*4.1. Desert-Oasis Ecotone and Land-Use Changes in the Tarim River Basin* 4.1.1. Desert-Oasis Ecotone Changes in the Tarim River Basin

Figure 3 shows the changes in the extent of the desert-oasis ecotone in the Tarim River Basin and its various sub-basins in 1990, 2000, and 2015. The desert-oasis ecotone of the Tarim River Basin declined during this period, decreasing from 67,642 km<sup>2</sup> in 1990 to 46,613 km<sup>2</sup> in 2015. At the same time, the area of the desert in the study area expanded, with the proportion of desert area increasing from 59.88% in 1990 to 63.36% in 2015.

**Figure 3.** The areal extent of the ecotone in 1990, 2000, and 2015 in each basin. The lower maps show changes in the local ecotone in (**1**) the Weigan River Basin, (**2**) the Keriya River Basin, and (**3**) the main stream of the Tarim River.

At the sub-basin scale, the area of the ecotone of each sub-basin also decreased by different degrees, among which the three basins with the most significant area reduction were the Kaidu-Kongque, Weigan and Keriya river basins; the areas of reduction were 2223 km<sup>2</sup> , 1704 km<sup>2</sup> , and 5090 km<sup>2</sup> , respectively (Table S5). While the ecotone decreased in size, the vegetation coverage also significantly decreased, as shown by a drop in the NDVI from 0.142 to 0.127. This indicates a deterioration in the desert-oasis ecotone in terms of area and quality between 1990 and 2015.

#### 4.1.2. Land-Use Changes in the Tarim River Basin

The shrinkage of the desert-oasis ecotone in the Tarim River Basin is closely related to the strong land-use changes in the basin in recent decades. Table 1 further describes the area and proportion of each land-use type in the Tarim River Basin in 1990, 2000, and 2015. Overall, from 1990 to 2015, the area of arable land, water bodies, industrial land, and unused land

increased. In contrast, the area of grassland decreased, whereas the area of forest land did not significantly change. Unused land was the dominant land-use type in this area (accounting for about 53%), increasing by 1419 km<sup>2</sup> in the preceding two decades. Grassland also represented a dominant land-use type in the area. Grasslands decreased from 35.97% in 1990 to 34.60% in 2015; the areal decrease was 8911 km<sup>2</sup> . These land-use patterns reflect the intensity and nature of human activities; in fact, the expansion of arable land far exceeds the increase in the area of industrial land. Overall, the arable land area expanded drastically from 1990 to 2015, increasing by 7125 km<sup>2</sup> , while the area of industrial land increased by only 66 km<sup>2</sup> . The area of water bodies increased from 1990 to 2000 and then slightly decreased from 2000 to 2015. Over the entire period, the area of water bodies increased by 283 km<sup>2</sup> . Forest land, which occupies a relatively small proportion of the total, exhibited insignificant changes in the area; its relative proportion was stable, at about 1.90%.


**Table 1.** Areas and proportions of land-use types in the study area in 1990, 2000, and 2015.

Figure 4 shows the spatial distribution of the interconversion between different land uses in the Tarim River Basin during the period 1990–2015. The increase in arable land in 1990–2015 was mainly in the periphery of the original arable land and oasis and extended to the unused land. The increase in the area of arable land was mainly at the expense of grass land, unused land, and forest land. The increase in unused land was mainly distributed near the location of arable land and original grassland, and its areal expansion mainly came from the degradation of some arable land, forest land, and grassland. The increase in water bodies was mainly distributed in the foothills of the southern edge of the Tarim Basin, such as in the upstream areas of the Hotan and Yarkand river basins, and it was mainly derived from the transformation of grassland (Table S6).

#### *4.2. Driving Force Analysis*

Changes in the desert-oasis ecotone of the Tarim River Basin are inextricably linked to natural and anthropogenic factors. Therefore, this study investigated the intrinsic causes and drivers of the changes in the desert-oasis ecotone by changes in climatic parameters and anthropogenic activities.

#### 4.2.1. Meteorological Factors

A total of 26 meteorological stations in the Tarim River Basin were selected, and temporal trends in mean annual temperature and annual precipitation were analyzed. Of the total stations, 22 stations exhibited an increase in temperature and 18 stations exhibited an increase in precipitation (Figure 1). Changes in dry and wet conditions were analyzed by calculating the SPEI for different time scales of drought in the Tarim River Basin from 1990 to 2015. At the 3-month, 6-month, 9-month, and 12-month time scales, average SPEI values exhibited a decrease, indicating enhanced aridification in the study area (Figure 5). Moderate droughts occurred in 2006 and 2008. Since 2000, the frequency and severity of droughts have become stronger, suggesting that droughts are an important reason for the accelerated decline in the ecotone after 2000. The results of the analysis also demonstrate that multi-scale SPEI can effectively show the degree of drought and drought duration in the Tarim Basin. SPEI of different scales show different degrees of interannual oscillations

and interannual variability, but the overall direction of change was the same; the study area became more arid after 1990.

**Figure 4.** Land-use transfer map of the Tarim River Basin from 1990 to 2015. The lower maps show the local changes in (**1**) the Weigan River Basin, (**2**) the Keriya River Basin, and (**3**) the main stream of the Tarim River.

**Figure 5.** Time series of the 3-, 6-, 9-, and 12-month SPEI values in Tarim River Basin from 1990 to 2015.

#### 4.2.2. Human Factors: Groundwater Changes

Within the Tarim River Basin, where precipitation is extremely low, groundwater is an important source of irrigation water. With the increase in arable land in the basin, the exploitation of groundwater has also increased. For example, groundwater monitoring data collected in the Yarkand, Kaxgar, and Weigan river basins show that groundwater levels have decreased during the 21st century (Figure 6). The most significant decline has been in the Weigan River Basin, where declines in groundwater levels of up to 2.48 m, 4.93 m, and 3.97 m have been observed. The expansion of oasis cultivation and irrigation in the basin has presumably caused a significant decrease in groundwater levels, which is a key hydrological element for the survival of natural vegetation and directly affects the growth and maintenance of natural vegetation in the desert-oasis ecotone.

**Figure 6.** Changes in the groundwater table in (**a**) the Kaxgar River Basin, (**b**) the Yarkand River Basin, and (**c**) the Weigan River Basin. The dots represent the observation points of the groundwater level, and the fold lines represent the change in the trend of the groundwater level. The color of the fold line is consistent with the color of the observation point of the groundwater level.
