The Impact of Geographical Location of Households’ Residences on the Livelihoods of Households Surrounding Protected Areas: An Empirical Analysis of Seven Nature Reserves Across Three Provinces in China

Abstract

China has effectively safeguarded biodiversity by building the world’s largest system of nature reserves, but the livelihoods of farmers near the reserves are often not guaranteed. This paper aimed to deeply explore the intrinsic relationship between the geographical location of households and their livelihood outcomes within seven nature reserves across three provinces in China. Innovatively, this study subdivided households’ livelihood outcomes into four patterns: high well-being with high dependency (H-H), high well-being with low dependency (H-L), low well-being with high dependency (L-H), and low well-being with low dependency (L-L), in order to comprehensively analyze the diversity of households’ livelihoods and further reveal the spatial logic and mechanisms underlying regional development imbalances. Methodologically, a combination of quantitative analysis and qualitative research was adopted. Representative villages in the protected area and outside the protected area were selected for semi-structured interviews with the village heads. Meanwhile, farmers were randomly selected in the villages for structured interviews and 1106 questionnaires were collected. Through variance analysis, the study first identified the unique advantages of H-H-pattern households in natural resource utilization. Subsequently, a multinomial logistic model was used to deeply analyze how geographical location (including whether a household was located within a protected area and the distance to markets) affected the transition mechanisms of the other three livelihood outcomes towards the H-H pattern. Based on this, marginal effect analysis was employed to further delineate the specific influence pathways of geographical factor changes on households’ livelihood outcome selection probabilities. The results showed that the geographical location of households’ residences had a significant impact on their livelihood outcomes. For households in the L-L and H-L patterns, proximity to markets could significantly increase the probability of their livelihood transitioning to the H-H pattern. Meanwhile, residing within protected areas significantly promoted the transition of L-L and H-L households to the H-H pattern but showed a certain inhibitory effect on L-H households. Marginal effects analysis further shows that both living in protected areas and reducing distance to markets increase the tendency of households to be highly dependent on natural resources for livelihood outcomes. Compensation policies should be designed according to local conditions, and subsidies for the development of ecotourism and other service industries should be increased for rural households in protected areas to ensure sustainable development rather than transfer payments.

1. Introduction

In the context of the development and protection of natural resources, China, as a country with abundant natural resources and a large agricultural population, is particularly concerned about the sustainable development of its rural areas [1]. As the state attaches more and more importance to the construction of ecological civilization, the establishment of nature reserves has become an important measure to maintain biodiversity and ensure ecological security [2,3]. However, the designation of these areas is often closely intertwined with the livelihood space of farm households, raising the issue of how to promote the sustainable development of farm households’ livelihoods while protecting the ecological environment [4,5,6]. Nature reserves in China are widely distributed in various ecologically sensitive areas, such as forests, wetlands, grasslands, and nature reserves, which are not only important components of natural ecosystems but also the basis for the survival and development of many farm households. However, the long-term resource-dependent livelihood model has caused some farm households to face the challenges and dilemmas of livelihood transformation after the implementation of conservation policies [7,8]. Differences in geographic location, such as proximity to the core area of the nature reserve, transportation access, and unequal distribution of natural resources, directly affect farm households’ access to natural resources and their capacity to develop alternative livelihoods, which subsequently exerts a significant influence on household livelihood outcomes [9]. Therefore, it is important to explore the relationship between the geographic location of farm households and their livelihood outcomes in nature reserves in order to balance ecological conservation and economic development and to promote the transformation of farm households’ livelihoods.

To examine the complex relationship between the geographical location of rural households within protected areas and their livelihood outcomes, a synthesis of multi-dimensional empirical research is necessary to establish a comprehensive understanding of current conditions. This encompasses the direct impact of geographical location on access to livelihood resources, the moderating role of government policies, the contribution of economic indicators to livelihood outcomes, the correlation between geographical feature-based well-being indicators and farmer households’ well-being, as well as farmers’ perceptions of and adaptation strategies to climate change [10,11,12,13]. Geographical location significantly influences farmers’ access to livelihood resources. While nature conservation areas are established to preserve biodiversity, they impact the livelihoods of surrounding farmer households [14]. Farmers near the core zones face resource scarcity due to stringent protection policies (such as resource extraction and land use restrictions) [15,16], which directly limits their economic sources like agriculture and animal husbandry, and indirectly affects access to market information and technical support [17]. In contrast, farmers in peripheral zones can leverage their geographical advantages to diversify livelihood activities (such as the underforest economy) to broaden income sources and enhance livelihood resilience, particularly evident in the ecological functional areas upstream of the Yellow River [18]. Government policies play a pivotal role in moderating the relationship between geographical location and livelihoods. Through measures like ecological compensation mechanisms and relocation resettlement, the government not only protects the ecology but also promotes the sustainable livelihood development of farmer households [19]. Ecological compensation provides economic compensation and livelihood transformation funds for impacted farmers [20,21], while relocation resettlement addresses livelihood challenges for farmers in core zones and promotes coordinated regional development [22,23]. Economic indicators are crucial in measuring the sustainability of farmers’ livelihoods. Financial resources and expenditures directly reflect farmers’ economic conditions, influencing livelihood stability and growth potential [24,25]. Farmers with good economic conditions are more resilient to disasters and market risks, maintaining livelihood growth [26,27]. Geographical location also indirectly affects economic conditions, with transportation convenience and proximity to markets reducing transaction costs and broadening market horizons and sales channels [28,29]. Geographic feature-based well-being indicators, such as distance to the county seat, and distance to main roads, significantly impact farmers’ livelihoods [30]. These indicators determine farmers’ ability to access public resources, influencing farmland yields and income. Villages with steep slopes and water scarcity have weak livelihood foundations [31], while villages with advantageous locations enjoy convenient services and market opportunities, enhancing farmers’ well-being [32,33]. In the context of global climate change, farmers in nature conservation areas face more severe challenges and need to address ecological changes and livelihood risks. In summary, a complex relationship exists between the geographic location of farm households in China’s nature reserves and their livelihood outcomes. Geographic location not only directly determines the accessibility of livelihood resources for farm households but also indirectly influences these outcomes through various pathways, including government policies, economic indicators, and geographic well-being indicators. Consequently, in designing nature conservation policies and facilitating sustainable livelihood development for rural households, it is imperative to account for the pivotal role of geographic location and its dynamic interplay with socio-economic and environmental factors.

Despite the comprehensive exploration of the relationship between household geographic location and livelihood outcomes surrounding protected areas in China, existing literature presents several limitations. Firstly, although the direct impact of geographical location on livelihood resources was discussed, multidimensional data such as government policy adjustment and livelihood capital have not been effectively controlled [29]. These factors will have an important influence on the results. Secondly, the research mostly focuses on specific regions, resulting in insufficient understanding of the universal laws across geographical and socio-economic environments [18]. Finally, the research is not in-depth enough. Geographical location is often one-dimensional and considered as a control variable, and it has not been studied in detail [33]. Therefore, it is necessary to widely collect data from different regions and at different times, integrate a more comprehensive dataset, and fully grasp the impact of family geographical locations on the livelihoods around protected areas. The objective is to uncover the mechanisms by which geographic factors influence livelihood diversification, income levels, and the quality of life among farm households. By gathering extensive primary data and utilizing a mixed-methods research approach, this study aims to furnish scientific evidence to policymakers, enabling them to craft more targeted and effective support policies.

Assessing the role of geographic factors in promoting livelihood upgrading: This research will further focus on identifying differences between harmonious and non-harmonious coexistence areas and highlight key factors sustaining harmonious livelihoods.

Proposing the path for improving people’s livelihood based on the optimization of geographical elements: The ultimate goal of this study is to put forward a series of practical and feasible suggestions for improving people’s livelihoods.

2. Research Framework

As a comprehensive analytical tool, the Sustainable Livelihoods Framework (SLF) is centered on systematically examining and addressing the diverse livelihood challenges faced by farm households [34,35]. The framework not only focuses on how farm households can skillfully deploy and utilize various livelihood resources in the midst of complex social, economic, and environmental changes but also aims to optimize the allocation of such resources in order to build a dual defense of sustainability and stability of the livelihood system [36,37]. The significance of this is to empower farm households to break away from overdependence on external assistance, enhance endogenous development dynamics, and improve the resilience of their livelihoods to withstand unknown risks and challenges.

By integrating geographical coordinates as a structural determinant, the analysis reveals that disparities in natural capital and market accessibility systematically govern livelihood trajectories and economic returns within agrarian communities [16,32]. At the same time, differences in geographic location can also act first on the accumulation and allocation of livelihood capital, which in turn indirectly affects the choice and ultimate effectiveness of livelihood strategies [18]. Based on their own resource endowments and subjective preferences, farmers can adopt or not adopt alternative livelihood strategies, thus forming four types of livelihood outcomes: high well-being and high dependence on natural resources (H-H), high well-being and low dependence (H-L), low well-being and high dependence (L-H), and low well-being and low dependence (L-L). The levels of well-being and the degrees of dependence on natural resources are not absolute values. Instead, they are relative concepts, referring to whether an individual is in the top 50% or the bottom 50% of the surveyed population.

The dynamics of livelihood outcomes, like a feedback mechanism in a closed-loop system, not only influence the accumulation of livelihood capital in the next cycle of farm households’ livelihoods but also affect every part of the reproduction cycle [38]. Policies significantly affect the accumulation and distribution of livelihood capital, particularly through ecological compensation and immigration policies. These policies encourage households to diversify their resource use and enhance livelihood resilience. Ecological compensation enables households to adopt more sustainable practices, optimizing their resource allocation and reducing environmental impacts, thereby increasing well-being while safeguarding ecosystems. By recognizing geographical variations and their impact on livelihood strategies, policies can empower farmers to transition from L-H models to H-H models, ensuring a win-win scenario for both households’ well-being and ecological conservation.

Particularly noteworthy is that this result also mirrors the subtle coupled symbiotic relationship between households’ well-being and natural ecosystems. Especially in the macro context of protected area conservation policies, H-H households demonstrate a model of harmonious coexistence between humans and nature with their efficient and environmentally friendly utilization of natural resources, which not only enhances their own well-being but also mitigates the negative impacts on the ecosystem [39]. In contrast, L-H households rely on traditional, more extensive patterns of agricultural production, resulting in relatively low levels of well-being and limited potential for sustainable development. This phenomenon underscores the critical need to incorporate spatially explicit factors, such as geographic location, and their substantial implications for shaping livelihood outcomes and advancing sustainable livelihood transitions among farm households, so as to develop more precise and effective policies and measures that promote win-win development for the well-being of both farm households and ecosystems.

3. Data Sources

The research area includes Dorbod Mongol Autonomous County in Heilongjiang Province in northeast China; Zhouzhi, Meixian, and Yangxian counties in Shaanxi Province in northwest China; and Gongshan and Mengla counties in Yunnan Province in southwest China. These nature reserves are representative of different regions of China. They are home to various rare species, and the government’s protection efforts for them are stronger. Serving as the data source for this paper, they can enhance the objectivity and scientific nature of the research and have more practical guiding significance. The study focused on seven key nature reserves within these regions and ensured the comprehensiveness and timeliness of data collection through fine-tuned scheduling and fieldwork. Specifically, the research team conducted surveys in the Taibai Mountain Nature Reserve in Zhouzhi County, Shaanxi Province, from July 2 to 12 in 2017; in-depth surveys in the Gaoligong Mountain Nature Reserve during the period of August 5 to 9 in 2018; in-depth surveys in the Xishuangbanna Nature Reserve in Yunnan Province during the period of August 12 to 17 in 2018; and surveys in the Shaanxi Province from July 19 to 25 in 2019 in the Changqing and Crested Ibis Reserve for a comprehensive survey, while the Dulbert Mongolian Autonomous County in the northeast was surveyed from September 15 to 22 of the same year.

However, it is important to note that the sudden outbreak of the novel coronavirus pandemic from late 2019 to early 2020 delivered an unprecedented shock to socio-economic activities globally, including in China. The tourism industry in nature reserves was not spared, suffering significant losses [40,41]. To ensure the purity and accuracy of the research data and to avoid potential disruptions caused by the pandemic, this study’s data collection was deliberately concluded in 2019, establishing a temporal benchmark that safeguards the objectivity and scientific rigor of the research conclusions.

Semi-structured interviews are mainly conducted with the village head, with the aim of grasping the current situation of a village as a whole and making a preliminary design for the subsequent structured interview questionnaire. The structured interview questionnaire surveyed the community residents around the national nature reserves in three provinces of China. Data collection covered the Zhouzhi Mountain, Taibai Mountain, Changqing and Zhumadian Reserves in Shaanxi Province; the Gaoligong Mountain and Xishuangbanna Reserves in Yunnan Province; and the Zhalong Reserve in Heilongjiang Province, spanning 72 administrative villages. The survey team was composed of researchers from the University of Chinese Academy of Social Sciences, local college students and teachers, as well as professionals from the nature reserves. They conducted one-on-one interviews to complete the questionnaires. In Yunnan, sampling was mainly concentrated in the Mengla section of the Xishuangbanna Reserve and four surrounding communities outside the Gaoligong Mountain Reserve. In Heilongjiang, the study focused on the communities around the Zhalong Reserve in Duobei Mongolian Autonomous County, including the experimental area and core area of the Zhalong Reserve and its surrounding areas. In Shaanxi, villages on the periphery of the Zhouzhi, Taibai Mountain, Changqing and Zhumadian Reserves were prioritized. A systematic sampling framework was adopted, with 5–10 administrative villages selected within and around the reserves, 20–30 households randomly chosen from each village. A total of 1261 questionnaires were distributed, with 1106 valid responses, resulting in an effective response rate of 87.7%. A cross-provincial database of farmers’ livelihoods in nature reserve communities was established.

4. Research Methods

4.1. Construction of Explanatory Variables

4.1.1. Households’ Well-Being

To measure the level of well-being of farm households, this paper utilizes the Sustainable Livelihoods Framework (SLF) proposed by the United Nations Millennium Ecosystem Assessment (MA), drawing on the works of Wang et al. (2024) [27] and Zhang et al. (2023) [18]. It categorizes households’ well-being into five dimensions: the households’ well-being score for life (HWS-life), the households’ well-being score for health (HWS-health), the households’ well-being score for safety (HWS-safety), the households’ well-being score for social relations (HWS-society), and the households’ well-being score for choice (HWS-choice). Researchers obtain each of the five scores by weighting 4–6 related questions using the entropy weighting method (

Table 1. Weights of indicators calculated by entropy method.

Primary Indicators	Weights	Secondary Indicators	Weights	Combined Weights
HWS-life	0.1070	A1: Degree of burden of access to food	0.1582	0.0169
		A2: Degree of ease of access to water	0.1606	0.0172
		A3: Degree of burden of access to other necessities	0.1598	0.0171
		A4: Degree of ease of access to VDC services	0.1797	0.0192
		A5: Satisfaction with household housing conditions	0.1728	0.0185
		A6: Satisfaction with household’s arable land resources	0.1690	0.0181
HWS-health	0.0875	B1: Physical health	0.2233	0.0195
		B2: Mental health	0.2139	0.0187
		B3: Mood	0.2439	0.0213
		B4: Rest	0.3189	0.0279
HWS-safety	0.0659	C1: Personal safety	0.1904	0.0125
		C2: Property security	0.1771	0.0117
		C3: Low crime rate	0.3360	0.0221
		C4: Government’s efforts to fight crime	0.2019	0.0133
		C5: Food safety	0.0946	0.0062
HWS-society	0.2253	D1: Neighborhood friendliness	0.0761	0.0171
		D2: Trusted community friends	0.1447	0.0326
		D3: Trustworthiness of community neighbors	0.1607	0.0362
		D4: Funeral and wedding participation	0.6185	0.1393
HWS-choice	0.5143	E1: Degree of freedom of access to resources	0.2550	0.1311
		E2: Equity in access to education	0.2483	0.1277
		E3: Degree of freedom to build infrastructure	0.2224	0.1144
		E4: Freedom of agricultural production	0.2743	0.1411

).

$${HWS}_{ih}={\sum }_{j=1}^{n1}{w}_j^1{z}_{ij}$$

(1)

$$z_{ij}={\displaystyle \frac{z_{ij}^{\prime }-min\left(z_{ij}^{\prime }\right)}{{max(z}_{ij}^{\prime })-min(z_{ij}^{\prime })}}$$

(2)

$${\eta }_{ij}={\displaystyle \frac{z_{ij}}{\sum _{i=1}^{n2}{z}_{ij}}}$$

(3)

$${\theta }_j=-{\displaystyle \frac{1}{\mathit{ln}\left(n2\right)}}{\sum }_{i=1}^{n2}{\eta }_{ij}\times \mathrm{l}\mathrm{n}({\eta }_{ij})$$

(4)

$$w_j^1={\displaystyle \frac{(1-{\theta }_j)}{\sum _{j=1}^{n1}(1-{\theta }_j)}}$$

(5)

$w_j^1$ is the weight of each question; z_ij is the standardized score of each question; and $z_{ij}^{\prime }$ is the initial score of each question, which is a total of 5 levels on the Likert-type impact factor table, from low to high, from 1 very low to 5 very high, respectively. Finally, the scores of the five aspects of the well-being of the five farm households were then weighted by the entropy method to obtain the total score of the well-being of the farm households.

$${HWS}_i={\sum }_{h=1}^5{w}_h^2{HWS}_{ih},$$

(6)

4.1.2. Natural Resource Dependence

Researchers obtain the degree of natural resource dependence by dividing each interviewed farm household’s income from protected area-related natural resources by their total household income. They assume that all markets are perfectly competitive and that each unit of resource generates equal income for the farm household; therefore, the share of protected area-related income to total income reflects the share of resource utilization in the protected area to total resource utilization. Farm households’ protected area-related natural resource income consists of three components: cash income or the economic value of protected area products [42,43,44], income from tourism activities [45,46,47], and loss compensation income [48,49,50]. Loss compensation income represents the additional economic value that farm households could obtain if they could freely utilize resources in the protected area, and including this compensation income more accurately reflects farm households’ dependence on natural resources.

$${NRD}_l={\displaystyle \frac{{RI}_l}{{TI}_l}}$$

(7)

NRD_l denotes natural resource dependence, RI_l natural resource income of farm households, TI_l total income of farm households, and l individual farm households.

4.2. Variable Selection

4.2.1. Core Explanatory Variables

In order to explore the relationship between farm household livelihood outcomes and the geographic location of their residence, this paper categorizes the geospatial information of farm households into two dimensions: the location of the farm household’s residence relative to the protected areas and its location relative to the nearest bazaar.

Protected areas are one of the locations where farm households access natural resources and ecosystem services. The ease of access to these resources and services may influence farm households’ livelihood outcomes [51,52]. At the same time, regulations related to protected areas may impose restrictions on farm households’ use of natural resources, limiting their freedom to utilize them. With the development of tourism in protected areas, tourism may provide additional income to farm households, potentially affecting their livelihood outcomes [53,54]. Therefore, the relationship between the farm household’s residence and the protected area is one of the explanatory variables selected in this study.

Bazaars provide farm households with access to production materials, opportunities to sell products, and labor services, making them vital sites for economic exchange and influencing farm households’ production and livelihoods [55,56,57]. Thus, the location of a farm household’s residence relative to the bazaar is also an important factor influencing livelihood outcomes.

4.2.2. Control Variable

To control for the influence of farm household livelihood capital, this paper selects a range of control variables. The individual characteristics include gender, age, health status, marital status, and ethnicity. Livelihood capital control variables include whether the household adopts alternative livelihood strategies. Household human capital is represented by the highest level of education, skill training, family population number, and the number of laborers. Social capital is represented by whether any household member holds a village cadre position. Natural capital is measured by the household’s land area. Physical capital is represented by the household’s housing and major assets, while financial capital is measured by the receipt of in-kind or cash income and total household income [58]. The livelihood strategy variable reflects whether alternative livelihood measures are adopted. The ease of transportation affects the cost of accessing resources and selling products; the more convenient the transportation, the lower the time cost, which in turn affects livelihood outcomes [59,60,61]. Additionally, since all the households in this study are located near protected areas with similar management practices and abundant natural resources, the influence of other variables is controlled.

4.3. Construction of the Model

4.3.1. Four-Quadrant System

In this paper, the four-quadrant method [32] was applied to categorize the livelihood patterns of farm households near protected areas into four categories, as shown in Figure 1, namely high well-being with high dependency (H-H), high well-being with low dependency (H-L), low well-being with high dependency (L-H), and low well-being with low dependency (L-L). Specifically, the well-being of farm households was first categorized into high and low well-being according to the median, with those above the median being high well-being farm households and those below the median being low well-being farm households. Farm households’ dependence on natural resources was further categorized into two modes of high and low dependence according to the median, with high dependence above the median and low dependence below the median. Finally, the two divisions were ranked and combined to obtain the final four livelihood outcome divisions.

4.3.2. MNL Model

After categorizing farm household livelihood outcomes, the probabilities of the four outcomes can be used as dependent variables, with various influencing factors as independent variables, analyzed through a multinomial logistic regression model. The multinomial logistic model is used when analyzing discrete data with more than two options, where the choice is related to individual characteristics rather than the options themselves. In this study, the dependent variables are discrete data, and the independent variables are all related to individuals, making the multinomial logistic model appropriate. The standardized model equation is as follows:

$${Prob}_k{\displaystyle \frac{e^{{\beta }_k{x}_l}}{1+\sum _{k=1}^3{e}^{{\beta }_k{x}_l}}}$$

(8)

$${Prob}_4{\displaystyle \frac{1}{1+\sum _{k=1}^3{e}^{{\beta }_k{x}_l}}}$$

(9)

In the equation, we define the four livelihood strategies of farmers as 1 = L-L, 2 = L-H, 3 = H-L, and 4 = H-H. We use Prob_k to represent the probability of a farm household achieving each livelihood outcome and x_l to denote the individual-level variables, including the explanatory and control variables in this study. β_k is the parameter to be estimated.

By dividing Equation (8) by Equation (9) and taking the natural logarithm of both sides, we obtain Equation (10); β_kl shows the change in the log-odds ratio for a unit change in the independent variable. We select the H-H group as the reference group; Prob₄ represents the probability of the H-H strategy, as follows:

$$ln\left({\displaystyle \frac{{Prob}_k}{{Prob}_4}}\right){\sum }_{l=1}^{n3}{\beta }_{kl}{x}_l{\sigma }_k$$

(10)

5. Results

5.1. Descriptive Statistical Analysis

5.1.1. Explanatory Variable

The explanatory variables in this paper are all discrete variables. According to

Table 2. Descriptive statistics of explanatory variables.

Variables	Options	Frequency	Percent
Whether the household is in a protected area	Inside protected areas	352	31.83%
	Outside protected areas	754	68.17%
Distance of the household from the bazaar	Within 5 km	182	16.46%
	5 km–10 km	450	40.69%
	Beyond 10 km	474	42.86%

, 31.83% of the farm households, totaling 352 households, live within the nature reserve, while 68.17% of the farm households, totaling 754 households, live outside the nature reserve. The data on the distance of farm households from the bazaar shows that 182 farm households, accounting for 16.46% of the total, are within 5 km of the bazaar; 450 households, accounting for 40.69%, are between 5 and 10 km from the bazaar; and 474 households, accounting for 42.86% of the sample, are more than 10 km away from the bazaar.

5.1.2. Descriptive Statistics of Control Variables

From the analysis of

Table 3. Descriptive statistics of continuous control variables.

Variables	Description of Indicators	Mean	Std.dev.	Min	Max
Age	Year	50.31	11.1957	20	78
Number of working people in the family	People	2.77	1.1445	0	6
Family population number	People	4.23	1.4621	1	10
Family land area	Acres	4.43	2.3026	0	28
Number of large assets in the household	Individuals	7.57	3.1093	0	30
Total income	CNY	37,931.18	44,669.32	120	722,570

, the youngest respondent is 20 years old, the oldest is 78 years old, and the average age is 50.31 years old. All respondents are adults, which enhances the reliability of the data. In the sample, the maximum number of family members is 10, the minimum is 1, and the average is 4.23. The minimum number of family labor forces is 0 (that is, there is no labor population), and the maximum is 6. The average is 2.77. By comparing the data of family population and labor population, it is found that the proportion of the labor population in the family population is relatively high, and the labor force resources are relatively abundant. The minimum land area of the family is 0 mu, the maximum is 28 mu, and the average is 4.43 mu per household, indicating that the family’s land resource conditions are relatively good. The maximum number of large household assets held by a household is 30, the minimum is 0, and the average is 7.57. The highest total annual household income is CNY 722,570, the lowest is CNY 120, and the average annual household income is CNY 37,931.18, showing that the overall income is relatively low.

Through the analysis of

Table 4. Descriptive statistics of discrete control variables.

Variables	Options	Freq.	Percent	Variable	Options	Freq.	Percent
Ease of travel	Inconvenient	740	66.91%	Highest education level of family members	Illiterate	12	1.09%
	Average	212	19.17%		Elementary school	93	8.42%
	Convenient	154	13.92%		Junior high school	411	37.19%
Gender	Male	623	56.33%		High school	362	32.76%
	Female	483	43.67%		Secondary school and vocational high school	115	10.41%
Health status	Fully healthy	754	68.17%		University undergraduate or above	112	10.14%
	Sick but able to work	270	24.41%	Skill training	Not attended training	862	77.94%
	Sick and unable to work	76	6.87%		Attended training	244	22.06%
	Unable to take care of oneself	6	0.54%	Whether family members are village cadres	Yes	295	26.67%
Marital status	Married	1047	94.67%		No	811	73.33%
	Unmarried	23	2.08%	Family housing	Earth (grass) wood	378	34.18%
	Divorced	7	0.63%		Brick wood	372	33.63%
	Widowed	29	2.62%		Brick and concrete	223	20.16%
Ethnicity	Han nationality	635	57.41%		Reinforced cement soil	133	12.03%
	Minority nationality	471	42.59%	Family receives cash	Not acquired	949	85.8%
					Acquired	157	14.2%

, it can be seen that 740 farm households, accounting for 66.91%, consider the transportation inconvenient; the farm households who consider the transportation to be average or convenient account for nearly one-third (33.09%) of the sample, indicating that most farm households consider the transportation inconvenient. Males account for 44% of the total number of respondents, and females account for 56%, reflecting a relatively reasonable gender ratio. Most respondents report being in good health (68.17%); only a part of the respondents have no working ability (7.41%). The vast majority of respondents are married (94.67%); the unmarried population accounts for 2.08% of the total; the proportion of widowed people is close to that of the unmarried (2.62%); the proportion of divorced people is the smallest (0.63%). Among the sample, there are 635 Han people, accounting for 57.41%; there are 471 ethnic minority people, accounting for 42.59%, and the proportion of the two is relatively close, indicating a reasonable data structure. The highest educational attainment of farm households’ families is concentrated in junior high school and senior high school (37.19% and 32.76%); those with an educational attainment above senior high school account for only 20.55%; those with an educational attainment below junior high school account for 9.51%. Farm households that have participated in skills training account for 22.06% of the total; while farm households that have not participated in skills training account for 77.94%, showing a relatively low proportion of those receiving skills training. Among the interviewed farm households, 73.33% of the families have no family members serving as village cadres, and 26.67% of the families have members serving as village cadres. Most farm households’ houses are of grass wood and brick wood structures, accounting for 34.18% and 33.63% of the total, respectively; brick and concrete structures account for 20.16%; the reinforced concrete structures account for the smallest proportion, which is 12.03%. The proportion of families receiving in-kind or cash subsidies is relatively small, only 14.20%, and 85.80% of the families have not received in-kind or cash subsidies.

5.2. Comparison of Livelihood Strategies Under Different Livelihood Outcomes

According to

Table 5. Comparison of results of different livelihood strategies.

Livelihood Outcomes	L-L	L-H	H-L	H-H	Total
Alternative livelihoods do not exist	N = 245	N = 188	N = 86	N = 190	N = 709
	34.56%	26.52%	12.13%	26.80%	100%
Alternative livelihoods exist	N = 112	N = 8	N = 110	N = 167	N = 397
	28.21%	2.02%	27.71%	42.07%	100%
Total	N = 357	N = 196	N = 196	N = 357	N = 1106

Pearson chi2(3) = 142.6100 Pr = 0.000. Abbreviations: high well-being with high dependency (H-H), high well-being with low dependency (H-L), low well-being with high dependency (L-H), and low well-being with low dependency (L-L).

, there are 709 farm households without alternative livelihoods, of which 245 farm households with livelihood outcome L-L account for 34.56% of the total; 188 farm households with livelihood outcome L-H account for 26.52%; 86 farm households with H-L outcome account for 12.13% of the total; and 190 farm households with livelihood outcome H-H account for 26.80% of the total. There are 397 households with alternative livelihoods, of which there are 112 households with L-L livelihoods (28.12% of the total), 8 households with L-H livelihoods (2.02% of the total), 110 households with H-L livelihoods (27.71% of the total), and 167 households with H-H livelihoods (42.07% of the total). The number of farm households without alternative livelihoods is 312 more than the number of farm households with alternative livelihoods, indicating that a large proportion of the interviewed farm households do not have alternative livelihoods. The general level of well-being (H-L + H-H) was higher among the farm households with alternative livelihoods, which accounted for 69.78% of the overall number. Meanwhile, only 2.02% of the group with alternative livelihoods had high dependence on natural resources and low well-being, while the proportion of farm households with high well-being and high dependence amounted to 42.07%, which indicates that farm households with alternative livelihoods generally utilize the natural resources in the protected areas more efficiently.

5.3. Analysis of Variance of Natural Resource Dependence Under Different Livelihood Outcomes

According to

Table 6. Comparison of differences in natural resource dependence.

	L-L	L-H	H-L	H-H	Mean	Adjusted R2
Protected Area Product Index	0.028 (2,4)	0.26 (1,2,4)	0.033 (2,4)	0.16 (1,2,3)	0.11	0.285
Protected Area Tourism Index	0.0012 (2,4)	0.033 (1,4)	0.0061 (4)	0.26 (1,2,3)	0.092	0.309
Compensation Index	0.00083 (4)	0.0078	0.0012 (4)	0.018 (1,3)	0.0078	0.013
NRD	0.030 (2,4)	0.30 (1,3,4)	0.040 (2,4)	0.44 (1,2,3)	0.21	0.502

The numbers in parentheses in the upper right corner, 1 = L-L, 2 = L-H, 3 = H-L, 4 = H-H, indicate that the mean value is different from the mean value of the other livelihood results at the 5% level of significance under the Scheffe test. Abbreviations: high well-being with high dependency (H-H), high well-being with low dependency (H-L), low well-being with high dependency (L-H), and low well-being with low dependency (L-L).

, there is a significant difference between the high well-being, high-dependence livelihood group and the other three livelihood outcomes in terms of their reliance on the tourism index for protected areas, which is greater than their reliance on the protected area product index and compensation index. This indicates that farm households with high well-being, high-dependence outcomes mainly rely on protected area resources through tourism. Farm households with low well-being, high-dependence outcomes, on the other hand, primarily depend on selling products from the protected area, which leads to greater damage to natural resources and limited improvements in well-being. Additionally, the compensation income index for low well-being, high-dependence farm households is similar to that of low well-being, low-dependence, and high well-being, low-dependence farm households, demonstrating that this production model causes significant harm to natural resources while receiving weaker compensation support. Table 6 shows that most of the income from protected area resources for high well-being, high-dependence farm households comes from tourism, indicating that their use of natural resources puts less pressure on the environment, aligning with the requirements for "well-being-ecological" coupled development. In contrast, the low well-being, high-dependence group utilizes protected area resources in more traditional ways, leading to development at the expense of natural resource consumption and environmental degradation, with weaker sustainable development capacity.

5.4. Analysis of Variance of Households’ Well-Being Under Different Livelihood Outcomes

It can be known from

Table 7. Comparison of differences in well-being of farm households.

	L-L	L-H	H-L	H-H	Mean	Adjusted R2
HWS-life	0.70 (2,3,4)	0.66 (1,3,4)	0.88 (1,2)	0.89 (1,2)	0.78	0.355
HWS-health	0.67 (3,4)	0.69 (3,4)	0.81 (1.2,4)	0.88 (1,2,3)	0.76	0.186
HWS-safety	0.75 (2,3,4)	0.80 (1,3,4)	0.86 (1,2,4)	0.91 (1,2,3)	0.83	0.217
HWS-society	0.31 (2,3,4)	0.37 (1,3,4)	0.68 (1,2,4)	0.80 (1,2,3)	0.55	0.591
HWS-choice	0.11 (3,4)	0.13 (3,4)	0.30 (1,2,4)	0.45 (1,2,3)	0.26	0.515
HWS	0.29 (2,3,4)	0.31 (2,3,4)	0.52 (1,2,4)	0.63 (1,2,3)	0.44	0.702

The numbers in parentheses in the upper right corner, 1 = L-L, 2 = L-H, 3 = H-L, 4 = H-H, indicate that the mean value is different from the mean value of the other livelihood results at the 5% level of significance under the Scheffe test. Abbreviations: high well-being with high dependency (H-H), high well-being with low dependency (H-L), low well-being with high dependency (L-H), and low well-being with low dependency (L-L).

that in terms of health, safety, social relationships, free choice, and total well-being, the scores of farmers with high well-being and high dependence are higher than those of the other three types of farmers, significantly at the 5% level. And these farmers often adopt sustainable livelihood models such as developing ecotourism, which indicates that the "benefit-ecology" coupling is effective.

5.5. Multinomial Logistic Regression Model

5.5.1. Multicollinearity Test

Before conducting the regression, the data was first tested for multicollinearity. In this study, the variance inflation factor (VIF) was used to detect multicollinearity. Generally, if VIF < 4, the problem of multicollinearity is not significant. The correlation between independent variables has a relatively small impact on the regression results, and the estimated parameters are relatively stable, and the standard error will not be overly inflated due to multicollinearity.

Table 8. Multicollinearity VIF test.

Variables	VIF	1/VIF
ln(total income)	2.02	0.494661
Number of working people in the family	1.91	0.524152
Livelihood strategy	1.75	0.570518
Family population number	1.62	0.616304
Age	1.39	0.718117
Number of large assets in the household	1.36	0.73483
Health status	1.3	0.76703
Highest education level of family members	1.16	0.864025
Family receives cash	1.16	0.865014
Gender	1.08	0.924438
Family land area	1.07	0.931487
Marital status	1.05	0.951954
Whether family members are village cadres	1.04	0.961914
Ethnicity	1.02	0.977433
Family housing	1.02	0.979836
Skill training	1.01	0.985596

shows that the maximum VIF value in this study is 2.02, which is less than 4. It can be considered that there is no problem of multicollinearity.

5.5.2. Analysis of Regression Results

To study the impact of geographical factors on farm households’ shift towards the “well-being-ecological” coupled development model (H-H), this paper uses multinomial logistic regression to obtain the results.

When exploring the livelihood strategies and outcomes of farm households, geographic location serves as a central variable with complex and far-reaching impact mechanisms. Based on Models 1–3 in

Table 9. Multinomial logistic regression results.

Vars	Model 1 (L-L)	Relative Risk Ratio	Model 2 (L-H)	Relative Risk Ratio	Model 3 (H-L)	Relative Risk Ratio
Whether the household is in a protected area	−2.284 ***	0.102 ***	1.010 *	2.744 *	-2.371 ***	0.0934 ***
	(0.293)	(0.0299)	(0.610)	(1.674)	(0.315)	(0.0294)
Distance of the household from the bazaar	−0.539 ***	1.715 ***	0.194	0.824	−0.404 *	1.497 *
	(0.192)	(0.329)	(0.249)	(0.205)	(0.216)	(0.324)
Ease of travel	−1.421 ***	0.241 ***	−1.010 ***	0.364 ***	−0.873 ***	0.418 ***
	(0.158)	(0.0381)	(0.212)	(0.0774)	(0.153)	(0.0638)
Individual characteristics control variables	Controlled		Controlled		Controlled
Livelihood strategy control variable	Controlled		Controlled		Controlled
Livelihood capital control variable	Controlled		Controlled		Controlled
Constant	1.924	6.849	10.517 ***	36,924.58 ***	−4.755 **	0.0861 **
	(1.810)	(12.394)	(2.192)	(80,936.15)	(2.131)	(0.0183)
Observations	1106		1106		1106

Standard errors in parentheses *** p < 0.01, ** p < 0.05, * p < 0.1. Figures in parentheses are standard deviations. Abbreviations: high well-being with high dependency (H-H), high well-being with low dependency (H-L), low well-being with high dependency (L-H), and low well-being with low dependency (L-L).

, this paper analyzes in depth the direct effect of geographic location, especially the distance of residence from protected areas and marketplaces, on the transformation of farm households’ livelihood outcomes by controlling for key variables such as livelihood strategies and livelihood capital, using the H-H (high well-being with high dependency) livelihood outcomes as a benchmark. The following is a detailed interpretation and extended analysis of this research.

First, there is the difference between inside and outside protected areas: the “threshold effect” on livelihood outcomes. The regression results in Model 3 show that households living inside protected areas exhibit significant trends in livelihood outcomes compared to those outside protected areas. Specifically, residing inside the protected area significantly narrows the probability ratio between H-L (high well-being-low dependence) and H-H, while promoting the magnitude of positive H-H change, an effect validated at the 1% significance level. This finding reveals a potential “threshold effect” of protected area policies on farm households’ livelihoods: once farm households cross the geographic boundary into the protected area, their livelihood strategies and outcomes may be positively transformed by factors such as policy guidance, resource protection, and external support.

Furthermore, according to the modeling results of this study, the distance of farm households’ residence from the bazaar is also an important factor influencing livelihood outcomes. Farm households who are close to the bazaar have a relatively greater likelihood of transforming their livelihood outcomes from H-L to H-H, but this change is significant at the 10% level of significance, indicating that the radiation effect of the bazaar as a center of economic activity still exists and that natural resources acquired by farm households within the reserve need to be traded in the bazaar for sale.

Second is the delicate balance between market distance and livelihood options. The results of Model 2 show a more complex relationship between market distance and livelihood outcomes. Specifically, the relationship between distance from residence to market and the formation of L-H (low welfare-high dependency) and H-H livelihood outcomes by farm households is not significant, suggesting that distance to market may not be the main factor influencing these two livelihood outcomes.

However, it is noteworthy that the probability of forming L-H was significantly higher for farmers living inside the PA compared to those outside the PA than for those in H-H, a finding supported at the 10% level of significance. This reflects the fact that farmers in protected areas are more likely to experience a decline in welfare. It was learned in the semi-structured interviews that farmers in the protected area are facing strict protection policies, which also increases the risk of conflicts with wild animals. If farmers fail to transform to other sustainable livelihood methods in a timely manner and maintain the traditional model, they are very likely to fall into development predicaments. Third, the polarization of livelihood strategies within protected areas and challenges to the presumed ‘honeypot effect’ in adjacent zones are corroborated by Model 1 regression results, which demonstrate a statistically significant association between geographic location and livelihood outcomes. Specifically, households living in protected areas that are also close to markets are significantly less likely to fall into L-L (low well-being-low dependence) than H-H livelihood outcomes. This finding emphasizes the “dual advantage” of location—the resource advantage of the PA and the economic radiation of the market.

It is worth noting, however, that although living in the protected area brought some improvement to the livelihoods of the farm households, it did not fully support the hypothesis of a “honeypot effect”, whereby living in the protected area automatically led to an overall improvement in the well-being of the farm households. Instead, the findings suggest that living in protected areas may have facilitated the transition from L-L and L-H to H-H patterns but also hindered further transition from L-H to H-H, suggesting that improvements in livelihood outcomes are not linear and unconditional but are influenced by a combination of factors.

5.5.3. Marginal Analysis

Analysis of the data in

Table 10. Results of marginal analysis of multinomial logistic model.

Vars	Options	L-L	L-H	H-L	H-H
Whether the household is in a protected area	Inside protected areas	0.575 ***	0.0114 *	0.286 ***	0.127 ***
		(0.0446)	(0.00662)	(0.0408)	(0.0235)
	Outside protected areas	0.252 ***	0.136 ***	0.112 ***	0.499 ***
		(0.0233)	(0.0236)	(0.0153)	(0.0281)
Distance of the household from the bazaar	Within 5 km	0.275 ***	0.129 *	0.148 ***	0.449 ***
		(0.0558)	(0.0715)	(0.0368)	(0.0701)
	5 km–10 km	0.338 ***	0.0794 ***	0.156 ***	0.426 ***
		(0.0310)	(0.0193)	(0.0227)	(0.0323)
	Beyond 10 km	0.457 ***	0.050 ***	0.198 ***	0.295 ***
		(0.0380)	(0.0140)	(0.0307)	(0.0321)

Standard errors in parentheses *** p < 0.01, * p < 0.1. Figures in parentheses are standard deviations. Abbreviations: high well-being with high dependency (H-H), high well-being with low dependency (H-L), low well-being with high dependency (L-H), and low well-being with low dependency (L-L).

reveals the significant impact of distance from inside and outside the protected area and from the settlement to the bazaar on the livelihood outcomes of the farm households. Farm households inside protected areas have significant advantages in terms of livelihood improvement, especially for those who are able to fully utilize natural resources and market opportunities. However, farm households outside the protected areas face greater challenges and uncertainty. In addition, the ease of market access is also one of the important factors affecting the livelihood outcomes of farm households. Through in-depth analysis of these data, we can gain a more comprehensive understanding of the mechanisms by which farm households develop different livelihood outcomes and the socio-economic and environmental factors behind them.

• Comparative analysis of livelihood outcomes within and outside protected areas

Figure 2 shows the marginal effect of whether the household live in the protected area. The first is the low well-being-low dependency (L-L) analysis. The data show that the probability of L-L livelihood outcomes for farmers outside the protected areas is 57.5%, much higher than the 25.3% in the protected areas. This significant difference suggests that farm households outside protected areas generally face greater challenges in terms of economic and asset accumulation. Possible reasons for this include the following: first, there are fewer restrictions on resource exploitation and utilization outside of protected areas, but at the same time there is a lack of effective guidance and support, making it difficult for farm households to develop a stable source of income; and second, farm households outside of protected areas may face more intense market competition, with relatively fewer opportunities for accessing resources and markets, which in turn limits the scope for improving their livelihoods.

Second is the analysis of the low well-being with high dependence (L-H) group. For the L-H livelihood outcome, the probability is 13.7% for farmers inside the protected area, which is significantly higher than the 0.989% for farmers outside the protected area. This reflects the high level of resource dependence among farm households within the protected area, despite their low well-being. This may be related to the ecological protection policies and resource management systems in place. A considerable portion of farm households within the protected area are still unable to fully leverage resource advantages to improve their wealth, and their participation in ecotourism and related industries remains insufficient.

The third is the high-well-being with low dependency (H-L) analysis. For the H-L outcome, the probability of a farm household outside the protected area is 29.4% compared to 11.4% inside the protected area. This indicates that some of the farm households outside the protected area may have obtained higher income through non-agricultural activities or migrant labor, etc., and their well-being has improved. On the contrary, farm households inside the protected areas have fewer opportunities to earn high incomes without relying on protected area resources due to resource and market constraints, and the well-being of farm households is closely related to the utilization of protected area resources.

Fourth, the high well-being with high dependency (H-H) analysis shows that the probability of H-H outcome for farmers in the protected area was 49.6%, much higher than 12.2% outside the protected area. This data strongly suggests that farm households within the PA have a significant advantage in fully utilizing the natural resources of the PA to improve their livelihoods. This may be due to the relative abundance of natural resources and policy support in the protected areas, which enable farm households to realize the win-win situation of ecological conservation and well-being enhancement through sustainable development methods such as ecotourism and green agriculture.

2.

Impact of distance from residence to marketplace on livelihood outcomes

Figure 3 shows the marginal effect of the distance from the household to the bazaar. The first is the effect of distance from residence to bazaar on L-L livelihood outcomes. When the distance from the farmers’ residence to the bazaar increases, the probability of the L-L livelihood outcome increases accordingly. Specifically, the probability of L-L for farmers living more than 10 km from the bazaar is as high as 46.5%, while the probability of L-L within 5 km from the bazaar drops to 25.6%. This suggests that transportation costs and ease of access to markets are important factors influencing the livelihood outcomes of farmers. Longer distances increase the difficulty for farm households to access means of production and sell their products, limiting the scope for livelihood improvement.

Second is the impact of the distance from the residence to the market on the L-H livelihood outcome. The closer the residence is to the market, the higher the probability that the farm households’ livelihood outcome will be L-H. When the distance to the market exceeds 10 km, the probability of an L-H livelihood outcome is 4.49%. When the distance is between 5 and 10 km, the probability increases to 7.44%, and when within 5 km, it further rises to 15.3%. The results suggest that the closer farm households live to the market, the more dependent they may be on natural resources. This is likely because lower transportation costs encourage farm households to engage in low-efficiency resource-based activities, such as harvesting, which keeps their well-being levels low and damages the environment.

The third is the effect of distance from residence to the bazaar on the H-L livelihood outcome. For H-L livelihood outcome, the probability decreases as the distance from residence to the marketplace decreases. This may be due to the fact that proximity to market access makes it easier for farm households to access off-farm employment opportunities and market information, thereby increasing the likelihood that they will improve their economic situation through diversified livelihood strategies. However, it is worth noting that this downward trend is not significant, which may imply that high well-being is not solely determined by ease of market access but is also influenced by a variety of other factors.

Fourth is the effect of distance from residence to bazaars on H-H livelihood outcomes. Unlike the L-L and L-H/H-L types, the probability of the H-H livelihood outcome tended to increase significantly with decreasing distance from residence to market. The probability of H-H is as high as 44% when the distance from the farmer’s residence to the market place is within 5 km, which is much higher than other distance segments. This is further evidence of the importance of ease of market access in achieving high well-being. It also reflects the higher sensitivity and capacity of farm households in the protected areas to utilize natural resources and market opportunities.

The relative location of the agglomerations to the nature reserves and bazaar had different marginal impacts on the livelihood outcomes of the farm households, which were subtly reflected in the four different livelihood outcomes. Specifically, farmers inside and outside protected areas are prone to L-H versus H-H livelihood outcomes, which are diametrically opposed to the changes in livelihood outcomes induced by farmers’ proximity to the bazaar.

Farmers in protected areas are more inclined to develop L-H rather than H-H livelihood outcomes than those outside protected areas. There are multiple underlying reasons behind this phenomenon. On the one hand, the establishment of protected areas often implies the abundance and diversity of natural resources in the area, which provides farm households with unique conditions that enable them to rely on natural resources more conveniently for their production and livelihoods, whether through direct collection and utilization (L-H) or efficient utilization combined with modern technology and management tools (H-H). Lower transportation costs and a better understanding of the distribution of resources enable farm households in protected areas to have a better grasp of resource information and to develop more efficient livelihood strategies.

It is worth noting, however, that while the natural environment of the protected areas offers many advantages to farm households, the challenges associated with it should not be overlooked. Potential threats due to high wildlife activity, lagging infrastructure, and some constraints on resource utilization increase the uncertainty of farm households’ livelihoods, which may prompt some farm households to shift to L-H. On the other hand, the proportion of farm households opting for L-H and H-H increased significantly as they moved closer to marketplaces, while the proportion of farmers with L-L and H-L outcomes decreased. Strategies decrease accordingly. This change reveals the far-reaching impact of market accessibility on farm households’ livelihood decisions. As the center of commodity exchange, the bazaar not only provides a broad platform for farm households to sell their products but also an important channel for them to obtain production materials and technical support. With the shortening of distance, the reduction in transportation cost makes it easier for farm households to bring agricultural products and tourism service products collected in the protected area to the market, and at the same time, they can obtain all the resources needed for production from the bazaar at a lower cost. This two-way convenience undoubtedly enhances farm households’ reliance on and efficiency in utilizing natural resources in the protected areas and promotes the diversification and advancement of their livelihood strategies.

3.

Analysis on the relations between resource utilization in protected areas and market opportunities

When analyzing the comparison of livelihood outcomes inside and outside protected areas, we must delve into the interactive analysis of resource utilization in protected areas and market opportunities. Firstly, regarding resource utilization, households within protected areas exhibit a high dependence on natural resources, but this dependency does not always translate to high well-being. For instance, in the L-H model, despite the high dependence on resources, the well-being level of households within protected areas is relatively low. This may be related to inefficient resource utilization methods and lack of diversity, such as over-reliance on traditional gathering industries while overlooking the potential of sustainable development methods like eco-tourism and green agriculture. However, in the H-H model, households within protected areas have successfully achieved a win-win situation in resource utilization and well-being improvement. This is attributed to their deep exploitation and efficient use of natural resources, as well as the development of industries like eco-tourism and green agriculture supported by policy. This indicates that the transformation of resource utilization methods and the application of modern technologies are crucial for enhancing the well-being of households within protected areas. Conversely, the impact of market opportunities on household livelihood outcomes cannot be overlooked. An analysis of the impact of proximity to settlements and markets on livelihood outcomes shows that ease of access to markets has a significant impact on optimizing livelihood outcomes and increasing levels of well-being. For households within protected areas, although they possess abundant natural resources, without effective market channels to sell these resources or their processed products, the value of these resources cannot be fully realized. Therefore, households within protected areas need to more actively explore market opportunities, such as developing e-commerce and establishing cooperatives to broaden sales channels and increase product value. At the same time, the government and relevant departments should increase support for households within protected areas, providing market information, technical training, and other forms of assistance to help them better integrate into the market, achieve effective resource conversion, and continuously improve well-being.

5.6. Robustness Test

In order to make the research conclusion robust, in this study, the well-being level of farmers was calculated by the mean of five scores: life, health, safety, social relationship, and free choice. The scores such as life were calculated based on the mean of their sub-items. The calculation formula is

$$HWS={\displaystyle \frac{HWS-life+HWS-health+HWS-safety+HWS-society+HWS-choice}{5}}$$

(11)

$$HWS-life={\displaystyle \frac{A1+A2+A3+A4+A5+A6}{6}}$$

(12)

$$HWS-health={\displaystyle \frac{B1+B2+B3+B4}{4}}$$

(13)

$$HWS-safety={\displaystyle \frac{C1+C2+C3+C4+C5}{5}}$$

(14)

$$HWS-society={\displaystyle \frac{D1+D2+D3+D4}{4}}$$

(15)

$$HWS-choice={\displaystyle \frac{E1+E2+E3+E4}{4}}$$

(16)

The recalculated levels of farmers’ well-being and their dependence on natural resources were reclassified into four (L-L, L-H, H-L, H-H) livelihood outcomes according to the four-quadrant method.

Firstly, an analysis of variance was conducted on the happiness and dependence indices of different farmers for the four newly classified livelihood outcomes, and

Table 11. Multivariate analysis of variance.

	L-L	L-H	H-L	H-H	Mean	Adjusted R2
Protected Area Product Index	0.029 (2,4)	0.26 (1,3,4)	0.031 (2,4)	0.16 (1,2,3)	0.11	0.281
Protected Area Tourism Index	0.0019 (2,4)	0.043 (1,3,4)	0.0050 (2,4)	0.26 (1,2,3)	0.092	0.292
Compensation Index	0.00087 (4)	0.0081	0.0011 (4)	0.018 (1,3)	0.0078	0.013
NRD	0.031 (2,4)	0.31 (1,3,4)	0.037 (2,4)	0.43 (1,2,3)	0.21	0.497
HWS-life	3.77 (2,3,4)	3.56 (1,3,4)	4.47 (1,2)	4.58 (1,2)	4.14	0.429
HWS-health	3.60 (3,4)	3.62 (3,4)	4.43 (1,2,4)	4.60 (1,2,3)	4.08	0.322
HWS-safety	4.03 (2,3,4)	4.22 (1,3,4)	4.58 (1,2,4)	4.7 (1,2,3)	4.38	0.323
HWS-society	2.82 (2,3,4)	3.08 (1,3,4)	4.19 (1,2,4)	4.52 (1,2,3)	3.66	0.597
HWS-choice	1.54 (2,3,4)	1.66 (1,3,4)	2.07 (1,2,4)	2.73 (1,2,3)	2.04	0.377
HWS	3.15 (2,3,4)	3.23 (1,3,4)	3.97 (1,2,4)	4.23 (1,2,3)	3.66	0.716

The numbers in parentheses in the upper right corner, 1 = L-L, 2 = L-H, 3 = H-L, 4 = H-H, indicate that the mean value is different from the mean value of the other livelihood results at the 5% level of significance under the Scheffe test. Abbreviations: high well-being with high dependency (H-H), high well-being with low dependency (H-L), low well-being with high dependency (L-H), and low well-being with low dependency (L-L).

of the results was plotted. By comparing Table 6 with Table 7, it can be found that the results are basically the same, with only minor differences in the two items of “HWS-choice” and “Protected Area Tourism Index”, but this does not affect the conclusion of the study. L-H farmers rely on material products within the protected area for their livelihoods, which is clearly distinguished from other types of farmers. The tourism index of H-H farmers was significantly higher than that of other farmers, and the levels of other well-being except for living well-being were significantly higher than those of other farmers. This indicates that their high well-being is related to the development of ecotourism.

Secondly, the regression results can also support the previous conclusion. As shown in

Table 12. Multinomial logistic regression results.

Vars	Model 1 (L-L)	Relative Risk Ratio	Model 2 (L-H)	Relative Risk Ratio	Model 3 (H-L)	Relative Risk Ratio
Whether the household is in a protected area	−2.322 **	0.0981 ***	0.555	1.742	−2.446 ***	0.0867 ***
	(0.299)	(0.0293)	(0.547)	(0.953)	(0.321)	(0.0278)
Distance of the household from the bazaar	−0.414 **	1.512 ***	−0.244	0.783	−0.501 **	1.650 **
	(0.193)	(0.292)	(0.252)	(0.197)	(0.220)	(0.363)
Ease of travel	−1.353 ***	0.259 ***	−0.912 ***	0.402 ***	−0.931 ***	0.394 ***
	(0.157)	(0.0405)	(0.209)	(0.0839)	(0.158)	(0.0622)
Individual characteristics control variables	Controlled		Controlled		Controlled
Livelihood strategy control variable	Controlled		Controlled		Controlled
Livelihood capital control variable	Controlled		Controlled		Controlled
Constant	3.802 **	44.804 **	11.962 ***	156,654.8 ***	−5.688 ***	0.0339 ***
	(1.885)	(84.444)	(2.237)	(350,457.8)	(2.223)	(0.0753)
Observations	1106		1106		1106

Standard errors in parentheses *** p < 0.01, ** p < 0.05. Figures in parentheses are standard deviations. Abbreviations: high well-being with high dependency (H-H), high well-being with low dependency (H-L), low well-being with high dependency (L-H), and low well-being with low dependency (L-L).

, within the protected area or closer to the market, it is beneficial for L-L and H-L farmers to convert to H-H results. However, for L-H farmers, this geographical advantage does not affect their conversion to H-H. However, different from the previous results, the livelihood results of farmers in the protected area may not change from H-H to L-H, which indicates that this result is not robust. Similarly, this result does not affect the core conclusion of this study.

6. Discussion

Building on the findings of the study, this paper further explores the livelihood outcomes of farm households in protected areas and the complex economic, social, and environmental factors behind them. These findings are placed within the broader academic context, comparing and supplementing the work of Guerbois, Fritz, Wang, and other relevant scholars, aiming to comprehensively reveal the diversity and dynamics of livelihoods within protected areas. Guerbois and Fritz [16] emphasized in their research that the establishment of protected areas often provides ecosystem services, enabling nearby residents to utilize protected area resources to improve their well-being, leading to what is known as the ‘honey pot effect’. However, the results of this study do not fully support this optimistic view. We found that farm households living inside the protected area are diversifying their use of protected area resources, such as through tourism-related services, to enhance their well-being. In contrast, due to inconvenient transportation and limited knowledge of the protected area, surrounding farm households outside the protected area have low participation in tourism, which limits further improvements in their well-being [16]. This situation reveals the broad and complex nature of ecosystem service functions, suggesting that the link between ecosystem service provision and farm households’ actual well-being is neither linear nor automatic but is moderated by various social, economic, and cultural factors. While Guerbois and Fritz [16] may have focused more on the positive impacts of ecosystem services under traditional resource use, this study, by incorporating data on tourism income, uncovers the complexity and diversity of farm households’ livelihoods at the micro level. Therefore, future research should focus more on how to combine ecosystem service protection with livelihood improvements for farm households, exploring win-win development pathways. This study shares some common ground with the conclusions of Wang [32]. Wang’s research pointed out that proximity to markets is one of the key factors affecting farm households’ livelihood outcomes, particularly in resource-dependent communities. This study echoes that finding, also identifying that farm households closer to markets are more likely to adopt high resource-dependence livelihood strategies, as they reduce transaction costs and enhance the market competitiveness of their products. This conclusion is further reinforced by the marginal effects analysis, which visualizes the close relationship between spatial distance and livelihood outcomes. However, this study reflects the impact of protected areas on farm households through changes in their well-being rather than solely through income. Wang’s research may focus more on the economic effects of market proximity, while this study incorporates multidimensional factors such as health, safety, social relationships, and freedom of choice to comprehensively analyze the complex causes of farm households’ livelihood outcomes. Additionally, this study highlights tourism as a potential pathway to mitigate the negative impacts of protected area construction, supplementing and deepening Wang’s research by offering new perspectives and strategies for the sustainable development of communities around protected areas.

In addition to the studies by Guerbois, Fritz, and Wang, many other scholars have focused on the similarities and differences in the livelihoods of farm households inside and outside protected areas, as well as the factors influencing them. For example, some studies have explored how policy interventions, such as ecological compensation policies, drive the livelihood transformation of farm households. Zhang et al. [62] pointed out that well-designed policies can effectively incentivize farm households to reduce their direct dependence on natural resources and shift towards more sustainable livelihood strategies. These studies align with the ideas presented in this research, as both emphasize the crucial role of policy in mediating the relationship between farm households’ livelihoods and ecological protection. However, this study focuses more on understanding the actual behavior of farmers and revealing the internal logic and external constraints behind their livelihood outcomes, thus providing more detailed and specific references for policy formulation. Moreover, some research focuses on farm households’ perceptions and understanding of the value of ecosystem services and how these perceptions influence their livelihood decisions [14]. These studies highlight the significant role of farm households as direct users and managers of ecosystems in ecological conservation. Although this research does not directly address the cognitive aspects of farm households, it indirectly reflects their trade-offs and decisions between ecological protection and livelihood development through an in-depth analysis of their livelihood strategies and the underlying reasons. Future studies could further combine data on farm households’ perceptions to explore how raising awareness of ecological protection could foster a more harmonious relationship between people and nature.

The competitive market, by activating the potential value of ecotourism, has opened up a characteristic path for the sustainable livelihoods of farmers that combines ecological protection and economic gain. Around nature reserves, farmers rely on unique ecological resource endowments and participate in tourism market competition to transform ecological value into economic value. Farmers focus on the supply of ecological services and develop supporting projects such as hiking guides, nature education, ecological picking, and homestays, forming a complementary service with the core scenic area. This market participation based on ecotourism has enabled farmers to shift from traditional resource-dependent livelihoods to service-based and experience-based economies, forming a virtuous cycle of “protection—development—value addition” [63,64]. The innovative drive and resource integration capabilities generated by the competitive market have made ecotourism a bridge connecting the ecological value of protected areas and the improvement of farmers’ livelihoods, providing a market-based solution to the contradiction between protection and development.

Compared with the studies of Guerbois, Fritz, Wang, and other scholars, this study validates the viewpoints of the previous scholars and at the same time puts forward new insights and challenges, emphasizing the multidimensionality and dynamics of the livelihood issues of farm households in the construction of protected areas. Future research should further focus on the following aspects: first, deepening the exploration of the relationship between farm households’ cognition and behavior and revealing its internal logic; second, strengthening interdisciplinary cooperation and integrating the multidisciplinary perspectives of economics, sociology, and ecology to comprehensively analyze the complex system of farm households’ livelihoods both inside and outside the protected areas; and third, exploring more effective policy tools and market mechanisms to promote the coordinated development of ecological conservation and farm households’ livelihoods. Through these efforts, we are expected to contribute more wisdom and strength to the sustainable development of protected areas and neighboring communities. Finally, this study also has some limitations. The research data were collected before the COVID pandemic and failed to take into account the impact of the COVID epidemic on farmers’ engagement in the tourism industry. Under the conditions where rural households lack assets and their livelihoods are generally fragile, the impact of this part cannot be ignored.

7. Conclusions and Policy Implications

Based on the degree of dependence on natural resources and the well-being level of households, this paper meticulously categorizes the livelihood outcomes of households into four patterns: high well-being with high dependence (H-H), high well-being with low dependence (H-L), low well-being with high dependence (L-H), and low well-being with low dependence (L-L). By utilizing the multinomial logistic model, the direct impact of households’ geographical location on their livelihood outcomes is thoroughly explored. The research findings reveal that, after excluding factors such as transportation convenience, individual characteristics, livelihood strategies, and livelihood capital, the influence of geographical location remains significant. Specifically, due to their geographical location advantage, households within protected areas can more easily access natural resources, tending to form livelihood patterns with high dependence, i.e., the probabilities of the L-H and H-H patterns increase. However, due to policy and regulation restrictions, the change rate of the L-H pattern is more pronounced compared to the H-H pattern. Furthermore, the production methods of households relying on natural resources are highly dependent on market mechanisms. The market not only provides households with production tools but also facilitates the commercialization of natural resources to meet their living needs. Therefore, the proximity of households to markets directly affects their likelihood of having livelihood outcomes in the L-H and H-H categories.

The H-H pattern, as a model of “well-being-ecology” coupling development for efficient natural resource utilization, has achieved initial success. For farmers capable of developing tourism (H-H), the government should encourage families to actively participate in tourism development, establish close cooperative relationships with surrounding communities, and jointly promote tourism prosperity. Measures such as providing employment opportunities and skills training should be taken to ensure that community residents benefit from tourism. As for the farmers (L-H) who are unable to develop tourism, the compensation mechanism should be improved to ensure the rational allocation of compensation funds and guarantee the living standards of these farmers first. However, it should be noted that while developing tourism, ecological protection principles should be strictly followed to ensure that tourism activities do not cause damage to the ecological environment. The following ecotourism management measures are proposed: First, set tourist flow restrictions to prevent overcrowding and environmental degradation. Second, strengthen tourism environmental protection regulations by fining littering and other destructive behaviors. Third, implement ecological impact assessments to monitor and mitigate any adverse effects of tourism on the natural environment. These measures will help balance conservation and development, ensuring the sustainable livelihoods of households surrounding protected areas.

In addition, the location of trading venues such as markets should be reasonably planned to facilitate information exchange and material exchange among households. During the planning process, the impact of market location on households should be fully considered to avoid inconveniences for households due to excessive distance from markets. Markets should preferably be set up in densely populated and easily accessible areas, and the service radius of the markets should be reasonably determined based on the travel habits and transportation conditions of local residents. Without compromising the ecological environment, transportation development should be strengthened to improve the convenience of households’ travel. Improvements in transportation conditions have a significant impact on the livelihood outcomes of households. During the planning and implementation of transportation projects, comprehensive environmental impact assessments should be conducted to ensure that the project location, design, and construction schemes minimize the impact on the ecological environment. Routes with minimal ecological impact should be prioritized to avoid crossing ecologically sensitive areas such as nature reserves, wetlands, and forests.