A livelihood comprises living capacities, assets, and necessary production activities (Chambers and Conway, 1992). Resilience refers to the ability to absorb, deal with and adapt to environmental change, highlighting a system’s response to extreme interference and continual pressure (Engle et al., 2013; Folke, 2016). Speranza et al. (2014) defined livelihood resilience as the ability of one’s livelihood to maintain or improve essential attributes and functions while protecting against stress and disturbance. It comprises individual and household strategies under stress and shock (Quandt, 2018; Liu et al., 2020a). Tanner et al. (2015) highlighted the role of human agency and empowerment and found that people’s livelihood opportunities and well-being are crucial to developing livelihood resilience. Thus, we define livelihood resilience as subjects utilizing their livelihood capital, exerting their initiative, and adopting adaptive livelihood strategies to diversify livelihoods and well-being in the face of external changes and interference.

Most recent studies are primarily based on the “livelihood resilience framework” of Speranza et al. (2014), which is used to explore the influence of buffer capacity, self-organizational capacity, and learning capacity on livelihood resilience. Nyamwanza (2012) considered resilience and the adaptive capacity of livelihoods. Bu et al. (2023) developed a framework of livelihood resilience in the context of ecotourism development and change comprising the three dimensions of changes in the tourism development environment, adaptive capacity, and transformation capacity. However, these studies emphasized the impacts of multiple livelihood capacities, livelihood capital, and tourism development environments on livelihood resilience, yet did not focused on the relationships among these factors. Moreover, the literature ignores the transmission pathways between the antecedents and livelihood resilience. Therefore, it does not fully reflect the requirements of current livelihood research and must be further explored.

The research was conducted in Yellow River Source Park in Sanjiangyuan National Park in Maduo County, Qinghai Province, China. The park covers 1.82×10⁴ km² and was one of the first national parks established in China. The Yellow River Source Park community has a population of 9765, of whom Tibetans account for more than 90%. It is an area of traditional animal husbandry production and has a highly fragile ecological environment. The fragile ecological environment and the construction of the national park system have constrained the regional pastoral output and the development of herders’ livelihoods. In recent years, the national park has incorporated the ecotourism industry into the park’s concession project, seeking to coordinate the relationship between ecological protection and community development. Ecotourism development and changes in the Yellow River Source Park have affected the local population’s traditional livelihood strategy of grazing and increased the vulnerability of residents’ livelihoods. Therefore, studying the livelihood of residents in this area and exploring the factors influencing their livelihood resilience and the path of livelihood construction are important to develop livelihoods in the region.

A bilingual Tibetan-Chinese survey questionnaire was used to collect the data. The respondents’ native tongue, Tibetan, was used in the questionnaire. To guarantee the equivalence of the items in Tibetan from Chinese, in producing the Tibetan-Chinese translation of the questionnaire, one Tibetan professor, five postgraduates, and six Tibetan undergraduates translated and proofread the scale items following the translation-back translation procedure. The research sample mainly comprised residents who participated in tourism development projects in the park or who engaged in tourism-related work. They also included government workers, education workers, and other workers in the research area. We selected residents of Zhalinghu township, Huanghe township, Machali town, and Huashixia town for the investigation. From May 6 to 10, 2021, 73 questionnaires were issued and recovered; 66 were valid, for an effective response rate of 90.4%.

A 5-point Likert-type scale was used to rate each item. The first section of the questionnaire collected basic information from respondents, including their place of residence, gender, age, ethnic group, education, income level, primary source of income, and means of livelihood. The second part is the measurement of variables related to livelihood resilience. All the measurement items adapted from previous studies, as shown in Table 1. Tourism development perception (TDP) was measured by three items related to living standards, economic benefits, and development chances (Gong and Yang, 2017). The items for measuring adaptive capacity (AC) dimensions (i.e., livelihood capital, dependence on natural resources, and young labour) were adapted from Thulstrup (2015) and Arhin et al. (2022). The items for measuring transformation capacity (TC) dimensions (i.e., education input, government support, social relations, transformation opportunities and traffic accessibility) were adapted from Zheng et al. (2020) and Wu et al. (2024). Five items measured the two dimensions of adaptive and transformation capacity. The items for measuring livelihood resilience (LR) dimensions (i.e., learning opportunity and capacity, self-organization capacity, and living environment) were adapted from Arhin et al. (2022) and Sina et al. (2019).

Two methods were employed to test our hypotheses: PLS-SEM using Smart PLS 4.0.8 and fs/QCA using fs/QCA v. 4.0. PLS-SEM is used to test the hypothetical relationship between causal structures and the net impact on one or more results. This method is suitable for forecasting applications and theory construction and increasingly become a popular technical tool in empirical research (Ali et al., 2018). However, it can only capture the linear relationship between structures, not unassumed reverse cases (Lee et al., 2022). According to Ragin et al. (2008a), the theoretical method of fs/QCA is used to examine the connections between antecedent conditional factors and outcome variables. This analytical approach, which has roots in set theory and Boolean logic, was created specifically to investigate social phenomena using small samples (Fainshmidt et al., 2020; Ruiz-Palomino et al., 2021). Its use may help with research issues regarding livelihoods. Therefore, to overcome the shortcomings of PLS-SEM and deepen the understanding of the complex causal situation of micro subjects’ livelihood resilience, this study uses asymmetric fs/QCA to supplement the research and combines several equivalent configuration schemes.

Fs/QCA consists of the following steps (Pappas and Woodside, 2021). 1) Calibration: This process involved calculating the average value of each scale item and converting the antecedents into fuzzy sets using fuzzy scores of 95%, 50%, and 5% (Ragin, 2008a; Fiss, 2011). 2) Necessary conditions analysis: This is used to check whether each antecedent condition is necessary to produce the result, which is a premise of fs/QCA. When analysing conditions, a condition is considered necessary if its consistency registers a significance above 0.9. A condition is close to being necessary if that value registers above 0.8 (Ragin, 2008b). 3) Truth table analysis: This involves the listing all conceivable combinations and ranking the cases using a minimum frequency of 1, a consistency threshold of 0.8, and a PRI consistency of 0.5 (Ragin, 2008b; Greckhamer et al., 2018).

When conducting exploratory research, N=50 is the most reasonable minimum value (De Winter et al., 2009). The social traits of the samples are displayed in Table 2. The most considerable sample size for residences was 59.1% in Machali Town. A total of 33.3% of people were female, while 66.7% were male. The primarily were aged between 26 and 35 (42.4%). The majority of the respondents were Tibetans (69.7%). More than 37% (37.9%) had a bachelor's degree/college degree or above, and 6.10% had no education. 33.3% of the respondents had a monthly income between 2001 and 3500 yuan. Self-employment and industry are the primary sources of income for most residents, with 15.2% of the income coming from raising livestock and 6.1% being tourism related. Additionally, most residents (90.9%) had only one source of income, while a small minority (0.9%) had two or more sources.

Table 6 presents the analytical findings from fs/QCA for the prerequisites for livelihood resilience. According to Ragin (2008a), a condition is necessary when its consistency score exceeds 0.9. The results indicate that tourism development perception, adaptive capacity, and transformation capacity do not constitute necessary conditions for livelihood resilience. After the necessity analysis, the fs/QCA results provided sufficient configurations that led to both high livelihood resilience (HLR) and low livelihood resilience (LLR). For HLR, two causal combinations (HLR1 and HLR2) emerged (consistency=0.866734, coverage=0.786602). HLR1 (Transformation capacity* ~Tourism development perception) ^①

① The asterisk (*) indicates “and” in Boolean or fuzzy-set logic.

indicates that under negative perceptions of tourism development, HLR can be promoted by residents with high transformation capacity, regardless of their level of livelihood adaptability. HLR2 (Adaptive capacity* Transformation capacity) indicates that, as long as residents have high adaptive capacity and transformation capacity, HLR can be formed under the combination of both, regardless of whether there is a positive impact of tourism development.

Table 7 shows that three causal combinations (LLR1, LLR2 and LLR3), result in pathways leading to LLR (consistency=0.820926, coverage=0.848531). LLR1(~Adaptive capacity* ~Transformation capacity) shows that residents' livelihood adaptive capacity and transformation capacity are low. Even if the perception of tourism development is positive, the generation of HLR will be inhibited. LLR2 (~Transformation capacity* Tourism development perception) and LLR3(~Adaptive capacity* Tourism development perception) indicate that even if residents’ perception of tourism development is positive, as long as residents’ livelihood adaptive capacity and transformation capacity are poor, their livelihood elasticity will remain at a low level.

The results supported all earlier hypotheses. As shown for HLR1 and HLR2, multiple configurations resulted in HLR, confirming the study’s first tenet (P1). The second principle of causal asymmetry (LLR) was also supported (P2), as configurations of the same conditions led to distinct livelihood resilience outcomes (i.e., HLR2 and LLR1) depending on the presence or absence of adaptation and transformation capability (HLR2). Finally, transformation capability was a core condition (•) in HLR1 and HLR2, representing HLR. Tourism development perception was essential for LLR2 and LLR3, representing LLR.

This study makes three contributions to the research on the livelihoods of national park residents. First, we expanded the hypothetical analysis scenarios of the livelihood resilience of national park residents. Previous studies have focused on agriculture-based livelihoods (Quandt, 2018; Quandt et al., 2018; Liu et al., 2022). However, the unique livelihood context of herding in Sanjiangyuan National Park has not been sufficiently investigated (Li et al., 2022; Bu et al., 2023). National park ecology conservation, ecological tourism development (Shekari et al., 2022), and the maintenance of residents’ traditional livelihoods (such as grazing) provide unique contexts for understanding livelihood resilience that may differ from those in other types of regions. Under climate change, resettlement, and other conditions, the impacts of buffering, self-organization, learning capacity, adaptability, and transformation capability on livelihood resilience have been verified (Thulstrup, 2015; Liu et al., 2020b; Amadu et al., 2021; Fan et al., 2022; Wu et al., 2024). This study adds to the dimension of tourism perception in livelihood resilience studies. Using tourism development perception as a critical factor, we highlight the mediating role of adaptive and transformation capacity in influencing the residents’ livelihood resilience of national parks.

Second, this study focuses on the influence paths that lead to livelihood resilience and reveals the internal dynamics between the different elements of livelihood resilience. Specifically, we confirm the influence of tourism development perception on livelihood proposed by Zhang et al. (2021) and demonstrate the relationships among tourism perception, adaptive capacity, and transformation capacity, which thus far have not been evaluated. We found that adaptive capacity does not significantly influence livelihood elasticity under the influence of tourism development perception. Our understanding of the underlying assumptions of livelihood resilience is furthered by rejecting H7a and supporting H7b. The core element of adaptation ability is livelihood capital (Bu et al., 2023). When a livelihood subject’s available livelihood capital is insufficient, these internal factors will directly affect his or her choice of livelihood strategy regardless of whether the perception of tourism is negative or positive. Zheng et al. (2020) emphasized the impact of the combination of buffering, adaptation, and transformation capacity on farmers' livelihood resilience. However, our study revealed that adaptive capacity positively influences transformation capacity and livelihood resilience but does not positively affect livelihood resilience through transformation capacity under tourism development perception. As previously mentioned, consistent with our findings, Zheng et al. (2020) and Bu et al. (2023) indicated that combining adaptive and transformation capacity might enhance livelihood resilience. Moreover, adaptive capacity and positive tourism development perception strengthened the positive association with transformation capacity, the critical mediating variable of livelihood resilience evaluations in national parks.

Third, we combined PLS-SEM and fs/QCA for data analysis, consistent with the current suggestions to supplement quantitative approaches with qualitative analytical techniques for better comprehension of the issues under study (Ruiz-Palomino et al., 2021). Unlike recent studies, most articles use regression equation models to study livelihood resilience (Quandt et al., 2019; Liu et al., 2020b). These methods can be used to effectively analyse the linear effect of variables on outcomes. However, they do not explore the causal asymmetry of antecedent conditions or the nonlinear effects of variables on outcomes. Therefore, we apply PLS-SEM and fs/QCA simultaneously to comprehend the linear and nonlinear relationships among the dimensions, analyse the paths affecting livelihood resilience through the lens of causal asymmetry and effectively identify the distinctiveness of antecedent configurations that lead to high and low levels of livelihood resilience. To date, few studies have attempted to evaluate livelihood resilience. However, this study contributes to livelihood research by combining both methods to identify livelihood resilience.

From the management viewpoint, the findings show that residents’ livelihood resilience in national parks must focus on how residents perceive its tourism development. Understanding each aspect of residents’ needs for tourism development enables national park to design tourism participation programs to enhance residents’ perceptions holistically by providing more job opportunities and increasing residents' income, reinforcing their willingness to participate. For the development of national parks, tourism development represents the essence of what distinguishes this kind of tourism development from others. This differentiation is essential for promoting the comprehension of residents’ perceptions of and affective connections with such a tourism development model. Suppose that residents’ perceptions of tourism development remain positive. In that case, residents will be willing to participate, which combines high adaptive and transformation capacities with their traditional livelihood strategies to promote livelihood resilience. From a sustainable livelihood perspective, people with a high level of participation in tourism use diverse livelihood strategies that are essential for resilience.

Livelihood resilience strategies should differentiate between residents who participate in tourism development and those who do not, given that these two types of residents evaluate tourism development, adaptive capacity, transformation capacity, and livelihood resilience differently. Increasing willingness to participate in tourism is an intelligent strategy for residents who do not participate by emphasizing tourism’ economic effects and learning opportunities and its positive influence on the living environment and living standards. For residents who participate in tourism development, it is equally crucial to reinforce tourism livelihoods to cope with disturbances, adapt to changes, and enhance diverse livelihood abilities.

Finally, residents’ livelihood strategy options are limited by their livelihood capital and ability. With respect to increasing perceptions of tourism development, we further suggest that governments should be based on cooperatives and focus on innovating community resource utilization and optimizing resource utilization to reduce residents’ high dependence on natural resources. A starting point to achieve this would be to improve the environmental management mechanism of national parks, the treatment and welfare of “ecological conservators” and the livelihood capital stock of local residents and their capacity to adapt to external disturbances. In addition, we should further improve the market competition mechanism of parks. Introducing multiple franchise enterprises could increase residents’ job chances and expand the channels for transforming their livelihoods. Realizing the competition mechanism of regional enterprise management helps improve operators’ awareness of the crisis.

This research has several limitations. First, the sample size is mainly concentrated in Machali town. As the secondary distribution centre of tourism development in Maduo County, the residents of Machali town are less engaged in animal husbandry but more engaged in self-employment. The sample size significantly impacts the livelihood resilience of residents in a region. Moreover, residents engaged in animal husbandry accounted for 18.2% of the sample, which is a relatively high average score compared with residents engaged in other occupations. In contrast, the average scores for tourism development perception and transformation capacity are relatively low. The livelihood of residents engaged in animal husbandry was relatively stable in the early stage of the development of the national park, but had been affected by the time of the sample collection. The livelihood vulnerability of residents engaged in tourism and migrant work is relatively high. Therefore, future studies should increase the number of samples, select other regions to demonstrate the results, and compare the level and factors influencing the livelihood resilience of different practitioners (Wang et al., 2023).

Second, this study considers only the three aspects of livelihood resilience of tourism development perception, adaptive capacity, and transformation capacity. Future research might consider further aspects, including multisource disturbance, ecological effects, and policy change. In addition, the perception of tourism development is a highly subjective evaluation, and residents of different livelihood strategies and living stands may have different perceptions and behaviours when the same object is evaluated (Zhang et al., 2021). Therefore, the relationship between tourism development perception and livelihood resilience should be assessed and verified in different types of national parks, as well as in different stages of tourism development in Sanjiangyuan National Park.