The Xiaolangdi–Gaocun reach in the lower part of the Yellow River is selected as the study area with a total length of 337 km (shown in blue in the Figure 1). The reach is wide and shallow, with dense sandbars and variable water flow, and the broad beaches along the river are rich in food organisms. These conditions provide fish with foraging and spawning habitats, making them ideal for studying the ecology of rivers under the dam. The Yellow River Carp National Aquatic Germplasm Resource Reserve (YRCNAGRR) in the Zhengzhou section of the lower Yellow River was established in 2008 and chosen as a part of our study area (shown in gray in the Figure 1). Its main protection objects are the Yellow River carp and its spawning grounds, feeding grounds, and wintering grounds, as well as the aquatic ecology and terrestrial ecosystems (Zhang and Han, 2016), among which the Huayuankou reach is one of the important spawning grounds. For our study area, it is necessary to select a sufficiently long enough reach downstream of the dam as the study area in order to explore the influence of the dam on the different reaches. Meanwhile, there should be many relatively stable spawning grounds in the study area, which could make it a representative reach of the lower part of the Yellow River for studying fish and their habitats. The natural hydrological cycle of the river has changed since the construction of the Xiaolangdi Dam at the end of 2001, thereby affecting the quality and scale of aquatic habitats and the breeding activities of fish in the study area.

In simulating the physical characteristics of river flows, one-dimensional hydrodynamic modeling can be used to calculate the hydrological and hydrodynamic parameters of different sections at various times and, therefore, provide data to gauge fish habitat suitability. In this study, a one-dimensional hydrodynamic model of the lower Yellow River was constructed based on MIKE11 software, which could simulate the vertical homogeneous water flow of the river according to the terrain and the boundary of the water flow. The Saint-Venant equations and the Abbott–Ionscu 6-point implicit scheme are applied for the numerical simulation. Thus, the governing equations for the one-dimensional hydrodynamic model are as follows: B ∂ Z ∂ t + ∂ Q ∂ x = q , (1) ∂ Q ∂ t + ∂ ∂ x Q 2 A + g A ∂ Z ∂ x + Q Q K 2 = 0 , (2) where x (m) is the distance along the river; t (s) is the time; Z (m) and Q (m³/s) are the water level and water flow of the section, respectively; A (m²) is the cross-sectional area; q (m²/s) is the lateral inflow per unit length; B (m) is the river width; g (m/s²) is gravitational acceleration; and K (m³/s) is the flow modulus.

The data files of the MIKE11 hydrodynamic model mainly include files of sections, river networks, boundaries, and model parameters. The HD (hydrodynamic) module is the basic calculation module of MIKE11, and its module parameters are mainly numerical and physical parameters. Numerical parameters are related to the iterative solution of the equations, such as the number of iterations and iterative calculation accuracy. The physical parameter is mainly the roughness of the river bed. There are 152 measured post-flood sections in the study area. After considering that the two tributaries in the study area have little influence on the model simulation results, no tributaries are added to the one-dimensional hydrodynamic model. The measured flows at the Xiaolangdi Hydrological Station and water levels at the Gaocun Hydrological Station are used as the upper and lower boundary conditions, respectively. The time step of the hydrodynamic model is 1 min. We adjust the roughness of the river bed as model calibrations to improve the simulation accuracy. The outputs of MIKE11 are hydrological elements, such as the water level, flow, and velocity at each moment for 152 sections of the study area.

The habitat suitability index (HSI) is for evaluating the suitability and quality of wildlife habitats, and it is the biological basis of habitat simulations (Boavida et al., 2015; Morales-Marín et al., 2019; Sharma et al., 2022). Habitat suitability curves present the relationship between a species’ growth and its influencing factors (Cho et al., 2012; Knudson et al., 2015), which is generally determined by the appearance frequency and distribution of the species. The HSI varies from 0 to 1 to quantitatively indicate the habitat suitability. In this study, the Yellow River carp is selected as the indicator species for the habitat in the lower reaches of the Yellow River. For the breeding period (including the parent fish spawning period and the larval fish development period), four important habitat factors are selected to determine the habitat suitability of the Yellow River carp. They are water depth h, flow velocity v, water temperature t, and water-level rise dz, and their definitions are as follows.

(1) Flow velocity v: The Yellow River carp is a kind of benthic omnivorous fish and mostly inhabits on the soft riverbed or in water with plentiful aquatic plants. It can lay sticky eggs in various water bodies but prefers slow-flowing or still water with aquatic plants. Its hatching time is 3–7 days in general. Then, its larval fish usually requires a stable water flow for 3–4 days to stay alive. The suitable flow velocity for larval development is in the range of 0–0.15 m/s (Wang et al., 2020).

To summarize, the four different habitat factors of parent fish and larval fish are obtained, and its suitability curves of each single index are represented by step functions (as shown in Figures 2, 3). Then, the parent fish spawning suitability index (HSI _pf ) is formulated of water depth h, flow velocity v, and water-level rise dz by Eq. 3, and the larval fish development suitability index (HSI _lf ) is formulated of the flow velocity v, water depth h, and water temperature t by Eq. 4. In this paper, we propose the Habitat Suitability Index HSI of the Yellow River carp by the mean values of HSI _pf and HSI _lf in Eq. 5 as follows. H S I p f = S I h + S I v + S I d z 3 , (3) H S I l f = S I h + S I v + S I t 3 , (4) H S I = H S I p f + H S I l f 2 , (5) where H S I p f ,   H S I l f ,   a n d   H S I represent the parent fish spawning suitability index, larval fish development suitability index, and habitat suitability index of the Yellow River carp, respectively, and S I h ,     S I v ,   S I t , and S I d z represent the water depth index, flow velocity index, water temperature index, and water-level rise index, respectively.

We have the water temperature t, and we calculate the water depth h and flow velocity v of the river cross section using the one-dimensional hydrodynamic model. Then, the values of S I h , S I v , S I d z , and S I t can be pointed out according to Figures 2, 3. It can be seen that even under the same hydrological conditions, the suitability of parent fish and larval fish is different, and the comprehensive habitat suitability index HSI is obtained by Eqs 3, 4, 5. Different HSI values represent the different suitability of the habitat (Table 1) (Jiang et al., 2019; Tang et al., 2022). We find that all river suitable sections with a HSI value greater than .5 are suitable for carp’s reproduction, according to Table 1. The reach between these two continuous suitable habitat sections is defined as a suitable spawning reach. The reach length is defined as L _sr , and the average HSI value of up and down reach sections is defined as HSI _sr . L _sr and HSI _sr are used to describe the scale and quality of a spawning reach, respectively.

In this study, the habitat suitability index is used to build the habitat suitability model of the Yellow River carp. Furthermore, it is combined with the one-dimensional hydrodynamic modeling for habitat factors to define the habitat suitability and assess the suitability of the habitat.

The Nash–Sutcliffe efficiency (NSE) and the root mean square error (RMSE) are selected to evaluate the accuracy of the hydrodynamic model. The NSE is a statistical parameter that describes the fitting degree between simulation and measurement. The more it closes to 1, the better the simulation of the hydrodynamic model is. The RMSE represents the deviation between the simulated and the measured value. The more it closes to 0, the smaller the model error is.

The flood from August 16 to September 11 in the year 2018 was chosen to calibrate the riverbed roughness in the study area, and another flood from September 15 to October 10, 2018, was chosen to verify the hydrodynamic model. If the model can well simulate the flood process, it shows that the model has good applicability at all water levels and can also well reflect the hydrodynamic characteristics during the spawning period.

Three hydrological stations (Xixiayuan, Huayuankou, and Jiahetan) were selected to calibrate and verify this model. The range of riverbed roughness in this study was determined as .020 to .032. The water-level calibration and verification results of the three hydrological stations are shown in Figure 4, which well-fitted the water flow under different incoming waters and captured the process of water-level fluctuation, such as the rise and fall in 8.28–9.4. We listed the NSE and RMSE in Table 2 for the results of Figure 4. The minimum NSE for calibration and verification was .769, and the maximum RMSE was .161 m. This means that the model can accurately reflect the process of hydrodynamic changes and can be used for the construction of habitat models.

April to June is the spawning period of the Yellow River carp every year. It is the most critical and vulnerable stage for the Yellow River carp to spawn and develop. To breed successfully, the Yellow River carp has extremely high requirements on the ecological and hydraulic conditions of the habitat during its spawning period. There are 16 spawning grounds marked by stars (Figure 5), which are measured spawning locations in the Henan section of the Yellow River studied by Hui et al. (2019). To verify the simulation of the habitat model, the habitat factors from April 1 to June 30 of the year 2014 were chosen. The daily HSI values of 152 sections were calculated by these habitat factors, according to Eqs 3, 4, 5, and then, the daily average HSI values of each section from April to June were obtained. There were 19 suitable spawning reaches shown in red lines (Figure 5). Their size L _sr and spawning habitat quality HSI _sr are displayed in Supplementary Table S1. Most of the 16 measured spawning grounds studied by Hui et al. (2019) (black stars) fell within the spawning reaches (red lines) simulated by the habitat suitability model (Figure 5). This indicates that the habitat suitability model of the Yellow River carp can simulate accurately the positions of spawning habitats. For a measured spawning ground, its HSI value is the HSI value of the nearest section among 152 sections. 87.5% of the 16 measured spawning grounds had their HSI values greater than .5 except grounds 7 and 10 (Table 3). The spawning area is simulated by the average value of HSI from April to June; the measured spawning grounds only need to meet their reproductive requirements for a long enough time in April–June rather than the whole 3 months, so the model simulation results have a certain deviation from the measured spawning ground. These spawning grounds are generally suitable for the Yellow River carp, which is consistent with Hui’s findings (Table 1). This indicates that the habitat suitability model of the Yellow River carp has a good simulation effect and can well represent the quality of spawning habitats.

Based on the one-dimensional hydrodynamic model and the habitat suitability model of the Yellow River carp, the simulation of the habitat suitability before the dam construction (1980–1990) and after the dam construction (2006–2018) shows that the habitat suitability index reduces generally (Figure 6A), and the scale of the Yellow River carp spawning reaches in the lower reaches of the Yellow River decreases after the dam construction (Table 6; Figure 9).

The impact of the construction and operation of the Xiaolangdi Dam on the river ecology is two-fold. The Xiaolangdi Reservoir has changed the hydrological regime of the river downstream. The notably changed flow velocity, water levels, water-level rises, and water temperatures result in a decline in the quality of the Yellow River carp reproduction and a reduction in the size of the spawning reaches. The release of the lower-temperature bottom water from the reservoir has a direct impact on the development and reproduction of fish downstream of the dam. Furthermore, the dam has blocked channels used by migratory species for feeding and reproduction, with sharp declines for these fishes. For instance, some fishes in the Yellow River, like the migratory Coreius septentrionalis (Nichols) and Coilia nasus, are now endangered (Han et al., 2013).

However, the completion and operation of the Xiaolangdi Dam also plays a role in improving the ecological environment to some extent. It now ensures a continuous river flow downstream and increases the guarantee rate of ecological flow (Gao et al., 2014). Meanwhile, the sediment from the reservoir constantly scours the riverbed downstream, the floodplain flux is increased, and the river morphology tends to be stable. All of these changes provide a more stable habitat for organisms. There are complex interactions among these factors which affect the fish growth, and the responses of fish are also unclear, which makes it more difficult to establish clear correlations between fish habitat factors and habitat suitability. Deep research on the impact of multiple factors on fish habitats is still merited.