Impacts of Cascade Hydropower Development on Aquatic Ecosystems in the Middle Jinsha River Basin: A DPSIR-Based Ecological Risk Assessment

Abstract

Cascade hydropower alters river hydrological regimes and threatens aquatic ecosystems, calling for robust ecological risk assessment (ERA). Conventional assessments often rigidly apply the full five-layer Driving Force–Pressure–State–Impact–Response framework, leading to indicator redundancy and unbalanced weighting. Single weighting methods also fail to reconcile expert judgment with data variability. To address these issues, we developed a three-layer (target–element–indicator) evaluation system embedding DPSIR logic without its full structure, focusing on hydrological regime, water environmental quality, and aquatic ecology with ten indicators. We used an improved group AHP-CRITIC coupling method for weighting: AHP aggregates expert judgments via geometric mean, and CRITIC integrates data variability and inter-indicator conflict. Multi-attribute utility theory normalized indicators into a unified security index, applied to four cascade stations in the middle Jinsha River using 66-year (1953–2018) hydrological and seven-year (2013–2019) in situ monitoring data. The evaluation obtained a comprehensive index of 0.71 to 0.74, which is generally safe. River connectivity loss was the primary limiting factor. Hydrological alteration was mild overall with a value of 0.139, while extreme flow decline rate variation reached a high level of 0.83. Weekly regulated stations achieved over 97% ecological flow guarantee, which is much higher than daily regulated stations. This streamlined framework improves interpretability for cascade basins and supports sustainable watershed management.

1. Introduction

Water resources are the core element sustaining human survival, development and ecosystem stability, and the fundamental guarantee for high-quality regional socioeconomic development [1]. With accelerating global population growth and industrialization, per capita available freshwater resources have declined continuously, making the water ecological security crisis caused by water scarcity, pollution and habitat degradation a global governance priority [2]. China’s per capita water resources are only 28% of the global average, and the contradiction between uneven spatiotemporal water distribution and intensifying human activities has become increasingly prominent, leaving the ERA of key basins in a severe situation [3]. As the main stem of the upper Yangtze River, the Jinsha River serves as a critical water conservation area and ecological barrier for the entire Yangtze River Basin, supporting exceptional freshwater biodiversity and acting as the core habitat for 15 endemic fish species such as Schizothorax wangchiachii and Coreius guichenoti [4,5]. Its middle reach, spanning Sichuan and Yunnan provinces, undertakes essential functions including regional water supply, hydropower generation and irrigation, and plays a pivotal regulatory role in the water resource balance and ecological security of the middle and lower Yangtze River [6].

However, the past three decades have witnessed the most intensive cascade hydropower development in China in this reach, with four large dams constructed sequentially. These projects have fundamentally transformed the river’s physical structure: longitudinal connectivity has been completely disrupted, converting the free-flowing river into a chain of disconnected lentic reservoir reaches [7]. As documented by Grill et al. [8], the Jinsha River Basin is among the most fragmented river systems in Asia, with less than 10% of its main stem remaining unobstructed. Such structural alterations have triggered irreversible changes in river regimes: the natural seasonal flow pattern has been flattened, extreme hydrological events attenuated, and daily flow fluctuations amplified by peaking power generation—directly violating the five core components of the natural flow regime paradigm (magnitude, frequency, duration, timing, rate of change) essential for aquatic ecosystem integrity [9,10]. Superimposed on this dominant pressure are co-occurring anthropogenic stressors including rapid urbanization, intensive mineral extraction and agricultural expansion, which have increased impervious surface area by 47% and elevated total nitrogen and phosphorus loads by 32% and 28% respectively since 2000 [11,12]. The cumulative effects of dam fragmentation and diffuse human disturbances have manifested as an insufficient ecological flow guarantee, degraded water quality and disrupted aquatic food webs, becoming a major bottleneck restricting the basin’s sustainable development and the implementation of the Yangtze River Economic Belt Ecological Protection Strategy.

Ecological risk assessment (ERA) is the core means to identify ecological risks and formulate targeted protection strategies, with its key lying in constructing a scientific evaluation framework and indicator system [13]. Over the past four decades, global scholars have developed four major categories of assessment methods, each with distinct advantages and limitations [14]. First, indicator system frameworks such as PSR and DPSIR offer clear causal logic but are prone to structural redundancy. Second, process-based simulation models such as SWAT and EFDC provide high mechanistic accuracy but require massive high-precision data. Third, comprehensive integration methods including TOPSIS and fuzzy evaluation are suitable for multi-attribute decision-making yet are limited by weighting bias. Fourth, machine learning approaches feature strong nonlinear fitting ability but suffer from poor interpretability. Among these, the DPSIR framework is the most widely adopted in river basin assessments due to its complete causal chain linking socioeconomic drivers to ecological responses [5,15]. However, traditional cascade system assessments strictly follow its five-layer structure, leading to indicator redundancy and uneven weight distribution. Single weighting methods, both subjective ones like AHP and Delphi and objective ones like entropy and CRITIC, also fail to fully reconcile expert judgment with objective data variability, potentially weakening assessment robustness [16]. To address these limitations, this study develops a customized three-layer evaluation system consisting of target, element, and indicator, which embeds the core causal logic of DPSIR without replicating its full structure. Focusing on three core elements, namely hydrological regime, water quality, and aquatic ecosystem, with ten quantifiable indicators, we adopt an improved group AHP-CRITIC coupling method for weighting. In this method, AHP aggregates multi-expert judgments via geometric mean to reduce individual bias, while CRITIC integrates data variability and inter-indicator conflicts. Multi-attribute utility theory is then applied to normalize heterogeneous indicators into a unified security index.

At present, research on the Jinsha River Basin’s ecological security has mainly focused on single aspects such as water resource utilization and pollution source control, lacking a systematic evaluation of the middle reach that considers the coupling of cascade hydropower development and other human activities. Furthermore, most existing studies use single weighting methods, which may lead to biased evaluation results and fail to accurately reflect the relative importance of different ecological factors in cascade hydropower basins [16]. In view of this, this paper applies the above customized evaluation system to systematically assess the current status and spatiotemporal evolution of ERA in the middle Jinsha River Basin, and identifies key influencing factors and main obstacles. The results are expected to provide scientific support for ecological regulation and sustainable management of cascade hydropower development in the basin, and offer a methodological reference for similar complex river basins worldwide.

2. Materials and Methods

2.1. Overview of the Study Area

The Jinsha River Basin is located at 99°30′ E–102°30′ E and 26°30′ N–29°30′ N, flowing through 15 districts and counties in Yunnan and Sichuan Provinces with a basin area of about 105,000 km². The region is located in the transition zone from the eastern edge of the Qinghai–Tibet Plateau to the Yunnan–Guizhou Plateau, with severe topographic relief and an altitude of 1000–4500 m. It has a subtropical plateau monsoon climate with an average annual temperature of 12–18 °C and an average annual precipitation of 600–1200 mm, concentrated in June–September.

The river system in the basin is well developed, with the main stream having a total length of about 1050 km. The main tributaries include the Yalong River, Longchuan River and Yupao River. The forest coverage rate is about 45%, and the main soil types are red soil and yellow soil. The population in the basin is about 3.132 million, with agriculture and industry as the leading industries, and cascade hydropower development is the core economic pillar. The water ecological protection objectives are defined as follows: the main stream in the Yunnan section is subject to Class II water quality in accordance with Chinese Environmental Quality Standards for Surface Water (GB 3838-2002) [17], while the Panzhihua section implements Class III standard. A total of 15 endemic fish species, including Schizothorax wangchiachii, Coreius guichenoti and Percocypris pingi, are listed as key protected species.

The study area covers four cascade hydropower stations distributed sequentially from upstream to downstream along the main stem of the middle Jinsha River (Figure 1). The uppermost site is LY Hydropower Station (weekly regulation), followed by AH Hydropower Station (daily regulation, with its dam located 5 km downstream of the Cuiyi River confluence), JAQ Hydropower Station (weekly regulation, situated in Lijiang City), and LDL Hydropower Station (weekly regulation) as the lowermost site included in this study.

2.2. Data

Daily hydrological records from 1953 to 2018 were obtained from Shigu and Panzhihua hydrological stations, representing the natural regime and the post-impact regime respectively. All data were preprocessed by outlier elimination, linear interpolation for missing values, and consistency checks to ensure reliability.

Systematic collection was performed for multi-source data of LY, AH, JAQ, and LDL hydropower stations, including aquatic habitat and ecological monitoring results, environmental impact assessment documents and approvals, environmental supervision reports, completion acceptance reports and approvals, dispatching and operation data, reservoir area socioeconomic data, river terrain data, and meteorological data. These data were supplemented by in situ water environment and ecological field surveys. Rationality and reliability tests were conducted to support robust assessment and diagnosis of water environmental protection and ecological security in the middle Jinsha River Basin.

Field sampling was implemented at 9 cross-sections along the main stem and tributaries, covering the reservoir tail, the middle reservoir, the dam front, the downstream reach, and tributary outlets across the study area. For phytoplankton and small zooplankton such as protozoans and rotifers, qualitative samples were collected using a 25# plankton net (mesh size 64 μm), and quantitative samples were collected with a 2.5 L water sampler; 2 L of water was fixed with Lugol’s solution. For macrozooplankton including cladocerans and copepods, samples were collected using a 13# plankton net (mesh size 112 μm) and concentrated from 10 L water samples. Macrobenthos were sampled using a Petersen grab or D-frame net, with a cumulative sampling area of 0.5–1 m² per site. Water samples for nutrients were collected from surface, middle, and bottom layers, stored in clean containers, and analyzed in the laboratory within 24 h. Water temperature, pH, dissolved oxygen, and electrical conductivity were measured in situ at a consistent time each day.

The complete dataset includes: (1) 66 years (1953–2018) of daily hydrological data from Shigu and Panzhihua stations; (2) 7 years (2013–2019) of in situ water quality and aquatic ecological monitoring data from LY, AH, JAQ, and LDL stations; and (3) statistical data on hydropower station dispatching, ecological discharge, and fish propagation and release. The dataset fully characterizes the hydrological regime, water environment, and aquatic biodiversity in the cascade development zone of the middle Jinsha River.

Instrument and Software Information: All plankton sampling nets (25# and 13#), 2.5 L water sampler, Petersen grab, and D-frame net for field sampling were supplied by Fushun Keruisi Instrument Co., Ltd., Fushun, Liaoning, China. The portable ProDSS multiparameter water quality sonde, capable of in-situ measurements of water temperature, pH and dissolved oxygen, was manufactured by YSI Inc. (Xylem Analytics, Yellow Springs, OH, USA). Microsoft Excel 2019 (Microsoft Corporation, Redmond, WA, USA) was used for data collation and statistical analysis; Origin 2021 (OriginLab Corporation, Northampton, MA, USA) was adopted for figure plotting.

2.3. Methods

2.3.1. Development of an Evaluation Index System

As a classic framework for environmental system assessment, the DPSIR model systematically analyzes the interaction mechanism between human activities and the ecological environment through the causal chain of “Driving Force–Pressure–State–Impact–Response”. When the selected indicators are comprehensive, the indicators can be better classified, and the comprehensive effects of various elements on the system can be considered [18,19]. Based on this, this study determined the core framework logic of the evaluation system: taking the “comprehensive index of basin water ecological security” as the target layer, focusing on three core element layers including hydrological regime, water environmental quality and aquatic ecosystem, and screening the indicator layer characterizing the key characteristics of each element, forming an evaluation system structure with clear levels and definite causal correlation (Figure 2).

Combined with the actual characteristics of the study basin, on the basis of analyzing key influencing factors such as land resources, water resources, human interference and environmental pollution, through literature review, expert consultation and data availability verification, an evaluation system including 3 element layers and 10 indicator layers was finally determined (

Table 1. Evaluation indicators for watershed ecological security.

Target Layer	Element Layer	Indicator Layer	Indicator Attribute
Comprehensive index of ecological security	Hydrological regime data	Change rate of Indicators of Hydrologic Alteration (IHA)	Quantitative assessment of the change degree of river hydrological regime
		Monthly average flow change rate	Guarantee status for maintaining the non-degradation of river ecosystem
		Ecological flow guarantee degree	Guarantee status for maintaining the non-degradation of river ecosystem
		River connectivity	An important indicator for evaluating the integrity of river ecosystem
	Water environment quality	$\mathrm{Permanganate} \mathrm{index} ({\mathrm{C}\mathrm{O}\mathrm{D}}_{\mathrm{M}\mathrm{n}}$) concentration	An important indicator reflecting water quality
		Ammonia nitrogen (NH3-N) concentration	An important indicator for evaluating water quality
		TN concentration	A key indicator for evaluating water eutrophication degree and water quality
		TP concentration	An important indicator for evaluating water eutrophication degree and water quality
	Water ecological security	Trophic state	An important indicator reflecting eutrophication and algae biomass
		Number of important fish species	An important indicator for evaluating aquatic ecology

).

The indicators in Table 1 were quantified according to the following formulas and specifications.

For an individual IHA indicator, its degree of alteration is calculated as the rate of change in the IHA indicator, quantified using the RVA method. The formula for the degree of alteration of a single hydrological indicator is as follows:

$$D_i={\displaystyle \frac{N_i-N_e}{N_e}}\times 100\%$$

(1)

$$N_e=0.5\times N_T$$

(2)

where $D_i$ is the degree of hydrological alteration of the i-th IHA indicator; $N_i$ is the number of years in which the post-alteration IHA value falls within the 25th–75th percentile target range; $N_e$ is the corresponding expected number of years; and $N_T$ is the total number of years after alteration.

The formula for the overall hydrological alteration degree is:

$$D_0=\sqrt{{\displaystyle \frac{1}{32}}{\sum }_{i=1}^{32}{D}_i^2}$$

(3)

where $D_0$ is the comprehensive overall hydrological alteration degree of the river basin; and 32 is the total number of IHA indicators.

The formula for the monthly average flow change rate is:

$$∆Q={\displaystyle \frac{Q_{out}-Q_{in}}{Q_{in}}}\times 100\%$$

(4)

where $∆Q$ is the monthly mean flow change rate (%); $Q_{out}$ is the average outflow of the reservoir for that month (m³/s); and $Q_{in}$ is the average inflow of the reservoir for that month (m³/s).

The formula for the ecological flow guarantee degree is:

$$E={\displaystyle \frac{T_s}{T_t}}\times 100\%$$

(5)

where E is the ecological flow satisfaction degree; $T_s$ is the number of days during which the ecological flow requirement is met; and $T_t$ is the total number of monitoring days.

The score of river connectivity was assigned as 0. This is because the study reach is intensively developed by cascade hydropower projects. Dams and reservoirs block the longitudinal hydrological connectivity of the river channel, interrupting fish migration, material transport and water exchange. The longitudinal connectivity of the river is almost lost, resulting in a standardized value of 0 for this indicator.

The ${\mathrm{C}\mathrm{O}\mathrm{D}}_{\mathrm{M}\mathrm{n}}$, NH₃-N, TN, and TP concentrations were scored based on the measured concentrations at each monitoring section, following a linear normalization procedure in accordance with the Chinese Environmental Quality Standards for Surface Water (GB 3838-2002). All these indicators are negative indicators, meaning that lower pollutant concentrations correspond to better water quality and thus higher evaluation scores. The Class I water quality standard was set as the optimal benchmark (scoring 1 point), and the Class V standard as the worst benchmark (scoring 0 point). The measured concentrations at each section were converted into standardized scores within the 0–1 interval via linear interpolation, thereby characterizing the water quality level of each section.

The trophic state of the water body was calculated using the comprehensive trophic level index (TLI) method:

$$TLI={\displaystyle \sum _{j=1}^m}{w}_j\times {TLI}_j$$

(6)

where TLI is the comprehensive trophic state index of the reservoir; $w_j$ is the weight of the j-th water quality indicator; ${TLI}_j$ is the individual trophic state index of the j-th indicator; and m is the total number of indicators involved in the calculation. In this study, five indicators are selected: BOD₅, COD_cr, TP, TN, and chl.a. The trophic state is classified as oligotrophic when TLI < 30, and as mesotrophic when 30 ≤ TLI ≤ 50. The water bodies of the LY, AH, JAQ, and LDL reservoir areas are in an oligotrophic to mesotrophic state, and no eutrophication phenomenon has occurred.

Reservoirs of LY, AH, JAQ and LDL have implemented effective aquatic organism conservation measures, including fish stock enhancement and release and tributary habitat protection. The released fish can basically adapt to the ecological environment of the basin, with their recapture rate increasing year by year. No reduction has been found in the number of confirmed protected fish species. Therefore, the score for the indicator of the number of important fish species is uniformly assigned as 1.

An indicator system that highlights the particularity of the basin water environment was constructed based on the DPSIR model. The improved group Analytic Hierarchy Process (GAHP) and the Criteria Importance Through Intercriteria Correlation (CRITIC) objective weighting method were combined to determine the indicator weights, and the indicator system and safety level of the comprehensive index suitable for the comprehensive assessment of river basin ecological security were determined. Different indicators have different degrees of influence and contribution to various ecological security factors. Finally, the expert scoring method was used to determine the safety grade of each single factor, and the values were assigned according to 5 grades of safe, relatively safe, less safe, unsafe and extremely unsafe: 0.9, 0.7, 0.5, 0.3 and 0.1, respectively.

2.3.2. Determination of Indicator Weights

To scientifically determine the weights of each evaluation indicator, this study adopts a strategy that combines an improved GAHP with the CRITIC objective weighting method. The rationale for using combined weighting is that a single subjective weighting method (e.g., Analytic Hierarchy Process (AHP)), while capable of incorporating expert experience and domain knowledge, is susceptible to personal bias. Conversely, a single objective weighting method (e.g., entropy weight method or CRITIC), although based on data variability, may deviate from actual ecological significance. Therefore, coupling the two approaches allows retaining the advantages of expert judgment while using the data’s own contrast intensity and conflict for correction, thereby enhancing the reliability of the weights and the scientific validity of the evaluation results [20,21].

• Improved Group AHP Method

The AHP, proposed by Saaty in the mid-to-late 1970s [22], decomposes complex problems into a hierarchical structure and employs quantitative analysis, thereby making the decision-making process more systematic and significantly improving its validity and reliability [23].

Traditional AHP typically relies on the judgment of a single expert, which carries the risk of subjective bias. To overcome this limitation, this study adopts an improved group AHP approach [24]. The specific improvements include: ① inviting five experts in the fields of watershed ecology, water resource management, and environmental engineering to independently construct judgment matrices; ② conducting consistency checks on each expert’s judgment matrix (CR < 0.1) and discarding those that fail to meet the consistency requirement; ③ aggregating the elements of the validated judgment matrices using the geometric mean method to form a composite judgment matrix—the geometric mean was chosen over the arithmetic mean because it better preserves the scale consistency of matrix elements and is less sensitive to extreme values. The arithmetic mean is easily influenced by an individual expert’s extreme scale values, disrupting the inherent structure of the ratio scale; in contrast, the geometric mean, by performing an equally weighted average in logarithmic space, effectively suppresses the bias of extreme opinions, thereby reducing the interference of individual expert subjectivity on the final weights [25]; and ④ calculating the weights based on the composite matrix. This approach effectively mitigates the adverse effects of extreme opinions and individual subjective bias. The formula is:

$$a_{ij}=({\prod }_{k=1}^5{a}_{ij}^{\left(k\right)}{)}^{{\displaystyle \frac{1}{5}}}$$

(7)

where $a_{ij}^{\left(k\right)}$ denotes the element in the i-th row and j-th column of the judgment matrix constructed by the k-th expert.

2.

CRITIC Method

The CRITIC method was proposed by Diakoulaki [26]. Its core principle is to account for both the internal variability of each indicator and the interrelationships among indicators when determining indicator weights. This method comprehensively measures the objective weights of indicators based on the contrast intensity and conflict among evaluation indicators. Contrast intensity is reflected by the internal variability of each indicator, expressed as the standard deviation. A larger standard deviation indicates that the indicator carries more information and thus should be assigned a higher weight. Conflict reflects the correlation among influencing factors, expressed using correlation coefficients. The larger the correlation coefficient, the lower the weight [27]. The specific formulas are as follows:

$$C_j={\sigma }_j{\displaystyle \sum _{j=1}^{ m}}(1-{\tau }_{\mathit{ij}}) (j = 1, 2\dots , m)$$

(8)

$${w_j=C}_j/{\displaystyle \sum _{j=1}^{ m}}{(C}_j) (j=1, 2\dots , m)$$

(9)

where $C_j$ represents the total amount of information conveyed by the j-th indicator; ${\sigma }_j$ is the standard deviation of the j-th indicator; ${\tau }_{\mathit{ij}}$ is the correlation coefficient between the i-th and j-th indicators; $w_j$ is the weight of the j-th indicator.

3.

Determination of Comprehensive Weights

Based on the principle of minimizing the total deviation, the combined weight is derived by coupling the AHP and CRITIC methods [28]. The calculation formula is as follows:

$$W_j={\displaystyle \frac{\sqrt{W_{AHP,j}\times W_{CRITIC,j}}}{{\displaystyle {\sum }_{j=1}^n}\sqrt{W_{AHP,j}\times W_{CRITIC,j}}}}$$

(10)

where $W_j$ denotes the final combined weight of the j-th indicator; $W_{AHP,j}$ denotes the subjective AHP weight of the j-th indicator; $W_{CRITIC,j}$ denotes the objective CRITIC weight of the j-th indicator; and n denotes the total number of evaluation indicators.

Compared with the arithmetic mean or linear weighting, the geometric mean effectively balances the magnitude differences between subjective and objective weights, preventing one type of weight from dominating excessively. Moreover, the geometric mean possesses multiplicative symmetry; when a particular weight is extremely small, the combined weight is correspondingly reduced, thereby suppressing biases caused by extreme values [29]. Furthermore, the geometric mean exhibits superior order preservation and stability when combining multi-source information, and its advantages have been confirmed by several ERA studies [30,31].

2.3.3. Comprehensive Evaluation Method

• Evaluation method

The multi-attribute utility theory was used for the comprehensive evaluation. The multi-attribute utility theory applies a mathematical model to convert the results of all indicators under different standards into a total utility value between [0, 1]. Its core formula is:

$$\mathrm{U}={\displaystyle \sum _1^{\mathrm{n}}}{\mathrm{w}}_{\mathrm{i}}{\mathrm{u}}_{\mathrm{i}}$$

(11)

In the formula, U is the total utility value of the evaluation object; w_i is the weight of the i-th (i = 1, 2, …, n) indicator; and u_i is the utility value of the i-th indicator, ranging from 0 to 1.

2.

Evaluation criteria

According to the above calculation formula, the utility values of each indicator in the evaluation index system of the protection effect of tributary habitat replacement were assigned in the range of [0, 1], and the comprehensive evaluation index of each layer was obtained. The value range of the comprehensive evaluation index was divided into different grade standards (

Table 2. Evaluation criteria of comprehensive evaluation index.)

Total Utility Value	1	[0.75, 1)	[0.6, 0.75)	[0.1, 0.6)	[0, 0.1)
Evaluation result	Safe	Relatively safe	Generally safe	Unsafe	Extremely unsafe

).

The ecological security status obtained from the single factor analysis only reflects the effect of a certain factor and fails to reflect the multi-faceted basin ecological security problems. Therefore, it is necessary to reflect each factor in the form of numerical values, and determine the ecological security grade by grading through calculating the ecological security index. The formula of the ecological security index is:

$${\mathrm{E}\mathrm{S}\mathrm{I}}_{\mathrm{j}}=\mathrm{n}\sqrt{{\displaystyle \prod _{\mathrm{j}=1}^{\mathrm{n}}}{\mathrm{C}}_{\mathrm{i}\mathrm{j}}}$$

(12)

In the formula, ESI_j is the ecological security index of the j-th ecological security problem; C_ij is the ecological security grade value of the i-th factor in the j-th ecological security problem; and n is the number of factors.

Since different ecological problems are independent of each other, to reflect the overall ecological security status of the basin, it is necessary to conduct a comprehensive evaluation of multiple ecological problems. The minimum value method was adopted to obtain the comprehensive ecological security index (ESSI) of each hydropower station in space: $\mathrm{E}\mathrm{S}\mathrm{S}\mathrm{I}=\mathrm{m}\mathrm{i}\mathrm{n}\mathrm{E}\mathrm{S}{\mathrm{I}}_{\mathrm{j}}$.

3. Results and Analysis

3.1. Variation in Hydrological Regime

3.1.1. Change Rate of IHA Indicators

The IHA indicator system initially consisted of 32 hydrological parameters and was later expanded to 33 indicators with the Range of Variability Approach (RVA), forming the integrated IHA-RVA assessment method [10]. To address the limitations of the IHA method in overall status assessment, Richter et al. [10] proposed the RVA in 1997. This method establishes a threshold system for hydrological indicator variability and introduces supplementary assessment parameters, thereby elevating the analysis from discrete hydrological indicators to a comprehensive status assessment. The method sets the 25th percentile and the 75th percentile as the lower and upper bounds of the RVA target thresholds, constructing a two-layer evaluation model consisting of single-indicator and overall hydrological alteration. For an individual IHA indicator, its degree of alteration is calculated as the rate of change in the IHA indicator, quantified using the RVA method. To comprehensively evaluate the overall hydrological regime alteration of a river, Shiau et al. [32] proposed the overall hydrological alteration degree, which integrates the IHA indicators using a weighted average method. The IHA indicators are divided into five categories: monthly average flow, annual extreme flow, occurrence time of extreme flow, frequency and duration of high and low flow, and flow change rate and frequency. The ecological significance of various hydrological indicators is shown in

Table 3. Variation Indicators of Hydrological Regime.

Hydrological Regime	Monthly average flow	Meeting the habitat needs of aquatic organisms, the needs of plants for soil moisture content, the water needs of terrestrial organisms with high reliability, the migration needs of carnivores, and the impacts on water temperature and dissolved oxygen
	Annual extreme flow	Meeting the needs of vegetation expansion, construction of river geomorphology and natural habitats, nutrient exchange between rivers and flood detention areas, and distribution of plant communities in lakes, ponds and flood detention areas
	Occurrence time of annual extreme flow	Meeting the needs of fish migration and spawning, cyclic reproduction of living organisms, habitat conditions during the biological reproduction period, and species evolution
	Frequency and duration of high and low flow	Generating the frequency and magnitude of soil moisture required by vegetation, meeting the support of flood detention areas for aquatic organisms, sediment transport, river channel structure, bottom disturbance, etc.

.

The RVA method was used to calculate the change degree of each IHA hydrological indicator, and the change degree of various eco-hydrological indicators calculated by the set weight coefficient was used as an indicator to evaluate the water system environment and ecological security in the middle reaches of the Jinsha River.

The correlation coefficients, RVA thresholds and hydrological change degrees of each IHA indicator of Panzhihua Hydrological Station from 2006 to 2018 were calculated (

Table 4. Correlation Coefficients, RVA Thresholds and Hydrological Alteration Degrees of IHA Indicators at Panzhihua Hydrological Station.

IHA Indicator Category		Correlation Coefficient Between Shigu and Panzhihua	RVA Threshold			Variation Degree
Monthly average flow	January average flow	0.961593	626.8	497	687	0.08
	February average flow	0.951701	566	465	644	0.04
	March average flow	0.940085	558.5	492	635	0.00
	April average flow	0.977992	648.3	662	1002	−0.19
	May average flow	0.871575	1032	895	1683	−0.19
	June average flow	0.893575	1902	1686	3079	−0.18
	July average flow	0.993853	3783	2283	5956	−0.08
	August average flow	0.992646	3933	2030	5762	0.01
	September average flow	0.974825	3869	2436	5741	−0.01
	October average flow	0.87326	2328	1588	3333	0.01
	November average flow	0.861473	1295	918	1372	0.09
	December average flow	0.937703	801.7	608	1027	0.08
Annual extreme flow	1-day minimum flow	0.922306	460.3	451	556	−0.10
	3-day minimum flow	0.940193	467.7	454	556	−0.10
	7-day minimum flow	0.958408	476.9	456	558	−0.08
	30-day minimum flow	0.954279	507.6	465	605	−0.05
	90-day minimum flow	0.938587	559.5	487	650	0.00
	1-day maximum flow	0.995237	6518	3864	9016	0.00
	3-day maximum flow	0.994388	6301	3784	8768	−0.01
	7-day maximum flow	0.997383	5851	3641	8163	−0.01
	30-day maximum flow	0.972984	4843	3184	6198	−0.02
	90-day maximum flow	0.963377	3928	2657	5379	−0.03
Occurrence time of annual extreme flow	Occurrence time of annual minimum flow/d/d	0.503466	34	32	61	0.42
	Occurrence time of annual maximum flow/d/d	0.789281	223	189	244	0.03
Frequency and duration of high and low flow	Number of low pulses/times	1	0	0	0	0
	Duration of low pulses/d	1	0	0	0	0
	Number of high pulses/times	−0.0625	4	2	5	0.35
	Duration of high pulses/d	0.494567	23	9	47	0.52
Flow change rate and frequency	Rise rate/(m3·s−1·d−1)	0.915828	150.2	68	116	0.50
	Fall rate/(m3·s−1·d−1)	0.897304	−132.3	−94	−50	0.83
	Number of reversals/times	0.107742	93	79	107	0.39

).

The weights and variation degrees of IHA indicators were determined based on the CRITIC method (

Table 5. Weight and Variation Degree of IHA Indicators.

IHA Indicator		Weight	Variation Degree	Overall Variation Degree	Comprehensive Variation Degree
Monthly average flow	January average flow	0.086	0.08	−0.035	0.139
	February average flow	0.091	0.04
	March average flow	0.083	0.00
	April average flow	0.094	−0.19
	May average flow	0.082	−0.19
	June average flow	0.098	−0.18
	July average flow	0.084	−0.08
	August average flow	0.084	0.01
	September average flow	0.094	−0.01
	October average flow	0.071	0.01
	November average flow	0.068	0.09
	December average flow	0.072	0.08
Annual extreme flow	1-day minimum flow	0.092	−0.10	−0.039
	3-day minimum flow	0.092	−0.10
	7-day minimum flow	0.095	−0.08
	30-day minimum flow	0.116	−0.05
	90-day minimum flow	0.118	0.00
	1-day maximum flow	0.101	0.00
	3-day maximum flow	0.097	−0.01
	7-day maximum flow	0.099	−0.01
	30-day maximum flow	0.097	−0.02
	90-day maximum flow	0.097	−0.03
Occurrence time of annual extreme flow	Occurrence time of annual minimum flow/d/d	0.5	0.42	0.225
	Occurrence time of annual maximum flow/d/d	0.5	0.03
Frequency and duration of high and low flow	Number of low pulses/times	0.5	0	0.442
	Duration of low pulses/d	0.5	0
	Number of high pulses/times	0.46	0.35
	Duration of high pulses/d	0.54	0.52
Flow change rate and frequency	Rise rate/(m3·s−1·d−1)	0.39	0.50	0.63
	Fall rate/(m3·s−1·d−1)	0.36	0.83
	Number of reversals/times	0.35	0.39

).

Among the four built hydropower stations in the middle reaches of the Jinsha River, LY, JAQ, LDL and GYY have weekly regulation performance, and AH has daily regulation performance. Therefore, the variation in monthly average inflow and outflow of the four hydropower stations is very small. The variation rates of monthly average inflow and outflow of four typical hydropower stations are shown in Figure 3.

3.1.2. Ecological Flow Guarantee Degree

According to the approval of the Yangtze River Commission, the minimum downstream ecological flow standards for each hydropower station are as follows: LY 300 m³/s, AH 350 m³/s, JAQ 350 m³/s, LDL 400 m³/s and GYY 350 m³/s.

The satisfaction status of each hydropower station from 2013 to 2019 was as follows: for LY Hydropower Station, the total monitoring days were 2555, with 53 days not meeting the standard, and the guarantee rate was 97.93%, concentrated in December to April of the next year, all caused by dispatching; for AH Hydropower Station, 362 days did not meet the standard under the EIA standard of 510 m³/s, with a guarantee rate of 85.83%, and 88 days did not meet the standard after adjusting the standard to 350 m³/s, with a guarantee rate of 96.6%, 60% of which was due to insufficient upstream inflow; for JAQ Hydropower Station, the monitoring days from 2017 to 2019 were 1095, with 28 days not meeting the standard, and the guarantee rate was 97.44%, concentrated in January to April; for LDL Hydropower Station, the total monitoring days were 2555, with 170 days not meeting the standard, and the guarantee rate was 93.35%, concentrated in November to June of the next year, 58.8% of which was due to insufficient inflow. The satisfaction status of downstream ecological flow of each hydropower station is shown in Figure 4.

3.2. Changes in Water Environmental Quality

Before the construction of each hydropower station, most of the water environment indicators of the main stream water body met the Class II surface water environmental quality standard required by the water environment functional zoning, and a few indicators such as COD_Cr, BOD₅, sulfide, mercury and lead exceeded the standard. After the completion of each hydropower station in the middle reaches of the Jinsha River, rainfall runoff in the flood season carries non-point-source pollutants along the bank into the reservoir area, which has a certain impact on the water quality of the reservoir area. Some indicators such as COD, fecal coliform, TP and TN exceeded the standard periodically, but the overall water quality in the year still maintained the Class II standard. The water bodies in the reservoir areas of LY, AH, JAQ and LDL were in an oligotrophic to mesotrophic state, and no water eutrophication phenomenon occurred.

3.3. Changes in Aquatic Organisms

LY, AH and other reservoirs have adopted good aquatic organism protection measures. Through artificial propagation and release and tributary habitat protection, the released fish can basically adapt to the ecological environment of the basin, and the recapture rate has increased year by year. No decrease has been found in the number of identified protected fish species.

The number of phytoplankton species ranged from 56 to 92, with Bacillariophyta accounting for the highest proportion, the LY reach having the most species and the AH reach the fewest (Figure 5a); the number of zooplankton species ranged from 15 to 69, with Protozoa accounting for the highest proportion, the LY reach having the most species (Figure 5b); the number of benthos species ranged from four to 18, with Arthropoda accounting for the highest proportion, the LY reach having the most species (Figure 5c); the number of fish resource species ranged from 21 to 46, with Cypriniformes accounting for the highest proportion, the JAQ-LDL reach having the most species (Figure 5d). The above results are summarized in Figure 5.

3.4. Assessment of Basin Water Ecological Security

The single-indicator scores of each hydropower station showed that the score of IHA indicator variation degree of the hydrological regime was 0.90 for all stations. According to the adopted scoring criteria, this score corresponds to a mild hydrological alteration level, indicating a high similarity between the post-impoundment and natural hydrological regimes, with only minor deviations from the natural flow patterns. The scores of COD_Mn concentration and ammonia nitrogen concentration in water environmental quality were both 0.90, and the score of the number of important fish species was as high as 1.00; the score of river connectivity was 0 for all hydropower stations, which was a common limiting factor; the score of the ecological flow guarantee degree of AH Hydropower Station was the lowest at 0.80 (

Table 6. Indicator scores of each hydropower station in the middle reaches of the Jinsha River.

Indicator Layer	LY	AH	JAQ	LDL
IHA indicator variation degree	0.90	0.90	0.90	0.90
Ecological flow guarantee degree	0.95	0.80	0.95	0.95
River connectivity	0	0	0	0
Chemical oxygen demand concentration	0.90	0.90	0.90	0.90
Ammonia nitrogen concentration	0.90	0.90	0.90	0.90
Total nitrogen concentration	0.80	0.90	0.90	0.85
Total phosphorus concentration	0.80	0.85	0.85	0.90
Trophic state	0.90	0.90	0.90	0.90
Number of important fish species	1.00	1.00	1.00	1.00

).

Combined with the calculation of combined weights (

Table 7. Weighted scores of each indicator for the four hydropower stations.

Indicator Layer	Weight	LY	AH	JAQ	LDL
IHA indicator variation degree	0.09	0.08	0.08	0.08	0.08
Ecological flow guarantee degree	0.17	0.16	0.13	0.16	0.16
River connectivity	0.22	0.00	0.00	0.00	0.00
Chemical oxygen demand concentration	0.07	0.06	0.06	0.06	0.06
Ammonia nitrogen concentration	0.03	0.03	0.03	0.03	0.03
Total nitrogen concentration	0.03	0.03	0.03	0.03	0.03
Total phosphorus concentration	0.04	0.03	0.03	0.03	0.03
Trophic state	0.07	0.06	0.06	0.06	0.06
Number of important fish species	0.28	0.28	0.28	0.28	0.28

) and the comprehensive evaluation results (Figure 6), the comprehensive ecological security score of the basin ranged from 0.71 to 0.74, indicating an overall “generally safe” level. Among them, JAQ and LDL Hydropower Stations had the highest score of 0.74, and AH Hydropower Station had the lowest score of 0.71. The driving mechanism of the safety grade was as follows: the loss of river connectivity and insufficient ecological flow dispatching caused by cascade development constituted the core pressure; mild hydrological variation, up-to-standard water quality and stable biodiversity constituted the main state; measures such as ecological flow regulation and artificial propagation and release partially offset the pressure impact and maintained a generally safe level.

4. Discussion

4.1. Ecological Mechanism of Cascade Hydropower Impacts

The comprehensive ecological security index of the middle Jinsha River basin ranges from 0.71 to 0.74, indicating an overall “generally safe” level. This result reflects the dual effects of large-scale cascade hydropower development and active ecological protection measures. The core driving mechanism is that the pressure from river connectivity loss and hydrological regime alteration caused by dam construction is partially offset by effective response measures, such as ecological flow regulation and artificial propagation and release. As a result, the basic stability of the ecosystem is maintained.

River connectivity loss is identified as the primary limiting factor, with a score of 0 for all hydropower stations. This finding is consistent with the global pattern of cascade hydropower development. The four large dams in the middle reaches have completely blocked the longitudinal connectivity of the main stream, interrupting the migration routes of endemic fish such as Schizothorax wangchiachii and Coreius guichenoti [5,33]. Although artificial propagation and release have maintained the number of important fish species, they cannot restore natural spawning and migration processes, leading to a simplified community structure and loss of genetic diversity [34]. Additionally, dam construction interrupts natural sediment transport, causing reservoir siltation and downstream riverbed erosion, which further damages the physical habitat of aquatic organisms [35].

The hydrological regime shows mild overall alteration (comprehensive variation degree 0.139) but significant local differences. The monthly average flow variation rate is less than 0.1, indicating that cascade reservoirs have effectively smoothed seasonal flow fluctuations. However, the extreme flow decline rate reaches 0.83, representing the most severely altered hydrological indicator. This high decline rate is primarily attributed to the daily regulation of the AH hydropower station, where rapid load adjustments for power generation induce sharp daily flow fluctuations. From an ecological perspective, such extreme flow declines affect benthic habitats and fish spawning grounds [36,37]: abrupt reductions in flow can reduce water depth and wetted perimeter, potentially exposing river margins and altering benthic microhabitats, while also disturbing spawning cues and reducing the availability of suitable spawning substrata for fish. The ecological flow guarantee rate of weekly regulated stations (>97%) is significantly higher than that of daily regulated stations (85.83% under the original EIA standard of 510 m³/s and 96.6% after adjustment to 350 m³/s). This confirms that larger reservoir regulation capacity is more conducive to meeting ecological water demand.

Water environmental quality generally maintains the Class II standard, and the reservoir areas are in an oligotrophic to mesotrophic state. This is mainly attributable to strict pollution control measures during construction and operation, which have effectively controlled point-source pollution. However, periodic exceedances of TN and TP concentrations still occur during the flood season, indicating that agricultural non-point-source pollution has become the main water pollution source in the basin [38,39].

4.2. Comparison with Similar Studies

The results of this study are generally consistent with previous studies on cascade hydropower basins, but there are also significant differences due to diverse evaluation frameworks and study areas.

In terms of ecological security level, the “generally safe” result of this study (0.71–0.74) is similar to the evaluation results of the upper Yangtze River basin (0.68–0.76) reported by Zhang et al. [40] and the Lancang River cascade basin (0.65–0.73) by Li et al. [41]. Meanwhile, it is higher than the results for more intensively developed river reaches. This is mainly because the lower Jinsha River has higher development intensity and more severe cumulative ecological effects, whereas the middle reaches have relatively intact tributary habitats and more effective ecological protection measures.

Regarding key limiting factors, river connectivity loss is identified as the core problem in almost all cascade hydropower basin evaluations. This factor exerts a prominent influence on basin ecological security risk, accounting for a weight of 22% in our research. In contrast, the study of the Yellow River basin by Qiu et al. [42] identified water pollution as the primary limiting factor. The generally good water quality in the middle Jinsha River reflects the differences in dominant ecological problems among basins.

In terms of evaluation methods, distinct differences in indicator selection exist between studies. The results of this study are more focused on ecological attributes compared with studies that use the traditional five-layer DPSIR structure. For example, the study of Shaanxi Province by Wang et al. [15] included 12 socioeconomic indicators in the driving force and response layers, resulting in a higher comprehensive security index (0.78–0.82) and a lower weight of ecological indicators. This confirms that the traditional five-layer DPSIR structure tends to overestimate ecological security due to the dilution effect of socioeconomic indicators, while the three-layer structure adopted in this study can more accurately reflect the actual ecological status.

Regarding weighting methods, the two studies adopt different weighting approaches, which leads to inconsistent ranking of indicator weights. The AHP-CRITIC coupling method used in this study assigns the highest weight to the number of important fish species (28%), followed by river connectivity (22%). This differs from the entropy weight method used by Zhang et al. [40], which gave the highest weight to water quality indicators (35%). The coupling method balances expert judgment on the ecological importance of indicators with objective data variability, thereby avoiding the bias of single weighting methods.

4.3. Limitations of the DPSIR Model and This Study

Although this study improves the DPSIR model by constructing a streamlined three-layer structure, the inherent limitations of the DPSIR framework still exist, and this study also has several shortcomings that should be addressed in future research.

First, the DPSIR model is based on an overly linear causal assumption, which presumes a one-way linear chain between driving forces, pressures, states, impacts, and responses. This assumption ignores the complex nonlinear feedback loops that are characteristic of ecosystems [43]. For instance, the degradation of aquatic ecosystems may in turn affect the sustainable development of hydropower (a driving force), and ecological flow regulation (a response) directly alters the hydrological regime (a state) rather than acting solely through pressure reduction. Consequently, this linear simplification may lead to an incomplete understanding of the interaction mechanisms between human activities and ecosystems, as noted in previous reviews of environmental assessment frameworks [44].

Second, the DPSIR model lacks consideration of dynamic feedback processes. It is essentially a static evaluation framework that cannot capture the time lag of ecological impacts or the long-term effects of response measures [45]. For example, the impact of dam construction on fish populations may take decades to fully manifest, and the effectiveness of artificial propagation and release also requires long-term monitoring to evaluate. The static evaluation based on cross-sectional data in this study is unable to reflect such dynamic changes.

Third, this study has certain limitations in data and indicator selection. The ecological monitoring data cover only seven years (2013–2019), which is insufficient to reveal the long-term evolutionary trends of the ecosystem. The indicator system focuses on longitudinal connectivity and the ecological status of the main stream, but it ignores lateral connectivity between rivers and floodplains, as well as indicators of terrestrial ecosystem changes caused by reservoir inundation. Moreover, this study does not quantify the cumulative effects of cascade hydropower stations, nor does it consider the impact of climate change on basin ecological security.

5. Conclusions

(1)

An ERA system suitable for river basins with cascade hydropower development was constructed, including three element layers and 10 indicator layers. The AHP and CRITIC methods were coupled to determine the weights, which took into account both subjective experience and objective data, and the evaluation results were scientific and reliable.

(2)

The hydrological regime in the middle reaches of the Jinsha River Basin presented mild variation, and the cascade hydropower stations significantly changed the seasonal distribution of flow; there were temporal and spatial differences in the ecological flow guarantee rate, and the guarantee effect of weekly regulating hydropower stations was better than that of daily regulating hydropower stations; the water quality generally maintained the Class II standard, and the reservoir area was in an oligotrophic to mesotrophic state; the number of important protected fish species did not decrease.

(3)

The overall ecological security of the basin was at a “generally safe” level with a comprehensive score of 0.71–0.74. Insufficient river connectivity was the core limiting factor, and the optimization of ecological flow dispatching and water environment management were the keys to improving the security level.