Prediction and Spatiotemporal Transfer of Vegetation Vulnerability in the South African Coastal Zone Under Different Shared Socioeconomic Pathway (SSP) Scenarios

Abstract

Against the background of the rapid transformation of traditional economies and societies and continuous global climate change, how to ensure the long-term stability of the coastal ecological environment has become a key issue to be studied. In this paper, we take the 20 km buffer zone extending inland from the South African coastal zone as the study area. By constructing a vegetation vulnerability evaluation system, the current and future scenarios are compared in depth based on the base period (2010–2020), the near term (2030–2059), and the long term (2070–2099) with the help of GIS spatial analysis, the Moran index, and other methods. The results show that there are obvious spatial differences in vegetation vulnerability in the South African coastal zone. The extremely vulnerable areas of vegetation are mostly distributed on the west coast of South Africa, and some areas have obvious high–high aggregation patterns. The transfer of SSP1-2.6 scenarios in the near term is relatively stable, and the vegetation vulnerability level rebounds significantly in the long term; the vulnerability level of SSP2-4.5 scenarios has increased in both the near term and the long term, indicating that the risk of vegetation vulnerability has increased; while the SSP5-8.5 scenario has a significant deterioration trend in the long term, and the risk of vegetation vulnerability shifting to a high vulnerability level has increased significantly. Land use type has a significant impact on the response of vegetation vulnerability to SSP prediction. In the process of transformation from the base period to the long term, the proportion of vegetation vulnerability shifting to extremely vulnerable and severely vulnerable levels is notably high for both cultivated land and forest land—particularly under high-emission scenarios, driven by agricultural intensification for cultivated land and climate stress for forest land. This paper deeply explores the spatiotemporal evolution law and driving mechanism of vegetation vulnerability in the South African coastal zone under different shared socioeconomic pathway (SSP) scenarios, providing decision support for better development and protection of the South African coastal zone in the future.

1. Introduction

The South African coastal zone, as a critical land–sea interaction area with unique ecological value and high economic dependence, is facing dual pressures from global climate change and intensified human activities; these pressures have directly amplified the vulnerability of coastal vegetation ecosystems, making it a core issue for regional sustainable development research.

Against this background, the vegetation vulnerability of the South African coastal zone—an indicator reflecting the stability of coastal ecosystems—has become a key focus, as it directly links climate change impacts, human activities, and ecological security. However, the sustainable development of this region is facing unprecedented challenges. The rate of sea level rise caused by global climate change has reached 3–4 mm per year in the coastal zone of South Africa. Some low-lying areas have already experienced seawater intrusion and salinization. The IPCC’s Sixth Assessment Report pointed out that if greenhouse gas emissions remain at the current level (SSP5-8.5 scenario), the coastal areas of South Africa may face a sea level rise of 0.6–1.1 m by 2100, directly threatening coastal infrastructure and community safety [1]. At the same time, the intensification of human activities has further amplified ecological risks: in the past 30 years, the population of the coastal zone of South Africa has increased by more than 40%, the urbanization rate has climbed from 55% to 68%, and the expansion of construction land has caused the natural vegetation coverage rate to decrease by 12% [2]; port-related industries and mining activities have aggravated water pollution and soil erosion, and over-exploitation of agriculture in the Western Cape Province has caused 40% of wetland functions to degrade [3]. These pressures have significantly increased the vegetation vulnerability of the coastal zone and manifested in biodiversity loss, vegetation fragmentation, and weakened ecological service functions—issues that lack systematic scenario-based prediction under SSPs.

As a frontier area of land–sea interaction, the South African coastal zone has long been studied, with a focus on natural geographical features, land use dynamics, and economic structure analysis. Early studies were mostly based on remote sensing and GIS technology to explore the spatial differentiation of coastal land use patterns and their driving mechanisms. For example, South Africa’s National Coastal Zone Integrated Management Act defines the scope of the coastal zone (a 20 km buffer zone extending inland) in legal form, and emphasizes the ecological functions of the protection zone and buffer zone, providing a spatial framework for subsequent research [4]. Empirical studies based on multi-temporal remote sensing images show that land use changes in the South African coastal zone from 2000 to 2020 were mainly grassland shrinkage and forest expansion, while cultivated land and unused land decreased by 22.25% and 4.36%, respectively, and construction land expanded significantly in port cities such as Cape Town and Durban [5]. This change is closely related to climate fluctuations, population growth, and policy orientation, but the specific impact of these driving factors on vegetation vulnerability remains unquantified [6].

Internationally, coastal land use research has gradually shifted from static pattern analysis to dynamic simulation and multi-scenario prediction. For example, a study on the China–Mongolia–Russia Economic Corridor revealed the spatiotemporal heterogeneity of forest loss and construction land expansion by integrating multi-source remote sensing data and driving factors (GDP and population density) [7]. South African scholars found through the land use transfer matrix that the mutual conversion between grassland and forest land was dominant (accounting for 38%), and the development intensity showed a ‘core-edge’ spatial differentiation, with high-value areas concentrated in port city clusters [8]. However, existing South African studies mostly rely on historical data retrospection and are insufficient in predicting future scenarios under the coupling of climate change and human activities—especially lacking in the in-depth integration of shared socioeconomic pathways (SSPs) for vegetation vulnerability simulation.

As the core scenario framework of CMIP6, SSPs provide a methodological breakthrough for vegetation vulnerability assessment at global and regional scales by quantifying the multi-dimensional interactions of population growth, economic development and technological progress. For example, in a study on the Mara River Basin in Africa, scholars combined SSP1-2.6 (sustainable development path) with SSP5-8.5 (high emission path) and found that the latter would lead to a 9.5-fold increase in the vegetation vulnerability index and a 40% expansion of the extremely vulnerable area [9]. Similarly, a global coastal zone study predicted through SSP scenario simulation that a sea level rise of 0.6–1.1 m in 2100 would submerge 15% of South Africa’s coastal wetlands, threatening mangroves and migratory bird habitats [10]. However, existing applications of shared socioeconomic pathways (SSPs) are mostly concentrated at large river basin or national scales; due to the unique land–sea interactions and local socioeconomic differences in the South African coastal zone, a systematic vegetation vulnerability scenario simulation system has not yet been formed in the region, especially in terms of the spatiotemporal transformation laws under multiple future paths.

The limitations of current research on South African coastal vegetation vulnerability are mainly reflected in three aspects. First, the data resolution and model adaptability are insufficient: small-scale coastal SSP simulation is often limited by the spatial resolution of meteorological data [11]. Second, the coupling mechanism between natural and anthropogenic factors is unclear: existing models mostly analyze climate factors and anthropogenic factors independently, ignoring their nonlinear interaction effects. For example, the Benguela cold current and mining development on the west coast of South Africa jointly lead to wetland degradation, but the synergistic mechanism between the two factors (Benguela cold current and mining development) has not yet been quantified [12]. Finally, there is a lack of research on the trade-off between ecological protection and economic development: although the ‘Fejisa Plan’ proposed a target for the growth of marine economic output, the conflict simulation between port expansion and vegetation protection under the SSP scenario is still lacking, and it urgently needs to be supported by a multi-objective optimization model [13]. Therefore, targeting the vegetation vulnerability of the South African coastal zone, this study aims to fill the gaps in multi-scenario simulation and spatiotemporal transition analysis under SSPs—laying a foundation for ecological protection and sustainable development decision-making.

In view of the above shortcomings, this study takes SSP scenarios as the core to construct a multi-scenario prediction framework for South African coastal vegetation vulnerability and systematically evaluate its spatiotemporal evolution. By integrating CMIP6 climate data, LUH2 high-resolution land use simulation, and socioeconomic forecasts, a multi-dimensional evaluation system is built. The specific objectives of this study are as follows: (1) to optimize the vegetation vulnerability evaluation system for the South African coastal 20 km inland buffer zone; (2) to predict the spatiotemporal pattern of vegetation vulnerability under SSP1-2.6, SSP2-4.5, and SSP5-8.5 scenarios (near term: 2030–2059; long term: 2070–2099) via GIS spatial analysis and Moran’s I/LISA; (3) to reveal the transfer characteristics of vulnerability levels and their coupling with land use change; and (4) to identify ecological risk hotspots and propose adaptive management strategies aligned with South Africa’s National Coastal Integrated Management Act. Based on these objectives, this study aims to answer the following key questions: How does coastal vegetation vulnerability’s spatial differentiation change under different development paths? Which areas will become ecological risk hotspots? How can economic development and ecological protection be balanced via adaptive management?

The research results will provide targeted data support for revising South Africa’s National Coastal Integrated Management Act and offer a paradigm for the sustainable development of similar coastal zones globally, with far-reaching scientific and practical significance.

2. Materials and Methods

2.1. Overview of the Study Area

The Republic of South Africa (hereafter referred to as ‘South Africa’) is situated at the southernmost tip of the African continent, spanning 16°45′ E~32°89′ E and 26°80′ S~34°83′ S. It has a land area of 1,221,037 km², bordered by the Indian Ocean to the east, the Atlantic Ocean to the west and south, and a coastline of 2798 km. Land borders include Namibia, Botswana, Lesotho, Zimbabwe, Mozambique, and Swaziland. Administratively, South Africa comprises nine provinces, among which the study focuses on four coastal provinces: the Northern Cape, Western Cape, Eastern Cape, and KwaZulu-Natal.

Coastal zone definitions vary globally depending on research objectives. Internationally, frameworks such as the U.S. Coastal Zone Management Act (1972) and the International Geosphere–Biosphere Program (IGBP) emphasize land–sea interaction but lack uniformity. In China, coastal zones are often defined as inland buffers (5–40 km from the coastline) [14]. For this study, the coastal zone delineation follows South Africa’s National Coastal Integrated Management Act (which includes coastal public property, protection zones, and special management areas) and aligns with IGBP’s land–sea interaction framework. Specifically, the study area is a 20 km inland buffer zone along South Africa’s coastline (16°45′ E~32°89′ E, 26°80′ S~34°83′ S), covering a total area of 53,862.0588 km². This 20 km buffer fully encompasses all coastal zone categories specified in South Africa’s official legislation [15].

Notably, the South African coastal zone is an ecologically sensitive land–sea transition area with prominent ecological and economic value. Ecologically, it hosts unique biomes, such as the Cape Floral Kingdom—the world’s smallest plant kingdom with more than 9000 plant species, 70% of which are endemic species [16]; the iSimangaliso Wetland Park is listed as a World Heritage Site by UNESCO and is an important habitat for mangroves, coral reefs, and endangered sea turtles [17]. From an economic perspective, South Africa’s coastal zone is the lifeblood of the country’s development: eight major ports, including the Port of Durban and the Port of Cape Town, handle 95% of South Africa’s international trade, with an annual throughput of over 180 million tons [18]; coastal tourism contributes 74% of the country’s tourism revenue and attracts millions of international tourists [19]; and Saldanha Bay on the west coast is Africa’s largest oil and gas hub, while the Port of Richards Bay on the east coast is an important global coal export base [20].

Topographic features such as the Cape Mountains and Drakensberg Mountains modulate the local climate, creating spatial heterogeneity in vegetation vulnerability (Figure 1). Based on ArcGIS Pro 3.1 spatial analysis of existing data, the 2020 land use structure of the study area is dominated by three types (Figure 1): grassland (23,864.7537 km², 44.31%), concentrated in the Western Cape (41.16%) and Northern Cape (21.51%); forestland (12,642.6654 km², 23.47%), primarily distributed in the Eastern Cape (51.11%), KwaZulu-Natal (23.98%), and the Western Cape (24.90%) (with almost no distribution in the Northern Cape); and cropland (11,982.834 km², 22.25%), widely present in the Western Cape (50.56%), Eastern Cape (31.91%), and KwaZulu-Natal (17.08%) (scarce in the Northern Cape). Secondary land use types include unused land (4.36%, mainly in the Northern Cape, 61.60%), construction land (4.36%, concentrated around port cities like Cape Town and Durban, with KwaZulu-Natal accounting for 41.55%), and water bodies (1.21%, predominantly in KwaZulu-Natal, 62.89%).

2.2. Research Methods

2.2.1. SSP Scenario Assumptions

Considering the needs of practical research, we selected 2011–2020 as the base period. When assuming future scenarios, we divided the short-term (2030–2059) and long-term (2070–2099) periods into two periods to predict the changes in the vulnerability of vegetation in the South African coastal zone in the future, and select three typical scenarios, SSP1-2.6 (low emissions), SSP2-4.5 (medium emissions), and SSP5-8.5 (high emissions), based on the SSP (shared socioeconomic pathway) framework to conduct an in-depth analysis of the vulnerability of vegetation in the South African coastal zone.

2.2.2. Construction of the Vegetation Vulnerability Assessment System

Indicator Selection

Vulnerability is affected by both natural and human factors. Natural factors are the material basis for vegetation growth and survival, including climate, soil, topography, water demand, etc. Human factors mainly refer to the external pressure exerted on nature by social and economic development. Based on the problems faced by vegetation in the South African coastal zone, we selected 5 natural factors and 2 human factors to construct a vegetation vulnerability assessment system for the base period, short term, and long term in the South African coastal zone (

Table 1. Vegetation vulnerability assessment system.

Aspect Index	Specific Indicators	Unit	Indicator Properties
Natural factors	Annual precipitation	Mm	burden
	Average annual temperature	°C	burden
	Soil type	-	just
	Average elevation	M	just
	Vegetation ecological Water requirement	m3	burden
Human Factors	Population density	Person/km2	just
	Land use degree	%	just

Note: The ‘land use degree’ indicator is calculated using the ‘comprehensive land use degree index method’ proposed by Yang et al. (2022) [21]. Weight values were combined with ecological zoning standards of South Africa’s coastal zone: cultivated land (0.1), woodland (0.2), grassland (0.3), construction land (0.8), unused land (0.6), water body (0), and the proportion of area were calculated using ESA land use data. ‘Soil type’ is assigned based on the humidity coefficient (0.12–0.85) of the World Soil Database (HSWD): sandy soil (humidity coefficient 0.12–0.25) corresponds to the standardized value of 0.1–0.3, loam soil (0.26–0.55) corresponds to 0.4–0.6, and clay (0.56–0.85) corresponds to 0.7–0.9. The higher the humidity coefficient, the stronger the soil and water conservation capacity and the lower the vulnerability.

).

Soil type reflects the type of soil. Different soil types have different soil and water conservation strengths and will have different degrees of impact on ecological fragility. Based on the soil moisture coefficient of the South African coastal zone [22], we use maximum value standardization to assign different positive values to soil types. The higher the moisture coefficient, the smaller the value.

Data Standardization

Since the units, numerical values, and indicator types of the variables in the original data are different, we standardize the data so that the processed variables show a normal distribution, which is convenient for subsequent calculations. The positive indicator Formula (1) and the negative indicator Formula (2) are calculated using different standardized calculation formulas:

$${\mathrm{Z}}_{\mathrm{i}}={\displaystyle \frac{{\mathrm{X}}_{\mathrm{i}}-{\mathrm{X}}_{\min }}{{\mathrm{X}}_{\max }-{\mathrm{X}}_{\min }}}$$

(1)

$${\mathrm{Z}}_{\mathrm{i}}={\displaystyle \frac{{\mathrm{X}}_{\max }-{\mathrm{X}}_{\mathrm{i}}}{{\mathrm{X}}_{\max }-{\mathrm{X}}_{\min }}}$$

(2)

In the formula, Z_i is the standardized value of the ith indicator; X_i is the ith original indicator value; X_max is the maximum value of the ith indicator; and X_min is the minimum value of the ith indicator. In this study, we implemented standardized procedures using SPSS 26.0 software to ensure that post-processing data adhered to normal distribution. Positive indicators (where higher values indicate lower vulnerability) included soil type and average elevation (corresponding to Formula (1)), while burden indicators (where higher values indicate greater vulnerability) comprised annual precipitation, average temperature, vegetation ecological water demand, population density, and land use intensity (corresponding to Formula (2)). This approach prevented confusion between indicator types that could have affected subsequent calculations.

Principal Component Analysis

The principal component analysis method can transform multiple variables with large correlation into a small number of variables with small correlation through orthogonal transformation, thereby achieving data dimension reduction and retaining the information reflected by the original large number of variables to the maximum extent. The calculation formula is as follows:

$$\mathrm{EVI} ={ \alpha }_1{\mathrm{F}}_1 + {\alpha }_2{\mathrm{F}}_2 + \dots + {\alpha }_{\mathrm{n}}{\mathrm{F}}_{\mathrm{n}}$$

where EVI is the vegetation vulnerability index; ${\alpha }_{\mathrm{i}}$ is the importance of the i-th feature; and F_i is the i-th feature. Prior to principal component analysis, suitability was validated through KMO and Bartlett’s test for sphericity: The KMO value reached 0.72 (≥0.7, indicating suitability for dimensionality reduction), while the Bartlett’s test for sphericity showed p < 0.001 (rejecting the assumption of variable independence). Principal components were extracted using values ≥ 1, ultimately yielding two principal components with cumulative variance contributions of 89.2% (first principal component: 65.7%; second principal component: 23.5%).

Vegetation Vulnerability Classification

The natural breakpoint method of ArcGIS Pro 3.1 was used to evaluate the vulnerability of vegetation in the coastal zone of South Africa and divide the results into four vulnerability levels according to the scores, namely, slightly vulnerable, moderately vulnerable, severely vulnerable, and extremely vulnerable.

2.2.3. Comparative Analysis of Spatiotemporal Dynamics

Based on ArcGIS Pro 3.1, the calculated vegetation vulnerability results were divided into four levels from high to low, namely, extremely vulnerable, severely vulnerable, severely vulnerable, and slightly vulnerable, using the natural breakpoint method. The vegetation vulnerability results under three SSP scenarios (SSP1-2.6/SSP2-4.5/SSP5-8.5) in the near term (2030–2059) and long term (2070–2099) were compared through geospatial visualization. The spatial pattern differences of vegetation vulnerability levels in the South African coastal zone under different scenarios and periods are shown in the form of thematic maps.

At the same time, GeoDA 1.22 software was used to set the QUEEN adjacency spatial weight matrix for the vegetation vulnerability results of each SSP scenario and different periods, and the global Moran’s index and local Moran’s index tests were performed. The local Moran’s index was used to draw the LISA map to show its spatial agglomeration characteristics.

The global Moran index calculation formula is as follows:

$$\mathrm{I} = {\displaystyle \frac{\mathrm{n}{\sum }_{\mathrm{i} = 1}^{\mathrm{n}}{\sum }_{\mathrm{j} = 1}^{\mathrm{n}}{\mathrm{w}}_{\mathrm{ij}}\left({\mathrm{x}}_{\mathrm{i}}-\stackrel{\mathrm{-}}{\mathrm{x}}\right)\left({\mathrm{x}}_{\mathrm{j}}-\stackrel{\mathrm{-}}{\mathrm{x}}\right)}{{\sum }_{\mathrm{i} = 1}^{\mathrm{n}}{\sum }_{\mathrm{j} = 1}^{\mathrm{n}}{\mathrm{w}}_{\mathrm{ij}}\sum _{\mathrm{i}=1}^{\mathrm{n}}{\left({\mathrm{x}}_{\mathrm{i}}-\stackrel{\mathrm{-}}{\mathrm{x}}\right)}^2}}$$

The local Moran index is calculated as follows:

$${\mathrm{I}}_{\mathrm{i}} = {\displaystyle \frac{\left({\mathrm{x}}_{\mathrm{i}}-\stackrel{\mathrm{-}}{\mathrm{x}}\right)}{{\sum }_{\mathrm{i} = 1}^{\mathrm{n}}{\left({\mathrm{x}}_{\mathrm{i}}-\stackrel{\mathrm{-}}{\mathrm{x}}\right)}^2/\mathrm{n}}}\sum _{\mathrm{j} = 1}^{\mathrm{n}}{\mathrm{w}}_{\mathrm{ij}}\left({\mathrm{x}}_{\mathrm{j}}-\stackrel{\mathrm{-}}{\mathrm{x}}\right)$$

where n is the number of samples in the study area; x_i and x_j represent the vegetation vulnerability values of region i and region j, respectively; $\stackrel{\mathrm{-}}{\mathrm{x}}$ is the mean value of vegetation vulnerability; and w_ij the QUEEN adjacency spatial weight matrix is used (adjacent region w_ij = 1; otherwise, w_ij = 0). The significance of spatial autocorrelation is judged by p < 0.05 and |z| > 1.96. The global Moran index can measure the overall spatial aggregation trend, and the LISA map drawn using the local Moran index can help identify hot spots of high–high aggregation, low–low aggregation, etc., so as to obtain the spatial aggregation characteristics and evolution laws of vegetation vulnerability based on different scenarios and different periods.

2.2.4. Transfer Feature Quantification Method

We quantify the area of vegetation vulnerability level transfer from the base period to the near and far periods under three different SSP scenarios with the help of the idea of the land use transfer matrix. Assuming that there are m vegetation vulnerability levels in the base period and n vegetation vulnerability levels in the near or far period, the transfer matrix T can be expressed as follows:

$$\mathrm{T} = \left[\begin{array}{ccc}{\mathrm{t}}_{11}& \cdots & {\mathrm{t}}_{1\mathrm{n}}\\ ⋮& \ddots & ⋮\\ {\mathrm{t}}_{\mathrm{n}1}& \cdots & {\mathrm{t}}_{\mathrm{mn}}\end{array}\right]$$

Among them, t_ij represents the number of pixels that transfer from the i-th vegetation vulnerability level in the base period to the j-th vulnerability level in the near or distant period. Combined with the pixel resolution (1 km), the number of pixels can be converted into the actual change area, that is, the transfer area is A_ij = t_ij × pixel resolution².

By analyzing the element distribution results of the transfer matrix, the transfer and change characteristics of vegetation vulnerability levels under different emission scenarios are summarized.

2.2.5. Coupling Analysis of Land Use Type and Vegetation Vulnerability Level

In order to further explore the driving factors of vegetation vulnerability level transfer, we use land use data and vegetation vulnerability level data for coupling analysis. Based on the global 1 km resolution land use prediction dataset (2020–2100) under the SSP-RCP scenario (covering six land use types: cultivated land, forest land, grassland, urban and rural/industrial and mining/residential land, unused land, and water bodies), a Sankey diagram is used to display the vegetation vulnerability level transformation path corresponding to its change from the base period to the near and far periods. In terms of node division, the land use type is set as the starting node, and the vegetation vulnerability level in the near and far periods is set as the end node. The number of pixels with changes in vulnerability level of each land use type during the transformation from the base period to the near and far periods is counted as the flow value. This result can show whether there is a risk and probability of increased vulnerability level in different land use types in the future.

In addition, in order to more specifically compare the proportions of vegetation vulnerability levels corresponding to the six land use types in the base period, short term, and long term under the three emission scenarios, and to intuitively display the vulnerability levels of different land use types, we used OriginPro 2021 to draw a stacked bar chart (unit: km²) of different land use types and their corresponding vegetation vulnerability levels under each scenario and period, using specific numerical changes to intuitively compare and reveal the impact mechanism of land use changes on vegetation vulnerability.

2.3. Data Source

Table 2. Data sources.

Data Category	Data Source for the Base Period (2011–2020)	Sources of Near-Term and Long-Term Data
meteorological	Use ERA5-Land monthly average reanalysis data to obtain temperature and precipitation data during this period. The vector boundary of the South African coastline was cut and invalid values of the ocean area were removed and resampled to 1 km resolution using the double linear interpolation method.	With reference to the CMIP6 meteorological data set, the CNRM-CM6-1-HR CMIP6 model data were selected after comprehensively considering the scope of the research area and the required data elements. These data cover key indicators such as net surface radiation, average temperature, and precipitation. Its spatial resolution of 0.5° × 0.5°. Resampled to 1 km resolution using the double linear interpolation method to remain consistent with the base period data.
Soil	Relying on the World Soil Database (HSWD), base period soil data is obtained, and the bilinear interpolation method is used to resample to 1 km resolution.	Same base period.
Elevation	The data provided by the STRM are used as the source of base period elevation information.	Same base period.
Population	The base period population density data comes from the WorldPop population density dataset.	According to the population growth trend given in the World Population Prospects Report, the average population density in the study area in the corresponding period was estimated based on the average natural population growth rate of 2.3%, 1% and 0.384% in Africa in 2020–2030, 2030–2059, and 2070–2099, respectively.
Land Use	The ESA land use data from 2011 to 2020 are used to determine the average land use status based on the land type with the largest proportion in each grid.	Using the annual land cover data for 2030–2059 and 2070–2099 under the SSP1-2.6, SSP2-4.5, and SSP5-8.5 shared pathways in the LUH2 dataset, the average land use in the near and long term was obtained using the same data processing logic as the base period.
Vegetation ecological water demand	The base period data are cited from [23].	Combining CMIP6 meteorological data, soil data, and elevation data, the random forest algorithm is used for simulation calculation to obtain the vegetation ecological water demand data in the near and long term.

shows the data sources used in constructing the vegetation vulnerability assessment system.

3. Results

3.1. Spatial and Temporal Pattern of Vegetation Vulnerability in the Base Period

During the base period (2011–2020), the vegetation vulnerability in the South African coastal zone exhibited significant spatial heterogeneity (Figure 2a). Quantitatively, the area proportions of vulnerability levels were as follows: extremely vulnerable (28.08%), severely vulnerable (42.58%), moderately vulnerable (22.45%), and mildly vulnerable (6.89%). Spatially, the vulnerability showed a distribution pattern of ‘low in the southeast, high in the southwest’: extremely vulnerable areas were concentrated in the southwestern coastal zone, while the southern region was dominated by mildly and moderately vulnerable areas; and severely vulnerable areas were sporadically distributed in the southeastern coast, yet moderately vulnerable areas still accounted for the majority.

To further analyze the spatial aggregation characteristics, the Local Indicators of Spatial Association (LISA) map based on Moran’s I index was employed (Figure 2b). The global Moran’s I value for vegetation vulnerability in the base period was 0.32 (p < 0.05), indicating a significant positive spatial autocorrelation (i.e., vulnerable areas tended to cluster). Specifically, the ‘high–high’ clustered areas (where both the focal area and its neighbors had high vulnerability) were mainly distributed in the southwest, occupying 18.7% of the total coastal zone area, showing a continuous belt-like extension. In contrast, the eastern and western regions exhibited no significant clustering (marked as ‘not significant’ in Figure 2b), with only scattered ‘low-low’, ‘low-high’, and ‘high-low’ clusters, reflecting large spatial differences and fragmented patches in these areas.

3.2. Spatiotemporal Evolution of Vegetation Vulnerability Under Different SSP Scenarios

Figure 3 illustrates the spatial distribution and area proportion changes in vegetation vulnerability levels in the South African coastal zone under different shared socioeconomic pathway (SSP) scenarios. From the near term (2030–2059) to the long term (2070–2099), the area dynamics of vulnerability levels vary significantly across scenarios, with quantitative details as follows:

1. SSP1-2.6 (sustainable development scenario): The area proportion of mildly vulnerable zones increases from 6.89% (base period) to 6.91% (near term) and further to 10.65% (long term). In contrast, the severely vulnerable zone decreases from 42.58% (base period) to 42.60% (near term) and 39.17% (long term). The extremely vulnerable (28.08%, 28.00%, 25.35% for base/near/long term) and moderately vulnerable (22.45%, 22.49%, 24.83% for base/near/long term) zones remain relatively stable. This trend indicates a benign shift in vegetation vulnerability under low-emission conditions.

2. SSP2-4.5 (intermediate emission scenario): The extremely vulnerable zone expands from 28.08% (base period) to 32.53% (near term) and 35.73% (long term). Meanwhile, the moderately vulnerable zone shrinks from 22.45% to 18.25% (near term) and 23.46% (long term), and the mildly vulnerable zone decreases from 6.89% to 4.82% (near term) and 3.37% (long term). The severely vulnerable zone shows a slight increase from 42.58% to 44.40% (near term) and 37.45% (long term).

3. SSP5-8.5 (high fossil fuel emission scenario): The ecological risk intensifies significantly. The extremely vulnerable zone expands rapidly from 28.08% (base period) to 25.35% (near term, with a temporary fluctuation) and then to 41.33% (long term). The severely vulnerable zone increases from 42.58% to 33.93% (near term) and 44.26% (long term). In contrast, the moderately vulnerable zone increases from 22.45% to 35.58% (near term, a temporary rise) and 11.67% (long term), and the mildly vulnerable zone decreases from 6.89% to 5.13% (near term) and 2.73% (long term). This pattern reflects that high emissions trigger extreme climate events and ecological degradation, pushing up vegetation vulnerability.

Spatially, the ‘high–high’ clustered areas where both the focal area and its neighbors have high vulnerability are concentrated in the western coastal zone of the Cape Province (Figure 4). To quantify aggregation characteristics, the global Moran’s I values and clustered area proportions under different scenarios were analyzed:

1. SSP1-2.6: The global Moran’s I value increases from 0.32 (base period) to 0.35 (near term) and 0.38 (long term), with the ‘high–high’ clustered area expanding from 18.7% (base period) to 21.2% (near term) and 23.5% (long term). The ‘low–low’ clustered area (distributed at the junction of the Western Cape and Eastern Cape) remains stable at ~12.5%.

2. SSP2-4.5: The global Moran’s I rises from 0.32 to 0.40 (near term) and 0.45 (long term). The ‘high–high’ clustered area expands to 25.8% (near term) and 30.2% (long term), while the ‘low–low’ clustered area shrinks to 10.3% (near term) and 8.7% (long term).

3. SSP5-8.5: The global Moran’s I shows a sharp increase to 0.42 (near term) and 0.51 (long term). The ‘high–high’ clustered area grows to 28.9% (near term) and 35.7% (long term), and the ‘low–low’ clustered area reduces to 9.1% (near term) and 6.3% (long term).

In summary, higher emission scenarios accelerate the expansion of vulnerable areas from the near to long term, especially extremely and severely vulnerable zones, with the Cape Province coastal zone being the most prominent. The ‘high-high’ clustered areas correspond to intense urbanization and industrialization, while ‘low-low’ clusters rely on sound ecological foundations.

3.3. Analysis of the Characteristics of Vegetation Vulnerability Level Transfer

Table 3. Transition matrix of vegetation vulnerability levels in the South African coastal zone.

Unit: km2		Base Period 2011–2020
		Extreme	Severe	Moderate	Mild
Near-future SSP1-2.6	Extreme	19,540	1	0	0
	Severe	64	29,547	1	0
	Moderate	0	40	15,530	0
	Mild	0	0	11	4860
Near-future SSP2-4.5	Extreme	17,672	0	0	0
	Severe	1932	25,297	0	0
	Moderate	0	4291	12,918	0
	Mild	0	0	2624	4860
Near-future SSP5-8.5	Extreme	12,177	10,441	0	0
	Severe	7427	19,147	4337	0
	Moderate	0	0	11,205	1469
	Mild	0	0	0	3391
Far-future SSP1-2.6	Extreme	19,604	5339	0	0
	Severe	0	24,249	1748	0
	Moderate	0	0	13,794	2489
	Mild	0	0	0	2371
Far-future SSP2-4.5	Extreme	17,673	0	0	0
	Severe	1931	21,619	0	0
	Moderate	0	7969	15,542	1252
	Mild	0	0	0	3608
Far-future SSP5-8.5	Extreme	18,340	10,441	0	0
	Severe	1264	19,147	10,255	0
	Moderate	0	0	5287	2934
	Mild	0	0	0	1926

presents the transition matrix of vegetation vulnerability levels in the South African coastal zone from the base period (2011–2020) to the near future (2030–2059) and far future (2070–2099) under different SSP scenarios (unit: km²). We analyze the transition characteristics by quantifying the area transfer between vulnerability levels:

1. SSP1-2.6 scenario: In the near future, the vegetation vulnerability level transfer remains relatively stable. For instance, the extremely vulnerable zone retains 19,540 km², nearly unchanged from the base period, and the severely vulnerable zone maintains 29,547 km² with only 1 km² transferring to extremely vulnerable. In the far future, high-level vulnerable zones exhibit a noticeable trend of transferring to lower levels: the extremely vulnerable zone transfers 5339 km² to the severely vulnerable zone, accounting for 2.78% of its base period area; the severely vulnerable zone transfers 1748 km² to the moderately vulnerable zone, accounting for 6.17% of its base period area; and the moderately vulnerable zone transfers 2489 km² to the mildly vulnerable zone, accounting for 3.77% of its base period area. This pattern indicates a benign recovery trend under the sustainable development scenario.

2. SSP2-4.5 scenario: Both near and far futures show bidirectional transfers between levels: the extremely vulnerable zone transfers 1932 km² to the severely vulnerable zone in the near future and 5339 km² in the far future; the severely vulnerable zone transfers 4291 km² to the moderately vulnerable zone in the near future and 1748 km² in the far future; and the moderately vulnerable zone transfers 2624 km² to the severely vulnerable zone in the near future and 1252 km² in the far future. Notably, the transfer probability in the far future is lower than that in the near future—for example, the transfer area from moderately vulnerable to severely vulnerable decreases by 1372 km²—suggesting a partial mitigation of vulnerability escalation over time.

3. SSP5-8.5 scenario: The vegetation vulnerability level transfer shows a sharp deterioration trend, especially in the far future. In the near future, the severely vulnerable zone transfers 10,441 km² to the extremely vulnerable zone; in the far future, this transfer area remains 10,441 km², and the moderately vulnerable zone transfers 10,255 km² to the severely vulnerable zone, accounting for 14.74% of its base period area. Additionally, the severely vulnerable-to-extremely vulnerable transfer probability reaches 15% in the far future. These data intuitively reflect that under the high-emission scenario, the coastal ecosystem of South Africa suffers severe impacts, with the vegetation vulnerability level deteriorating sharply.

3.4. Analysis of the Coupling Relationship Between Land Use and Vegetation Vulnerability

Figure 5 illustrates the Sankey diagrams of land use–vegetation vulnerability level transfer in the South African coastal zone under different SSP scenarios (base period → near future, base period → far future). Quantitatively, natural vegetation types (forest land, grassland) show distinct transfer trends to vulnerability levels:

Forest land exhibits distinct vulnerability transfer trends under different SSP scenarios. Under SSP1-2.6, in the near future, a certain proportion of forest land transfers to mild vulnerability, with a larger proportion shifting to moderate and severe vulnerability and a small proportion moving to extreme vulnerability. In the far future, the proportion of forest land transferring to mild vulnerability decreases, while the proportion transferring to extreme vulnerability increases, and moderate and severe vulnerability remain the dominant types. Under SSP5-8.5, in the near future, only a very small proportion of forest land transfers to mild and moderate vulnerability, while most shifts to severe and extreme vulnerability. In the far future, the proportion of forest land moving to severe and extreme vulnerability further expands, with mild and moderate vulnerability accounting for an extremely small proportion.

Grassland also shows scenario-dependent vulnerability transfer patterns. Under SSP1-2.6, in the near future, a notable proportion of grassland transfers to mild and moderate vulnerability; in the far future, these proportions show a slight increase. Under SSP5-8.5, in the near future, a certain proportion of grassland transfers to severe and extreme vulnerability; in the far future, these proportions rise further.

For cultivated land, its vegetation vulnerability distribution is scattered: under SSP2-4.5, in the near future, it distributes across all vulnerability levels (mild, moderate, severe, and extreme); in the far future, the distribution pattern shows minor changes. Construction land (urban/rural, industrial/mining, and residential land) exhibits high vulnerability: under all SSP scenarios, a large proportion of construction land remains at severe or extreme vulnerability, with little variation across scenarios.

Figure 6 presents the stacked bar charts of vegetation vulnerability level distribution for each land use type under different SSP scenarios. Key quantitative features include:

SSP1-2.6: Cultivated land maintains a high proportion in mild (28.4–30.1%) and moderate (31.2–32.5%) vulnerabilities across the base period, near future, and far future. Forest land shows similar stability, with mild (29.7–31.2%) and moderate (30.5–32.1%) proportions changing slightly.

SSP2-4.5: Grassland’s moderate vulnerability proportion increases from 35.2% (base period) to 40.1% (near future) and then decreases to 32.7% (far future).

SSP5-8.5: Unused land’s severe and extreme vulnerability proportion rises significantly from 45.3% (base period) to 58.7% (near future) and 67.2% (far future).

Water bodies: Across all scenarios, water bodies predominantly remain in mild vulnerability, with a stable proportion (75.2–77.8%) and no obvious changing trend.

In summary, different land use types exhibit distinct transformation paths to short-term and long-term vulnerability levels. Natural vegetation types (forest land, grassland) show larger fluctuations in vegetation vulnerability under different SSP scenarios, while construction land maintains persistently high vulnerability. Different land use patterns thus lead to varying sensitivities of vegetation vulnerability to environmental changes.

4. Discussion

4.1. Driving Mechanism of the Spatiotemporal Evolution of Vegetation Vulnerability Under Different SSP Scenarios

Different SSP scenarios significantly affect the evolution of vegetation vulnerability through differentiated socioeconomic pathways. Under the SSP5-8.5 scenario, high carbon emissions and rapid urbanization have aggravated the pressure on vegetation degradation. The transfer matrix shows that the area of high vulnerability has expanded by more than 40%, confirming the negative effects of the imbalance between rapid economic development and ecological protection. This is consistent with the findings of Frazee et al. (2003), who noted that in the Cape Floristic Region, human activities such as urban expansion and agricultural encroachment have led to a 35% reduction in natural vegetation cover over the past three decades, particularly in the Western Cape’s coastal areas—aligning with our observation of high vulnerability in port cities and construction land clusters (Figure 6) [24].

Under the SSP1-2.6 scenario, low-carbon development and ecological restoration policies have effectively slowed down the trend of vulnerability escalation. This result is consistent with the IPCC AR6 (2022) prediction of increased ecological risks in high-emission scenarios and also supported by Afuye et al. (2024)’s study on the Amathole District Municipality—they found that low-emission policies combined with targeted vegetation restoration led to a 19% reduction in moderately vulnerable vegetation areas over 10 years, reflecting the positive role of sustainable development pathways [25].

Previous studies have shown that socioeconomic factors and policies such as land use change, carbon emissions, and population growth are significantly different under different SSP scenarios, which plays an important role in the prediction and simulation of environmental change. For instance, Opiyo et al. (2025) observed in western South Africa that under SSP5-8.5, urbanization-driven construction land expansion (increasing by 28% from 2010 to 2022) and agricultural intensification contributed to a 42% increase in vegetation browning areas, while under SSP1-2.6, land use conversion to natural reserves reduced vulnerability by 15% [26]. Mbuqwa et al. (2024) further confirmed in the Eastern Cape’s Amahlathi Local Municipality that SSP scenario differences in agricultural expansion rates (1.2%/year under SSP5-8.5 vs. 0.3%/year under SSP1-2.6) directly led to divergent vegetation vulnerability trajectories, with high-emission scenarios accelerating the conversion of low-vulnerability grasslands to high-vulnerability croplands [27].

Vegetation vulnerability responds significantly to the above-mentioned human activity factors. Some recent studies have found that human activities have a much stronger impact on ecosystem vulnerability and vegetation vulnerability than climate change. IPCC AR6 (2022) emphasized that in coastal zones globally, human activities contribute 60–80% to vegetation vulnerability. This supports our result that land use change is the core driver (evidenced by the Sankey diagram of land use–vegetation vulnerability transfer, Figure 5) [28]. In South Africa’s coastal zone specifically, Mpanyaro et al. (2024) found that along the Buffalo River Catchment, human-induced land use change explained 73% of vegetation vulnerability changes, while climate variability (precipitation/temperature anomalies) accounted for only 27% [29]. The evolution path of vegetation vulnerability in the South African coastal zone is thus highly dependent on the choice of socioeconomic development path, leading to substantial differences in spatiotemporal evolution under different SSP scenarios.

Under the SSP1-2.6 scenario, countries actively adopt sustainable development policies and invest heavily in clean energy and environmentally friendly technologies. In this scenario, the vulnerability transfer matrix shows the characteristics of a small-scale transfer from high-level vulnerability to low-level vulnerability, reflecting the positive restoration trend of the ecosystem under low emissions and ecological protection. However, in the long term, this trend has reversed, with vegetation vulnerability levels showing a reverse escalation. This phenomenon may be related to the existence of a certain threshold in the ecosystem; when cumulative environmental changes exceed ecological thresholds, negative feedback mechanisms are triggered, offsetting earlier restoration gains [30].

In contrast, under the SSP2-4.5 scenario, socioeconomic development is at a medium level, and technological progress is relatively slow. Extremely vulnerable areas shift to severely vulnerable areas, severely vulnerable areas shift to moderately vulnerable areas, moderately vulnerable areas shift to mildly vulnerable areas, and the area of mildly vulnerable areas decreases—overall showing a trend of gradual degradation of vulnerability levels, but with a larger transfer scale compared to SSP1-2.6. This means that under the medium-emission scenario, the ecosystem still has a certain self-regulation ability, but the reduction in the area of mildly vulnerable areas also reflects weakened ecosystem stability. For example, in the Eastern Cape Province, agricultural expansion and tourism development have accelerated the conversion of low-vulnerability areas to high-vulnerability areas [31].

Under the SSP5-8.5 scenario, characterized by high carbon emissions, industrialization, and urbanization, vegetation degradation has become increasingly severe. The transfer matrix shows that although there is a shift from extremely vulnerable areas to severely vulnerable areas, the trend of transfer from low-level to high-level areas dominates. From the base period to the near and far future, the area of extremely and severely vulnerable levels increased the most (nearly 30%), and the transfer area from moderately vulnerable areas to mildly vulnerable areas also reached 4.22%, leading to rapid deterioration of vegetation vulnerability. This confirms the negative effects of fossil fuel-driven development models on ecosystems [32]. This result also aligns with the IPCC’s prediction that ecological risks will be significantly aggravated under high-emission scenarios, further indicating that socioeconomic pathways are key factors determining the evolution of coastal vegetation vulnerability.

4.2. Impact of Land Use Change on Vegetation Vulnerability Shift

In the process of ecosystem evolution, land use change is also an important factor affecting the spatiotemporal distribution of vegetation vulnerability and one of the core drivers of ecosystem change [33]. This is particularly prominent in South Africa’s coastal zone, where human-induced land use conversion directly reshapes vegetation cover structure and ecological resilience, consistent with the IPCC AR6 (2022) conclusion that land use change contributes 60–80% to coastal vegetation vulnerability, surpassing the impact of climate variability alone [34]. To clarify the impact of different land type evolutions on vegetation vulnerability transfer, in this study, we integrated Sankey diagrams (Figure 5), stacked bar charts (Figure 6), and transfer matrix methods to explore the intrinsic link between land use and vegetation vulnerability, further confirming that land use change is the core driver of vegetation vulnerability transfer in the study area.

According to the Sankey diagram and stacked distribution diagram, the vegetation vulnerability evolution of natural vegetation is deeply constrained by land use change. Under high-intensity human disturbances, such as agricultural expansion in the Eastern Cape and urban sprawl in the Western Cape, natural vegetation patches are fragmented, their ecosystem structure and function are degraded, and this drives a consistent shift toward higher vulnerability levels. This aligns with findings from Von Maltitz, G.P. et al., who noted that in southern African coastal biomes, natural vegetation loss due to land use conversion accounts for 65% of vegetation vulnerability escalation, with lowland fynbos and riparian woodlands being the most sensitive to such changes [35].

In contrast, human-dominated land (urban, industrial, and residential areas) exhibits relatively small variations in vegetation vulnerability across different SSP scenarios. This is attributed to intensive human management measures, such as artificial irrigation for urban park green spaces, regular pruning of industrial vegetation buffers, and pest control, which mitigate the impacts of external environmental stress. The research by Du Plessis, S et al. further confirmed that in South Africa’s coastal urban agglomerations, managed green spaces within construction land have 30% lower vulnerability than unmanaged natural patches under the same climate stress, which supports our observation of stable vulnerability in human activity land [36].

Cropland, as a land use type shaped by both natural factors and human activities, shows distinct vulnerability trajectories across SSP scenarios. Although cropland inherently has high vulnerability due to monoculture planting and frequent tillage disturbances, its vulnerability is further amplified under the SSP5-8.5 scenario. This is because high-emission scenarios accelerate agricultural intensification, such as extended cultivation seasons, increased chemical fertilizer use, and expansion into marginal ecological zones, which degrades soil organic matter and reduces vegetation resistance to environmental stress. Nel, J.L. reported that in the Eastern Cape’s croplands, SSP5-8.5 scenarios led to a 22% increase in moderately vulnerable areas compared to SSP1-2.6, which is consistent with our transfer matrix result that cropland vulnerability expanded by 18.7% under SSP5-8.5 [37].

Notably, the interaction between land use change and biodiversity also modulates vegetation vulnerability dynamics. Some research found that mixed land use patches have slower vulnerability escalation than monocultures, as higher biodiversity enhances ecosystem stability [38]. This provides additional context for our conclusion: land use change drives vulnerability not only through ‘type conversion’ but also through ‘compositional diversity’, a factor that has been overlooked in previous coastal vulnerability studies but is critical for South Africa’s biodiversity-rich coastal zones.

Our study takes the period from 2011 to 2020 as the base period. Although it is not yet 2030, we have now reached 2025. We can roughly identify the specific impacts of land use change on vegetation vulnerability transfer in some areas by comparing the conditions of the base period (2011–2020) and 2025. We selected four typical case sites in the Northern Cape, the western part of the Western Cape, the eastern part of the Western Cape, and KwaZulu-Natal, and compared the historical images of 2011 with the latest available historical images, which can reveal the specific impacts of land use change on vegetation vulnerability from a short-term perspective and echo the previous results of the paper.

The case area of the arid coastal zone in the Northern Cape is shown in Figure 7. The image in 2011 shows that the land use is dominated by unused land and grassland, with scarce forest land. Driven by the arid climate and land types with low ecological resilience, it was already in extreme vulnerability in the base period. The image in 2024 shows that although there is no large-scale conversion to cropland or construction land, the natural vegetation patches dominated by drought-tolerant shrubs and herbs have significantly degraded, and the exposed area of sandy land has expanded. By comparing the historical images from 2011 to 2024, although there is no obvious large-scale land type conversion such as cropland or construction land, the natural vegetation patches dominated by drought-tolerant shrubs and herbs have significantly degraded, and the exposed area of sandy land has expanded. The slight decrease in vegetation coverage density is still sufficient to maintain the extreme vulnerability of vegetation in any scenario. This phenomenon shows the high sensitivity of arid coastal ecosystems to climatic stress and implicit human impacts. Even in areas with low human activity intensity, the qualitative decline of vegetation coverage can still be the core driver for the persistence or intensification of vulnerability. It confirms that the degradation of natural vegetation quality is a driving factor of vulnerability and also reflects the high sensitivity of arid coastal ecosystems to stress.

The case area of the coastal zone in the Western Cape (north of Cape Town) is shown in Figure 8. The image in 2011 shows that the land use is dominated by urban construction land, scattered cropland around, and natural vegetation. Driven by the high-intensity human activities from the port and urban development of Cape Town, it was already in severe–extreme vulnerability in the base period. The image in 2025 shows that construction land, such as port terminals and residential areas, has expanded to the surrounding areas. The natural vegetation patches, such as lowland fynbos originally distributed at the edge of the city, have been occupied, and the degree of fragmentation has intensified. Some areas have even been converted into urban-affiliated green spaces. Although there is vegetation coverage, they belong to non-native vegetation under artificial management. By comparing the historical images from 2011 to 2025, it can be roughly determined that the development scenario of this area belongs to the SSP2-4.5 scenario. The explicit expansion of construction land and the implicit degradation of natural vegetation occur simultaneously. Even with the supplement of artificial green spaces, the structure and function of the native vegetation ecosystem have still significantly declined, which means that the vegetation vulnerability of this area has continuously remained at a severe level or above. This phenomenon reflects that in the coastal zones with dense human activities, like the Western Cape, the direct occupation of natural vegetation through the expansion of construction land and the indirect weakening of ecosystem services are the core drivers for maintaining high vulnerability. It shows that the expansion of construction land dominated by humans will lead to the degradation of natural vegetation and the maintenance of high vulnerability and also forms an intuitive echo with the ‘Spatial distribution of vegetation vulnerability in the South African coastal zone in the near and long term under the different SSP scenarios’ shown in Figure 3.

The eastern side of the Western Cape is a coastal zone case of low vegetation vulnerability in the study area, as shown in Figure 9. The image in 2011 shows that the land use is dominated by natural vegetation such as forest land, grassland, and wetland, with few traces of human activities. There is a large area of continuous green vegetation coverage, and the ecosystem around the water body is complete. Driven by the good ecological background, this area was in moderate–mild vulnerability in the base period. The image in 2024 shows that small-scale tourism facilities appear around some wetlands and at the edge of forest land, and scattered cropland or pastures appear. Although there is no large-scale fragmentation of natural vegetation patches, the continuity has decreased, and the vegetation buffer zone of the water body has also been compressed due to the construction of artificial facilities. By comparing the historical images from 2011 to 2024, although there is no large-scale expansion of construction land, low-intensity human activities have disturbed the natural vegetation. This makes the vegetation vulnerability of this area still show a trend of transferring from mild to moderate even under the relatively mild development scenario, such as SSP1-2.6. The slight disturbance of the native ecosystem weakens its buffering capacity against environmental changes. This phenomenon reflects that even in coastal zones with good ecological foundations and low human activity intensity, low-intensity human disturbances to natural vegetation will also promote the evolution of vulnerability to a higher level. This shows that even under the SSP1-2.6 scenario, if management is improper, the areas with good ecological foundations will still degenerate.

The coastal zone case study on the northeast side of KwaZulu-Natal is shown in Figure 10. The image in 2011 shows that the land use is dominated by coastal forests, wetlands, and native grasslands. It has a tropical savanna climate, and due to the influence of the coastal warm current and terrain, this area also has the characteristics of a tropical humid climate, with an excellent ecological background. Driven by this, it was in mild–moderate vulnerability in the base period, and due to the strong disturbance resistance of the ecosystem, the initial disturbance of natural vegetation by human activities was extremely weak. The image in 2024 shows that only a small number of small residential areas and tourism facilities appear at the edge of coastal forests and around wetlands, and some native grasslands have been sporadically transformed, but the core areas of natural vegetation patches have not been massively damaged, and the overall continuity is still good. The compression degree of the vegetation buffer zone of the water body is also relatively slight. By comparing the historical images from 2011 to 2024, the disturbance of low-intensity human activities to the native ecosystem is small in scope and shallow in degree. This makes the variation range of vegetation vulnerability in this area relatively smaller and more stable under the three future scenarios of SSP1-2.6, SSP2-4.5, and SSP5-8.5. It not only reflects the ‘innate resilience’ of areas with good ecological foundations but also indicates that low-intensity and non-destructive human activities have a more limited impact on their vulnerability dynamics. This phenomenon further confirms that the quality of the ecological background is one of the core prerequisites for determining the stability of vegetation vulnerability. When an area has both the characteristics of an excellent ecological foundation and weak human disturbance, even under multiple scenarios, the vulnerability is more likely to maintain a relatively stable state.

4.3. Limitations of the Study and Future Research Directions

In terms of data sources and selection, although in this study we used future scenario data from CMIP6 and multi-source datasets, limitations remain regarding data timeliness, dynamics, and spatial precision. The base period (2011–2020) has a relatively short time span, making it difficult to fully capture long-term vegetation response patterns under historical climate and land use changes. Meanwhile, the lack of micro-scale vegetation data restricts in-depth analysis of interactions between vegetation and environmental factors. Additionally, the 1 km spatial resolution of the adopted data is insufficient for small-scale vulnerability assessments, especially in regions with complex terrain or fragmented vegetation cover, which may lead to the omission of fine-scale vulnerability hotspots [23,39,40]. In the future, data accuracy could be enhanced by integrating high-resolution remote sensing data and establishing a long-term in situ monitoring network, thereby improving the comprehensiveness and dynamics of data support for vulnerability assessment.

At the model and method level, in this study, we adopted SSP scenarios to reflect the combined impacts of climate and socioeconomic change, but the generalized nature of these scenarios fails to fully account for the unique geographical, ecological, and socioeconomic characteristics of the South African coastal zone. Regional systems exhibit inherent complexity derived from their specific biophysical backgrounds and human activity patterns, and generic global or continental-scale scenarios often overlook region-specific drivers that shape vegetation vulnerability [41,42,43]. Furthermore, the study lacks sufficient model validation and comparative analysis, which may introduce uncertainties into the predicted results. Future research should integrate regional characteristics into scenario optimization and adopt multi-model comparison approaches to improve the reliability of assessment outcomes.

In terms of factors influencing vegetation vulnerability, while in this study we considered both natural and human drivers, it still has limitations in addressing the specificity of the South African coastal zone. On one hand, the ecosystem service functions of vegetation were not incorporated into the vulnerability assessment framework, leading to an incomplete understanding of how service loss amplifies vulnerability [44]. On the other hand, the study insufficiently quantified the interactive effects between different drivers, which are critical for explaining nonlinear changes in vegetation vulnerability [45,46]. Future work could develop an integrated assessment model that incorporates ecological service indicators and quantifies driver interactions, thereby providing a more comprehensive understanding of vegetation vulnerability dynamics in the study area.

5. Conclusions

5.1. Core Findings

This study evaluates vegetation vulnerability in the South African coastal zone (20 km inland buffer) across three periods using spatial analysis, Moran’s I/LISA, transfer matrix, and Sankey diagram methods, integrating CMIP6 high-resolution climate data (CNRM-CM6-1-HR), LUH2 land use projections, and multi-source ecological datasets. The core findings, scientific contributions, and practical implications are summarized as follows:

(1) Spatial pattern of vulnerability: The base period vegetation vulnerability in the South African coastal zone exhibits a distinct ‘high in the west, low in the east’ gradient. Extreme vulnerability is concentrated in arid coastal areas of the Northern Cape and southwestern coastal zones of the Western Cape, with 18.7% of these areas forming high–high-agglomeration zones as identified through LISA analysis. In contrast, the Eastern Cape and northeastern KwaZulu-Natal—characterized by better ecological backgrounds —maintain mild-to-moderate vulnerability, with low–low agglomeration accounting for 12.3% of the total area.

(2) Differentiated responses to SSP scenarios: Under SSP1-2.6, long-term vegetation vulnerability shows a recovery trend, with the extreme vulnerability area decreasing by 9.46% compared to the near term. Under SSP5-8.5, the long-term extreme vulnerability area expands by over 40% relative to the base period, dominated by transfers from severe to extreme vulnerability. SSP2-4.5 exhibits moderate deterioration, with the total high-vulnerability area increasing by 7.65% in the long term, reflecting a ‘gradual degradation’ trajectory between the two extreme scenarios.

(3) Land use–vegetation vulnerability coupling: Natural vegetation and cropland are the most sensitive land use types to vulnerability changes. Under SSP5-8.5, woodland vulnerability escalates significantly, with 58.8% shifting to severe and 34.7% shifting to extreme vulnerability; cropland vulnerability increases by 18.7% due to agricultural intensification. In contrast, construction land maintains stable vulnerability across all scenarios, while water bodies retain mild vulnerability due to microclimate buffering effects.

5.2. Scientific Contributions

This study fills regional gaps in SSP-based vegetation vulnerability assessments by (1) constructing a multi-scenario evaluation framework tailored to the South African coastal zone, integrating 0.5° CMIP6 data (higher precision than 1° datasets used in previous African studies) and 1 km LUH2 land use data to improve small-scale vulnerability mapping accuracy; (2) quantifying the coupling mechanism between land use change and vulnerability transfer, revealing that human activities contribute 60–80% to vulnerability dynamics—consistent with IPCC AR6 conclusions but refined for coastal-specific contexts; and (3) identifying ‘ecological thresholds’ in arid coastal ecosystems, where cumulative environmental stress triggers reverse degradation even under low-emission scenarios, providing new insights for semi-arid coastal vulnerability research.

5.3. Practical Implications

Aligned with South Africa’s National Coastal Integrated Management Act, targeted management strategies are proposed: (1) For high-risk western zones, strictly control port expansion and mining activities, and drought-resistant vegetation restoration should be implemented; (2) for cropland-dominated areas, promote eco-friendly practices such as crop rotation and fallow to reduce soil degradation and vulnerability; and (3) for ecologically sensitive zones, establish protected buffers around wetlands and coastal forests to prevent low-intensity tourism/residential development from disrupting ecosystem stability.

5.4. Limitations and Future Directions

This study is constrained by 1 km land use data resolution, which limits the capture of small-scale vulnerability hotspots. Future research could integrate 10 m Sentinel-2 remote sensing data to refine local-scale assessments and incorporate biodiversity indices to further explore the role of ecological community structure in vulnerability regulation.