Spatial Association Networks and Factors Influencing Ecological Security in the Yellow River Basin

Abstract

The Yellow River Basin (YRB) is an important ecological security barrier in China, playing an irreplaceable role in soil and water conservation, climate regulation, and biodiversity maintenance, and it is related to the stability and security of the ecosystem. Exploring the spatial correlation networks and factors influencing ecological security in the YRB can provide new ideas for cross-domain collaborative governance, promote efficient cooperation among regions, and optimize resource allocation. Using a quantitative approach to assess the YRB’s ecological security, we employed an adjusted gravity model, social network analysis, and quadratic assignment procedure analysis to understand the spatial connection dynamics. The results indicate the following: (1) Ecological security in the YRB continued to improve from 2005 to 2019, but the overall level was low. The degree of the dispersion of the ecological security status among cities constantly increased, and there were significant regional differences in the level of ecological security in the YRB. (2) From 2005 to 2019, the number and density of network connections among cities within the YRB increased significantly, and the ecological security links gradually strengthened. The Shandong Peninsula city cluster and the Hubao–Eyu City cluster are not only located at the core of the network but also play the role of “bridge intermediary”, exhibiting strong control. (3) Among all variables, economic development and geographic proximity increased significantly in terms of their correlation with the YRB’s ecological security. The study of spatial connectivity networks and their influencing factors in the YRB provides new ideas for inter-regional collaborative governance.

1. Introduction

As economic globalization advances and humans intensify the utilization of natural resources and environments, our ecosystems face threats and damage on an unparalleled scale [1]. The goal of urbanization is to improve humans’ living environment and quality of life; however, the process of urban development has increased emissions of pollutants such as PM2.5, which poses a threat to public health [2]. The consequences of these adversities manifest as diverse ecological challenges, such as habitat destruction, dwindling biodiversity, aggravated soil degradation, and the depletion of carbon reservoirs [3,4,5]. Such repercussions alter the inherent structure, dynamics, and functioning of ecosystems [6]. Crucially, ecosystems have a bounded capacity for self-regulation. When human-induced perturbations surpass the ecosystem’s natural ability to recover, this may induce irreversible damage, jeopardizing both regional ecological security and broader sustainable developmental goals [7]. As ecological security patterns are based on the interactions between landscape patterns and ecological functions, as well as processes, they are crucial for the provisioning of ecosystem services and, thus, the maintenance of ecological sustainability [8].

Marked as the most populous developing nation, China has experienced significant strides in socio-economic realms following its period of reform and opening up. However, this rapid advancement has incurred substantial environmental costs, including natural resource overexploitation, industrial and domestic pollutant emissions, climate change impacts, and social disparities [9]. Consequently, when pursuing economic growth and social progress, environmental protection must be prioritized, with the adoption sustainable development models that balance economic and social development with ecological security. Acknowledging these pressing ecological challenges, China has integrated ecological security as a crucial component of its comprehensive national security strategy [10]. The Yellow River Basin (YRB), holding strategic importance in China’s ecological security and economic development, requires ecological security assessment to enhance regional ecological protection and promote sustainable development.

Much of the ecological security research has concentrated on assessment method innovation and refinement. These studies frequently utilize conceptual frameworks such as pressure–state–response (PSR) [2], driver–pressure–state–impact–response (DPSIR) [11], and driver–pressure–state–impact–response–management (DPSIRM) to establish ecological security evaluation structures. A wide array of techniques has been employed for these assessments, ranging from the entropy weight–TOPSIS and fuzzy object element models to the variation coefficient method [12,13]. Some have harnessed the entropy–comprehensive evaluation approach or the advanced Super-SBM model [14,15]. Analytic hierarchy processes (AHPs) have been used as well [7]. In addition, the improved ecological footprint methodology allows for a more comprehensive assessment of ecological security and an enhanced understanding of ecological conditions [16]. In terms of the research content, which mainly focuses on evaluations of ecological security, the evaluation of ecological vulnerability and the spatial and temporal patterns of ecological security have also attracted attention, with research objects including tourism [17], cities [18], grasslands [19], water [20], cultivated land [21], and forests [22]. From these studies, it can be concluded that artificial ecosystems are more vulnerable compared to natural ecosystems, such as urban ecosystems. Some scholars have also studied special areas, such as old industrial areas with severely damaged ecosystems due to various kinds of pollution [23]. To summarize, these studies evaluate ecological security at various scales, such as national, provincial, and municipal levels, from the perspective of statistical data or raster spatial data. However, there is no clear definition of urban ecological security. In this study, we define urban ecological security as the state and response of urban ecosystems under environmental–social–economic multi-source pressure. It reflects the sustainable development capability of the city. In addition, although various scholars focus on different aspects of ecological security, the underlying logic mostly starts from the idea of pressure–state–response. The PSR framework effectively depicts system linkages and demonstrates adaptability and systematic qualities, making it particularly suitable for urban ecological security assessment.

Research on ecological security’s spatial distribution patterns has employed various methods to identify trends and effects. Spatial autocorrelation techniques and hot-spot investigations have proved valuable, alongside coefficient of variation approaches and models combining standard deviational ellipse with gravity centers. Additionally, spatial econometric models, such as the Spatial Durbin Model (SDM) [18,24,25], enhance our understanding of spatial aggregation, clustering, and distribution patterns [26]. Research mainly focuses on the ecological security pattern, and research topics include urban areas [27], rural areas [28], tourism [29], and other subject that are closely linked to ecological security. Yet, measurements rooted solely in “attribute data” often only illuminate agglomeration and isolation based on geographical closeness, neglecting a more holistic regional association framework and intricate micro-links on a broader scale [16]. Social Network Analysis (SNA) effectively addresses these analytical limitations. Emerging from sociological foundations, SNA employs mathematical techniques and graph theory to quantitatively analyze “relational data.” Its primary strength lies in revealing intricate associative patterns among network participants and their attributes. Through a detailed analysis of these relationships, SNA offers valuable insights that mitigate the limitations commonly associated with solo entity research, traditional metric studies, and attribute-centric inquiries [30]. Some studies have utilized social network analysis and modified gravity modeling to examine the evolutionary characteristics and formation mechanisms of spatial correlation networks in forest ecological security, coordinated urban agglomeration development, and ecological welfare performance in China. Illustratively, some studies have used social network analysis and modified gravity modeling to investigate the evolutionary characteristics and formation mechanisms of spatial correlation networks involved in forest ecological security, the coordinated development of urban agglomerations, and ecological welfare performance in China [31]. The spatial association network abstractly represents complex ecological security relationships among cities, where node numbers, connecting edges, structural characteristics, and connectivity critically influence overall network robustness [32].

The aforementioned studies provide comprehensive assessments of ecological security and enhance our understanding, yet certain limitations persist. First, most studies focused on spatial features. This research was based primarily on geographic proximity, overlooking the evolving dynamics and spillover effects of spatial networks, which extend beyond geographical boundaries. Second, limited attention has been directed toward the determinants influencing the spatiotemporal evolution of ecological security patterns. While traditional metrics predominate current studies seeking to understand these influences and their underlying mechanisms, there remains a notable absence of research examining ecological security’s spatial association networks through network theory analysis. Third, advances in digital communication and modern transportation systems have reduced traditional geographical constraints, enabling the unrestricted movement of production components, including population, capital, knowledge, and information dynamics [33]. This inter-regional transfer transforms the relationship between economic growth, resource utilization, and ecological conservation, creating complex spatial networks. Given the increasingly interconnected nature of ecological security, examining the YRB’s spatial ecological network through a comprehensive perspective becomes essential.

In this study, the ecological security level of the YRB is measured from multiple perspectives. Considering the modified gravity model and social network analysis method, the evolution of the ecological spatial correlation network structure and its driving mechanism are studied in depth, providing a new way of thinking for the study of urban ecological security correlation. Specifically, drawing on data from 78 cities within the YRB spanning 2005–2019, a three-pronged approach was adopted. The PSR framework was employed to sculpt an index system tailored for assessing ecological security. Subsequently, the entropy weight–TOPSIS was harnessed to quantify the ecological robustness of the YRB. To delineate the spatial interconnectedness of ecological security within the YRB, a refined gravity model was utilized. Furthermore, the intricacies of the overall and individual network attributes were unraveled through the lens of SNA. Quadratic assignment procedure (QAP) analysis offered insights into how diverse factors shaped the spatial network of ecological security. In light of these findings, actionable strategies for fostering regional collaborative governance were proposed.

The innovations of this study are as follows: (1) The existing spatial metrology models often have inherent defects, and the understanding of spatial relations is relatively limited. The innovative integration of SNA in this study not only successfully overcomes these defects, but also greatly enriches the understanding dimension of spatial relations. This breaks the traditional framework for ecological security research, opens up a new research path with great potential, and expands and deepens the research perspective and methods. (2) Current studies often do not fully explore the key role of cities in the collaborative governance of ecological security. This study focuses on the evolution characteristics of the spatial network and the spatial spillover effect, emphasizes the important role of cities in the ecological security network, lays a new foundation for the ecological security collaborative governance of the YRB, fills the gap in the existing research in this respect, and provides strong theoretical support for subsequent collaborative governance practice.

2. Materials and Methods

2.1. Overview of the Research Area

Emerging from the Qinghai–Tibet Plateau, the Yellow River spans nine provinces—Qinghai, Sichuan, Gansu, Ningxia, Inner Mongolia, Shaanxi, Shanxi, Henan, and Shandong—covering an expansive 75 × 10⁴ km. The terrain of the YRB is high in the west and low in the east. As an important ecological barrier in China, the Yellow River not only provides support for regional economic development, but also plays a key role in national strategy, which is of great significance to the security and stability of China’s ecosystem. However, at present, its ecological security is facing many challenges: the ecosystem is fragile, more than three-quarters of the area is in a moderately fragile state, the ecological flow rate is low, and soil erosion and other problems have not been eradicated. Water pollution is worsening under the influence of industrial structures, especially some tributaries in the middle reaches. The situation is severe, and it is urgent to strengthen protection and management. Our study takes the nine provinces covered by the YRB as the study area and tries to ensure the completeness of the provincial administrative regions. Considering the missing statistics of some cities, 78 cities were finally identified to construct the ecological security network (Figure 1).

2.2. Data Sources

The raw data are from the statistical reports of cities in the YRB from 2005 to 2019. GDP, population, the total retail sales of consumer goods, the per capita disposable income of urban residents, the proportion of fiscal expenditure on science and technology, the proportion of fiscal expenditure on education, and the proportion of value added by the tertiary industry to GDP all come from the National Economic and Social Development Communique. The total water resources data came from the water resources bulletin. Industrial sulfur dioxide emissions, industrial wastewater emissions, industrial smoke (powder) dust emissions, and total carbon emissions data are from the energy statistical yearbook. The data on population growth rate, the annual registered unemployment rate, the number of university students, the number of doctors, the number of hospital beds, the number of electric buses, the rate of the harmless treatment of domestic garbage, the rate of the centralized treatment of urban domestic wastewater, and the rate of the comprehensive utilization of general industrial solid wastes are all from the China Urban Statistical Yearbook. The data on population density, per capita road area, urbanization rate, construction land area, and the green coverage rate of built-up areas are all from the Statistical Yearbook of Urban Construction in China. Due to the lack of carbon emission data for prefecture-level cities, this paper draws on related studies and simulates the estimation based on long time series of nighttime lighting data.

2.3. Methodology

Initially, we employed the PSR model to establish the criteria for ecological security evaluation. Subsequently, the entropy weight technique was harnessed to gauge the ecological security standards within the YRB. To comprehend the spatial and temporal fluctuations in ecological security, we integrated descriptive statistics with GIS. Next, by adapting the gravity model, we formulated a spatial connection framework, focusing on ecological security within the same basin. This framework’s overarching and individual characteristics were then dissected using SNA. Finally, the QAP analysis was executed to delve into the determinants influencing the spatial connection framework for ecological security (Figure 2). Specifically, the following sections are included.

2.3.1. Construction of Assessment System

The United Nations Environment Programme (UNEP) introduced the PSR conceptual model, which has since gained significant traction in both national and global ecological security evaluations. This diagnostic tool unravels the intrinsic causal associations within ecosystems, establishing a linkage between human endeavors and the ensuing ecological consequences [34]. Pressure delineates the immediate environmental impact of human actions, state depicts the prevailing or emerging environmental conditions, and response encapsulates human-initiated interventions addressing environmental challenges. Drawing from an extensive corpus of literature [35,36], we have tailored the PSR framework to devise an ecological security evaluation matrix specifically for the YRB, detailed in

Table 1. The assessment indicators system of ecological security.

First-Level Indicator	Second-Level Indicator	Third-Level Indicator	Measurement Units and Type	Weight
Pressure	Social pressure	Population growth rate	% (−)	0.0015
		Population density	person/km2 (+)	0.0511
		Number of registered urban unemployed at the end of the year	person (−)	0.0032
	Economic pressure	GDP growth rate	% (−)	0.0033
		GDP	CNY 108 (−)	0.0786
	Environmental pressure	Industrial sulfur dioxide emissions	t (−)	0.0044
		Industrial wastewater discharge	104 t (−)	0.0033
		Industrial smoke (dust) emissions	t (−)	0.0002
		Total carbon emissions	104 t (−)	0.0461
State	Social state	Road area per capita	m2 (+)	0.0306
		Urbanization rate	% (+)	0.0192
	Economic state	GDP per capita	CNY (+)	0.0659
		Total Social Retail Goods	CNY 108 (+)	0.1075
		Urban disposable income per capita	CNY (+)	0.0320
	Environmental state	Total water resources	108 m3 (+)	0.1430
		The area of construction land	km2 (+)	0.0658
		Greenery coverage in built-up areas	% (+)	0.0078
Response	Social response	Number of college students per 10,000 people	person (+)	0.1088
		Number of doctors per 1000 people	person (+)	0.0169
		Number of hospital beds per 10,000 people	(+)	0.0216
		Number of public electric bus per 10,000 people	(+)	0.0715
	Economic response	Proportion of fiscal expenditure on education	% (+)	0.0063
		Proportion of fiscal expenditure on science and technology	% (+)	0.0603
		Proportion of tertiary industry in GDP	% (+)	0.0141
	Environmental response	Centralized treatment rate of urban sewage	% (+)	0.0095
		Harmless treatment rate of household garbage	% (+)	0.0126
		General industrial solid waste comprehensive utilization rate	% (+)	0.0151

.

Pressure indicators use the natural population growth rate to represent the population growth trend; use population density to represent the spatial distribution of the population; and use the number of registered unemployed people in urban areas at the end of the year to represent the employment situation. Economic pressure mainly comes from the regional economic scale and the initiative of regional economic development, which are, respectively, represented by the regional gross domestic product and the growth rate of the regional gross domestic product. The total emissions of sulfur dioxide from industry, the total discharge of industrial wastewater, the total discharge of industrial smoke (dust), and the total carbon emissions are used to represent the pressure on the ecological environment caused by the pollutants generated by social and economic development.

The state indicators include per capita road area to represent the level of urban construction; urbanization rate to represent the aggregation state of the population; per capita gross domestic product to represent the overall state of the regional economy; total retail sales of consumer goods to represent the regional consumption level; and the per capita disposable income of urban residents to represent the regional living standard. Total water resources represent the state of the water-holding function of the ecological environment; the area of construction land represents the utilization of land resources; and the green coverage rate of built-up areas represents the greening status of the city.

The response indicators include the number of college students per 10,000 people to represent regional intellectual capital. The numbers of doctors and hospital beds per 1000 people represent the regional medical level. The numbers of public buses (electric vehicles) per 10,000 people represent the regional transportation situation. The proportions of education expenditure and science and technology expenditure in fiscal expenditure reflect the investment capacity of the regional government and the attention it pays to ecological security. The proportion of the tertiary industry in GDP represents the positive actions taken to optimize the industrial structure and control pollution. The centralized treatment rate of urban sewage and the harmless treatment rate of domestic waste represent the positive responses made at the social level in terms of ecological governance. The comprehensive utilization rate of general industrial solid waste represents the positive governance measures taken against industrial pollution.

The entropy weight method is used to determine the weight of each evaluation factor and the TOPSIS method is used to measure the comprehensive proximity of each evaluation object to the positive and negative evaluation objects and finally evaluate the ecological security level of the city.

2.3.2. Entropy Weight–TOPSIS

The entropy weight–TOPSIS approach builds upon the foundational TOPSIS method, integrating the entropy weight technique to objectively assign weights to each index based on its variability [37]. TOPSIS quantifies the Euclidean distances between the evaluation entity and both the optimal and the most undesirable entities. This assists in distinguishing the rank and relative quality of the entity under evaluation, leading to a holistic evaluation outcome [38]. The core computational steps involved in the entropy-weighted TOPSIS include the following:

(1)

Normalization—included to mitigate variations due to size and scale disparities among indices. The extremum approach is harnessed for standardization [39].

Positive indicator:

$$y_i={\displaystyle \frac{x_{ij}-\mathrm{m}\mathrm{i}\mathrm{n}\left(x_{ij}\right)}{\max \left(x_{ij}\right)-\mathrm{m}\mathrm{i}\mathrm{n}\left(x_{ij}\right)}}$$

(1)

Negative indicator:

$$y_i={\displaystyle \frac{\max \left(x_{ij}\right)-x_{ij}}{\max \left(x_{ij}\right)-\mathrm{m}\mathrm{i}\mathrm{n}\left(x_{ij}\right)}}$$

(2)

where $x_{ij}$ is the raw data value of the $j$ index of the $i$ city.

(2)

Entropy weighting.

$$w_j={\displaystyle \frac{1-\left[-\frac{1}{\ln n}{\sum }_{i=1}^n\left(\frac{y_i}{{\sum }_{i=1}^n{y}_i}\right)\ln \left(\frac{y_i}{{\sum }_{i=1}^n{y}_i}\right)\right]}{{\sum }_{j=1}^m\left\{1-\left[-\frac{1}{\ln n}{\sum }_{i=1}^n\left(\frac{y_i}{{\sum }_{i=1}^n{y}_i}\right)\ln \left(\frac{y_i}{{\sum }_{i=1}^n{y}_i}\right)\right]\right\}}}i=\mathrm{1,2},\dots ,n\left(2\right);j=\mathrm{1,2},\dots m;w_j\left[0,\right.\left.1\right]$$

$$r_{ij}=w_j\times y_i$$

(3)

where $W_j$ is the weight of the $j$ index, $y_i$ is the value of index of the $i$ city after pre-processing, $r_{ij}$ is the weighted value of the $j$ index of the $i$ city, $n$ is the number of cities, and $m$ is the number of evaluation indexes. It should be noted that evaluation indicators will have zero values after standardization. Therefore, in order to avoid the occurrence of zero values that cause errors in the calculation of $W_j$, a value tending towards zero is added to all evaluation index values.

(3)

The TOPSIS technique posits that the most favorable option should not only be in closest proximity, in terms of Euclidean distance, to the optimal solution but should also be the furthest from the least favorable one [40].

The positive ideal ($L^{+}$) and negative ideal ($L^{-}$) solutions, respectively, represent the optimal and worst solutions in the indicator vector. These are calculated as follows:

$$C^{+}=\mathrm{m}\mathrm{a}\mathrm{x}(r_{1m,}{r}_{2m,}\dots r_{nm,})$$

(4)

$$C^{-}=\mathrm{m}\mathrm{i}\mathrm{n}(r_{1m,}{r}_{2m,}\dots r_{nm,})$$

(5)

It is necessary to determine the Euclidean distance between each evaluation scheme and both the positive and negative ideal solutions.

$$D_i^{+}=\sqrt{\sum _{j=1}^m{\left({\mathrm{C}}^{+}-r_{ij}\right)}^2}$$

(6)

$$D_i^{-}=\sqrt{\sum _{j=1}^m{\left({\mathrm{C}}^{-}-r_{ij}\right)}^2}$$

(7)

At last, the relative closeness $T_i$ ($0\le T_i\le 1$) of the ecological security of each city in the YRB is calculated [41], and the level of ecological security of each city is expressed by $T_i$. The formula for the relative closeness $T_i$ is as follows:

$$T_i={\displaystyle \frac{D_i^{-}}{{D_i^{+}+D}_i^{-}}}$$

(8)

The value of $T_i$ is between 0 and 1. The larger the value of $T_i$ is, the closer the evaluation object is to the ideal object, and vice versa.

2.3.3. Spatial Association Network

To conduct SNA, establishing a spatial association network is vital [42]. Currently, the prevalent methods to establish this network include the gravity model and the VAR Granger causality test [43]. The latter, however, is not suited for cross-sectional data and is sensitive to lag order, potentially undermining accuracy when inscribing spillover relationships. The traditional gravity model, on the other hand, lacks directionality and solely illustrates the intensity of association [44]. Given these considerations, this study employs an adapted gravity model, influenced by past research, to pinpoint the ecological security’s association network in the YRB.

$$Y_{ij}=k_{ij}\times {\displaystyle \frac{T_i\times T_j}{{\left(\frac{D_{ij}}{g_i-g_j}\right)}^2}},k_{ij}={\displaystyle \frac{T_i}{T_i+T_j}}$$

(9)

where $Y_{ij}$ represents the ecological security connection strength between two cities. $K_{ij}$ acts as the gravitational coefficient. Both $T_i$ and $T_j$ denote the ecological security measures of city $i$ and city $j$, correspondingly. The term $D_{ij}$ indicates the geometric central distance separating these cities. Similarly, $g_i$ and $g_j$ depict the real GDP per capita for city $i$ and city $j$, in that order. The expression $D_{ij}/(g_i-g_j)$ provides insight into the economic geographical separation of the cities in question. We calculated the spatial correlation network matrix of ecological security based on the modified gravity model, and used the gravitational mean value of ecological security of each row as a judgment criterion. If a value is greater than the mean, it is marked as 1, indicating the existence of spatial correlation; if it is less than the mean value, it is marked as 0, indicating the non-existence of spatial correlation.

2.3.4. Social Network Analysis

Recognized as a versatile instrument, SNA finds application across diverse domains, offering a nuanced understanding of spatial associative configurations and the distinct roles each node assumes. Given its potential, this study seeks to harness SNA to delve into the spatial connectivity traits of ecological security within the YRB. The exploration is bifurcated into holistic network attributes and individualistic network features. The former encompasses four pivotal metrics: network density, connectedness, hierarchy, and efficiency. These portray the bond intensities, structural resilience, node-based asymmetric accessibility, and superfluous links, respectively. The latter introduces three salient indicators, namely, centrality, closeness centrality, and between centrality, elucidating a city’s prominence within the network, its autonomy from its counterparts, and its capacity to exert influence over other cities. The specific formula is outlined below [45].

(1)

Holistic network attributes

Network density is an indicator of the closeness of ecological security connections, which is equal to the ratio of the actual number of existing relationships to the theoretical maximum number of relationships. The higher the network density, the closer the spatial correlation between ecological security levels. If $n$ is the number of cities and $m$ is the actual number of related relationships, the calculation formula for network density $D$ is as follows:

$$D=m/\left[\left(n|\left(n-1\right)\right)\right]$$

(10)

Network connectedness is an indicator used to measure the robustness of ecological security spatial correlation networks. If the correlation degree is 1, it indicates that all cities are in the overall network and have good network robustness. Otherwise, at least one city is considered outside the overall network. If $n$ is the number of cities and $v$ is the paired number of cities that cannot be connected, the formula for calculating the network correlation degree $C$ is as follows:

$$C=1-v/\left[n\left(n-1\right)/2\right]$$

(11)

Network hierarchy is used to characterize the degree of asymmetric accessibility of each city in the ecological security spatial association network. The higher the hierarchy, the greater the gap in the status of each city in the ecological security spatial association network. A few cities are at the center of the network, while the rest of the provinces are at the edge. If $t$ is the number of pairs of symmetric reachable relationships, $max\left(t\right)$ is the maximum possible value of the number of pairs of symmetric reachable relationships, and the calculation formula for the network level $H$ is as follows:

$$H=1-t/\mathit{max}(t)$$

(12)

Network efficiency is used to describe the degree of redundant connections in the ecological security spatial correlation network. Network efficiency is inversely proportional to the redundant connections in the network. If network efficiency is low, there are more redundant connections. The more ecological security spatial correlation paths there are, the more stable the network is. If $e$ is the number of redundant association relationships, $max\left(e\right)$ is the maximum possible number of redundant association relationships, and the calculation formula for network efficiency $E$ is as folows:

$$E=1-e/\mathit{max}(e)$$

(13)

(2)

Individualistic network features

Measures of centrality can indicate which city is important in the ecological network; these measures are centrality, closeness centrality, and betweenness centrality [46]. Centrality can measure the number of paths by which a city directly connects with other cities in the network, reflecting the importance of the city in the network. The higher the centrality, the closer the city is to the central position in the overall network. If n is the number of cities, $m_1$ and $m_2$ are the point-in and point-out degrees, respectively, and the calculation formula for the centrality $C_{RDi}$ is as follows:

$$C_{RDi}=\left(m_1+m_2\right)/\left(2n-2\right)$$

(14)

Closeness centrality is used to characterize the shortest path distance between a city and all other cities, indicating the city’s ability to achieve rapid communication and the degree of synergy with all other cities. The higher the centrality, the less likely the city is to be dominated by other cities in the ecological security spatial association network, and the easier it is to establish connections with other cities. If $d_{ij}$ is the shortcut distance between city $i$ and city $j$, the formula for calculating the closeness centrality $C_{APi}$ is as follows [47]:

$$C_{APi}=\sum _{j=1}^n{d}_{ij}$$

(15)

Between centrality is used to measure the degree of control of a city over the relationships between other cities in a spatial correlation network. The greater the degree of intermediary centrality, the stronger the control of the city over the ecological security relationships between other cities, indicating that the city’s intermediary role in the spatial correlation network is more prominent. If $n$ is the number of cities and $b_{jk}\left(i\right)$ is the ability of city i to control the correlation between city $j$ and the ecological security level of city $k$, the mediating centrality $C_{RBi}$ [47] is as follows:

$$C_{RBi}={\displaystyle \frac{2{\sum }_{j=1}^n{\sum }_{k=1}^n{b}_{jk}\left(i\right)}{n^2-3n+2}}$$

(16)

2.3.5. QAP Analysis

As both the spatial association matrix and related influences constitute matrices derived from relational data, the potential for multicollinearity among these variables arises. This situation, if overlooked, can introduce biases into the derived estimations. Unlike traditional methods that hinge on mutual independence among predictors, QAP stands out as a nonparametric alternative, exhibiting greater resilience [48]. In the context of our research, we employ QAP to delve into the determinants shaping the spatial connections within ecological security networks, shedding light on the underlying mechanisms fostering inter-city associative patterns.

The YRB’s ecological security spatial association network stems from the intricate interplay of various determinants. Understanding these factors illuminates the network’s foundational mechanisms and offers vital insights for its foundational optimization. Relying on previous research, this study identified nine distinct variables: geographical closeness, gauged by the Rook adjacency weight matrix (with adjacent cities valued at 1, non-adjacent ones at 0); economic growth disparities, using GDP per capita as a metric; urbanization variance, deduced from urban population proportions; industrial structure differentials, quantified by tertiary industry ratios; population density disparities; intellectual capital variances; high-tech differentials, assessed by local tech spending as a share of general public budgets; governmental oversight, estimated by ratios of fiscal expenditure to GDP; and environmental guidelines, gauged through parameters like industrial wastewater and emissions of sulfur dioxide, smoke (dust), and carbon. Other metrics include centralized sewage treatment rates, domestic waste processing rates, and industrial waste management ratios. We synthesize a comprehensive environmental regulatory index using the entropy weight method.

3. Results

3.1. The Descriptive Statistical Characteristic of Ecological Security

Table 2. Descriptive statistics of ecological security in the YRB.

Value	2005	2010	2015	2019
average value	0.1253	0.1600	0.1869	0.2088
standard deviation	0.0497	0.0673	0.0777	0.0853
kurtosis coefficient	1.843	2.883	2.943	3.508
skew coefficient	1.306	1.656	1.742	1.865

outlines the ecological security dynamics of the YRB spanning 2005–2019. The data suggests a consistent uptrend, with the ecological security mean value transitioning from 0.1253 in 2005 to the higher value of 0.2088 by 2019. This growth trend indicates that the overall ecological security level of the YRB has been effectively improved in the past 15 years. This may be due to the continuous strengthening of ecological protection measures in the basin, increasing the management of environmental pollution, promoting ecological restoration projects, and promoting the concept of green development, gradually improving the ecological environment.

The standard deviation reflects the degree of dispersion of the data distribution. The standard deviations of the ecological security level of the cities in the YRM in 2005 and 2019 are 0.0497 and 0.0853, respectively. This indicates that the ecological security level is relatively close between the cities in 2005 and the ecological security situation is relatively stable. But in 2019, the gap between the ecological security levels of cities gradually increased. This may be due to the fact that different cities have different economic development rates, ecological protection inputs, and policy implementation, leading to inconsistent improvement in the level of ecological security. The kurtosis coefficients are 1.843 and 3.508 in 2005 and 2019, respectively. The increase in the kurtosis coefficient signifies a more extreme distribution of the level of urban ecological security and an increase in the number of cities with similar levels. The gap between cities with good and poor levels of ecological security is increasing. The skewness coefficients are 1.306 and 1.865 in 2005 and 2019, respectively, and the data show a clear right skew. This indicates that the level of ecological security is gradually improving and the number of cities is increasing. Meanwhile, cities with poor ecological security levels face great ecological pressure and urgently need balanced development.

3.2. The Spatiotemporal Evolution Characteristic of Ecological Security

In order to clarify the spatial differences in YRB’s ecological security, we classified the ecological security status of the YRB based on unified standards. Based on the calculated ecological security level results, 0.2006, 0.3030, and 0.4054 were established as classification breakpoints. The status of 2005, 2010, 2015, and 2019 was divided into four categories: low, medium-low, medium-high, and high (Figure 3). The results show that the regional differences in ecological security in the YRB are very significant. From the distribution of ecological security level, it can be seen that there are only a small number of cities with high or medium-high ecological security levels, while cities with low or medium-low ecological security level are more widely distributed. This phenomenon fully shows that the ecological condition of the whole YRB is not ideal. The widespread occurrence of cities with low ecological security levels may be attributed to prolonged reliance on traditional practices and resource overexploitation, resulting in substantial environmental degradation, coupled with insufficient investment in ecological restoration and protection measures.

During the period from 2005 to 2019, the ecological security level of provincial capitals such as Xi’an, Zhengzhou, Taiyuan and Jinan was higher than that of other cities in the same period. The cities realized the transition from low to high, and the ecological security level was gradually improved. This was largely due to their political, economic and geographical advantages. Political advantages enable these cities to gain more support for policy formulation and implementation. Economic advantages allow them to have sufficient funds to invest in ecological and environmental governance. The location-specific advantage is conducive to attracting more talents and technology, and improving the scientific and technological level of ecological protection. With strong economic and social growth, these cities are better able to offset ecological pressures and achieve coordinated development between ecology and economy.

The ecological security of Hohhot, Baotou, Ordos, Jinzhong, Linfen, Xinxiang, Jiaozuo and other provincial capital cities and the Shandong Peninsula city cluster has generally transited from low-grade to intermediate- or even high-grade, and the ecological security is at a medium level. These cities can achieve improvements in ecological security level. On the one hand, because they are driven by the radiation of provincial cities, they benefit from industrial transfer, technological exchange, and other aspects, promote the optimization and upgrading of economic structure, and reduce the damage to the ecological environment. On the other hand, the Shandong Peninsula city cluster itself has a relatively developed economy, and has certain capital and technical strengths, allowing it to invest in ecological protection. At the same time, it also actively responds to the national ecological policy and strengthens the ecological environment governance in the region.

Low to medium-low ecological security is mainly distributed in Gansu, Shaanxi, Shanxi, Inner Mongolia and other regions. Although these cities have shown improvements in their ecological security index, they have not achieved substantial advancements and remain at low levels. This condition primarily stems from their inherently fragile ecological environment, facing challenges such as soil erosion and land desertification. Additionally, these regions experience relatively slow economic development, maintain simple industrial structures, and possess limited capacities for ecological protection investment, resulting in gradual ecological restoration progress. Despite recent improvements, supported by national policies, significant challenges persist in achieving substantial ecological security enhancement.

From the perspective of geographical distribution, the regions of Shanxi, Henan, Shandong demonstrate relatively favorable ecological security levels, with gradually improving ecological conditions. This improvement can be attributed to their advanced economic development, substantial ecological protection investment, and accelerated industrial restructuring towards green and low-carbon patterns. However, central and western regions, including Shaanxi, Gansu and Inner Mongolia, exhibit low ecological security levels with slow recovery rates, showing marked disparities compared to eastern regions. Beyond their fragile ecosystems and delayed economic development, these central and western regions face challenges from weak infrastructure, limited ecological protection technology, and insufficient expertise. Addressing these regional disparities requires enhanced cooperation, increased policy support, and capital investment in central and western regions to achieve comprehensive ecological security improvements across the YRB.

3.3. Characteristic Analysis of Spatial Association Network

Within this study, we employed an adapted gravity model to map out the spatial interrelations of ecological security across 78 cities within the YRB. In order to deeply portray the spatial correlation path and intensity of ecological security in the YRB, this paper uses ArcGIS10.8 software to visualize the spatial correlation network of ecological security in 2005, 2010, 2015 and 2019, and divides the gravitational values into four levels by using the natural breakpoint method. The first-level network represents the weak connection effect, and the network connection strength of the other levels increases sequentially, as shown in Figure 4. Upon examination, the intricacies of the basin’s ecological security emerge as a multifaceted network. Significantly, urban areas are beginning to transcend their conventional geographical confines, giving rise to inter-regional connectivities. From an overall point of view, the number and density of network connections at all levels in the YRB increased significantly from 2005 to 2019, which indicates that the ecological security linkages between regions in the YRB gradually strengthened. In 2005, the YRB primarily exhibited a first-level network dominance, with limited second- and third-level networks, while fourth-level networks only existed between selected cities in the northern Shandong Peninsula. In terms of the spatial characteristics, the network structure of the upstream and middle reaches is relatively sparse, and the downstream area, especially with the urban agglomeration of the Shandong Peninsula as the core distribution is relatively dense. In 2010 the first-level network was gradually developed and perfected and with the urban agglomeration of the Shandong Peninsula and Ordos and other regional central cities as the core, the second-level and third-level networks began formation. The year 2015 saw extensive second-level network formation, establishing a primary network-dominated structure supplemented by secondary and tertiary networks. By 2019, the spatial correlation network featured widespread first-level distribution, predominantly featured second-level networks, and supported third- and fourth-level structures, with fourth-level networks radiating from the Shandong Peninsula urban agglomeration and provincial capitals like Ordos.

3.3.1. Overall Network Characteristic of Spatial Association Network

Figure 5 illustrates the computation of overarching network attributes from 2005 to 2019 using Ucinet software (Version 6.560). The network connectivity score maintained a constant value of 1 throughout the study period, demonstrating comprehensive spatial linkages of ecological security among all cities without isolation. This indicates the existence of significant spatial interrelations and spillover effects, confirming that there was a robust network structure. The network hierarchy value of 0 reflects the balanced influence of each city in terms of the spatial interconnectedness of ecological security, highlighting their individual contributions to the spatial linkages. Over the past 15 years, the ecological security network density in the YRB has exhibited a complex “up–down–up” fluctuation pattern, though the overall trend shows decline. The network density reached its peak of 0.1920 in 2007, when cities within the YRB prioritized ecological conservation collaboration and actively participated in cross-regional ecological project exchanges. However, the index decreased to its lowest value of 0.1850 in 2014, possibly due to cities prioritizing individual economic over ecological safety cooperation, resulting in reduced inter-city interactions. In 2019, the value of the index increased to 0.1880, but is still well below the median of 0.5, indicating that spatial interactions for ecological security in the YRB are still very limited. The low network density reveals deficiencies in ecological resources sharing, protection technology exchange, and the coordinated management of ecological issues among cities, emphasizing the urgent need for enhanced inter-city cooperation and regional coordination.

It is worth noting that the efficiency of the ecological security network in the YRB maintains a consistently high level, exceeding 0.7, and demonstrates an oscillating upward trend. This improved network efficiency indicates reduced redundant connections between cities, enabling more efficient coordination of ecological resource allocation, information transmission, and protection actions. Several cities have enhanced resource utilization efficiency in ecological protection project cooperation. However, despite high network efficiency, the overall network stability requires strengthening. This may be attributed to insufficient stability in the ecological security situation of key node cities, where ecological crises could significantly impact the entire network’s operation.

Analysis of network density, hierarchy, connectivity and efficiency reveals that resource circulation in YRB cities operates efficiently, with cities being successfully integrated into the spatial network, establishing a foundation for ecological security protection. However, the current low network density necessitates the urgent enhancement of the network structure.

3.3.2. Individual Network Characteristic of Spatial Association Network

In examining the overarching network attributes, this study delves deeper into the singular network of the YRB’s ecological security. It particularly focuses on metrics like centrality, closeness centrality, and betweenness centrality over four distinct periods of 2005, 2010, 2015, and 2019. The objective is to decipher each city’s position, contribution, and role within the ecological security association network.

Centrality indicates how closely a city is linked to the ecological security of other cities. The spatial distribution reveals cities of high and medium centrality, mainly distributed in the Shandong Peninsula urban cluster and the Hohhot–Baotou–Ordos–Yulin Urban Group, alongside provincial capitals like Taiyuan, Zhengzhou, Hohhot, and Xi’an. Notably, cities like Taiyuan, Zhengzhou, Yan’an, and Baoji displayed a downtrend in centrality, indicating their drift from the network’s core (Figure 6). In contrast, cities like Pingliang and Xi’an are drawing nearer to the network’s center. Regions exhibiting low to medium-low centrality are predominantly found in Gansu’s Hexi, Shanxi’s Jinzhong and Jinbei, and in parts of Henan, Shandong, Ningxia, and Inner Mongolia. Specifically, Henan and Shandong’s low centrality have seen a diminishing pattern over time, whereas Ningxia and Gansu display an uptick. The subdued centrality in these regions can be attributed to their marginal ecological security and the absence of developmental pivots. This necessitates a future focus on strengthening metropolitan constructions, especially with the rising centrality in areas like Zhengzhou and Xi’an. Embracing cross-regional collaboration, fostering resource exchange among cities, and nurturing new growth centers can enhance holistic network connectivity.

Closeness centrality reflects the proximity of a city to other cities, emphasizing the ease of rapid interaction. The spatial and temporal distributions of closeness centrality and centrality are very similar for all cities, except for the intersection of Shaanxi and Gansu (Figure 7). In this intersection, the closeness centrality gravitates towards median values, which implies the region’s potential to foster spatial affiliations via capital, technological advances, and managerial strategies. Enhancing the ecological security here can play a pivotal role in augmenting overall regional ecological stability. Geographically, regions exhibiting diminished or marginally low closeness centrality span across territories in Shanxi, Henan, and Gansu. Over the studied interval, Shaanxi and Gansu’s areas with such values have expanded, signifying that the ecological betterment exerts minimal influence on neighboring cities. Furthermore, their progress is not substantially propelled by adjacent cities, relegating them to “peripheral entities” within the ecological security spatial network. The distance between these areas and the central city, combined with the lack of intermediary hubs, ultimately leads to a lower closeness centrality.

Betweenness centrality assesses the influence of each city within the ecological security’s spatial association matrix. Figure 8 illustrates these findings. The spatial analysis reveals a distinct polarized pattern in the distribution of betweenness centrality. Cities exhibiting significant medium to high values in this metric are limited in quantity and spatially consistent. This primarily encompasses developed cities such as Zhengzhou, Xi’an, and Jinan, which function as crucial bridges and intermediaries in the YRB’s ecological safety network. These cities represent critical junctions or nodal points in this spatial framework, exerting substantial influence over the entire network. Disruptions in these hubs could generate vulnerabilities, creating ”structural voids”. Conversely, areas displaying lower or marginally elevated betweenness centrality values are extensive and demonstrate growth tendencies. This suggests that the YRB lacks sufficient cities serving as transmission mediators for ecological security, thus impending the development of robust spatial associations.

3.4. QAP Analysis

3.4.1. QAP Correlation Analysis

Using Ucinet, a correlation study was analysis was performed, employing 5000 random permutations to calculate correlation coefficients between the ecological security association matrix and individual influencing factors, as shown in

Table 3. QAP correlation analysis results.

	2005	2007	2009	2011	2013	2015	2017	2019
Geographical proximity	0.231 ***	0.226 ***	0.221 ***	0.226 ***	0.232 ***	0.232 ***	0.220 ***	0.225 ***
Difference in economic development	0.454 ***	0.452 ***	0.459 ***	0.456 ***	0.453 ***	0.455 ***	0.460 ***	0.459 ***
Difference in environmental regulation	0.047 *	0.023	0.018	0.069 **	0.031	0.043	0.064 **	0.055 *
Difference in industrial structure	0.077 **	0.088 **	0.058 *	0.098 ***	0.090 **	0.055 *	0.056 *	0.062 *
Difference in intellectual capital	0.243 ***	0.240 ***	0.238 ***	0.212 ***	0.164 ***	0.163 ***	0.159 ***	0.157 ***
Difference in government regulation	−0.012	0.040	0.015	0.017	0.032	0.025	0.008	0.030
Difference in population density	−0.048 **	−0.021	−0.047 **	−0.034	−0.018	0.005	−0.010	−0.022
Difference in urbanization	0.242 ***	0.221 ***	0.211 ***	0.203 ***	0.244 ***	0.234 ***	0.231 ***	0.246 ***
Difference in high-tech	0.164 ***	0.150 ***	0.167 ***	0.165 ***	0.165 ***	0.182 ***	0.151 ***	0.156 ***

Note: *, **, and *** represent significant levels at 10%, 5%, and 1%, respectively.

. This outcome demonstrates clear patterns. Regarding geographical proximity, resource flow and information sharing costs decrease significantly when regions are geographical close, facilitating the establishment of strong ecological connections. For instance, neighboring regions can implement joint ecological protection initiatives, share ecological monitoring data in real time, and collectively address various ecological challenges, effectively enhancing the spatial correlation of ecological security networks. Regarding urbanization differences, regions with varying urbanization levels exhibit clear complementarity in ecological resource requirement and protection approaches. Highly urbanized areas typically possess advanced environmental protection technology and adequate capital, while less urbanized areas maintain abundant natural ecological resources. Through collaborative exchanges, both parties can enhance the ecological security network and strengthen spatial linkages. Concerning differences in industrial structure, regions with distinct industrial compositions can achieve collaborative development. For example, industrial and agricultural areas can cooperate in resource recycling and ecological product supply, effectively reducing environmental pressure and enhancing the correlation of the ecological security network correlation. Technological disparities significantly influence ecological security spatial correlation. Technologically advanced cities can disseminate environmental management, protection, and monitoring technologies to developing cities. Intellectual capital difference also plays a crucial role. Regions rich in intellectual capital can provide expertise and knowledge to areas with limited intellectual resources, jointly addressing ecological security challenges and enhancing spatial correlation. Economically growing regions typically possess more ecological protection resources and can stimulate surrounding areas through financial assistance and technology transfer, promoting inter-regional ecological security cooperation and improving spatial correlation.

Furthermore, throughout the analysis process, these factors demonstrated statistical significance, indicating that their high differentiation positively influences the spatial correlation of the ecological security network. However, population density differences and the environmental regulatory matrix only show significance in specific years. Initial research indicates that greater environmental regulation differences facilitate spatial association. This occurs because substantial differences in environmental regulation reflect regional variations in ecosystem regulation capacity and ecological protection strategies. These differences promote mutual learning between regions, such as sharing water resource management and forest protection experiences, jointly optimizing ecological security networks, and strengthening spatial correlation. Conversely, smaller population density differences favor spatial correlation. Regions with similar population densities share comparable ecological resource-bearing pressures and ecological demands, enabling easier consensus-building and the implementation of ecological cooperation projects. This reduces ecological resource distribution inequalities and ecological conflicts arising from population density differences, supporting spatial correlation.

At the same time, the difference matrix related to government regulation has never passed the significance test, which indicates that the impact of government regulation on the spatial correlation of ecological security is still uncertain. This may be due to the fact that government regulation involves many complex factors such as policy formulation, enforcement intensity, and scope of regulation. The modes and effects of regulation differ in different regions at different stages and on different ecological issues, and a stable and observable influence model on spatial correlation of ecological security has not yet been formed, so it is difficult to show a significant correlation in this study.

3.4.2. QAP Regression Analysis

Between 2005 and 2019, comprehensive QAP regression analyses were conducted using a dataset of 5000 randomized permutations (

Table 4. QAP regression analysis results.

	2005	2007	2009	2011	2013	2015	2017	2019
Geographical proximity	0.2546 ***	0.2534 ***	0.2476 ***	0.2551 ***	0.2615 ***	0.2620 ***	0.2489 ***	0.2582 ***
Difference in economic development	0.5202 ***	0.4365 ***	0.4322 ***	0.4459 ***	0.4512 ***	0.4464 ***	0.4601 ***	0.4465 ***
Difference in environmental regulation	−0.0476 ***	−0.0596 ***	−0.0580 ***	−0.0678 ***	−0.0571 ***	−0.0432 ***	−0.0237	−0.0337 **
Difference in industrial structure	−0.0555 ***	−0.0405 **	−0.0279 *	0.0079	0.0183	−0.0182	−0.0116	−0.0141
Difference in intellectual capital	0.1373 ***	0.1478 ***	0.1464 ***	0.1210 ***	0.1120 ***	0.1171 ***	0.1186 ***	0.0973 ***
Difference in government regulation	−0.0313 *	−0.0034	−0.0074	−0.0131	−0.0121	−0.0221	−0.0263	−0.0189
Difference in population density	−0.0005	−0.0027	−0.0270 **	−0.0180	0.0004	−0.0035	−0.0200	−0.0307 **
Difference in urbanization	0.0149	0.0408 **	0.0421 **	0.0382 **	0.0421 **	0.0503 **	0.0506 ***	0.0605 ***
Difference in high−tech	−0.1238 ***	0.0075	0.0308 *	0.0470 **	0.0145	0.0263	−0.0146	0.0392 **
R2	0.305	0.295	0.300	0.299	0.296	0.296	0.292	0.297
Adj−R2	0.304	0.294	0.299	0.298	0.295	0.295	0.291	0.296

Note: *, **, and *** represent significant levels at 10%, 5%, and 1%, respectively.

). The adjusted R2 consistently ranged from 0.291 to 0.304, achieving significance at the 1% level. These results indicate that the selected determinants explain between 29.1% and 30.4% of the variance in spatial ecological security interconnections within the YRB, demonstrating robust model fit. Analysis of standardized regression coefficients revealed that economic development and geographical proximity emerged as dominant factors among the determinants, significantly influencing spatial ecological security relationships within the basin. Geographical proximity consistently exhibited positive coefficients, indicating that spatial proximity facilitates the formation of ecological networks. The YRB’s inherent characteristics, including limited water availability, high sand content, and unpredictable rivers, strengthen this relationship. These natural features create substantial barriers to resource movement, fostering stronger spatial connections among adjacent cities. Economic development disparities consistently demonstrate positive coefficients, suggesting that economic differences between cities strengthen spatial interconnections. This pattern relates to the spatial diffusion and redistribution of ecological security determinants. Environmental regulation variations showed statistically significant negative coefficients, except in 2017, indicating that similar environmental guidelines across cities promote spatial network development. Cities with comparable economic capabilities and resource endowments tend to adopt similar environmental measures, reinforcing their interconnections. Industrial structure coefficients displayed cyclical patterns, with statistical significance being greater during periods of negative values, suggesting that there are periodic inhibitory effects on spatial ecological network development, particularly in later study periods. Intellectual capital variations maintained a positive trend, reflecting the uneven distribution of academic and intellectual resources across the basin, with resource-rich regions maintaining advantages. Governmental regulation’s influence was only significant in 2005, suggesting its diminished contemporary role in shaping ecological security spatial patterns. Population density coefficients only achieved statistical significance in 2009 and 2019, indicating its limited impact on spatial ecological networks. While urbanization variations showed non-significant coefficients in 2005, they demonstrated increasing positive influence in subsequent years. Urban hubs, with their elevated urbanization levels, can channel crucial resources such as technological innovation and skilled personnel, indispensable for ecological resilience. The technological prowess of regions, gauged by the high-tech differential coefficient, displayed episodic significance. The initial phases of the study, characterized by economic infancy and minimal technology innovation disparities, evidenced a more harmonized regional collaboration. However, as time progressed and the technological innovation divide widened, regions with advanced capabilities began exerting a more pronounced influence on their less advanced counterparts.

4. Discussion

4.1. Spatial and Temporal Evolution Characteristics of Ecological Security

Data analysis of the ecological security level of the YRB from 2005 to 2019 shows that the overall ecological security level of the YRB has steadily improved. This improvement is attributed to the transformation of the regional economic development mode and the intensification of ecological protection policies. This aligns with previous research findings [49,50,51]. Cities exhibiting high ecological security levels are predominantly located in the Shandong Peninsula city cluster and provincial capital cities. These economic growth centers attract capital, technology, and talent through resource advantages, efficient utilization, and effective social response mechanisms, resulting in reduced environmental pressure from economic development and higher overall ecological security. The Shandong Peninsula urban agglomeration demonstrates superior economic development and technological innovation capabilities compared to other basin urban clusters, leading to reduced ecological impact from production activities. Conversely, lower ecological security levels are primarily observed in Gansu, Ningxia, and select cities in Shanxi, Shaanxi, and Inner Mongolia. These regions typically exhibit lower economic development, relying on less sophisticated growth models and suboptimal industrial structures, resulting in increased environmental pressure and diminished ecological security [52,53]. Some studies, emphasizing mine management funds and nature reserve areas as key indicators, concluded higher ecological levels in the western YRB’s coal-mining regions [54]. City rankings fluctuate annually, with Ordos showing remarkable improvement from 44th position in 2005 to 10th, primarily due to its adoption of new industrialization approaches since the 21st century, enhancing industrial economy quality and market competitiveness. Following the 18th National People’s Congress of China, Ordos has advanced industrial restructuring and transformation, significantly improving ecological security. Baotou City, a rapidly developing economic center in Inner Mongolia, has similarly pursued transformation and upgrading strategies.

4.2. Characterization of Spatial Correlation Network

Based on closeness centrality and other network measures, the cities that serve critical roles in the network can be identified [46]. Spatial correlation network analysis reveals that cities with medium and high values of centrality and other indicators are primarily distributed across the Shandong Peninsula city cluster, the Hubao-Eyu city cluster, and Taiyuan, Zhengzhou, Hohhot, Xi’an, Yan’an City, and Baoji City. These cities maintain stronger connections with other cities and occupy core positions within the spatial correlation network. These urban centers possess robust economic foundations, advanced industrial structures, strong innovation capabilities, and significant advantages in infrastructure and resource allocation, enabling them to exercise leadership in maintaining and enhancing regional ecological security. Additionally, their frequent exchanges and interactions with other cities in the network strengthen their central position, facilitating the coordinated development of inter-regional ecological security. In contrast, cities such as Bayannaoer, Binzhou, Dezhou, and Weifang consistently ranked low across various indicators during the study period. Analysis reveals that remote geographical locations and fragile ecological environments significantly constrain these cities’ development potential. Furthermore, these cities face challenges including slow economic growth, insufficient capital investment, limited technological innovation capacity, and outdated management practices, diminishing their competitiveness in the regional ecological security network. Regarding spatial distribution patterns, low-value areas are primarily concentrated in Gansu Province’s Hexi region, Shanxi Province’s Jinzhong and North Shanxi regions, and select cities in Henan, Shandong, Ningxia, and Inner Mongolia, exhibiting low-low agglomeration characteristics. Notably, low-value areas in Henan and Shandong demonstrated a decreasing trend, potentially attributable to increased ecological protection investments and industrial structure optimization. Conversely, Ningxia and Gansu showed increasing low-value trends, suggesting heightened challenges in ecological security maintenance, requiring enhanced policy support and resource allocation. The middle and low regions, primarily distributed across parts of Henan, Shaanxi, Gansu, and Shanxi, exhibit relatively low ecological security levels and lack strong regional growth poles, resulting in low centrality and paracentricity indices. This outcome correlates closely with regional economic development, ecological protection capacity, and inter-city cooperation levels [55].

4.3. Policy Implications

Economic development and ecological construction are intrinsically linked. Cities with developed economies tend to have high levels of ecological security. Examples include the Shandong Peninsula urban agglomeration, Taiyuan, Hohhot, and Xi’an. Cities facing problems such as weak economic growth and ecological pollution should abandon the backward development model of “development first, governance later”. They should instead leverage their status as late bloomers to catalyze transformative changes, underpinned by innovation and a solid ecological base for economic ascendancy. Developed cities such as the Shandong Peninsula urban agglomeration, Taiyuan, Hohhot and Xi’an, as bastions of both economic and ecological vitality, should continue to adhere to green and low-carbon development strategies. These urban centers must enhance their metropolitan influence, direct substantial economic resources toward ecological preservation, and maintain balanced economic and ecological systems.

While industrial advancement plays a pivotal role in spurring economic and societal progress, it underscores the palpable strain and detrimental impact that human activities exert on our natural surroundings, as highlighted by Li et al. [56]. The core issue with ecological security stems from the discord between the pace of industrial growth and the ecological threshold. Traversing from mid-to-late phases of industrialization, the YRB manifests significant challenges tied to an oversimplified, rigid, and standardized industrial matrix [57]. Predominantly, urban industrial progression hinges on competitive strategies for economic expansion, yielding fragmented urban industrial landscapes and limited inter-city collaboration. The primary objective for cities now involves identifying regional strengths and constraints to target economic and ecological development goals effectively. Furthermore, cooperation mechanisms, including multilateral agreements, can facilitate industrial restructuring and the development of an economically concentrated spatial layout. This approach will enhance industrial interconnectivity between cities, increase regional productivity, and strengthen competitive advantages.

Economic development and geographical proximity are the primary factors driving significant changes in ecological security within the YRB. Regarding economic development, upstream cities should prioritize eco-industry development and establish eco-tourism, specialized agriculture, and animal husbandry, leveraging their unique natural landscapes and ethnic cultural heritage to convert ecological advantages into economic benefits. Midstream cities and downstream cities have a strong industrial base, therefore, they should transform and upgrade traditional industries, cultivate new industries, increase technological change to develop new industries, create industrial clusters, attract talents and capital, and promote the diversification of industrial structure. Additionally, due to high population density, substantial agricultural demand exists. Downstream cities must develop efficient agricultural practices, enhance service industries, and promote ecological, urban, and intelligent agriculture. This includes improving agricultural production efficiency and quality, strengthening agricultural product branding, developing advanced processing methods, and increasing agricultural value addition. Furthermore, these regions should utilize the Yellow River’s cultural and ecological resources to develop cultural–creative industries, tourism, healthcare, and other service sectors, thereby increasing the service sector’s economic contribution and promoting sustainable economic development. Regarding geographical proximity, cities across the YRB must enhance collaboration in water resource utilization and protection by establishing joint water resource monitoring and allocation mechanisms, implementing cross-regional pollution monitoring systems, developing platforms for ecological protection coordination, improving upstream water conservation practices, and establishing appropriate ecological compensation mechanisms between upstream and downstream cities.

Policy implementation requires the establishment of cross-regional and cross-departmental collaborative mechanisms, the enhancement of communication between the upper, middle, and lower reaches, and the facilitation of information sharing, resource integration, and policy coordination. These efforts aim to collectively improve the YRB’s ecological security level, promote coordinated development between economic society and ecological environment, and optimize the spatial correlation network of ecological security.

4.4. Research Insufficiency and Prospect

On the basis of constructing an ecological security evaluation system, this paper further portrays the spatial and temporal changes, spatial correlation characteristics, and factors influencing ecological security in the YRB from 2005 to 2019. But there are still several limitations that can be improved in the future.

(1)

Due to the lack of data on relevant indicators of ecological security evaluation at the county level, this paper only studied the characteristics of spatial correlation of ecological security at the municipal level, and the research scale is too rough for a more precise analysis. Therefore, the data could be further mined in the future research to construct the ecological security evaluation index system at the county level to improve the application value of the research.

(2)

The correction of the gravity model needs to be further improved. With the popularization of various modes of transportation, the limitation of spatial distance has been greatly weakened and time costs have been saved. In this paper, only the geospatial distance and the economic distance are considered; the time cost is not considered. Therefore, in the future, the distance measurement should be continuously improved in order to measure the distance more accurately.

5. Conclusions

Utilizing the entropy weight–TOPSIS method, our research assessed the ecological security status across 78 cities within the YRB from 2005 to 2019. The findings shed light on the distinct spatiotemporal progression of ecological security. Leveraging a refined gravity model, we further deciphered the spatial interconnection patterns of the YRB’s ecological security. The intrinsic features of both the overarching and individual networks came under scrutiny through a social network analysis approach. Furthermore, the QAP assisted in probing the determinants that influence the YRB’s ecological security, encompassing both correlation and regression analyses. This comprehensive approach aimed to elucidate the underlying dynamics that give rise to the interconnected network of ecological security within the YRB. The main conclusions are as follows:

(1)

From 2005 to 2019, the ecological protection of the YRB continued to strengthen, but the difference in the level of ecological security between cities became larger and the spatial difference was obvious. The number of cities with high levels of ecological security remains limited, while cities with medium and low levels are more common.

(2)

From 2005 to 2019, all cities in the YRB were in an associated network, and there were no isolated cities. The cities had a balanced position in the network, but the density of the network was low, and the stability of the network needed to be further strengthened. In terms of individual attributes, Shandong Peninsula, Hubao-Eyu, Taiyuan, and Zhengzhou were not only hub nodes, but also had close relationships with neighboring cities, effectively playing the role of hub connection and intermediary. On the contrary, cities such as Gansu, Ningxia, Shanxi, Shaanxi and Henan were marginal in their connections with other cities.

(3)

The standardized regression coefficients of economic development level and geospatial proximity are much higher than other factors; these are the main factors contributing to the differences in ecological security associations in the YRB.