Evaluation of Coastal Ecological Security Barrier Functions Based on Ecosystem Services: A Case Study of Fujian Province, China

Abstract

Constructing coastal ecological security barriers is beneficial for preventing environmental degradation and enhancing resilience to natural disasters. This study examines the functionality of these barriers from an ecosystem service perspective, developing an Ecosystem Security Barrier Function (ESBF) index to analyze its spatiotemporal variations. From 2000 to 2020, habitat quality in the study area experienced a slight decline, while water supply capacity initially increased and then decreased. Water purification capacity hit its lowest point in 2015 before improving. The ESBF generally ranged from moderate to high levels, with higher values in the northwest and lower values in the southeast, showing strong spatial autocorrelations. Despite mild degradation in some areas, overall stability was maintained with frequent transitions between ESBF levels. Utilizing the Multiscale Geographically Weighted Regression (MGWR) model, we conducted a grid-scale analysis of the driving mechanisms behind ESBF. We found that precipitation, elevation, and the Normalized Difference Vegetation Index (NDVI) positively correlated with ESBF, whereas population density, land use, and nighttime lights negatively correlated. The relationship between temperature and ESBF showed a “north-positive, south-negative” pattern. The study recommends enhancing coastal wetland restoration, strengthening protective forest construction, and effectively controlling pollutant sources entering the sea to safeguard and improve the ecological security barrier function.

1. Introduction

Climate change and human development are rapidly transforming coastal regions worldwide. These areas, known for their dense populations, advanced economies, and rich biodiversity, provide significant ecological services [1]. However, frequent extreme weather events and rapid urbanization have led to recurrent natural disasters in coastal areas [2] and a continuous reduction in ecological land, making it increasingly difficult to balance ecological protection and economic development [3]. In response to ecological degradation, China initiated several large-scale ecological restoration projects in late 1990s, significantly mitigating the issue [4]. Despite these efforts, ecological protection remains a long-term challenge. In 2020, the Chinese government issued the “Master Plan for Major Projects in the Protection and Restoration of Important National Ecosystems (2021–2035)”, outlining nine key projects, including those focusing on coastal zones. This plan emphasizes improving the ecological security barrier system to enhance environmental quality and maintain stable ecosystem development [5]. As the first line of defense against marine disasters, constructing coastal ecological security barriers is both a crucial component of China’s national strategy and a practical necessity for ecosystem protection [6].

The term “ecological barrier” has various interpretations, referring to both natural geographic units like mountains and rivers, and artificial structures like bridges and cities. These barriers can impede the exchange of ecosystems, species, and genes between geographic units. This obstructive function is crucial for preventing species invasion, pollution dispersion, and other environmental threats [7]. An ecological security barrier is a composite concept encompassing both spatial and functional dimensions. It refers to marginal areas that protect internal ecosystems from external environmental disturbances and destruction [8]. These barriers function within specific regions to ensure ecological security and promote sustainable development [6]. Ecosystem service functions are important tools for reflecting the functionality of ecological barriers. Ecosystem services refer to the various benefits and products provided by ecosystems through their natural processes and functions. Daily [9] defined ecosystem services as “the benefits and products that humans derive directly or indirectly from the natural environment”. Costanza [10] further proposed a widely used classification framework for ecosystem services, dividing them into four major categories: provisioning services, regulating services, supporting services, and cultural services. The study of ES balance has become a focal point in current ecological and environmental research and has been widely applied in the monitoring and management of regional ecological security [11].

Ecological security barriers are often viewed as the supply side of ecosystem services [12]. From an ecosystem service perspective, these barriers serve as supply areas while the regions they protect are the demand areas [13]. Therefore, quantifying and spatially analyzing ecosystem service functions can illustrate the strength of ecological barrier functions [6]. Land use change, a key factor in human and ecosystem change, significantly impacts ecosystem service functions [14]. Consequently, understanding how land use influences these ecosystem service functions is a primary focus of current research [15]. In recent years, the development of tools such as InVEST has made the spatial assessment and management of ecosystem services more refined and actionable [16,17]. Various spatial regression models, such as Geographically Weighted Regression (GWR), Ordinary Least Squares (OLS), and Multiscale Geographically Weighted Regression (MGWR), are commonly employed to analyze spatial dependencies and correlations between research targets and potential factors [18]. While the OLS model quantifies the influence of independent variables on dependent variables, the GWR model addresses spatial non-stationarity but has limitations in explaining spatial heterogeneity [19]. The MGWR model, an extension of GWR, overcomes these limitations by elucidating the multiscale relationships between variables, making it increasingly popular in spatial research [8].

This study focuses on the coastal areas of Fujian Province, which are vital to China’s economic development and biodiversity conservation [20]. Rapid industrialization, urbanization, agricultural modernization, and frequent extreme weather events have severely threatened the ecological services of these coastal areas, leading to significant reductions in terrestrial ecological resources and adverse impacts on the marine environment, such as declining water quality and habitat degradation [21]. As a transitional zone between terrestrial and marine ecosystems, Fujian’s coastal areas serve as crucial ecological barriers. Therefore, investigating the spatial differentiation, temporal changes, and driving mechanisms behind the ecological security of this region is essential.

Ecological security barriers are regions of unique ecological value, and coastal areas serve as critical buffers between the ocean and inland regions [8]. This study adopts an ecosystem services perspective to quantify water yield, water purification capacity, and habitat quality in Fujian Province from 2000 to 2020. The aim is to enhance habitat quality, reduce marine pollution, and meet flood prevention and disaster mitigation needs. Based on these data, we construct an Ecological Security Barrier Function (ESBF) index and analyze its spatial distribution pattern and spatial autocorrelation. Additionally, we use the MGWR model to investigate the primary driving factors and intrinsic relationships influencing the regional ESBF. The results provide scientific support for ecological protection planning in Fujian’s coastal areas and offer a reference for assessing the ecological barrier functions of other coastal regions. For the convenience of readers, a list of acronyms used in this paper, along with their full forms, is provided in the Abbreviations.

2. Study Area and Data Sources

2.1. Study Area

Fujian Province, located in southeastern China, borders Zhejiang, Jiangxi, and Guangdong provinces. The study area encompasses the coastal zone of Fujian Province (23.5671 E~27.4430 E, 116.9221 E~120.7220 N), including 27 districts. This area covers the cities of Zhangzhou, Quanzhou, Putian, Fuzhou, Ningde, and Xiamen, with a total area of approximately 27,097 km² (Figure 1). The topography primarily consists of coastal plains, coastal wetlands, and hilly mountains, characterized by a subtropical monsoon climate influenced by the ocean. The region experiences a warm and humid climate, with an annual average temperature of 17~22 °C. Precipitation is abundant and concentrated in the spring and summer seasons, with red soil, lateritic soil, and paddy soil being the predominant soil types. The coastal zone serves as an ecological barrier for inland areas, playing a crucial role in protecting the inland environment, regulating the climate, and defending against natural disasters. However, economic development and population growth in coastal areas have placed significant pressures on the coastal ecosystem [22,23].

2.2. Data Sources

In this study, land use data are classified into six primary categories: cultivated land, forestland, grassland, water areas, construction land, and unused land, along with 21 secondary categories. All spatial data are projected using the WGS 1984 UTM_49 coordinate system, and the raster datasets are resampled to a spatial resolution of 30 m × 30 m, with preprocessing performed on the ArcGIS software (Version 10.8.2) platform. NDVI data are downloaded using the GEE platform. The other data types and their sources are listed in

Table 1. Data used in the paper and their sources.

Data	Data Resources	Description
Land Use Dataset	Resource and Environment Science and Data Center (https://www.resdc.cn) (accessed on 4 February 2024).	Raster, 30 m × 30 m
Population Spatial Distribution		Raster, 1 km × 1 km
Watershed		shapefile
River Network		shapefile
Annual Mean Precipitation		Raster, 1 km × 1 km
Annual Mean Temperature		Raster, 1 km × 1 km
Soil type		Raster, 30 m × 30 m
Soil clay content		Raster, 1 km × 1 km
Soil organic carbon content		Raster, 1 km × 1 km
Nighttime Light Data	Harvard Dataverse (https://dataverse.harvard.edu/dataverse/harvard/?q=nighttime+light) (accessed on 12 February 2024).	Raster, 30 arc s
DEM (Digital Elevation Model)	Geospatial Data Cloud (https://www.gscloud.cn/) (accessed on 12 February 2024).	Raster, 30 m × 30 m
NDVI (Normalized Difference Vegetation Index)	NASA LP DAAC at the USGS EROS Center (https://lpdaac.usgs.gov/products/mod13a1v006/) (accessed on 12 February 2024).	Raster, 500 m × 500 m
Evapotranspiration Data	https://doi.org/10.5194/essd-11-1931-2019 (accessed on 13 February 2024).	Raster, 1 km × 1 km
China Bedrock Depth Dataset	http://globalchange.bnu.edu.cn/research/cdtb.jsp (accessed on 13 February 2024).	Raster, 100 m × 100 m
Available Water Capacity (SoilGrids 2017 AWC)	https://data.isric.org/geonetwork/srv/eng/catalog.search#/metadata/e33e75c0-d9ab-46b5-a915-cb344345099c (accessed on 14 February 2024).	Raster, 250 m × 250 m

.

3. Methods

Figure 2 illustrates the technical process of this study. Firstly, by collecting multi-period data on land use, soil, meteorological conditions, topography, and NDVI, this study analyzes the supply and demand of ecosystem services in the study area. Using the Nutrient Delivery Ratio, Annual Water Yield, and Habitat Quality modules of the InVEST model, the study examines the water purification, water supply, and habitat maintenance capabilities in the coastal areas of Fujian Province. Based on this analysis, the study constructs the Ecosystem Security Barrier Function (ESBF) index and examines its spatiotemporal variations. Finally, by comparing the OLS, GWR, and MGWR models, the study selects the optimal model to reveal the main influencing factors and driving mechanisms behind ESBF, then visualizes the results.

3.1. InVEST Modle

The InVEST model is an open-source tool for assessing ecosystem services, mainly used to evaluate the contributions of natural capital to human well-being. It assesses the benefits of various ecosystem services in mitigating natural disasters and maintaining ecological balance [24], providing scientific support for ecological protection policy and planning.

3.1.1. Habitat Quality

Habitat includes the resources and conditions necessary for the survival of individuals and populations. Habitat quality reflects the overall suitability and health of the environment where biological communities live. It indicates the biological support capacity and living conditions of the areas that biological communities depend on. Areas heavily disturbed by human activities pose threats to the ecological environment. In this study, paddy fields, drylands, rural residential areas, urban land, and other construction land are selected as threat factors. The formula for the habitat quality index is as follows:

$$Q_{xj}=H_j\left[1-\left({\displaystyle \frac{D_{xj}^{\mathrm{z}}}{D_{xj}^{\mathrm{z}}+k^2}}\right)\right]$$

(1)

where Q_xj is the habitat quality of grid x in land use type j, ranging from 0 to 1; H_j is the habitat suitability of land use type j; k is the half-saturation constant; and D_xj is the degree of habitat degradation. The parameter values for habitat quality are determined based on the model manual and relevant studies [23,25,26], as shown in

Table 2. Weights and maximum influence distance of threat factors.

Threat	Max_dist	Weight	Decay
Dry Land	3	0.6	Linear
Urban Land	5	0.7	Exponential
Paddy Field	3	0.6	Linear
Other Building Land	6	0.9	Exponential
Rural Settlements	10	1	Exponential

and

Table 3. Habitat suitability and sensitivity to threat factors for each land use type.

Land Use Types	HABITAT	Paddy Field	Dryland	Urban Land	Rural Settlements	Other Building Land
No Data	0	0	0	0	0	0
Paddy Field	0.4	0.3	0.3	0.5	0.4	0.5
Dry Land	0.4	0.3	0.3	0.5	0.4	0.5
Forested land	1	0.8	0.8	0.9	0.8	0.8
Shrubland	0.8	0.4	0.4	0.8	0.7	0.7
Open woodland	0.6	0.85	0.85	0.9	0.8	0.8
Other woodland	0.4	0.9	0.9	0.9	0.8	0.8
High Cover Grassland	0.7	0.4	0.4	0.6	0.5	0.6
Medium Cover Grassland	0.5	0.45	0.45	0.65	0.55	0.65
Low Cover Grassland	0.3	0.5	0.5	0.7	0.6	0.7
Rivers	0.9	0.65	0.65	0.85	0.75	0.8
Lakes	1	0.7	0.7	0.9	0.8	0.7
Reservoirs and Ponds	0.8	0.7	0.7	0.9	0.8	0.7
Mudflat	0.6	0.75	0.75	0.95	0.85	0.7
Beaches	0.6	0.75	0.75	0.95	0.85	0.7
Urban Land	0	0	0	0	0	0
Rural Settlements	0	0	0	0	0	0
Other Building Land	0	0	0	0	0	0
Marshland	0.9	0.7	0.7	0.8	0.75	0.6
Bare Land	0	0	0	0	0	0
Rocky Land	0	0	0	0	0	0
Ocean	0	0	0	0	0	0

.

3.1.2. Annual Water Yield

Water yield refers to the total volume of water resources acquired from natural rainfall and groundwater recharge over a specific period [27]. This metric is crucial for assessing the hydrological cycle and the efficient use of water resources, impacting agriculture, industrial production, and daily life. The formula for calculating water yield Y_(x) is as follows:

$$Y_{(x)}=(1-{\displaystyle \frac{AE{T}_{(x)}}{P_{(x)}}})\times P_{(x)}$$

(2)

where P_(x) represents the annual precipitation (mm) on grid cell x, and AET_(x) is the actual evapotranspiration (mm) for different land use types. The actual evapotranspiration is calculated as:

$${\displaystyle \frac{AE{T}_{xj}}{P_x}}={\displaystyle \frac{1+{\omega }_x{R}_{xj}}{1+{\omega }_x{R}_{xj}+1R_{xj}}}$$

(3)

$${\omega }_x=Z{\displaystyle \frac{PAW{C}_x}{P_x}}$$

(4)

$$Rxj={\displaystyle \frac{k_{xj}\times E{T}_0}{P_x}}$$

(5)

where R_xy is the Budyko dryness index for land use type j on grid cell x, and ω_x reflects the ratio of available water for vegetation to expected precipitation. Z is the Zhang coefficient, and K_ij denotes the ratio of crop evapotranspiration to reference evapotranspiration.

3.1.3. Nutrient Delivery Ratio

The NDR module quantitatively simulates nutrient sources and migration processes within a watershed, assessing the nutrient retention capacity (nitrogen and phosphorus) of different land covers. The outputs for total nitrogen (TN) and total phosphorus (TP) indicate the water purification function, with higher outputs signifying poorer purification capacity [28]. Input parameters for the model include Digital Elevation Model after depression filling, land use data, watershed boundaries, nutrient runoff proxy, threshold flow accumulation (default 1000), Borselli k (default 2), and a biophysical table. Annual average precipitation can substitute for the nutrient runoff proxy. Given that nutrient transport occurs across the entire watershed, input data spans several watersheds, including the Minjiang, Hanjiang, and coastal rivers of Zhejiang and Fujian. Post-modeling, results are masked and clipped to derive study-area-specific outcomes. Parameter settings for N and P nutrient load coefficients and retention efficiencies are detailed in

Table 4. Biophysical table of water purification module.

Landuse Types	Load_p *	eff_p *	Crit_len_p *	Load_n *	eff_n *	crit_len_n *
Cultivated land	1.22	0.35	25	19.4	0.25	25
Forest land	0.15	0.8	300	2.12	0.72	300
Grassland	0.2	0.48	150	3.2	0.4	150
Water area	0.01	0.05	150	0.01	0.05	150
Construction land	2.1	0.05	10	12	0.05	10
Unuseed land	0.05	0.05	150	1.45	0.05	10

* load_n and load_p represent the N/P output load coefficients, respectively, with units in kg/(hm²·a). eff_n and eff_p denote the maximum retention efficiency of N and P by vegetation. crit_len_N and crit_len_p indicate the maximum distance for retaining N and P nutrients for each type of land use, measured in meters (m).

, based on geographic similarity and model manual references [29,30,31].

3.2. Ecological Security Barrier Function Index

To evaluate the comprehensive barrier function of the coastal areas of Fujian, we define the Ecological Security Barrier Function (ESBF) Index (I_sf) based on habitat maintenance, water supply capacity, and water purification. This index ranges from 0 to 1, with higher values indicating a stronger ESBF and greater protection for affected areas. The calculation of I_sf is as follows:

$$I_{sf}={\displaystyle \sum _{i=1}^n{I}_{sf,std}\times {\omega }_i}$$

(6)

where I_sf is the barrier function index, I_sf,std is the standardized value of the i-th barrier function, and w_i is the weight of the i-th barrier function.

The Analytic Hierarchy Process (AHP) was used to determine the weights. First, by constructing the judgment matrix, the largest eigenvalue, λ_max, was calculated to be 4.0036. The consistency ratio was 0.0013, which is less than 0.1, indicating that the consistency test was satisfied. The final weights were determined as follows: habitat quality (0.2968), water yield (0.2605), TN (0.2201), and TP (0.2226). Given the positive and negative impacts of different indicators on the ecological environment, habitat quality and water yield are set as positive indicators, while TN and TP outputs are negative indicators. Standardization is performed using Equations (7) and (8):

$$X_{ij}-\min (X_i)/\left(\max (X_i)-\min (X_i)\right)$$

(7)

$$\max (X_i)-X_{ij}/\left(\max (X_i)-\min (X_i)\right)$$

(8)

where X_ij represents the j-th indicator in the i-th sample, and Max X_j and Min X_i denote the maximum and minimum values of the j-th indicator, respectively.

3.3. Spatial Heterogeneity Analysis

In this study, we employed Moran’s I index to evaluate the spatial autocorrelation of the ESBF distribution in the coastal areas of Fujian, indicating the spatial clustering trend [32]. Additionally, we utilized the Getis-Ord G index [26] to reveal the distribution characteristics of local spatial clustering, accurately identifying local clustering areas, namely “Hotspots” (areas with high ecological quality) and “Coldspots” (areas with lower ecological quality). This facilitated further analysis of driving factors to understand the internal mechanisms of the ecological environment.

3.4. Multiscale Geographically Weighted Regression (MGWR) Model

Based on literature from regions with similar natural conditions and socio-economic backgrounds [13,33,34,35], we selected nine representative factors from five aspects: climatic conditions, topography, vegetation, soil and human activities. These factors include elevation, slope, annual average temperature, annual average precipitation, NDVI, population density, nighttime light intensity, soil organic carbon content, clay content, proportion of construction land, and the proportion of cultivated land. Using ArcGIS software, we created a 2 km × 2 km fishnet grid for the spatial sampling of the ecological barrier function index and various indicators, collecting a total of 8108 sample data points.

We applied the Multiscale Geographically Weighted Regression (MGWR) model, which integrates geographic space with morphological characteristics, accounting for spatial differences and reflecting relationships between data at different locations [18]. This model helps to precisely identify the main driving factors affecting the ESBF. The linear regression model is presented as follows:

$$Y_i={\beta }_0(U_i,V_i)+{\displaystyle \sum j{\beta }_{bwj}}(U_i,V_i)X_{ij}+{\epsilon }_i$$

(9)

where Y_i represents the ESBF index, X_ij denotes the indicator factors, (U_i,V_i) are the geographical coordinates of the sample points, ε_i represents the random error, and β_bwj is the local regression coefficient. The larger the absolute value of this coefficient, the stronger its impact on the ecological barrier function index.

4. Results

4.1. Dynamic Change of Land Use

As a key factor in the evolution of the ecological environment, land use types in Fujian’s coastal areas vary significantly with altitude. As shown in Figure 3a and Figure 4, forests and cultivated land dominate the regional land use structure. Forests are primarily found in the northeastern and southwestern parts of the study area, while grasslands are mostly scattered in the coastal plain areas. Over the past 20 years, both forests and grasslands have experienced degradation, with decreases of 1.03% and 0.77%, respectively. Cultivated land is mainly concentrated in the coastal plains and has shown a continuous decline, with a total reduction of 4.51%. Construction land is mainly in the central coastal areas. As a key area of China’s reform and opening-up, cities like Xiamen and Fuzhou have experienced rapid socio-economic development leading to a significant expansion of urban land, which increased by 5.96% over 20 years, contributing to the reduction of cultivated land.

4.2. Habitat Maintenance Services

The results from the InVEST model indicate that the average habitat quality in 2000 was 0.527, followed by 0.472 (2005), 0.501 (2010), 0.484 (2015), and 0.481 (2020). Between 2000 and 2020, the average habitat quality decreased by 0.046, indicating an overall declining trend in habitat quality in the study area. Using the equal interval method [36], we subdivided habitat quality into five levels (Figure 3b and Figure 5), revealing significant spatial differences. High-quality habitats are mainly located in the northeastern regions, whereas the central coastal plain areas have relatively lower habitat quality. Overall, the spatial distribution pattern of habitat quality shows higher quality in the north and west and lower quality in the south and east. Specifically, the “Poor” category occupies the largest proportion, primarily in the densely populated central coastal areas. Although the area of the “Poor” category has contracted over the past 20 years, the “Bad” category increased by 6.1%, the “High” quality area decreased by 5.97%, and the medium quality areas remained relatively stable, leading to an overall decline in habitat quality. To improve the habitat quality across the study area, more focus should be placed on the low-quality regions.

4.3. Water Supply Service

The water yield is an important parameter representing hydrological and ecological processes such as land cover, precipitation, evapotranspiration, and runoff. Greater water yield depth indicates stronger water production capacity in the region [28]. As shown in Figure 6, the spatial differentiation pattern of water yield depth in the coastal areas of Fujian Province has remained roughly the same over different periods, generally decreasing from northeast to southwest. Over the past 20 years, the maximum water yield depth initially increased and then decreased, reaching a peak of 1657.97 mm in 2010. Spatially, there is significant heterogeneity in water yield depth, with high-value areas mainly concentrated in the northwest of the study area. This distribution pattern of water yield depth is significantly related to local precipitation and temperature distribution patterns [27]. The precipitation intensity and total precipitation in the study area decreased gradually from north to south [37], while the temperature increased gradually from north to south. Compared with the southern part of the study area, the precipitation in the northern part of the study area is large, the temperature is relatively low, and the increase is weak, so the water production in the northeastern part is higher.

4.4. Water Purification Service

The NDR module simulates the output of total nitrogen (TN) and total phosphorus (TP) in the watershed, which reflects regional water purification capabilities. Higher TN and TP outputs indicate poorer water purification capacity [29]. Figure 7 and Figure 8 show the output of TN and TP per unit area in the study region. Throughout the study period, the central region consistently had high TN outputs, while the northern and southern regions exhibited lower TN outputs. From 2000 to 2015, TN output showed a gradual increasing trend, peaking in 2015, before declining in 2020. The TP output pattern is similar to TN, with high TP output areas mainly in the central and southern regions. This suggests that the water purification capacity of the coastal region decreases from inland areas towards the coastline. The central coastal plain, a primary area for farmland and cities, experiences increased nitrogen and phosphorus output due to agricultural fertilization and urban sewage. Conversely, the northern region, mostly covered by forests and grasslands, has less nitrogen and phosphorus transport due to vegetation absorption and fixation.

4.5. Dynamic Change Analysis of Ecological Safety Barrier Function (ESBF)

The average ESBF index for the coastal region of Fujian Province was 0.6042 (2000), 0.5929 (2005), 0.5971 (2010), 0.5873 (2015), and 0.5631 (2020), indicating a gradually decreasing trend. The ESBF is categorized into five levels with intervals of 0.2, as depicted in Figure 9. The ESBF in Fujian’s coastal region is primarily classified into “Moderate” and “Good” levels, with the “Good” level being the most widespread, accounting for over 42.864%. However, its proportion has decreased from 52.168% in 2000 to 42.864% in 2020. In contrast, the “Moderate” range has slightly increased. Notably, although the “Poor” level has risen, the proportions of the “Bad” and “Poor” levels remain low, indicating that the ecological environment is generally stable.

The spatial distribution of the ESBF has remained relatively consistent, with high-value areas mainly in the northwest, while the central coastal areas exhibit weaker ecological barrier functions (Figure 10). To analyze the evolutionary trends of the ESBF, we assigned index levels from 1 to 5, in ascending order, and calculated the differences for each area. Significant, moderate, slight, and no change in ESBF are indicated by absolute differences of ≥3, 2, 1, and 0, respectively. Figure 11 shows that from 2000 to 2010, most areas showed no change in ESBF, with 13.3% showing an increase and 12.53% showing a decrease, primarily with slight changes. From 2010 to 2020, the areas with degraded ESBF increased, mainly concentrated in the northern part of the study area, while areas with improved ESBF were relatively fewer. Overall, from 2000 to 2020, 73.77% of the areas showed no change in ESBF, with a slight degradation trend, and areas with moderate to high changes accounted for less than 5%.

Examining the specific transfer directions and quantities of ESBF levels (Figure 12) shows that different degrees of transitions occurred between various levels, mainly concentrated between the “Good”, “Moderate”, and “High” levels. Approximately 1113.8 km² of “High” level areas transitioned to “Good”, followed by a degradation of 1086 km² from a “Good” to “Moderate” ESBF level.

4.6. Analysis of the Driving Factors of ESBF

4.6.1. Spatial Autocorrelation Analysis

The MGWR model excels in capturing the variations in relationships between variables across different geographical locations, making it effective for evaluations with strong spatial heterogeneity. Using ESBF sample data from a 2 km × 2 km grid, spatial autocorrelation analysis was performed with ArcGIS software. The global spatial autocorrelation results show that the Moran’s I for ESBF in five periods are: 0.5649, 0.6179, 0.6173, 0.5773, and 0.5472, with Z-values all exceeding 175 and P-values less than 0.01, indicating strong spatial autocorrelation and obvious spatial clustering of ESBF. Figure 13 illustrates the spatial clustering of ESBF in the coastal areas of Fujian Province, highlighting that hot spots (high-value clusters) are mainly in the mountainous northern regions and offshore central areas, while cold spots (low-value clusters) are mostly in the coastal plains of the central and southern regions. Over time, these spatial patterns of cold and hot spots have remained relatively stable.

4.6.2. Analysis of Driving Factors Based on the MGWR Model

If the selected indicators have multicollinearity, it can lead to unstable regression coefficients and reduce the explanatory power of the results. Multicollinearity is typically indicated by a tolerance (Tol) value less than 0.1 or a variance inflation factor (VIF) greater than 10. For the eleven selected indicators, OLS calculations showed a maximum VIF value of 3.94 and Tol values greater than 0.25, indicating no multicollinearity issues (

Table 5. Collinearity Diagnostics and Model Results Comparison.

Parameters	Collinearity Diagnostics		MGWR (Bandwidth: 71–8107)			GWR (Bandwidth: 295)			Globle OLS
	Tol	VIF	Min	Mean	Max	Min	Mean	Max
Temperature	0.25	3.94	−0.16	−0.02	0.20	−0.56	−0.05	0.45	−0.12
Precipitation	0.40	2.51	0.04	0.31	1.02	−0.20	0.34	1.76	0.31
Population	0.77	1.30	−0.31	−0.07	0.12	−0.89	−0.06	0.79	−0.01
NDVI	0.38	2.66	−0.13	0.22	0.72	−0.35	0.04	0.57	0.02
Night time light	0.47	2.15	−1.34	−0.17	0.20	−1.31	−0.15	1.42	−0.09
Construction Land Ratio	0.42	2.37	−0.40	−0.15	0.01	−1.15	−0.16	0.22	−0.14
Cultivated Land Ratio	0.68	1.48	−0.61	−0.28	−0.04	−0.62	−0.29	−0.07	−0.31
Slope	0.60	1.67	0.11	0.16	0.20	−0.16	0.07	0.33	0.07
Elevation	0.45	2.23	−0.25	0.14	0.33	−0.22	0.18	1.00	0.11
Soil Clay Content	0.96	1.04	−0.09	0.03	0.21	−0.17	0.03	0.43	0.03
Soil Organic Carbon Content	0.73	1.37	−0.32	−0.02	0.12	−0.45	−0.03	0.20	−0.05
R2	-	-	0.758	-	-	0.769	-	-	0.691
Adjusted R2	-	-	0.744	-	-	0.750	-	-	0.690
AICc	-	-	12,454	-	-	12,503	-	-	13,240

).

To ensure the representativeness of the results, average values of ESBF and each indicator over five periods were calculated. Using these average samples, MGWR, GWR, and OLS models were compared for their data fitting capabilities (Table 5). The R² and Adjusted R² values showed similar performance for MGWR and GWR, with GWR performing slightly better and OLS relatively worse. The AICc value, which is often used to measure the goodness of fit and complexity of statistical models, is smallest for the MGWR model, indicating it performs best in explaining the data [38]. The absolute values of the mean regression coefficients for each indicator are ranked as follows: Annual precipitation (−0.31) > Construction land ratio (−0.28) > NDVI (0.22) > Night time light (−0.17) > Slope (0.16) > Cultivated land ratio (−0.15) > Elevation (0.14) > Population (−0.07) > Soil clay content (0.03) > Temperature (−0.02) > Soil organic carbon content (0.02).

From a climate perspective, the regression coefficient for precipitation with ESBF ranged from 0.04 to 1.02, indicating a positive correlation. The central areas of the study region, including Hanjiang and Xianyou counties, showed the most significant impact of precipitation, while the northern and southern regions had weaker correlations. Figure 14 shows that temperature effects on ESBF varied significantly, with a positive effect in the northern regions and a negative effect in the southern regions.

Regarding topographical conditions, slope showed a positive correlation with ESBF, with regression coefficients between 0.11 and 0.20, indicating a relatively minor influence. The southernmost areas, such as Zhao’an, Yunxiao, and Dongshan counties, were the least affected by slope. Elevation also had a positive correlation with ESBF, with significant positive correlations in the central coastal cities of Jinjiang and Shishi (coefficients > 0.23), while the northernmost areas like the Fuding and Xiapu counties had weaker correlations (coefficients < 0.15). NDVI predominantly showed a positive correlation with ESBF, with regions of high NDVI impact (coefficients > 0.4) mainly in the northern areas like Fuding and Xiapu, as well as the central region of the study area.

From the perspective of human activity factors, population and nighttime light predominantly have a negative impact on ESBF. The negative impact of population is most evident in the southernmost areas like the Zhao’an and Dongshan counties (coefficients ≤ −0.2), while the spatial distribution of the regression coefficients for nighttime light is more dispersed. Construction land ratio and arable land ratio, representing the most intensive human activity land types, mainly show negative correlations with ESBF, with the most prominent negative correlations in the northern regions of the study area.

In addition, from the perspective of soil factors, soil clay content has a positive effect on ESBF in general, but it is mainly negative in the northern bay area. The soil organic carbon content in the south of the study area was mainly negative in Nanan District and Zhangpu County.

5. Discussion

5.1. Construction of Ecological Safety Barrier Function Index

Since the early 21st century, rapid socio-economic development in China’s coastal regions, including Fujian Province, has led to significant population influx, increasing pressure on land resources and ecosystems [22]. Fujian’s coastal area, located in southeastern China, acts as a vital barrier protecting inland regions from marine erosion and reducing the inflow of pollutants into the sea [21]. To address the ecosystem service demands of this barrier, we constructed an ESBF index for Fujian’s coastal area. This index evaluates the effectiveness of the ecological environment in resisting environmental pressures and human activities, providing a basis for formulating ecological protection and management strategies. Habitat quality, water purification, and water supply services independently play roles and collectively maintain and enhance the overall function of the ecosystem through complex interactions [13].

Habitat quality is a key indicator of ecosystem health, directly influencing its stability and resilience [36]. It is fundamental to constructing an ecological security barrier [39]. Compared to inland areas of Fujian Province, the coastal area’s habitat quality is moderate and declining [23,34]. High-intensity human activities have degraded habitats, particularly in the nearshore areas of Xiamen and Quanzhou, where the quality is poor compared to offshore inland areas [17].

Healthy ecosystems provide stable water resources through functions such as water conservation and hydrological cycle regulation [40]. Ecosystems with high water yield also tend to have high biodiversity and ecological health. Over the past 20 years, water yield in Fujian’s coastal area has decreased from north to south, with high-value areas in mountainous regions with abundant precipitation and high altitudes. While vegetation and soil regulate the hydrological cycle by intercepting precipitation, reducing surface runoff, and increasing soil infiltration [41], these areas exhibit lower water yield capacity. However, densely vegetated areas contribute to long-term ecosystem stability and health.

Recent rapid socio-economic development in coastal areas has led to pollution levels exceeding the self-purification capacity of land and water, causing eutrophication and frequent red tides [16,42]. Maintaining robust water purification capabilities enhances ecosystem resistance to environmental changes and human disturbances, improving resilience and aiding marine ecosystem protection [43]. The water purification capacity in Fujian’s coastal area decreases from the inland towards the coastline. Arable lands have higher N/P output levels, with soil absorbing some nutrients, but fertilization also leads to non-point source pollution [44]. Land use changes and river confluence effects transport agricultural runoff, industrial emissions, and urban sewage to coastal plains, reducing water purification capacity. Fujian’s abundant coastal wetlands, mainly in estuaries and bays, are crucial for water purification and habitat maintenance. However, the rate of wetland loss has greatly increased in recent years, with wetlands becoming increasingly fragmented [3]. Strengthening the restoration and management of coastal wetlands is key to enhancing the ecological security barrier [45].

The “Ecological Restoration Planning for Fujian Province Land Space (2021–2035)” emphasizes the need to control pollutant discharge into the sea, prevent marine disasters like red tides, and improve ecosystem quality to establish a robust ecological security barrier system. This plan supports constructing such a barrier from a macro perspective. From 2000 to 2020, the ESBF index shows a declining trend, with higher values in the north and lower values in the south, indicating significant spatial autocorrelation. The dynamic balance of coastal wetlands’ ecosystem services functions contributes to water purification and habitat quality improvement [46,47]. Limiting urban expansion, reducing forest and arable land loss due to urban growth, and strictly controlling pollutants entering rivers and the sea can enhance the ESBF’s effectiveness [45].

5.2. Analysis of Driving Mechanism

The Ecological Security Barrier is vital for protecting the environment from external disturbances and ensuring ecosystem stability and sustainability. Understanding the factors influencing ESBF is essential for comprehending its formation mechanisms and spatial variations. Unlike traditional spatial analysis models like GWR and OLS [48], we employed the MGWR model to analyze the impact of eleven major factors on ESBF, providing detailed spatial analysis and visualization of local regional characteristics.

Global warming, driven by human activities, accelerates the water cycle and causes abnormal precipitation, impacting ecosystem services [49]. In the study area, precipitation is positively correlated with ESBF, showing a gradient from coastal to inland regions, with the most sensitive areas in the central part. The impact of temperature on ESBF varies significantly between the north and south, with a positive correlation in the north and a negative correlation in the south.

Elevation changes significantly affect temperature conditions, limiting vegetation growth and distribution, thus impacting the ecosystem [50]. Slope, an important topographical indicator, influences urban planning and nutrient diffusion [17]. Both elevation and slope have mainly positive impacts on ESBF in the study area. NDVI, which measures surface vegetation greenness, reflects regional vegetation growth and coverage [33]. In the central region of the study area, the regression coefficients of NDVI show a distribution pattern of high values surrounding a low center. High-value areas (coefficients > 0.4) are distributed in the forested and grassy northern and central regions, displaying a strip-like distribution. This indicates that vegetation plays a crucial role in maintaining ecological environment quality and constructing habitat barriers. Vegetation supports and improves regional ecological security by enhancing biodiversity and regulating hydrological cycles and climate conditions [51,52].

Located in China’s southeastern coastal zone, the study area has developed economies and convenient transportation [53]. Since 2000, thanks to its advantageous geographical location and policy conditions, the industrial sector has rapidly developed, attracting a large influx of population. Since 2000, industrial sector development and population influx have led to rapid urban expansion, increasing the demand for transportation and industrial land and reducing arable land. Urbanization indicators, such as construction land ratio [20] and nighttime lights [54], reflect human activity intensity and have predominantly negative effects on ESBF, especially in the northern regions where ESBF is more sensitive.

In summary, the ESBF is weaker in coastal plain areas, showing a limited ability to resist external disturbances. This is influenced by frequent human activities and topographical conditions like elevation and slope. Coastal areas, densely populated with numerous industries, require future efforts to enhance coastal protective forests, expand afforestation, protect estuarine regions, improve wetland ecological functions, and strengthen ESBF. These measures will reduce marine disaster impacts on inland areas and minimize pollutants entering the sea, thereby improving the marine ecological environment.

5.3. Limitations

Although this study has delved deeply into the practical application of constructing coastal ecological security barriers, several issues remain unresolved due to objective limitations. Prior research on ecological security barriers has primarily been theoretical, with few practical application studies. This paper approaches the construction of ESBF from the perspective of ecosystem supply and demand. However, building an ecological security barrier is a vast subject, and future research can explore this system from more perspectives. Furthermore, Fujian’s coastal areas are significantly impacted by marine disasters such as typhoons. Due to difficulties in obtaining data, factors like typhoons and wind speeds were not included in this study’s driving factor analysis. Future research should pay more attention to these issues.

6. Conclusions

This study constructs the Ecological Security Barrier Function (ESBF) index based on the InVEST model, emphasizing the supply and demand of ecosystem services such as habitat quality, sea pollutant reduction, flood mitigation, and water purification. We conducted an in-depth analysis of the spatial distribution and changing patterns of this index. Additionally, the MGWR model was utilized to examine the spatial factors influencing ecological security barriers at a grid scale, offering valuable insights for the sustainable development of coastal ecological environments.

Urban expansion increased construction land demand, significantly reducing arable land in the study area. Over the past 20 years, the average habitat quality index of the study area was 0.492, indicating a medium level, with a slight downward trend in habitat maintenance function. The region’s water resource supply capacity initially increased and then decreased, peaking in 2010. Water purification capacity weakened yearly, reaching its lowest in 2015 before showing some improvement.

The mean ESBF index gradually declined, generally ranging from good to moderate levels. The ESBF index was higher in the northwest and lower in the southeast. Over the years, frequent transitions between ESBF levels occurred, with mild degradation in some areas. However, 73.77% of the region maintained stable ESBF levels, indicating overall ecological stability.

The spatial distribution of ESBF exhibited strong spatial autocorrelation, with distinct cold and hot spots. The driving mechanisms of ESBF in Fujian’s coastal areas showed significant spatial heterogeneity due to factors such as human activities, topography, vegetation, and climate conditions. Overall, precipitation, slope, elevation, and NDVI correlated positively with ESBF, whereas population, construction land ratio, arable land ratio, and nighttime lights correlated negatively with ESBF. The relationship between temperature and ESBF varied, being positive in the north and negative in the south. Soil clay content and soil organic carbon content had relatively little influence on ESBF, and were mainly positively correlated. Effective measures to maintain and gradually enhance regional ESBF stability include strengthening coastal wetland restoration, enhancing coastal protection forest systems, limiting urban expansion, and controlling pollutants entering the sea.