The Pearl River is the largest river in southern China, with a total length of 2,214 km and a mean annual discharge of 3.3 × 10¹¹ m³. The sampling location, i.e., Zhaoqing section, is in the lower reach of the Pearl River, about 180 km upstream from the estuary (Figure 1). The annual average air temperature, annual rainfall, and annual discharge of this section are about 22.0°C, 1,644.5 mm, and 6,810 m³/s, respectively (Lu, 1990). There are 76 fish species recorded in the lower reach of the Pearl River, excluding the estuary area (Zhang et al., 2020). Over the past century, 18 spawning grounds of silver carp have been reported in the middle and upper reaches of the Pearl River (Committee of Pearl River Fishery Resources Investigation, 1985), and approximately 26 dams had been built near the spawning grounds by 2007. Our long-term monitoring results (unpublished data), along with a recent study (Gao et al., 2020), indicate that only the spawning grounds in the lower reaches of the tributary Liujiang River and the Laibin–Pingnan section of the mainstream still exist after 2007.

For our samples, the spawning grounds are located in the Guiping–Pingnan section, estimated by the development periods of captured larvae and water velocity (Pearl River Water Resources Commission, unpublished data). Eleven dams have been constructed in the mainstream of the Pearl River. All these dams were completed before 2005 and are located at the upper reaches, except for two downstream dams: Changzhou Dam and Datengxia Dam (Chen et al., 2018). The Changzhou Dam is a low head dam (height < 30 m, highest water level = 16 m) built in 2007 with only a daily discharge regulation. The Datengxia Dam began to impound in 2020, with a maximum water level of 61 m expected in 2023. Taken together, all large dams were located in the headwater and upper reaches of the Pearl River before 2019, far away from the current silver carp spawning grounds. Therefore, the impoundment would not influence the water temperature in the current spawning grounds during the study period (2006–2019).

We sampled three times a day (06:00–08:00, 13:00–15:00, and 19:00–21:00) every other day throughout the year. Larval fish were collected using a pyramid trap net (mesh size = 0.5 mm, length = 6 m). The net had a rectangular mouth (width × height: 1.5 m × 1.0 m), and the end opened to a larval collection cage (0.8 m × 0.4 m × 0.4 m) (Shuai et al., 2018). A cup-type flowmeter (LS45-2) was fixed at the net mouth center to measure the water volume flowing through the net. The net was submerged in the water with mouth open and facing the river flow. The net position would change slightly with the rise and fall of the water level, but it was kept 10–15 m away from the bank, where the flow velocity was approximately 0.3 m/s. Sampling was initially conducted at S1 (23°2′40″ N, 112°27′5″ E) from 2006 to 2013. But the location was changed to S2 (23°9′54″ N, 112°16′58″ E, about 28 km upstream of S1) from 2017 to 2019, because cargo ships anchored near S1 (Figure 1). No samples were collected from 2014 to 2016.

Larvae samples were kept in a 5% formalin solution immediately after collection. The larvae of silver carp were identified (based on morphological characteristics) and counted by an experienced researcher in the laboratory (Yi et al., 1988). The abundance of larvae was calculated as the number of larvae per 24 h. The collected silver carp larvae were mainly at two developmental stages: the emergence of air bladder and one-chamber air bladder. Spawning dates of larvae were estimated using back-calculation methods from the captured dates (Yi et al., 1988). The spawning date of larvae was defined as the day of spawning occurrence.

We collected daily water temperatures and discharges at the Dahuangjiangkou (DHJK) gauge station to represent the environmental conditions of the spawning ground from 2006 to 2019 (Figure 1). Daily water temperature was collected by an automatic water temperature monitor, which recorded the temperature at least four times a day. Daily river discharge (Dis) was obtained from the Pearl River Water Resources Commission.

We back-calculated the earliest birth date of the captured larvae, date of the spawning peak, and the end date of spawning by subtracting their development periods from the capturing dates of the larvae. These variables were used to describe the spawning period across our study. A previous study reported that the minimum discharge required to observe the occurrence of silver carp larvae was 10,000 m³/s in the Pearl River (Shuai et al., 2018). Thus, we used the first date of the discharge increased to 10,000 and 13,000 m³/s to index the possible flow-related reasons for changes in spawning timing. The minimum water temperature for silver carp spawning was 18°C (Yi et al., 1988). We counted the number of days with a temperature not lower than 18°C (hereafter, N_18) prior to the spawning season (from January to March). The threshold temperature for initial gonad development of adult silver carp was 15°C. We thus regarded the date when the water temperature reached 15°C during spring as the start date of stage IV (Wang et al., 2014). The end date of stage V denoted the date when larval abundance reached 5% of the total observed annual abundance. The cumulative temperature from stages IV to V was calculated by the sum of daily water temperature from stages IV to V in the year. The above variables and the mean monthly water temperature of the spawning ground were used to describe the changes in water temperature of spawning and cumulative temperature. Using the day of year calendar, the date is a sequential day number starting with day 1 on January 1.

The discharge suitability index (DSI) and the temperature suitability index (TSI) were applied to define the discharge and temperature suitability for silver carp’ spawning (Yu et al., 2018). The DSI was calculated as DSIi=DiDm⁢a⁢x, where DSI_i is the discharge suitability index of the discharge bin i, D_i is the value of larvae abundance weighted by the number of spawning occurrences of the discharge bin i, D_max is the maximum value of larvae abundance weighted by the number of spawning occurrences across all discharge bins. The TSI was calculated as TSIj=TjTm⁢a⁢x, where TSI_j is the temperature suitability index of the temperature bin j, T_j is the value of larvae abundance weighted by the number of spawning occurrences of the temperature bin j, T_max is the maximum value of larvae abundance weighted by the number of spawning occurrences across all temperature bins.

All datasets of discharge, temperature, and abundance of larvae have normal distributions. The analysis were performed using Shapiro–Wilk tests in the “stats” package in R 3.6.1 (R Core Team, 2019). Pearson correlation analysis was used to determine the statistical significance of correlations between the spawning timing, discharge increase, and monthly temperature variables. The correlation analysis was performed using SPSS 13.0. Data for daily temperature (from January to March, August to September) was converted to weekly time-series for trend analysis using the “stats” package in R. The structural change in a time-series was detected by the chow test/F statistic using the “strucchange” package in R to plot the corresponding p-values (Zeileis et al., 2002). Differences were considered statistically significant at p < 0.05.

The silver carp started to spawn when the river discharge reached 2,310 m³/s (Figure 2A). The occurrence of spawning gradually increased with increased discharge before 7,000–9,000 m³/s, and then decreased (Figure 2A). However, the highest discharge suitability index (DSI) was observed at a discharge of 13,000–15,000 m³/s (Figure 2A). Moreover, there was a relatively lower peak of DSI at a discharge of 19,000–21,000 m³/s (Figure 2A).

The lowest temperature observed for the occurrence of silver carp spawning was 20.7°C (Figure 2B). The occurrence of spawning gradually increased with increasing temperature before 27–28°C, slowly declined after that, and then plummeted (Figure 2B). The temperature suitability index (TSI) stayed at a low level when the temperature was lower than 22°C, then sharply increased to the peak at 25–26°C. Then TSI rapidly declined and remained at a relatively stable level at temperatures of 26–28°C after that; finally, it sharply decreased at temperatures above 28°C (Figure 2B).

Spawning environment suitability of fish in rivers is determined by hydrodynamic, geomorphological, and biological interactions that generate complex and localized linkages (Alvarez-Mieles et al., 2019). The same species distributed over a variety of environmental conditions could probably demonstrate different ecological and hydrological requirements for spawning. The silver carp is distributed all over the world and has high local adaptation in reproduction characteristics. The initiation of silver carp spawning without changes in discharge was noted in the Wabash River, but the occurrences of eggs and larvae of this fish seemed to be triggered by flood pulses in the Yangtze River and Pearl River (Coulter et al., 2016; Shuai et al., 2018; Yu et al., 2018). The most suitable discharge and temperature of silver carp’ spawning habitats in the middle and lower reaches of the Yangtze River were 15,000–21,300 m³/s and 21–24°C, respectively (Yi et al., 1988; Yu et al., 2018). In contrast, the current study demonstrated that spawning in the Pearl River was initiated when discharge was much lower, approximately 2,310 m³/s, indicating that the fish can start to spawn without significant increases in discharge. The most suitable discharge for silver carp’ spawning in this study (13,000–15,000 m³/s) was also lower than that in the Yangtze River. The distance between the middle and lower reaches of the Yangtze River and those of the Pearl River is approximately 800 km, and there is a relatively high genetic distance between these two silver carp populations (Ji et al., 2009). Therefore, different hydrological requirements for silver carp spawning in those two large rivers in China could probably attribute to local adaptation.

Our study provides some new insights about the mechanisms of how river discharge affects the spawning characteristics of silver carp. First, our findings suggest that the spawning peak of silver carp, rather than the initial spawning, is probably triggered by flow increase in the Pearl River. This was evidenced by the fact that the start date of spawning was not significantly correlated with high discharge (Ddis_1 and Ddis_13), while the spawning peak was significantly and positively correlated with high discharge (Ddis_1). Second, we showed that spawning occurrence of silver carp occurred at a relatively low discharge level, but with relatively lower daily larvae abundance. This finding is likely because only part of the mainstream is suitable for fish spawning when the discharge is at a low level. Then as the discharge gradually increases, some hydro-fluctuation zones become suitable for spawning, which explains why the highest average daily larvae appeared at high discharge. Flow pulses can probably create more suitable spawning niches for fish.

We indicate that the minimum and most suitable temperatures for silver carp’ spawning in the Pearl River were higher than those in the Yangtze River (Yi et al., 1988). Temperature can both affect the development of gonad and trigger spawning of fish (Morgan et al., 2013). In the present study, silver carp spawned at 20.7–30.1°C in the Pearl River, while they spawned at 18–28°C in the Yangtze River (Chen et al., 2009). Fish are very capable of finding habitat patches with suitable spawning temperature (Górski et al., 2010), which might explain why silver carp can spawn at a wide range of temperatures. The lowest temperature for silver carp spawning is 18°C indicating that water temperature is not a limiting factor in the Pearl River where average annual water temperature in spawning grounds is approximately 23°C (Yi et al., 1988). Although there is a wide range of temperatures for silver carp spawning in the present study, the most suitable temperature was closely related to the flood season.

We showed that the spawning period of silver carp has been gradually shortened, given the start date of spawning fluctuated with a slight tendency to delay, while the spawning peak and end date obviously occurred earlier over the years. The start date of spawning of silver carp in the Pear River was largely regulated by water temperature, and warmer temperature prior to the spawning season can delay the spawning (Warren et al., 2012). The temperature-related changes in spawning timing are evident in a large number of freshwater fish species. For example, elevated water temperatures caused earlier spawning of rainbow smelt (Osmerus mordax) in the Great Lakes (O’Brien et al., 2012). Warmer temperatures prompted earlier phenology that led to 17 species of larval fishes occurring earlier in the year in southern California (Asch, 2015). In the Yangtze River, a water temperature decline of 2–4°C between March and May resulting from the Three Gorges Dam impoundment led to a delay in silver carp spawning (Wang et al., 2014). However, the rising temperature can also delay fish spawning. It has been reported that the elevated summer temperature caused delayed spawning of brook trout (Salvelinus fontinalis) and reduce spawning activities in Rock Lake (Warren et al., 2012). Likewise, in our study, the results suggest that rising temperature from January to March (prior to the spawning season) may delay the spawning start date and decreases the occurrence of spawning. Given the uncertainty between 2014 and 2016 and the limited time-series points, our analyses and conclusions on the trend of Dspawn should be viewed with caution.

We indicated that water temperature increases before the spawning season may be the principal reason for the initial spawning delay in the present study. However, an observed delay in silver carp spawning in the Yangtze River was caused by water temperature declines before spawning (Wang et al., 2014). This could attribute to the reasons that when the mean temperature before the spawning season is lower than 15°C (threshold temperature for gonad development of silver carp), the increase of temperature may favor body growth and promoted gonad development (Yi et al., 1988). However, when the mean temperature substantially exceeds 18°C (threshold temperature for silver carp spawning), we speculated that the elevated temperature would increase metabolism, diverting energy from the fat accumulation, which is needed to complete gonad development, ultimately causing spawning delay (Luksiene and Svedang, 1997; Pankhurst and Munday, 2011). The mean water temperature before and during the spawning season in the spawning ground (DHJK) in the Pearl River were approximately 18 and 26°C, respectively. However, in the spawning grounds of silver carp in the Yangtze River, the mean water temperatures before and during the spawning period were approximately 13 and 21°C, respectively (Wang, 2016). These differences in ambient temperatures might be the principal reason for the divergence in warming effects on the onset of spawning. Although temperature-dependent adaptation allows parents to adjust the spawning time so that the larvae match the timing of the food source (Neuheimer et al., 2018), most studies consider that temperature-related spawning time variations can be attributed to changes in effective accumulated temperature (Robinson et al., 2010; Chezik et al., 2014; Hansen et al., 2017). The initiation of spawning in the present study did not significantly correlate with cumulative temperature. Although cumulative degree-day is an important factor affecting spawning initiation, it is not the main limiting factor in a subtropical river with high temperatures (Wang et al., 2014; Hansen et al., 2017).

The average temperature in August and September was significantly and negatively correlated with the end of spawning, indicating that high temperature in August and September promotes the earlier termination of silver carp spawning. The mean water temperature in the Pearl River between August and September is approximately 28.6°C, which is higher than mean temperatures in other months. The optimum temperature for embryo development of silver carp is 22–28°C (Chen et al., 2009). Therefore, it is likely that silver carp in the Pearl River complete spawning early to avoid the adverse effects of rising temperature on embryo development in a freshwater system with limited thermal refugia (McCullough et al., 2009; Warren et al., 2012).

We revealed that the temperature before the spawning period (from January to March) and near the end of spawning (August and September) were increased during the study period. According to the relationships between the timing of spawning and thermal regime, this warming trend was likely the principal reason for the decreasing trend of the spawning time of silver carp in the present study. Given there was no large dam near the spawning ground during the sampling period, the temperature variation could not be attributed to dam impoundment. It has been reported that the Pearl River basin has been undergoing air temperature increases since the 1950s caused by climate change (Zhang et al., 2012; Tian and Yang, 2017). Therefore, the upward trend in the water temperature in the present study is most likely due to global warming.