetan communities is the lowest among those of the major ethnic groups in Qinghai (1.95 ± 6.13 °C) (Fig. 4, Table 2),   Table 1. One-way ANOVA of the mean annual temperature and mean annual aridity threshold of the communities of major ethnic groups in Qinghai. a Degree of freedom (between groups). b Degree of freedom (within groups). * Via One-way ANOVA analysis, significant at 0.05 level (P < 0.05).
owing to their high altitude locations (3,517 ± 1,255 m a.s.l.) (Fig. 3). Statistically, their mean annual temperature is significantly lower than that of the Han, Tu, Hui, and Salar communities (p < 0.05) ( Table 2). The Mongolian communities are associated with the second lowest mean annual temperature (2.25 ± 2.78 °C) (Fig. 4, Table 2). Although the temperature is a bit higher than that of the Tibetan communities (1.95 ± 6.13 °C, it is significantly lower than that of the other four ethnic groups' communities (5.14 ± 3.06 °C) (p < 0.05) (Fig. 4, Table 2). On the other hand, the mean aridity threshold of Mongolian communities (16.46 ± 88.77) is significantly higher than that of other ethnic groups' communities (6.37 ± 166.47) (p < 0.05) (Fig. 5, Table 2). This may be caused by the extreme aridity threshold values associated with those communities in the Qaidam Basin. We excluded all the communities in the Qaidam Basin and then recreated our analysis. No significant difference of the communities' aridity threshold values among all ethnic groups was observed (p > 0.05) ( Table 2).
We further focused on the Hehuang Region ( Fig. 1b), because 88.7% of the communities in Qinghai are located there. Based on the type of the communities, six ethnic groups were combined to four groups for further statistical comparison, namely agricultural (Han and Tu), agri-nomad (Hui and Salar), pastoral (Mongolian), and plateau (Tibetan). We found that the communities' mean annual temperature and mean aridity threshold among the above groups are significantly different (p < 0.05) ( Table 3). The mean annual temperature (6.06 ± 2.14 °C) and the mean aridity threshold (4.15 ± 1.14) of the agri-nomad communities are higher than the agricultural ones (5.11 ± 3.08 °C and 3.86 ± 2.64) (p < 0.05) ( Table 4). Agri-nomad communities mainly scatter around the southern slope of the Laji Mountain in the Yellow River Valley ( Fig. 2c and d). The more intense solar radiation and stronger evaporation there result in relatively higher temperature and aridity threshold values in the Yellow River Valley than those in the Huangshui River Valley, where the agricultural communities locate (Fig. 5, Table 4).

Discussion
Linkage between environmental factors and the geographic distribution of ethnic groups. Owing to the unique geographic setting of the Tibetan Plateau, located at the transition zone between semi-humid and semi-arid climate, Qinghai is traversed by the ecotone between agriculture and animal husbandry 20 . Association between human settlement patterns and subsistence strategies in our study region in pre-historic [15][16][17] and historic periods 21 has been evidenced by archaeological and historical records. In the present study, we found that the geographic pattern of the communities of various ethnic groups in Qinghai still exhibits a close connection with physical environmental factors. Also, the connection matches very closely with the current subsistence strategies of various ethnic groups.
The Tibetan is a plateau race. Tibetans live in the cold high-altitude region in Qinghai, and provide for themselves with highland barley, yak, dzo, Tibetan sheep, and Tibetan goat. Also, in Tibet, Gansu, Sichuan, and Yunnan, Tibetans mainly locate in cold high-altitude regions. Highland barley, the primary subsistence crop of the Tibetans, is suitable in the high-altitude zone owing to its genetic tolerance to low temperature (with a semi-lethal low temperature of −4.8-18.4 °C) 22 . Yak 23 , dzo, Tibetan sheep 24 , and Tibetan goat 25 are cold-resistant livestock complimentary to highland barley, supplying meat, eggs, and dairy products to the Tibetans. Tibetans carry a unique gene (HIF2A/EPAS1 and EGLN1) adapted to low temperature 26,27 perhaps as a result of the limited food resources. Such genetic make-up probably enables the Tibetans to survive in the cool high-altitude regions. Still, a biological threshold can be observed. There are very few people living in the coldest part of Qinghai, and −5 °C is probably the minimum mean annual temperature value allowing for large-scale human communities there (Fig. 4).
The Mongolians subsist with animal husbandry on grassland or meadow, and are widely distributed in central and eastern Inner Mongolia. Such vegetation, which is grown in lower temperatures and relatively arid environments, fits into the animal husbandry activities of the Mongolians. There are eight Mongolian communities distributed in the grassland, six in the meadow, five in the desert, and one in the shrub-land (Fig. 6). The grassland vegetation in the Henan Mongolia Autonomous County is similar to the one in the middle and the eastern parts of Inner Mongolia (Fig. 6). Vast grasslands support a large number of sheep and cattle, sustaining the husbandry of the Mongolian communities. In addition, some Mongolian communities spread along areas of alpine meadow vegetation, where the cattle and sheep can be fed on abundant pasture.
The diversity of ethnic groups in the Hehuang Region may be attributable to the substantial regional variation of the aridity threshold. Within the region, the relatively humid environment in the Huangshui River Valley ensures crop growth and hence, a large amount of cultivated land has been opened up in the valley (Fig. 6). The arid environment in the Yellow River Valley does not meet the water demand of crops and is only suitable for dry grassland, as well as animal husbandry activities (Fig. 6). The above disparity paves the way for the differentiation of subsistence strategies within the Hehuang Region. Agricultural communities plant wheat, potato, rape and other crops and settle in relatively humid flatland in the Huangshui River Valley. As agri-nomadic communities engage in both agriculture and animal husbandry, they occupy relatively arid grassland in the Yellow River Valley, where the husbandry that supplies red meat to the Hui and Salar is also supported.
In addition, the Huangshui River is a tributary of the Yellow River. Its erosion datum surface (1,580 m a.s.l.) locates where the Huangshui River feeds into the Yellow River. The Huangshui River bends in the valley owing to the small fall (4,200-1,580 m) and tardier down-erosion. Because of steady lateral erosion, the Huangshui River  Table 2. Tukey HSD test of the mean annual temperature and mean annual aridity threshold of the communities of major ethnic groups in Qinghai. *Via Tukey HSD test, significant at 0.05 level (P < 0.05).

Figure 5. Mean annual aridity threshold and the distribution of the communities of major ethnic groups in
Qinghai. The aridity threshold data were derived from the evaporation and precipitation datasets (evaporation divided by precipitation) that were retrieved from China Meteorological Administration (http://data.cma.cn/). The map was generated in ArcGIS version 10.1 (www.esri.com/software/arcgis).  Table 3. One-way ANOVA of the mean annual temperature and mean annual aridity threshold of the communities of major ethnic groups (combined) in Hehuang Region. a Degree of freedom (between groups). b Degree of freedom (within groups). *Via One-way ANOVA analysis, significant at 0.05 level (P < 0.05). Valley is more flat (Fig. 1b) and more appropriate for crop cultivation in comparison to the Yellow River Valley (Fig. 6).  (Table S1). Furthermore, the habitat of the six major ethnic groups in Qinghai corresponds to their subsistence strategies. In terms of the number of communities, agricultural ethnic groups share about 38.24%, agri-nomadic ethnic groups share about 10.55%, pastoral ethnic groups share about 4.40%, while plateau ethnic groups share about 46.81%. The Plateau (Tibetan) ethnic group is mainly distributed in the high attitude area (above 4,000 m a.s.l.). As Tibetans have accustomed themselves to the cold-humid rather than the cold-arid environment, they are not dominant in Xinjiang and Inner Mongolia (Table S1). On the other hand, although the cold-arid environment is suitable for animal husbandry, it is not favorable for steppe, leading to the scarcity of the pastoral (Mongolian) ethnic group in Gansu, Sichuan, and Tibet (Table S1). The agricultural (Han, Tu, Kazak, and Yi) ethnic group is distributed in all of the provinces in northwestern China. Yet, subject to the environmental constraints of agriculture, it is hard for agricultural ethnic group to live in areas with high elevation. Hence, only 8.17% of the population in Tibet belongs to agricultural ethnic group (Table S1). In addition, the agri-nomadic (Hui, Uyghur, Dongxiang and Salar) ethnic group is scarce in Sichuan, Mongolia, and Tibet (Table S1) for different reasons. Southward migration of the pastoral ethnic group to Sichuan is hampered by the humid environment, while the development of low-altitude agriculture in Mongolia and Tibet is restricted by the cold environment. In addition, a unique ethnic group called Manchu lives in the humid forest area, and they sustain their living by hunting and gathering. The four types of subsistence strategies in Qinghai probably arise from the variety of physical environments there. Different habitats are characterized by different elevation, temperature, and aridity threshold, while the presence or absence of certain geographical features in some regions probably makes certain subsistence strategies more or less feasible there. Previous studies have built on environmental determinants such as precipitation, vegetation, and pathogens to account for cultural diversity at the global and continental levels 1,28 . Our empirical findings show that the geographic distribution of ethnic groups in terms of their communities is related to physical environmental factors such as temperature, aridity threshold, and vegetation cover at the provincial level, indicating the possibility for the relationship to be downscaled to lower geographic levels.

Ethnic group (combined) Agricultural
At the global level, the present cultural diversity is a rich variety of hunter-gathers, pastoralists, agriculturalists, and more complex societies, revealing the possibility that the process of cultural diversification may be independent of subsistence 1 . Yet, at lower geographic levels, the landscape diversity in Qinghai is associated with the diverse subsistence strategies of ethnic groups and consequently determines their geographic distribution and variety. This is substantiated by the analogy of human genetics. People of some ethnic groups such as Tibetans and Han Chinese are basically genetically homogenous [29][30][31] , implying that they may have common origins. But, when they live in different environments, they equip themselves with different subsistence strategies for food selection and environmental adaptation, and have subsequently evolved into different cultures.

Methods
Data. Although the territories and the population size of communities have been changing over time, their ethnic grouping has been rather stable in the long-term. In this study, the ethnic grouping of communities was made according to their most dominant population stated in chronographs. Coordinates of the communities were obtained from Google Earth. The altitude of each community is represented by the elevation of its administrative center.
To quantify the physical environmental parameters of Qinghai, the mean annual temperature, precipitation, and evaporation data from 39 meteorological stations in Qinghai in 1980-2010 (http://data.cma.cn/) and the 1:4,000,000 vegetation map (http://westdc.westgis.ac.cn) (Fig. S1) were employed. And the mean annual aridity threshold was calculated by dividing the precipitation from the evaporation. In addition, the coordinates for those meteorological stations were obtained from Google Earth. Based on the above data inputs, the mean annual temperature, precipitation, and evaporation associated with each community were obtained via the Kriging Interpolation function in ArcGIS. The mean annual aridity threshold index of each community was derived from the resultant values of evaporation and precipitation (i.e., evaporation divided by precipitation).

Statistical analysis.
Based on the coordinates of the communities of each of the major ethnic groups, the spatial analytical tool Point Density in ArcGIS version 10.1 was employed for density analysis and the making of density maps. Under this method, a cluster (i.e., geographic concentration of communities) is determined by the number of communities of a particular ethnic group which fall within a neighborhood around each cell. In this study, the radius parameter of neighborhood was set as 2 km in calculating the geographic concentration of communities.
One-way ANOVA was used to test for significant differences between the means over a single dependent variable for two or more groups divided by an explanatory variable (i.e., independent factor). If F is significant, it can be concluded that there are differences in group means and thereby, the explanatory variable is proven to have an effect on the dependent variable 32 . In the present study, One-way ANOVA was employed to test whether the communities of various ethnic groups could be demarcated by environmental parameters (i.e., temperature and aridity threshold). The ethnic grouping of communities was entered as the independent ANOVA factor, while the value of each type of environmental parameter was entered as a dependent ANOVA factor. In addition, Tukey HSD test was used in conjunction with One-way ANOVA to identify which pairs of groups show statistically significant mean differences.