Ecological Risk Assessment of Watersheds Based on Adaptive Cycling Theory—A Case Study of Poyang Lake Ecological and Economic Zone

Abstract

Under the global urbanization context, irrational land use patterns have exacerbated ecosystem imbalance. Developing watershed ecological risk assessment methods based on adaptive cycle theory holds significant scientific importance for flood risk prevention. This study established a watershed ecological risk assessment framework within the adaptive cycle framework, focusing on the Poyang Lake Ecological Economic Zone in the middle-lower Yangtze River Basin. The results revealed that high-risk ecological areas clustered around the Poyang Lake water system with scattered urban distribution, while medium-risk zones dominated the study area. Low-risk regions primarily concentrated in the Yuanhe Plain of southwestern region. The system exhibited significant spatial heterogeneity in “exposure” and “disturbance” risks. Medium–high exposure pixels accounted for 43.3% with a dispersed distribution, whereas disturbance pixels concentrated in Poyang Lake waters and developed urban areas (64.34%), indicating that disturbance exerted a stronger influence on risk assessment outcomes. Governance practices demonstrated that policy preferences may introduce biases into watershed ecological risk evaluations. Multi-scenario simulations using an Ordered Weighted Averaging (OWA) algorithm identified risk-uncertain zones in southeastern hilly areas and northern Poyang Lake waters, while distinguishing stable high/low-risk regions unaffected by decision-making influences. These findings provide critical references for formulating sustainable watershed management strategies.

1. Introduction

Along with the global urbanization process, the continuous expansion of human activities is seriously affecting the structural and functional stability of ecosystems, and unsustainable land development and utilization patterns have led to environmental degradation and a significant increase in the risk of local floods and droughts [1,2,3,4]. In June–August 2024, the Yangtze River Basin was hit by torrential rains, resulting in major regional flooding disaster. According to the Jiangxi Flood Control and Drought Relief Command Center, the flooding of Poyang Lake in Jiangxi caused 1,669,000 people to be affected, with a direct economic loss of CNY 2.7 billion, and the security of the ecosystem in the watershed should not be ignored.

A watershed ecosystem is a natural–economic–social composite ecosystem at the watershed scale, linked by water and composed of ecological elements and demographic, social, and economic elements [5,6,7]. This ecosystem is large and comprehensive, with interactions between internal elements and strong, long-term anthropogenic disturbances during its evolution [8]. With socio-economic development, the watershed is subjected to increasingly serious problems of pollution by human activities and natural disasters, leading to a structural shortage of system resources in the lake and its watershed section, and the conflict of interest between the two makes the watershed a regional system full of challenges within the relationship between society and ecology [9]. Currently, social–ecological system theoretical modeling has become a central tool for assessing the properties of complex systems and has received widespread attention in the field of regional ecological risk assessment research [10].

From the late 1990s to the early 2000s, scientists developed research models for watershed-scale studies. They also began attempting ecological risk assessments at the watershed level [11]. Most global research on watershed ecological risks has focused on how natural disasters and external risks affect higher-level ecosystems and their parts. However, fewer studies have examined risks from internal ecosystem factors or human activities. In 1996, Hooper and Duggin created a flood risk zoning model based on ecological features. They then proposed land use restriction policies using this model [12]. In 2007, Zhang studied soil erosion risks in the Yanhe River Basin, analyzing hazard severity and loss impacts [13]. In 2008, Liu et al. built on Zhang’s work to develop a flood disaster risk assessment framework for the Huai River Basin. Their framework used three dimensions—hazard, stability, and vulnerability—to create an indicator system for disaster risk assessment [14]. These assessments mainly focus on damage from single natural disasters. They do not account for human activities or unexpected events. Additionally, the evaluation indicators used are often complex, and a standard indicator system has not yet been established.

Based on this, this study takes Poyang Lake Ecological and Economic Zone as the research object, uses GIS Pro 3.0 spatial software analysis to introduce the three-dimensional framework of the adaptive cycle within the ecological risk evaluation system of watershed landscapes, and adopts Ordered Weighted Assessment (OWA) to simulate the spatial distribution of risk on different preferences of decision-makers in order to satisfy the needs of landscape planning and future development and protection.

2. Materials and Methods

2.1. Study Area

The Poyang Lake Eco-Economic Zone (114°29′–117°42′ E, 27°30′–30°06′ N) is located in northern Jiangxi Province, encompassing 38 counties and cities, including Nanchang, Jiujiang, and Jingdezhen, with a total area of 51,200 km² (Figure 1). Centered around Poyang Lake—China’s largest freshwater lake—the zone lies within a subtropical humid monsoon climate region, characterized by abundant water resources and unique hydrological dynamics. The lake’s basin spans 162,200 km², accounting for 9% of the Yangtze River Basin. As a critical freshwater ecosystem, agricultural production base, and ecological–economic hub in the middle-lower Yangtze region, the area plays a dual role in ecological preservation and economic development. Balancing environmental protection with urban growth is vital for ensuring sustainable progress in central China.

Unlike the broader Poyang Lake Basin, the Poyang Lake Eco-Economic Zone functions as an integrated development region, incorporating not only natural ecological components but also socioeconomic and cultural dimensions. This integration offers diverse datasets for multi-variable and multi-level environmental risk assessments, making the zone an ideal study area for analyzing complex human–environment interactions.

2.2. Data Sources and Preprocessing

This study employs spatial analysis software, including ArcGIS Pro 3.0 and Fragstats 4.2, to process and overlay multiple datasets, such as elevation, climate, the Enhanced Vegetation Index (EVI), land use, and the nighttime light index. Slope data were derived from the 90 m resolution SRTM DEM dataset provided by the Geospatial Data Cloud. Climate data were obtained from the National Tibetan Plateau Data Center. EVI data were extracted using Google Earth Engine (GEE) to acquire long-term MODIS 1 km annual maximum composite data. Land use data for 2023, with a spatial resolution of 1 km, were sourced from the Resource and Environment Science and Data Center of the Chinese Academy of Sciences. Based on the “Current Land Use Classification” standard (GB/T 21010-2017) [15], land use types were categorized into six classes: cropland, forest, grassland, water bodies, built-up areas, and unused land. Nighttime light index data were batch-processed using GEE to extract NPP VIIRS nighttime light remote sensing data. All layers were clipped and reprojected to the unified WGS84 geographic coordinate system.

2.3. Methodology

2.3.1. Constructing an Ecological Risk Assessment Framework Based on Adaptive Cycles

The adaptive cycle theory, developed as an extension and refinement of traditional views on ecosystem succession [16], provides a holistic and dynamic approach to understanding system interactions and risk responses [17]. Holling C.S. introduced the concept of resilience, supplementing traditional ecosystem transitions with two dynamic processes: gradual reorganization and rapid release [18]. This framework posits that systems undergo four phases—exploitation (r), conservation (K), release (Ω), and reorganization (α)—forming an adaptive cycle through the interplay of three attributes: potential, connectedness, and resilience [19] (Figure 2). The system’s inherent properties correspond to the potential layer; interactions among its components are reflected in the connectedness layer; and its capacity to adapt to environmental risks constitutes the resilience layer [17,19]. These attributes drive the emergence of new elements, structures, and functions, including self-adaptation and evolutionary reorganization at the cycle level [20].

Disaster risks align with the logic of system evolution, characterized by exposure and disturbance effects (Figure 2). Exposure refers to the internal characteristics of a system, while disturbance represents external environmental conditions, including human interventions and climatic factors. The interaction between the watershed spatial system and exposure–disturbance risks is fundamentally a dynamic process of spatial element reorganization, structural adjustment, and functional transformation. As external conditions change, the continuous input of disaster risk factors can disrupt or damage the watershed ecosystem, leading to changes in spatial elements and watershed structures.

Compared to ecosystem conservation, the social production and development of the Poyang Lake Eco-Economic Zone are equally critical. Thus, strategies focusing solely on ecological protection are insufficient to address its complexity. The watershed spans multiple administrative regions, involving diverse governance structures [21]. Given decision-makers’ preferences for risk trade-offs, a more systematic and comprehensive understanding of adaptive risk changes in the Poyang Lake Eco-Economic Zone is essential. Existing risk studies on the Poyang Lake Basin primarily focus on ecological resilience assessment, land use ecological risk evaluation, and quantitative analysis of land use changes.

However, they lack a holistic assessment of factors influencing the entire system. For instance, Lai et al. analyzed the spatiotemporal evolution of ecological resilience in the Poyang Lake Basin [22]; Zhao conducted ecological risk analysis based on land use changes [23]; Zhang developed a remote sensing ecological index to comprehensively evaluate ecological quality changes over time and space [24]; and Huang examined the spatiotemporal evolution of landscape ecological risks over a decade from a landscape pattern perspective [25]. Current assessment methods do not fully capture the complex interactions between ecological and social systems in the watershed [26]. Additionally, the risks and corresponding mitigation strategies for the Poyang Lake socio-ecological region are not well understood [27].

While the adaptive cycle theory reveals the feedback mechanisms between systems and risks, it lacks the capacity to quantify data indicators and spatially localize them. Existing regional ecological risk assessment systems classify risk sources into “exposure” and “disturbance” categories, relying on static overlays that overlook the dynamic evolution of system adaptation to risks. To address this gap, this study integrates adaptive cycle theory with a regional ecological risk assessment framework to develop a model that incorporates watershed-specific adaptive features.

2.3.2. Constructing the “Potential–Connectedness–Resilience” Ecological Risk Indicator System

The rapid development of the Poyang Lake Eco-Economic Zone has introduced ecological disturbances, while the watershed remains vulnerable to frequent floods and droughts. To address these dual challenges, risk sources are classified into three dimensions—potential, connectedness, and resilience—based on their exposure or disturbance impacts (

Table 1. Ecological risk indicator system for watersheds based on adaptive cycling theory.

Criteria (Weights)	Risk Layer	Indicators (Weights)	Normalize
Potential risk (0.1634)	Exposing	Slope (0.0958)	+
		Land use type (0.3181)	+
		Enhanced vegetation cover (0.233)	−
	Disruptions	Average annual temperature (0.0577)	+
		Rainfall erosivity (0.1495)	+
		Population density (0.146)	+
Connectivity risk (0.297)	Exposing	Shannon Diversity (0.1863)	−
		Spread index (0.3149)	−
		Landscape separation (0.1305)	+
	Disruptions	Weighted average proximity factor for built-up land and water body area (0.0774)	+
		Distance to building site (0.2114)	−
		Distance from road (0.0795)	+
Resilience risk (0.5396)	Exposing	Trends in vegetation cover (0.1554)	−
		Landscape structural stability index (0.2128)	−
	Disruptions	Trends in nighttime light intensity (0.2834)	+
		Land use stability index (0.2498)	−
		Trend in mean annual temperature (0.0987)	+

). Ecological risk source factors cover multiple fields. To ensure scientific and objective research results, this study uses hierarchical analysis to build a hierarchical structure model. First, six experts from geography, ecology, land use, and landscape architecture were selected. Questionnaires (Appendix A) were distributed to them to assess each factor’s impact level on ecological risk and score pairwise comparisons. This process constructed a judgment matrix. Yaahp multi-criteria decision software determined and calculated weights for regional ecological risk evaluation indicators. A hierarchical model was created, and expert scores were used to generate the judgment matrix. After the matrix passed a consistency test, the arithmetic mean method calculated weights for ecological factors. In the standard layer, the potential risk is 0.1634, the connectivity risk is 0.297, and the resilience risk is 0.5396.

Potential risks reflect the system’s intrinsic heterogeneity in spatial structure and dynamic succession processes, which shape its response to regional ecological stressors [19]. In contrast, disturbance effects arise from natural hazards and human development activities, creating compounded threats to ecosystem stability.

Exposure indicators include the slope, land use type, EVI, Shannon diversity, contagion index, landscape division, vegetation cover trend, and landscape structural stability index. Slope values (thresholds: 6°, 10°, 15°, and 25°) were normalized to assess soil erosion risks [28]. Land use types indicate ecological threats across development stages. The Enhanced Vegetation Index (EVI) quantifies vegetation health and greenness [29]. Shannon diversity measures landscape richness, positively correlating with habitat interactions [30]. The contagion index evaluates patch aggregation [31], while the landscape division index assesses spatial segregation patterns. Landscape metrics were calculated using Fragstats 4.2 with a 3 km moving window to map spatial distributions [32]. Trend analysis of the annual maximum EVI (2013–2023) reveals ecosystem succession under risk conditions. Landscape structural stability (SIi) quantifies land use stability [33,34,35].

Disturbance indicators include the annual mean temperature, rainfall erosivity, population density, distance to built-up areas/roads, weighted adjacency coefficient of built-up and water areas, nighttime light intensity trends, land use stability (LUSi), and annual temperature trends. Extreme climate events increase the risks of floods and droughts [36]. Rainfall erosivity indicates the potential for soil erosion [28,37]. Population density represents human activity intensity. Euclidean distances to built-up areas and roads, derived via ArcGIS [38], quantify anthropogenic encroachment on ecological zones [39]. The weighted adjacency coefficient of built-up and water areas inversely correlates with system connectivity [25]. Nighttime light intensity trends (2019–2023) reflect urbanization and energy consumption patterns [40]. Land use stability (LUSi) captures landscape patch responses to external disturbances [27]. Theil–Sen Median trend analysis and Mann–Kendall tests were applied to assess temporal changes in these indicators [41,42].

2.3.3. Constructing the Watershed Adaptive Ecological Risk Index

To standardize the process, this study normalizes individual indicator data when calculating relevant indicators and their spatial distributions. This conversion uses Equations (1) and (2) to turn the data into relative measures that reflect watershed ecological risk. The goal is to simplify subsequent calculations [43].

$$\mathrm{Positive} \mathrm{indicators}: X_p={\displaystyle \frac{X_i-X_{min}}{X_{max}-X_{min}}}$$

(1)

$$\mathrm{Reverse} \mathrm{Indicator}: X_n={\displaystyle \frac{X_{max}-X_i}{X_{max}-X_{min}}}$$

(2)

where X_p is the normalized value of the forward indicator; X_n is the normalized value of the reverse indicator; X_min is the minimum value of the risk indicator; X_max is the maximum value of the risk indicator; the range of the normalized indicator image element is 0–1; and the risk value of “potential–connectivity–resilience” is obtained as a 3-level risk value [44]:

$$R_i=\sum _{i=1}^n(W_i\times I_i)$$

(3)

In the formula, R_i represents the potential, connectivity, and resilience risk values. AERISES stands for the adaptive ecological risk index. The normalized risk indicators are called I_i. And W_i shows the weight allocation of each indicator compared to the guideline layer. By combining the three levels of superimposed analysis, we can finally obtain the adaptive ecological risk index:

AERISES = W_p × R_p + W_c × R_c + W_r × R_r

(4)

2.3.4. Ecological Risk Scenario Setting for Watershed Landscape Based on OWA Methodology

To make watershed ecological risk assessment more useful and practical for decision-making, this study calculates the ecological risk of the Poyang Lake Ecological and Economic Zone. It uses the ordered weights from the OWA algorithm. Then it conducts environmental risk assessments under multiple risk scenarios. The goal is to provide a more comprehensive decision-making strategy. Here is how scenario assessment works. First, all indicators are sorted based on the weights from hierarchical analysis. Next, the 17 indicators are ordered from highest to lowest. Then the OWA operator calculates the positional weights for these 17 indicators under different scenarios. Finally, the sorted indicators are multiplied by their corresponding positional weights. This process generates the ecological risk values for each scenario. To determine the positional weights in the OWA operator, the study uses the Yager-defined operator [45,46]. The formula for this is as follows:

$$w_j=Q_{RIM} (\frac{j}{n})-Q_{RIM} (\frac{j-1}{n}), j=1, 2, \dots , n, Q_{RIM} (r)=r^{\alpha }$$

(5)

In this formula, j stands for the rank order of indicators. w_j represents the weight assigned to each rank, and n is the total number of indicators included. The variable r is the independent input value, while α serves as the power index that defines the scenario type. When α is less than 1, lower-ranked indicators carry more weight. This means decision-makers adopt an optimistic stance, expecting landscape ecological risk to gradually decrease. Conversely, when α is greater than 1, higher-ranked indicators have more influence. This reflects a pessimistic stance, indicating that landscape ecological risk is expected to increase over time. When α equals 1, there is no preference for any rank order, meaning a neutral stance in risk assessment.

3. Results and Discussion

3.1. Spatial and Temporal Distribution of Adaptive Ecological Risk Indices

The value-at-risk calculations for each indicator yielded a spatial distribution pattern for the 17 indicators (Figure 3). The analysis of the weights of the criteria after the combined stacked ordered weighting (Figure 4) shows that the spatial distribution is characterized by inconsistent differences in the layers of evaluation criteria.

The potential risk (Figure 4a) presents the system’s own characteristics, and the risk value of each prefecture-level city is prominent (risk index > 0.8), and the evaluation result grade is close to 1, but the area distribution is small, at only 0.69% of the total area (Figure 5), in which the potential risk of urban land is a more serious threat to the whole region; the medium risk area (risk index 0.4–0.6) is the watershed of Poyang Lake, Jinxi Lake, Jingshan Lake, and the southern side of Qinglan Lake. The ecological source of the watershed has an area distribution accounting for 33.35% of the whole area, and the surrounding areas of the watershed are gradually tilted to medium and low risk (risk index < 0.4), accounting for 58.06% of the total area. Land use types and enhanced vegetation cover are the factors influencing risk values through weighting. The overall ecological risk evaluation shows a trend: low risk to high risk, followed by medium risk, from the perimeter to the center. Urban pixels prominently exhibit high-risk patterns. The study area is surrounded by forested land, while the central Poyang Lake region has extensive cultivated and construction land. This explains the consistent trend in human activity range results. Areas with high-enhanced vegetation cover are primarily forested, while water areas with low vegetation cover drive changes in risk values.

The connectivity degree (Figure 4b) shows the system’s spatial structure. High-risk areas (risk > 0.6) cover a relatively small area, forming a belt-shaped distribution. These areas mainly lie in the water body regions of the study area. Scattered urban and rural built-up areas fall into the medium-high risk category (risk 0.6–0.8). Their distribution area is small, accounting for only 12.67% of the total area. In contrast, medium–low risk areas (risk 0.2–0.6) occupy a larger portion, making up 76.09% of the area. These areas mainly include cultivated and forested lands around the study region.

Low-risk zones (risk < 0.2) represent just 9.79% of the total area, mainly located in the northwestern and northeastern water bodies. The boundary of Poyang Lake Reservoir neighbors unbuilt land or farmland, presenting a clear high-risk pattern. By comparison, the water bodies of Zhelin Reservoir, Junmin Reservoir, and Luxi River are adjacent to large forested areas, showing a lower-risk profile. This suggests that water system areas connected to human activities pose greater threats. In contrast, ecological spaces bordering forest land are assessed as more secure. This implies that the threat is higher in areas where water system connectivity is associated with human activities. Among them, areas dominated by construction land and cultivated land have a relatively high intensity of human activities, and ecological restoration can be prioritized to stabilize the environmental security pattern. On the contrary, the ecological space bordering forest land is evaluated to be more secure. Water and forest land are two types of land use with low intensities of human activity. Both of them play an important role in guaranteeing regional habitat quality. The study on the construction of the ecological security pattern in Southern Jiangsu Province also reached the same conclusion [47].

The distribution pattern of the toughness layer (Figure 4c) is most similar to that of the final risk layer (Figure 4f). Medium- and high-risk areas mainly cluster in the north of Poyang Lake, Nanchang City, Jingdezhen, northern mountainous and hilly regions, the south of Zhelin Reservoir, and the southern plateau areas of Yingtan. These areas make up 22.74% of the whole study region. What these places have in common is a lot of industrial building and infrastructure development. As the economy grew, the natural environment became damaged faster, and the ecological pressure went up. The other medium-high risk areas are spread out across the study area, centered around individual counties. The toughness layer shows how the Poyang study area’s development changed in response to watershed ecological risks over the past ten years.

In 2022, Jiangxi Province released the Implementation Program for Deepening Poyang Lake Ecological Protection and Compensation Mechanism. This program included many environmental protection measures. In recent years, as these policies were put into action and environmental rules improved, damage to the environment slowed down. The overall ecological situation in the Poyang Lake Basin has improved. But the rising ecological risks show that traditional ways of assessing ecological risks have problems. These methods cannot fully understand the complex and changing risks in socio-ecological systems. Also, they do not think enough about how well the basin can handle risks on its own and whether decision-making plans are a good fit for dealing with these risks.

The spatial distribution of risk coefficients reveals clear differences between exposure and disturbance risks. Exposure risk pixels spread across the entire study area, while disturbance risk pixels cluster in Poyang Lake waters and developed urban areas. This suggests that external disturbances are the main factor affecting deviations in risk assessment results. Looking at exposure risk distribution (Figure 4d), medium- and high-risk areas mainly concentrate in the transitional zones between urban construction areas and water bodies. Notable examples include Nanchang, Jingdezhen, and the central region of Poyang Lake, which show a ring-like pattern. Among these, ecological spaces such as Poyang Lake Wetland National Nature Reserve, Poyang Lake Nanji Wetland National Nature Reserve, and Duchang Migratory Bird Provincial Nature Reserve have been invaded by human construction. These changes in surface cover type correlate positively with wetland exposure risk intensity, leading to higher ecological risks in these areas. Low ecological risk values appear in the Ganjiang River Basin zone, accounting for 6.43% of the total area. Disturbance risk distribution (Figure 4e) relates closely to spatial variations in influencing factors like urban development, agricultural activities, and climate change. In 2009, more integrated sub-watershed management sites were established in the study area’s southwestern region. These sites featured favorable climatic conditions and flat terrain, resulting in lower disturbance risks that cover 35.66% of the overall area.

A comparison of the spatial distributions of exposure and disturbance risks reveals an important pattern. Areas with higher disturbance risk cover a wider scope. This indicates that the ecosystem of the Poyang Lake Ecological and Economic Zone is more vulnerable to disruptions from human social development. The study’s conclusion highlights key priorities. When protecting and restoring the watershed, decision-makers must pay closer attention to how human activities affect the watershed ecosystem. They need to implement effective measures to reduce disturbance risks. This will help maintain the local ecosystem’s health and ensure its sustainable development. Looking at different stages of system change, the following is clear: During the transition from the restructuring stage to the protection stage, exposure levels suggest that policy measures can actively restore the original ecological state. In contrast, moving from the release stage to the utilization stage, disturbance risks signal potential threats. These threats come from both human activities and natural environmental factors in the area.

3.2. Analysis of Different Risk Scenarios

The study calculated simulated predictions of watershed risk under different scenarios. It used Equation (5) to determine the order weights for each indicator layer in every scenario (

Table 2. Results of the calculation of indicator order weights.

Decision Risk Coefficient (Vj)	α = 0.4	α = 0.5	α = 0.8	α = 1	α = 1.5	α = 2	α = 3	α = 10	α = 20
Trends in nighttime light intensity (V1)	0.868	0.243	0.104	0.059	0.051	0.044	0.033	0.005	0.000
Land Use Stability Index (V2)	0.031	0.100	0.077	0.059	0.055	0.051	0.043	0.013	0.002
Stability indexof landscape structure (V3)	0.018	0.077	0.069	0.059	0.056	0.053	0.048	0.020	0.005
Area-Weighted mean proximity Index (Built-up land and water body area) (V4)	0.013	0.065	0.065	0.059	0.057	0.055	0.051	0.027	0.009
Fractional VegetationCover Trend (V5)	0.010	0.057	0.061	0.059	0.058	0.057	0.054	0.034	0.014
Landscape Division Index (V6)	0.009	0.052	0.059	0.059	0.058	0.058	0.056	0.040	0.020
Shannon’s diversity index (V7)	0.007	0.048	0.057	0.059	0.059	0.059	0.058	0.047	0.028
Annual Average Temperature Variation Trend (V8)	0.006	0.044	0.055	0.059	0.059	0.060	0.060	0.054	0.036
Land Use Type (V9)	0.006	0.042	0.054	0.059	0.060	0.060	0.061	0.060	0.046
Contagion index (V10)	0.005	0.039	0.053	0.059	0.060	0.061	0.063	0.066	0.057
Enhanced Vegetation Index (V11)	0.005	0.037	0.052	0.059	0.060	0.062	0.064	0.072	0.068
Rainfall Erosivity (V12)	0.004	0.036	0.051	0.059	0.061	0.062	0.065	0.079	0.081
Population Density (V13)	0.004	0.034	0.050	0.059	0.061	0.063	0.066	0.085	0.095
Euclidean Distance (Road) (V14)	0.004	0.033	0.049	0.059	0.061	0.063	0.067	0.091	0.110
Euclidean Distance (Construction Site) (V15)	0.003	0.032	0.049	0.059	0.061	0.064	0.068	0.097	0.126
Elevation (V16)	0.003	0.031	0.048	0.059	0.061	0.064	0.069	0.103	0.143

). Then, the indicators were ranked from the most optimistic to the most pessimistic based on the scenario conditions. Finally, the ecosystem services in the study area under each scenario were divided into five categories.

A comparison of watershed landscape ecological risk assessments under different selection preferences (Figure 6) reveals distinct patterns. In the optimistic scenario (α < 0.5), nearly the entire Poyang Lake Ecological and Economic Development Zone falls into the medium–low risk category. This makes the risk of early warning less meaningful. Under a more optimistic scenario (0.5 ≤ α < 1), the central-western landscape of the Poyang Lake Plain Area becomes a medium-risk zone. Only Xinyu and parts of the Yuanhe Navigation, Jinjiang, and Fuhe River Basins remain in the low-risk zone. When decision-makers adopt a more pessimistic stance, Nanchang’s main urban area is mostly classified as high risk. The main lake area of Poyang Lake, along with Yongxiu, Yugan, and Duchang Counties, is identified as a higher-risk area. In contrast, parts of Wuning County, Ruichang, Xinyu City, Xinguan County, and Fuzhou show lower-risk landscapes. Under highly or extremely pessimistic assessment results (α ≥ 10), high-risk areas concentrate in wetland landscapes, lakeshore green zones within watersheds, Lushan Mountain, Wulaofeng, and the northern side of Dongxiang District. These areas face significant ecological coercion risks.

Decision-makers weigh various factors for landscape planning policies, which are not based on extreme preferences and conditions. Therefore, within the range of the above preferences, the trend of the evaluator’s expected future tendency, α, was selected for consideration in the range of 0.8 to 1.5 (Figure 7). When α is 1, the calculation results reflect the ecological risk of the current social–ecological system landscape. These results use the indicator weighting method in this study. They correspond to the “normal risk scenario” and match the traditional weighted linear combination (WLC) method. In this method, all indicator weights are equal. Each weight has a value of 1/17 or 0.059. The decision-maker adopts an unbiased decision-making approach. No single indicator determines the final ecological risk evaluation results. In this scenario, the decision-maker shows no bias, and no indicator dominates the final evaluation outcome. When α reaches 0.8, it means that the risk pixels in the outer part of the study area are reduced to a low-risk value, i.e., a “risk-negligent scenario”. When α reaches 1.5, it means that some of the pixels in the medium-risk area are given higher importance, i.e., “risk-emphasized scenario”. The trends of ecological risk values of watershed landscapes under the three different environmental simulation scenarios were generally convergent.

The high-risk areas all accounted for less than 1% and were mainly distributed in Nanchang. Nanchang provides a useful model for organizing different risk levels in risk management. The city falls into a high-risk category needing special attention. Nighttime light patterns and population density (Figure 3) show very active human development in Nanchang. This activity coincides with a highly sensitive natural environment. This combination matches the “disturbance” level in disaster risk systems. Areas like this give greater importance to certain land uses. Urban building land deserves particular focus. The city should encourage efficient and compact use of its developed areas. Increasing green space coverage inside the city is important. Limiting the outward expansion of urban building land is also key. This efficient land use supports natural ecosystem development. The Gan River flows directly through Nanchang. This river brings rich wetland wildlife and unique natural scenery. Any management plans here must fully protect the river’s natural functions first. Protecting these functions allows maximum environmental benefits. At the economic level, the area can use its environmental resources. Developing nature-based tourism makes good sense. Building waterside wetland parks represents a strong option for this region. Nanchang also possesses deep historical roots. The city produced many notable historical figures. It contains numerous cultural sites and monuments. The area should fully use its rich local history and culture. Creating a distinctive tourism industry chain, reflecting local character and heritage, is recommended. Other medium-sized and smaller cities share similar high-risk profiles. These places often have dense populations and strong economies. Yet they frequently pay less attention to ecological protection. These cities require similar high-risk category management.

The northern, southern, and eastern parts of the Poyang Lake Ecological Economic Zone have the lowest ecological risk levels in the watershed landscape. These areas account for 49.37% to 57.11% of the total area. However, the risk coefficients for different scenarios vary. Overall, watershed planning needs to balance “development” and “protection.” As a result, the final outcome of this balance, based on the risk coefficient, can change depending on the watershed’s development stage.

When a risk factor of 0.4 is set as the threshold for classifying watershed protection areas, 57.11% of the study area is ecologically safe under a low-priority risk scenario (α = 0.8). But this proportion drops to 49.37% under a high-priority risk scenario. The main changes occur in medium-risk and high-risk thresholds. Cultivated areas in northern Jingdezhen, the eastern hilly area of Pengze County, and southeastern Yingtan—all currently low-risk—are likely to become medium-risk. Meanwhile, medium-risk cultivated areas in southern Yuganshen County along the Xinjiang River, north and south of Poyang County’s Bintian Reservoir, lakeside green zones in Lushan and Duchang counties, and South De’an County’s Zhelin Reservoir are prone to becoming high-risk and need attention.

In simpler terms, it is harder to tell if medium-risk landscapes are transitioning from restructuring to development when decision-makers’ priorities shift. Identifying these uncertain risk elements is the aim of scenario analysis. Decision-makers should focus on Jingdezhen, Yingtan, Xunyang, De’an, and Pengze counties, where the increase rate of ecological risk area exceeds 5% (Figure 8). They need to decide whether “developing” or “protecting” these elements best guides the study area’s future.

The key in specifying “development” or “protection” is to define the area’s future direction. Since the original purpose of setting assessment priorities is to reduce uncertainty, decision-makers should use the floating range of α as a result of risk simulation in practical judgments. They should also adjust planning goals according to different stages of watershed development and changes in the natural environment.

3.3. Risk Classification Management Strategy

Future management can sort areas by high, medium, and low risk levels.

Areas facing high-risk threats need careful land planning and strict controls. This applies to Nanchang City, Jingdezhen City, the main part of Poyang Lake, Yongxiu County, Yugan County, and Duchang County. These places require attention whether their situation achieves a “risk-negligent scenario” or “risk-emphasized scenario”. Monitoring land use changes is also vital in specific high-change zones. Key areas include the southern Xinjiang River in Yugan County, the northern and southern parts of Bintian Reservoir in Poyang County, the lakeside green areas near Lushan City and Duchang County, and the southern farmland near Zhelin Reservoir in De’an County. People can carefully use groundwater resources. This approach helps make up for water shortages when lakes dry up.

Medium-risk areas cover most of the region. Managing these zones is very important. Future work should focus on key natural areas around Poyang Lake, Jinshi Lake, Junshan Lake, and southern Qinglan Lake. Special attention belongs to the hilly land north of Jingdezhen City, east of Pengze County, and the farmland southeast of Yingtan City. These spots may become medium-risk zones. Forests and farms dominate these areas. Actions should increase plant cover. Creating flood buffer zones near water bodies is necessary. Using wetlands fully helps manage floodwaters.

Low-risk ecological areas need protection. Keeping water levels stable in Baiyangdian Lake and nearby rivers is essential. This protects these safer zones. The smaller waterways around Poyang Lake and northwest town–country areas deserve particular focus. Land use stability appears uneven here, and safety is fragile. A “protection-first” approach fits best. Keeping farmland and forest areas intact is crucial. Maintaining healthy natural systems in these places prevents damage to wetlands and waterways.

4. Conclusions

From a social–ecological system viewpoint, the Poyang Lake Ecological and Economic Zone serves as the assessment focus. The adaptive cycle theoretical framework is used to rethink the concept of watershed ecological risk assessment. The research shifts from individual landscape risk indicators to systematic and comprehensive landscape assessment and simulation.

The spatial layout and dynamic changes in the watershed system are thoroughly analyzed by using potential connectivity and resilience indicators. OWA scenario simulation is then applied. This application provides decision-makers with scientific evidence and practical guidance for formulating planning strategies.

• The ecological risk of the study area under the social–ecological system follows a specific spatial pattern. It centers around Poyang Lake’s water system, with scattered distributions in urban areas. Medium-risk areas make up the largest portion, while low-risk areas mainly lie in the southwest Yuan River plain region. Connection degree and connectivity degree stand as the most critical indicators. The risk ranges of connectivity and resilience align with the general trend of disturbance distribution.
• The system shows clear spatial differences in exposure and disturbance risk. The main water area of Poyang Lake has a “high exposure–disturbance” state. This is because of natural disaster impacts and intense human activities. In contrast, the Gan River, Yuan River, and Xin River basins have “low exposure–disturbance” conditions. Their social systems develop more slowly, and their economic resilience is weaker.
• Using the ordered weighted average method, the study reorganizes and simulates the ecological risk of the watershed. It identifies high- and low-risk areas less affected by policy changes: Nanchang City and Zhelin Reservoir are high-risk zones while surrounding watershed areas are low-risk. Focusing on medium- and high-risk regions, the simulation of α’s change rate shows a more than 5% increase in risk values across Jingdezhen, Yingtan, Xunyang District, De’an County, and Pengze County. Among these, Xunyang District experiences a 17.14% shift from medium to higher risk. This helps meet the decision-making needs for regional sustainable development under different policy scenarios.