Contribution of the reservoir backflow to the eutrophication of its tributary: a case study of the Xiangxi River, China

Eutrophication of inland waters was a topic of widespread interest in the 1960s and 1970s (Ai et al. 2015; Huang et al. 2020). Several studies have shown that nutrient indicators such as nitrogen and phosphorus are important nutrients that determine the degree of eutrophication in water bodies (Gilbert & Burford 2017). This is also an important condition for controlling the eutrophication system of water bodies (Sepehri & Sarrafzadeh 2018). In recent decades, following the development of hydropower, the construction of dams has caused discontinuity between river sediments and hydrological and biological communities (Conley et al. 2009). The decrease of flow velocity in reservoir areas will expand the rivers in these areas into lakes, and the phytoplankton communities in their waters will evolve from river-type (such as diatoms and dinoflagellates) to lake-type communities (such as toxic cyanobacteria and green algae) (Andersen et al. 2019). The construction of a reservoir drives local economic development. However, a large amount of nutrients such as nitrogen and phosphorus will enter the reservoir with the tributaries, causing serious eutrophication effects and harmful algal blooms in the reservoir. As examples, N limitation in the Western South Pacific Ocean, as well as the Atlantic Ocean, has been reported (Hale et al. 2016). However, P limitation has been reported in the western Dutch Wadden Sea (Ly et al. 2014) and N limitation in acidic lake Caviahue, Argentina (Beamud et al. 2010). Nutrient limitation transitions are related to phytoplankton succession and the variation of nutrient composition that is sensitive to quite a number of factors, such as extreme weather like rainstorm and flood, nutrient retention within dams, different vertical mixing depths, and seasonal variations (Abell & Hamilton 2015; Ran et al. 2017). Some examples include the Miyun Reservoir and the Three Gorges Reservoir of China (Ye et al. 2007; Su et al. 2014).

Among them, the Three Gorges Reservoir is the most prominent. After the completion of the Three Gorges Reservoir, the tributaries in the backwater-inundated area of the Three Gorges Reservoir will be underpinned by the Yangtze River. This results in slow flow movement in local waters, the velocity of which is between 0.02 and 0.04 m/s (Cao 2010), below the critical velocity threshold of harmful algae bloom disappearance (0.05 m/s) (Mitrovic et al. 2003). At this status, harmful algae blooms easily develop when the nutrient concentrations in the water bodies are high. A typical example of eutrophication in the tributaries of the Three Gorges reservoir is the Xiangxi River. According to Ping Zhang's research (Li et al. 2007), the overall water quality of the Xiangxi River has reached a state of serious eutrophication. Several harmful algae blooms have occurred in the Xiangxi River in recent years. However, scholars have increasingly realized that the main cause of eutrophication in the Xiangxi River is related to the Three Gorges Reservoir. The impoundment of the Three Gorges Reservoir slows down the velocity of the Xiangxi River, which is conducive to the occurrence of harmful algae blooms (Cao et al. 2015). On the other hand, the regulation of reservoir water level and the frequent exchange of water between the Yangtze River and the Xaingxi River will introduce nutrients into the Xiangxi River and may also lead to the occurrence of harmful algae blooms (Zhang et al. 2012).

Although many scholars have studied the eutrophication of the Xiangxi River, their research focuses more on nutrient input and migration in the upper reaches of the Xiangxi River and its tributaries. The contribution of the high nutrient load of the Three Gorges Reservoir to the eutrophication of the Xiangxi River is generally ignored. In Xinqin Han's research on the changes in the concentrations of chlorophyll in the Xiangxi Bay, the results showed that the nitrogen in the water bodies is closely related to the occurrence of harmful algae blooms of the Xiangxi River (Han et al. 2006). Guangjie Zhou's evaluation of algae diversity in tributaries showed that the harmful algae blooms of the Three Gorges Reservoir are seasonal and can be managed by controlling external pollution and increasing water flow velocity (Zhou et al. 2006). However, the construction of dams destroys the continuum of rivers, interrupting the migration of nutrients. We should consider the influence of nutrient backflow on the eutrophication of the tributaries. Based on the above situation, combined with the convective diffusion principle of river dynamics, we suggest that the backwater caused the backflow of the Three Gorges Reservoir, which is inseparable from the eutrophication of the Xiangxi River. It is evident from the literature that most scholars in the study of the reservoir backflow work from the perspective of river sediment dynamics and river hydrology. For example, Yanan Huang's research is on the variations of stratified density currents in the Xiangxi Bay in the flood season (Huang et al. 2018), and Xiaoxiang Feng's research is on the sediment at the tail of tributaries under conditions of reservoir backflow (Feng et al. 2005). Few studies on the reservoir backflow come from the nutrient perspective.

In this study, we take the Xiangxi River as the research object, combined with the diachronic water quality indices of the Xiangxi River, also using established formulas for estimating the nutrient flux and the reservoir backflow intensity of the Xiangxi River. We aim to clarify the contribution of the backflow of the Three Gorges Reservoir to the eutrophication of the Xiangxi River and to provide a scientific theoretical basis for the prevention and control of the harmful algae blooms in the Xiangxi River.

The Xiangxi River is the largest tributary of the Three Gorges Reservoir, located in Hubei province, China. It originates in the Shennongjia forest region and flows through the Xingshan County and the Zigui County, and the main stream is 94 km long with a basin area of 3,091 km². Thus, it has three main tributaries, including the Nanyang River, the Gufu River and the Gaolan River. It belongs to the subtropical continental monsoon climate, and due to the large differences in terrain height and complex topography, the vertical variation of temperature is obvious. The annual average temperature is 16.6 °C, the annual average flow is 40.18 m³/s, and the annual average rainfall is 1,015.6 mm (Ye et al. 2003; Tang et al. 2004). In this study, we chose the Xiangxi Section (estuary), the Xiakou Section, and the Gaoyang Section of the Xiangxi River as sampling sites. The layout of the geographic information of the Xiangxi River is shown in Figure 1.

The sampling time in this study was from January 2014 to May 2016, and the sampling frequency was once a month. During the sampling process, we used the velocimeter to measure the velocity of flow at the Xiangxi Section of the Xiangxi River and measured at intervals of 5 m until reaching the bottom of the Xiangxi Section. We used a 1.5 L glass sampler to collect 1.5 L water samples. Sampling started at the surface (0 m), and then it was taken at the 5 m depth. Sampling depth increased exponentially, with each sample taken at a depth that was twice that of the previous depth until the bottom was reached. We divided the collected water samples into three parts (each part of 500 mL) in the laboratory. One part was filtered by Whatman GF/10 nucleopore, and then we leached the chlorophyll a (Chl-a) from the filter paper using 90% acetone. The filtered water sample was used to measure the concentration of soluble reactive phosphorus (SRP) and ammonium ion (⁠⁠), and another part of the water sample was used for measuring the concentration of total nitrogen (TN), total phosphorus (TP), and chemical oxygen demand (COD). The remaining sample was retained as a spare. We then used the alkaline potassium peroxydisulfate spectrophotometric method to measure the concentration of TN, the Nessler reagent spectrophotometric method to measure the concentration of ⁠, the ammonium molybdate spectrophotometric method to measure the concentration of TP and SRP, the potassium dichromate method to measure the concentration of COD and the spectrophotometric method to measure the concentration of Chl-a.

In this study, we believe that the degree of water exchange between the tributary bay and the reservoir is the main factor determining the change of nutrient flux. Therefore, in the following study, we will establish a quantitative evaluation method for the water exchange rate between tributaries and reservoirs, as well as a calculation method for changes in nutrient flux at the estuary of tributaries and reservoirs. Based on the above research results, the corresponding relationship between the water yield exchange rate and the nutrient flux was constructed, and the rule was verified by measured data.

In order to study the contribution of the reservoir backflow to the eutrophication of the Xiangxi River, it is necessary to analyze the reservoir backflow situation at the Xiangxi Section of Xiangxi River. At the intersection of the Xiangxi River and the Yangzte River, there will be a tributary gate, as shown in Figure 2(a). The flow pattern and flow structure in the Xiangxi Section depend on the degree of holding of hydrodynamic forces between the water flow of the Xiangxi River and that of the Yangzte River. Based on Zhang Sheng's research (Zhang et al. 2009), it was found that different forms of reservoir backflow exist in the Xiangxi River in different seasons. On this basis, the middle backflow occurs in spring and summer, and the surface backflow occurs in autumn and winter. The backflow form is shown in Figure 2(b).

As both the Yangzte River and the Xiangxi River are open channel flows, the energy of water flow is the sum of kinetic energy and potential energy. Because the riverbed in this area exists in a natural state of cohesion, there has been a little change of riverbed morphology. The potential energy here is relatively stable, and the water flow energy can be expressed by kinetic energy, and also momentum. We choose momentum as the research target in this study. According to the momentum mechanism of the reservoir backflow of tributaries, the intensity of the reservoir backflow can be defined as in Equation (5):

Considering that the change in the reservoir capacity will affect the reservoir backflow, it not only changes the area of the estuary section but also affects the discharge and velocity of the tributary, and the instability of the reservoir backflow will change with the slight change of hydraulic conditions. It is necessary to calculate the backflow intensity of the Three Gorges Reservoir with more stable indicators. Therefore, we further deduce and simplify Equation (6) through two aspects such as flow and flow velocity.

We averaged the water quality data obtained from different sampling points at different water depths and then put them in Figures 4 and 5. The focus of this study is the backflow of tributaries, so we used vertical dotted line markers to mark the fluctuation of water quality indices of the Xiangxi Section.

According to Figure 4, we found that the concentration of TN in each section is stable, and the differences between them were small during the monitoring period, except for the fluctuation of the Xiangxi Section in July 2014, which was 3.69 mg/L. The maximum value of the concentration of TN in the Gaoyang Section was 3.87 mg/L. The changes in the concentration of were dramatic, fluctuating during the whole water quality monitoring period. There is not much difference between the concentrations of of each group, except for the maximum value of the concentration of of the fluctuate of the Xiakou Section, which was 1.28 mg/L, and that of the Xiangxi Section, which was 1.10 mg/L. Like the changes in the concentration of TN, the changes in the concentration of TP in each section were stable, except for one violent fluctuation between December 2014 and March 2015, during which all sections reached a higher value in the range of 0.52–0.58 mg/L. While the changes in the concentration of SRP were gentle, it only fluctuated in March 2014 and April 2016; the maximum value of the concentrations of SRP of the two fluctuates is of the Xiangxi Section, which was 0.37 mg/L.

According to Figure 5, the concentration of Chl-a at each sampling section was under 5 mg/L at most times, but there were fluctuations in May and December of each year. The maximum value of the concentration of Chl-a in the Xiangxi Section during the fluctuation period was 152.24 mg/L, which was significantly higher than that of the other two sections. The changes in the concentration of COD are similar to that of Chl-a at each section; it changes between 1 and 3 mg/L and had some fluctuations in June of each year, but there was a big fluctuation in September 2015. The Xiangxi section displayed the maximum value of the COD concentration in that fluctuation, which was 5.18 mg/L.

We calculated the average estuary flow velocity data at different water depths, as shown in Figure 6(a). The average nutrient concentration level at different depths in the Xiangxi Section is shown in Figure 6(b). The results show that most of the reservoir backflow is surface-middle layer backflow during the monitoring period, while the upper water body at the Xiangxi Section showed an obvious reservoir backflow phenomenon. The velocity of the upper water body is of a triangular distribution, and the maximum value is 0.08 m/s at 5 m water depth. While the lower water body flow is a normal flow, it has a wedge-shaped distribution, gradually increasing with the water depth, with a maximum value of 0.11 m at 65 m water depth. Regarding the nutrient concentration distribution, the TN concentration increased sharply between 0 and 10 m, then stabled at about 1.4 mg/L with little fluctuation until the bottom of the estuary. Changes in the concentration of TP were similar to TN, but stabled at about 0.25 mg/L, and increased with water depth.

We calculated α and F according to Equations (3) and (14), as shown in Figure 7. We used dash-dot lines to mark both α= 1 and F= 1. The results showed that the value of α was less than 1 at most time, but F has several values greater than 1, with a cycle of about 3 months.

In order to verify the contribution of the reservoir backflow to the eutrophication of the Xiangxi River, we used a linear correlation analysis between the backflow intensity and the nutrient flux coefficient, the results of which are shown in Figure 8. The r of the linear correlation analysis is −0.52, and there exists a negative correlation between α and F.

According to Section 3.1, we found that the TN concentration of the Xiangxi Section is between the Xiakou Section and the Gaoyang section most of the time, because a nitrogen cycle exists in the river, and as it flows there will be more nitrogen deposited into the sediment in the lower reaches of the river, which drops the concentration of TN. As the widening of lower reaches of the river increases the surface water volume, algae environmental capacity will increase, and its nitrogen fixation capacity can stabilize TN to a certain level. According to Zhe Li's research (Li et al. 2009), the large fluctuation of the concentration of TN is related to reservoir backwater regulation, because the TN concentration of the Xiangxi section is larger than the other two sections during the fluctuation period (expect the fluctuation on February 2016). The concentration of is the same as that of TN, but it is more significant, considering that is a strong reducing substance with poor stability which may trigger autoxidation used by nitrifying bacteria in water bodies (Liu et al. 2017). But there exists a phenomenon of the sudden increase of the concentration of in the Xiangxi Section, decreasing from downstream to upstream. Based on the TN concentration changes, the above phenomenon provides strong evidence of the N backflow from the Three Gorges Reservoir into the Xiangxi River. The change in the concentration of TP in the Xiangxi Section was different from that in the concentration of TN, which was far higher than the remaining two sections during the whole monitoring period, except in February 2015 (the average concentration of TP of the Xiangxi Section is 0.22 mg/L, the Xiakou Section is 0.15 mg/L, and that of the Gaoyang Section is 0.13 mg/L). Considering that P is an important nutrient element of planktonic algae activities (Qin et al. 2013), it will drop rapidly during harmful algae blooms. But the concentrations of Chl-a and TP both fluctuated many times in the same period (March 2014, early 2015 and April 2016). The variation of the concentration of SRP in each section is similar to that of TP, but the fluctuation is smaller; the average concentration of SRP of the Gaoyang Section is 0.05 mg/L, the average concentration of SRP of the Xiakou Section is 0.08 mg/L and the average concentration of SRP of the Xiangxi Section is 0.12 mg/L. Elimination of the weak effect of atmospheric P deposition on the concentration of TP (Fan et al. 2010), combined with the high TP concentration in the Xiangxi Section, shows that reservoir backflow is an important factor leading to increasing concentrations of TP in the Xiangxi River. The changes in the concentrations of COD and Chl-a are similar. They fluctuate at the same time when the concentrations of N and P series indices fluctuate, where the most prominent fluctuation is from the Xiangxi Section. The rest of the time they both achieve a stable status, especially the concentration of Chl-a. Both proved the change in the eutrophication degree of the Xiangxi River at the observed level.

According to Figure 6(a), combining Kebin Dong's research on the relationship between velocity and algae growth (Dong 2010), if the flow velocity of the surface water at the Xiangxi Section creates a good external environment for algae growth, harmful algae blooms will occur with nutrient backflow, consistent with the changes in the nutrients. Figure 6(b) indicates that there was a fluctuation of the concentration of TN at 40 m water depth, where reservoir backflow and normal flow cross. This shows that nutrients can be fed back into the Xiangxi River by reservoir backflow. As seen in Figure 7, the change in the SRP is relatively mild, with an average value of α= 0.48, which indicates that there is a steady backflow of SRP from the Yangtze River into the Xiangxi River. The fluctuation of TN changed greatly, and its peak value was α = 4.35, but its average value was α = 0.78, which indicates that the change of TN is still mainly caused by the reservoir backflow. However, the changes in the value of α of and TP are similar. After several periods of small-scale normal flow (the maximum value of is α= 2.79, the maximum value of TP is α= 3.26), the main change pattern was still attributed to reservoir backflow, and the average value of was α= 0.61, while the average value of TP is α= 0.45. The changes of F in Figure 7 show the occurrence of many reservoir backflows during the monitoring period. According to the relevance analysis in Figure 8, we conclude that there is a significant correlation between reservoir backflow and eutrophication of the Xiangxi River. In general, the main cause of eutrophication of the Xiangxi River was found to be the reservoir backflow from the Three Gorges Reservoir.

Regarding nutrient flux, we define the product of the average velocity, the average nutrient concentration, and the area of the river section as the instantaneous nutrient flux of the section. Then, the nutrient flux in a certain period of time can be obtained by factoring the time. However, the accuracy of the nutrient flux values is affected if the convection velocity is averaged and then used for calculation. But after derivation, we found that because the flow velocity is a vector, the flow velocity will be contrary to the flow direction of the river due to the influence of the changes of the hydraulic conditions of the river bed, such as the disturbance of submerged plants, the obstruction of the pebble bed structure, and the viscous force of suspended sediment. When calculating the stratified velocity data, we need to use its module to calculate the nutrient flux. The calculated nutrient flux will be significantly larger than the nutrient flux obtained by averaging the flow rate. Moreover, the width of tributaries will gradually increase along the lower reaches, and the stratification of nutrient and flow velocity will gradually become more complex; the differences between nutrient flux values calculated by two different methods will become larger and larger. If the nutrient flux is still reversed at this time, then we can draw the conclusion of nutrient backflow. Therefore, our calculation method for measuring the nutrient flux of the Xiangxi River is reasonable.

In the calculation of backflow intensity, based on the law of conservation of energy, we choose kinetic energy as the research object in this study because the elevation at the estuary remains unchanged and the solid potential energy remains unchanged. Based on the characteristics of instantaneous changes in water flow, we study two aspects such as flow and flow velocity. Regarding the flow aspect, because of the influence and occurrence of reservoir backflow, we use the water exchange rate to study flow from the observation of volume change. Because the reservoir and its tributaries can be seen as a unified whole, we can separate the tributary from the rest of the reservoir. According to the constant variation of water volume between them and the spatiotemporal continuity of their flow, we use the water exchange rate to replace flow, which simplified the calculation process. Regarding the flow velocity aspect, we used the one-dimensional section of the estuary as the research point, dividing the flow velocity into two parts – normal flow and backflow. Taking the constant length of the one-dimensional sections of the two parts as the contact point, we get the ratio and then deduce the calculation formula of the reservoir backflow intensity.

In general, in the calculation of the reservoir backflow intensity, we replace the experimental variable with a stable variable by the equivalent substitution method, which not only simplifies the calculation but also increases the stability of the calculation results. The research methods we used in this study are reasonable and normative, and they have strong universality. They can be applied to practice, such as the calculation of the backflow intensity of river estuaries and the study of water exchange in lakes.

Previous studies have shown that the harmful algae blooms of the tributary require a specific and suitable growth environment. In consideration of the research on water temperature, sunshine, and algae species of Yuling Huang, Yanqing Zhang and Jingsong Guo (Huang et al. 2009; Guo et al. 2010; Zhang et al. 2014) and the findings of Section 4.1, it can be argued that reservoir backflow is obviously an important inducement for harmful algae blooms in tributaries. This is because it pours a large number of nutrients into tributaries and contributes to the growth of the primary producer communities in rivers. The primary producers in the rivers are divided into two parts: one is planktonic algae in the water body and the other is periphyton in the riverbed and wet periphery. Periphyton will grow with the increase of nutrient concentrations, but due to the limitations of the adherent area, it can only maintain the community at its own environmental capacity level, while planktonic algae are not restricted and will grow in large quantities, leading the emergence of harmful algae blooms.

In order to solve the algae problem, it is necessary to destroy the suitable conditions for the formation of harmful algae blooms. The most affordable and efficient way to accomplish this is to change the hydrodynamic conditions between the Xiangxi River and the Yangtze River. Because the Xiangxi River has both a normal flow process in the upper reaches and the reservoir backflow process in the estuary, the flow structure of the estuary is restricted by the balance between the kinetic energy of the flow of the tributary and the mainstream, which is easy to change. Changing the water level and the flow frequency of the Yangtze River at the estuary of the Xiangxi River can affect the steady state of the estuary flow, water temperature, nutrient concentration, and the accumulation of algae and can also change the distribution pattern of the reservoir backflow along the water depth direction at the estuary. Changing the direction and flow of the reservoir backflow can create an environment for reciprocating water exchange in the estuary, cutting off the steady supply of nutrients from the Yangtze River to the Xiangxi River. Efficient movement of the water body can also destroy the stable supply of light energy, while changing the hydraulic conditions of the two currents at the confluence of the Xiangxi River and the Yangtze River can adjust the relative scale of the true light layer and the mixed layer of the river, as well as reducing the accumulation space of algae in the surface water body the contribution of nutrient backflow in the Xiangxi River to the growth environment of harmful algae blooms. The above measures can be applied to prevent harmful algae blooms, but when harmful algae blooms are present, the drainage and concurrent power generation of the Three Gorges Reservoir is an effective way to control the algae blooms, because it can increase the flow velocity of the Xiangxi River, rejecting planktonic algae and nutrient. This can also change the poor nutrition of the lower reaches caused by dam interception and increase the economic benefits of the reservoir.

At present, China has not established a set of standard operation specification for nutrient flux measurement in the reservoir area based on water volume change. This would be detrimental to the management of the water environment in reservoir tributaries and bays, leading to high ecological risk and uncontrollable eutrophication. Therefore, in order to grasp the influence law of nutrient flux changes in the reservoir area more conveniently, it is a good choice to use the assessment of water exchange rate. In this study, a mathematical relationship between the water exchange rate and nutrient flux changes was established. It realizes the improvement of operability and standardization of nutrient flux quantification in the reservoir bays of tributaries in the Three Gorges reservoir area, which is conducive to the accurate observation of eutrophication in all reservoir bays of tributaries in the reservoir area under a unified standard.

Based on the analysis of the diachronic water quality data of the Xiangxi River, we defined and calculated the nutrient flux and the backflow intensity of the Three Gorges Reservoir. There is a significant negative correlation between these factors, which proves that the backflow of the Three Gorges Reservoir has a great contribution to the eutrophication of the Xiangxi River. Moreover, we find that changing the hydraulic conditions of the backflow of the estuary can destroy the suitable flow environment for algae survival and prevent harmful algae blooms.

We have given guidance and help to the collection of sediments in our experiment. W.L. was responsible for the whole research work, including the experimental work and the analysis of the results. X.L. designed the main content framework of the manuscript and assisted in the analysis of the experimental results. J.L. assisted in the preparation and operation of the experiment. M.B. assisted in some data analysis and the writing of this manuscript.

This study is supported by the State Key Laboratory of Eco-hydraulics in Northwest Arid Region, Xi'an University of Technology (Grant No. 2018KFKT-7), the National Natural Science Foundation of China (Grant No. 52109099), and the Fundamental Research Funds for the Central Universities (Grant No. 2042020kf0004), and the National Natural Science Foundation of China (Grant No. 52079094).

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this manuscript.

Data cannot be made publicly available; readers should contact the corresponding author for details.