Nonlinear Impact of Population Shrinkage on Urban Ecological Resilience: A Threshold Effect Analysis Based on City-Level Panel Data from the Yangtze River Economic Belt, China

Abstract

In the context of rapid urbanization and demographic transition, the implications of population shrinkage for urban sustainable development have attracted increasing scholarly attention. Nevertheless, empirical evidence on the relationship between population change and urban ecological resilience remains limited. Drawing on the Pressure–State–Response (PSR) framework, this study constructs a comprehensive indicator system to assess urban ecological resilience in 110 cities along the Yangtze River Economic Belt (YEB) over the period of 2012–2021. Furthermore, a panel threshold regression model is employed to examine the nonlinear effects of population shrinkage on urban ecological resilience. The findings indicate that urban ecological resilience exhibits an overall upward trend in YEB, characterized by pronounced spatial disparities. Eastern cities have a higher level of resilience than cities in the western region in YEB. The number of cities with shrinking populations is gradually increasing, and these shrinking cities are mainly small and medium-sized cities. The empirical results show that the impact of population shrinkage on urban ecological resilience is distinctly nonlinear, and regional economic development plays a moderating role in this nonlinear relationship. At lower levels of economic development, population shrinkage does not significantly moderate urban ecological resilience. As the economy reaches a moderate stage, population shrinkage exerts a stronger modulatory effect on ecological resilience. When economic development advances to a higher level, however, population shrinkage tends to inhibit ecological resilience. Overall, this study provides a scientific basis for the population–ecological policies tailored to local conditions and offers valuable insights to promote urban sustainable development under conditions of population shrinkage.

1. Introduction

Against the backdrop of rapid globalization and urbanization, population growth and urban spatial expansion have long been considered the core drivers shaping the spatial structure of cities, and are often regarded as key factors stimulating economic vitality and technological progress [1,2,3]. However, contemporary urban development is increasingly confronted with a “polycrisis”—a set of interconnected crises occurring simultaneously, including climate and environmental deterioration, economic instability, regional inequalities, and structural challenges such as population shrinkage and aging—which have been significantly accelerated under the ongoing impact of the COVID-19 pandemic [4]. These multifaceted challenges not only promote population shrinkage worldwide, but also fundamentally alter the operating conditions of urban systems, reshaping the mechanisms through which urban adaptive capacity and ecological resilience are formed [5]. In this context, population shrinkage, as a significant indicator of contemporary urban transformation, exerts a dual effect on urban ecosystems. On one hand, it can reduce resource consumption and environmental pressure, thereby providing opportunities for urban ecological recovery [6]. On the other hand, population shrinkage may lead to declining economic vitality, underutilization of public infrastructure, reduced investment in ecological governance, and weakened land management capacity, exposing urban ecosystems to new risks [7]. Consequently, the issue of urban ecological resilience under population shrinkage has become a critical and urgent topic in research on sustainable urban development [8,9,10].

In recent years, ecological resilience has emerged as a key indicator for assessing a city’s capacity to maintain ecosystem functions, safeguard residents’ well-being, and support sustained socioeconomic development. It has thus become an important theoretical tool for building resilient cities and advancing ecological civilization [11]. Most scholars define urban ecological resilience as the ability of urban ecosystems to maintain a stable state and respond rapidly and effectively to internal and external disturbances, thereby preserving the integrity of their structure and functions. Existing studies span multiple spatial scales, including the national level regional level, urban agglomerations, and ecologically fragile areas [12,13,14,15]. Research in this field has primarily focused on clarifying theoretical frameworks [16], constructing indicator systems [17], measuring resilience levels [18], and exploring influencing mechanisms [19]. The findings suggest that factors such as economic growth [20], industrial structure upgrading [21], and the level of foreign direct investment [22] exert positive effects on urban ecological resilience, whereas population growth [23] and environmental regulation [24] are found to have negative impacts on resilience levels. However, as an emerging demographic dynamic, the mechanisms through which population shrinkage affects urban ecological resilience, as well as its potential non-linear characteristics, remain insufficiently explored [25]. Specifically, existing studies exhibit two main limitations. First, population change is often treated merely as an external driving force or a control variable, rather than being conceptualized as a core determinant of urban ecological resilience, which constrains the ability to explain the ecological consequences of demographic change under China’s rapid urbanization and structural transformation. Second, whether the impact of population shrinkage on urban ecological resilience is non-linear, and how this relationship is contingent upon contextual factors such as the level of economic development, has yet to be subjected to systematic empirical examination.

The Yangtze River Economic Belt spans eastern, central, and western China and occupies a pivotal position in China’s national regional development strategy. Data from national population censuses show that 51 out of 110 major cities in the YREB experienced population shrinkage between 2000 and 2020, accounting for 46.36% of the total [26]. Given the representative demographic dynamics and strategic importance of cities along the YREB, this study takes the Yangtze River Economic Belt as the research area to examine the spatiotemporal evolution of population shrinkage and urban ecological resilience, and to identify their non-linear relationships. This study makes three main contributions. First, it enriches the theoretical framework of urban ecological resilience by incorporating population shrinkage into the core analytical mechanism shaping resilience formation. Second, it advances understanding of the non-linear population–ecology relationship by revealing the context-dependent role of economic development levels in mediating the effects of population shrinkage on ecological resilience. Third, it provides empirical evidence and policy-relevant insights to inform sustainable urban development strategies under conditions of population shrinkage, not only for the Yangtze River Economic Belt, but also for other regions worldwide facing similar demographic transitions.

2. Theoretical Framework

2.1. Impacts of Population Shrinkage on Urban Ecological Resilience

A large population can provide cities with abundant labor, fiscal revenues, and innovative capacity, thereby supporting investment in green infrastructure and ecological governance and enhancing cities’ ability to withstand external shocks. However, rapid population growth may also lead to excessive resource consumption, disorderly spatial expansion, and the accumulation of environmental pressures, resulting in heightened ecological vulnerability. By contrast, when population size begins to shrink, reduced resource demand, lower pollutant emissions, and decreased land-use intensity may, at certain stages, alleviate urban ecological pressures and enhance the recovery and adaptive capacity of urban ecosystems [6,27]. However, as population shrinkage intensifies, its adverse effects gradually become more pronounced. Persistent population outflows result in labor shortages and a contraction of consumer markets, weakening the endogenous growth momentum of the economic system and thereby reducing the capacity of urban systems for self-recovery and sustainable development [7]. Meanwhile, declining fiscal revenues constrain municipal investment in environmental governance, the maintenance of green infrastructure, and ecological conservation, further undermining the self-repair and long-term operational capacity of urban ecosystems [28]. In addition, population shrinkage may lead to infrastructure underutilization, reduced resource-use efficiency, and the degradation of urban service systems, rendering urban ecosystems more vulnerable when confronted with sudden shocks or prolonged environmental pressures [29]. Based on the above analysis, the impact of population shrinkage on urban ecological resilience is likely to exhibit nonlinear characteristics. Accordingly, this study proposes the following hypothesis:

Hypothesis 1.

There exists a nonlinear relationship between population shrinkage and urban ecological resilience.

2.2. Threshold Effects of Regional Economic Development Level

The level of regional economic development—reflected in the maturity of industrial structure, fiscal capacity, and innovation support—determines whether cities can sustain ecological governance investment and maintain system adaptability in the face of demographic change [30]. In regions with relatively low levels of economic development or those at an early stage of structural transformation, moderate population shrinkage may facilitate the exit of resource-intensive industries, alleviate environmental pressures, and promote economic restructuring through the release and reallocation of factors such as land and capital. Such resource reallocation effects can enhance the efficiency of ecological governance and the restoration of ecosystem functions, thereby strengthening urban ecological resilience [25]. By contrast, in regions with higher levels of economic development, well-established industrial systems and stronger fiscal capacity can help buffer the adverse effects of population shrinkage. Through sustained investment in green infrastructure, the promotion of high-quality economic transformation, and improvements in governance capacity, such regions are better able to maintain—and potentially enhance—the adaptive and recovery capacity of urban ecosystems [31]. However, this buffering effect has inherent limits under conditions of severe population shrinkage. Once population loss exceeds a certain threshold, economic performance and governance capacity may still experience systemic shrinkage, ultimately leading to a reduction in ecological resilience [8]. In sum, the level of regional economic development partly determines both the direction and magnitude of the impact of population shrinkage on ecological resilience. Accordingly, this study proposes the following hypothesis:

Hypothesis 2.

The regional economic development level exerts a threshold moderating effect on the relationship between population shrinkage and ecological resilience.

3. Research Methods and Data Sources

3.1. Description of the Study Area

The Yangtze River Economic Belt spans eastern and western China, encompassing 11 provinces and municipalities: Shanghai, Jiangsu, Zhejiang, Anhui, Jiangxi, Hubei, Hunan, Chongqing, Sichuan, Guizhou, and Yunnan (Figure 1). Covering approximately 2.0523 million square kilometers—21.4% of China’s total land area—it is among the core regions driving the nation’s high-quality economic development. As a key locus of national strategic initiatives and an important economic corridor, the economic belt both fosters regional connectivity and safeguards the ecology. According to 2022 figures, the economic belt has a permanent population of 608 million, accounting for 43.1% of China’s total population, with a regional gross domestic product (GDP) of 55.98 trillion yuan, or 46.5% of the national total. However, rapid industrialization and urbanization have intensified regional resource and environmental constraints, worsening the tension between ecological conservation and urban development. Thus, urban ecological resilience has garnered increasing attention [24]. Based on natural geographical divisions, the study area is divided into three sections: the lower reaches (Shanghai, Jiangsu, Zhejiang, and Anhui), middle reaches (Jiangxi, Hubei, and Hunan), and upper reaches (Chongqing, Sichuan, Yunnan, and Guizhou).

3.2. Research Methodology

3.2.1. Development of the Urban Ecological Resilience Indicator System

The concept of resilience has its roots in the physics field, where it is defined as a system’s capacity to return to its initial state following external disturbances [32]. In the 1970s, ecologist Holling introduced resilience into ecological research [33]. He defined it as the ability of an ecosystem to return to its original state or maintain its functional and structural stability after external disturbances. With research progress, resilience has gradually expanded from natural ecology to social ecology, and further into urban studies, giving rise to a new theoretical framework for studying the adaptability and sustainability of complex urban systems. Currently, the academic community generally defines urban resilience as the capacity of a city to quickly recover and maintain normal functioning following stress, shocks, or challenges [34]. Moreover, urban ecological resilience, a critical subsystem, is garnering extensive attention from scholars across various fields. From the perspective of resilience science, this study uses the pressure-state-response (PSR) model to objectively evaluate urban ecological resilience in the economic belt. The PSR model, developed by the Organization for Economic Co-operation and Development and the United Nations Environment Programme, follows a logical framework in which human production and livelihood activities exert pressure on environmental systems, engendering changes in ecological conditions, to which society then responds [35]. The model is widely applied across various research domains, including ecological security assessment [36], sustainability evaluation of fishery resources [37] and research on farmland abandonment [38]. Additionally, some scholars have integrated this model into studies on urban resilience [13,39].

The mechanism of urban ecological resilience is illustrated in Figure 2. Based on the Pressure–State–Response (PSR) framework, this study characterizes the dynamic interactions within urban ecological systems across three dimensions—pressure, state, and response—reflecting the interplay between external disturbances, internal conditions, and governance feedbacks. Within the pressure dimension, per capita industrial wastewater discharge captures the ecological stress imposed by industrial effluent on urban ecosystems. Per capita sulfur dioxide emissions and per capita industrial soot and dust emissions represent air pollution pressures arising from industrial production activities, collectively reflecting the primary exogenous disturbances borne by urban ecological systems. The state dimension reflects the environmental foundations and carrying capacity of urban ecological systems under existing pressures. Specifically, the green coverage rate in built-up areas represents the self-purification capacity of the urban environment, while per capita park green space area captures the basic conditions of ecological space provision and environmental conservation within cities and per capita built-up area reflects urban land resource allocation and spatial carrying capacity, and higher values imply greater flexibility for ecological space allocation and environmental governance, thus constituting a positive indicator [40]. In the response dimension, this study focuses on cities’ governance and regulatory capacities when confronting ecological pressures and environmental changes. The harmless treatment rate of domestic waste reflects the level of environmental remediation and pollution control; the comprehensive utilization rate of general industrial solid waste indicates the city’s capacity for integrated ecological resource utilization; and the centralized treatment rate of wastewater treatment plants captures the adequacy of infrastructure provision and governance capacity in responding to pressures such as wastewater discharge. Urban ecological resilience is gradually formed through the continuous interaction of pressure–state–response processes. External pressures affect urban resilience by disturbing the state of the ecological system, while the system’s state directly determines the city’s capacity to respond to and buffer such pressures. In turn, governance responses—through institutional interventions and technological inputs—feed back into the pressure transmission pathways and, to some extent, reshape the structural configuration of the urban ecological system. These three dimensions interact dynamically and jointly shape urban ecological resilience (

Table 1. Evaluation index system for urban ecological resilience.

Target Level	Primary Indicator	Secondary Indicator	Tertiary Indicator	Unit	Properties
Urban ecological resilience	Pressure	Wastewater discharge	Per capita industrial wastewater discharge	t	−
		Industrial waste	Per capita sulfur dioxide emissions	t	−
		Environmental pollution	Per capita industrial soot and dust emissions	t	−
	State	Self-purification	Green coverage rate in built-up areas	%	+
		Environmental conservation	Per capita park green space area	km2	+
		Land resource	Per capita built-up area	km2	+
	Response	Environmental remediation	Harmless treatment rate of household waste	%	+
		Resource recycling	Comprehensive utilization rate of general industrial solid waste	%	+
		Sewage treatment	Centralized treatment rate of wastewater at sewage treatment plants	%	+

).

3.2.2. Entropy Weight-TOPSIS Method

The entropy weight method determines the weight of each indicator based on the degree of data dispersion, ensuring objectivity and rationality in weight allocation. Building on this foundation, we introduce the TOPSIS method to identify alternatives by measuring their relative distance from the optimal and worst solutions [41]. The specific steps are as follows:

The first step involves constructing the original data matrix X:

$$X=\left[\begin{array}{cccc}{x}_{11}& x_{12}& \dots & x_{1m}\\ x_{21}& x_{22}& \dots & x_{2m}\\ ⋮& ⋮& \ddots & ⋮\\ x_{n1}& x_{n2}& \dots & x_{nm}\end{array}\right]$$

(1)

where X presents the data matrix, n is the number of cities, and m is the number of indicators.

The second step involves eliminating potential dimensional effects among the data by applying a standardization process, resulting in the following standardized matrix:

$$\mathrm{Positive} \mathrm{indicator}: X_{ab}={\displaystyle \frac{y_{ab}-y_{ab\min }}{y_{ab\max }-y_{ab\min }}}$$

(2)

$$\mathrm{Negative} \mathrm{indicator}: X_{ab}={\displaystyle \frac{y_{ab\max }-y_{ab}}{y_{ab\max }-y_{ab\min }}}$$

(3)

where X_ab denotes the standardized value of city a for indicator b, and y_abmin and y_abmax represent the minimum and maximum values of indicator b for a city, respectively. To avoid the occurrence of negative or zero values during the dimensionless normalization process, a small constant (0.0001) is added to the results of the above two equations. Accordingly, the adjusted value is defined as Y_ab = Z_ab + 0.0001.

The third step is to calculate the proportion of the indicator b for the city a within that indicator, denoted as P_ab:

$$P_{ab}=Y_{ab}/{\displaystyle \sum _{a=1}^m{Y}_{ab}}$$

(4)

The fourth step is to calculate the information entropy of the indicator b for the city a, denoted as e_b:

$${\mathrm{e}}_b=-{\displaystyle \frac{1}{\ln m}}\times {\displaystyle \sum _{i=1}^m{P}_{ab}\times \ln }{P}_{ab}$$

(5)

The fifth step is to calculate the weight of each indicator, denoted as W_b:

$$W_b=1-e_b/{\displaystyle \sum _{b=1}^n(1-e_b)}$$

(6)

The sixth step is to construct the weighted matrix. Each column of the standardized matrix Z is multiplied by its corresponding indicator weight W_b:

$$Z^{\ast }=\left[\begin{array}{cccc}{\omega }_1\ast z_{11}& {\omega }_2\ast z_{12}& \dots & {\omega }_m\ast z_{1m}\\ {\omega }_1\ast z_{21}& {\omega }_2\ast z_{22}& \dots & {\omega }_m\ast z_{2m}\\ ⋮& ⋮& \ddots & ⋮\\ {\omega }_1\ast z_{n1}& {\omega }_2\ast z_{n2}& \dots & {\omega }_m\ast z_{nm}\end{array}\right]$$

(7)

The seventh step is to determine the positive ideal solution (PIS) and negative ideal solution (NIS). The PIS is composed of the maximum values of each column in the weighted standardized matrix, while the NIS is composed of the minimum values of each column:

$$\mathrm{Positive} \mathrm{ideal} \mathrm{solution}: Z^{+}=(Z_1^{+},Z_2^{+},\dots ,Z_m^{+})$$

(8)

$$\mathrm{Negative} \mathrm{ideal} \mathrm{solution}: Z^{-}=(Z_1^{-},Z_2^{-},\dots ,Z_m^{-})$$

(9)

The eighth step is to calculate the Euclidean distance between each object and the PIS and NIS:

$$D_a^{+}=\sqrt{{\displaystyle \sum _{a=1}^n{(Z_a^{+}-Z_{ab})}^2}},D_a^{-}=\sqrt{{\displaystyle \sum _{a=1}^n{(Z_a^{-}-Z_{ab})}^2}}$$

(10)

Finally, a comprehensive evaluation index U_ab is calculated to determine the relative performance of each city:

$$U_{ab}={\displaystyle \frac{D_a^{-}}{D_a^{+}-D_a^{-}}}$$

(11)

where U_ab represents the overall score of the city a, ranging between 0 and 1. A higher U_a value signifies stronger urban ecological resilience, whereas a lower value indicates weaker resilience performance.

3.2.3. Population Shrinkage Indicator

Existing studies generally tend to define population shrinkage as a phenomenon in which regions with relatively independent socioeconomic systems experience a sustained shrinkage in total population, labor force, and related demographic scales over a given period [42]. Building on previous studies that define the concept of population shrinkage, we use the annual average population change rate over a specified period within the study unit as the measurement indicator. This variable is used to identify areas of population shrinkage and assess the degree of shrinkage. Following common practice and China’s five-year planning cycle, a five-year period is adopted to capture sustained demographic change while reducing the influence of short-term fluctuations in population statistics. The calculation formula is as follows [43]:

$$R_{(t_1,t_2)}=\sqrt[n]{{\displaystyle \frac{P_{t_2}}{P_{t_1}}}}-1$$

(12)

where t₁ denotes the initial year of the study, t₂ represents the final year, and n is the number of years between t₁ and t₂, which is 5 in this study. P_t1 and P_t2 denote the resident population of the study area in years t₁ and t₂, respectively. P_(t1,t2) represents the average annual rate of population change over the period from t1 to t₂. When P_(t1,t2) < 0, population shrinkage has occurred during this period, with the absolute value of P_(t1,t2) reflecting the severity of the shrinkage. The larger the absolute value, the greater the extent of population shrinkage. Conversely, when P_(t1,t2) ≥ 0, no trend of population shrinkage in the study area has been seen over the given period.

3.2.4. Threshold Effect Model

(1)

Selection of variables for the threshold effect model

Amid China’s falling fertility rates and regional population migration, population shrinkage has become increasingly pronounced, exerting profound long-term effects on urban ecological development. GDP per capita is a widely used indicator representing the level of regional economic development and reflecting differences in development stages, fiscal capacity, and industrial maturity across cities. Therefore, this study sets population shrinkage as the core explanatory variable to examine its influence on urban ecological resilience. Urban ecological resilience is selected as the dependent variable, and regional economic level serves as the threshold variable, represented by per capita GDP.

To mitigate estimation bias resulting from omitted variables, we incorporate a series of control variables into the model. Drawing from relevant literature, we select the urbanization level, governmental regulatory capacity, industrial upgrading, technological investment, and degree of openness to external markets as control variables.

(1) Urbanization level [44]: Urbanization drives urban development through resource concentration and spatial restructuring while simultaneously putting pressure on urban ecosystems. This study uses the proportion of the permanent urban population as a representative measure. (2) Government regulatory capacity [45]: Government regulatory capacity is a key factor in urban resource allocation and governance system functioning. In this study, we use government fiscal expenditure as a percentage of GDP as a proxy for government regulatory capacity. It should be noted that this indicator reflects the overall fiscal regulatory capacity of the government, whose effect primarily operates indirectly by influencing the level of public services and governance effectiveness, thereby shaping the response capacity of urban ecological resilience. (3) Advanced industrial structure [21]: The industrial structure acts as a critical nexus between economic activities and the ecological environment. In this study, the level of industrial advancement is signified by the ratio of tertiary sector output to secondary sector output. (4) Technological investment [24]: Enhancing technological innovation creates new avenues for urban ecological conservation and the minimization of resource consumption. This study quantifies technological investment using the proportion of government expenditure on science and technology relative to regional GDP. (5) Degree of openness to external markets [46]: The degree of openness reflects the extent and intensity of a city’s interactions with external environments, influencing resource flows, environmental carrying capacity, and ecosystem stability. This study uses the ratio of foreign investment to regional GDP as a proxy for openness.

(2)

Model assumptions

The threshold effect model (i.e., the threshold regression model), introduced by Hansen [47], is used to analyze nonlinear relationships between variables. Its core premise is that the effect of an explanatory variable on the dependent variable sees a structural shift once a specific critical value (threshold) has been attained. In this study, the threshold effect model is used to examine the relationship between the degree of population shrinkage and urban ecological resilience. The model is built using regional economic development levels as the threshold variable, while the urbanization level, government regulatory capacity, industrial advancement, technological investment, and openness to foreign trade are control variables. The model assumptions are as follows:

$$\begin{array}{c}UER=\alpha +{\beta }_{1}P{S}_{it}\cdot I(ED{L}_{it}<{\gamma }_1)+{\beta }_{2}P{S}_{it}\cdot I({\gamma }_1\le ED{L}_{it}<{\gamma }_2)+\dots +\\ {\beta }_{nt}P{S}_{it}\cdot I(ED{L}_{it}\ge {\gamma }_n)+\rho X_{it}+u_i+{\epsilon }_{it}\end{array}$$

(13)

where I(·) represents the threshold function, which takes values of either 1 or 0. The expression in the parentheses is the threshold variable. That is, when the specified condition has not been met, the function has a value of 0, whereas when the condition has been satisfied, it has a value of 1. The parameters γ₁, γ₂, and γ_n are the threshold values. The dependent variable, UER, represents urban ecological resilience, while the core explanatory variable, PS, refers to population shrinkage. The threshold variable, EDL, captures the regional level of economic development. The coefficients β₁, β₂, and β_n measure the explanatory power or effect intensity of population shrinkage on urban ecological resilience under different threshold conditions. Additionally, ρ denotes the coefficient of the control variable, μi denotes individual fixed effects, and ε_it follows an independently and identically distributed structure. To mitigate the influence of individual fixed effects μ_i, we ensure that the threshold effect model undergoes a within-group mean transformation. Subtracting this mean-adjusted equation from the original yields the following revised formulation:

$$\begin{array}{c}{Y}_{it}^{*}={\beta }_{1}P{S}_{it}^{*}\cdot I(ED{L}_{it}<{\gamma }_1)+{\beta }_{2}P{S}_{it}^{*}\cdot I({\gamma }_1\le ED{L}_{it}<{\gamma }_2)+\dots +\\ {\beta }_{nt}P{S}_{nt}^{*}\cdot I\left(ED{L}_{it}\ge {\gamma }_n\right)+\rho X_{it}+{\epsilon }_{it}^{*}\end{array}$$

(14)

The threshold value and the squared residuals are computed for Equation (14). Subsequently, an F-test is completed to identify the statistical significance of the threshold effect. The null hypothesis underlying the threshold effect is as follows:

$$H_o:{\beta }_1={\beta }_2=\dots ={\beta }_{nt}$$

(15)

First, we substitute the indicator data into Equation (15) to test for the presence of a threshold effect. If the null hypothesis is not rejected, there is no threshold effect; otherwise, a threshold effect is present. Next, we conduct an Likelihood Ratio test on the threshold value to evaluate its statistical significance. The p-value is calculated through the Bootstrap method, and the threshold effect is verified as significant only if the p-value reaches statistical significance, thus communicating whether the threshold value is within a confidence interval. Last, we compare γ_n with different threshold values to identify the extent of the core explanatory variable’s effect on the dependent variable.

3.3. Data Sources

Factoring in the timing of policy implementation, as well as data continuity and availability, we select 110 cities across 11 provinces in the economic belt as the research sample. The relevant data are mainly pulled from the China Urban Statistical Yearbook, the China Urban Construction Statistical Yearbook, and the statistical yearbooks and social development reports of individual cities for the corresponding years. For years with certain indicators missing, data are supplemented using linear interpolation.

4. Results

4.1. Temporal and Spatial Evolution Characteristics of Urban Ecological Resilience

4.1.1. Temporal Trends

A statistical analysis is performed on the overall level of urban ecological resilience across cities in the economic belt from 2012 to 2021 (Figure 3). Overall, we observe an increase in the average urban ecological resilience from 0.527 in 2012 to 0.604 in 2021, translating to an annual growth rate of 0.77%. This steady upward trend reflects a continuous enhancement of urban ecological resilience, evidencing an increased capacity of cities to resist both internal and external ecological pressures. From a regional distribution perspective, urban ecological resilience along the Yangtze River Economic Belt exhibits a clear spatial pattern of lower reaches > middle reaches > upper reaches, reflecting differences in natural endowments, economic development, and industrial structure. Cities in the lower reaches demonstrate significantly higher resilience, supported by a strong economic base, advanced technological capacity, and well-developed infrastructure, which enhance their ability to respond effectively to environmental and socio-economic uncertainties. In contrast, cities in the middle reaches face ecological constraints arising from the relocation of high-pollution, high-resource-consumption industries, which intensifies environmental pressure and limits resilience. Meanwhile, cities in the upper reaches, characterized by complex topography, ecological sensitivity, limited economic development, and weaker infrastructure, exhibit lower resilience due to reduced capacity for ecological risk prevention and adaptive management.

4.1.2. Spatial Variation

Selecting 2012 and 2021 as representative time points, urban ecological resilience levels were classified into five categories using the natural breaks method in ArcGIS 10.8, low (U < 0.45), medium–low (0.45 ≤ U < 0.50), medium (0.50 ≤ U < 0.55), medium–high (0.55 ≤ U < 0.60), and high (U ≥ 0.60), and a spatiotemporal variation map was generated (Figure 4). Overall, urban ecological resilience along the Yangtze River Economic Belt exhibits a “high in the east, low in the west” spatial pattern, with resilience levels shifting from low to medium–high over time and regional disparities gradually narrowing. In 2012, low and medium–low resilience areas were primarily located in the upper and southern middle reaches, including Bijie, Baoshan, and Zhaotong, whereas medium–high resilience areas were concentrated in the lower reaches, such as Yichun, Fuzhou, and Xuancheng, reflecting the spatial imbalance in economic development, resource allocation, and ecological governance capacity. By 2021, resilience in the upper and middle reaches had improved markedly, with low-resilience areas transitioning to medium and medium–high categories, indicating progress in ecological governance and adaptive capacity in central and western regions. High-resilience areas remained concentrated along the downstream “Ganzhou–Xuzhou” corridor, with resilience levels further strengthened.

4.2. Spatiotemporal Evolution Characteristics of Population Shrinkage

4.2.1. Temporal Variation

Following relevant literature [43], this study uses an annual population decrease of 1% as the threshold to categorize population change into three types: severe shrinkage (R ≤ −1%), mild shrinkage (−1% < R < 0%), and non-shrinkage (R ≥ 0%). From a temporal perspective, population shrinkage in the Yangtze River Economic Belt has intensified over time (

Table 2. Number of population shrinkage units in different periods in the economic belt.

Type	2012–2016				2017–2021
	Overall	Upstream Region	Midstream Region	Downstream Region	Overall	Upstream Region	Midstream Region	Downstream Region
Mild shrinkage	24	11	7	6	33	6	17	33
Severe shrinkage	24	7	7	10	39	18	10	24
Total	48	18	14	16	72	24	27	48

). Between 2012 and 2016, 48 cities experienced population shrinkage, accounting for 43.64% of the sample. In the period of 2017–2021, the number of cities experiencing population shrinkage increased to 72, representing 65.45% of the sample, with severely declining cities rising from 24 to 39, indicating a significant aggravation of population shrinkage.

4.2.2. Spatial Variation

From a spatial distribution perspective, between 2012 and 2021, cities across the upper, middle, and lower reaches of the Yangtze River Economic Belt experienced varying degrees of population shrinkage (Figure 5). During 2012–2016, population shrinkage was primarily concentrated in the upper and middle reaches, with relatively few cities in the lower reaches affected. In contrast, during 2017–2021, both the extent and intensity of population shrinkage increased markedly. Severely shrinking areas were mainly clustered around urban agglomerations, where factors such as insufficient economic vitality, delayed industrial transformation, and outmigration exacerbated population loss. Lightly shrinking areas were more dispersed, predominantly located in the middle reaches, where cities faced moderate net population outflows without severe shrinkage. Areas experiencing no population shrinkage were largely concentrated in the economically developed eastern coastal regions and some central cities, which continued to attract population growth due to higher economic development levels, advantageous geographical locations, and well-established infrastructure. Overall, population shrinkage is more pronounced in small and medium-sized cities, whereas large cities rarely experience such shrinkage, reflecting the relative advantage of major cities in attracting and retaining population.

4.3. Impacts of Population Shrinkage on Urban Ecological Resilience

4.3.1. Threshold Effect Model Test

Given the complex relationship between population shrinkage and urban ecological resilience, this study introduces regional economic development as a threshold variable. Using a panel threshold effect model, we examine the nonlinear moderating role of population shrinkage on urban ecological resilience. The Bootstrap resampling method, with 1000 iterations, is used to test the threshold model. The results are presented in

Table 3. Threshold effect test of population shrinkage on urban ecological resilience.

Threshold Variable	BS Count	Threshold Type	p-Value	Threshold Value	95% Confidence Interval of the Threshold	Critical Value
						10%	5%	1%
EDL	1000	Dual threshold	0.0320	4.5396	[4.5280, 4.5470]	15.8795	19.4562	28.0396
	1000	Triple threshold	0.0475	4.5948	[4.5865, 4.6101]	15.7971	19.5250	27.9416

. To avoid potential multicollinearity and endogeneity among the selected control variables, the relevant variables were logarithmically transformed. Accordingly, the threshold values reported in Table 3 correspond to the log-transformed regional economic development level (EDL). If the test result passes, subsequent dual-threshold and triple-threshold tests are conducted in sequence. The results indicate that the study passed the double-threshold test, identifying two significant thresholds (γ₁ = 4.540, γ₂ = 4.595), based on which the sample was divided into three economic development regimes. In the threshold regression, the 120 samples are distributed across these regimes as follows: 25% in the low regime (EDLit ≤ γ₁), 40% in the medium regime (γ₁ < EDLit ≤ γ₂, including 48 samples), and 35% in the high regime (EDLit > γ₂), providing sufficient data support. These findings demonstrate that regional economic development exhibits a double-threshold effect in moderating the impact of population shrinkage on urban ecological resilience.

4.3.2. Threshold Regression Results

As the degree of population shrinkage intensifies, its effect on urban ecological resilience shows a significant interval effect, as shown in

Table 4. Panel threshold regression of population shrinkage on urban ecological resilience.

Threshold Reversion Effect	Results for the Full Sample
Single threshold value	4.540
Double threshold value	4.594
UER (EDLit ≤ γ1)	0.505 (0.166)
UER (γ1 < EDLit ≤ γ2)	0.616 ** (0.049)
UER (EDLit > γ2)	−0.502 ** (0.013)
_cons	0.255 (0.098)
R2	0.630
F test	F (8109) = 32.9 Prob > F = 0.000

Note: ** represent statistical significance at the 0.01 levels, respectively.

. When regional economic development falls below the threshold of 4.540, an increase in population shrinkage tends to enhance urban ecological resilience, albeit not to a statistically significant extent. This result suggests that a certain economic threshold must be reached before population shrinkage can have a positive influence on urban ecological resilience. When the regional economic level is between 4.540 and 4.595, a 1% increase in population shrinkage corresponds to a 0.616% improvement in urban ecological resilience. However, when the regional economic level exceeds 4.595, a 1% increase in population shrinkage results in a 0.502% shrinkage in urban ecological resilience. Overall, the relationship between population shrinkage and urban ecological resilience along the economic belt is not linear but demonstrates a significant “inverted U-shaped” pattern, influenced by varying levels of economic development. Therefore, this study confirms that Hypothesis 1 and Hypothesis 2 are supported.

5. Discussion

Amid ongoing population shrinkage, cities face challenges including resource reallocation, land-use changes, and infrastructure adjustments, placing higher demands on the stability and adaptability of urban ecosystems. To identify the effect of population shrinkage on urban ecological resilience, this study, based on the resilience theory framework, uses the PSR model to quantify and evaluate the level of urban ecological resilience in the economic belt and its spatiotemporal variation characteristics. Moreover, a threshold regression model is introduced to explore the nonlinear impact mechanisms of population shrinkage on urban ecological resilience.

5.1. Spatiotemporal Dynamics of Population Shrinkage and Urban Ecological Resilience

Over the past decade, population shrinkage in the economic belt has exhibited a trend of temporal intensification and spatial expansion. From 2012 to 2016, our data show that areas experiencing population shrinkage are primarily concentrated in the upper and middle reaches of the Yangtze River, with relatively fewer cases in the lower reaches. However, between 2017 and 2021, the extent of population shrinkage expands significantly, with severe shrinkage zones predominantly located around urban agglomerations, particularly in resource-based cities and regions with insufficient economic vitality. Contrastingly, non-shrinking areas are mainly found in economically developed eastern coastal cities and key central cities. This result aligns with Liu’s [48] research on the spatiotemporal evolution of population shrinkage in China, which also indicates an accelerating trend of population shrinkage, particularly in the middle and lower reaches of the Yangtze River, where shrinkage zones have expanded markedly. However, this study emphasizes the regional differentiation of population shrinkage within the economic belt. By employing a more granular temporal framework, it offers a detailed analysis of how population shrinkage patterns evolve in specific regions at different time points.

From the perspective of urban ecological resilience, there is an increasing trend of the overall resilience level of cities in the economic belt from 2012 to 2021, along with a gradual reduction in regional disparities. Regarding spatial distribution, there is a notable “higher in the east, lower in the west” pattern, with low-resilience areas progressively transitioning into medium-to-high-resilience zones. Notably, the resilience level in the upper reaches of the Yangtze River shows a significant improvement. Compared with existing studies, this research employs the PSR model to assess urban ecological resilience, further revealing its dynamic evolution and regional disparities. Additionally, by emphasizing improvements in the middle and upper reaches of the Yangtze River, this study supplements the existing literature and emphasizes progress in regional coordination and balanced governance.

5.2. Effect of Population Shrinkage on Urban Ecological Resilience

Traditionally, population shrinkage has been associated with economic decline and the weakening of urban functions. However, recent studies increasingly suggest that population shrinkage does not necessarily imply stagnation in urban development [33]. Across global urban systems, population shrinkage has emerged as a structural phenomenon affecting both developed and emerging economies, with its ecological and governance consequences exhibiting pronounced context-specific variations [49]. This study finds that the effects of population shrinkage on urban ecological resilience exhibit pronounced stage-specific characteristics under different levels of economic development. Specifically, in cities with a moderate level of economic development, population shrinkage can create favorable conditions for enhancing urban ecological resilience through resource reallocation, optimization of land-use structure, and industrial restructuring. However, once the level of economic development further increases and surpasses a critical threshold, the effect of population shrinkage on urban ecological resilience shifts from being promotive to inhibitory. These findings indicate that the ecological effects of population shrinkage do not accumulate linearly but instead depend on a city’s development stage and institutional context. This pattern is consistent with international studies documenting nonlinear relationships among city size, economic maturity, and environmental performance [50,51].

From the perspective of city type, the coexistence of high economic development and population shrinkage is not an isolated phenomenon. It is primarily observed in cities with relatively mature economic systems that are undergoing industrial restructuring or transformation. Cities such as Wuhu, Panzhihua, and Jingmen exhibited both high levels of economic development and notable population shrinkage during the sample period. These cities typically maintain relatively high per capita economic output; however, factors such as industrial upgrading, rising living costs, and the siphoning effects of neighboring central cities contribute to ongoing population outflows. In this context, population shrinkage negatively affects urban ecological resilience by intensifying fiscal pressures, underutilizing infrastructure, and weakening environmental governance capacity. Therefore, identifying and overcoming critical thresholds is crucial for transforming population shrinkage into an opportunity for enhancing urban ecological resilience.

This finding not only supports existing theoretical frameworks regarding the impacts of population shrinkage on resources and the environment but also highlights the mechanistic role of regional economic development level as a key moderating factor. The enhancement of urban ecological resilience depends not only on economic growth itself but also on the adaptive transformation of resource allocation, spatial structure adjustment, and governance capacity under the context of demographic change. These insights provide valuable policy implications for shrinking cities worldwide in terms of ecological governance and the design of sustainable transition pathways.

6. Conclusions

This study, based on panel data from cities in the economic belt from 2012 to 2021, has constructed a comprehensive urban ecological resilience evaluation index within the PSR model framework to evaluate the spatiotemporal differentiation of population shrinkage and urban ecological resilience. Additionally, a threshold effect model is used to evaluate the relationship between population shrinkage and urban ecological resilience. The key findings are as follows:

(1)

Urban Ecological Resilience: The overall level of urban ecological resilience in the economic belt shows a gradual improvement, albeit with significant regional disparities, exhibiting a spatial pattern of “higher resilience in the east and lower resilience in the west.” Moreover, the downstream region demonstrates the highest resilience, benefiting from a developed economy, abundant resources, and advanced technological conditions. The midstream region reflects a moderate level of resilience owing to the cumulative effects of industrial pollution and ecological pressures. Contrastingly, the upstream region, characterized by high ecological sensitivity and insufficient infrastructure, shows relatively weaker resilience.

(2)

Population Shrinkage Trends: The extent of population shrinkage in the economic belt shows a trend of increasing intensity, with its spatial scope expanding. Severe population shrinkage is mainly concentrated in upstream resource-based cities and economically underdeveloped areas. Contrastingly, economically developed downstream regions and central cities still experience population growth, reflecting a broader trend of population and resource concentration in large urban centers.

(3)

Effect of Population Shrinkage on Urban Ecological Resilience. The influence of population shrinkage on urban ecological resilience shows an inverted U-shaped nonlinear pattern, moderated by regional economic development levels. In areas with low economic development, the positive effects of population shrinkage on ecological resilience are insignificant. At moderate economic development levels, population shrinkage improves ecological resilience through resource optimization and economic restructuring. However, at high economic development levels, population shrinkage leads to economic stagnation, infrastructure underutilization, and a decline in ecological governance capacity, significantly weakening ecological resilience.

Based on the spatiotemporal differentiation characteristics of urban ecological resilience and population shrinkage in the economic belt, as well as the nonlinear relationship between them, cities must formulate appropriate population and ecological management policies customized for regional characteristics and stages of economic development. Given the broader context of population shrinkage, it is essential to harness its potential benefits while mitigating its adverse effects. To promote the coordinated development of demographic changes and urban ecological resilience and to advance the construction of sustainable cities, we propose the following policy recommendations. (1) Region-specific strategies to address population shrinkage and enhance urban ecological resilience—To effectively respond to population shrinkage and sustain improvements in urban ecological resilience, policymakers should implement measures tailored to the economic conditions of different regions. (2) Strengthening regional cooperation and resource flows to promote coordinated urban development amid population shrinkage—To foster coordinated development in the economic belt under the pressures of population shrinkage, policymakers should reinforce cooperative linkages between the upper, middle, and lower reaches of the region and create a transregional joint prevention and control system for ecological protection. (3) Smart empowerment for urban development: addressing the challenges of population shrinkage—In the context of the accelerating trend of population shrinkage and the wave of digital transformation, the development of smart cities poses a crucial avenue for enhancing urban ecological resilience. By leveraging digital platforms and big data technologies, we can establish a comprehensive monitoring system incorporating key elements such as population dynamics, resource utilization, and ecological conditions.