© EDP Sciences, 2017

1 Introduction

The grass carp Ctenopharyngodon idellus (Valenciennes), is a fish species of the Cyprinidae family native to Eastern Asia (Guillory and Gasaway, 1978). Adult individuals can be found in lakes, ponds, pools, and the backwaters of large rivers (Shireman and Smith, 1983). It prefers large, slow-flowing, or standing water bodies and tolerates temperature ranging from 0 to 38 °C (Froese and Pauly, 2017). It is a typical herbivorous species worthy of commercial cultivation because of its proficiency in consuming phytoproteins and carbohydrates (Wang, 2000; Froese and Pauly, 2017). As one of the major freshwater species in the Yangtze River, its production accounted for 25% of all freshwater fish production in 2011 in China. The natural resources of C. idellus were abundant in the Yangtze River before the 1960s (Liu et al., 2004), but water pollution, habitat degradation, and over-exploitation have caused wild fish resources to decline sharply since the 1990s (Gui, 2003). Since 2002, an annual program for supplementing C. idellus was implemented in the Yangtze River to restore and maintain this fish resource, but the potential consequences remain unknown.

Hatchery-based restocking programs are well known and widely implemented worldwide to conserve wild fish populations (Fraser, 2008; Kostow, 2009). It is estimated that over 300 fish species were released worldwide every year (Welcomme and Bartly, 1998). Although habitat restoration should always be the first choice when possible, rearing fish in hatcheries and releasing them into the wild can be useful in maintaining sensitive natural populations along with sustainable fish production (Einum and Fleming, 2001; Araki et al., 2007; Johnsson et al., 2014). As young fish that enable to survive the early stages of life are also likely to survive to adult size, many endangered species recovery programs rely on the release of hatchery-reared individuals to ensure long-term population viability (Brown and Day, 2002).

Although hatchery rearing techniques are progressing, hatchery-reared fish often display behavioral deficits and suffer high post-release mortality rates in the wild (Brown and Day, 2002). Indeed, behavioral traits in particular are modified by hatchery conditions. Fish living in the natural environment are exposed to complex habitats and natural challenges (Armstrong et al., 2003). In contrast, hatchery conditions eliminate potential stressors and modify the experiences and learning conditions of reared fish. Matsuzaki (2009) and colleagues found that feral strains of the common carp Cyprinus carpio (Linnaeus) behave differently to domestic strains. Of the two, the feral strains are better at detecting prey and are more cautious of predator attacks (Matsuzaki et al., 2009). Apart from altering behavior, hatchery programs have been documented to impact the genetic and ecological structure of wild C. idellus (Liu et al., 2009; Zhao et al., 2011).

Considering hatchery shortfalls, Chinese fisheries scientists have had concerns about the potential impacts of stocking programs on fish conservation for more than a decade (Shen, 2002; Chen, 2003; Gui, 2003). A great number of studies have investigated the impacts of hatcheries on salmonids' behavioral traits. By comparison, studies of cyprinids are scarce, especially those focused on C. idellus, despite its considerable biological and economic importance.

In this study, we compared the behavioral responses of hatchery-reared and wild C. idellus to predators after visual and olfactory contact. We also compared their survival rate after being placed in contact with an actual predator. With this study, we hope to shed light on potential ecological risks caused by hatchery-reared C. idellus in stocking programs and to improve hatchery conditions to better match natural conditions.

2 Material and methods

2.1 Experimental fish

In this study, hatchery-reared (n = 160) and wild (n = 160) C. idellus were used. Wild C. idellus fry were caught in the Yangtze River in China (Honghu, Wuhan; 114.02 E; 30.08 N) using floating trap nets. Fry were collected instead of juveniles because wild C. idellus juveniles are sparse in the Yangtze River, making it difficult to catch enough juveniles for this study. Hatchery-reared C. idellus, F₄ generation, were obtained from a local hatchery. Ancestors (P₀) of hatchery-reared fish were wild C. idellus caught from the Yangtze River ranging from Honghu (114.02 E; 30.08 N) to Huanggang (114.87 E; 30.45 N). The ancestors' offspring (generations F₁–F₄) were hatchery-reared (Fig. 1). The F₁ generation was directly obtained from 60 to 80 pairs of wild brood fish. These brood fish were selected from more than 1000 wild-caught P₀ C. idellus. About 1000 F₁ individuals were reared to maturity in captivity, and 60–80 pairs of F₁ were then selected as brood fish for F₂. Analogously, the F₃ and F₄ generations were acquired. This rearing method obtained hatchery-reared F₄ individuals with great genetic diversity (Wang, 2000). In the hatchery, brood fish (10–15 kg in body weight, 300–450 ind·ha⁻¹ in density) were reared in ponds 1.5–2.0 m in depth, and the female:male ratio was 1:1.2. On a daily basis, they were provided with commercial feed, aquatic plants (common duckweed Lemna minor), wheat seedlings, and vegetables. Artificial reproduction began when the water temperature rose to 23–26 °C, between April and June. Brood fish were injected with luteotropin releasing hormone analogue at 10 μg·kg⁻¹ for females and 5 μg·kg⁻¹ for males. Oocytes were fertilized using the dry method for 2 min. The fertilized eggs were then transferred to the fry incubator for 4–5 days until hatched.

Both hatchery-reared F₄ fry and wild fry were then transferred into respective outdoor ponds in similar conditions. The fry were reared at a density of 50 ind·m⁻², and they fed primarily on natural zooplankton during the early phase (30 days) of culture. During the later phase (30 days), artificial food pellets (36.0% protein, 12.1% lipid, 11.3% ash, and 18.2 J·mg⁻¹ energy) were provided. After two months, their mean (±S.E.) total length (L_T) was 6.21 cm (±0.02 cm) and mass was 2.35 g (±0.02 g), and they were transferred to indoor tanks for acclimation. The two groups of juveniles were kept separate in circular fiberglass tanks (φ, 150 cm; volume, 1766 L) with a water recycling culture system. The tanks were filled with gravel and plants to minimize stress. During acclimation, water temperature was maintained between 25.0 and 27.5 °C under a 12-h light (daytime), 12-h dark (night) cycle. They were fed commercial food pellets (36.0% protein, 12.1% lipid, 11.3% ash, and 18.2 J·mg⁻¹ energy) twice daily.

Fig. 1 Overview of the experimental design. Wild-caught and hatchery-reared fish originated from the same region of the Yangtze River, between Honghu (114.02 E; 30.08 N) and Huangang (114.87 E; 30.45 N). Hatchery-reared fry were kept in hatchery conditions for 3 generations. Wild-caught and hatchery-reared F4 fry were kept in outdoor ponds until they became juveniles, at which point they were transferred to indoor tanks for acclimation (two weeks) before experimentation.

2.2 Anti-predator behavioral experiment

Pair-contrasts methods were performed to compare the anti-predator behavior of hatchery-reared and wild C. idellus in the presence of mandarin fish Siniper cachuatsi (Basilewsky), a natural enemy that is common and widely distributed in the Yangtze River and lakes along the river (Liang et al., 1998). Samples of hatchery-reared (n = 40, L_T = 7.62 ± 0.04 cm, mass = 3.69 ± 0.08 g) and wild (n = 40, L_T = 7.67 ± 0.05 cm, mass = 4.11 ± 0.09 g) C. idellus were randomly selected and tested, and no significant difference was found for L_T (t-paired test, P  > 0.05) between the two groups. Eight paired-test trials were conducted wherein the behavior of five hatchery-reared and five wild C. idellus were tested. Predators (n = 8, L_T = 26.9 ± 5.1 cm, mass = 315.4 ± 29.8 g) were captured in Liangzi Lake using trap nets. Prior to the experiment, eight predators were reared under the same condition as that used to acclimate C. idellus. The test tank (100 × 30 × 40 cm, Fig. 2A) was filled with dechlorinated water (120 L). A transparent perforated Plexiglas™ partition was used to divide the tank into two equal compartments. The partition allowed visual and odor cues to be exchanged between the two species. One compartment contained a predator and the other five C. idellus (either hatchery-reared or wild). The compartment containing C. idellus was divided into eight cells, each 25 × 10 cm in size (Fig. 2A). It was also divided into two equal parts marked by a vertical line on the back wall of the tank. The part closer to the partition was defined as a risky area while the other part, containing two stones, was defined as a non-risky area (Fig. 2A). The compartment chosen to contain the predator was alternated on a daily basis to avoid spatial biases (Duan et al., 2013). Additionally, the individual predator used for testing was changed regularly to avoid potential biases caused by activities unique to individuals. The rear and lateral walls of the tank were covered by black sheets to reduce visual disturbances and facilitate video-tape analysis. An 80-W lamp illuminated the tank above the water. The lighting regime was maintained as in the acclimation period, and the water temperate, dissolved oxygen, and pH were 25 ± 0.5 °C, 7.5 ± 0.4 mg · L⁻¹, and 8.19 ± 0.2, respectively.

All trials were carried out between 1000 and 1400 h every day. Each trial consisted of a paired comparison that included two test tanks, one with five randomly-selected hatchery-reared C. idellus, and one with five wild C. idellus. The C. idellus were introduced and acclimatized for 2 h. Their behavior was subsequently filmed for 1 h using a video camera (HDR-PJ790E; Sony). Then, a predator was introduced through a sliding tunnel (Fig. 2A), and filming continued for another hour. The sliding tunnel was fixed and extended from the outside of the tank to the surface of the water inside (Fig. 2A). This allowed introduction of the predator with minimal disturbance to the C. idellus. Once a trial was completed, all C. idellus and the predator were transferred to other holding tanks. The test tank was then prepared for reuse by deep cleaning and refilling with fresh water to remove potential alarm odor cues.

Because fish movement along the third dimension (from tank front to back) was not filmed, only two-dimensional images were analyzed. Videos taken from 5 min before to 5 min after predator introduction were analyzed. A frame was extracted every 20 s from each 5-min period, yielding 15 temporal frames per period. Three indicators reflecting axes of anti-predator behavior were measured, finding the mean (±S.E.) for each 5-min period for each group in each trial.

One indicator was the time spent in the risky area, reflecting the sensitivity of prey in detecting and collecting information about predators (Magurran and Seghers, 1990; Murphy and Pitcher, 1997). Another indicator was shoal cohesion, measured using the dispersion index (I_D), calculated as the ratio between the variance and the mean of the distribution of individuals in a defined area (Malavasi et al., 2004). For every temporal frame, fish were counted in each of the eight cells so that the mean and variance of the fish distribution across the cells could be determined. A completely random dispersion is when I_D = 1. When I_D > 1, regular aggregation of the shoal is indicated. Finally, inspection rate was measured. During an inspection event, C. idellus maintains visual contact with the predator and positions itself facing the predator while slowly swimming (Seghers, 1973, Kelley et al., 2003). Of the five C. idellus in the tank, those displaying this behavior as inspection rate were counted.

Fig. 2 Test tanks for measuring anti-predator behavior (100 × 30 × 40 cm; A) and for survival rate with exposure to a predator (B). The former containing (a) predator mandarin fish; (b) a transparent partition; (c) a compartment containing 5 wild or hatchery-reared C. idellus selected at random; (d) gravel shelter; (e) light source, and (f) fixed tube used to introduce a predator. The tank was divided into a risky area (g) and a non-risky area (h). The latter containing (a) test tank (φ, 150 cm; H, 100 cm); (b) two predator mandarin fish; (c) 10 wild and 10 hatchery-reared C. idellus selected at random (size-matched); (d) aquatic plant shelter; (e) gravel shelter; and (f) light source.

2.3 Predation experiment

Paired-contrasts methods were performed to compare anti-predator ability between the hatchery-reared and wild C. idellus. Four circular fiberglass tanks (φ, 150 cm; height, 100 cm; water depth, 80 cm) were supplied with temperature-controlled, flow-through water, and the floors were covered with gravel and aquatic plants. Two mandarin fish of similar size (L_T = 29.4 ± 3.9 cm, mass = 395.9 ± 22.3 g) were introduced in the tank and were allowed to prey on C. idellus.

Hatchery-reared (n = 120) and wild (n = 120) C. idellus were selected for the tests. Because the size of the individual is a key factor affecting a fish's chance of survival (Nilsson and Brönmark, 1999), the C. idellus were divided into two groups. The big-sized group weighed 7.97 g (±0.05 g) and measured 9.67 cm (±0.01 cm) in length (t-paired test of L_T, P > 0.05), and the small-sized group weighed 1.95 g (±0.01 g) and measured 5.86 cm (±0.01 cm) in length (t-paired test of L_T, P > 0.05). Six replicate trials were conducted for both groups, and each trial included 10 hatchery-reared and 10 wild fish. Feeding was suspended for all individuals (predators and prey) 24 h before each trial. Hatchery-reared C. idellus were marked by clipping a small part of the right ventral fin; wild fish were similarly marked on the left ventral fin (Zhang et al., 2014). Each trial started when size-matched hatchery-reared (n = 10) and wild (n = 10) fish were introduced via a net cage to a circular tank for 30 min acclimation. The net cage was then removed to expose the C. idellus to the predators. The trial terminated when approximately 50% of the C. idellus were eaten by the predator, then the remainder was counted (Zhang et al., 2014). The survival rate, was calculated as S/10 ∗ 100, where S is the number of surviving fish.

2.4 Statistical analyses

To compare the anti-predator behavior of hatchery-reared vs. wild C. idellus, linear mixed-effects models in the R package “nlme” (Crawley, 2007) were used to account for the non-independence of repeated measures taken on the same group of fish before and after predator exposure (Zuur et al., 2009). The three anti-predator behavior indicators were the response variables. Group identity was included as a random factor. Fixed effects were fish origin (wild or hatchery), time relative to predator exposure (before or after), and their second-order interaction. The fish were size-matched, eliminating size from the model. When a significant interaction between the origin and time of exposure was found, post hoc mixed models were applied separately before and after predator exposure. The significance threshold was 0.05.

To compare the survival rate of hatchery-reared and wild C. idellus in the predation experiment, the generalized linear mixed-effects model from the “lmer” package was used. This model can evaluate the random effect of fish group, the fixed effects of origin and size, and their second-order interaction. Using this model, the effects of origin and total size on survival rate (binomial distribution) were tested. All data were analyzed using R (R Development Core Team).

3 Results

3.1 Anti-predator behavior

Hatchery-reared and wild C. idellus avoided the risky area differently, as shown by the significant origin-by-exposure on the time spent in the risky area (Tab. 1, P < 0.01). Post hoc tests showed no difference in avoiding the risky area before predator exposure (P = 0.089), but avoidance of the area was stronger for hatchery-reared fish compared to wild fish after predator exposure (P < 0.001, Fig. 3a).

Hatchery-reared and wild C. idellus changed their shoaling behavior in the same way after predator exposure, as indicated by the non-significant origin-by-exposure (Tab. 1; P = 0.69). Fish shoals were more aggregated after predator exposure (Tab. 1; P = 0.02; Fig. 3b). Specifically, hatchery-reared fish aggregated less than the wild fish both before and after predator exposure (Tab. 1; P < 0.001, Fig. 3b).

Hatchery-reared and wild C. idellus changed their inspection rate differently after predator exposure, as shown by the significant origin-by-exposure (Tab. 1; P < 0.001). Post hoc tests showed that wild fish increased their inspection rate after exposure to the predator, but hatchery-reared fish reduced them after predator exposure (before predator exposure, P = 0.70; after predator exposure, P < 0.001, Fig. 3c).

Fig. 3 Comparisons of indicators measured in the anti-predator behavioral experiment. (a) Mean (±S.E.) time spent in the risky area by wild and hatchery-reared fish across eight replicates (n = 5 fish per trail). (b) Mean (±S.E.) shoal cohesion (aggregation index) of wild and hatchery-reared fish across eight replicates. (c) Mean (±S.E.) inspection rate (number of fish facing the predator) of wild and hatchery-reared fish across eight replicates. Asterisks above bars indicate significant differences (P < 0.05) after post hoc tests. (d) Mean (±S.E.) survival rate of wild and hatchery-reared fish after a predation event across six replicates. Asterisks above bars indicate significant differences (P < 0.05) between groups.

3.2 Survival rate

C. idellus of different body sizes had similar survival rates (effect of length, z = 0.38, P = 0.70), but wild fish survived better than hatchery-reared fish when directly exposed to the predator (effect of origin on survival rate, z = 3.71, P < 0.001; Fig. 3d).

4 Discussion

Taken together, these results suggest that hatchery-reared C. idellus retained some anti-predator abilities, consistent with previous experimental studies of other species, such as cod Gadus morhua (Linnaeus), European sea bass Dicentrarchus labrax (Linnaeus), and guppies Poecilia reticulata (Peters) (Malavasi et al., 2004; Meager et al., 2011; Swaney et al., 2015). However, hatchery-reared C. idellus displayed lower shoaling and inspection behavior when exposed to predator cues compared to wild fish, suggesting their anti-predator behavior is less efficient than their wild counterparts. Accordingly, hatchery-reared C. idellus had a lower survival rate than wild fish when exposed to an actual predator.

In addition, hatchery-reared C. idellus were more cautious in the risky area than wild fish. Such risk-averse behavior was also documented in hatchery-reared D. labrax and G. morhua (Malavasi et al., 2004; Nordeide and Svasand, 1990). To our best understanding, behavior of the hatchery-reared C. idellus observed in this study have not been observed in other studies of carp. Wild C. carpio were usually reported to be more cautious to predator cues, compared with its domestic counterparts (Matsuzaki et al., 2009). However, studies using carp still remain scarce, and further explorations of the mechanisms underlying behavioral divergence are required in the future.

Hatchery-reared C. idellus displayed lower shoaling cohesion and inspection rate than their counterparts in this study. In general, group formation is a widespread phenomenon in animal populations. This is thought to be an efficient defense strategy against predators (Krause and Ruxton, 2002). Prey shoals are reportedly rarely attacked by bluegill sunfish Lepomis macrochirus (Rafinesque) because the predator prefers individuals that are not in groups (Ioannou et al., 2012). Accordingly, the wild C. idellus achieved higher survival rate than hatchery-reared fish when exposed to predators in this study, indicating that wild fish benefited from their good shoaling and inspection rate.

More specifically, more hatchery-reared C. idellus were eaten than wild C. idellus when exposed to actual predator attacks. This could have resulted from the altered anti-predator behaviors of the hatchery-reared C. idellus. Studies of steelhead trout fry Oncorhynchus mykiss (Walbaum), and juvenile coral reef damselfish Pomacentrus wardi (Bonaparte) demonstrated that fish fed in artificial conditions without predator experience were predator-naive and vulnerable to benthic predators (Berejikian, 1995; Lönnstedt et al., 2012). Predator-naive D. labrax juveniles were observed to acquire anti-predator abilities through training with predators (Malavasi et al., 2004). Hence, hatchery-reared C. idellus could be trained using complex habitats (e.g., water structures, shelters, live predators) to strengthen their anti-predator abilities in future reintroduction to natural conditions where high predation risks exist.

In this study, only early life conditions (incubation and approximately 10 days after hatching) differed between hatchery-reared and wild fry. The wild fry were kept in the hatchery for two months before the experiment, the domestication could partly influence their response to predator signals. In spite of this, the present study show that wild C. idellus outperformed their hatchery-reared counterparts in anti-predator response and survival rate, suggesting that early stage exposure to predators or predator cues in nature could play important roles in behavior divergence. However, we must also acknowledge that bias owing to our experimental design should not be overlooked. The mechanisms underlying the behavioral differences between the hatchery-reared and wild C. idellus thus remain to be elucidated in future studies.

5 Conclusion

Compared to wild C. idellus, hatchery-reared C. idellus exhibited altered anti-predator behavior and lower survival rate when exposed to a actual predator. This suggests that hatchery-reared C. idellus could pose potential ecological risks when released in nature for conservation purposes. With this study, we hope to motivate further work on hatchery rearing consequences and to improve success in stocking programs.

Acknowledgements

This study was financially supported by the R&D Project (Grant No. 2015BAD13B02) of the Ministry of Science and Technology of China; the National Natural Science Foundation of China (Grant No. 31670542); the Research Project (Grant No. 2014FBZ04) from State Key Laboratory of Freshwater Ecology and Biotechnology, CAS; and the Special Fund (Grant No. 201203081) for Agro-scientific Research in the Public Interest. In addition, Lijun Tang was supported by the LOTUS+ project for study abroad. Lisa Jacquin was funded by a Idex starting grant from the Université Fédérale de Toulouse, France (Grant No. V5R28JACQU). The EDB laboratory in Toulouse is part of the “Laboratoire d'Excellence” TULIP (ANR-10-LABX-41).

References

All Tables

All Figures

Fig. 1 Overview of the experimental design. Wild-caught and hatchery-reared fish originated from the same region of the Yangtze River, between Honghu (114.02 E; 30.08 N) and Huangang (114.87 E; 30.45 N). Hatchery-reared fry were kept in hatchery conditions for 3 generations. Wild-caught and hatchery-reared F4 fry were kept in outdoor ponds until they became juveniles, at which point they were transferred to indoor tanks for acclimation (two weeks) before experimentation.

In the text

Fig. 2 Test tanks for measuring anti-predator behavior (100 × 30 × 40 cm; A) and for survival rate with exposure to a predator (B). The former containing (a) predator mandarin fish; (b) a transparent partition; (c) a compartment containing 5 wild or hatchery-reared C. idellus selected at random; (d) gravel shelter; (e) light source, and (f) fixed tube used to introduce a predator. The tank was divided into a risky area (g) and a non-risky area (h). The latter containing (a) test tank (φ, 150 cm; H, 100 cm); (b) two predator mandarin fish; (c) 10 wild and 10 hatchery-reared C. idellus selected at random (size-matched); (d) aquatic plant shelter; (e) gravel shelter; and (f) light source.

In the text

Fig. 3 Comparisons of indicators measured in the anti-predator behavioral experiment. (a) Mean (±S.E.) time spent in the risky area by wild and hatchery-reared fish across eight replicates (n = 5 fish per trail). (b) Mean (±S.E.) shoal cohesion (aggregation index) of wild and hatchery-reared fish across eight replicates. (c) Mean (±S.E.) inspection rate (number of fish facing the predator) of wild and hatchery-reared fish across eight replicates. Asterisks above bars indicate significant differences (P < 0.05) after post hoc tests. (d) Mean (±S.E.) survival rate of wild and hatchery-reared fish after a predation event across six replicates. Asterisks above bars indicate significant differences (P < 0.05) between groups.

In the text