The data used in this study encompass land cover, climate, and digital elevation model (DEM). Land cover data of the KRB in 1990 and 2015 with a resolution of 30 m were provided by the Land Change and Regional Adaptation Research Group of the Qinghai-Tibet Plateau, Institute of Geographic Sciences and Natural Resources Research, Chinese Academy of Sciences (CAS). Interpretation accuracy of the land cover type with land cover data of the KRB was 83.36%. The meteorological data used included station data and raster data. The station data were taken from the five meteorological stations within the Chinese KRB, covering the period between 2000 and 2018 and the station data of the Nepalese KRB region were taken from Department of Hydrology and Meteorology, Government of Nepal. Raster meteorological data encompass 0.5° spatial resolution Climate Research Unit (CRU) TS 4.03 data released by this group at the University of East Anglia (https://www.cru.uea.ac.uk/ data). Additional interpolated meteorological data included average temperature, maximum temperature, and minimum temperature as well as average precipitation. DEM data at a 30-m resolution (Aster GDEM) were downloaded from the United States Geological Survey (USGS) (http://earthexplorer.usgs.gov/).

Land cover types in the KRB mainly include cropland, forest, shrubland, grassland, sparse vegetation, waterbody, built-up area, wetland, bare land, glacier and snow cover. It is obvious that the area and proportion of land cover types are different with the increase of elevation. Results show that cropland, forest, shrubland, grassland, sparse vegetation, bare land and glacier and snow cover are distributed from low to high elevations, generally consistent with the natural belt distributional trends. However, vertical distribution characteristics on both slopes are quite different.

Peak values for the distribution of main land cover types (i.e., forest land, shrubland, grassland, sparse vegetation, and bare land) within each elevation range also rose with elevation (Figures 2a and 2b). The peak value is the highest distribution ratio for one land cover type at a certain altitude. As the Himalayas block the warm and humid air flow of the Indian Ocean, the northern slope experiences little precipitation, and the climate is cold and arid, which has the features of the typical continental plateau. Most parts of the northern slope is over 4000 m, resulting in the loss of the vegetation zone below this elevation. In terms of the number of peak values, most land cover types conform to a unimodal distribution (i.e., a single peak value) with the increase of elevation. On the northern slope, land cover types that follow a unimodal distribution mainly include cropland (peak at 4200 m), forest (3700 m), grassland (4700 m), sparse vegetation (5100 m), waterbody (4000 m), wetland (4100 m) and glacier and snow cover (5800 m). It is also the case that other types of land cover exhibit bimodal distributions as built-up area, shrubland and bare land for they have two peaks. One peak of built-up area is distributed at 4100 m, which is the maximum peak, and the other one at 4400 m; one peak of shrubland is distributed at 4100 m, and the other one at 4500 m; one peak of bare land is distributed at 5000 m, and the other one at 5300 m, which is the maximum peak. However, hydrothermal conditions on the southern slope are especially abundant due to the strong influence of warm and moist air currents from the Indian Ocean. The relative height difference from the foothill to the mountain top is large, indicating that the combination of water and heat varies greatly with the elevation. Land cover types that follow a unimodal distribution on the southern slope mainly include cropland (peak at 1400 m), forest (2000 m), grassland (4900 m), sparse vegetation (5300 m), built-up area (1500 m), bare land (5500 m), and glacier and snow cover (5700 m) (Figure 2b). These peaks also appear successively with elevation. Shrubland, waterbody and wetland follow a bimodal distribution. One peak of shrubland is at 2300 m and the other one at 3100 m; one peak of waterbody is at 300 m and the other one at 5300 m; one peak of wetland is at 300 m and the other one at 4600 m (Figure 2b).

The areas and proportions of different land cover types are disparate across the same range of elevation. Statistical analyses of the area and proportion of each land cover type in every 100-m gradient in elevation, which makes it possible to determine the dominant land cover types at different elevations, referred to as the dominant belt. It is therefore composed of all the dominant land cover types across the whole elevation range. The dominant belts on the northern slope included forest dominant belt at the elevation between 2200 m and 3900 m, shrubland dominant belt between 3900 m and 4200 m, grassland dominant belt between 4200 m and 5100 m, sparse vegetation dominant belt between 5100 m and 5500 m, bare land dominant belt between 5500 m and 6000 m, and glacier and snow cover dominant belt above 6000 m (Figure 3a). The dominant belt on the southern slope included cropland dominant belt at the elevation between 100 m and 1100 m, forest dominant belt between 1100 m and 3900 m, bare land dominant belt between 3900 m and 5700 m, and glacier and snow cover dominant belt above 5700 m (Figure 3b).

Data showed that land cover changes remained relatively stable on both northern and southern slopes of the KRB between 1990 and 2015, although some land cover types have changed obviously. Land cover types with obvious area reduction on the northern slope are glacier and snow cover (-176.72 km²) and grassland (-163.83 km²), while shrubland (-485.22 km²) and glacier and snow cover (-121.19 km²) on the southern slope reduced obviously (Figure 4a). The reduction of glacier and snow cover was mainly occurring within an elevation range between 5000 m and 6000 m. The loss of area on the northern slope was greater than on the southern slope, and was accompanied by a concomitantly higher rate of decline. Results showed that 88% of reduced glacier and snow cover on the northern slope was converted to bare land, while 9% turned into waterbody (i.e., glacier lakes and river) (Figure 4b). In contrast, the retreated glacier and snow cover was converted to bare land (77%), grassland (11%), and sparse vegetation (8%) on the southern slope (Figure 4c). The reduced area of grassland on the northern slope was second only to that of glacier and snow cover within a range of 4000 m to 5000 m. The reduced area of grassland was mainly converted to bare land (53%) and sparse vegetation (37%) (Figure 4b). In contrast, the reduced area of shrubland was mostly distributed on the southern slope, and its reduction rate was large (up to 2.8%) and was mainly converted to cropland (63%) and forests (18%) (Figure 4c). The area of wetland on the southern slope also tended to decrease with the highest decrease rate of 3.7%, being mainly converted to cropland (56%) and bare land (17%).

In terms of the increase of land cover type area, the area of bare land (1196 km²) increased markedly on the northern slope, while waterbody and cropland only slightly enlarged (Figure 4a). Reductions in bare land area have mainly occurred in the elevation range of 5000 m to 6000 m with an annual average decrease rate of 0.19%. The largest contributors to the increases in bare land area were glacier and snow cover (38%), grassland (31%), and sparse vegetation (30%) (Figure 4b). Results showed that cropland (increased by 437 km²) and forest (336 km²) are the main forms of land cover increase on the southern slope (Figure 4a). The increase of cropland mainly occurred below 2000 m, and the increase of forest mainly occurred between 1000 m and 2000 m, with annual growth rates of 0.77% and 0.63%, respectively (Figure 4c).

Changing trends for each land cover type versus elevation have been different over time, especially between northern and southern slopes (Figure 5). For the last 25 years, the distribution peak for cropland on the southern slope tended to move to lower elevations over time from 1500 m to 1200 m. While the cropland area on the northern slope remained small and relatively stable in vertical distribution.

Vertical changes in forest area were quite dramatic, especially on the southern slope. The distribution of forest exhibited two peaks in 1990, one at 1000 m and the other at 2000 m. While a single peak at 1600 m was found in 2015, and from 1990 to 2015 the forest area increased markedly between 1000 m and 2000 m. During this period, there was no obvious change in the forest distribution curve on the northern slope.

Change of grassland on the southern slope was more obvious than that on the northern slope. Results showed that grassland area increased obviously at about 4000 m, which was probably the result of returning cropland to grassland. However, the change of grassland on the northern slope was not obvious, which may be due to the overconcentration at a range of 4100 m to 5000 m.

The built-up area changed obviously in this period. On the southern slope, built-up area was concentrated between 1500 m and 1600 m in 1990, while its distribution was between 1100 m and 2000 m in 2015, showing an obvious trend of moving to lower elevations. In contrast, the built-up area on the northern slope was mainly distributed between 4300 m and 4400 m in 1990. As built-up area increased in this period, its distribution was concentrated between 4200 m and 4600 m in 2015, showing a trend of moving to higher elevations. The peak distribution of glacier and snow cover on both slopes tended to move to higher elevations in this period. It is clear that the movement of glacier and snow cover on the northern slope was greater than that on the southern slope. Glacier and snow cover generally tended to move to higher elevations. The peak of glacier and snow cover moved from 5500 m to 5600 m on the southern slope and from 6000 m to 6100 m on the northern slope over the past 25 years.

Shrubland on the southern slope tended to move generally towards higher elevations from 1990 to 2015. Results showed that the peak elevation of sparse vegetation on the northern slope did not change despite an increase in area and proportion. Bare land on the southern slope moved to higher elevations and became more concentrated around the peak, but bare land on the northern slope became more dispersed. The peak of wetland on the southern slope was not obvious, and the distribution was uniform with elevation.

The data presented in Table 1 showed the vertical changes in land cover types on both slopes. Elevation range in this case denotes the elevation distribution range for each land cover type, while the core distribution zone denotes the concentrated elevation distribution range (i.e., the sum of area more than 60% of the total area) in each case. The dominant belt refers to the land cover type occupying the largest percentage at this elevation gradient.

The data showed that the dominant belt comprising cropland, forest, and bare land as well as glacier and snow cover all tended to expand on the southern slope. The cropland dominant belt range was between 100 m and 700 m in 1990. By 2015, this range expanded to between 100 m and 900 m, with the upper limit of the cropland dominant belt expanding 300 m to a higher elevation. During this period, the lower limit of the forest dominant belt also rose by 300 m, while the upper limit of its dominant belt expanded by 1900 m to a higher elevation. The lower limit of the bare land dominant belt dropped by 100 m to a lower elevation over this time, while its upper limit rose by 300 m to a higher elevation. The lower limit of the glacier and snow cover dominant belt also rose by 300 m to a higher elevation.

This was in contrast to the situation of the northern slope, where the upper limit of the forest dominant belt decreased by 200 m to a lower elevation. The upper boundary of the grassland dominant belt retreated by 200 m over this period and replaced by sparse vegeta-tion between 3900 m and 4100 m. The lower boundary of glacier and snow cover increased by 100 m over this period.

Changes in vegetation types are mainly determined by climate change. Changes in temperature, precipitation, and sunlight all modify the original environments in which vegetation grows and causes movement to more suitable locations (Guo et al., 2019; Peters et al., 2019). Therefore, due to the great differences in the geographic environment within the KRB, there are significant differences in the influencing factors of the vertical belt changes on both slopes (Table 2). The observations showed that cropland and forest were mainly found on the southern slope, so the change of cropland is significantly related to climatic factors. Expansions in cropland are directly proportional to precipitation change. That is to say, with the increase of precipitation, cropland also showed a trend of expansion. The probability of cropland expansion increases 1.148 times with every 1 mm increase in precipitation. Results showed that cropland tended to expand, which was positively correlated with changes of average temperature, and negatively correlated with changes of minimum and maximum temperatures. The upper and lower boundaries of the forest also developed towards higher elevations, and the northward movement of the forest was mainly related to changes of precipitation and minimum temperature, which was negatively correlated with changes of precipitation. As precipitation decreases, the probability of forest change increases, which was positively correlated with changes of minimum temperature. The probability of forest change also increases in concert with precipitation. Glacier and snow cover on the southern bare land tended to expand, and most glacial retreated areas were converted to bare land. slope tended to retreat over time, while Glacial retreat was mainly related to temperature change, which was positively correlated with changes in average and maximum temperatures, and negatively correlated with changes in minimum temperature. This is in line with our conventional understanding of these phenomena, which is that the factors influencing the expansion of bare land were the same as those controlling the retreat of glacier and snow cover, but the correlation coefficient is opposite.

Grassland is the main ecosystem type on the northern slope of the KRB, followed by glacier and snow cover which are sensitive to climate change. The results showed that the main changes in grassland were the receding of the upper boundary of the grassland dominant belt and the degraded area was replaced by sparse vegetation. Changes in grassland were mainly influenced by temperature change, which were negatively correlated with the changes of average and minimum temperatures.

The huge elevation differences within the KRB mean that hydrothermal composition varies with elevation, which led to vertical and north-south ecosystem differentiation throughout the study area. Indeed, due to differences in water and heat, ecosystems on the northern and southern slopes of the KRB have developed their own unique vertical combinations (Cidan, 1997). In this context, early research to assess the vertical belt in the KRB began in 1973, while data collected by the scientific investigation team to the Mt. Qomolangma in the Chinese territory between 1959 and 1960 allowed Zhang Jingwei and others to study the region’s vertical vegetation zonation as well as its relationships to the horizontal zones (Zhang and Jiang, 1973). The vertical vegetation patterns across Mt. Qomolangma region have not changed obviously since 1968 (Zhang and Jiang, 1973; Zhang et al., 2020). However, due to increases in temperature and human activities, the vertical belt has experienced local changes, including changes in internal structure and quality (Wu et al., 2020).

Although there has been a limited number of studies on changes and driving forces controlling the vertical belt of Mt. Qomolangma in recent years, yet a great deal of research has been done on the surrounding mountains. The Qinling Mountains are considered to be an important boundary between wet and dry climates for research purposes, rather than a temperature boundary (Fang and Yoda 1989). Therefore, these mountains influence the dry-wet pattern across the north-south transitional zone (Zhao et al., 2020). Studies on climate change and vegetation coverage within the Qinling-Daba Mountain region showed that the changes of vegetation coverage were positively correlated with precipitation and negatively correlated with temperature (Liu et al., 2016). These results indicated that the influence of precipitation on vegetation growth across the Qinling Mountains was stronger than temperature. Changes in vegetation coverage within the Daba Mountains were correlated with temperature to some extent, but not to precipitation (Ren et al., 2012). Those results also showed that it is more appropriate to use the Qinling Mountains as the temperature boundary between the warm temperate zone and the north subtropical zone than the Daba Mountains (Zhao et al., 2020). In terms of each vertical belt, results showed that alpine grasslands on the southern part of the Qinghai-Tibet Plateau are more sensitive to climate variability overall even though the alpine grassland has obvious spatial heterogeneity. Areas in the south are more sensitive to temperature change, while areas in the northeast are more responsive to precipitation. The central part of the region is mainly influenced by radiation and temperature changes (Li et al., 2019). The sensitivity of alpine grasslands to climate change, especially temperature change, increases significantly with the increase of elevation (Li et al., 2019). Similarly, continuous warming across the KRB has directly and indirectly influenced changes of the local cropland. As a result of increases of average temperature and changes of monsoonal rainfall dates, agricultural farming history has also changed greatly, resulting in significant variations in cropland area of the KRB (Paudel et al., 2016). Climate change has also exerted greater negative effects on cropland, since only a small number of crops is resistant to high temperatures, but most are not. As a result of the negative impact of temperature increase, the susceptibility to various crop diseases has also increased, and another clear impact on cropland production and changes were evident (Paudel et al., 2016; 2020). The alpine forest belt ecotone within this land cover type provides a simplified model for studying global ecology and climate change. Dynamic spatiotemporal changes of tree lines can be reconstructed on this basis. Data showed that the upward movement of tree lines was a response to recent climate warming, and that the rate of change was mainly regulated by spring precipitation (Sigdel et al., 2018). Indeed, the rate of increase in this case has been higher across the eastern part of the wettest Himalayas, suggesting that spring precipitation has contributed (Sigdel et al., 2018).

Human activities change land use pattern, which is also the main reason for the rapid changes within the vertical belt. Cropland is the ecosystem most affected by human activities. In the early period of rapid population growth, more people needed food and resources, which was related to the expansion of cropland. The statistical results of cropland driving changes within the KRB showed that population density has exerted a significant impact on cropland expansion, with the greatest impact from 1978 to 1992 and 1992 to 2010 (Paudel et al., 2016). Indeed, over the period between 1992 and 2010, cropland area decreased slightly and was converted to fallow land or grassland. There is also a direct relationship between young Nepalese migrant workers and cropland change (Paudel et al., 2016; 2020). Across this region, forests within the lower elevation were prone to be disturbed by human activities. Declines in forest area may also be related to the increases of cropland and built-up area, while the distribution range of bare rock and gravel lands has also increased markedly. The expansion of bare rock and gravel lands to lower elevation has mainly been the result of grassland degradation by overgrazing (Zhang et al., 2013). The expansion of bare rock and gravel lands to higher elevation areas are mainly the result of high-altitude glacier retreats caused by climate warming (Nie et al., 2010; 2013). Research has shown that climate warming, precipitation, and soil moisture all influenced changes in alpine grasslands in areas where human activities have been weak. In contrast, in areas where human activity is intense, overgrazing is the main factor leading to alpine grassland degradation (Li et al., 2018). Changes in glacier and snow cover are also the main result of climate factors. However, in areas where tourism and mountaineering activities are developed, such as the famous Mt. Qomolangma region, glacier and snow cover has been severely damaged by human activities. In addition, previous literatures have emphasized the holy land of plodded around mountains, waters and Buddhist temples in Tibet. Kangrinboqe is recognized to be the center of the world by Hinduism, Tibetan Buddhism, Bon religion and ancient Jainism in Tibet, and the existence of the holy lakes such as Lake Manasarovar also increases the pressure on the ecological environment. All the changes discussed here have influenced, or will continue to influence the ecosystem services provided by alpine ecosystems and highlight the huge ecological and environmental challenges in cross-border areas, particularly the border between China and Nepal. The vertical belt distribution and change of land cover analyzed in this study reflect the joint influence of climate change and human activities on land and vegetation.