Identification of Priority Areas for Ecological Restoration at a Small Watershed Scale: A Case Study in Dali Prefecture of Yunnan Province in China

Abstract

Conducting ecological restoration has emerged as a critical governance strategy for enhancing ecosystem diversity, stability, and sustainability. The scientific identification of priority restoration areas is a prerequisite for effective ecological restoration projects. Current research on identifying priority restoration zones predominantly relies on administrative-scale frameworks, and the reliability and scientificity of the identified results are somewhat insufficient. To address this gap, this study selected Dali Prefecture in Yunnan Province, a region characterized by dense river networks, as the research area to identify the priority areas of ecological restoration. In view of the application of the InVest model in watershed-scale restoration, biodiversity assessment, and other fields, we utilize sub-watershed units and the InVEST model, and five key ecosystem services—water conservation, water purification (N/P), habitat quality, climate regulation, and soil retention—were quantified. Temporal changes in these services from 2015 to 2020 were analyzed alongside ecological risk assessments and restoration zoning. Priority areas were further identified through Ordered Weighted Averaging (OWA) operators under varying decision-making preferences. The optimal threshold for watershed delineation was determined as 11.04 km², resulting in 1513 refined sub-watershed units after correction, with 71.59% concentrated in the 10–50 km² range. A spatial analysis revealed an east-to-west gradient in ecosystem service distribution, where eastern regions consistently exhibited lower values compared to central and western areas. From 2015 to 2020, soil retention per unit area increased by 5.09%, while water purification for N and P showed marginal improvements of 0.97% and 0.39%, respectively. Conversely, water conservation declined significantly by 10.00%, with carbon sequestration and biodiversity protection experiencing slight reductions of 1.74% and 1.92%, all within a 2% variation margin. Ecological risk zoning identified low-risk areas (grades 1–3) predominantly in western and northeastern Dali, encompassing 1094 sub-watersheds (77.36% by count and 73.92% by area), while high-risk zones (grades 4–5) covered 386 units (26.08% by area). Integrating ecological quality and risk levels, the study area was classified into four functional zones: Zone I (high quality, high risk), Zone II (low quality, high risk), Zone III (low quality, low risk), and Zone IV (high quality, low risk). With increasing risk tolerance, the priority restoration areas expanded from eastward to central regions. Based on the scenario simulations under ecological priority, status quo, and development-oriented policies, the critical restoration areas include the Sangyuan River Basin, mid-reach of the Juli River, and upper Miyu River. This methodology provides a theoretical and technical foundation for ecosystem service enhancement and degraded ecosystem rehabilitation in Dali Prefecture and similar regions.

1. Introduction

The accelerated processes of urbanization and industrialization have precipitated severe environmental degradation marked by irrational resource exploitation, manifesting in critical challenges such as environmental pollution [1], biodiversity collapse [2], and compromised ecosystem functionality [3]. These issues now constitute major constraints on socioeconomic development, garnering global attention. Since the 21st century, ecological restoration has emerged as a critical strategy for maintaining ecosystem stability and biodiversity conservation [4], receiving prioritized attention from international bodies and national governments. In response to global ecosystem degradation, the United Nations launched the “United Nations Decade on Ecosystem Restoration (2021–2030)” to reverse these trends. China faces persistent ecological challenges, including water quality deterioration and declining biodiversity indices [5], underscoring the urgent need for integrated, systematic, and holistic approaches to regional ecological restoration as a cornerstone of its ecological civilization construction [6]. A pressing current issue lies in translating systemic and integrative theoretical frameworks into localized restoration practices—identifying ecological vulnerabilities; delineating priority zones; and implementing comprehensive protection, restoration, and management strategies [7]—to align with China’s sustainable development imperatives.

The National Major Project Plan for the Ecosystem Conservation and Restoration of Critical Regions (2021–2035), promulgated in June 2022, explicitly advocates transitioning from single-element remediation to systemic and holistic ecosystem governance [8], emphasizing the imperative to address ecological challenges through integrated landscape management [9]. This strategic shift underscores that scientifically delineating priority ecological restoration zones serves as the foundation for advancing synergistic regional development and safeguarding ecological security. Current methodologies for identifying territorial ecological restoration priorities predominantly rely on administrative jurisdictional units as spatial frameworks. Illustrative examples include the work of Zhang et al. [10], who developed a multi-categorical restoration zoning system in Dayu County, Jiangxi Province, integrating mining rehabilitation, watershed management, farmland consolidation, rural settlement restructuring, and transportation corridor optimization. Similarly, Fu et al. [11] employed an ecological security pattern analysis to demarcate conservation–restoration hotspots in Hezhou City, Guangxi Zhuang Autonomous Region, demonstrating the application of spatial prioritization techniques at subnational scales.

However, administrative unit-based approaches exhibit systemic limitations, including fragmented interdepartmental planning, the disjointed implementation of sector-specific projects, the disarticulated management of interconnected ecological processes, and localized restoration outcomes that fail to coalesce into landscape-scale resilience. These challenges highlight the urgent need to adopt ecosystem-based spatial governance frameworks, which operationalize whole-territory system analysis and cross-scale coordination mechanisms. Prior research often lacks fine-grained spatial resolution due to administrative boundaries, masking intra-regional variability. By contrast, our 1513 sub-watershed units (average size 19.46 km²) capture micro-ecological processes and reveal gradients (e.g., west–east declines in water yield) that administrative units obscure. This spatial precision supports targeted restoration planning, offering actionable insights for local policymakers. Such frameworks align with global initiatives like the UN Decade on Ecosystem Restoration, advocating for harmonized strategies that transcend administrative boundaries to achieve ecological integrity and functional connectivity.

A watershed constitutes a fundamental unit of both physical geography and socioeconomic development, integrating forest, river, lake, agricultural, and urban subsystems into a complex socio-ecological network [12,13,14]. Functioning as a geographically bounded unit, watersheds inherently serve as optimal mediators for coordinating multi-scale spatial interactions and bridging urban–rural planning elements. Such basin-scale planning frameworks inherently preserve ecosystem integrity while fostering human–nature synergies [15,16].

Recent advancements highlight the paradigm shift toward watershed-based ecological restoration transcending administrative boundaries, emerging as a critical frontier in restoration science [16,17,18]. Small watershed-scale analyses prove particularly effective in elucidating ecological process mechanisms, with contemporary research demonstrating dual trajectories: intensified micro-level investigations and synthesized macro-level integrations. This methodological progression significantly enhances the scientific validity and operational reliability of restoration zone identification. Nevertheless, current watershed-scale studies exhibit three critical gaps: (1) insufficient depth in mechanistic understanding, (2) fragmented analytical frameworks lacking systemic coherence, and (3) a predominant focus on hydrological elements rather than integrated natural–social–economic systems. Addressing these limitations requires developing transdisciplinary watershed management models that explicitly incorporate geomorphological specificity, socioeconomic drivers, and ecological thresholds—a pivotal challenge demanding urgent scholarly attention.

Dali Bai Autonomous Prefecture, situated at the ecotone between the Western Yunnan Plateau (Dianxi Gaoyuan) and the Hengduan Mountains, serves as the headwater region of the Red River (Yuanjiang) and is traversed by three major international rivers: the Nu (Salween), Lancang (Mekong), and Jinsha (Upper Yangtze). This biogeographically strategic location endows the region with exceptional ecological value juxtaposed with acute environmental fragility [19], characterized by diverse climatic zones, heterogeneous vegetation communities, and globally irreplaceable endemic biodiversity. Critical ecological challenges include accelerating species depletion, severe soil erosion, and compounding water quality deterioration across interconnected watershed systems. This study establishes Dali Prefecture as a pivotal case for advancing watershed-scale ecological restoration methodologies. By analyzing remotely interpreted land use data (2015–2020), we implement an integrated analytical framework combining ArcGIS-based hydrological modeling for optimal sub-watershed delineation, InVEST-driven ecosystem service quantification, ecological risk assessment through landscape disturbance indices, and spatial prioritization using Ordered Weighted Averaging (OWA) operators. The methodology systematically identifies critical thresholds for sub-watershed classification while elucidating spatiotemporal patterns of ecosystem service degradation and ecological risk propagation mechanisms. The findings provide spatially explicit insights into eco-geomorphic interactions at the watershed scale, generate decision-support matrices for restoration priority allocation, and establish technical protocols that harmonize ecological integrity with socioeconomic feasibility. This interdisciplinary approach offers actionable strategies for enhancing regional ecosystem service capacities and informs adaptive governance in ecologically fragile montane regions.

2. Study Area Overview

2.1. Physiographic Characteristics of the Study Area

Geospatial Context

Dali Prefecture, the sole Bai Autonomous Prefecture in China, is situated in west–central Yunnan Province, bounded by the Lancang River to the east and the Jinsha River to the south. Spanning the geographic coordinates 98°52′ E to 101°03′ E longitude and 24°40′ N to 26°42′ N latitude, it covers a total area of 29,459 km², ranking as the fifth-largest administrative division in Yunnan Province by territorial extent. The prefecture extends 320 km at its maximum east–west axis and 270 km north–south. Administratively, it comprises one city, eight counties, and three ethnic autonomous counties: Dali City; Xiangyun, Binchuan, Midu, Yunlong, Jianchuan, Eryuan, Heqing, and Yongping Counties; and Weishan Yi and Hui Autonomous County, Nanjian Yi Autonomous County, and Yangbi Yi Autonomous County.

Dali Prefecture borders Chuxiong Prefecture to the east, Pu’er City and Lincang City to the south, Baoshan City and Nujiang Prefecture to the west, and Lijiang City to the north. As a strategic corridor in Yunnan Province for northward access to Sichuan and Tibet and southward connectivity to Myanmar’s maritime outlets, it occupies a pivotal position in Yunnan’s international transportation network linking Southeast and South Asia. Concurrently, Dali serves as the prefecture with the most advanced transportation infrastructure in western Yunnan and functions as a regional economic and logistical hub. The substantial passenger flow, cargo flow, and information exchange generated by its connectivity endow it with extensive transportation influence across Southwest China, Southeast Asia, and South Asia, solidifying its status as a strategic stronghold in China’s southwestern frontier and a critical transportation nexus in western Yunnan. The geographic context of the study area is illustrated in Figure 1.

3. Research Methodology and Data Sources

3.1. Research Methodology

3.1.1. Hydrological Analysis

This study employs the Hydrology toolsuite in ArcGIS 10.2 to derive hydrological features through sequential geoprocessing: depression filling, flow direction determination, flow accumulation computation, stream network extraction, and watershed delineation. The resulting watershed units were subsequently refined through land use data integration to enhance hydrological spatial accuracy.

Depression filling was applied to the Digital Elevation Model (DEM) prior to a flow direction analysis to mitigate computational errors in subsequent hydrological processes. The flow direction analysis was conducted on the depression-free DEM using the D8 algorithm (eight-direction pour point method), which determines flow paths based on the maximum slope descent. The flow accumulation analysis assumes that each raster cell conveys a unit of water flow, calculating the Number of Inflow Paths (NIP) for each cell based on topographic gradients (i.e., flow from higher to lower elevations). Cells with elevated NIP values correspond to valley regions, whereas NIP = 0 indicates watershed divides. Surface runoff initiation is defined by a user-specified threshold (Y), where stream networks are extracted under the condition NIP > Y. This threshold critically governs the spatial accuracy of watershed boundary delineation. Sub-basin delineation involved tracing all upstream cells contributing to valley cells within the flow accumulation matrix. Key natural hydrological parameters, including total drainage area, number of drainage units, stream network length, and drainage density, were quantified to characterize the watershed system.

The watershed units delineated through the DEM-based analysis exhibited artificially linear boundaries or parallel contours in the reservoir and lake catchments. To better represent hydrological realism, this study refined the watershed units using 2020 land use data while preserving topographic-driven flow continuity (i.e., discharge from higher to lower elevations) and maintaining hydrological integrity within naturally defined boundaries.

3.1.2. Ecosystem Service Quantification

This study employs the widely adopted InVEST (Integrated Valuation of Ecosystem Services and Trade-offs) model, a multi-modular framework renowned for its comprehensive capacity in ecosystem service assessment. Characterized by its systematic integration of biophysical and socioeconomic parameters, the model enables the spatial quantification of ecosystem services—a critical advancement that resolves historical limitations in service characterization and visual representation [20]. Particularly suited for small watershed applications, its spatially explicit outputs facilitate precise identification of service provision hotspots and degradation thresholds, establishing it as an indispensable tool for watershed-scale ecological spatialization studies [21].

Watersheds function as integrated systems where small catchments serve as optimal natural management units for addressing soil erosion and water pollution control. Considering the ecological challenges within the study area, the ecosystem services prioritized for ecological restoration projects include water conservation, purification, habitat quality maintenance, climate regulation, and soil retention.

• Water yield regulation 

The water storage capacity plays a pivotal role in hydrology, influencing both peak discharge and water supply [22]. Water conservation services were quantified using the Water Yield module of InVEST 3.10.2, which implements a water budget approach [23,24]. This methodology mechanistically integrates key hydrological drivers, including precipitation inputs, soil depth profiles, root zone characteristics, and partitioned surface evaporation and vegetation transpiration fluxes.

The calculation formula is as follows:

$$Y_{xj}=(1-{\displaystyle \frac{{AET}_{xj}}{P_x}})P_x$$

(1)

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

(2)

$$R_{xj}={\displaystyle \frac{k\times {ET}_0}{P_x}}[k=\mathrm{m}\mathrm{i}\mathrm{n}(1,{\displaystyle \frac{LAI}{3}}\left)\right]$$

(3)

$${\omega }_x=Z\times {\displaystyle \frac{{AWC}_x}{P_x}}$$

(4)

$${AWC}_x=\mathrm{m}\mathrm{i}\mathrm{n}(S_{dmax},R_d)\times PAWC$$

(5)

$$R=\mathrm{m}\mathrm{i}\mathrm{n}(1,{\displaystyle \frac{249}{V}})\times \mathrm{m}\mathrm{i}\mathrm{n}(1,{\displaystyle \frac{0.9\times TI}{3}})\times \mathrm{m}\mathrm{i}\mathrm{n}(1,{\displaystyle \frac{K}{300}})\times Y$$

(6)

where

Y_xj represents the total water yield of grid unit x in land class j;

P_x is the annual average precipitation of grid unit x;

AET_xj denotes the annual mean evapotranspiration of grid unit x in land class j;

R_xj is the Budyko aridity index of grid unit x in land class j;

ω_x is the vegetation-available water coefficient;

k is the vegetation coefficient;

LAI denotes the leaf area index;

Z is the seasonal rainfall index (ranging from 1 to 30);

AWC_x represents the available water capacity;

PAWC is the plant available water capacity (ranging from 0–1);

S_dmax is the maximum soil depth;

R_d indicates the root distribution depth;

R refers to the water conservation capacity per grid unit (mm);

K is the saturated hydraulic conductivity (cm/d);

V represents the watershed coefficient;

TI is the topographic index;

Y denotes the water yield.

b.

Water purification 

Water purification services were assessed through the NDR (Nutrient Delivery Ratio) module of InVEST 3.10.2 to quantify total nitrogen (TN) and total phosphorus (TP) retention capacities. The magnitude of these retention metrics inversely reflects regional water purification efficacy, where elevated TN/TP retention levels indicate diminished service performance [25,26]. The module mechanistically simulates ecosystems’ capacity to intercept diffuse-source pollutants (N/P), explicitly modeling retention processes across heterogeneous landscapes.

$${ALV}_i={\mathrm{H}\mathrm{S}\mathrm{S}}_{\mathrm{i}}·{\mathrm{p}\mathrm{o}\mathrm{l}}_{\mathrm{i}}$$

(7)

$${HSS}_i={\displaystyle \frac{{\gamma }_i}{{\gamma }_{\omega }}}$$

(8)

$${\gamma }_i=\mathrm{l}\mathrm{g}\sum _i{R}_i$$

(9)

where

ALV_i = the adjusted nitrogen (phosphorus) load value of grid cell i (kg);

pol_i = the output coefficient of grid cell i;

HSS_i = the hydrological sensitivity score of grid cell i;

γ_i = the runoff index of the study area for grid cell i;

γ_ω = the average runoff coefficient of grid cell i;

$\sum _{\mathrm{i}}{\mathrm{R}}_{\mathrm{i}}$ = the sum of the water yield across all grid cells.

c.

Habitat quality 

Habitat quality operationalizes the capacity of ecosystems to sustain viable conditions for species persistence and population viability, serving as a composite metric that quantifies both regional biodiversity integrity and habitat-type resilience to anthropogenic disturbances [27,28]. The assessment was conducted using the Habitat Quality module (v3.10.2) of InVEST, which leverages a threat-sensitivity framework that systematically weights habitat suitability against landscape degradation pressures. The computational framework is formalized as follows:

$$D_{xj}=\sum _{r=1}^R\sum _{y=1}^{Y_r}\left({\displaystyle \frac{w_r}{{\sum }_{r=1}^R{w}_r}}\right)r_y{i}_{rxy}{\beta }_x{s}_{jr}$$

(10)

$$i_{rxy}=\left(1-{\displaystyle \frac{d_{xy}}{d_{rmax}}}\right) \text{(Linear)}$$

(11)

$$i_{rxy}=\mathrm{e}\mathrm{x}\mathrm{p}(-{\displaystyle \frac{-2.99d_{xy}}{d_{rmax}}}) \text{(Exponential)}$$

(12)

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

(13)

where

R = the number of threat factors;

Y_r = the total grid count across all threat layers on land type soil layers;

W_r = the weight of threat factors;

r_y = the number of instances for a given threat factor;

i_rxy = the impact of threat factor in grid cell x’s habitat on grid cell y;

d_xy = the distance between grid cell x and grid cell y;

d_rmax = the maximum influence distance of threat factors;

β_x = the protection level of grid cell x;

H_j = ecological suitability;

k = half-saturation constant.

d.

Climate regulation 

Climate regulation services were quantified through the Carbon Storage and Sequestration module (v3.10.2) of InVEST. This framework operationalizes ecosystem carbon stock estimation by synergizing land use/land cover (LULC) data with stratified carbon pool dynamics. The model stratifies total carbon storage into four biophysically distinct reservoirs: (1) aboveground biomass (AGB), representing live vegetation carbon in standing biomass; (2) belowground biomass (BGB), quantifying root system carbon stocks; (3) soil organic carbon (SOC), encompassing stabilized organic compounds within soil matrices; and (4) dead organic matter (DOM), accounting for necromass and litter-derived carbon fluxes. The computational algorithm is formalized as follows:

$$C_i=C_{i-above}+C_{i-below}+C_{i-dead}$$

(14)

$$C_{total}=\sum _{i=1}^n{C}_i{S}_i$$

(15)

where

i = the land use type index;

C_i = the total carbon density of land use type i;

C_i-above = the aboveground vegetation carbon density of land use type i;

C_i-below = the belowground live root carbon density of land use type i;

C_i-soil = the soil carbon density of land use type i;

C_i-dead = the vegetation litter carbon density of land use type i;

C_total = the aggregated total carbon density.

e.

Soil conservation 

Soil conservation capacity was quantified using the Sediment Delivery Ratio (SDR) module (v3.10.2) of InVEST. This framework operationalizes retention dynamics by stratifying total soil conservation into two mechanistically distinct components: (1) sediment retention volume, representing the physical trapping of eroded particulates, and (2) erosion mitigation volume, quantifying reductions in soil loss attributable to conservation engineering measures or natural landscape-mediated interception processes [29,30]. The computational architecture is formalized as follows:

$${rkls}_i=R_i\times K_i\times {LS}_i$$

(16)

$${usle}_i=R_i\times K_i\times {LS}_i\times C_i\times P_i$$

(17)

$${Sedr}_i={Se}_i\sum _{j=1}^{i=1}{Usle}_j\prod _{z=j+1}^{i=1}(1-{Se}_z)$$

(18)

$${Sedret}_i={rkls}_i-{usle}_i+{sedr}_i$$

(19)

where

rkls_i = the potential soil erosion of grid cell i;

R_i = the rainfall erosivity of grid cell i;

K_i = the soil erodibility of grid cell i;

LS_i = the slope length gradient factor of grid cell i;

usle_i = the actual soil erosion of grid cell i;

C_i = the vegetation cover factor of grid cell i;

P_i = the conservation practice factor of grid cell i;

Sedr_i = the sediment retention capacity of grid cell i;

Se_i = the sediment retention efficiency of grid cell i;

Usle_j = the sediment yield from upslope grid cell j;

Sedret_i = the soil conservation capacity of grid cell i.

The spatiotemporal variations of five ecosystem services at the watershed scale were analyzed using the Zonal Statistics tool in ArcGIS 10.2. An overlay analysis of the five ecosystem service types for 2015 and 2020 was performed via the Raster Calculator to quantify watershed-scale change magnitudes between these years. Areas were classified into three categories based on variation trends: ecosystem service reduction zones (change <0), stability zones (change = 0), and enhancement zones (change >0). To precisely identify ecological degradation hotspots, severity thresholds were applied: severe degradation (>20% decline), moderate degradation (10–20% decline), minor degradation (5–10% decline), ecological stabilization (−5–5% variation), and ecological improvement (<−5% decline).

$$∆{ES}_{ij}={ES}_{2020ij}-{ES}_{2015ij}$$

(20)

$$∆D_{ij}={\displaystyle \frac{({ES}_{2020ij}-{ES}_{2015ij})}{{ES}_{2015ij}}}$$

(21)

where

ES_ij = the change in the type i ecosystem service value for watershed unit j;

ES_2020ij = the value of the type i ecosystem service in watershed unit j for 2020;

ES_2015ij = the value of the type i ecosystem service in watershed unit j for 2015;

ΔD_ij = the degradation degree of the type i ecosystem service in watershed unit j.

3.1.3. Integrated Ecosystem Service Index (ESI) Analysis

The composite ecosystem services index (ESI) was developed by synthesizing the ecological restoration index framework established by Song, W. et al. [31,32]. Elevated ESI values indicate enhanced composite ecological integrity, with higher magnitudes reflecting superior systemic functionality across coupled ecological processes.

$$ESI=\sum _1^n{\phi }_i\times E_i$$

(22)

$$E_i={\displaystyle \frac{{ES}_i-{ES}_{i\mathrm{m}\mathrm{i}\mathrm{n}}}{{ES}_{i\mathrm{m}\mathrm{a}\mathrm{x}}-{ES}_{i\mathrm{m}\mathrm{i}\mathrm{n}}}}$$

(23)

where

φ_i = the weight assigned to a type i ecosystem service;

E_i = the standardized value of a type i ecosystem service;

ES_i = the measured value of a type i ecosystem service;

ES_imin = the minimum observed value of a type i ecosystem service;

ES_imax = the maximum observed value of a type i ecosystem service.

3.1.4. Ecological Risk Assessment (ERA)

The ecological risk assessment framework was developed following the methodology established by Fu M D et al. [33], which integrates dynamic changes in ecosystem services through a coordinated development lens. This model stratifies risk quantification into systemic perturbations caused by service imbalances, formalizing the computational architecture as follows:

$$R=P\times D$$

(24)

$$P=\sum _i^{\mathrm{n}}{p}_i$$

(25)

$$D_i={ES}_{2020i}-{ES}_{2015i}$$

(26)

where

R is the ecological risk coefficient;

P is the probability of a negative transformation in ecosystem services;

p_i = the probability of a negative transformation for the i-th ecosystem service (0 ≤ p_i ≤ 1);

n = the total number of assessed services (e.g., n = 5).

Threshold examples:

P = 1 when all 5 services undergo negative transformations;

P decreases in 0.2 increments per service preservation (0.8, 0.6,...,0);

D = the total loss magnitude from negative transformations;

D_i = the quantified loss of the i-th ecosystem service.

The ecological degradation risk stratification protocol was operationalized following the methodological framework of Pan, X. et al. [34], which aggregates reverse-transformed ecosystem service fluxes across watershed units. Degradation severity tiers were classified by cumulative service flux thresholds: watersheds ranking within the top 10%, and 10th–20th and remaining percentiles were designated as light, moderate, and severe loss regimes, respectively. This stratification yielded a systemic service depletion assessment matrix (

Table 1. Ecosystem service-based ecological degradation grading assessment matrix.

Risk Level	Mild Impairment	Moderate Loss	Severe Loss
High Probability (1, 0.8)	3	4	5
Moderate Probability (0.6)	2	3	4
Low Probability (0.4, 0.2)	1	2	3

).

3.2. Data Source

The study area data encompassed ecosystem service outputs derived from the InVEST (Integrated Valuation of Ecosystem Services and Trade-offs) model, and geospatial processing results from IDRISI 17.0. Land use/cover (LULC) data were obtained from the China Multi-period Land Use Remote Sensing Monitoring Dataset (CNLUCC) hosted by the Resource and Environment Science and Data Center, Chinese Academy of Sciences (https://www.resdc.cn). Through radiometric correction, watershed-specific cropping, and supervised classification with an overall accuracy exceeding 90% (kappa coefficient >0.85), we generated 2015 and 2020 LULC maps at 100 m × 100 m spatial resolution. Biophysical coefficient tables and model parameters for the InVEST modules were calibrated using field-collected data from the study area, supplemented by guidelines from the InVEST 3.13.0 User’s Guide. Watershed accumulation thresholds and catchment area thresholds were determined through a DEM-based hydrological analysis (8-flow direction algorithm) in ArcSWAT 2012. All model inputs and data sources are systematically cataloged in

Table 2. Data inputs and sources for the InVEST model.

Dataset	Format	Source and Unit	Service Type
Maximum Rooting Depth (cm)	TIFF	Harmonized World Soil Database (HWSD) v2.0, jointly developed by Food and Agriculture Organization (FAO) and International Institute for Applied Systems Analysis (IIASA), mm	Water Yield Regulation
Annual Precipitation (mm)	TIFF	Resource and Environment Data Cloud Platform (http://www.resdc.cn), mm	Water Yield Regulation
Plant Available Water Capacity (PAWC) (%)	TIFF	Calculated from soil texture analysis within the study area, ranging 0–1	Water Yield Regulation
Annual Mean Reference Evapotranspiration(ET0) (mm)	TIFF	National Earth System Science Data Center (http://www.geodata.cn/), mm	Water Yield Regulation
Current Land Use/Land Cover (LULC) Map	TIFF	Institute of Geographic Sciences and Natural Resources Research, Chinese Academy of Sciences (https://www.resdc.cn)	Water Yield Regulation, Water Purification, Climate Regulation, Soil Conservation, Habitat Quality
Watershed Accumulation Threshold	shp	Threshold: 12,268	Water Yield Regulation, Water Purification, Soil Conservation
Water Yield Biophysical Coefficient Table	csv	Determined through integration of regional conditions and existing studies with reference to InVEST model parameters	Water Supply Provision
Seasonal Constant	constant	Derived from daily precipitation records (2015–2020) within the study area: Seasonal constant κ = 6 (2015), κ = 5.6 (2020)	Water Supply Provision
Digital Elevation Model (DEM)	TIFF	Geospatial Data Cloud (http://www.gscloud.cn/)	Water Purification, Soil Conservation
Water Purification Biophysical Coefficient Table	csv	Determined through integration of regional conditions and existing studies with reference to InVEST model parameters	Water Purification
Carbon Pool	csv	Determined through integration of regional conditions and existing studies with reference to InVEST model parameters	Climate Regulation
Rainfall Erosivity Factor (R)	TIFF	National Earth System Science Data Center (MJ·mm/(ha⋅hr))	Soil Conservation
Soil Erodibility Factor (K)	TIFF	Calculated based on soil properties of the study area, unit: t·ha·hr/(MJ·ha·mm)	Soil Conservation
Biophysical Coefficient Table	csv	Determined through integration of regional conditions and existing studies with reference to InVEST model parameters	Soil Conservation
Kb, IC0, SDRmax	constant	Adopted default valuesL: Kb = 2, IC0 = 0.5, SDRmax = 0.8	Soil Conservation
Threat Factor Data	csv	Determined through integration of regional conditions and existing studies with reference to InVEST model parameters	Habitat Quality
Threat Source Data	TIFF	Determined through integration of regional conditions and existing studies with reference to InVEST model parameters	Habitat Quality
Half-Saturation Coefficient	TIFF	K default: 0.5	Habitat Quality

, adhering to TRUST principles for transparent data reporting.

The input data for the IDRISI 17.0 model originated from InVEST model outputs. The 2020 ecosystem services’ quantification results (five service types) were processed in ArcGIS 10.2 and converted to ASCII format and subsequently transformed into IDRISI’s native .rst format through the CONVERT module. OWA (Ordered Weighted Averaging) simulations were implemented under multiple risk decision scenarios, with the module’s data requirements as detailed in

Table 3. Data input formats and accuracy specifications for the IDRISI 17.0 OWA model.

Decision Factor		Format	Precision
Water Conservation Service		IDRISI	100 m × 100 m
Water Purification Service	N Removal Service	IDRISI	100 m × 100 m
	P Retention Service	IDRISI	100 m × 100 m
Habitat Quality Service		IDRISI	100 m × 100 m
Climate Regulation Service		IDRISI	100 m × 100 m
Soil Retention Service		IDRISI	100 m × 100 m

:

4. Results and Discussion

4.1. Threshold Determination and Feature Analysis of Watershed Units

4.1.1. Threshold Impacts on River Network Extraction Accuracy

The watershed unit threshold refers to the defined value range for extracting river network information within specific research areas. Significant discrepancies in extracted river network data emerge under varying thresholds due to the region’s dense hydrological system. Rational determination of this threshold facilitates a comprehensive and accurate understanding of watershed hydrological characteristics. As an integrated parameter reflecting watershed attributes, drainage density demonstrates pronounced sensitivity to threshold variations. Introducing drainage density to identify the optimal threshold effectively minimizes subjective bias, thereby providing a scientific basis for precise watershed delineation. Given the region’s complex river networks and synthesizing previous studies, thresholds within the range (0–4500 km²) were systematically applied to extract watershed unit river network data, enabling a quantitative analysis of threshold impacts on hydrological parameters such as watershed unit characteristics.

As shown in Figure 2, the total watershed unit area exhibits a stepwise decreasing trend with increasing thresholds. The threshold–area fitting curve reveals that the total area initially decreases sharply and subsequently stabilizes. Within the 0–180 range, the total watershed unit area remains constant despite threshold increases, matching the study area’s total extent. Between 180 and 2700, the watershed unit area undergoes significant reduction with threshold increments, while in the 2700–4500 km² range, the total area continues to decline but gradually stabilizes.

An analysis of the area–threshold relationship across varying watershed thresholds reveals that the total areal coverage of hydrological units meets the research requirements (i.e., matches the total area of the study region) when the threshold is ≤180 km².

4.1.2. Determination of the Optimal Threshold for Hydrological Units

Drainage density serves as a comprehensive descriptor of watershed characteristics, reflecting heterogeneity across different basins. To determine the optimal threshold for Dali Prefecture, we introduced drainage density (defined as total stream length per unit area) to quantify threshold’s impacts on hydrological feature extraction. As shown in Figure 2, when the threshold increases from 0.9 km² to 9 km², drainage density decreases sharply from 75.77 m/km² to 25.68 m/km², accompanied by a steep slope in the curve. Between thresholds of 18 km² and 108 km², drainage density declines from 17.15 m/km² to 7.10 m/km², with gradually slowing curve slopes indicating diminished sensitivity to threshold increments. At thresholds exceeding 108 km² up to 180 km², drainage density stabilizes, decreasing marginally from 7.10 m/km² to 5.95 m/km². Power function fitting of the threshold–density relationship yields an R² value of 0.99834, approaching unity and confirming excellent model fidelity.

The point of tangency between the power function and the linear function represents the inflection point, mathematically termed the stability point. This study employs the tangent method of power-law and linear functions to identify inflection points. Before the inflection point, the threshold change rate (Δx) is smaller than the drainage density change rate (Δy). Beyond the inflection point, Δx > Δy. At the inflection point, Δx = Δy, indicating the threshold at which drainage density stabilizes with respect to threshold variations.

Thus, the following system of equations is established:

$$\mathrm{y}=-\mathrm{x}+\mathsf{\alpha }$$

(27)

MATLAB (https://www.mathworks.com/products/matlab.html) calculations demonstrate that when a = 33.5544, the system yields a unique solution of x = 11.0412. No valid solutions exist for a < 33.5544, while dual solutions emerge when a > 33.5544. Therefore, x = 11.0412 is identified as the optimal threshold.

4.1.3. Watershed Unit Delineation Refinement and Statistical Results

Using the optimal threshold of 11.0412 km², the preliminary watershed unit delineation is shown in Figure 3a:

The study area exhibits complex hydrographic features, including 152 river channels with a total length of 37.55 km and a reservoir/lake area of 380.68 km². To ensure the integrity of the watershed units, the preliminary delineation was cross-validated against 2020 land use inventory data and remote sensing imagery. The refined delineation resulted in 1513 watershed units (Figure 3b), with the detailed statistical outcomes summarized in

Table 4. Stratified area classification statistics of watershed units in Dali Prefecture.

Area Class (km2)	Number of Watershed Units	Percentage (%)
A < 10	395	26.11%
10 ≤ A < 20	626	41.37%
20 ≤ A < 30	259	17.12%
30 ≤ A < 40	116	7.67%
40 ≤ A < 50	67	4.43%
A < 50	50	3.30%
Total	1513	100.00%

.

4.2. Decadal Shifts in Ecosystem Services at Small-Watershed Scale: A Case Study in Dali Prefecture

4.2.1. Spatiotemporal Dynamics of Ecosystem Service Provisioning in Sub-Watershed Systems

(1)

Water Yield Regulation Service

The total water conservation capacity in the study area was 1484.80 × 10⁶ mm in 2015 and 1462.69 × 10⁶ mm in 2020, with watershed unit averages of 533.06 mm/ha and 479.77 mm/ha, respectively. Both the total capacity and per-unit-area values revealed a statistically significant decline (2015–2020), marked by a 10.00% reduction in water conservation services per hectare.

Spatially, water conservation capacity in Dali Prefecture exhibited a persistent west-high, east-low, and central-moderate distribution pattern during both years (Figure 4a,b), closely mirroring annual precipitation gradients. Distinct classification boundaries were observed: high-value zones concentrated in the northwestern sectors, medium-value zones formed a belt-like distribution, and low-value zones dominated the eastern regions.

Notably, 68.94% of the study area experienced declining water conservation capacity, with the most severe reductions clustered in the central and southern zones. Conversely, the northeastern and western regions displayed increasing trends (Figure 4c), suggesting spatially heterogeneous responses to climatic or anthropogenic drivers.

(2)

Water Purification Service

The total TN export in the study area was 15.80 × 10⁶ kg in 2015 and 15.72 × 10⁶ kg in 2020, with watershed-averaged values of 6.18 kg/ha and 6.12 kg/ha, respectively. Correspondingly, TP export decreased from 6.56 × 10⁶ kg to 6.53 × 10⁶ kg, with per-unit-area values declining from 2.57 kg/ha to 2.56 kg/ha over the same period. Both TN and TP exports exhibited decreasing trends (2015–2020), with reductions of 0.97% and 0.39% in per-unit-area export rates, respectively. These declines inversely reflect improvements in water purification services for nitrogen (N) and phosphorus (P), corresponding to 0.97% and 0.39% increases in nutrient retention efficiency.

Spatially, water purification services for N and P displayed a north-high, south-low, and central-low distribution pattern in both years, though classification boundaries remained indistinct. High-capacity zones clustered in northwestern Dali Prefecture, aligning with major river systems (e.g., Heihui River, Lancang River, and Yongping Great River). Low-capacity zones dominated the southern and central regions, particularly in upstream areas of the Misha River, Binju River, Sangyuan River, and Yuanjiang-Red River, exhibiting fragmented spatial patterns.

From 2015 to 2020, 54.48% and 55.48% of the study area experienced declines in N and P purification services, respectively. Degraded zones (covering 54.48% of the area) concentrated in central Heihui River, Luoluo River, Qingshui River, and Dada River basins. Conversely, enhanced purification services clustered in the central and western regions, including the Erhai Lake, Lancang River, Gonglang River, and Lixian River watersheds (Figure 5).

(3)

Habitat Quality Service

The habitat quality index in the study area was 1.48 × 10⁶ (dimensionless) in 2015 and 1.45 × 10⁶ in 2020. The watershed-averaged habitat quality index decreased from 0.52 to 0.51 during this period. Both total habitat quality services and per-unit-area indices exhibited a declining trend (2015–2020), with a 1.92% reduction in habitat quality per unit area.

As shown in Figure 6a,b, habitat quality displayed a western-high, central-low, and southeastern-low spatial pattern in both years. Low-value zones occurred in the Liandong River, Binju River, Daying River, midstream Juli River, and upstream Yuanjiang-Red River regions, where intensive anthropogenic disturbances correlate with higher socioeconomic development levels. Moderate-quality areas dominated central Dali Prefecture, while high-quality zones formed contiguous clusters in the western regions, demonstrating significant spatial aggregation.

From 2015 to 2020, 62.96% of the study area experienced habitat quality degradation, primarily concentrated in the northern and western watersheds (e.g., Heihui River, upstream Bijiang River, and Luoluo River). Conversely, habitat quality improvements focused on the Erhai Lake-centered block (Figure 6c).

(4)

Climate Regulation Service

The total carbon storage in the study area was 287.63 × 10⁶ t in 2015 and 282.62 × 10⁶ t in 2020, with watershed-averaged values of 100.50 t/ha and 98.75 t/ha, respectively. These metrics indicate a 1.74% decline in per-unit-area carbon sequestration capacity during 2015–2020, reflecting a downward trend in climate regulation services.

As illustrated in Figure 7a,b, climate regulation services exhibited a central-low, peripheral-high spatial pattern in both years, with overall service capacity remaining relatively robust. Low-capacity zones formed contiguous clusters, primarily along the Zhongying River, Daying River, Sangyuan River, Zhonghe River, and Henggou River. Moderate-capacity zones extended linearly from the Miju River through Erhai Lake to the Luwo River, while high-capacity areas dominated eastern Dali Prefecture, clustering around the Lancang River, Heihui River, Bijiang River, and Shunbi River watersheds.

Figure 7c reveals that 65.07% of the study area experienced declining climate regulation services from 2015 to 2020, with pronounced degradation in the western and northeastern regions. Affected areas included the upper Lancang River and the mid-reaches of the Shunbi River, Tulü River, Baishitou River, Duomei River, and Houshan River. Limited improvements occurred sporadically, concentrated in upstream Erhai Lake and its immediate periphery.

(5)

Soil Conservation Service

The total soil retention in the study area measured 300.15 × 10⁸ t in 2015 and 315.44 × 10⁸ t in 2020, with watershed-averaged values of 10,994.26 t/ha and 11,554.17 t/ha, respectively. Despite a 5.09% increase in per-unit-area soil retention services during 2015–2020, the overall soil retention capacity exhibited a declining trend at the ecosystem scale.

As depicted in Figure 8a,b, soil retention capacity followed a western-high, eastern-low, central-deficient spatial pattern in both years, with clearly demarcated classification boundaries. High-value zones formed linear concentrations along the Lancang River, Bijiang River, and Daoliu River in western Dali Prefecture. Moderate-capacity areas were sporadically distributed across the region, notably along the Duomei River and Houshan River. Low-capacity zones aligned with the Misha River–Heihui River flow axis, spanning northern and southern sectors of the prefecture, while minimal-capacity clusters dominated the eastern and central areas, exhibiting block-like distributions along the upper Yuanjiang–Honghe River, Sangyuan River, Zhonghe River, Liandong River, and Libidian River.

Notably, only 1.21% of the study area experienced reductions in soil retention services during 2015–2020 (Figure 8c).

4.2.2. Integrated Analysis of Ecosystem Services

(1)

Ecological Composite Index Assessment

Through standardized calculation and a spatial overlay analysis of five ecosystem services at the small-watershed scale, we derived the Ecosystem Services Composite Index (ESI) for 2015 and 2020. The ESI values were classified into five tiers using the natural breaks method (Jenks)—low (<1.84), moderately low (1.84–2.74), medium (2.74–3.38), moderately high (3.38–3.92), and high (>3.92)—as illustrated in Figure 9a,b. Temporal changes in ESI during 2015–2020 were quantified using raster-based calculation tools, with the results visualized in Figure 9c.

The Ecological Composite Index (ESI) serves as a critical indicator of regional ecological quality, where higher ESI values denote superior ecological conditions and vice versa. Our analysis reveals an overall declining trend in ecological quality across Dali Prefecture during 2015–2020.

Spatially, both the 2015 and 2020 ESI distributions exhibited a pattern of eastern low, western high, and central inferiority. Low-value zones (low and moderately low tiers) predominantly clustered in the central and eastern regions, notably within the Erhai Lake Basin, Yibin River, and Zhong River catchments. Moderate-value areas were primarily concentrated in the south, with scattered pockets elsewhere. High-value zones (moderately high and high tiers) dominated the western regions, displaying pronounced spatial aggregation. Notably, 58.06% of the prefecture’s area exhibited ESI reduction during 2015–2020, predominantly concentrated in the eastern regions with historically higher ecological quality.

As illustrated in Figure 10, the ESI tiers within the study area during both 2015 and 2020 were predominantly classified as moderate and relatively high, demonstrating favorable baseline ecological conditions across the region. However, the ESI exhibited an overall declining trend during 2015–2020, characterized by an expansion of low-tier ESI areas and the contraction of high-tier zones, while moderate-tier coverage exhibited a statistically significant increase. This pattern may be attributed to intensified anthropogenic disturbances resulting from tourism industry expansion and urban built-up land encroachment in Dali Prefecture, driving transitions from high-tier to moderate-tier classifications and concurrent proliferation of low-tier areas.

(2)

Ecological Risk Assessment

The ecological risk assessment in this study conceptualizes five types of ecosystem service degradation as risk receptors, employing a “Risk = Degradation Probability × Loss Magnitude” framework. As depicted in Figure 11, the spatial distribution of ecological risk tiers in Dali Prefecture reveals distinct geographical patterns. Low-risk zones (Tiers 1–2) are primarily concentrated in the central and northwestern regions, including the Erhai Lake Basin, upper Lancang River, and lower Bijiang River catchments. Moderate-risk areas (Tier 3) dominate the northern and southern sectors, such as the upper Heihui River, Juli River, and Leqiu River watersheds. High-risk zones (Tiers 4–5) exhibit clustered distribution in the northeastern and western parts of the prefecture, with scattered occurrences elsewhere. Notably, regions with elevated ecological risk tiers paradoxically overlap with areas of superior ecological quality—these zones, characterized by favorable natural endowments, have experienced cumulative degradation risks due to intensified developmental activities in recent decades. This spatial incongruity underscores the tension between ecological preservation and anthropogenic pressures, where historically robust ecosystems now face heightened vulnerability from expanding tourism infrastructure and urban land-use conversion.

As illustrated in Figure 12, excluding Tier 0 ecological risk, both the areal coverage of risk zones and the number of watershed units exhibit a marked decline with ascending risk tiers, a trend that intensifies progressively. A quantitative analysis reveals that ecological risk zones span 28,059.72 km² within the study area, predominantly classified as Tiers 1–3. These lower-risk tiers encompass 1094 watershed units, accounting for 73.92% of the total risk area and 77.36% of all watershed units. This spatial-statistical pattern demonstrates the predominance of lower-risk tiers across the region, suggesting limited systemic shifts in overall ecological risk profiles and indicating a generally stable ecological security status.

4.3. Delineation of Priority Ecological Restoration Zones in Dali Prefecture

4.3.1. Restoration Zoning and Optimization Through Trade-Offs Between Ecological Quality and Risk

To scientifically delineate the ecological restoration priorities and operationalize trade-off mechanisms between ecological quality and risk, watershed units were stratified into four typological zones based on spatial congruence (Figure 13): I, High-Ecological-Quality, High-Risk Zones; II, Low-Ecological-Quality, High-Risk Zones; III, Low-Ecological-Quality, Low-Risk Zones; and IV, High-Ecological-Quality, Low-Risk Zones.

(1)

Zone I (High Quality, High Risk)

Clustered predominantly in eastern Dali Prefecture, with scattered distributions elsewhere, this zone encompasses 207 watershed units (3805.90 km², 13.29% of the prefecture). Characterized by high baseline ecological quality yet elevated risk exposure, strategic interventions should prioritize sustainability-oriented adaptive management frameworks. The conservation of critical ecological spaces—including forests and high-coverage grasslands—must be reinforced through policy-supported restoration programs. Concurrently, tourism infrastructure development requires rigorous ecological connectivity planning to align with surrounding landscapes, fostering positive ecosystem succession and mitigating escalating risk trajectories.

(2)

Zone II (Low Quality, High Risk)

Aggregated in northeastern Dali with fragmented southern/northern occurrences, this zone spans 296 watershed units (4540.27 km², 15.86%). Marked by degraded ecosystems compounded by high-risk indices, it demands priority restoration governance. Regulatory measures must constrain urban expansion intensity while implementing targeted remediation protocols (e.g., erosion control and riparian buffer rehabilitation) to decouple land-use pressures from ecological degradation. The systematic integration of risk mitigation thresholds into regional spatial planning is imperative to reverse current trajectories.

(3)

Zone III (Low Quality, Low Risk)

Centered in mid-Dali with radial extensions from Erhai Lake Basin eastward, this zone covers 362 watershed units (7656.69 km², 26.74%). Despite minimal risk exposure, suboptimal ecological quality necessitates systemic quality enhancement initiatives. Urban renewal programs should prioritize retrofitting inefficient land uses and constructing multifunctional green infrastructure, including peri-urban protective forest belts and lakeside ecological corridors. Enhancing greenway networks and residential greening coverage will concurrently elevate ecological functionality and human livability in this densely populated sector.

(4)

Zone IV (High Quality, Low Risk)

Concentrated in northern/western Dali and overlapping with protected areas (e.g., nature reserves, national forest parks), this zone comprises 648 watershed units (12,628.63 km², 44.11%). Functioning as the prefecture’s ecological stabilization core, its high-quality/low-risk profile underscores the success of existing conservation regimes. To perpetuate this status, institutional reinforcement is critical: upgrading monitoring infrastructure, formalizing cross-jurisdictional governance protocols, and strictly regulating anthropogenic perturbations. Proactive buffering of protected area boundaries will further insulate these zones from external destabilizing factors.

4.3.2. Simulation-Based Identification of Priority Ecological Restoration Areas Using OWA Operators

(1)

Risk Trade-Offs in Ecosystem Service Scenario Simulations

The decision risk coefficient α embodies decision-makers’ risk preferences through the differential weighting of influencing factors, with varying α values reflecting distinct prioritization logics and strategic emphases. When α < 1, greater weights are assigned to higher-ranked factors, representing a pessimistic stance that amplifies perceived ecological risks. Conversely, α > 1 prioritizes lower-ranked factors, reflecting an optimistic stance associated with diminished landscape ecological risk projections. A neutral weighting scheme (α = 1) indicates no explicit preference. To identify priority areas for ecological restoration, five ecosystem services were incorporated as decision factors, and Ordered Weighted Averaging (OWA) operators were employed to simulate comprehensive ecosystem service assessments under varying risk tolerance levels in Dali Prefecture. To ensure the comprehensiveness of the risk assessment, we selected representative risk coefficients across different stages—specifically 0.00001, 0.1, 0.2, 0.5, 1, 2, 5, 10, and 1000—to simulate various scenarios, targeting pessimistic attitudes (α < 1), optimistic attitudes (α > 1), and neutral preferences (α = 1). This approach ensures that sufficient scenarios are available to support an analysis under different attitudinal perspectives. So, nine discrete risk coefficients were evaluated: α₀ (0.00001), 0.1, 0.2, 0.5, 1, 2, 5, 10, and α→∞ (α = 1000). The ordinal weights and corresponding risk coefficients for these scenarios are systematically presented in

Table 5. Decision risk coefficients and associated weighting parameters.

α		0.00001	0.1	0.2	0.5	1	2	5	10	1000
Scenario		Scenario 1	Scenario 2	Scenario 3	Scenario 4	Scenario 5	Scenario 6	Scenario 7	Scenario 8	Scenario 9
Ordinal Weights	ω1	1.0000	0.8823	0.7784	0.5345	0.1667	0.0816	0.0019	0.0000	0.0000
	ω2	0.0000	0.0551	0.1003	0.1892	0.1667	0.1927	0.0375	0.0016	0.0000
	ω3	0.0000	0.0295	0.0562	0.1214	0.1667	0.2358	0.1465	0.0330	0.0000
	ω4	0.0000	0.0178	0.0347	0.0807	0.1667	0.2245	0.2767	0.1795	0.0000
	ω5	0.0000	0.0104	0.0206	0.0501	0.1667	0.1723	0.3209	0.3999	0.0000
	ω6	0.0000	0.0049	0.0097	0.0241	0.1667	0.0930	0.2165	0.3861	1.0000
Trade-off Index		0	0.14	0.26	0.53	1	0.83	0.68	0.54	0
Risk Attitude		Optimistic	Mildly Optimistic	Moderately Optimistic	Strongly Optimistic	Neutral	Strongly Pessimistic	Moderately Pessimistic	Mildly Pessimistic	Pessimistic

:

The Trade-off Index reflects the weight distribution among decision factors, where higher values indicate more balanced weighting across variables. As shown in Figure 14, the Trade-off Index initially increases and then decreases with rising decision risk (α). At the extremes of α₀ (0.00001) and α→∞ (α = 1000), the Trade-off Index reaches 0, corresponding to strategies that prioritize a single ecosystem service under extreme optimism (Scenario 1) and extreme pessimism (Scenario 9), respectively. When α = 1, all decision factors receive equal weights, achieving the maximum Trade-off Index (neutral risk attitude, Scenario 5). Among the remaining six scenarios, the Trade-off Index values follow this descending order: Scenario 6 (0.84), Scenario 7 (0.69), Scenario 8 (0.54), Scenario 4 (0.53), Scenario 3 (0.26), and Scenario 2 (0.14). This pattern highlights the nonlinear relationship between risk tolerance and equity-efficiency trade-offs in ecosystem management, with critical thresholds at α = 1 (neutral balance) and extreme α values (single-service dominance).

(2)

Identification of Priority Areas for Ecological Restoration under Different Scenarios

This study processed standardized raster maps of five ecosystem services (2020 data) by ranking them according to their normalized mean values in descending order: ω₁ for nitrogen purification, ω₂ for phosphorus purification, ω₃ for habitat quality, ω₄ for climate regulation, ω₅ for water conservation, and ω₆ for soil retention. Sequence-dependent order weights were integrated into the Ordered Weighted Averaging (OWA) module of IDRISI 17.0. The simulation outcomes for each scenario were classified into five distinct categories using the natural breaks method in ArcGIS 10.2: low and moderately low (ecological restoration zones), moderate, moderately high (general ecological zones), and high (well-preserved ecological zones), as illustrated in Figure 15.

From Scenario 1 (α₀ = 0.00001) to Scenario 9 (α→∞ = 1000), the spatial distribution of ecological restoration zones exhibited a progressive east-to-center expansion within Dali Prefecture as the decision risk coefficient (α) increased, accompanied by enhanced spatial clustering of restoration priorities. Scenario 1, characterized by minimal restoration zones, and Scenario 9, marked by disproportionately extensive priority areas, both demonstrated significant deviations from realistic conditions. Scenarios 2 (α = 0.001) to 4 (α = 1) predominantly targeted restoration efforts in the upper Sangyuan River basin, the mid-reaches of the Zhonghe River, and the upstream regions of the Yuanjiang–Honghe River system. In contrast, Scenarios 5 (α = 2) to 8 (α = 10) prioritized the Erhai Lake watershed, mid-Miju River basin, upper Sangyuan River basin, Zhonghe River basin, and upper Juli River basin.

The simulations revealed a consistent spatial pattern: ecological restoration zones were predominantly concentrated in densely populated and highly accessible areas, with restoration extents displaying a notable outward expansion trend. This spatial dynamic highlights the critical interplay between anthropogenic activity intensity and ecosystem service optimization in restoration planning, emphasizing the need for adaptive strategies that reconcile ecological objectives with socioeconomic realities.

Scenario 1 has a relatively small ecological restoration area, while Scenario 9 features an excessively large ecology prioritized restoration area—both scenarios deviate significantly from reality. By comparing different trade-offs and considering the current status of Dali Prefecture alongside relevant policy documents, Scenarios 3, 5, and 8 are not only representative of different stages but also align more closely with existing conditions and policy plans. Through a comparative analysis of trade-off magnitudes (Levels 3/5/8) within multi-objective optimization frameworks and grounded in Dali Prefecture’s current socio-ecological conditions (with reference to Yunnan Province’s Territorial Spatial Planning Implementation Guidelines 2021–2035), three policy-driven scenarios were formalized: (1) the Ecological Priority Scenario (S3), reflecting decision-makers’ emphasis on ecosystem rehabilitation through enhanced environmental governance frameworks; (2) the Status Quo Scenario (S5), maintaining prevailing socioeconomic development strategies with adaptive policy refinements; and (3) the Development Priority Scenario (S8), prioritizing economic growth through intensified resource utilization, whereas systematically externalizing ecological costs. These scenario outcomes are spatially configured in Figure 16, where graduated color ramps explicitly visualize the spatial autocorrelation of ecological trade-offs (RGB: 34-139-34 for conservation priority zones; RGB: 255-215-0 for development hotspots).

Under the Ecological Priority Scenario (S3), areas with high ecological integrity constituted the largest proportion, where priority restoration zones covered 1838.94 km² (6.42% of the prefecture’s territory), exhibiting spatially fragmented distribution around the Erhai Lake Basin. The Status Quo Scenario (S5) demonstrated moderate ecosystem conditions overall, with critical rehabilitation areas expanding to 2856.82 km² (10.98% coverage), predominantly clustered in the eastern and downstream sections of the Erhai watershed. In contrast, the Development Priority Scenario (S8) revealed substantially degraded ecological baselines, requiring intervention across 8122.04 km² (28.37% territorial share), concentrated in central and eastern administrative divisions of Dali Prefecture.

It is important to note that the three policy scenarios designed in this study aim to simulate the spatial evolution of ecosystem services under different policy orientations, rather than providing a universally optimal restoration strategy. In practical applications, decision-makers should select among the nine OWA scenarios based on local governance capacity, budget availability, and ecological conservation objectives. For instance, regions with limited financial resources but strong ecological mandates may prefer the Ecological Priority Scenario (S3), which identifies relatively small-scale restoration zones and thus lowers restoration pressure. Conversely, areas with robust economic capacity and a strong focus on regional development may adopt the Development Priority Scenario (S8), which allows for broader restoration coverage while accommodating growth demands. The Status Quo Scenario (S5) offers a more balanced trade-off between ecological protection and development needs, making it suitable for regions with moderate governance and fiscal conditions.

5. Conclusions

This investigation systematically examines ecosystem service transitions in Dali Prefecture, Yunnan Province, through a sub-watershed scale analysis, utilizing Ordered Weighted Averaging (OWA) operators to model decision-making preferences for ecological restoration prioritization. Watershed delineation identified an optimal unit threshold of 11.04 km². Spatial patterns revealed progressive westward enhancement of ecosystem service capacity across sub-watersheds from 2015 to 2020. Low ecological risk zones (primarily Classes 1–3) concentrated in western and southern sectors, whereas high-risk areas encompassed 386 watershed units spanning 6351.62 km². Priority restoration efforts should focus on water purification services demonstrating extensive spatial degradation with severe impairment levels.

An integrative assessment of ecological quality and risk parameters categorized the landscape into four distinct typologies: high-quality, high-risk (HQ-HR); low-quality, high-risk (LQ-HR); low-quality, low-risk (LQ-LR); and high-quality, low-risk (HQ-LR) zones. The LQ-HR classification emerges as the critical target for immediate restoration interventions. Scenario simulations incorporating OWA operators revealed that restoration priority areas predominantly coincide with densely populated regions and areas of heightened socioeconomic accessibility, exhibiting marked spatial expansion tendencies. A comparative analysis of three policy scenarios (ecological priority, status quo maintenance, and development orientation) identified the Sangyuan River Basin, mid-reach Juli River, and upper Mujü River Basin as paramount restoration targets.

These geospatial simulations demonstrate substantial applicability for regional sustainable planning, though pragmatic operational adaptations remain imperative during implementation phases. The methodology aligns with IPBES conceptual frameworks while incorporating social–ecological system theory to interpret landscape openness dynamics, ensuring compliance with Landscape and Urban Planning’s scenario development standards. Future studies should consider the potential synergies and trade-offs among ecosystem services. For example, improving soil retention through vegetation may reduce water yield. These interactions highlight the need for integrated management strategies that account for service interdependencies when designing restoration plans.