*2.1. Study Area Overview*

The westernmost county in the Changji Hui Autonomous Prefecture is Manas County, which is part of the Xinjiang Uygur Autonomous Region and is situated in the Manas River Basin. Its location is between 43◦21 21" and 45◦20 N, and 85◦40 to 86◦31 32" east. See Figure 1 for details. Manas County's overall in 2021 was 1.102 million hectares, including 163,000 hectares of irrigated arable land, 13 townships, 81 administrative villages, 24,311 farmers, and 43,586 rural employees, while 34,669 rural jobs can be found in the countryside (28,718 agricultural workers). The principal industry in Manas County accounted for 475.202 million yuan of the county's 1534.769 million yuan GDP in 2021 [35]. Situated on the Tianshan Mountains' northern side, Manas County is a significant agricultural production area. By the end of 2020, Manas County has passed various types of transfer. The current rural land transfer area of Manas County exceeds 25,000 hectares, and the number of households participating in land transfer is 6827, with a transfer rate of 56%, showing great agricultural development potential and research value [36].

**Figure 1.** Location and overview map of the study area.

#### *2.2. Data Source and Sample Characteristics*

Socioeconomic statistics and survey data comprise the majority of this study's data. The questionnaire data originated from the visiting survey that the research group undertook in Manas County between June and September 2021. The social and economic statistics data were taken from the "Xinjiang Statistical Yearbook" and the "Manas County Statistical Yearbook data" (Table 1). This study selected Xiazhuangzi Village, Zhangjiazhuang Village, Dawanzi Village, Sifuzhuang Village, Hongliukeng Village, Xigou Village, Dongmaidi Village, and Xibibibi Village in Letuyi Town, Lanzhouwan Town, Beiwucha Town, and Baojiadian Town, with consideration given to the difficulty of data acquisition and data integrity. A total of 600 households were chosen at random, including 200 families in nearby villages that were not involved in land transfers and 400 homes in the transferaffected township itself. A participatory farmer assessment approach [37] was employed to allow in-depth discussions with farm households, focusing mostly on the fundamental circumstances of families, such as income, savings, and educational attainment of family members. In all, 571 valid questionnaires were collected covering the two topics involved in the transfer: form of land transfer and transfer area. The 95.2% effective return rate satisfied the study's data criteria. Through surveys and interviews, the characteristics of the sample farmers were compiled (Table 2).

**Table 1.** Distribution of questionnaires in the study area.


**Table 2.** Descriptive statistics of sample farmers in the Study Area.


#### *2.3. Research Methodology*

#### 2.3.1. Division of Farmers and Land Transfer Mode

The farmers in the study area were divided into four groups based on the ratio of their non-agricultural income to total income: agricultural farmers, agricultural part-time farmers, non-agricultural farmers, and non-agricultural part-time farmers. The percentages of their non-agricultural income are shown in Table 3 as less than 10%, 10%–50%, 50%–90%, and more than 90%.

**Table 3.** Classification of farmers by type and criteria.


Based on the existing studies, four typical townships in Manas County were chosen as representatives, and two villages in each township were chosen to suggest the four most prevalent land transfer modes in the county: the farmers' spontaneous mode, the centralized continuous mode, the joint-stock cooperative model, and the leaseback and re-contracting mode. The land transfer modes were classified according to the differences in the operating agents after the land transfers. Different land transfer strategies were categorized according to how the land was managed. The farmers' spontaneous mode was categorized as individual operation, the centralized continuous mode as family operation, and the joint-stock cooperative model and leaseback and re-contracting mode were defined as collective operation. To compare variations in the changes in farmers' livelihood capital under various land transfer models, the meanings, transfer modalities, and characteristics of the various models were compiled and studied, as shown in Table 4.

#### 2.3.2. Quantitative Model of Livelihood Capital

In their research evaluating farmers' livelihoods, local and foreign academics have in recent years proposed a range of assessment index systems [41]. This present study adopted the sustainable livelihood framework (SLF), currently the most popular framework, proposed by the United Kingdom International Development Agency (DFID), taking into account a combination of economic, social, and ecological positions. This study integrated the research findings of Zhang et al. and further separates physical capital into productive capital and living capital, in order to more fully depict the influence of land on farmers' livelihoods [42]. As a result, six different types of capital were considered in this study: natural capital, financial capital, human capital, social capital, production capital, and living capital. In this article, 18 evaluation elements from six categories were chosen in accordance with the framework and survey data from cities and villages in the Manas River Basin. Table 5 details the material and assignment requirements:



**Table 5.** Farmers' livelihood capital indicator system.

The methods of determining the indicator weights were primarily hierarchical analysis, expert scoring, and the entropy method [43]. In order to eliminate subjectivity in the assignment and the repetitiveness of the indicator attributes, this study used the entropy method to determine the weights. The specific calculation process was as follows.

First, the indicator data were invariantly steeled [44]. This selection used the extreme difference standardization method to standardize the replicated data to eliminate the effect of different data magnitudes; the formula is as follows:

$$M\_{ij} = \left(X\_{ij} - \min X\_j\right) / \left(\max X\_j - \min X\_j\right) \tag{1}$$

where *Mij* is the standardized value of item i under the jth indicator, *Xij* is the value of livelihood capital assigned to the i-th farmer under the jth indicator, *minXj* and *maxXj* are the maximum and minimum values of the jth livelihood capital assigned, respectively.

The *Mij* was normalized by the formula:

$$M\_{\bar{i}\bar{j}} = M\_{\bar{i}\bar{j}} / \sum\_{i=1}^{m} M\_{\bar{i}\bar{j}} + 0.001 \tag{2}$$

where *Iij* is the normalized value, m is the farmer's value, and 0.001 is the overall shift to the right to prevent the presence of a 0 value and to facilitate subsequent calculations.

Next, the entropy values and entropy weights *ej, Wj*, of each indicator were calculated with the following equations:

$$e\_j = -1 / \ln m \sum\_{i=1}^{m} I\_{ij} \ln I\_{ij} \tag{3}$$

$$\mathcal{W}\_{\dot{\jmath}} = 1 - \mathfrak{e}\_{\dot{\jmath}} / \sum\_{j=1}^{n} (1 - \mathfrak{e}\_{\dot{\jmath}}) \tag{4}$$

where, *ej* (0 ≤ *ej* ≤ 1) is the entropy value of the jth indicator, −1/Inm is the information entropy coefficient, *Wj* is the entropy weight of the jth indicator, and n is the number of livelihood capital indicators.

Finally, the value of the livelihood capital indicator of farm households was calculated, with the magnitude of the value reflecting the level of livelihood capital. The formula is:

$$B\_{ij} = \sum\_{i=1}^{n} I\_{ij} \mathcal{W}\_{\hat{j}} \tag{5}$$

where *Bij* is the value of each livelihood capital indicator of the ith farmer, *Iij* is the normalized value of each livelihood capital indicator of the farmer, and *Wj* is the weight of each livelihood capital indicator.
