Natural Resource Dependence and Household Adaptive Capacity: Understanding the Linkages in the Context of Disaster Resettlement

Abstract

The largest disaster reduction and relocation project was conducted in Shaanxi Province, China, in an effort to reduce the threat of natural disasters and preserve the environment. Although the literature has attempted to assess the economic and ecological impacts of the project quantitatively, there is currently a dearth of research on the connection between resource dependence and adaptive capacity at the rural household levels. Using survey data from southern Shaanxi, China, this study evaluated and quantified natural resource dependence (NRD) and household adaptive capacity (HAC) in the context of disaster resettlement. Simultaneously, we explored the differences in NRD and HAC among various groups and relocation characteristics. An ordinary least squares regression model was used to specifically examine the relationship between them. The results indicated that, first, NRD was significantly and positively related to HAC. Second, the dependence of relocated households on energy, food, and income had a significantly positive correlation with HAC. Third, compared to local, involuntary, and scattered resettlement households, the HAC of relocated households, voluntary relocated households, and centralized resettlement households is substantially lower. The aforementioned findings have significant policy implications for rural China and other developing nations, as they can help reduce natural resource dependence and increase adaptive capacity.

1. Introduction

In recent years, social changes have had a profound impact on the living environment and lifestyle of the population, and resettlement occurs frequently at the international level. The World Bank defines resettlement as the process of relocating those who have been affected by instability or disadvantage, with the core objective of improving or at least restoring their income and standard of living [1]. Resettlement is a unique form of mobility. It is not just a physical spatial move but more like a social and emotional transformation that involves adapting to a new environment and making new connections. And the why, where, and how people move are well thought out and planned before they relocate [2]. In recent decades, there has been a growing number of resettlement projects and their justifications. Among them, some are relocations necessitated by development, with the aim of promoting sustainable economic and social development in the region [3]. Meanwhile, poverty alleviation relocation, as an important policy tool, aims to achieve poverty alleviation goals by relocating poor people from remote areas to better-developed regions [4]. Ecological migration is receiving increasing attention [5]. It reduces ecological vulnerability by relocating residents to more habitable places, effectively minimizing the interference of human activities on the natural environment and avoiding over-exploitation and destruction. In addition to the several types of relocation mentioned above, frequent natural disasters bring many economic, material, and even spiritual challenges to local residents. The issue of post-disaster resettlement is becoming increasingly prominent, and proper resettlement of disaster victims is becoming particularly important [6,7]. To address the multiple dilemmas of environmental protection, livelihood improvement, and poverty eradication, the Chinese government has completed millions of household resettlement projects. Among them, Shaanxi Province launched the largest-ever disaster resettlement project in 2011, covering 2.4 million people in 28 counties in three prefectures [8,9]. Previous research has concentrated on the demographic and ecological elements of disaster resettlement [10,11]; however, there is a relative lack of discussion related to reducing the natural resource dependence (NRD) of households and enhancing their adaptive capacity. On the one hand, NRD provides an important research perspective for us to deeply understand and explore the social impacts of disaster resettlement programs. For example, Liu (2024) empirically found in the context of disaster resettlement that participation in disaster resettlement programs is more likely to achieve low NRD [12]. On the other hand, adaptive capacity provides an important entry point for exploring livelihood reconstruction and restoration of affected rural households [13]. It has been suggested that migrant resettlement programs lead to a decrease, not an increase, in households’ adaptive capacity [6,14]. Previous studies have rarely discussed the role of households’ dependence on natural resources in influencing their adaptive capacity, which is a possible contribution of this study.

As a result of the concept of ‘natural capital’, academics are now concentrating on the connection between human activity and natural resources [15]. The economic and ecological value of natural resources has been widely recognized [16,17,18]. With the growing popularity of sustainable development in recent years, the effective utilization of natural resources and the reduction of NRD have become topics of widespread concern in the academic community [19]. The percentage of a rural household’s total income from natural resources is commonly used to assess the degree of dependence on these resources [20]. Specifically, it falls into direct and indirect categories, with direct dependence referring to the physical objects that individuals and households obtain directly from natural resources, such as fuel wood for cooking and heating, and indirect dependence includes the purchase of products derived from these natural resources [21]. Previous studies have used various methods to calculate and measure NRD in different contexts. Wu et al. (2023) focused on three pastoralist districts with different natural resource utilization patterns and empirically analyzed them using the Tobit model to conclude that pastoralist households mostly rely on natural resources to meet their income demands [22]. Liu (2024) combined NRD with rural household well-beings and found that households involved in disaster resettlement were more likely to achieve lower NRD [12]. Balbi (2019) modeled NRD with the use of elements such as household size and gender to construct a measure of overall NRD [23]. In addition, studies have found that the number of laborers, level of education, and geographic location can explain variations in NRD [24,25]. This study examines the degree to which rural households rely on NRD in the context of previous research.

Natural resources are a primary source of livelihood for rural households. Owing to the frequency of disasters and poor living conditions in their places of residence, rural households often rely heavily on local resources to meet their basic needs. Therefore, balancing NRD with household adaptive capacity (HAC) is a priority in disaster resettlement. The study of adaptive capacity originated in the field of climate change, with previous research focusing on conceptualization. Some typical definitions, such as that of Engle (2011), consider adaptive capacity as a potential characteristic of an individual, community, or social-ecological system in response to a threat or opportunity [26]. Maldonado (2014) stated that a family has the capacity to foresee, manage, and reduce disturbances brought about by humans or natural causes, as well as to bounce back from the effects of those disturbances [27]. Furthermore, Parsons’s study of resilience suggests a distinction between coping capacity and adaptive capacity. Coping capacity refers to the utilization of available resources and opportunities to maintain operations, whereas adaptive capacity refers to adjusting through learning, adaptation, and transformation to maintain resilience [28]. Many scholars have argued that adaptive capacity and vulnerability are inextricably linked and that assessing adaptive capacity should focus on those who are most vulnerable [29,30]. A considerable amount of adaptation-related literature has been framed in terms of resilience and vulnerability, with adaptation bridging the gap between them [31,32,33]. The adaptive capacity is complex, multidimensional, and dynamic. To measure this, scholars have explored various approaches and developed many frameworks and indicators. Mbaziira (2023) adopted and modified the Local Adaptive Capacity framework to assess the adaptive capacity of agro-pastoralists and found that innovation, educational attainment, non-agricultural activities, and loans had a significant effect on adaptive capacity [34]. The five capitals in the livelihoods framework are commonly used to assess adaptive capacity. Drawing on the five capitals, Nursey-Bray (2021) developed socioeconomic indicators such as poverty, household well-being, health, and education levels to measure adaptive capacity [35]. This study utilized and refined the HAC framework, introduced by Acosta (2013) [36] and Li (2017) [37], to conduct an empirical analysis of rural households’ adaptive capacity for disaster resettlement.

As mentioned above, the existing literature focuses on natural resource dependence and adaptive capacity in different contexts but has rarely explored the relevant impacts of resettlement projects at the rural household scale from the perspective of combining NRD and adaptive capacity. Therefore, this study examines the linkages between disaster resettlement, NRD, and the HAC to provide lessons for sustainable development in the region. The potential contribution of this study is that it discusses and validates the changes in NRD and their effect on adaptive capacity in an attempt to complement research on the relationship between the two and provide new directions for subsequent research. Based on the above considerations, this study aims to answer two crucial questions: (1) What is the current situation of the relocated households in the disaster resettlement areas in southern Shaanxi, and how can their ability to adapt to a new life be improved? (2) How does NRD in the region affect HAC? Drawing on prior research and the extant literature, this study followed the following steps to achieve this goal: First, we calculated the HAC by selecting 13 specific indicators from three dimensions: awareness, capacity, and action. Second, three aspects were used to compute and measure NRD: energy, food, and income. Subsequently, the degrees of NRD and HAC of several rural household types and samples were compared. Finally, using an ordinary least squares regression, the effect of natural resources on HAC was examined for several samples by incorporating various relocation characteristics. The remaining sections are arranged as follows: The research methodology and data sources are introduced in the second section, the findings are presented in the third section, the fourth section summarizes the discussion, and the fifth provides the conclusion.

2. Data Sources and Research

2.1. Research Area

Ankang Prefecture (Figure 1) is one of the three counties in Shaanxi where disaster resettlement is carried out, and this study was carried out there [38]. It is situated in the southern part of Shaanxi Province, located in Qinba’s hinterland, and has many national nature reserves within its boundaries, where ecological protection is under great pressure and natural disasters are frequent, hampering economic and social development and putting rural households’ livelihoods at greater risk. To completely solve the constraints imposed on rural households in the fragile ecological environment, Shaanxi Province started a large-scale migrant relocation project in the Qinba mountainous area, where Ankang Prefecture is located, as early as 2011, covering 28 counties in the province of Shaanxi and 2.4 million people [39]. Through migration and relocation projects, local poverty and other problems have been effectively solved, and the living standards of local residents have improved. Overall, relocation work in Ankang started early, was a heavy task, and encompasses various types of households; therefore, it has a good representativeness for conducting this research.

2.2. Data Collection

Data collection was performed on rural household livelihoods in many counties and districts of Ankang Prefecture, Shaanxi Province, which provided the data used in this study. For this survey, our team utilized a multi-stage stratified sampling method to conduct questionnaire surveys among rural households and communities. The specific operational approach was as follows: Firstly, considering the dispersed rural households and inconvenient transportation in the Ankang region, which posed significant challenges for field surveys and interviews, three key counties—Hanbin, Ziyang, and Ningshan—were selected for the survey based on their gross domestic product (GDP). Secondly, administrative villages were sampled from the selected sample counties, specifically including 4 townships and 8 administrative villages in Ningshan County and Hanbin District, as well as 3 typical resettlement communities in Ziyang County. Each village sample was further categorized into three types: local households without relocation, resettlement villages composed entirely of relocated households, and mixed resettlement villages with both relocated and local households. Thirdly, based on the list of villager groups provided by the village committees, 2 villager groups were randomly selected from each administrative village. However, during the actual survey process, we found that a large proportion of households in the selected villager groups were found to be engaged in migrant work, resulting in their absence from the villages during our survey period. Lastly, we conducted face-to-face household surveys with all permanent residents in the sampled villager groups during the survey period. In the end, 657 valid questionnaires were collected, consisting of 198 from non-relocated and 459 relocated households.

The questionnaire comprised the following main sections: (1) Fundamental details regarding rural households, including gender, age structure, health status, household size, average education level of household members, and income composition. (2) Natural, social, financial, and material components of household livelihood capital, such as the amount of land that can be farmed, the number of households that can receive assistance, per capita net income, and housing structure [41,42]. (3) NRD in rural households, particularly on energy, food, and income [20]. (4) Relocation-related information, i.e., whether is the household was relocated, when it was relocated, the type of relocation, the nature of the relocation, and the method of resettlement. Drawing from previous research [8,38], in this study, households were divided into two categories: relocated and non-relocated (local), and into two types of relocation: centralized and scattered resettlement. The nature of relocation was categorized as voluntary or involuntary. The questionnaires were entered into a dataset and processed.

2.3. Indicator Construction

2.3.1. Household Adaptive Capacity (HAC)

Adaptive capacity is the potential that an individual, community, or social-ecological system has in response to a threat or opportunity [26]. In terms of livelihoods of rural households, following Acosta (2013) [36] and Li (2017) [37], this study categorized HAC into 3 dimensions: awareness, ability, and action, and selected 13 specific indicators from 6 determinants to construct the HAC index (Figure 2). Awareness is described by experience, which responds to rural households’ adaptation to detrimental alterations in the environment [29,43]. It is specifically constituted by the head of household’s age and previous work experience. The more experience, the better the judgment a household can make when facing external change. The head of the household is the nucleus of the family. The age of the head of the household affects to some extent the choices made by the household in the face of external shocks. Meanwhile, the ability dimension is determined primarily by material resources, infrastructure, and technology. Material resources typically reflect a household’s economic status, as represented in this study by the housing structure [8] and cultivated land area [44]. Infrastructure evaluation considers physical capital and distance to the main road [43]. Closer proximity to the main road facilitates household travel, and a richer accumulation of physical capital facilitates changes and improvements in household livelihoods. Technology can facilitate rural households’ adaptation to change, with choices measured by skills training [45] and the number of village leaders [13]. Rural households that participated in the training had better choices of adaptation strategies than those that did not. Village cadres are equipped with new ideas and technologies that reduce risk when facing external shocks. Finally, economic resources and flexibility determine the actions. Agricultural income [38] and housing value [9] are commonly used measures of household economic resources. The ability of a household to endure and recover in the face of external challenges is referred to as flexibility and is mostly related to household size [43] and non-farm income [46]. Non-farm income represents the flexibility of household livelihoods; the higher the non-farm income, the more flexible and adaptable the household income. We used the methodologies of earlier research by Erenstein (2010) and Quandt (2018) [44,47]. First, the raw data were standardized for polar deviation to compare data from different units. Second, for every respondent, the individual indicator scores were averaged to provide a composite determinant index for each of the six determinants. The average score for each adaptive capacity component was calculated. Finally, the three-component results for each household were averaged to determine the total household HAC index. Consequently, this index represents the full range of HAC measures for the surveyed households and ultimately serves as a proxy for overall HAC.

2.3.2. Household Natural Resource Dependence (NRD)

Natural resources can sustain rural household livelihoods and increase their resilience to risks. Natural resource dependence is used to assess the extent of a household’s dependence on natural resources and is usually measured as the percentage of a rural household’s total income from natural resources, which is commonly used to assess (Figure 3) [20]. This study captures three components: the proportion of natural resource utilization in a household’s food supply (referred to as food dependence), the portion of directly used natural resources in a household’s energy usage (referred to as energy dependence), and whether the source of income includes gains from the utilization of natural resources (referred to as income dependence). First, the percentage of the NRD portion of rural household food expenditure was used as the degree of food dependence, which was calculated as the self-sufficient portion of rural household food divided by the total household food expenditure [24]. Second, energy dependence was calculated as the percentage of the natural resource portion of energy consumed directly by rural households [42]. Third, income dependence was quantified by computing the percentage of all household income from agriculture, forestry, and livestock [48]. Finally, the overall HAC was determined by averaging the three indices that assessed dependence on energy, food, and income [49].

2.4. Regression Analysis Model

This study created a multivariate linear regression model and used the ordinary least squares (OLS) method to evaluate the relationship between a household’s NRD and HAC. NRD and HAC served as independent and dependent variables, respectively, and whether the household relocated, type of relocation, and nature of the relocation served as control variables. The specific calculation formula is as follows:

$$\gamma ={\beta }_0+{\beta }_1{X}_1+{\beta }_2{X}_2+\cdots +{\beta }_n{X}_n+\mu$$

where ${\beta }_0$ is a constant term; ${\beta }_1$, ${\beta }_2$, $\cdots$ ${\beta }_n$ are regression coefficients; and $\mu$ is the random error term.

Based on previous literature [6,42], the regression was divided into two sections. First, we divided the sample into three categories (relocated households, local households, and the entire sample) and used varying sample sizes to examine the impact of natural resource dependence on HAC. Second, to evaluate the influence of NRD on HAC under various relocation features, we introduced relocation, the type of relocation, and the nature of relocation throughout the entire sample.

Table 1. Variables used in the regression analysis.

Variables	Description	Mean	SD
Whether relocated	Household relocation = 1; non-relocation = 0	0.699	0.459
Relocation type	Resettlement that is centralized = 1; scattered = 0	0.758	0.429
Relocation nature	Relocation that is voluntary = 1; involuntary = 0	0.862	0.345
Household size	The number of people living in the household (persons)	4.496	1.608
Dependence ratio	The ratio of youngsters and the elderly to the household’s total labor force (%)	0.277	0.225
Education level	Average years of education for household members (years)	6.180	2.654
Experience	Types of experiences for household members (1. Village leaders; 2. Technicians, educators, and doctors; 3. Workers at enterprises; 4. Military personnel; and 5. No previous experience)	0.498	0.841
Phone charge	Household members’ last-month phone charge (CNY)	228.122	357.361
Loan	The loan’s potential (certainly = 1, probably = 2, generally = 3, less likely = 4, and certainly cannot = 5)	3.430	1.348
Social support	The total amount of aid funds available (CNY)	61.072	185.960

Note: USD 1 = CNY 6.2284 in the survey year.

lists the variables used in the regression analysis.

Relocation characteristics are mainly categorized into whether a household relocated, the type of relocation, and the nature of relocation. Among the 459 households that relocated, 395 (86.057%) relocated voluntarily. Voluntary relocation refers to relocation owing to poverty, disasters, or other reasons that render the original residence unviable. By contrast, involuntary relocation is forced by an external coercive force, where migrants are subjectively unwilling to relocate. Compared with the resistance of involuntary relocated households, voluntary relocated households show more active acceptance and thus have better adaptive capacity and recovery ability. There are two primary types of resettlement methods: centralized resettlement (354 households) and scattered resettlement (43 households). Centralized resettlement is led by the government, which centrally manages relocated households [50]. Centralized resettlement communities tend to be larger in size, and related industrial support is not yet complete, prompting households to select their sources of livelihood. Scattered resettlement households tend to have better economic conditions and higher levels of off-farm livelihoods. The education variable measures the educational level of a household. Higher educational levels were associated with more flexible employment options. The average household size in the survey area was 4.496, and the size of the household directly affected the demand for natural resources [43]. The dependence ratio describes the ratio of youngsters and the elderly to the household’s total labor force. A higher labor force share value indicates that the lower the demographic burden, the more resilient the household is to risk. Telephone bills and social support belong to the social capital of a household, and when households encounter shocks, their social network can act as a buffer to help alleviate their difficulties. In the questionnaire, experiences contained five options (1. Village leaders; 2. Technicians, educators, and doctors; 3. Workers at enterprises; 4. Military personnel; and 5. No previous experience). Previous literature [8,42] found that households with experiences 1–4 have greater livelihood sustainability and a lower likelihood of falling into poverty. Finally, loans represent a household’s financial capital, primarily in terms of the likelihood of obtaining them.

3. Results

3.1. Comparing the NRD of Different Types of Households

Table 2. NRD of different types of households.

Indices	Whether Relocated			Relocation Type			Relocation Nature
	Yes	No	t-Test	Centralized	Scattered	t-Test	Voluntary	Involuntary	t-Test
Total dependence	0.087 (0.006)	0.238 (0.130)	12.089 ***	0.071 (0.006)	0.142 (0.016)	5.204 ***	0.076 (0.006)	0.162 (0.019)	5.096 ***
Energy dependence	0.097 (0.011)	0.363 (0.022)	12.365 ***	0.066 (0.010)	0.195 (0.029)	5.486 ***	0.073 (0.010)	0.248 (0.038)	5.993 ***
Food dependence	0.032 (0.006)	0.101 (0.147)	5.259 ***	0.023 (0.006)	0.068 (0.015)	3.316 ***	0.034 (0.006)	0.035 (0.013)	0.052
Income dependence	0.131 (0.010)	0.252 (0.022)	5.758 ***	0.124 (0.011)	0.162 (0.022)	1.635	0.122 (0.010)	0.204 (0.035)	2.913 ***

Note: *** indicate that the t values are significant at the 1% significance level.

displays the level of NRD of various household types. The overall dependence of households that relocated (0.087) was notably less than that of households located (0.238). Compared to local households, there was a significant decrease in the dependence on energy, food, and income. In terms of resettlement type, the total dependence on centralized resettlement (0.071) was significantly lower than that on scattered resettlement (0.142). Furthermore, compared to households that relocated involuntarily (0.162), the total dependence of voluntary relocation (0.076) was lower. This may be because households that have relocated, especially those that relocated voluntarily and settled in a centralized resettlement, have higher expectations of relocation and are more willing to adapt to a series of changes in production and life brought about by relocation, are therefore in a better position to receive help from the government and the community, and have relatively lower NRD.

3.2. Comparing the HAC of Different Types of Households

The HAC of relocated households (3.529) was substantially lower than that of the local households (3.824;

Table 3. Comparing the level of HAC under various relocation characteristics.

Indices	Whether Relocated			Relocation Type			Relocation Nature
	Yes	No	t-Test	Centralized	Scattered	t-Test	Voluntary	Involuntary	t-Test
HAC	3.529 (0.034)	3.824 (0.063)	4.420 ***	3.477 (0.036)	3.748 (0.084)	3.423 ***	3.500 (0.036)	3.833 (0.101)	3.348 ***
Awareness	0.463 (0.011)	0.478 (0.017)	0.747	0.460 (0.013)	0.486 (0.022)	1.029	0.465 (0.012)	0.474 (0.024)	0.275
Ability	1.934 (0.023)	2.316 (0.044)	8.387 ***	1.882 (0.023)	2.112 (0.057)	4.547 ***	1.893 (0.023)	2.224 (0.071)	5.143 ***
Action	1.132 (0.0200	1.030 (0.037)	−2.607 ***	1.135 (0.022)	1.143 (0.047)	0.170	1.137 (0.022)	1.136 (0.057)	−0.024

Note: *** indicate that the t values are significant at the 1% significance level.

). The awareness and ability of relocated households were noticeably inferior to those of local households. The awareness and action aspects of voluntary relocation were largely consistent with those of households that relocated against their own will. Furthermore, the ability dimension of voluntarily relocated households was much lower than that of involuntarily relocated households. Meanwhile, the HAC of centralized relocation households (3.477) was substantially lower than that of scattered relocation households (3.746), which was mainly reflected in the ability dimension, with less difference between awareness and action. The disaster resettlement program greatly lowers the external danger to displaced households as a proactive preventative strategy against natural disasters. The reduced adaptive capacity of relocated households is due to the significant changes made to their previous living environments, in terms of production and living, and they require a longer period of recovery and reconstruction.

In addition, to visually analyze the differences in HAC under different livelihood strategies, this study compared the adaptive power of rural households using kernel density function plots (Figure 4). From the comparison of different livelihood types, the density distributions of HAC of pure, non-farm, and diversified rural households were slightly different, with similar and steep curves, indicating that the fluctuation in the adaptive capacity of these three groups was extremely small. Simultaneously, HAC values were concentrated at the mean. However, the curve was steeper for the non-agricultural type, which implies that the fluctuations in the adaptive capacity of this type of household were small. By contrast, the curves for pure agriculture and diversified livelihood types were relatively flat, suggesting that the fluctuations in HAC were more pronounced. In terms of peaks, the non-agricultural type had a significantly higher peak than the purely agricultural and diversified types, suggesting that the data on HAC were more intensive here. Furthermore, all three livelihood types showed a distribution pattern with an elongated right tail, indicating increased variation in HAC.

3.3. Analysis of NRD and HAC

This study classified households into three categories of NRD (high, medium, and low) using k-means cluster analysis to further examine the level of NRD. The three categories of NRD were used as the horizontal coordinates and the HAC as the vertical coordinates to create a box line plot (Figure 5). With a significance level of 0.000 and an F-statistic of 2274.55, the results demonstrated a substantial disparity between all three groups. These values were as follows: high dependence (0.500), medium dependence (0.249), and low dependence (0.034). The median of low dependence is in the center of the box, the upper and lower quartiles and the upper and lower truncation points are close to each other, and the symmetry of the data is strong, which belongs to the standard normal distribution. These findings suggest that dependence on natural resources is distributed evenly. The medians for medium and high dependence were skewed towards the bottom of the box, and the upper and lower cutoffs spanned a larger distance, showing a skewed distribution.

Meanwhile, in order to explore the relationship between NRD and HAC, primary fitting graphical analysis, secondary fitting graphical analysis, and kernel density regression fitting graphical analysis were used for discussion purposes. The results of the graphical fitting are shown in Figure 5b, where NRD shows an overall upward trend on HAC, and this force gradually increases as NRD increases.

3.4. The Influence of NRD on HAC

To investigate how NRD affects HAC, HAC was utilized as the response variable, and NRD (total dependence) and its three components (energy, food, and income dependence) were utilized as explanatory variables. Each family was categorized into local, relocated, and total samples using Models 1–12, as shown in

Table 4. The effect of NRD on HAC across several samples.

Variables	Local Households				Relocated Households				Total Sample
	Model 1	Model 2	Model 3	Model 4	Model 5	Model 6	Model 7	Model 8	Model 9	Model 10	Model 11	Model 12
Total dependence	−0.004				0.913 ***				0.606 ***
Energy dependence		0.078				0.474 ***				0.351 ***
Food dependence			−0.093				0.547 **				0.316 **
Income dependence				−0.030				0.305 **				0.211 **
Household size	0.136 ***	0.138 ***	0.137 ***	0.137 ***	0.194 ***	0.196 ***	0.198 ***	0.197 ***	0.183 ***	0.186 ***	0.183 ***	0.182 ***
Dependence ratio	−0.171	−0.181	−0.160	−0.170	−0.190	−0.188	−0.199	−0.203	−0.181	−0.177	−0.163	−0.157
Education level	0.030	0.031	0.030	0.030	0.039 ***	0.039 ***	0.406 ***	0.038 ***	0.042 ***	0.041 ***	0.044 ***	0.045 ***
Experience	0.382 ***	0.385 ***	0.380 ***	0.382 ***	0.246 ***	0.236 ***	0.239 ***	0.247 ***	0.2928 **	0.290 ***	0.286 ***	0.288 ***
Phone charge	0.000	0.000	0.000	0.000	0.000 *	0.000 *	0.000	0.000 *	0.000 **	0.000 **	0.000 *	0.000 *
Loan	−0.139 ***	−0.139 ***	−0.140 ***	−0.139 ***	−0.076 ***	−0.073 ***	−0.079 ***	−0.082 ***	−0.104 ***	−0.104 ***	−0.106 ***	−0.107 ***
Social support	0.002 **	0.002 **	0.002 **	0.002 **	−0.000	−0.000	−0.000	−0.000	−0.000	−0.000	−0.000	−0.000
Constant	3.147 ***	3.107 ***	3.156 ***	3.157 ***	2.529 ***	2.547 ***	2.584 ***	2.589 ***	2.686	2.670 ***	2.748 ***	2.740 ***
R2	0.426	0.427	0.426	0.426	0.408	0.405	0.393	0.392	0.407	0.406	0.400	0.400
N	193	193	193	193	450	450	450	450	643	643	643	643

Note: ***, **, and * indicate that the t values are significant at the 1%, 5%, and 10% significance levels, respectively.

and

Table 5. The effect of NRD on HAC in various relocation characteristics.

Variables	Model 13	Model 14	Model 15	Model 16	Model 17	Model 18	Model 19	Model 20	Model 21	Model 22	Model 23	Model 24
Total dependence	0.459 ***				0.835 ***				0.778 ***
Energy dependence		0.263 ***				0.424 ***				0.385 ***
Food dependence			0.213				0.480 **				0.534 **
Income dependence				0.140				0.285 **				0.248 *
Whether relocated
Relocated households	−0.126 **	−0.127 **	−0.182 ***	−0.179 ***
Relocation type
Centralized resettlement					−0.112 *	−0.113 *	−0.148 **	−0.158 **
Relocation nature
Voluntary relocation									−0.230 ***	−0.232 ***	−0.291 ***	−0.275 ***
Household size	0.184 ***	0.186 ***	0.185 ***	0.184 ***	0.194 ***	0.196 ***	0.180 ***	0.196 ***	0.191 ***	0.193 ***	0.194 ***	0.193 ***
Dependence ratio	−0.195	−0.191	−0.189	−0.184	−0.179	−0.177	−0.182	−0.185	−0.177	−0.176	−0.180	−0.185
Education level	0.038 ***	0.037 ***	0.037 ***	0.037 ***	0.038 ***	0.038 ***	0.040 ***	0.038 ***	0.037 ***	0.070 ***	0.038 ***	0.036 ***
Experience	0.290 ***	0.288 **	0.285 ***	0.286 ***	0.243 ***	0.235 ***	0.237 ***	0.245 ***	0.245 ***	0.237 ***	0.239 ***	0.246 ***
Phone charge	0.000 **	0.000 **	0.000 *	0.0000 **	0.000 *	0.000 *	0.000 *	0.000 *	0.000 *	0.000 *	0.000	0.000 *
Loan	−0.100 ***	−0.101 ***	−0.101 ***	−0.102 ***	−0.073 ***	−0.071 ***	−0.075 ***	−0.078 ***	−0.080 ***	−0.078 ***	−0.085 ***	−0.086 ***
Social support	−0.000	−0.000	−0.000	−0.000	−0.000	−0.000	−0.000	−0.000	−0.000	−0.000	−0.000	−0.000
Constant	2.807 **8	2.816 ***	2.900 ***	2.892 ***	2.613 ***	2.632 ***	2.670 ***	2.700 ***	2.776 ***	2.800 ***	2.882 ***	2.874 ***
R2	0.411	0.411	0.406	0.406	0.414	0.410	0.400	0.400	0.420	0.416	0.411	0.408
N	643	643	643	643	450	450	450	450	450	450	450	450

Note: ***, **, and * indicate that the t values are significant at the 1%, 5%, and 10% significance levels, respectively. The reference group for “Whether relocated”, “Relocation type”, and “Relocation nature” is local households, scattered resettlement, and involuntary relocation.

. To further examine the role of relocation features in the influence of NRD on HAC, Models 13–24 incorporate characteristics such as relocation, nature of relocation, and type of relocation.

Table 4 illustrates that Models 1, 3, and 4 demonstrate a negative correlation between HAC and total dependence as well as food and income dependence of local households. Model 5 demonstrates a strong positive correlation between total dependence and HAC at the 1% level. Similarly, in Models 6–8, the relocated households’ dependence on energy, food, and income were all positively correlated with HAC. This finding suggests that relocated households can enhance their HAC by fully utilizing their natural resources at the three levels of energy, food, and income. For the full sample, total dependence and energy dependence were positively associated with HAC at the 1% level, and food and income dependence were significantly positively associated with HAC at the 5% level (Models 9–12).

Table 5 presents relocation features to further explore the effect of NRD on HAC under the influence of relocation. According to the findings, there is a substantial negative link between relocated households, voluntary relocation, centralized resettlement, and HAC. Among the control variables, education level and household size showed significant positive correlations with HAC at the 1% level. HAC increases with the mean number of years of schooling, and a larger household size is linked to a higher probability of improved adaptive capacity. Second, the number of family members’ experiences was significantly and positively related to HAC. Additionally, we discovered that loans significantly harm HAC and that the excessive financial burden from loans affects the improvement of HAC. In addition, except for Model 23, the telephone bills were significantly and positively correlated with HAC.

Robustness tests typically evaluate the effectiveness of variables and approaches in the context of changes in parameters. To test for robustness, we used HAC as an ordered category and conduct regression analyses using an ordered logit model (

Table 6. Robustness test of the effect of NRD on HAC in various relocation characteristics.

Variables	Model 13	Model 14	Model 15	Model 16	Model 17	Model 18	Model 19	Model 20	Model 21	Model 22	Model 23	Model 24
Total dependence	1.400 ***				2.515 ***				2.236 ***
Energy dependence		0.725 ***				1.222 ***				1.009 **
Food dependence			0.588				1.295 *				1.492 **
Income dependence				0.505 *				0.946 **				0.827 **
Whether relocated
Relocated households	−0.270	−0.292 *	−0.438 ***	−0.415 **
Relocation type
Centralized resettlement					−0.305	−0.319	−0.413 **	−0.433 **
Relocation nature
Voluntary relocation									−0.805 ***	−0.807 ***	−0.976 ***	−0.922 ***
Household size	0.555 ***	0.557 ***	0.551 ***	0.552 ***	0.642 ***	0.649 ***	0.648 ***	0.650 ***	0.647 ***	0.652 ***	0.653 ***	0.651 ***
Dependence ratio	−0.674 **	−0.666 *	−0.617 *	−0.613 *	−0.744 *	−0.719 *	−0.675	−0.714 *	−0.736 *	−0.712 *	−0.684	−0.711 *
Education level	0.105 ***	0.105 ***	0.105 ***	0.105 ***	0.112 ***	0.112 ***	0.116 ***	0.115 ***	0.110 ***	0.112 ***	0.114 ***	0.113 ***
Experience	0.765 ***	0.754 ***	0.744 ***	0.750 ***	0.725 ***	0.692 ***	0.686 ***	0.714 ***	0.722 ***	0.691 ***	0.689 ***	0.711 ***
Phone charge	0.000 *	0.000 *	0.000	0.000 *	0.000 **	0.000	0.000	0.000	0.000	0.000	0.000	0.000
Loan	−0.274 ***	−0.275 ***	−0.271 ***	−0.276 ***	−0.212 ***	−0.207 ***	−0.210 ***	−0.219 ***	−0.228 ***	−0.225 ***	−0.230 ***	−0.238 ***
Social support	−0.000	−0.000	−0.000	−0.000	−0.000	−0.000	−0.000	−0.000	−0.000	−0.000	−0.000	−0.000
R2	0.039	0.039	0.039	0.039	0.043	0.043	0.041	0.042	0.045	0.044	0.048	0.044
N	643	643	643	643	450	450	450	450	450	450	450	450

Note: ***, **, and * indicate that the t values are significant at the 1%, 5%, and 10% significance levels, respectively. The reference group for “Whether relocated,” “Relocation type,” and “Relocation nature” is local households, scattered resettlement, and involuntary relocation.

). The results demonstrate that although the coefficients and significance of the ordered logit and OLS models fluctuate slightly but trend in the same direction, it can be stated that the results of the study are robust.

4. Discussion

Disaster resettlement is an important means of improving livelihoods and reducing natural hazards with significant economic and social impacts [51], as well as a solution for households to achieve sustainable livelihoods [6]. This study used a disaster resettlement project as a backdrop to examine how NRD affects HAC at the micro-rural household level. There was a strong and favorable relationship between the NRD of relocated households and HAC. Moreover, the relocated households’ energy, food, and income dependence had significant positive effects on HAC. The HAC of relocated households improved with an increase in NRD. This result is consistent with earlier studies that demonstrated the beneficial functions of natural resources [38,52]. This phenomenon occurs because, even though the original infrastructure of relocated households has improved and their livelihood activities have shifted from agriculture to non-agriculture, the resettlement sites are less supportive of the relocated people’s livelihood transformation and subsequent development. Our field survey found that fuelwood, crops, and other traditional industries that depend on natural resources are still the key factors for their survival. However, some scholars have different views on this topic. For example, household dependence on natural resources is detrimental to human capital accumulation [16,53], and resource overdependence may negatively affect the quantity and quality of education and health.

Additionally, our findings indicate a noteworthy distinction in NRD between local and relocated households. Undoubtedly, the way displaced households use natural resources will change as a result of moving from the natural environment on which they rely to their new surroundings. This will affect their daily livelihoods and may change their understanding of and reliance on natural resource utilization. Consistent with the findings of previous studies, disaster resettlement plays a crucial role in transforming the original livelihood patterns and reducing dependence on natural resources [39,42]. After resettlement, the original living environment of rural households is improved, livelihood activities are transformed from agriculture to non-agriculture, and infrastructure improvements (e.g., the use of clean energy) provide a realistic basis for the reduction of NRD of relocated households [54]. Our findings also show that relocated, voluntarily relocated, and centrally resettled rural households have considerably lower HAC than local, involuntarily relocated, and scattered resettled rural households. This is due to the fact that disaster resettlement efforts pose many challenges to the livelihoods and future development of relocated households [55,56]. Compared to local households, relocated households are highly vulnerable to difficulties in their livelihoods after resettlement because of the loss of livelihood assets during the disaster. This is reflected in difficulties such as low levels of education that prevent people from finding non-farm jobs, poor health of family members, and excessive household burdens.

Meanwhile, relocation characteristics have a substantial impact on HAC. First, the findings showed that relocation, voluntary relocation, and centralized resettlement were significantly and negatively associated with HAC. This finding is consistent with earlier findings showing that most relocated households do not adapt well to their new lives [6,14]. According to several studies, the enormous expenses associated with relocation put greater financial strain on relocated households, which in turn makes them more vulnerable [8,57]. Second, in disaster resettlement programs, government workers generally hold the concept that “once the relocated people arrive at the resettlement area, the work is over,” disregarding the significance of follow-up assistance. Although living conditions and essential services of households have improved significantly as a result of centralized resettlement, post-relocation support has not been fully effective in solving household livelihood adaptation and reconstruction problems. Returning to poverty remains a possibility for some households after the one-time relocation payment ends. Finally, in terms of the nature of relocation (categorized as voluntary or involuntary depending on the specifics of the relocation project), voluntary relocation significantly reduces HAC, a phenomenon that is simple to explain. Although voluntary relocation enables rural households to be more proactive in adapting to a new life, low levels of education, poor health, and narrow social networks [42,58], inevitably have a direct negative impact on HAC. Therefore, it is crucial to fully understand the impact of individuals’ behavioral patterns in different contexts on HAC in subsequent support policies.

Our study had certain limitations. First, the HAC evaluation indicators were derived from a framework created abroad, which has some limitations in the selection of indicators. The indicators may differ in different contexts and environments and require further validation and improvement. Second, owing to the characteristics of the surveyed regions, our assessment of NRD applies mainly to rural areas and lacks comparisons between different regions and different stages of development. Thus, it may not be possible to explain how NRD affects HAC in other situations. Third, this study examined only cross-sectional data and did not analyze how NRD affects HAC over time. To strengthen the persuasiveness of the findings, future research should consider broadening the scope of the study to collect survey data from various periods and regions.

5. Conclusions

Disaster resettlement is not merely the spatial relocation of populations and the physical reconstruction of communities; it profoundly affects the redistribution of populations, the optimal allocation of resources and the environment, and the overall adjustment of the economy and society. It is an important means of improving livelihoods and reducing natural hazards, as well as a solution for households to achieve sustainable livelihoods. From the standpoint of disaster resettlement, this study concentrates on how NRD affects HAC by referring to Acosta [36] and Li’s [37] research framework that categorizes HAC into three dimensions: awareness, ability, and action, and by drawing on Wang’s [20] study that assesses NRD in four categories: total dependence, energy, food, and income. This study used a t-test and introduced box-and-line plots, kernel density function plots, and OLS regression analysis to explore the impact of NRD on HAC. The results show that NRD is significantly and positively related to HAC. Meanwhile, energy, food, and income dependence were significantly and positively related to HAC. Additionally, the effect of NRD on HAC for various relocation characteristics was investigated. Compared to involuntary relocation (0.162) and scattered resettlement households (0.142), we discovered that the NRD of voluntary relocation (0.076) and centralized resettlement households (0.071) was markedly lower. In contrast, local (3.824), involuntarily relocated (3.833), and scattered resettled households (3.748) had substantially higher HAC than relocated (3.529), voluntarily relocated (3.500), and centralized resettled (3.477).

The contribution of our study is that the above findings can help rural households better integrate into the migrant resettlement sites, improve their household adaptive capacity to the new life, and enhance their sustainable livelihoods. In addition, these findings provide a reference for subsequent support strategies and policy formulation for rural resettlement in China. Natural resource dependence has a positive impact on rural households’ adaptive capacity, and the research in this study corroborates this finding. Although regional variations may cause variations in the study’s conclusions, changing the high adaptive capacity resulting from a high dependence on natural resources is a long-term process, and policymakers and stakeholder groups must collaborate to address and solve these challenges. When formulating policies, governments should prioritize harmonizing the relationship between protecting resources and increasing adaptive capacity. Our study has some limitations. On the one hand, the assessment methods and indicators we have selected may vary in different contexts and environments; on the other hand, our assessment applies mainly to rural areas and lacks comparisons across regions and stages of development. This study is only an attempt to highlight the issue of natural resource dependence and adaptive capability, and further empirical and theoretical research on this topic is expected. Clear guidance can be provided on how to translate low resource dependence and high adaptation into practical strategies that can help reduce natural resource dependence and enhance rural household adaptive capacity in a sustainable way.