Post-relocation livelihood adaptation among resettled households in karst mountainous areas: outcomes, associated factors, and configurations in Guizhou Province, China

1 Introduction

Poverty eradication, narrowing inequality gaps, and promoting sustainable development are core issues of the United Nations 2030 Agenda for Sustainable Development (SDGs) (Li et al., 2022; Wang et al., 2025). Karst landforms account for approximately 15 to 20% of the global terrestrial area, supporting around 1 billion people, the vast majority of whom live in developing countries and ecologically fragile areas (Parise et al., 2004). Compared with other ecologically fragile areas, karst regions face rigid constraints including scarcity of cultivated land driven by rock desertification and severe surface water leakage, which make the man-land system conflict particularly acute. As a pioneer in global poverty governance, China has implemented the relocation project for poverty alleviation that aligns with its national conditions, contributing Chinese wisdom and solutions to addressing global poverty-related challenges (Liu and Zhao, 2024). By the end of 2020, China had completed the task of complete poverty eradication on schedule, securing historic achievements in the poverty alleviation campaign. Relevant studies show that relocation for poverty alleviation has achieved remarkable effects in improving farmers’ livelihoods and promoting regional development. However, some relocated populations still face problems such as unstable livelihood foundations, poor social integration, and weak self-development ability, with overall livelihood adaptation outcomes at a relatively low level and major hidden concerns regarding sustainable livelihoods (Jing et al., 2025). The situation is particularly acute in karst areas, characterized by significant rock desertification, fragmented cultivated land, and severe shortages of water and soil resources, all of which complicate the post-relocation livelihood reconstruction process. This has created an urgent need to enhance livelihood adaptation outcomes. Livelihood adaptation outcomes represent the essential capacity of farmers to navigate changes in their post-relocation environment and to withstand various risk shocks. The level of these outcomes directly influences the ability of relocated farmers to achieve stable poverty alleviation and sustainable development. During the transitional period aimed at effectively aligning the consolidation and expansion of poverty alleviation campaign outcomes with rural revitalization, addressing how to improve the livelihood adaptation outcomes of resettled farming households in karst regions has emerged as a pressing practical issue. In this context, systematically evaluating the livelihood adaptation outcomes of these households, thoroughly identifying their key associated factors and configurational paths, and proposing targeted support measures not only aligns with the requirements for regular targeted support outlined in the 2026 Central No. 1 Document but also possesses both theoretical and practical significance for enhancing regional sustainable development capacity and promoting stable poverty alleviation in karst areas.

In recent years, scholars have examined migrant livelihoods in considerable depth, with research covering livelihood capital and livelihood strategies (Caulfield et al., 2021; Huang et al., 2023), livelihood resilience (Ningi et al., 2025), livelihood vulnerability and risk (Al Mamun et al., 2023; Md et al., 2024; Yu et al., 2024), and livelihood adaptation outcomes (Hoque et al., 2018; Zhang et al., 2022a; Wu et al., 2023b). Most studies draw on the SLF developed by the U. K. Department for International Development (DFID) and approach the topic from the perspective of capital (Ehteshammajd et al., 2023; Rehman et al., 2026). Adaptive capacity has now become a prominent area of research in global change and regional response. Existing studies have mainly addressed the conceptual definition and measurement of livelihood adaptation outcomes, while giving limited attention to psychological capital. The principal measurement methods include the equal weight method, the analytic hierarchy process, and the entropy method. Engle (2011) defines adaptive capacity as the process through which individuals, when facing external disturbances, mobilize their reserve resources and make dynamic adjustments to maintain their current state. Adger et al. (2005) argue that adaptive capacity is rooted in vulnerability and resilience, and that it becomes particularly important when the external environment changes. Liu et al. (2026) quantified livelihood adaptation outcomes and dependence on ecosystem services by constructing an indicator system based on awareness, action, and capacity. From the perspective of social-ecological networks. Xiao et al. (2024) used a regression model to examine the livelihood adaptation outcomes pathways of farmers in Minqin County, along with their influencing factors and key network structures. Liu et al. (2018) drawing on an adaptive capacity framework, developed an evaluation indicator system and assessment model for livelihood adaptation outcomes capacity, and used a multinomial logistic model to analyze how migrated households’ adaptive capacity affects their livelihood adaptation outcomes strategies. Dovie and Pabi (2023) conducted a study in the Greater Accra Region, a coastal city in Ghana, using the community risk screening tool—adaptation and livelihoods (CRiSTAL), developed by the international institute for sustainable development, to analyze experiences with climate-related disasters and the overall livelihood conditions of the population. A review of existing research shows that scholarly discussion of farming households’ current livelihood adaptation outcomes still has three main areas requiring further development. First, the systematic measurement of the adaptive capacity dimension remains inadequate. Existing indicators mainly emphasize material and economic factors, with insufficient attention to farming households’ psychological identification and sense of belonging after relocation. In addition, previous literature has predominantly treated economic, social, and psychological adaptation as either parallel processes or sequential stages. However, whether these dimensions progress synchronously under specific ecological migration contexts remains a critical gap. This study, therefore, shifts focus from measuring overall levels to diagnosing inter-dimensional asynchrony. Second, analyses of the mechanisms influencing livelihood adaptation outcomes are often restricted to the linear and symmetric net effects identified by regression models, making it difficult to capture nonlinear interactions and synergistic effects among multiple factors. At present, there are few precedents for directly using quantitative measures derived from micro-level survey data as outcome variables in fsQCA, and there is also a lack of coupling with an optimal-parameter geodetector. In addition, no systematic analysis has yet been identified of the potential explanatory pathways through which concurrent multiple factors and conditional configuration pathways shape livelihood adaptation outcomes. The fsQCA views cases as configurations of multiple antecedent conditions. This analytical approach identifies shared membership relationships in the data by comparing samples rather than relying on variable correlations, thereby revealing complex explanatory configurations associated with the outcome and offering greater robustness than traditional net-effect methods (Fiss, 2011). Third, current research on Livelihood adaptation outcomes are largely dominated by qualitative approaches, with relatively few studies integrating quantitative and qualitative methods at the household level; studies focusing on farmers’ adaptive capacity to new environments after relocation in karst settings are even rarer.

This study focuses on farm households in karst mountainous areas and utilizes the SLF along with adaptive capacity theory to conduct the following research. The main contributions and potential marginal advancements of this work are as follows: (1) it integrates psychological adaptation outcomes into the analytical framework, constructs a comprehensive evaluation indicator system encompassing five dimensions—economic, employment, living, social, and psychological—and employs the CRITIC composite index to assess farm households’ livelihood adaptation outcomes and analyze the status of each dimension; (2) it utilizes the optimal parameters-based geographical detector (OPGD) to identify the associated effects of individual factors and their interactions on adaptation outcomes, thereby elucidating the nonlinear enhancement mechanisms among these factors; and (3) it incorporates fuzzy-set qualitative comparative analysis (fsQCA) to identify, from a configurational perspective and through a combined qualitative and quantitative approach, multiple equivalent conditional pathways that lead to high adaptation outcomes. It aims to answer the following questions: (1) What is the level of multidimensional livelihood adaptation outcomes among migrated households in karst mountainous areas, and what differences exist? (2) Which factors are independently and interactively associated with farmers’ livelihood adaptation outcomes, and what characteristics do these interactive associations exhibit? (3) Are there any single-factor necessary conditions for high livelihood adaptation outcomes among farmers, and which configurations of conditions are associated with high livelihood adaptation outcomes? This study aims to provide a scientific reference for consolidating the achievements of poverty alleviation through relocation in karst mountainous areas. It supports local governments in formulating targeted policies and promotes the sustainable development of the livelihoods of relocated households. Additionally, it offers empirical evidence for similar ecologically fragile areas worldwide in the coordinated advancement of SDG₁ and SDG₁₀.

2 Materials and methods

2.1 Theoretical analysis

The SLF developed by the UK Department for International Development (DFID), grounded in the perspective of livelihood capital, has been widely adopted (Scoones, 2009; Adger, 2010). It seeks to strengthen the ability of vulnerable groups to withstand risks, meet basic living needs, and accumulate assets, while also supporting the formulation of effective interventions through the identification of key factors affecting livelihoods (Fitawok et al., 2022). Nevertheless, the SLF has inherent limitations: its characterization of livelihood systems tends to emphasize static forms of capital while giving insufficient attention to dynamic processes of adaptation outcome (Maldonado and Moreno-Sánchez, 2014). Adaptive Capacity Theory addresses this limitation by focusing on the ability of a system to maintain its core functions through learning and reorganization under conditions of disturbance (Engle, 2011; Batenga et al., 2024). Based on these theories, and considering the complexity and regional characteristics of migrated farmers in Guizhou Province, this study deconstructs livelihood adaptation outcomes into five dimensions: economic, occupational, living, social, and psychological adaptation outcomes. This framework retains the SLF’s attention to multiple capital dimensions while incorporating the adaptive capacity literature’s emphasis on dynamic adjustment and subjective perception (Figure 1). Critically, this integration generates three novel insights. First, it enables the diagnosis of asynchronous adaptation across dimensions, rather than treating adaptation as a homogeneous process. Second, it shifts the analytical focus from independent net effects to configurational equifinality, allowing us to examine how different combinations of conditions may be associated with similarly high adaptation outcomes. Third, it extends the SLF by explicitly incorporating psychological capital as a constitutive dimension of adaptation outcomes, thereby addressing a long-standing gap in the framework.

Under the multiple pressures of a new environment, migrated farmers face significant challenges in reconstructing and developing their economic, occupational, living, social, and psychological adaptation outcomes. They must therefore develop livelihood adaptation outcomes that facilitate adjustment to a relatively unfamiliar livelihood system. These dimensions are interrelated and mutually supportive. Economic adaptation outcomes serve as the core foundation for ensuring the livelihood stability of migrated farmers, as they determine the material conditions necessary to maintain basic living standards in the new environment. Occupational adaptation outcomes provide essential support for economic adaptation, as stable employment generates sustained income and acts as a driving force for enhancing livelihood adaptation outcomes. Living adaptation outcomes represent the external manifestation of farmers’ integration into the new community, reflecting the supportive role of post-relocation lifestyles and living conveniences in their livelihoods. Social adaptation outcomes are crucial for the reconstruction of social networks and relate to farmers’ ability to obtain livelihood resources through new social ties. Psychological adaptation outcomes involve the rebuilding of positive expectations for the future and recognition of self-worth, thereby consolidating the internal motivation for long-term settlement. The five dimensions interact with one another, enabling households to achieve sustainable livelihoods. Additionally, the measurement of livelihood adaptation outcomes are used to identify the adaptation outcome levels of different households, categorizing them into three groups: high adaptation outcomes, moderate adaptation outcomes, and low adaptation outcomes. This classification provides a basis for governments and individuals to formulate differentiated policy management measures. Such processes further enhance households’ livelihood adaptation outcomes in terms of development capacity, income structure, human settlement well-being, and community integration, thereby contributing to a positive cycle between livelihood outcomes and sustainable development.

This outcome-oriented framework provides the conceptual basis for the subsequent OPGD and fsQCA analyses, which examine the associated factors and configurational patterns linked to high livelihood adaptation outcomes.

2.2 Overview of the study area

Located in western Guizhou Province, the prefectures of Bijie, Liupanshui, and Anshun are situated in a descending terrain from west to east, characterized by rugged and highly fragmented topography, as well as extensive typical karst landforms. All three prefectures fall within a subtropical monsoon climate zone, characterized by mild winters and cool summers, and form a contiguous area of severe karst rocky desertification. The inherently fragile ecological baseline, compounded by acute scarcity of arable land and extreme land fragmentation, has long established this region as a priority zone for China’s poverty alleviation and relocation programs. By the end of 2025, Bijie had a permanent resident population of 6.61 million and a gross regional product (GRP) of 254.11 billion CNY; Liupanshui had 3.03 million residents and a GRP of 177.78 billion CNY; and Anshun had 2.47 million residents and a GRP of 123.24 billion CNY. The three prefectures are also multi-ethnic settlement areas, with Han, Miao, Yi, and Bouyei populations intermingled, resulting in pronounced cultural diversity. During the 13th Five-Year Plan period, each prefecture undertook large-scale relocation efforts: Bijie alone relocated 0.28 million people, while Liupanshui and Anshun faced equally urgent post-relocation development challenges. The extent to which relocated households can adapt their livelihoods to the receiving environment directly determines their long-term welfare, the effectiveness of ecological restoration in the depopulated source areas, and the depth of the ongoing rural revitalization strategy. Therefore, we selected this region as the study area (Figure 2) due to its representativeness in revealing the livelihood adaptation patterns of relocated populations in contiguous karst rocky-desertification zones, and its direct policy relevance in refining targeted follow-up support measures.

2.3 Data sources

This study used semi-structured interviews to conduct field research and obtain primary data, combining the breadth of quantitative inquiry with the depth of qualitative analysis. The research team first conducted a pilot survey and then revised the questionnaire. The official survey data was collected in July 2024 from respondents selected from 12 representative resettlement communities in Bijie, Liupanshui, and Anshun. The selection of these 12 communities was sufficient to meet the diversity requirements for comparative analysis and fully reflected the heterogeneity of resettlement types and geographic environments in the karst region. The detailed screening procedure, the candidate pool, and the basic characteristics of the 12 selected communities are presented in Appendix A. The selection criteria included: (a) resettlement type (county-level and non-county-level resettlement communities); (b) community size; and (c) survey accessibility. This purposeful sampling design ensures that the selected communities reflect, from an analytical perspective, the diverse real-life circumstances faced by the relocated families. In each community, we obtained a complete list of all relocated households from community administrators. We then applied a stratified sampling design to each community, with strata based on city (Bijie, Liupanshui, and Anshun) and resettlement location (county-level and non-county-level communities), resulting in a total of six strata. Approximately 20–30 households were sampled from each community. Within each stratum, random sampling was conducted using Excel’s RAND function. The survey was conducted through face-to-face interviews lasting 20–30 min, with the assistance of local interpreters to overcome language barriers. A total of 352 questionnaires were distributed across 12 communities. After rigorous data cleaning, 46 questionnaires were excluded for the following reasons: (1) Incomplete responses (n = 24); (2) Logical inconsistencies (n = 12); (3) Responses exhibiting linear or formulaic patterns (n = 10). Ultimately, 306 valid responses were obtained, resulting in a response rate of 86.93%. The head of each selected household or the primary decision-maker regarding household livelihood participated in the survey as a respondent. To ensure the validity and reliability of the questionnaire data, reliability and validity tests were conducted. In this study, SPSS 27.0 was used to test the reliability and validity of the questionnaire. The overall Cronbach’ s alpha coefficient of the questionnaire was 0.714; the KMO statistic was 0.773, and Bartlett’ s test yielded p = 0.00. These results indicate that the questionnaire had good reliability and construct validity. The questionnaire covered the following content: ① basic household information on rural households, including gender, age, household size, and educational attainment; ② livelihood assets, including household income, areas of cropland and forestland, and community networks; and ③ post-relocation adaptation, including satisfaction with various aspects such as sense of community belonging and integration of ethnic customs. In the subsequent results section, we use a coding system to illustrate representative cases to illustrate typical configurational patterns, codes R1-R6 represent representative samples selected from all 12 communities, F1-F3 represent the rural household respondents, and C1-C2 represent the cadre respondents. Table 1 presents the basic characteristics of the surveyed farming households.

Table 1. Characteristics of the farmers interviewed.

2.4 Research methods

2.4.1 Construction of the indicator system

2.4.1.1 Construction and description of the livelihood adaptation outcomes measurement indicator system

In this study, we differentiate among three interrelated yet distinct concepts: adaptive capacity, adaptation process, and adaptation outcomes. (a) Adaptive capacity refers to the latent and dynamic ability of households to learn, reorganize resources, and respond to disturbances; it is challenging to measure directly using cross-sectional survey data. (b) The adaptation process denotes the temporal sequence of adjustments following relocation, necessitating longitudinal data for observation. (c) In contrast, adaptation outcomes pertain to the current post-relocation conditions of households, encompassing economic status, employment status, living conditions, social integration, and psychological well-being. The composite index developed in this study assesses multidimensional post-relocation livelihood adaptation outcomes, rather than adaptive capacity itself. Therefore, throughout the manuscript, the term livelihood adaptation outcomes are employed to denote the measured construct, while adaptive capacity is reserved for theoretical discussions and is explicitly acknowledged as not being directly measured.

Grounded in the SLF and adaptive capacity theory, and tailored to the actual conditions of relocated farmers in karst mountainous areas, this study characterizes farmers’ livelihood adaptive capacity across five dimensions: economic adaptation outcomes occupational adaptation outcomes, living adaptations, social adaptation outcomes, and psychological adaptation outcomes (Table 2). Drawing on the research of Zhao and Ren (2022) and Mahedi et al. (2026), (1) economic adaptation outcomes represents a critical condition for relocated farmers to sustain basic livelihoods and accumulate assets in resettlement areas, thereby reducing their dependence on external assistance. Within this dimension, annual household income per capita (EAO₁) reflects farmers’ economic productive capacity and serves as a core indicator of livelihood adaptive capacity; Types of household income (EAO₂) indicate income diversity. Highly diversified income sources are associated with greater economic stability and may reduce reliance on a single agricultural sector, although this study does not directly measure the capacity to withstand market shocks; and household consumption levels (EAO₃) captures both actual purchasing power and shifts in living standards. Referencing studies by Mushtaq et al. (2026) and Zou et al. (2020), (2) occupational adaptation outcomes measure the flexibility of relocated households in integrating into the labor market and securing stable income, which directly influences income growth and risk resilience. Here, the non-farm employment rate (OAO₁) reflects the transition of relocated farmers from traditional agriculture to modern economic sectors; the location of employment areas (OAO₂) indicates job accessibility and commuting distance, noting that while out-of-province employment may yield higher income, it also entails greater social costs; and employment stability (OAO₃) denotes the capacity of households to secure long-term, steady income, which is fundamental to establishing sustainable livelihoods. Drawing on studies by Zhang et al. (2022b) and (Hu et al., 2026), (3) living adaptation focuses on relocated farmers’ satisfaction with their residential environment, infrastructure, and public services, which collectively determine daily convenience and comfort. Specifically, comfort of the residential environment (LAO₁) influences how well migrated farmers accept their new community; level of infrastructure development (LAO₂) assesses physical facilities such as water, electricity, roads, and telecommunications; and accessibility of public services (LAO₃) encompasses satisfaction with healthcare, education, transportation, shopping, and community services, directly determining how efficiently farmers’ basic needs are met. Drawing on the work of Liu et al. (2025) and Sun (2022), (4) social adaptation outcomes reflect how farmers establish social ties through neighborhood interaction, community participation, and social trust to secure information and support. In this regard, frequency of neighborhood interactions (SAO₁) indicates daily socialization, forming the foundation of primary social networks; community event participation (SAO₂) reflects farmers’ willingness to integrate into public life, facilitating access to information and collective resources; and level of social trust (SAO₃) measures their sense of reliance on and security with neighbors and community staff, as a high-trust environment fosters cooperation and interaction among relocated households. Informed by the studies of Tang et al. (2026), (5) psychological adaptation outcomes capture relocated farmers’ sense of community belonging, life satisfaction, and confidence in future development shaped during spatial reconstruction and identity transformation, serving as a key adaptation factor for sustaining long-term mental health and positive behaviors. Within this dimension, sense of belonging in the community (PAO₁) is a core indicator of identity formation and emotional attachment to the resettlement site; sense of well-being (PAO₂) reflects a comprehensive, objective evaluation of overall living conditions, serving as a direct measure of welfare; and confidence in future development (PAO₃) represents positive expectations for the long-term prospects of families and children, which motivates continued engagement in livelihood-enhancing activities.

Table 2. Construction of a livelihood adaptation outcomes indicator system.

2.4.1.2 Selection and description of associated factors

To investigate the associated factors and configurational structure of livelihood adaptation outcomes, this study comprehensively selects multidimensional factors affecting farm households’ livelihood adaptation outcomes as detection factors and configurational factors. By running the OPGD and fsQCA models, it conducts an in-depth analysis of the associated factors of farm households’ livelihood adaptation outcomes and the synergistic effects of different factor configurations on farm households’ livelihood adaptation outcomes. Drawing on existing research and proceeding from internal capital and external structure (Tu et al., 2024; Xu et al., 2024), the study selects nine variables covering family characteristics, economic conditions, geographical environment, community governance, social security, and cultural integration (Table 3). It is worth noting that the X₄ indicator measures households’ subjective satisfaction with their income levels; this indicator is conceptually distinct from the objective per capita income included in the outcome index. However, since both variables reflect the economic dimension of livelihoods, caution should be exercised when interpreting their relationship with the outcome index.

Table 3. Variable selection and description.

2.4.2 Measurement of farmers’ livelihood adaptation outcomes

This study first applies range standardization to the measurement indicators to prevent result bias caused by dimensionality issues. The CRITIC method determines the effect of indicator weights by comprehensively considering the internal variability of evaluation indicators and the degree of association among indicators (Pan and Chen, 2026). Compared with subjective weighting methods such as the analytic hierarchy process, its weight allocation is more objective; compared with the entropy weight method, which assigns weights entirely based on the degree of data dispersion, the CRITIC weighting results are also more robust. Therefore, this study constructs a CRITIC composite index to comprehensively measure the livelihood adaptation outcomes of migrated farmers.

W j = C j ∑ j = 1 n C j = S j × R j ∑ j = 1 n ( S j × R j ) = ∑ i = 1 m ( x ij − x j ) 2 m − 1 × ∑ k = 1 n ( 1 − r jk ) ∑ j = 1 n [ ∑ i = 1 m ( x ij − x j ) 2 m − 1 × ∑ k = 1 n ( 1 − r jk ) ] (1)

LAOI i = ∑ j = 1 n x ij W j (2)

In Equations 1, 2, m denotes the number of samples to be evaluated; n represents the total number of evaluation indicators; r jk is the Pearson correlation coefficient between indicator j and indicator k; S j , R j , C j signify the variability, conflict, and information quantity of the j-th indicator, respectively; W j stands for the assigned weight of the j-th indicator; x ij represents the dimensionless value of the j-th indicator for the i-th sample; and LAO I j denotes the comprehensive livelihood space index for the i-th household.

2.4.3 Optimal parameter-based geographical detector

Compared with the conventional GeoDetector, the optimal parameters-based geographical detector (OPGD) effectively identifies the optimal combination of spatial discretization methods and scale parameters, providing a robust scientific basis for accurately revealing the associations between dependent and independent variables (Wang et al., 2024). By offering diverse discretization strategies for continuous variables, this method substantially reduces subjective intervention, thereby enhancing the objectivity and accuracy of the analytical results (Song et al., 2020). In this study, continuous associated factors were discretized using the equal interval, quantile, natural breaks, and standard deviation methods. An OPGD model was then constructed using the “GD” package in R to determine the optimal parameter combination, which was subsequently applied to identify the factors influencing farmers’ livelihood adaptation outcomes and their interactions.

q = 1 − ∑ h = 1 L N h δ h 2 N δ 2 (3)

In Equation 3, the q statistic represents the explanatory power of an independent variable on the LAOI; a higher q value indicates stronger explanatory power. The variable h denotes the stratification of the independent variable, where L is the total number of strata. N and N h represent the number of sample units in the entire study area and in stratum h, respectively. Similarly, σ , σ 2 denote the variance of the LAOI across the entire study area and within stratum h, respectively.

The interaction detection analysis examines whether the joint influence of two factors on the dependent variable operates independently, is strengthened, or is attenuated. Based on the resulting interactions, five categories are identified: nonlinear weakening, univariate weakening, bivariate enhancement, independent, and nonlinear enhancement (Table 4).

Table 4. Interaction judgment rules and results.

2.4.4 Fuzzy-set qualitative comparative analysis

Farmers’ livelihood adaptation outcomes are influenced by a confluence of multiple factors rather than a single determinant (Fiss, 2011). This study employs fuzzy-set Qualitative Comparative Analysis (fsQCA) to perform a multilevel and multidimensional configurational analysis grounded in configurational thinking, Boolean logic, and counterfactual analysis. The objective is to elucidate the sufficiency of various factor configurations in enhancing farmers’ livelihood adaptation outcomes within karst mountainous regions and to identify the corresponding configurational pathways, thereby offering novel research perspectives. Additionally, fsQCA necessitates the assessment of consistency and coverage to ascertain optimal configurational patterns.

Consistency ( X ≤ Y ) = ∑ min ( x i , y i ) / ∑ x i (4)

Coverage ( X ≤ Y ) = ∑ min ( x i , y i ) / ∑ y i (5)

In Equations 4, 5, X and Y denote the sets of condition variables and the outcome variable, respectively, while x i , y i represent the fuzzy-set membership scores of the variables within sets X and Y.

3 Results

3.1 Overall description of farmers’ livelihood adaptation outcomes levels

The overall level of farmers’ livelihood adaptation outcomes varies considerably (Figure 3), with a maximum value of 0.876, a minimum value of 0.374, and a mean index of 0.646. The original values are relatively concentrated and evenly distributed. The mean of the livelihood adaptation outcome indices is lower than the median, indicating a left-skewed distribution. This pattern suggests that while a small number of households exhibit extremely low adaptation outcomes that skew the mean downward, the majority of farmers achieve relatively high adaptation outcomes. Consequently, 55.89% of the sample falls within the higher range of adaptation outcomes, whereas only a limited number of farmers exhibit low adaptation outcomes.

At the dimensional level, Figure 3 illustrates that economic and occupational adaptation outcomes are relatively low compared to the other three dimensions. To further assess the statistical significance of this finding, a repeated-measures ANOVA was performed on the scores across the five adaptation outcome dimensions. Mauchly’s test of sphericity indicated a violation of the sphericity assumption (p < 0.05), necessitating the application of the Greenhouse–Geisser correction. The results demonstrated a significant main effect of adaptation outcome dimension, F_{(4, 1,220)} = 319.906, p < 0.001, partial η² = 0.512. Post-hoc pairwise comparisons with Bonferroni correction revealed that all pairwise differences among the five dimensions were statistically significant. Notably, psychological adaptation outcomes were significantly higher than those of all other dimensions (p < 0.001). Additionally, living and social adaptation outcomes were significantly higher than both economic and occupational adaptation outcomes (p < 0.001). Economic and occupational adaptation outcomes were identified as the lowest among the five dimensions (p < 0.001). This suggests that most migrated farmers recognize the current relocation and resettlement policies and report relatively high satisfaction with their new living environment. Through local acquaintance-based social networks, they can obtain basic livelihood support, maintain social connections, and receive psychological comfort. Even so, clear weaknesses remain in income accumulation and access to non-farm employment opportunities. Among these dimensions, occupational adaptation outcomes show the greatest individual-level variation and has the highest interquartile range, indicating marked differentiation among rural households in this respect. The survey found that some rural households in the study area are able to earn income through migrant work, while others cannot secure stable income support because they lack the labor capacity required for non-farm employment. In the original values, economic adaptation outcomes are relatively concentrated, whereas the other dimensions are more dispersed, suggesting imbalance in their internal structures and a relatively high proportion of low values in occupational adaptation outcomes. As presented in Figure 4, the mean score of all indicators under the three dimensions of psychological, living, and social adaptation outcomes is 0.26. In contrast, the mean value across the two dimensions of economic adaptation outcomes and occupational adaptation outcomes is 0.17, with particularly significantly lower scores observed for per capita annual household income and the non-farm employment rate. This is the main reason why their economic adaptation outcomes and occupational adaptation outcomes indice remain low, indicating that rural households’ livelihood adaptation outcomes continues to face clear constraints. Most rural households have not yet made a stable transition from traditional agricultural livelihoods to non-farm livelihoods; their income sources remain relatively limited, and the threshold for obtaining stable non-farm employment remains relatively high. The survey found that most middle-aged and older residents in the resettlement areas lack the labor capacity needed for non-farm employment. During the busy farming season, they can only return to their places of origin to cultivate crops such as corn and rice, and the low income and long production cycles associated with this farming contribute to weak household income. After relocation, farmers developed a strong sense of community belonging, indicating that relocation improved their living conditions and, to some extent, preserved familiar social ties by creating interaction spaces with ethnic cultural symbols, such as drum towers and cultural squares. Overall, their integration at the psychological and living levels has proceeded relatively smoothly.

3.2 Analysis of differences among farmers under different livelihood adaptation outcomes levels

To examine the overall characteristics of livelihood adaptation outcomes among migrated farmers, this study employed K-means cluster analysis to categorize farm households in the region into three distinct groups based on their livelihood adaptation outcomes index: the relatively advanced type, the stable development type, and the needs improvement type. Subsequently, a one-way ANOVA was conducted to evaluate the validity of this classification. The results revealed an F-statistic of 820.565 with a significance level of p = 0.000, indicating statistically significant differences among the three groups. The relatively advanced households constituted 34.97% of the sample, with a mean value of 0.743; stable development households made up 44.44%, with a mean value of 0.639; and needs improvement households represented 20.59%, with a mean value of 0.497. These findings suggest that the overall livelihood adaptation outcomes of farm households in the region is relatively high. At the same time, a clear gradient gap exists between low-adaptation households and those with medium or high adaptation outcome (Figure 5a), indicating that some households have adapted more slowly to the new environment after relocation. The three groups differ only slightly in economic adaptation outcomes, living adaptations, social adaptation outcomes, and psychological adaptation outcomes, but they show more pronounced variation in occupational adaptation outcomes. The occupational adaptation outcomes score of low-adaptation households are only 0.281, the lowest across all dimensions, suggesting that employment remains the main constraint on improving the livelihood adaptation outcomes of low-adaptation farm households.

Differences were observed in the mean values of indicators such as per capita annual household income, non-farm employment rate, and social trust relationships among farm households with different levels of livelihood adaptation outcomes (Figure 5b). A gap was observed between relatively advanced households and households in need of improvement, which may help explain why the livelihood adaptation outcomes index of the latter tended to be lower than that of the former. In our sample, households with low adaptation levels are disproportionately concentrated in township centers or central villages, accounting for 70% of this subgroup. Although this pattern has not undergone formal statistical testing, it is supported by field observations, suggesting a potential relationship between the resettlement site and adaptation outcomes. The majority of these households rely primarily on crop cultivation for their livelihoods. In some cases, their inputs are not matched by corresponding outputs, resulting in weak and unstable incomes. They also have difficulty obtaining the same level of resource provision as medium- and high-adaptation households resettled in county seats, face limited access to employment information, and have fewer opportunities for vocational skills training. These conditions further restrict their potential to shift into nonfarm industries and widen the gap among households with different levels of adaptation.

3.3 Exploring the associated factors of farmers’ livelihood adaptation outcomes

To provide an in-depth explanation of the correlations and configurational relationships between various influencing factors and farmers’ livelihood adaptation outcomes, this study integrates the OPGD and fsQCA models. Combining these two models helps overcome the limitations of a single method by enabling the independent assessment of the relative importance of each factor while also deeply uncovering the complex associated factors and configurational patterns linked to high livelihood adaptation outcomes that emerge from multifactor interactions.

3.3.1 Results of the optimal parameter-based geographical detector

3.3.1.1 Single-factor detection

The single-factor detection results (Figure 6a) indicate the influence intensity of each driving variable on the livelihood adaptation outcomes of migrated farmers. Overall, the selected factors show strong explanatory power, although their driving intensities vary. Household income level (X₄) is the primary core factor, followed by the comprehensiveness of medical and pension insurance coverage (X₈) and family health status (X₂), with q-values of 0.390, 0.379, and 0.307, respectively. These three factors are the main associated factors currently linked to farmers’ livelihood adaptation outcomes. Income is closely associated with farmers’ consumption capacity, production capacity, and ability to withstand risks at the resettlement sites after relocation, and may be relevant to adaptation to the new environment. The completeness of medical and old-age insurance coverage reduces the risk that farmers will fall into poverty because of illness or loss of working capacity in old age, and therefore provides an institutional basis for long-term livelihood stability. Household health status is closely related to the working capacity of the main labor force; poor health not only weakens income-generating capacity but also occupies the labor time of other household members. Taken together, these three factors suggest that economic capital, social capital, and human capital provide the core support for farmers’ livelihood adaptation outcomes. In addition, the q values for the intensity of community public safety enforcement efforts (X₇) and the degree of integration into community living habits and cultural customs (X₉) both exceeded 0.20, indicating relatively strong explanatory power for livelihood adaptation outcomes. By contrast, the q values for slope (X₆), the householder’s level of education (X₁), and whether family members’ positions in the community or government agencies (X₃) were relatively low, suggesting weaker explanatory power. The elevation of the resettlement site (X₅) showed no significant association with farmers’ livelihood adaptation outcomes (q = 0.093, p > 0.5). This non-significant finding may be attributed to the relatively limited variation in elevation among the sampled resettlement sites, all situated within the karst mountainous region. Furthermore, government-led resettlement programs often standardize infrastructure and public service provision, potentially diminishing the impact of natural topographic factors such as elevation on adaptation outcomes. These results align with prior studies that suggest socio-economic and policy factors generally exert a greater influence than physical geographic factors in determining post-relocation adaptation (Hu and Zhao, 2025). Within the study area, except for the R1 community in Bijie and the R2 community in Liupanshui, which are located at relatively high elevations, the elevations of the other resettlement sites do not differ substantially.

3.3.1.2 Interaction effect of factors

Figure 6b reports the interaction detection results. Most interaction pairs show enhanced explanatory power relative to their respective single factors, indicating bivariate enhancement or nonlinear enhancement. Nonetheless, not all interactions exceed the single factor with the highest q-value. These findings imply that livelihood adaptation outcomes are jointly associated with multiple factors, with interactions generally strengthening explanatory power but not uniformly across all pairs. Specifically, the q-values for the interactions between household income level (X₄) and all other factors exceed 0.4, suggesting that X₄, as a fundamental core factor, substantially enhances the explanatory power for livelihood adaptation outcomes when coupled with other variables. Consequently, X₄ is identified as the most important interactive explanatory factor of farmers’ livelihood adaptation outcomes, corroborating the single-factor detection result that X₄ had the strongest individual explanatory power. The interaction X₄∩X₆ demonstrates an enhanced explanatory power, indicating that households with higher incomes, situated on gentler slopes, are more likely to achieve favorable adaptation outcomes. Higher income may allow households to select resettlement sites that are characterized by gentler slopes and improved conditions. Conversely, relocating to areas with gentler slopes may facilitate income growth by enhancing transportation and production conditions. Given the cross-sectional design, the observed interaction along these two pathways should be interpreted as a pattern of association. Gentler slopes imply flatter terrain, which may be associated with easier agricultural operations and lower costs of supporting infrastructure. For example, households in flatter areas may engage in larger-scale crop cultivation through land transfers, while households in steeper areas may rely on specialty crops or local tourism to generate employment and income. This interactive pattern further illustrates the complex, potentially bidirectional relationships among factors associated with livelihood adaptation outcomes.

The interaction values of X₇, X₈, and X₉ with other variables all exceeded 0.2 (Figure 6b), indicating diverse interaction types, including bivariate enhancement, nonlinear enhancement, and univariate weaken. This complexity indicates that the combined effect of these variables can enhance the joint explanatory power in some combinations, but may lead to a decline in explanatory power in others. Despite this, the consistently high q value emphasizes that variables such as community safety management, medical and elderly care coverage, and integration into local lifestyles and cultural customs are interaction conditions significantly associated with livelihood adaptation outcomes. For instance, R3 community in Bijie city has set up a one-yuan lunch canteen for elderly residents aged 60 and above. Meanwhile, the R4 community in Bijie city, the R5 community in Liupanshui city and the R6 community in Anshun city offer targeted volunteer services such as free haircuts, medical care and after-school tutoring for the elderly and children. Figure 6b also reveals that the interactive q-values of altitude (X₅) with other factors are generally higher than the individual q-value of X₅. This indicates that while altitude exerts limited direct influence on its own, its impact on livelihood adaptation outcomes are substantially amplified when intersecting with other resources and conditions. Only when integrated with complementary factors, such as local industrial policies and transportation infrastructure, can the role of altitude in shaping the livelihood adaptation outcomes of relocated farmers be fully realized.

3.3.2 Results of fuzzy-set qualitative comparative analysis

Livelihood adaptation outcomes are a systemic process shaped by the interaction and interconnection of multiple factors. Using the fsQCA method, this study examines the sufficiency of different configurations of conditional variables in the process of livelihood adaptation outcomes, thereby identifying the multiple factors associated with its full realization. The livelihood adaptation outcomes index is used as the outcome variable, and X₁-X₉ are used as the condition variables. Following the existing literature (Ding et al., 2026; Rasoolimanesh, 2026), the direct calibration method was applied in fsQCA 4.1, with the thresholds for full membership, the crossover point, and full non-membership set at 95, 50, and 5%, respectively. These percentile-based anchors were selected for two primary reasons. First, they reflect the empirical distribution of each condition variable in our sample, rather than relying on arbitrary theoretical values. This method minimizes the impact of extreme outliers on membership scores. Second, in comparison to less extreme thresholds, such as the 75 and 25% anchors, the 95 and 5% anchors provide a clearer distinction between cases that are fully within or fully outside a condition. This reduces the proportion of cases with ambiguous membership around the crossover point, thereby enhancing the interpretability of the resulting configurations. The consistency threshold for necessary conditions, PRI, and the case frequency threshold were set at 0.8, 0.8, and 2, respectively.

3.3.2.1 Necessity analysis of individual conditions

To prevent necessary conditions from being misclassified as logical remainders during truth table analysis and subsequently eliminated from the parsimonious solution, a necessary condition analysis must precede the configurational analysis. Following the established necessity analysis threshold (consistency>0.9) recommended in the fsQCA literature (Schneider and Wagemann, 2012), we assessed whether any individual condition qualifies as necessary for achieving high livelihood adaptation outcomes. As shown in Table 5, the consistency values for the nine core variables and their negations are all below 0.9, indicating that no single variable independently constitutes a necessary condition for the full realization of migrated farmers’ livelihood adaptation outcomes. This finding confirms that the reshaping of livelihood adaptation outcomes in the study area is not determined by a single factor, aligning with the OPGD detection results. Consequently, all variables must be integrated into the fsQCA model to systematically reveal the multiple conditional configurations and underlying mechanisms driving the full realization of livelihood adaptation outcomes.

Table 5. Results of the necessity analysis.

“~“denotes the logical operator “not”; Consistency measures the extent to which each solution is a subset of the set of results; coverage indicates the extent to which each solution covers the results. A condition is considered necessary if its consistency exceeds 0.9 (Schneider and Wagemann, 2012).

3.3.2.2 Multivariate configurational path analysis

As shown in Table 6, there are six configurations associated with high livelihood adaptation outcomes, and the consistency of each valid path exceeds 0.9, meeting the consistency standard for configuration analysis. Among all cases that satisfy these 6 valid configurations, more than 90.6% of the outcomes are generated through the combined effect of nine conditional variables, indicating a high level of consistency and constituting a sufficient condition. The overall coverage of the model is 0.537, suggesting that these six configurations can effectively explain 53.7% of the cases with high livelihood adaptation outcomes, while also revealing the important reasons behind the realization of livelihood adaptation outcomes as a result variable. Representative instances from typical cases are selected for further analysis, and the driving configuration paths are summarized as follows:

Table 6. Configuration of associated factors for fully achieving livelihood adaptation outcomes among households.

Solid dark blue circle on a white background. Indicates that the core condition is present; Image is too blurry to identify specific content or details. No distinguishable objects or text are visible. indicates that the boundary condition is present; Red circle with a bold red X through the center, commonly used as a symbol to indicate something is not allowed, incorrect, or not selected. indicates that the core condition is missing; Green recycling symbol featuring three arrows forming a circle with four white spaces between them, representing the process of recycling and environmental sustainability. indicates that the boundary condition is missing; a space indicates that the presence or absence of the condition has no effect on the result.

Path H1 comprises three third-order equivalent configurations: H1a, H1b, and H1c. This path is classified as a policy social capital-associated type. Its core conditions include family members’ positions in the community or government agencies and household income levels, while the peripheral conditions encompass community public safety enforcement efforts, the comprehensiveness of medical and pension insurance coverage, the extent to which community lifestyles and cultural customs are integrated. Constrained by resource scarcity and underdeveloped markets in the Karst mountainous region, households achieve livelihood adaptation outcomes by embedding themselves within grassroots governance networks to secure policy information and administrative support. As Respondent F1 stated, “I am older, have little education, and possess no specialized skills. The community cadre arranged for me to work as a security guard, which provides a stable monthly income to help support my family.” Similarly, community cadre C1 remarked, “My family was a poor household relocated here, and I currently serve as the community secretary. For families facing difficulties and possessing limited labor, we proactively coordinate with relevant subdistrict departments to actively secure public welfare positions, helping them solve their most practical problems and ensuring they have a stable source of income.” For households reluctant to engage in farming, vocational skills training facilitates their transition to non-agricultural employment. This transition not only broadens their income streams but also fosters the reconstruction of social networks in their new environment, thereby enhancing livelihood stability. Ultimately, this configuration demonstrates that when households possess robust social networks and a baseline income, they can leverage policy information and resources to rebuild their livelihood adaptation outcomes.

Path H2 represents a balanced, comprehensive capital type, featuring the householder’s level of education, family members’ positions in the community or government agencies, household income level, elevation, and slope as core conditions. Meanwhile, the intensity of community public safety enforcement efforts, the comprehensiveness of medical and pension insurance coverage, and the extent to which community lifestyles and cultural customs are integrated serve as peripheral conditions. Farming households within this configuration possess a relatively comprehensive capital portfolio; they not only enjoy economic advantages and social connections but also transform elevation and slope from obstacles to survival into productive resources. Respondent F2 noted: “I studied agronomy in secondary vocational school. In the past, my parents always grew corn on the land in front of our house back home. The land was prone to waterlogging during the rainy season, constantly leading to poor harvests. Later, I switched to growing Sichuan pepper and sold it online, and our current income is much higher than before.” The presence of slope and elevation as core conditions does not signify a geographic disadvantage; rather, given a certain level of cognitive capacity, these factors can be successfully converted into resource advantages. Ultimately, this configuration demonstrates that a robust combination of human, social, economic, and perceived natural capital facilitates farming households’ attainment of high-level livelihood adaptation outcomes.

Path H3, representing the institutionally embedded type, features the householder’s level of education, household income level, and community public safety enforcement efforts as core conditions; family health status and slope as core absent conditions; and the comprehensiveness of medical and pension insurance coverage, alongside the extent to which community lifestyles and cultural customs are integrated, as peripheral conditions. Although certain relocated households in karst mountainous areas suffer from labor shortages due to disabilities or chronic illnesses, others leverage educational resources and online channels to access policy information, relying on community security and social safety nets to secure their livelihoods. As Respondent F3 stated, “I injured my back when working outside the province before and could no longer do heavy physical labor. My wife is also in poor health, so I can only work at a nearby supermarket. Coupled with the support of the minimum living allowance and medical insurance, we can still get by.” This configuration demonstrates that when households possess a solid economic foundation and education level, they can successfully rely on institutional safety nets and community support to maintain high livelihood adaptation outcomes, even if their health capital and resettlement environment are constrained.

Path H4, representing the vulnerability-breakout type, features family health status and family members’ positions in the community or government agencies as core present conditions; householder’s level of education, elevation, and slope as core absent conditions; and the community public safety enforcement efforts, the comprehensiveness of medical and pension insurance coverage, and the extent to which community lifestyles and cultural customs are integrated as peripheral conditions. This pathway contains the highest number of absent conditions, exhibiting the typical characteristics of households in deeply impoverished regions. Community cadre C2 noted: “Many households relocated to our community previously lived deep in steep mountainous areas, where transportation to the county seat was highly inconvenient, and most heads of households had only completed primary school. Since their relocation, the community has provided them with policy support, such as employment training, public welfare jobs, and annual dividends, which has significantly enhanced their overall living standards and well-being.” This configuration indicates that despite the farmers’ low education levels, unfavorable geographical conditions, and weak livelihood foundations, having family members in community or government positions secures crucial policy and resource advantages. Supported by good family health status as a foundational pillar, these households are able to proactively overcome vulnerable livelihood conditions and gradually enhance their adaptation outcome.

3.3.3 Robustness testing

In fsQCA model results can be sensitive to the specific configurations of conditions and cases; therefore, conducting a robustness check is essential. Referring to relevant research (Shi et al., 2025; Ding et al., 2026). This paper employs consistency levels and calibration anchors to conduct robustness tests. (1) When the consistency level was increased from 0.8 to 0.85, the results obtained with the adjusted consistency level configuration remained highly consistent with the original configuration results presented in Table 6. This finding indicates that the results of this study exhibit a high degree of robustness. (2) Robustness tests were performed by adjusting the calibration anchors from 95, 50, and 5% to 90, 50, and 10%. The adjusted results are displayed in Table 7, which demonstrates that configuration paths 2–4 remain consistent with the findings outlined in Table 6. Minor changes were observed in configuration path H1: the role of householder’s level of education in path H1a shifted from unclassified to marginal; Elevation in path H1b changed from unclassified to marginal. These minor adjustments suggest that adjusting the calibration point has a limited impact on the configuration paths H1a and H1b, thereby confirming the robustness of the findings of this study.

Table 7. The configuration path used to adjust calibration points.

Solid dark blue circle on a white background. Indicates that the core condition is present; Image is too blurry to identify specific content or details. No distinguishable objects or text are visible. indicates that the boundary condition is present; Red circle with a bold red X through the center, commonly used as a symbol to indicate something is not allowed, incorrect, or not selected. indicates that the core condition is missing; Green recycling symbol featuring three arrows forming a circle with four white spaces between them, representing the process of recycling and environmental sustainability. indicates that the boundary condition is missing; a space indicates that the presence or absence of the condition has no effect on the result.

4 Discussion

This paper utilizes case studies from Bijie, Liupanshui, and Anshun in Guizhou Province to integrate adaptive capacity theory within a sustainability framework. It systematically evaluates the levels and structural characteristics of livelihood adaptation outcomes across five dimensions, particularly focusing on capital stocks and psychological attribution. The findings reveal not only the varying levels, associated factors, and configurational structures of livelihood adaptation outcomes but also provide new insights into the sustainable livelihood challenges encountered by relocated households in karst regions.

4.1 Major findings and interpretation

The CRITIC measurement results show that the outcomes of livelihood adaptation in ecologically fragile karst areas following resettlement exhibit significant asynchrony. Although overall adaptation outcomes are at a moderately high level, a breakdown by dimension reveals a clear spatiotemporal mismatch: psychological and living adaptation have reached relatively high levels, while economic and occupational adaptation continue to lag significantly behind. This pattern contrasts sharply with the conventional sequence of relocation adaptation, in which economic recovery generally precedes psychological integration, and also differs from previous findings that “farmers’ livelihood adaptation outcomes are a process of co-evolution between multidimensional capital and capabilities” (Wu and Duan, 2022; Zhao et al., 2025). We define “asynchronous adaptation” as the lack of synchrony in the recovery trajectories across different functional dimensions of livelihood following a disruptive relocation event (Bourai et al., 2026). In traditional migration theory, economic recovery is typically viewed as a prerequisite for psychological integration and life satisfaction. However, in the context of karst regions, our findings challenge this sequence. This reversal can be attributed to government-led resettlement models, which prioritize immediate improvements in physical infrastructure, housing quality, and access to public services. Improvements at this level can rapidly alleviate material anxieties, foster a sense of security, and consequently enhance psychosocial well-being at an early stage. Conversely, economic-employment recovery remains sluggish because it is contingent upon slow-changing structural factors, such as regional labor market maturity, the transferability of farmers’ human capital, and market accessibility. Unlike the top-down efficiency of infrastructure provision, economic reintegration is a market-driven and path-dependent process that necessitates prolonged capacity-building.

The OPGD results identify household income level, completeness of medical and pension insurance coverage, and household health status as the factors with the strongest explanatory power for variation in adaptation outcomes. This pattern is consistent with the SLF, which emphasizes the foundational roles of financial, social, and human capital (Nasrnia and Ashktorab, 2021; Wu et al., 2023a). Our findings indicate that post-relocation livelihood adaptation outcomes result from a complex interaction between structural and individual-level factors, rather than being shaped by any single determinant. This is consistent with the findings of Kanak Pervez et al. (2026), who found that land-use structure and ecological factors are better predictors of agricultural output than climate variability alone. In particular, the significant explanatory power of insurance coverage and health status suggests that social protection and household human capital may partially compensate for economic limitations in the short term. Interaction detection further shows that pairwise combinations of these factors generally yield enhanced explanatory power, indicating that no single factor alone is sufficient to explain high adaptation outcomes. For instance, the interaction between income level and insurance coverage often exceeds the explanatory power of either factor alone. This implies that when economic resources are constrained, the presence of comprehensive social security and good health can buffer some of the negative associations with low income. This is consistent with the fsQCA configurations, where some pathways achieve high adaptation outcomes without high income as a core condition; instead, social security and community governance appear as core or contributing conditions. Therefore, our results suggest that social protection and community support may function as compensatory mechanisms in the short term by providing basic security, reducing vulnerability to shocks, and maintaining social stability, thereby supporting overall adaptation outcomes even when economic recovery is incomplete.

However, the compensatory role of social protection and community support should not be interpreted as guaranteeing long-term livelihood independence. The fsQCA results show that configurations lacking strong economic conditions tend to have lower raw coverage, suggesting that such configurations are less prevalent among households with high adaptation outcomes. In other words, while non-economic resources can buffer economic deficiencies in specific contexts, they do not replace the need for sustainable income generation and employment stability. Social protection and community support are external buffers; they provide a safety net but do not by themselves build productive capacity. Long-term livelihood independence requires the accumulation of financial capital, stable employment, and diversified income sources. Our cross-sectional data capture adaptation outcomes at a single point in time and cannot demonstrate whether the observed compensation persists over time. Longitudinal research is needed to examine whether households relying primarily on social support can transition to self-sustaining livelihoods.

4.2 Policy implications

Enhancing the livelihood adaptation outcomes of migrated farmers is a crucial step in transitioning resettlement areas from a focus on “maintaining stability” to one of “achieving prosperity,” while simultaneously providing essential support for their long-term sustainable development (Wang et al., 2022). Policy interventions should aim to bridge the gap between the relatively rapid psychological and living adaptation processes and the more gradual economic and occupational recovery processes. The results from OPGD and fsQCA indicate that high livelihood adaptation outcomes are linked to the coordinated interaction of multiple factors, rather than the adjustment of any single factor. Consequently, policy interventions should adopt a systemic perspective that integrates economic, social, and institutional support in a context-specific manner, rather than relying on isolated measures. Based on these research findings and Figure 7, we propose corresponding policy implications at both the government and household and individual levels.

• (1) Narrowing the economic adaptation gap through income diversification and social protection. The OPGD results indicate that household income level, the completeness of medical and pension insurance coverage, and household health status are the factors with the strongest explanatory power for variations in adaptation outcomes. This suggests that policies should prioritize the enhancement of these foundational conditions. At the governmental level, support should be tailored to the resource endowments of different karst regions; however, the primary focus should be on enabling households to diversify their income sources and reduce reliance on a single sector. Instead of prescribing specific industries or crops, policies should foster an enabling environment for diversified livelihood activities, including off-farm employment and local economic development, while ensuring respect for ecological constraints. Concurrently, expanding the coverage and accessibility of medical and pension insurance can mitigate livelihood vulnerability and buffer the negative impacts of economic limitations, as indicated by the compensatory patterns observed in the fsQCA configurations. At the household level, farmers may benefit from actively pursuing diversified income opportunities, engaging in relevant skill development programs, and strategically allocating household assets to enhance their economic adaptation outcomes.
• (2) Promoting adaptive governance through multi-factor coordination. The results of the fsQCA analysis indicate that achieving high levels of livelihood adaptation does not require meeting any single condition; rather, there are multiple equivalent final configurations. This implies that policy packages should be flexible and tailored to household characteristics and local conditions. For households achieving high adaptation outcomes through high income combined with favorable non-economic conditions, the priority should be to maintain economic vitality while further strengthening community governance to sustain social integration, preventing regression in the non-economic dimensions. For households relying on social capital to compensate for limited income, policies should focus on preserving and mobilizing existing social networks—such as through community organizations and mutual-aid groups, while gradually building economic capacity through targeted skill development and off-farm employment support. For households depending on policy and social security to compensate for economic constraints, the immediate focus should be on robust social protection and public service delivery, but this must be explicitly linked to a longer-term plan for income generation, otherwise short-term compensation may become permanent dependence. For households with balanced multi-factor configurations, the emphasis should be on monitoring and reinforcing existing strengths to prevent deterioration. At the governmental level, employment support should be tailored to specific pathway types rather than delivered as a uniform set of training programs. For households with limited labor capacity, policy-based safety nets must be effectively implemented. At the household level, farmers are encouraged to pursue lifelong learning and continuously adapt their skills to evolving market demands. Digital tools and platforms can facilitate market access; however, their effectiveness is contingent upon local infrastructure and digital literacy, which require complementary public investment.
• (3) Fostering social integration and community support as compensatory mechanisms. The discussion above highlights that psychosocial integration can occur faster than economic recovery and that social protection and community support can partially compensate for economic limitations in the short term. However, this compensation does not guarantee long-term livelihood independence. Therefore, policy should not only address economic capacity but also strengthen the social fabric that supports adaptation. At the governmental level, this can be achieved by supporting community governance structures and neighborhood mutual-aid mechanisms. For example, in communities where social trust and neighborhood interaction are already strong, local governments can use these networks to organize job information sharing, skill demonstration activities, and mutual support for job-seeking, thereby converting social capital into economic opportunities. For communities with weaker social integration, priority should be given to building basic community governance structures first, before expecting social networks to play an economic role. Such initiatives can help cultivate social trust, a sense of belonging, and informal safety nets, which are particularly important in the early stages of resettlement. At the household level, farmers may benefit from proactively expanding their social networks, participating in community organizations, and engaging in public affairs. Households should be encouraged to use their social connections not only for emotional support but also for concrete economic purposes, such as accessing market information or employment referrals.

4.3 Limitations and future research

The study also has several limitations: (1) the cross-sectional design prevents causal inference. Although OPGD and fsQCA identified factors with strong explanatory power and configurations associated with high livelihood adaptation outcomes, these results represent statistical associations and set-theoretic patterns rather than causal mechanisms. Future research using longitudinal or panel designs is needed to examine causal pathways. (2) the outcome index includes per capita income as one of its dimensions, while one explanatory variable (X₄) measures household income satisfaction. Although objective income and income satisfaction are conceptually distinct, they are empirically correlated. Therefore, the observed association between X₄ and the outcome index may partly reflect construct overlap rather than an independent predictive relationship. Accordingly, findings should be interpreted as diagnostic associations rather than causal effects. (3) All data were collected via self reported questionnaires, which may have introduced both common method bias and social desirability bias. Because both outcome and explanatory variables were obtained from the same respondents at a single time point, the magnitude of some associations may be inflated. Furthermore, respondents may have tended to give socially desirable answers, particularly on subjective constructs such as happiness, community belonging, and confidence in the future. To mitigate these limitations, future studies could incorporate objective indicators or multi source data to enhance the validity of the findings. (4) The study area is limited to a specific karst mountainous region in Guizhou Province. The findings may not be generalizable to other resettlement areas with different ecological, economic, or institutional conditions. Future research should include cross-regional and cross-context comparisons to test the robustness and applicability of the findings.

5 Conclusion

This study evaluates the livelihood adaptation outcomes of relocated farmers in karst mountainous areas and examines the factors associated with these outcomes. Based on the empirical results and discussion, the following conclusions are drawn.

• (1) The overall level of livelihood adaptation outcomes among the surveyed farmers is above average; however, notable disparities exist across various dimensions. Psychological and living adaptation outcomes are relatively high, whereas economic and occupational adaptation outcomes remain comparatively low. This pattern indicates that limited economic capital accumulation and inadequate employment support are significant factors contributing to lower adaptation outcomes.
• (2) Farmers classified into high, medium, and low adaptation levels exhibit significant differences across all dimensions, with occupational adaptation outcomes displaying the largest disparity. This finding indicates that vocational skills and access to employment information are closely associated with overall livelihood adaptation outcomes.
• (3) Household characteristics, economic conditions, geographic environment, community governance, social security, and cultural integration are associated with livelihood adaptation outcomes to varying degrees. Among these, household income, coverage of medical and old-age insurance, and family health status demonstrate the strongest explanatory power. Pairwise interactions among these factors generally exhibit either dual-factor or non-linear enhancement, suggesting that their combined effects are more substantial than those of individual factors.
• (4) The factors associated with farmers’ livelihood adaptation outcomes are complex and characterized by equifinality. No single optimal configuration or individual factor is universally linked to high adaptation outcomes. Instead, different factors play distinct roles across multiple configurational pathways. High adaptation outcomes are associated with specific combinations of conditions formed through complex, non-linear interactions, and these configurations represent different pathways leading to the same outcome.

Generative AI statement

The author(s) declared that Generative AI was not used in the creation of this manuscript.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher’s note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Supplementary material

The Supplementary material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fsufs.2026.1922524/full#supplementary-material

Supplementary material Supplementary material at the source

Footnotes

Edited by: Jie Xu, Xi'an University of Finance and Economics., China

Reviewed by: Yang Chen, Southwest Jiaotong University, China

Sri Hery Susilowati, National Research and Innovation Agency (BRIN), Indonesia