Impacts of Ecological Migration on Land Use and Vegetation Restoration in Arid Zones

Abstract

Poverty and disasters are globally prevalent in ecologically fragile areas. Ecological migration is regarded as an effective way to address these issues. This paper investigates the spatial pattern of ecological migration and the corresponding spatiotemporal changes in land use and vegetation restoration in Gulang County, located in northwest China, between 2010 and 2018. For this purpose, we calculated three indicators: the transfer matrix of land use, normalized difference vegetation index (NDVI), and vegetation restoration degree (VRD). We found that ecological migrants in Gulang County moved from the Qilian Mountain National Park to the intersection between flat area and desert. The spatial patterns ranged from high-altitude to low-altitude, and the slopes became less steep. The distribution of the resettlements is more clustered and shaped by the traffic conditions and guided by the local governments. Unused land in the whole-village migration area and construction land in the resettlement area were dramatically impacted by ecological migration (625% and 279.3%, respectively). The cropland and construction land in the outmigration areas were mainly replaced with grassland and forest. In contrast, the grassland and unused land in the resettlement area were transferred to cropland and construction land. After ecological migration, the mean NDVI and VRD in Gulang County significantly increased, indicating that the vegetation in the outmigration areas quickly recovered. Moreover, the VRD in the whole-township migration areas was greater than that in the whole-village migration areas (121% > 68%). The main contribution to the increase in NDVI was the conversion of forest to grassland, accounting for 33%. In addition, the transition from other types of land to grassland made a larger contribution to the NDVI than conversions to forest.

1. Introduction

Along with ecological degradation, ecological migration has been given much attention worldwide [1,2]. During the period from 2008 to 2018, global ecological degradation caused nearly 265 million people to migrate. Moreover, global climate change leads to increasingly severe ecological crises; thus, it is predicted that 143 million people will be exposed to ecological degradation by 2050 [3]. Furthermore, in these areas, the change in demographics and the resulting changes in the socioeconomic systems will exacerbate impacts on the local ecological environment [4]. One can therefore imagine that ecological migration will occur more frequently in the future.

As a consequence of and also a solution to ecological degradation, ecological migration not only minimizes the ecological damage caused by human activities but also facilitates ecological restoration in the outmigration areas, which has great importance for improvements in the regional ecological environment [5]. A growing number of developing countries have found that ecological migration is an effective way to protect the ecological environment and reduce poverty. In the policy aspect, China initiated an ecological migration program in 2001 [6,7], aiming to protect the fragile ecological environment and reduce poverty. Such a program is also helpful to decrease natural disasters such as desertification, drought and landslides, thus avoiding damage to human beings [1,8]. To some extent, assessing the impact of ecological migration and proposing feasible plans for environmental improvement are conducive for the government to ensure environmental governance and reduce poverty via policies [9].

Researchers have agreed that ecological migration is the process by which an impoverished population in ecologically fragile areas, areas with an important ecological function and uninhabitable areas moves to other areas due to the severe natural environment [10,11,12,13,14]. Many scholars have used field survey data or official statistical data to assess the impact of ecological migration on the social economy or reducing poverty [7], local attachment [15] and farmers’ satisfaction [16] and the willingness to partake in ecological migration using a structural equations model [6]. They found that compensation for resettlement significantly affects farmers’ willingness and satisfaction. Existing studies also show that ecological migration helps protect the environment in outmigration areas and accelerates urbanization as well as reducing poverty [9,17]. However, previous studies have not reached a consensus on the impact of ecological migration. Some argue that ecological migration reduces human activities in the outmigration areas, thus protecting the ecological environment and, in turn, decreasing natural disasters [18]. Others point out that for the immigration areas, a larger number of migrants will deteriorate the local ecological environment [9]. Moreover, overcapacity caused by resettlement certainly has a negative impact on the natural and socio-economic systems, inducing secondary ecological damage. With the development of satellite technology, some scholars have used remote sensing data to obtain a more comprehensive assessment, which can more objectively examine the changes in land cover, agricultural land abandonment [19,20] and urban expansion [21] caused by ecological migration. They found that rural–urban migrants promoted urban expansion in the immigration areas while exacerbating the abandonment of rural cropland and forest in the outmigration areas. Other scholars used the normalized difference vegetation index (NDVI) [22] or adopted the land use transfer matrix [23] to explore the vegetation recovery and ecological risks in earthquake zones [24], the Three Gorges reservoir area [25,26,27] and ecological migrant resettlement areas [28]. However, these studies focused only on the effects of ecological migration on the ecological environment in the immigration areas. Few studies paid attention to the spatiotemporal variation in ecological restoration in both the immigration and outmigration areas before and after ecological migration, especially in arid zones with a seriously fragile ecosystem. In addition, there are some limitations of the methodologies or datasets in the previous literature on the effects of ecological migration. Moreover, some key issues have yet to be explored. For example, in arid zones, what is the spatial pattern of ecological migration? How does ecological migration impact land use and vegetation coverage? What are the spatio and temporal variations in vegetation restoration in both immigration and outmigration areas?

In this paper, we explore these issues by focusing on Gulang County, a typical ecological migration area in China. The southern part of Gulang County is part of the Qilian Mountain National Park which is an important ecological security barrier in northwest China. However, the area has a fragile ecological environment and is home to numerous populations in poverty areas. In the pursuit of environmental protection and poverty reduction, the local government implemented a large-scale ecological migration project in 2010, gradually transferring 62,400 farmers and herdsmen living in deep mountainous areas to lower altitude areas. By the end of 2018, good results had been achieved. Given this is a complex ecosystem covering a largely arid area and the scale of the human migration which has occurred, we regard Gulang County as a representative study area.

In this context, we conducted field research and used the land use transfer matrix, NDVI and vegetation recovery degree (VRD) to systematically analyze the spatial pattern of ecological migration, spatial and temporal changes in land use and vegetation recovery in Gulang County from 2010 to 2018. Exploring the environmental impacts of ecological migration can provide some policy implications for subsequent planning and development of ecological conservation in such arid zones.

2. Materials and Methods

To investigate the impacts of ecological migration on land use and vegetation restoration in the arid area, we followed the research process shown in Figure 1. There are three steps: (a) the collection and collation of relevant data of the study area, (b) the selection of key indicators and calculation and (c) statistical analysis and visualization.

2.1. Study Area

Gulang County, located in the Hexi Corridor and connecting to the Tengger Desert in the north and the Qilian Mountain National Park in the south, is one of the counties in Wuwei city, Gansu province, China. The county’s topography is high in the south but low in the north, with an elevation between 1583 and 3703 m (Figure 2) and a slope of 0–65.72°. As a typical northwestern arid zone, the annual precipitation is about 300 mm in Gulang County. The county’s long-term poverty and ecological fragility are mainly attributed to natural disasters and poor transportation conditions. The ecological resettlement policy in Gulang County was implemented in 2010, with the relocation of migrants starting in 2015 and a phased ending at the end of 2018.

In the process of ecological migration, people moved from the Qilian Mountain National Park in the south to the Tengger Desert in the north, a special location with numerous migrants. The immigration areas comprise 12 resettlement villages (Luzhouxiaochengzhen, Xinming, Yangguang, Fuming, Fuyuan, Ganen, Yuanmeng, Weiming, Kangle, Aiming, Huiming and Leming), with a total of 62,400 immigrants, accounting for 20.2% of the total population in Gulang County. According to the masterplan, between 2018 and 2035 in Gulang, the ecological migration area is to be divided into a whole-township migration area (WTM), a whole-village migration area (WVM) and a migrant settlement area (MS). The WTM area includes three townships, i.e., Hengliang, Gancheng and Xinbao. The WVM area consists of the Shibalipu, Huangyangchuan, Heisongyi, Dingning, Peijiaying, Mingquan, Gufeng and Dajing townships. The MS area includes the Huanghuatan and Xijing townships. Considering the original families’ situations, the immigrants have all been resettled in towns and communities with good transportation conditions and flat terrains. They can enjoy the improved infrastructure and public services, although they cannot engage in the agricultural sector.

2.2. Data Sources

Multiple data sources were used in this study, including remote sensing data products, land use data, statistical yearbooks and ecological migration relocation data (

Table 1. Summary of data collection.

Data Type	Time Periods	Data Sources	Usage
MODIS NDVI	2010, 2015, 2018	NASA’s Earth Observing System MODIS/Terra Vegetation Indices 16-Day L3 Global 250 m	Reflect the vegetation condition and detect its trends
Land use	2010, 2015, 2018	Resource and Environmental Science Data Center of the Chinese Academy of Sciences (http://www.resdc.cn (accessed on 23 May 2022))	Analyze the area changes and shifts in land use
Socioeconomic statistical data	2010–2019	Government statistical bulletins and yearbook	Analyze the changes in socioeconomic development
Ecological migration data	2010–2018	Government Administration	Analyze the size of ecological migration and the location of resettlement

). The multi-temporal MODIS NDVI 16-day synthesis product was obtained from the Computer Information Center of the Chinese Academy of Sciences (http://www.gscloud.cn (accessed on 23 May 2022)). The product has a spatial resolution of 250 m and a temporal resolution of 16 d. We accordingly calculated the maximum NDVI value for each month and used its average from July to September to show the vegetation coverage for the growing seasons in 2010, 2015 and 2018. Note that the best growing months in arid zones are July, August and September; hence, we chose these three months as our baseline. The land use data were obtained from the Resource and Environmental Science Data Center of the Chinese Academy of Sciences (http://www.resdc.cn (accessed on 23 May 2022)). We used land use data with a resolution of 30 m from 2010, 2015 and 2018. Six types of land use were included: cropland, forest, grassland, water, construction land and unused land. In addition, we referred to statistical data from the Statistical Yearbook (2010–2019) as well as paperwork on poverty alleviation, relocation projects and ecological migrants, provided by the Gulang County Bureau of Statistics. The dataset contains the resident population in each township, the number of ecological migrants and the resettlement sites within Gulang County.

2.3. Methods

2.3.1. Land Use Transfer Matrix

The land use transfer matrix was used to reflect the structural change in regional land use as well as the land use trends caused by human activities [29]. The method was derived from systems analysis, which quantitatively depicts the system state and state transfer from moment T to moment T + 1. Hence, it could be used to reflect the spatial and temporal variations in land use. The specific formulation can be seen in Equation (1):

$$S_{ij}=\left[\begin{array}{cccc}{S}_{11}& S_{12}& \dots & S_{1n}\\ S_{21}& S_{22}& \dots & S_{2n}\\ \dots & \dots & \dots & \dots \\ S_{n1}& S_{n2}& \dots & S_{nn}\end{array}\right]$$

(1)

where S_ij is the area of land (km²) transferred from type i to type j; n is the number of types of land use and i and j are the types of land use at the beginning and the end, respectively.

2.3.2. Land Use Dynamic Degree

For different types of land use, the changes in land use caused by human activities are involved in the changing rate, the transfer direction and the degree of utilization [29]. The land use dynamic degree (K) can be used to reflect the change in the number of land use types [30]. Equation (2) shows the calculation of K:

$$K=\frac{U_{\mathrm{b}}-U_{\mathrm{a}}}{U_{\mathrm{a}}T}\times 100\%$$

(2)

where K is the land use dynamic degree within a certain period; U_a and U_b show the land area (km²) at the start and the end, respectively; and T is the time interval between the certain period.

2.3.3. Normalized Difference Vegetation Index (NDVI)

NDVI is the normalized ratio of the difference value between the red and near-infrared bands in remote sensing imagery. NDVI relates to the green leaf density, photosynthetically active radiation, vegetation productivity and cumulative biomass, with a linear or quasi-linear relationship. NDVI is widely regarded as one of the most effective indicators reflecting the state of vegetation cover and its growing state [31,32,33]. The difference in reflectance at different radiation wavelengths is captured to identify healthy or sparse vegetation areas. Compared to sparse vegetation, healthy vegetation can reflect more near-infrared radiation. The NDVI value is close to 1 in the healthiest woodlands and 0 on the bare ground. The NDVI value is calculated as in Equation (3):

$$NDVI=\frac{NIR-R}{NIR+R}$$

(3)

where NIR is the DN value of near-infrared bands and R represents the DN value of the red band.

2.3.4. Vegetation Restoration Degree (VRD)

Vegetation restoration greatly contributes to ecological restoration [34]. The VRD is an effective indicator of the degree of vegetation restoration in a certain period. A higher VRD means better restoration [28]. The VRD is the average NDVI between two periods, which is expressed as Equation (4):

$$VRD=\frac{NDV{I}_{T_2}-NDV{I}_{T_1}}{NDV{I}_{T_1}}$$

(4)

where NDVI_T1 represents the vegetation conditions at the initial period T₁ and NDVI_T2 represents the vegetation conditions at period T₂.

2.3.5. Contribution to NVDI Increment by Land Use (CNL)

The CNL shows the contribution of different land use categories to the NDVI change and is used to clarify how different categories and their changes contribute to changes in vegetation [28]. The formula for CNL is shown by Equation (5):

$$CN{L}_{ij}=\frac{CNDV{I}_{ij}}{{{\displaystyle \sum }}_{j=1}^{6}CNDV{I}_{ij}}$$

(5)

where i represents the initial type of land use and j is the type at the end. There are six types of land use: i, j = 1, 2, 3, 4, 5, 6. CNDVI_ij is the aggregate of the NDVI change value when the type of land use transfers from i to j.

3. Results

3.1. Spatial and Temporal Variations in Ecological Migration

Between 2010 and 2018, the population of ecological migrants in Gulang County first quickly increased and then slowly increased. In terms of spatial distribution, the ecological migration occurred from high to low elevation and from a steep to a gentle slope. Migrants settled together alongside the transportation networks (Figure 3). Between 2010 and 2015, according to the “government’s instruction and voluntary registration”, 34,700 residents moved from the WTM and the WVM areas to 12 migration resettlement sites in the north. They were from 73 poor villages in 11 poor townships located at a high elevation of 3703 m in the south. Between 2015 and 2018, the government implemented the poverty reduction policy and guided 27,700 residents in the southern mountainous areas (a slope of 65.72) to the MS area in the north, with flat terrain and good transportation and public facilities. The MS area is adjacent to a railway or highway, with convenient transportation. Furthermore, the MS area has better medical care and education. Migrants resettled in the MS area could thus improve their livelihoods. Moreover, the high concentration of resettlement increased the land use efficiency. In turn, the negative impact of human activities on the ecological environment decreased, thus improving vegetation restoration.

Before the ecological migration, people living in the southern WTM and WVM areas in Gulang County were in poverty due to the poor transportation system and the severe or fragile ecological system. The residential area had a dispersed distribution and was located on a complicated terrain. For these reasons, assistance from the local government or village cadres did not have a good effect on poverty reduction. Furthermore, there were bigger challenges regarding disaster relief. After the ecological migration, a total of 62,400 migrants resettled at the 12 resettlement sites in the north MS area. The highest emigration was to the HSY township, with 13,979 migrants, followed by the GC township with 8729 migrants. The most popular was the LZXCZ township, with 17,697 migrants (Figure 4a).

The main outmigration area was the WVM area in the National Qilian Mountain Ecological Reserve. As seen in Figure 4b, the total number of migrants that moved out of the WVM area was nearly 40,000 between 2010 and 2018, accounting for 68% (the WTM area is 32%). In addition to the goal of ecological protection, poverty reduction was also a key aim. A total of 28,812 people were in poverty during the first period of migration (2010–2015), which reduced to 20,650 in the second period (2015–2018), accounting for 79.27% of the total migrants (Figure 4c).

3.2. Spatial and Temporal Evolution of Land Use

3.2.1. The Change in Land Use Area

Changes in the land use area show the direct impact of ecological migration. As seen in Figure 5a,b, in Gulang County between 2010 and 2018, the area of cropland experienced the biggest change, while the forest land experienced the smallest change. In Figure 5c, we can see that in the WTM area, the area of forest decreased while the grassland area increased. However, the changes in forest and grassland areas were the opposite in the WVM area (Figure 5d). In the MS area, the construction land and cropland experienced a continual increase but the unused land decreased (Figure 5e). In summary, the magnitude of change in these six types of land use had the following trend: cropland > grassland > unused land > construction land > water > forest.

We calculated the land use dynamic degree (K) based on Equation (2).

Table 2. Changes in land use dynamic degree, 2010–2018.

	Region	Period	Cropland	Forest	Grassland	Water	Construction Land	Unused Land
K	Gulang	2010–2015	0.05%	−0.04%	−0.10%	0.47%	3.15%	0.01%
		2015–2018	2.98%	−0.35%	−1.68%	1.25%	3.25%	−1.57%
		2010–2018	1.15%	−0.15%	−0.69%	0.77%	3.38%	−0.59%
	WTM	2010–2015	0.01%	−0.08%	0.00%	1.88%	0.00%	−0.43%
		2015–2018	−0.07%	−24.04%	0.15%	−2.96%	0.90%	0.88%
		2010–2018	−0.02%	−9.03%	0.05%	−0.04%	0.34%	0.05%
	WVM	2010–2015	−0.01%	−0.01%	−0.01%	0.99%	0.77%	1088.89%
		2015–2018	0.69%	0.59%	−0.81%	−1.92%	−0.90%	−2.67%
		2010–2018	0.26%	0.21%	−0.31%	−0.14%	0.13%	625.00%
	MS	2010–2015	0.43%	0.00%	−0.67%	0.00%	287.50%	0.30%
		2015–2018	66.75%	−2.51%	−12.17%	85.71%	17.28%	−1.31%
		2010–2018	25.85%	−0.94%	−4.83%	32.14%	279.30%	−0.31%

Notes: K represents the land use dynamic degree.

shows the results. The magnitude of K among these six types was as follows: construction land > cropland > water > grassland > unused land > forest. The cropland area first steadily increased and then quickly increased. The grassland area slowly decreased and then steeply increased. The water and construction land areas slowly increased. The forest area also slowly decreased. The unused land first slowly increased and then experienced a steep drop.

Specifically, cropland increased by 157.22 km², from 1703.16 km² in 2010 to 1860.38 km² in 2018. The grassland reduced by 127.91 km². Between 2015 and 2018, the cropland area increased by 152.9 km²; the area of grassland and unused land decreased by 116.11 and 41.29 km², respectively; the area of water and construction land increased by 0.92 and 12.06 km², respectively; and the area of forest reduced by 1.12 km².

The types of land use varied in the different migration areas. In the WTM area (Figure 5c), cropland and water first increased and then decreased, while grassland and unused land showed the opposite trends—first declining and then increasing. Forest continually declined. Between 2015 and 2018, in the WTM area, the forest area dramatically declined from 2.44 to 0.68 km². Its land use dynamic degree was −24.04%. These changes occurred because parts of the ill forest disappeared due to a lack of manual maintenance or enough precipitation.

Meanwhile, in the WVM area (Figure 5d), cropland and forest first decreased and then increased, and water and construction land as well as unused land first increased and then decreased. As shown in Table 2, between 2010 and 2015, the unused land area increased from 0.09 to 4.99 km², with a land use dynamic degree of 1088.89%. In the process of migration, those leaving had to abandon their croplands or grasslands, which became unused lands. In addition, the vacant houses, especially the unsafe or old ones, were destroyed; thus; the unused land further increased.

Finally, in the MS area (Figure 5e), cropland and water as well as construction land showed an increasing trend, while forest and unused land first increased and then decreased. The grassland area continually decreased. Between 2010 and 2015, the land use dynamic degree for construction land was 287.5%, and that of cropland and water in 2015–2018 was 66.75% and 85.71%, respectively (as seen in Table 2). We provide some explanations for this. With an increase in new migrants, housing demand also increases, so grasslands or forests in low-quality or unused land are exploited. Moreover, the water supply has to be increased to meet the demands of agricultural irrigation and residents. Although the additional needs for cropland and construction land (for housing or infrastructures) caused the grassland or forest to decrease, the local government implemented some policies, such as urban greening and village beautification, to increase the vegetation coverage. All this rapidly improves the local ecological environment.

3.2.2. The Change in Land Use Transfer

We next investigated the impact of ecological migration on the changes in land use transfer. Between 2010 and 2018, the land use transfer showed significant changes (Figure 6a). In the first stage of ecological migration, between 2010 and 2015, housing construction and cropland use greatly increased in the resettlement area. Conversely, in the WTM and WVM areas, the cropland was mainly converted to forest and grassland. The most obvious land use change in this transformation was from grassland to cropland and construction land. The land use transfer between cropland and grassland is the largest of the many types of transfer.

In the second stage of ecological migration, from 2015 to 2018, the main tasks were the relocation of migrants and the reclamation of homesteads as well as the implementation of the national ecological protection policy of restoring farmland to grassland and forests. Therefore, the land use was transferred from grassland, unused land and construction land to cropland and from cropland and unused land to grassland. The total additional amount of land transferred is ranked as follows: cropland > grassland > unused land > construction land > forest land > water. The amount of transferred land is ranked as follows: grassland > cropland > unused land > construction land > forest land > water. The conversion from grassland to cropland was the largest, and the conversion from unused land to grassland and cropland was the largest from 2010–2015. The area converted from urban to cropland increased to 20.88 km² compared to the period of 2010–2015.

The land use transfer among the six types of land in the WTM area was the same as that in the WVM area (Figure 6b,c). Within the MS area (Figure 6d), the land use transfer showed different trends between the first and second stages of ecological migration. Between 2010 and 2015, 1.59 km² of unused land and 3.02 km² of grassland were converted to construction land. Between 2015 and 2018, 66.36 km² of grassland, 15.55 km² of unused land and 2.76 km² of construction land were transferred to cropland. During this period, the transfer of cropland was the largest (Figure 6d). However, due to the rapid outmigration, the grassland and forest lacked artificial maintenance and watering and could not adapt to the severe environment. Finally, this was turned into unused land (Figure 6c).

Between 2010 and 2015, some of the unused land and grassland was rapidly transferred to construction land to provide housing and public facilities. To improve the livelihood of 62,400 immigrants, a large amount of unused land, construction land and grassland was converted to cropland. Some of the converted construction land was from the original housing in response to the policy of “vacating residential land”. This met the additional needs of the increasing number of migrants to a certain extent.

3.3. Change in Vegetation

3.3.1. Spatial and Temporal Variations in NDVI

Between 2010 and 2018, in Gulang County, the NDVI showed a continuous increasing trend (Figure 7a). Across different areas (Figure 7b), the NDVI in the WTM and the WVM areas increased, while that in the MS area first declined and then slightly increased. The box graph shows that the change in NDVI in the MS area was insignificant compared with that in the other two areas.

The outmigration in the WTM and WVM areas reduced the impact of human activities on the vegetation. In addition, the improvement in household income and livelihood, as well as industrial adjustment, reduced the reliability of agriculture. Hence, the local ecological system received relief with the growth in areas of grassland, forest and water. Furthermore, the government implemented the policy of restoring farmland to forest and grassland, promoting the conversion of cropland and unused land to high-quality grassland and forest. This was the reason for the continuous rising trend of NDVI in Gulang County between 2010 and 2018. In the MS area, the decline in NDVI between 2010 and 2015 was due to the number of migrants, which increased the construction land demands within a short period of time. Hence, parts of the forest, grassland and unused land were converted to construction land.

3.3.2. Spatial and Temporal Variations in Vegetation Restoration

We next investigated the change in vegetation restoration in Gulang County. As shown in

Table 3. Changes in NDVI average and vegetation restoration degree, 2010–2018.

Region	Mean NDVI			Vegetation Restoration Degree (VRD)
	2010	2015	2018	2010–2015	2015–2018	2010–2018
Gulang	0.24	0.27	0.34	0.13	0.26	0.42
WTM	0.19	0.24	0.42	0.26	0.75	1..21
WVM	0.31	0.35	0.52	0.13	0.49	0.68
MS	0.16	0.20	0.28	0.25	0.40	0.75

, the average NDVI and VRD values showed an increasing trend between 2010 and 2018. Spatially, the vegetation rapidly recovered in the WTM and WVM areas (Figure 8). More than 86.3% of the raster had positive VRD values, showing the improved vegetation restoration.

The NDVI value increased from 0.19 to 0.42 in the WTM area, from 0.31 to 0.52 in the WVM area and from 0.16 to 0.28 in the MS area between 2010 and 2018. The VRD values increased from 0.26 to 0.75, 0.13 to 0.49 and 0.25 to 0.40 in the respective areas. In addition, the VRD value was higher in all three areas between 2015 and 2018 than it was in 2010–2015. Between 2010 and 2018, the VRD was 42% in Gulang County, 121% in the WTM area, 68% in the WVM area and 75% in the VRD area.

According to Figure 8, the vegetation recovery in the WTM and WVM areas was more significant than that in the MS area, although this area showed some recovery. Accordingly, we can infer that the relocation and ecological migration projects had a profound impact on protecting the ecological environment. Note that there were some negative VRD values in the MS area, as migrants moving into this area caused a shortage in water resources, providing the need to convert grassland into a reservoir.

3.4. Spatial and Temporal Changes in Vegetation Restoration among Different Land Use Types

To explore the detailed spatial distribution of vegetation restoration, we calculated the land use transfer matrix to obtain the CNL value. As depicted in Figure 9, grassland and forest showed an obvious increase in the WTM and WVM areas. Grassland increased by 2.05 km² in the WTM area and forest increased by 0.92 km² in the WVM area. In the WTM area, the land use changes included a decrease in forest and increase in grassland. However, the changes in land use in the WVM area included a decrease in grassland and an increase in forest. Due to the increasing short-term living demands that occurred after the migration, a large amount of grassland was converted to cropland in the concentration resettlement sites (the MS area). In the MS area, adjacent to the Tengger Desert, grassland and forest significantly increased. This is because the local government employed farmers to regularly plant grassland and forest. In doing so, the increase in grassland and forest could reduce the danger of desertification and improve the ecological environment in the arid zone.

We next show the CNL values in Figure 10. The transfer from forest to grassland made the largest contribution to NDVI growth, with a CNL of 0.33. The unchanged forest had a CNL level of 0.31, and grassland, following the transformation of cropland, had a level of 0.26. In addition, some types of land, when converted to grassland, made the largest contribution to NDVI growth. Other types of land, when converted to grassland, contributed the least to the growth in NDVI. All the types of land that were changed to grassland made a greater contribution to the growth in NDVI than the land transferred to forest.

During field research and interviews, we found that in the outmigration areas, the local government increased the process of restoring farmland to forest and grassland and demolished or reclaimed old housing, finally planting vegetation on the cropland. Meanwhile, the local government strictly restricted grazing around the Qilian Mountain National Park in the WVM area. Accordingly, farmers gradually stopped grazing. Reducing grazing activities improved the vegetation restoration in that area. The main reason for the increase in the NDVI of croplands was the change in crop varieties. The migrants reduced the area planted with cereals and tuber crops while increasing the area with a large amount of maize crops. Maize has a long growth cycle and relatively wide leaf cover, leading to a significant increase in the NDVI of cropland.

4. Discussion

Ecological migration has a positive effect on vegetation restoration in integral ecosystems [35]. In this case, outmigration helped to reduce human activities in the fragile ecological area. Furthermore, the land use altered as the vacant houses were demolished; thus, the grassland or forest areas increased. However, the impact of human activities on land use or vegetation restoration has two sides [36]. On the one hand, as was found in the field research, the resettlement of migrants accelerated urbanization, leading to the encroachment of numerous croplands and forests. This eventually led to a significant decrease in vegetation cover. This finding is consistent with other research [37]. We also found that the vegetation cover increased due to outmigration and the programs restoring farmland to forest and grassland. Other research also obtained a consistent conclusion [4,38].

Meanwhile, ecological resettlement had a positive impact on poverty reduction [39], as indicated by the reduction in the poor population after migration. In remote or mountainous areas, residents obtained a chance to move out of the fragile environment and live in other areas with good housing conditions, public services and facilities. With the development of electrical facilities and the rapid spread of electrical products, people reduced the cutting of traditional timber. This was an indirect means of environmental protection [28].

Ecological migration is a complex process, and assessing its impact on land use and vegetation restoration requires comprehensive consideration [9]. Previous studies have only unilaterally used land use data to study the impact of ecological migration on the ecological environment, without examining specific environmental changes through field research. However, based on the field survey, we found that mass migration in the short term would cause some risks to livelihood or ecology in the resettlement areas. In addition to reducing harm to the ecological environment [5,40,41,42], ecological migration can lead to some shocks in the social network in the resettlement areas. Noteworthily, we also found that the contribution of grasslands’ land use change to the NDVI in the dry zone was larger than that of woodlands, which may be a novel research finding. Although ecological migration would help the restoration of grassland or forests in these outmigration areas in the short term, grassland or forest areas were also reduced in the resettlement areas. If the migration programs do not respect the environmental characteristics in the arid zone, this could lead to more serious ecological problems in the long term [43]. Specifically, in the outmigration areas, artificial grassland or forest will be lost due to the lack of necessary manual maintenance. Furthermore, in the resettlement areas, the increasing demand for cropland and construction land will decrease the areas of grassland or forest, thus causing harm to the ecological environment. New migrants would increase the employment pressure in the resettlement areas. Therefore, the local government should provide more job opportunities and increase non-agricultural skills training. The government should adjust the agricultural structure and consumption structure to reduce the dependency on the land [28]. This would help to alleviate the ecological and environmental pressure.

Ecological migration aims to reduce poverty and protect the environment [28]. The migration desires of the local residents are related to poverty and the fragile ecological environment [8]. A higher poverty rate, a vulnerable ecological environment, natural disasters and poor infrastructure and public services are all determinants for outmigration [44]. In addition, the government’s compensation is the preliminary element ensuring willingness to move (Figure 11). We found that many migrants were concerned that their agricultural skills would not be sufficient for them to adapt to their new settlements. The new burdens regarding housing and high consumption provided further pressure. In addition, the inconsistency in compensation policies dramatically reduced residents’ satisfaction with the government (Figure 11).

Therefore, we put forward some suggestions to complete the ecological migration. First, before ecological migration takes place, the government should consider the existing resources and the differences in residents’ wealth. Accordingly, the government should set proper goals in combination with ecological migration and ecological protection. When the government advocates for their relocation policy, they should allow for willingness to migrate. Second, in the course of the migration, the government should support those residents who cannot afford the new housing, even offering a free house, to avoid “poverty again” [4]. Third, after ecological migration, the potential risks to the ecological system should be considered, and the government should improve the ecological compensation system [27,45]. Fourth, the government should learn more about the needs of new residents, such as their new means of livelihood. New residents should acquire more training for jobs and then upgrade their own traditional skills to establish a better life rather than relying on natural resources. Finally, the government should improve the relationship between human beings and the natural land through adjustments to the industrial structure or social structure [28,46]. In addition, the government could adopt dynamic supervision of ecological systems to achieve sustainable development.

5. Conclusions

It is of great importance that the response of land use and vegetation restoration to ecological migration is examined. We used remote sensing technology and field research data to analyze the temporal–spatial variations in ecological migration as well as its impact on land use and vegetation restoration between 2010 and 2018 in Gulang County, Gansu province, China. We found the following: (1) An increasing number of migrants moved from high-elevation to low-elevation areas and from steep areas to flatter ones. Most migrants relocated along the transportation network and became more densely clustered together. (2) Ecological migration dramatically affected the land use areas as well as the land use transformation. These impacts differed among different migration areas (the WVM, WTM and MS areas). In the WVM and WTM areas, cropland and construction land were converted to grassland and forest. In the MS area, grassland and unused land were converted to cropland and construction land. Ecological migration interfered with the unused land in the WVM area by up to 625% and the construction land in the MS area by up to 279.3%. (3) The vegetation restoration remained stable. The NDVI and VRD significantly increased; the VRD reached 121% in the WTM area and 68% in the WVM area. In addition, forest that was converted to grassland made the largest contribution to NDVI growth (0.33). The contribution of other land types that were transferred to grassland was bigger than the contribution of other lands that were transferred to forest.

We conclude that ecological migration could reduce the impact of human activities on the ecological environment, although artificial grassland or forest would be lost due to lack of maintenance after outmigration. We accordingly suggest that it is necessary to consider the natural conditions and the willingness of migrants to move out before migration. The government should provide more forest maintenance in the outmigration areas and more non-agricultural job opportunities after migration. Only with a better consideration of both environmental protection and poverty reduction can ecological migration help to achieve sustainable development in both outmigration and resettlement areas.

There are some limitations to our work, due to some unobservable factors. For example, many other indicators could be used to measure vegetation restoration, such as the enhanced vegetation index (EVI). Taking these indicators into consideration may help us to find more interesting results. Other than ecological migration, the policy of restoring cropland to forest or grassland also impacts vegetation restoration; thus, their impact should be formally considered. Future works could be carried out from these perspectives.