Optimization of the flow conditions in the spawning ground of the Chinese sturgeon (Acipenser sinensis) through Gezhouba Dam generating units

Chinese sturgeon (Acipenser sinensis) is a critically endangered species, and waters downstream from Gezhouba Dam are the only known spawning ground. To optimize the velocity conditions in the spawning ground by controlling the opening mode of Gezhouba Dam generator units, a mathematical model of Chinese sturgeon spawning ground was established in FLOW-3D. The model was evaluated with velocity measurements, and the results were determined to be in good agreement. By inverting the 2016–2019 field monitoring results, the model shows that the preferred velocity range for Chinese sturgeon spawning is 0.6–1.5 m/s. Velocity fields of different opening modes of the generator units were simulated with identical discharge. The suitable-velocity area was maximal when all units of Dajiang Plant of Gezhouba Dam were open. For discharges below 12,000 m3/s, most of the area was suitable; for discharges above 12,000 m3/s, the suitable area rapidly decreased with increasing discharge. A comparison of suitable areas under high-flow showed that at discharges of 12,000–15,000 m3/s, opening 11–13 units on the left side was optimal. For discharges above 15,000 m3/s, all units should be open. We used these results to recommend a new operation scheme to support the conservation of Chinese sturgeon.

The Gezhouba Dam Project is the first hydropower station on the Yangtze River. The dam is intended to provide multiple benefits to society, including power generation and prevention of flooding or droughts. However, dams may also change the transport of water and sediment 1 , which affects fish habitats as in the case of the required habitats of Chinese sturgeon (Acipenser sinensis) 2,3 . The Chinese sturgeon is a large anadromous fish, a national first-class protection animal and a critically endangered species 4 . Before the construction of Gezhouba Dam, the spawning ground of the Chinese sturgeon were mainly in the lower reaches of the Jinsha River and upper reaches of the Yangtze River 5 . After the closure of Gezhouba Dam, the Chinese sturgeon selected a new spawning ground in the waters downstream of Gezhouba Dam, which now represents the only known Chinese sturgeon spawning ground 6 . According to the results of continuous monitoring in recent years, the reproductive geographic range and amounts of Chinese sturgeon have substantially decreased 4,7 .
Efforts to conserve the Chinese sturgeon have focused on determining suitable hydrologic or hydraulic conditions of the spawning ground. One of the hydraulic environmental variables thought to be important for reproduction is the flow velocity in the spawning habitat. A previous study has shown that the Chinese sturgeon actively selects flow velocity conditions that are beneficial to its habitat and reproduction 8 , and flow velocities exceeding the maximum tolerable velocity of the fish affect the normal habitat 9 . Research on the flow velocity of the spawning ground of the Chinese sturgeon has mainly focused on two aspects: historical data of field measurements and numerical simulation. One study based on field measurements concluded that the Chinese sturgeon chose an area with a flow velocity of 0.62-1.16 m/s when spawning 10 . The hydrological data and measured velocity of the spawning ground of the Chinese sturgeon downstream of Gezhouba Dam were analysed to calculate the velocity range of 1.0-2.0 m/s 11  www.nature.com/scientificreports/ on site and concluded that the suitable discharge range for Chinese sturgeon was 1.07-1.65 m/s 12 . In another study, researchers measured the velocity in the spawning ground by ADCP and found that the average velocity of the spawning ground was 0.73-1.75 m/s 13 . A numerical modelling study retrieved the flow field during historical detection days and concluded that the suitable velocity range of Chinese sturgeon was 1.1-1.7 m/s 14 . In a hydrodynamic simulation of the spawning ground of Chinese sturgeon, the authors concluded that the most suitable velocity range was 0.97-1.48 m/s 15 . In another numerical modelling study, researchers simulated the water level and velocity by a two-dimensional (2D) hydraulic model and found that a velocity of 1.06-1.56 m/s was suitable for the spawning of Chinese sturgeon 16 . These results confirmed the preference of Chinese sturgeon for a range of flow velocity while spawning, but the estimated suitable ranges differ due to differences in research precision and methods. In addition, the effect of the dam operation on the characteristics of water flow in the spawning ground has been studied 17,18 , but few improvement measures and methods have been proposed. The waters downstream from Gezhouba Dam are the only known spawning ground of Chinese sturgeon, but the sensitivity of the velocity fields to the operating modes of Gezhouba Dam is unclear. The purpose of this study is to determine how to optimize releases from the dam to improve the spawning habitat. We continuously monitored the spawning ground of Chinese sturgeon every year from 1982 during the prospective spawning period. To effectively model links from dam operations to the spawning habitat, we used a combination of numerical simulation and field monitoring. Chinese sturgeon is a type of benthic fish. After its eggs are washed away by water, they adhere to the cracks in the bottom and hatch. Therefore, the bottom hydraulic parameters are more significant than the 2D vertical average hydraulic parameters 8 . A mathematical model of the three-dimensional (3D) hydrodynamics of the spawning ground of Chinese sturgeon was established. The velocity field experienced by Chinese sturgeon from 2016 to 2019 was determined by comparing the hydroacoustic locations of fish with the model results. From this comparison, we estimated the preferred velocity range for spawning of the Chinese sturgeon. Furthermore, the optimal scheme of different generating units of Gezhouba Dam was simulated and analysed to form recommendations for alternative release procedures to support the reproduction of the sturgeon.

Results
Flow velocity threshold. There were 92 Chinese sturgeon signals from 2016 to 2019, which were identified with the DIDSON dual-frequency video sonar system. The distribution map of Chinese sturgeon signals was shown in Fig. 1. The number of monitored signals in 2016 was significantly higher than in 2017-2019. The latest wild reproduction of the Chinese sturgeon occurred in 2016. Overall, most Chinese sturgeon signals were distributed within 500 m downstream from Gezhouba Dam, and there were more in the right side(facing downstream) than in the left side. The flow field of each sturgeon signal was simulated by the model, and the velocity of each signal location was obtained. According to the statistical analysis of the flow velocity values, the frequency of the sturgeon signal at different flow velocity values was shown in Fig. 2. The results show that most signals were concentrated in areas with flow velocities of 0.6-1.5 m/s, which accounted for 88.1% of the signals; areas with flow velocities below 0.6 m/s accounted for 4.3% of the signals, and areas with flow velocities above www.nature.com/scientificreports/ 1.5 m/s accounted for 7.6%. Therefore, 0.6-1.5 m/s was selected as the preferred flow velocity range of the Chinese sturgeon for spawning activity. This result was approximately consistent with the ranges proposed by most other researchers. The low limit of the velocity range was lower than that of other researchers. There may be two reasons for this result: the first was that the bottom velocity we analysed was lower than the surface velocity and vertical average velocity under the same conditions; the second was that our research time was after 2016, and the discharge during the spawning period was relatively low, so the velocity of the Chinese sturgeon signal was also relatively low.  Figure 3 shows the flow fields of the spawning ground under different opening modes with identical discharge. By comparing the areas with a velocity threshold range of 0.6-1.5 m/s in different cases, the most favourable opening mode was determined. In case 1, the velocity at the outlet of the units was higher than the 1.5 m/s velocity threshold, but the discharge of each unit was only 516.6 m 3 /s, so the high-velocity range was limited, and most areas were suitable. In case 2 and case 3, there was a large difference in proportions of suitable area. Because the left side was deeper than the right side, the flow velocity on the right side was higher under the same  www.nature.com/scientificreports/ discharge, and case 3 more easily exceeded the flow threshold, which resulted in a larger unsuitable area. Case 2 was more suitable than case 1, which also demonstrated that opening the left-side units was more favourable. In Once the design discharge is exceeded, the spillway on Erjiang River discharges water, and the velocity distribution of the study area is not affected. Therefore, case 1 represents the lowest discharge of 5590 m 3 /s, and case 2 represents a discharge of 6000 m 3 /s. For each subsequent case, the discharge was increased by 1000 m 3 /s to case 13 with the highest flow of 17,930 m 3 /s. In case 14, all units reached the design discharge, and the discharge of each unit was 825 m 3 /s 19 . Figure 4 shows the proportion of suitable-velocity area with all units open under different discharges. According to the calculation results, the proportion of suitable area slightly fluctuated at approximately 96.2% for discharges of 5590-11,000 m 3 /s. Because the discharge of each unit was low, the velocity of the unit outlet was low, and most areas were within the velocity threshold. Therefore, it is advantageous to open all units when the discharge is low. After the discharge reached 12,000 m 3 /s, the proportion of suitable area rapidly decreased. Because the discharge of each unit was high, on the right side of Dajiang River, the velocity of the unit outlet exceeded the velocity threshold and increased with increases in discharge, and the range of effect gradually increased. In the last case, the proportion of suitable area was only 6% when the units reached the designed discharge of 825 m 3 /s. Because the discharge of each unit was too high, almost all areas exceeded the velocity threshold except for small areas on both sides. Therefore, at discharges below 12,000 m 3 /s, opening all units is favourable, and at discharge above 12,000 m 3 /s, a higher discharge corresponds to more unfavourable conditions.
Optimal scheme under high-flow conditions. High-flow conditions at Gezhouba Dam are considered those that exceed 12,000 m 3 /s because of the substantive decline in suitable habitat area at higher discharges. Because opening the units on the left side of the Dajiang Plant provides a more uniform, suitable habitat, we evaluated 20 cases with a left-side opening mode under different discharge, as shown in Table 3 Figure 5 shows the proportions of suitable area for different opening modes under high-flow conditions. The calculation results show that when the discharge was 12,000 m 3 /s, 13,000 m 3 /s, and 14,000 m 3 /s, the proportion of suitable area showed a parabolic trend with the increase in number of units. When the discharge was 12,000      Figure 7 shows daily discharge of the Chinese sturgeon during the spawning day downstream of Gezhouba Dam. The highest discharge of the first spawning was 27,290 m 3 /s on October 15, 1990, and the lowest discharge was 5810 m 3 /s on November 23, 2009. The highest discharge of the second spawning was 18,170 m 3 /s on November 1, 2000, and the lowest discharge was 5590 m 3 /s on December 2, 2012. Since the spawning date gradually became delayed, the discharge of the first spawning showed a downward trend overall. The spawning discharge was less than 12,000 m 3 /s after 2002. Most spawning dates featured discharges above 12,000 m 3 /s before 2002, which accounted for 75%; discharges above 15,000 m 3 /s accounted for 55%, and discharges above the design discharge of 17,930 m 3 /s accounted for 25%. The second spawning had a lower discharge than the first spawning; discharges above 12,000 m 3 /s accounted for 52.9%, those above 15,000 m 3 /s accounted for 17.6%, and a discharge above 17,930 m 3 /s occurred only once on November 1, 2000.    www.nature.com/scientificreports/ from Gezhouba Dam to Xiaoting, and spawning mainly concentrated in the approximately 12 km reach between Gezhouba Dam and Yanzhiba Islet 6 . During 1996-2007, the spawning ground was the main channel of Yangtze River from Gezhouba Dam to approximately 2 km upstream of Yanzhiba Islet, and the main spawning site was within the approximately 4-km reach from Gezhouba Dam to Miaozui 25,26 . The spawning area could be divided into two parts: upstream spawning area and downstream spawning area; the spawning frequency and scale of the downstream spawning area were obviously larger than those of the upstream spawning area 12,26 . Because the spawning date was mainly concentrated in October, the spawning discharge was high, the suitable area of the upstream spawning area was small, and the upstream spawning area did not feature favourable locations for Chinese sturgeon to perch; thus, the Chinese sturgeon primarily selected to spawn in the downstream spawning area.

Case no Discharge (m 3 /s) Opening mode of units Discharge of each unit (m 3 /s) Proportion of suitable area (%)
All natural reproduction of Chinese sturgeon has occurred in the upstream spawning area since 2008 27,28 , which was also the main research area of this paper. Since 2008, the natural reproduction date of Chinese sturgeon has been postponed to middle and late November, or even early December, when the discharges were less than 10,000 m 3 /s. The suitable area of the upstream spawning ground was large. Because Chinese sturgeon migrate as far upstream as possibly for reproduction, they selected to reproduce in the upstream spawning area. The change of spawning ground may be due to changes in discharge.
Factors affecting spawning of the Chinese sturgeon. According to current research, the riverbed topography, bottom substrate, velocity, water temperature, water level, discharge, sediment content and other factors are thought to affect the spawning of Chinese sturgeon. Some researchers emphasized the important role of the water level 22 . Other researchers suggested that the changes in riverbed bottom substrate might have caused positional changes in the critical spawning ground of Chinese sturgeon 27,29 . Some researches indicated that the discharge and water temperatures were necessary conditions for Chinese sturgeon spawning and hatching 21 . Some researchers believed that the delay in the decrease in water temperature caused by the Three Gorges Reservoir and low numbers of reproductively mature individuals have contributed to the failure in natural breeding 30,31 .
Although many variables may contribute to the quality and quantity of the spawning habitat, we focused on the velocity as a key metric because of the direct effect of the velocity on the fish spawning habitat. In addition, the flow velocity can be managed through changes in reservoir operation and the opening mode of dam units. In contrast, the temperature, turbidity, and substrate are difficult to manipulate through management actions. The multi-variate nature of the habitat implies that velocity manipulation alone may not be sufficient, and the operation would not be effective unless combined with water temperature and sediment factors, so other factors should also be evaluated in the future because they may work together.

Conclusions
Based on the field monitoring results of 2016-2019, the FLOW-3D model was used to simulate the flow field of monitored sturgeon signals, and it was concluded that the preferred velocity range for Chinese sturgeon was 0.6-1.5 m/s. Under a given discharge, the suitable-velocity area was maximal when all units of the Dajiang Plant of Gezhouba Dam were open, and the conditions were more favourable when the units on the left side were open. When the discharge was less than 12,000 m 3 /s, the proportion of suitable area slightly fluctuated at approximately 96.2%; when the discharge was 12,000 m 3 /s, the suitable area rapidly decreased with increasing discharge. Moreover, for different opening modes at high flows, at discharges of 12,000-13,000 m 3 /s, opening 11-12 units on the left side was the best; at a discharge of 14,000 m 3 /s, opening 12-13 units on the left side was the best; when the discharge reached 15,000 m 3 /s, opening 14 units was the best. The optimal scheme for the opening mode of the units at different discharges was analysed, and the results provide new ideas for the protection and ecology conservation of Chinese sturgeon.

Methods
Study area. Field surveys have shown that the only known spawning ground is located in the section between Gezhouba Dam and Miaozui, which is approximately 4 km downstream of Gezhouba Dam 12 . Therefore, the area between Gezhouba Dam and Miaozui was selected for investigation in this study, as shown in Fig. 9. Figure  This area was divided into several cross-sections (Fig. 9c), and the velocity of the cross-section was measured with a 300-kHz acoustic Doppler velocity profiler (ADCP). In addition, acoustic imaging sonar monitoring was performed in this area. According to the monitoring results from 2016 to 2019, most of the Chinese sturgeon signals appeared within 700 m below the Dajiang Plant units of Gezhouba Dam, as shown in the red box in Fig. 1c. Moreover, according to the field investigation, the area of the spawning ground further decreased in recent years because most of the spawning behaviour of Chinese sturgeon has occurred in the red box in Fig. 9c since 2008 27,28 . Hence, the range of 700 m downstream of the Dajiang Plant units was the key area in this study. This area is shown in Fig. 9d, which shows a bathymetric map of this area, and the colour shading and contours represent the water depth when the water level downstream from the dam was 41. Numerical model. In this study, we used numerical model FLOW-3D, which is a commercial CFD package based on the finite volume method (FVM) that solves the Reynolds-averaged Navier-Stokes equations 32 . It can effectively estimate the flow structure and velocity distribution in different water layers 33 .
Boundary and initial conditions. The upstream boundary condition used the known discharge based on releases from Gezhouba Dam. The pressure boundary was used for the dowstream boundary and set to the water level. The water surface was a free surface, using a pressure boundary, given standard atmospheric pressure. The wall boundary was used for the solid boundary of the bottom and both sides. The initial condition was the water level, and the initial velocity was 0.
Mesh construction. A hexahedral orthogonal grid was used to mesh the model, which can iteratively define a base mesh to fit the surface geometries. The finite volume method was used to discretize the governing equation, and the GMRES algorithm was used to solve the equations 34 . The mesh sizes were selected to respect the requirements of the grid convergence index (GCI) method to test the spatial convergence 35 . The X-axis direction and Y-axis direction mesh sizes were 3-8 m, and the Z-axis direction mesh sizes were 1-2 m.  Fig. 10. From Fig. 10, the distribution of flow velocity of each cross-section was in good agreement, especially in the Dajiang River area, where the latest spawning ground of the Chinese sturgeon was located. The errors of the model and measured values were generally less than 0.2 m/s, and the maximum error was 0.43 m/s, which appeared next to the dividing dike in cross-section 3. The two-tailed t-test permutation of the model and measured values showed no significant difference: P = 0.45 > 0.05. Therefore, the model simulation was reasonable and acceptably simulated the water flow characteristics of the spawning ground of Chinese sturgeon.
Acoustic monitoring. Acoustic imaging sonar is a fast and effective method to study Chinese sturgeon because it can monitor the number and distribution of fish without approaching and harming the fish 36 . For acoustic monitoring, this paper used a Dual-Frequency Identification Sonar (DIDSON) system, which is widely used in fishery management, structural detection, pipeline leakage identification, underwater monitoring, underwater searching, underwater security inspection, etc. 37 . The main monitoring area was approximately 4 km long between Gezhouba Dam and Miaozui. During the investigation, the sonar transmitter was fixed to the side of the survey vessel and located 0.3 m below the water surface. A GPS device produced by the Garmin company was used for navigation and positioning. We continuously performed monitoring every day from October to January of the following year for 3-4 h with a zigzag survey pattern to ensure full coverage of the spawning ground. The monitoring results were saved in the form of video images by acquisition software (DIDSON V5.25), and Chinese sturgeon signals were confirmed by measuring the full length, swimming behaviour, and body shape. To reduce the error of judgement and obtain highaccuracy Chinese sturgeon signals, each monitoring signal was confirmed by at least two different researchers.

Data availability
Data are available from the corresponding author upon request.