Analysis of changes in the ecohydrological situation and its driving forces in the Li River Basin

Climate change and human activities have altered global and regional hydrological cycle processes (Dimitriadis et al. 2021). Climate change can directly affect the hydrological cycle by altering precipitation or evapotranspiration. For instance, warmer temperatures promote surface evaporation, which releases more vapor in the air, which may increase the probability of rainfall events (Olafsdottir et al. 2021; Su et al. 2024). However, human activities have altered the hydrological cycle in various ways, with land use and land cover change being one of the most direct ways (Guan et al. 2023). With climate change and intensified human activities rising, many rivers worldwide show a significant downward trend in runoff (Bissenbayeva et al. 2021). These changes threaten global or regional water security.

The driving mechanisms affecting runoff changes vary considerably at different time scales. Quantifying the impacts of climate change and human activities on hydrological situations at different time scales based on different models (Zeng et al. 2014; Huang et al. 2016). Runoff change characteristics are critical indicators for maintaining the balance between supply and demand and ecological stability in a watershed, and it is of practical significance to analyze changes in the hydrological situation of the watershed as a whole and its driving studies (Huang & Qiu 2022; Guo et al. 2022b).

For the analysis of the hydrological situation, most scholars have qualitatively analyzed the hydrological conditions from the perspective of overall hydrological changes in the basin (Wang et al. 2022). Previous studies have focused on three main aspects of analyzing runoff-driving mechanisms. The first is the quantitative attribution of runoff changes based on different analytical formulations of the Budyko hypothesis. For example, Lv et al. (2019) studied the impact analysis of a typical tributary of the Yellow River (Fen River Basin) on climate change and changes in watershed characteristics based on Budyko's hypothesis and found that the contribution of lower watershed surface changes to runoff changes was 92.27% during the period 1951–2010. Zheng et al. (2021) an attribution analysis of runoff changes in the Kuye River Basin based on three Budyko methods showed that the characteristic parameters of the catchment water surface were the most critical factors influencing runoff changes in the Kuye River. However, the Budyko model could only quantitatively analyze the relative contribution of runoff change drivers on the yearly scale and could not quantitatively isolate the impacts of the runoff change drivers on different time scales. The second is the reconstruction of natural runoff based on a data-driven model, which analyses the causes of runoff changes based on simulated and actual runoff. Guo et al. (2023) introduced an extended- and short-term memory model to model meteorological flows in the Xiangjiang River Basin and used a separation framework to quantify the impacts of anthropogenic disturbances and climate change on ecological flows at multiple time scales. The third is to quantify the relative contributions of climate and sub-humans through physical hydrological modeling, including the Soil and Water Assessment Tool (SWAT) model, the variable infiltration capacity (VIC) model, by restoring natural runoff sequences and comparing measured and simulated runoff. Wang et al. (2024) assessed the sensitivity of several climate change scenarios and land use changes to runoff in the Qin River Basin through the SWAT model, and the simulation results were sufficiently accurate to study the effects of regionally varying environmental conditions on runoff response. Yang et al. (2020) analyzed runoff changes in the Weihe River Basin using a VIC model and found that climate change is dominant in causing runoff reductions. In contrast, runoff reductions due to human activities are relatively small.

Although the river basin is the fourth largest in Hunan Province, it is an important base for grain, cotton, and other crops and livestock breeding commodities in Hunan Province and China (Hu et al. 2011). However, most of the previous hydrological situation studies are based on a single model analyzing the drivers of runoff change in a catchment on an annual scale, needing more detailed time scales (seasonal and monthly scales) (Zeng et al. 2014). The seasonal precipitation distribution and reservoirs' scheduling and operation over a year can alter the intra-annual runoff processes in a watershed (Ning et al. 2022). Therefore, quantitative attribution analyses of runoff at seasonal and monthly scales are essential. This study constructed an analytical framework to quantify the drivers of watershed runoff changes at different time scales. This multi-scale analytical approach provides a more detailed and comprehensive perspective on understanding the dynamics of watershed hydrological cycles and improves the accuracy and reliability of the analyses through the mutual verification of the results between models. The research results can provide new ideas for a comprehensive analysis of the basin's ecohydrology and the influencing factors behind it, as well as for the management of water resources and the restoration of the ecosystem in the Li River Basin.

In order to address the lack of research on the changes in the hydrological situation of the Li River Basin, this study is divided into the following three main steps: (1) comprehensive quantitative assessment of the degree of change in the overall hydrological situation of the basin using the indicators of hydrologic alteration (IHA)–RVA method; (2) the elasticity coefficient method based on the Budyko assumption describes the contribution of the drivers that cause changes in annual runoff; (3) construct a separation framework under the monthly scale of the ABCD model to quantitatively analyze the impacts of climate change and human activities at different time scales (seasonal and monthly) under the ecohydrological changes in the watershed. The quantitative results of the ABCD model were compared with those derived from the traditional method (Choudhury–Yang formula).

Li River is located at 109°30′–112°E longitude and 29°30′–30°12′N latitude, with a total length of 388 km in its mainstream. Its basin is mainly in the northwestern part of Hunan Province, and the overall terrain is high in the northwest and low in the southeast. Finally, it is injected into China's second largest freshwater lake, Dongting Lake, from Xiaoduokou, which has the highest runoff of the river in Hunan Province. The region has a subtropical monsoon humid climate due to the influence of topography, often producing cyclone rain. Frontal rain, rainfall intensity, rainfall, the distribution of precipitation in the watershed of the upper reaches of more than the lower reaches of the distribution of more than the hills than the plains, especially to the northwest of the alpine region of the precipitation is large. Land use types in the watershed are dominated by woodland, grassland, and cropland. Li River is the fourth largest river in Hunan Province and belongs to the Xiangxi Hydropower Base, one of the 12 largest hydropower bases in China. Nowadays, two large reservoirs, Jiangya Reservoir and Zaoshi Reservoir, have been built, greatly improving the flood control and regulation of the river's lower reaches.

The cumulative distance level method is the cumulative value of the difference between the annual mean hydrological data and the multi-year annual mean hydrological data, and the extreme point where the cumulative amount exists is selected as the hydrological mutation point. Specific principles are described in reference (Pei et al. 2022).

These three methods have been widely used to test the mutability of hydrological data series, so to further understand the evolution of runoff in the Li River Basin, we used the M–K sliding t-test with the cumulative distance level method.

Each of the four parameters in this model, a, b, c, and d, has a different range of values and physical significance. The parameter's value ranges from 0 to 1, indicating the tendency for runoff to occur before the soil is fully saturated. The value of parameter b ranges from 0 to 1,000 and represents the upper limit of the sum of soil moisture storage and total actual evapotranspiration in the watershed. The value of parameter c ranges from 0 to 1 and indicates the proportionality coefficient of the recharge of the soil water layer to the groundwater. The value of parameter d ranges from 0 to 1 and represents the groundwater outflow coefficient (Xin et al. 2019). The above ABCD model needs to set the initial values of initial groundwater storage G₀ and soil storage S₀ and rate the four parameters a, b, c, and d. We used a genetic algorithm to rate the ABCD model parameters (Ji et al. 2021). Where G₀ is taken as 150 mm, S₀ is taken as 80 mm, a is taken as 0.99, b is taken as 692, c is taken as 0.91, and d is taken as 0.99.

To simulate the basin runoff reasonably and to better test the model's parameters, the model was tested and validated at the beginning of the simulation to decide whether it was suitable for this catchment. Previous authors have often used periods of low human activity in the first 10–20 years of the sequence as a test period for model performance assessment (Wu et al. 2019; He et al. 2022). Wu et al. (2019) and others argue that the ABCD model has been widely used as a hydroclimatic model to study the response of watersheds to climate change, using the period of low human activity as the test period. He et al. (2022) found that 1960–1980 were the years in the whole of China that were less affected by human activities and could be considered as runoff in a natural state.

We used the M–K test, sliding t-test, and cumulative distance level method to analyze the Li runoff series mutation trend from 1961 to 2019 at the Shimen station of the river. The results of the test are shown in Table 2, which shows that it is evident that 1983 is the mutation year jointly tested by these tests. In order to quantitatively evaluate the hydrological changes in the basin before and after the mutation year, the extended-time series of runoff in the Li River Basin 1961–2019 was divided into two time periods using the mutation year 1983 as the demarcation point: 1961–1983 as the natural period, and 1984–2019 was regarded as the change period.

This paper also calculated the degree of change of each group of indicators and the overall hydrological alteration degree of the Li River Basin based on the RVA method (see Table 3). After the mutation, the degree of change was more minor for the group of monthly flow averages (Group I) and the group of time of occurrence of annual extreme flows (Group III), which were 11.16 and 22.15%, respectively. In contrast, the annual extreme flow group (Group II), the high and low pulse group (Group IV), and the flow alteration rate group (Group V) all had moderate levels of alteration. Therefore, we determined that the overall hydrological alteration of the Li River Basin is moderate (36.52%). This suggests that the natural hydrological state of the river basin is being disturbed in a nontrivial way after the abrupt change and that identifying the reasons behind the change will allow us to better deal with the different needs of humans and nature.

Huang et al. (2023) analyzed the evolution of basin runoff characteristics due to climatic and anthropogenic changes in the Kuno River Basin based on the time-varying Budyko framework for the period 1981–2018, and it was found that the Kuno River's runoff volume decreased by 30.65 mm during the change period. Wang et al. (2021) based on the sensitivity analysis of runoff changes in the Baiyangdian watershed from 1960 to 2010 based on the Budyko hypothesis, the runoff volume decreased by 43.41 mm according to the sensitivity analysis. However, the elasticity coefficients and parameter changes before and after the sudden change of the river basin in this paper are shown in Table 4; compared with the natural period, we can find that precipitation, potential evapotranspiration, and runoff depth in the change period have been reduced to different degrees, in which runoff depth has been reduced the most, by 51.31 mm, while the potential evapotranspiration has been reduced by 41.25 mm, the precipitation has been reduced by 8.36 mm, and the subsurface parameter has been increased by 0.09 mm. In addition, the average annual aridity index (ET₀/P) and the average annual runoff coefficient (R/P) decreased by 0.03. However, both coefficients are more significant than 0.5, suggesting that there is still abundant water flow in the Li River Basin. ε_P, ε_ET0, and ε_n increased by 0.002, decreased by 0.003, and increased by 0.039 after the mutation, respectively. This suggests that runoff is negatively correlated with potential evapotranspiration (ET₀) and positively correlated with precipitation (P), and subsurface parameters (n). For every 1% increase in P, ET₀, and n, the runoff depth increased by 1.31%, decreased by 0.44%, and decreased by 0.53%, respectively.

We calculated the effect of precipitation, potential vapor dispersion, and subsurface changes on runoff, and the results are shown in Table 5. Compared to the natural period, when we study the climatic factors (precipitation and potential evapotranspiration), we can find that the average runoff depth, R, decreases by 51.31 mm during the mutation period; multi-year average precipitation P decreased by 8.36 mm, and the resulting change in runoff increased by 64.01 mm, contributing 64.51%; potential evapotranspiration ET0 decreased by 41.25 mm, which drove an increase in runoff change of 15.21 mm, contributing 15.21%, and the subsurface parameter n increased by 0.09 mm, which drove a decrease in runoff change of 20.01 mm, contributing 20.16%. So, we can calculate the change in runoff due to climate change (precipitation and potential evapotranspiration) as 79.22 mm, contributing 79.84%.

From the perspective of watershed ecosystem integrity, natural runoff conditions are critical to maintaining the health and integrity of river ecosystems (Luo et al. 2018). Climate change and human activities work together to affect river runoff. Our study shows that the impact of human activities in the river basin is significantly higher than that of climate change and is the main driver of changes in the hydrological situation in the basin (Bissenbayeva et al. 2021). It can be argued that human activities and climate change pose threats to the river's ecosystem stability in many ways (Koch et al. 2023). Therefore, discussing the potential ecological impacts of the changing hydrological situation in the Li River Basin is necessary.

Wu et al. (2024) assessed the contribution of climate change and human-induced runoff in the Black River Basin at different times through the Budyko coupled equilibrium equations, and analyzed the drivers of runoff depletion by applying the WTC/Multivariate Wavelet Coherence (MWC) and Generalized Additive Model (GAM) models, and the results showed that changes in subsurface characteristics caused by human activities were the main factors affecting runoff in the study area. As shown in Dang (2021), the runoff volume of the upper Gan River trough watershed is influenced by the combination of precipitation and human activities, and human activities have an increasing impact on the runoff. Li et al. (2022) used the Budyko equation and the elasticity coefficient method to calculate the contribution of different factors to the runoff changes in the Jialing River Basin. The results showed that human activities are the primary factor in the runoff and vegetation changes in the Jialing River Basin. However, this paper presents the joint test of three tests, the M–K test, the sliding t-test, and the cumulative distance level method, which shows that 1983 was the year of the sudden hydrological change of the Li River. Before the mutation (before 1983), the runoff of the Li River Basin was stable; after the mutation (after 1983), the natural hydrological conditions of the river changed significantly, with runoff reduction showing a downward trend and the unevenness of mean annual precipitation becoming more prominent. Based on the IHA indicator, we can find that, in comparison with the pre-hydrological variability, the hydrological extremes of flow are reduced after the variability. Most of them remain in a low degree of alteration, but the degree of alteration of the annual average 1 day minimum flow reaches 85.14%. At the same time, 5 groups of 32 ecohydrological indicators were moderately altered. Among them, changes in the number of high- and low-flow pulses negatively affect the structure of aquatic organisms' food chains (Roach & Winemiller 2015; Luo et al. 2018). In addition, we found that the rate of flow increase in the river basin was moderately altered. The number of reversals was highly altered, which may exacerbate the potential risk of drought stress on local vegetation. This could lead to a further reduction in vegetation cover and also affect the safety and security of water requirements of mobile organisms along the river banks (Nkiaka & Okafor 2024).

In addition, it was found that the extreme flow events in the river basin are no longer in a stable range, mainly because the construction of terrace reservoirs can vastly change the hydrological processes of natural rivers, resulting in an artificial and constant environment that lacks natural extremes, which leads to a greater degree of alteration of the hydrological indicators, which brings about an impact on the ecosystem of the Li River Basin (Su et al. 2023). As shown by Shakarami et al. (2022), extreme flow alterations caused by dam construction affect river flow regime changes to a certain extent, while climate change has a role in altering river channel morphology. The construction and operation of reservoirs affect the redistribution of water resources in both time and space; for the reservoir area upstream of the dam, it will lead to an increase in the water level, a decrease in the flow rate, and non-seasonal fluctuations in the water level of the reservoir with the operation of reservoir scheduling, thus changing the natural hydrological pattern; in addition, for the downstream area below the dam, the daily regulation of the hydropower plant may lead to the downstream river water level and flow rate fluctuations, which may adversely affect the habitat of aquatic organisms (Zhang et al. 2024). Meanwhile, as stated by Wang et al. (2021), human activities such as damming and interception have global impacts on river flow and sediment conditions. Interception will lead to an increase in the gradient of water flow, which degrades the upstream riverbed and silts the downstream reaches, affecting the geomorphology of the riverbed.

Watershed runoff processes are nonlinear and highly complex (Nia et al. 2023). On the one hand, the Budyko model was used to study the drivers of runoff changes. Based on the Choudhury–Yang equation, potential evapotranspiration is more sensitive to runoff changes than precipitation, with 18.61%. At the same time, human activities are the leading cause of runoff changes, contributing 72.46%. Thus, human activities' contribution can be seen to be more significant. On the other hand, different models and methods can tap different information in making runoff predictions. Each method has its limitations and applications, and choosing the appropriate method according to the purpose of the study is an effective research tool (David & Schmalz 2020; Jahangir et al. 2023). Comparison of the ABCD model results with the annual-scale quantification results derived from the elasticity coefficient method of Budyko's hypothesis reveals that they are essentially the same, demonstrating the reliability of the two models in analyzing the drivers of runoff variability in the Li River Basin. This section discusses the Budyko and ABCD models' results to further describe the variability of river runoff-driving mechanisms across time scales.

We assess the drivers of runoff variability at multiple time scales using the ABCD model and compare the results with the annual-scale quantification derived from the elasticity coefficient method based on Budyko's assumptions. We find that the results are essentially the same, thus confirming the reliability of the ABCD model. Moreover, based on this, the model enables quantitative attribution of runoff changes on monthly and seasonal scales. Li (2021) analyzed the evolution of runoff in the Li River Basin from 1967 to 2016 based on the SWAT model and the double cumulative curve method and found that the contribution of climate change and human activities to runoff change is basically equal. However, in most cases, climate change has a greater impact, and the decrease in precipitation is the main factor for the decrease in net flow in the Li River Basin. Xiao (2015) analyzed the impacts of climate change and human activities on runoff in the four water basins of Dongting Lake from 1960 to 2010 using the Butyko model and the Geomorphology-Based Hydrological Model (GBHM) distributed hydrological model. Both models showed that climate change was the main factor leading to changes in runoff in the four water basins, with changes in rainfall playing a major role. Our study not only supports the results on annual scales but also further extends the study on more detailed time scales and finds clear differences in the impacts of climate change and human activities on runoff on seasonal and monthly scales.

On a seasonal scale, spring and winter runoff are mainly influenced by human activities, contributing 59.74 and 60.77%, respectively, while summer and autumn are mainly driven by climate change, contributing 59.26 and 61.98%, respectively. On a monthly scale, climate change had a reducing effect on runoff throughout the year (except January, February, March, and July), with runoff changes ranging from 0.53 to 6.89 mm from January to March, and the enormous change in runoff of 43.26 mm in July. In April–September (except for April and June) human activities drive runoff changes lower, and climate change contributes more to runoff changes, which suggests that climate change has intensified during the flood season. The degree of change is higher than human activities, mainly since floods in the river basin are concentrated from April to September. The river terrace reservoirs play a function of storing water and preventing floods in this period. Human activities contributed even more in April and June, driving a reduction in runoff of 16.37 and 20.03 mm, respectively, due to anthropogenic impacts. This phenomenon is caused by upstream and downstream dispatching in reservoir complexes (Wu et al. 2022). In order to prevent flooding, reservoir complexes store water at the end of the flood season, which is used to cope with the shortage of water for people's living and production during the non-flood period. Precipitation during the flood season is also more concentrated, which makes the interaction between climate change and human activities reach its maximum during the flood season (Zhou et al. 2018; Li et al. 2023).

We quantitatively assessed the changes in the Li River hydrological situation on multiple time scales, found the abrupt change years of the river runoff by using M–K, sliding t-test, and cumulative distance level method, analyzed the periodicity of the river runoff sequence by using complex Morlet wavelet analysis, analyzed the potential ecological impacts of the changes of the river hydrological situation by using the IHA–RVA method, and analyzed the potential ecological impacts of the changes of the river hydrological situation based on the Budyko assumptions and the ABCD model. The contribution of climate change and human activities to runoff changes at different time scales was quantitatively separated. The study showed that 1983 was the point of abrupt change in the runoff volume of the river in the natural runoff state. After the mutation, the annual runoff was reduced by 5.27%. The 32 ecohydrological indicators in the Li River Basin were dominated by low- to medium-degree changes, and the overall hydrological situation change was moderately altered by 36.52%. This not only has a certain degree of negative impact on the reproduction and survival of local organisms but is also not conducive to the ecosystem restoration and protection of the downstream Dongting Lake wetland. In addition, human activities are the main driver of runoff changes, contributing 72.46%. The impacts of climate change and human activities on runoff show significant differences on seasonal and monthly scales. The analytical framework constructed in this study can quantify the drivers of watershed runoff change at annual, quarterly, and monthly scales to achieve a comprehensive assessment of the Li River Basin.

Although these models have been successfully applied to quantify and assess the effects of climate change and human activities on hydrological regime changes, they still have some uncertainties and limitations (Samantaray et al. 2021). For traditional physical models, the input data requirements for runoff prediction and reconstruction are very stringent; however, the lack of some of this data can lead to the model's failure. The data-driven class of models is centered on black-box models, which are less explicit about the principles of runoff simulation. These models do not reflect the hydrological processes of runoff variability due to precipitation, potential evapotranspiration, and other factors and are not conducive to the quantitative analysis of the drivers of runoff variability (Jakeman & Hornberger 1993; Sajikumar & Remya 2015; Oyebode & Stretch 2019). Therefore, future studies should further refine the study area and analyze the differences in hydrological processes and ecological responses in different regions within the basin in multiple directions in order to gain a more comprehensive understanding of the complexity and diversity of the basin ecosystem. Moreover, future research should be combined with hydrological models to analyze the river basin water resources system and provide a further basis for water resources management and river ecosystem protection.

The authors would like to thank their brothers and sisters at the North China University of Water Resources and Electric Power for their comments and help with this study.

Y.L. conceived the study and wrote the first draft, Y.L., S.C., and W.Y. collected and analyzed the data, X.B., Y.M., Z.Y., L.Y., and B.W. analyzed the methodology, and H.W. and W.G. supervised the thesis, and all the authors provided inputs and assistance on the first few versions of the manuscript. All authors read and approved the final manuscript.

This study was supported by the Basic Research Project of Key Scientific Research Projects of Colleges and Universities of Henan Province (24ZX007).

The authors declare there is no conflict.

Data available on request from the authors. The data that support the findings of this study are available from the corresponding author upon reasonable request.