Journal of Fish Biology (2014) 84, 1527–1538 doi:10.1111/jfb.12381, available online at wileyonlinelibrary.com
Rapid growth increases intrinsic predation risk in genetically modified Cyprinus carpio: implications for environmental risk L. Zhang*†, R. E. Gozlan‡, Z. Li*, J. Liu*, T. Zhang*§, W. Hu* and Z. Zhu* *State Key Laboratory of Freshwater Ecology and Biotechnology, Institute of Hydrobiology, Chinese Academy of Sciences, Wuhan, 430072, China , †Graduate School University of Chinese Academy of Sciences, Beijing, 100049, China and ‡Unité Mixte de Recherche Biologie des Organismes et Écosystèmes Aquatiques (IRD207, CNRS 7208, MNHN, UPMC), Muséum National d’Histoire Naturelle, 75231, Paris Cedex, France (Received 9 September 2013, Accepted 7 February 2014) The intrinsic effect of feeding regime on survival and predation-induced mortality was experimentally tested in genetically modified (GM) Cyprinus carpio and wild specimens. The results clearly indicate a knock-on effect of the GH gene (gcGH) introduction into the C. carpio genome on their vulnerability to predation. The experiments unequivocally showed that it is the genetic nature of the C. carpio rather than its size that affects the risk of predation. In addition, fed C. carpio were more susceptible to predation risk. Thus, the study characterizes the existence of a trade-off between somatic growth and predator avoidance performance. Current research in Europe suggests that high uncertainty surrounding the potential environmental effects of escapee transgenic fishes into the wild is largely due to uncertainty in how the modified gene will be expressed. Understanding variables such as the cost of rapid growth on antipredator success would prove to be pivotal in setting up sound risk assessments for GM fishes and in fully assessing the environmental risk associated with GM fish escapees. © 2014 The Fisheries Society of the British Isles
Key words: carp; Ctenopharyngodon idellus; GM; growth hormone; invasive; risk assessment; transgenic.
INTRODUCTION At a time where genetically modified (GM) fishes will inevitably find their way to the dinner table (Britton & Gozlan, 2013; Ledford, 2013), the environmental implications of farming GM fishes remain. From a commercial viewpoint alone, faster growth characteristics of transgenic fishes have been a central consideration as they allow faster attainment of market size and optimization of production cost (Zhu, 1992; Devlin et al., 1995). From an ecological perspective, it is currently challenging to understand the effects of such GM fishes on wild fish populations and overall on aquatic ecosystem function. There have been many studies on a range of species that show the potential transfer of genes from engineered plants and animals to their wild counterparts but §Author to whom correspondence should be addressed. Tel.: +86 27 6878 0369; email: [email protected]
1527 © 2014 The Fisheries Society of the British Isles
1528
L . Z H A N G E T A L.
there is still a lack of information about the ecological interactions of GM organisms released into native aquatic communities. Most GM fish aquaculture research has focused on salmonid species owing to the extent of this aquaculture production in countries that possess the transgenic technology (Breton & Uzbekova, 2000; Takeuchi et al., 2001; Devlin et al., 2004; Mori et al., 2007). With the emergence of other economic partners such as China into the world stage, a whole new market for GM fishes has opened up with the use of cyprinid species (Zhu et al., 1985). With a current annual freshwater fish production of c. 54 Mt (Zhao et al., 2011), China is becoming the most significant contributor to world fisheries (i.e. 17% of the global capture) and aquaculture (i.e. 61⋅4% of global aquaculture production). Recent work has focused on GM common carp Cyprinus carpio L. 1758 as it constitutes one of the most farmed species in Asia and eastern Europe and has shown the growth potential of GM C. carpio over wild specimens (Guan et al., 2008; Duan et al., 2009), with, under hatchery conditions, c. 1⋅9 times the growth of their wild counterparts (Wang et al., 2001; Li et al., 2007). At the same time recent studies have shown that a large number of invasive non-native fish species are cyprinid species such as topmouth gudgeon Pseudorasbora parva (Temminck & Schlegel 1846), nasus Chondrostoma nasus (L. 1758) or C. carpio itself (Gozlan, 2008; Gozlan et al., 2009, 2010). The reason for this is likely to be found in the manner that cyprinid and in particular C. carpio farming is performed. Because of the relatively low market value of C. carpio production when compared, for example, with salmonid species, facilities are far less biosecure often consisting of earthen ponds on floodplains or net cages in lakes, thus limiting infrastructure costs. These types of farming infrastructures are not considered as closed systems and the introduction of GM C. carpio into these environments would inevitably lead to escapees into natural open systems such as lakes or rivers (Maclean & Laight, 2000). As most GM fish research focuses on enhancing individual growth, it is vital to characterize the existence of behavioural and ecological trade-offs associated with such rapid growth (Lima & Dill, 1990). Several studies have revealed a direct increased risk of predation in fishes showing naturally faster growth rate in wild populations (Gotthard, 2000; Lankford et al., 2001; Biro et al., 2006). Given that rapid growth in fishes is linked to higher growth-hormone levels that stimulate appetite, a difference in boldness from fast-growing individuals when compared with slow-growing ones has been measured as a behavioural syndrome driving individuals to maintain feeding activity even under increasing predation risk (Lima & Dill, 1990; Biro et al., 2006; Ioannou et al., 2008; Nyqvist et al., 2013). Thus, there is a natural trade-off against selection of fast-growing individuals (Arendt, 1997; Sundström et al., 2004a, b), showing that rapid growth cannot be used as a surrogate for fitness as is often suggested (Stearns & Koella, 1986; Perrin & Rubin, 1990; Stearns, 1992). Optimal foraging and game theories, which are thought to provide a reliable basis for prediction, are based on the assumption that individuals within animal populations always behave to maximize their own chances of survival and reproduction (i.e. maximizing their fitness), no matter how much the environment changes (Sutherland, 1996). Individuals that grow faster should reap the benefits of large size earlier in life and experience higher survival and thus fitness (Roff, 1992; Stearns, 1992; Sogard, 1997). In fishes, there are numerous well-known benefits of rapid growth, including enhanced resistance to starvation during the winter, lower susceptibility to predators and higher competitive ability (Arendt, 1997; Devlin et al., 1999). If faster somatic growth was truly better, natural
© 2014 The Fisheries Society of the British Isles, Journal of Fish Biology 2014, 84, 1527–1538
P R E D AT I O N R I S K I N G M C Y P R I N U S C A R P I O
1529
selection would be expected to have driven growth rates to the maximum possible allowed by physiological or phylogenetic constraints (Stephen & Arnott, 2005), which is not the case as many animals, including fishes (Ali et al., 2003), display growth rates in nature that are kept below the physiological maximum (Arendt & Wilson, 1997, 1999; Gotthard, 2000). This indicates that the fitness advantage of rapid growth is naturally balanced against other life-history trait costs (Arendt, 1997) such as decreased escape performance (Sibly et al., 1985; Lankford et al., 2001; Munch & Conover, 2003), less vigilance and more willingness to risk predation exposure when foraging (Johnsson et al., 1996; Abrahams & Pratt, 2000; Tymchuk & Devlin, 2005a, b; Nyqvist et al., 2012) and greater predation mortality (Biro et al., 2004). Until now, theoreticians have overlooked the intrinsic underlying physiological cost of rapid growth (Stephen & Arnott, 2005) as most studies have only looked at the increased risk of predation incurred by individuals when foraging (Lima & Dill, 1990). Here, the intrinsic effect of feeding status on survival and predation-induced mortality was experimentally tested in GM C. carpio and wild specimens. Understanding variables such as the cost of rapid growth on antipredator success would prove to be pivotal in setting up sound risk assessments (RA) for GM fishes (Devlin et al., 2006) and in fully assessing the environmental risk associated with GM fish escapees.
MATERIALS AND METHODS SOURCE OF C. CARPIO The all-fish gene of growth hormone transgenic carp construct pCAgcGH contains the grass carp Ctenopharyngodon idellus (Valenciennes 1844) GH gene (gcGH) driven by the 𝛽-actin gene promoter in C. carpio. P0 , F1 , F2 , F3 , F4 and F5 all-fish GH-transgenic C. carpio were produced as reported by Guan et al. (2008). In May 2010 the F5 GH-transgenic C. carpio (T) were produced by cross-fertilization between F4 transgenic males and normal non-transgenic C. carpio (NT) wild-type females captured in the floodplain of the Yangtze River. Non-transgenic C. carpio were produced by cross-fertilization of males and females in F4 sibling fishes to minimize the effects of genetic differences and maternal and paternal effects. Both NT and T groups of C. carpio were reared in the contained ponds in Wuhan Duofu High-Tech Company Ltd. At the start of July, at least 1 month before the start of the experimental trials, both groups of C. carpio were transferred to an indoor water-recycling fish-rearing facility and reared in circular fibreglass tanks (diameter 150 cm and volume 1000 l). Cyprinus carpio were acclimated under a 12L:12D photoperiod regime and were fed twice daily at 0900 and 1600 hours with frozen Chironomus sp. larvae. Water temperature ranged between 27⋅9 and 31⋅7∘ C, dissolved oxygen was 6⋅75–7⋅35 mg l−1 and pH ranged between 8⋅27 and 8⋅45. In addition, the Mandarin fish Siniperca chuatsi (Basilewsky 1855) species was selected as a natural predator of C. carpio. It is a demersal piscivorous predator, feeding exclusively on live fishes and widely distributed in many rivers and lakes in China (Liang et al., 1998). All predators were captured from the wild in the Yangtze floodplain to maximize natural predatory behaviour and taken back to the indoor facility at the same time in July and reared in the same fibreglass test tanks as the C. carpio. Siniperca chuatsi were given at least 1 month before the start of the experiment to acclimate to the experimental environment and were fed once a day during that period with live juvenile C. carpio. E X P E R I M E N TA L P R O T O C O L A paired-contrast design was performed to test the effect on population of genotype (T v. NT), and food consumption (fed v. unfed) on vulnerability to predation. Thus, 24 h before
© 2014 The Fisheries Society of the British Isles, Journal of Fish Biology 2014, 84, 1527–1538
1530
L . Z H A N G E T A L.
any treatment all test individuals (predator and prey) were deprived of food. Thirty minutes before some trials, groups of C. carpio were allowed an unrestricted meal and thus labelled as fed C. carpio in these respective trials. Each trial consisted of a set of three predators of similar sizes per tank [mean ± s.e. mass = 283 ± 33⋅57 g; mean ± s.e. total length (LT ) = 26⋅41 ± 0⋅90 cm] and one of the four groups of prey. The four groups of tested C. carpio were set as follows: (1) NT fed (n = 30) and T fed (n = 30), (2) NT unfed (n = 30) and T unfed (n = 30), (3) NT unfed (n = 30) and NT fed (n = 30) and (4) T unfed (n = 30) and T fed (n = 30). Each trial was replicated at least four times and a maximum of six times. All C. carpio used in the trials were selected to be of similar mass and LT (NT C. carpio mean ± s.e. LT = 4⋅99 ± 0⋅35 cm; mean ± s.e. mass 2⋅02 ± 0⋅35 g and T C. carpio mean ± s.e. LT = 4⋅82 ± 0⋅42 cm; mean ± s.e. mass 1⋅81 ± 0⋅43 g). Schools of C. carpio were introduced into the net cage within each trial arena and allowed 15 min acclimation. Net cages were then removed and S. chuatsi were allowed to prey. All fish were observed and recorded visually during each trial and the experiment was stopped when c. 50% of C. carpio were captured. The time of each trial was ranged between 1⋅5 and 2 h. For each trial, each specific group was marked by fin clipping of either right or left ventral fins. Ventral fin side for each marked group was randomly selected between replicated trials to avoid confusion. All C. carpio were anaesthetized with Eugenol at the concentration of 20 mg l−1 before marking (Velisek et al., 2005), weighing and measuring. Fin clipping was performed as it generates extremely limited pain when compared with external tagging in the fish muscle and it is particularly suitable to the small size of C. carpio. In addition, clipped fins re-grow fairly rapidly and the effect of clipping the ventral fin on swimming ability is limited as ventral fins in C. carpio are not used for propulsion and escape behaviour. No other source of mortality apart from predation was noticed during the whole period of experimentation. All ethical considerations for the welfare of C. carpio during the regulated procedure were taken into account according to the institutional ethical regulations in place. In each experiment, a cut-off point of 50% mortality was used to minimize C. carpio suffering; a lower cut-off point would have represented an arbitrary end point rather than true predator preference. GM C. carpio have a strong desire to feed so they experience an increased risk of predation during foraging. A bare environment was used in each experiment to represent the natural situation in most Chinese lakes that, as a result of extensive fisheries exploitation, are fairly shallow, flat and characterized by the absence of macrophytes or overhanging vegetation on the banks. Thus, the lack of shelter was environmentally realistic and ensured that the experiments tested susceptibility to predation of C. carpio, not the antipredator ability (Munch & Conover, 2003).
S TAT I S T I C A L A N A LY S E S The relative vulnerability of C. carpio from contrasting treatments was estimated using frequencies of captured and non-captured individuals during the various trials. The frequency of captured individuals into mortality estimates was then translated. Mortalities from contrasting treatments were then analysed as 2 × 2 contingency tables (𝜒 2 -test in cross table) to test treatment effects. A significance level of P ≤ 0⋅05 was used for all tests. Statistical analyses were performed using SPSS 16.0 (www-01.ibm.com/software/analytics/spss/downloads/) and graphics were made with Excel and origin75 software. The number of C. carpio in each trial was based on the natural density of C. carpio in Chinese lakes. The number of replicates (between four and six) was the minimum number of replicates to statistically characterize significant differences based on the intrinsic variability (i.e. noise) in the data whilst minimizing the ethical impact of the study (i.e. reduction). D I F F E R E N T I A L V U L N E R A B I L I T Y O R P R E D AT O R PREFERENCE? As prey selection from predators could have resulted more from a preference for fed C. carpio that are visibly fatter than the unfed ones, rather than from the ability of prey to avoid predation, in a separate no-choice predation trial experiment, C. carpio evasiveness was measured independently from predator selectivity. In this evasiveness experiment the predators were
© 2014 The Fisheries Society of the British Isles, Journal of Fish Biology 2014, 84, 1527–1538
P R E D AT I O N R I S K I N G M C Y P R I N U S C A R P I O
1531
allowed to forage exclusively on either NT or T C. carpio (NT mean ± s.e. LT = 6⋅89 ± 0⋅35 cm; mass = 4⋅78 ± 0⋅92 g and T mean ± s.e. LT = 6⋅86 ± 0⋅33 cm; mass = 4⋅85 ± 0⋅80 g). Trials were initiated after the paired-contrast experiment by stocking each replicate tank with three new S. chuatsi that were deprived of food for 24 h. A school of each type of C. carpio (n = 20) was introduced into a net cage within each tank and allowed 15 min to acclimate. Net cages were then removed and S. chuatsi were allowed to prey for 100 min. In order to minimize interference with the running of the experiment, every 5–10 min, each tank was photographed to count the number of test C. carpio left in each trial. Thus, survival curves were established to characterize the mortality trade-off for each C. carpio type. Kaplan–Meier (K-M) survival probabilities were used to construct survivorship functions for each C. carpio treatment. Statistical analyses were performed with the SURVIVAL programme (Steinberg & Colla, 1988; Li et al., 2008). Non-parametric Mantel–Cox log-rank tests were then used to test whether survivorship curves obtained by the K-M method differed between the four groups of carp (Li et al., 2007). Median lethal time (tL50 ) and maximum-likelihood estimates of survival probability were calculated for each treatment using the non-parametric K-M product-limit method (Kaplan & Meier, 1958). Z-tests were used to test whether √ significant differences existed in tL50 of each treatment C. carpio: z = [S1 (t) − S2 (t)] ( {x2 [S1 (t)] + x2 [S2 (t)]})−1 , where x = s.e. and z is the value that could define whether there is a significant difference as the function of P-value. There is a significant difference when z < 0⋅05, whereas there is no significant difference when z ≥ 0⋅05; Si (t) is the estimated time of the tL50 of the test C. carpio in group i through the SURVIVAL analysis. Statistical analyses were performed again using SPSS 16.0 and graphics were made with Excel and origin75 software.
RESULTS P R E D AT I O N M O RTA L I T I E S
Size-matched groups of T and NT C. carpio genotypes differed significantly in their ability to avoid predators under both fed and unfed conditions, with fast-growing T genotypes being highly vulnerable to predation relative to slow-growing NT genotypes. In the trial with the unfed C. carpio, mortality rate of T C. carpio was significantly higher than that of the NT C. carpio [𝜒 2 = 5⋅842, d.f. = 1, P < 0⋅05; Fig. 1(a)]. Similar results were found in trials with fed T and NT C. carpio, where T C. carpio were more likely to be captured by predators than the NT C. carpio [𝜒 2 = 7⋅065, d.f. = 1, P < 0⋅01; Fig. 1(b)]. T and NT C. carpio that consumed meals suffered significantly higher predation mortality than their unfed counterparts. In T C. carpio, fed individuals were more likely to be captured by the predators than their unfed T C. carpio [𝜒 2 = 6⋅559, d.f. = 1, P < 0⋅05; Fig. 1(c)]; feeding also significantly reduced escape performance in NT C. carpio attacked by predators [𝜒 2 = 5⋅376, d.f. = 1, P < 0⋅05; Fig. 1(d)]. D I F F E R E N T I A L V U L N E R A B I L I T Y O R P R E D AT O R PREFERENCE?
At the population level, unfed T C. carpio were captured faster by predators than unfed NT C. carpio [𝜒 2 = 5⋅281, d.f. = 1, P < 0⋅05; Fig. 2(b)]. Similar results were found when C. carpio were allowed to feed before the trial [𝜒 2 = 6⋅237, d.f. = 1, P < 0⋅05; Fig. 2(a)], suggesting that T C. carpio are less able to escape predation than NT individuals (Fig. 2). Thus, survival times for fed T individuals were significantly lower than those for the unfed individuals [𝜒 2 = 7⋅225, d.f. = 1, P < 0⋅01; Fig. 2(d)]
© 2014 The Fisheries Society of the British Isles, Journal of Fish Biology 2014, 84, 1527–1538
1532 1·0
L . Z H A N G E T A L.
(a)
1·0
Predation mortality
0·9 0·8
(b)
1·0
0·9
*
0·8
(c)
1·0 0·9
0·9
**
*
0·8
0·7
0·7
0·7
0·6
0·6
0·6
0·6
0·5
0·5
0·5
0·5
0·4
0·4
0·4
0·4
0·3
0·3
0·3
0·3
0·2
0·2
0·2
0·2
0·1
0·1
0·1
0·1
0
0
0 T and fed NT and fed
*
0·8
0·7
T and fed NT and fed
(d)
0 T and fed NT and fed
T and fed NT and fed
Fig. 1. Effect of Cyprinus carpio genotype (i.e. transgenic, T or natural, NT) on predation vulnerability when either unfed for 24 h before the experiment or with unlimited feeding 30 min before the experiment. Experimental set-up included three Siniperca chuatsi of similar sizes, and either (a), (b) mixed schools (n = 60) of transgenic and non-transgenic individuals with different feeding status [(a) n = 4; (b): n = 5] or (c), (d) single genotyped schools (n = 60) also of different feeding status [(c) n = 5; (d) n = 4]. Mortalities from contrasting treatments were then analysed as 2 × 2 contingency tables (𝜒 2 -test in cross table) to test treatment effects. Significant differences in mortalities between treatments were tested using a 𝜒 2 -test (*P < 0⋅05 and **P < 0⋅01). Values are mean ± s.e.
and fed NT C. carpio [𝜒 2 = 19⋅546, d.f. = 1, P < 0⋅001; Fig. 2(c)]. Survival times of fed and unfed C. carpio indicated that consumption of meals reduced evasiveness in both genotypes (Fig. 2) and unfed T C. carpio were captured faster by the predators than fed NT C. carpio even though the difference between them was not significant [𝜒 2 = 1⋅821, d.f. = 1, P > 0⋅05; Fig. 2(e)]. The tL50 of the four treatment groups NT fed, T fed, NT unfed and T unfed was 20, 10, 50 and 30 min, respectively (Table I), with the tL50 of NT fed individuals being significantly lower than that of the unfed individuals (P < 0⋅001); the same for the T fed C. carpio (P < 0⋅001). Finally, the tL50 of T fed C. carpio was also significantly lower than that of the NT fed (P < 0⋅001) and the same for T unfed v. NT unfed C. carpio (P < 0⋅01). Analysis of failure times indicates that selection against C. carpio that display genetic potential for rapid growth could be at least partially explained by lower escape potential.
DISCUSSION The results clearly indicate a knock-on effect of the GH gene (gcGH) introduction into the C. carpio genome on their vulnerability to predation. The experiments unequivocally show that it is the genetic nature of the C. carpio rather than its size that affects the risk of predation. In addition, the satiety level of the C. carpio also increased the predation risk regardless of any specific genetic makeup. Thus, the study characterizes the existence of a trade-off between somatic growth and predator avoidance performance (Lankford et al., 2001; Li et al., 2007). Similar results were obtained in brown trout Salmo trutta L. 1758, in which simple injection of growth hormone influenced antipredator behaviour (Johnsson et al., 1996; Jönsson et al., 1998), making fast-growing S. trutta more risk-prone, significantly decreasing their critical swimming speeds and thus making them more vulnerable to predation (Farrell et al., 1997). Here,
© 2014 The Fisheries Society of the British Isles, Journal of Fish Biology 2014, 84, 1527–1538
1533
P R E D AT I O N R I S K I N G M C Y P R I N U S C A R P I O
(a)
(b)
(c)
(d)
1·0 0·8 0·6
Survival probability
0·4 0·2 0
1·0 0·8 0·6 0·4 0·2 0 0
20
40
60
80
100
10
20
30
40
50
60
70
80
(e) Survival probability
1·0 0·8 0·6 0·4 0·2 0 0
20
40
60
80
100
Survival time (min) Fig. 2. Survival curves of Cyprinus carpio genotype (i.e. transgenic, T or natural, i.e. non-transgenic, NT) on predation vulnerability when either unfed for 24 h before the experiment or with unlimited feeding 30 min before the experiment. In each of the four experiments, three Siniperca chuatsi of similar sizes were allowed to prey on a monotyped school (n = 20) of either transgenic fed (n = 4 replicates) or transgenic unfed (n = 4) or non-transgenic fed (n = 6) or non-transgenic unfed (n = 5) C. carpio. Comparison of the survival times between (a) T and fed ( ) and NT and fed ( ); (b) T and unfed ( ) and NT and unfed ( ); (c) NT and fed ( ) and NT and unfed ( ); (d) T and fed ( and T and unfed ( ) and (e) T and unfed ( ) and NT and fed ( ).
© 2014 The Fisheries Society of the British Isles, Journal of Fish Biology 2014, 84, 1527–1538
1534
L . Z H A N G E T A L.
the fed–unfed trials suggested that large meals represented a hidden additional cost of growth for the C. carpio. Previous research has argued that high rates of food consumption and rapid growth in fishes drive metabolism to its aerobic maximum, thereby producing a conflict with allocation of oxygen to other functions such as swimming (Priede, 1985; Wieser & Medgyesy, 1990; Wieser, 1991). Other studies on Salmo salar L. 1758 and Ictalurus punctatus (Rafinesque 1818) show that the overexpression in growth hormone-stimulated fishes has enhanced forage motivation, lowered resolution of the food selection, increased activity level and reduced clustering trends (Abrahams & Sutterlin, 1999; Dunham et al., 1999; Sundström et al., 2003, 2004a, b; Tymchuk & Devlin, 2005a, b). It is key to differentiate between farmed fishes and wild populations of C. carpio as both environments have separate selection pressures. Most research on growth-enhanced fishes, either through the use of growth hormones or the use of a modified genome, has an economic applied endpoint, which is to produce marketable fishes at a faster pace, thus lowering production cost. In such farm environments, fishes are not subjected to predation and thus, as seen in rainbow trout, Oncorhynchus mykiss (Walbaum 1792), for example, growth rates in domestic strains can reach three times those observed in wild strains (Johnsson et al., 1996). This differential in growth is achieved by a greater motivation to feed with elevated foraging rates and also increased aggression (Fleming et al., 2002). On the contrary, in natural environments, foraging competition is an important evolutionary pressure, so is predation. Thus, in the wild, the selection for rapid growth is counter selected by individuals with better predator avoidance capabilities (Arendt, 1997; Post & Parkinson, 2001; Munch & Conover, 2003; Stoks et al., 2005) and this trade-off explains the absence of individuals with maximized growth rates in nature (Werner & Gilliam, 1984; Werner & Anholt, 1993; Anholt & Werner, 1995; Gotthard, 2000; Biro et al., 2003a, b, 2004, 2006; Stoks et al., 2005). In light of the results of this study, the key environmental risk lies with the potential for transgenic fishes to escape aquaculture farms and contaminate wild natural stocks. Because of the relatively low economic value of C. carpio compared with salmonid species, their culture facilities tend to be less biosecure and often consist of earthen ponds located on the floodplain or fish cages in reservoirs, as these limit infrastructure costs. Such farming infrastructures are not closed systems and the use of transgenic Table I. Mean survival time and median lethal time (tL50 ) expressed as the survival time of 50% of a given school of Cyprinus carpio genotype (i.e. transgenic, T or natural, i.e. non-transgenic, NT) either unfed for 24 h before the experiment or with unlimited feeding 30 min before the experiment. The experimental set-up included three Siniperca chuatsi of similar size. The superscript lowercase letters in the median estimate column represent significant differences between the two values Mean Identification A B C D
Type
Median(tL50 )
Estimate s.e.
NT and fed NT and unfed T and unfed T and fed
27⋅40 49⋅44 35⋅83 20⋅47
2⋅98 4⋅67 3⋅85 2⋅33
95% c.l. 21⋅56–33⋅25 40⋅27–58⋅61 28⋅28–43⋅37 15⋅89–35⋅88
Estimate s.e. 20⋅0a 50⋅0b 30⋅0c 10⋅0d
95% c.l.
Z-value Pair
2⋅92 14⋅26–25⋅73
Rapid growth increases intrinsic predation risk in genetically modified Cyprinus carpio: implications for environmental risk L. Zhang*†, R. E. Gozlan‡, Z. Li*, J. Liu*, T. Zhang*§, W. Hu* and Z. Zhu* *State Key Laboratory of Freshwater Ecology and Biotechnology, Institute of Hydrobiology, Chinese Academy of Sciences, Wuhan, 430072, China , †Graduate School University of Chinese Academy of Sciences, Beijing, 100049, China and ‡Unité Mixte de Recherche Biologie des Organismes et Écosystèmes Aquatiques (IRD207, CNRS 7208, MNHN, UPMC), Muséum National d’Histoire Naturelle, 75231, Paris Cedex, France (Received 9 September 2013, Accepted 7 February 2014) The intrinsic effect of feeding regime on survival and predation-induced mortality was experimentally tested in genetically modified (GM) Cyprinus carpio and wild specimens. The results clearly indicate a knock-on effect of the GH gene (gcGH) introduction into the C. carpio genome on their vulnerability to predation. The experiments unequivocally showed that it is the genetic nature of the C. carpio rather than its size that affects the risk of predation. In addition, fed C. carpio were more susceptible to predation risk. Thus, the study characterizes the existence of a trade-off between somatic growth and predator avoidance performance. Current research in Europe suggests that high uncertainty surrounding the potential environmental effects of escapee transgenic fishes into the wild is largely due to uncertainty in how the modified gene will be expressed. Understanding variables such as the cost of rapid growth on antipredator success would prove to be pivotal in setting up sound risk assessments for GM fishes and in fully assessing the environmental risk associated with GM fish escapees. © 2014 The Fisheries Society of the British Isles
Key words: carp; Ctenopharyngodon idellus; GM; growth hormone; invasive; risk assessment; transgenic.
INTRODUCTION At a time where genetically modified (GM) fishes will inevitably find their way to the dinner table (Britton & Gozlan, 2013; Ledford, 2013), the environmental implications of farming GM fishes remain. From a commercial viewpoint alone, faster growth characteristics of transgenic fishes have been a central consideration as they allow faster attainment of market size and optimization of production cost (Zhu, 1992; Devlin et al., 1995). From an ecological perspective, it is currently challenging to understand the effects of such GM fishes on wild fish populations and overall on aquatic ecosystem function. There have been many studies on a range of species that show the potential transfer of genes from engineered plants and animals to their wild counterparts but §Author to whom correspondence should be addressed. Tel.: +86 27 6878 0369; email: [email protected]
1527 © 2014 The Fisheries Society of the British Isles
1528
L . Z H A N G E T A L.
there is still a lack of information about the ecological interactions of GM organisms released into native aquatic communities. Most GM fish aquaculture research has focused on salmonid species owing to the extent of this aquaculture production in countries that possess the transgenic technology (Breton & Uzbekova, 2000; Takeuchi et al., 2001; Devlin et al., 2004; Mori et al., 2007). With the emergence of other economic partners such as China into the world stage, a whole new market for GM fishes has opened up with the use of cyprinid species (Zhu et al., 1985). With a current annual freshwater fish production of c. 54 Mt (Zhao et al., 2011), China is becoming the most significant contributor to world fisheries (i.e. 17% of the global capture) and aquaculture (i.e. 61⋅4% of global aquaculture production). Recent work has focused on GM common carp Cyprinus carpio L. 1758 as it constitutes one of the most farmed species in Asia and eastern Europe and has shown the growth potential of GM C. carpio over wild specimens (Guan et al., 2008; Duan et al., 2009), with, under hatchery conditions, c. 1⋅9 times the growth of their wild counterparts (Wang et al., 2001; Li et al., 2007). At the same time recent studies have shown that a large number of invasive non-native fish species are cyprinid species such as topmouth gudgeon Pseudorasbora parva (Temminck & Schlegel 1846), nasus Chondrostoma nasus (L. 1758) or C. carpio itself (Gozlan, 2008; Gozlan et al., 2009, 2010). The reason for this is likely to be found in the manner that cyprinid and in particular C. carpio farming is performed. Because of the relatively low market value of C. carpio production when compared, for example, with salmonid species, facilities are far less biosecure often consisting of earthen ponds on floodplains or net cages in lakes, thus limiting infrastructure costs. These types of farming infrastructures are not considered as closed systems and the introduction of GM C. carpio into these environments would inevitably lead to escapees into natural open systems such as lakes or rivers (Maclean & Laight, 2000). As most GM fish research focuses on enhancing individual growth, it is vital to characterize the existence of behavioural and ecological trade-offs associated with such rapid growth (Lima & Dill, 1990). Several studies have revealed a direct increased risk of predation in fishes showing naturally faster growth rate in wild populations (Gotthard, 2000; Lankford et al., 2001; Biro et al., 2006). Given that rapid growth in fishes is linked to higher growth-hormone levels that stimulate appetite, a difference in boldness from fast-growing individuals when compared with slow-growing ones has been measured as a behavioural syndrome driving individuals to maintain feeding activity even under increasing predation risk (Lima & Dill, 1990; Biro et al., 2006; Ioannou et al., 2008; Nyqvist et al., 2013). Thus, there is a natural trade-off against selection of fast-growing individuals (Arendt, 1997; Sundström et al., 2004a, b), showing that rapid growth cannot be used as a surrogate for fitness as is often suggested (Stearns & Koella, 1986; Perrin & Rubin, 1990; Stearns, 1992). Optimal foraging and game theories, which are thought to provide a reliable basis for prediction, are based on the assumption that individuals within animal populations always behave to maximize their own chances of survival and reproduction (i.e. maximizing their fitness), no matter how much the environment changes (Sutherland, 1996). Individuals that grow faster should reap the benefits of large size earlier in life and experience higher survival and thus fitness (Roff, 1992; Stearns, 1992; Sogard, 1997). In fishes, there are numerous well-known benefits of rapid growth, including enhanced resistance to starvation during the winter, lower susceptibility to predators and higher competitive ability (Arendt, 1997; Devlin et al., 1999). If faster somatic growth was truly better, natural
© 2014 The Fisheries Society of the British Isles, Journal of Fish Biology 2014, 84, 1527–1538
P R E D AT I O N R I S K I N G M C Y P R I N U S C A R P I O
1529
selection would be expected to have driven growth rates to the maximum possible allowed by physiological or phylogenetic constraints (Stephen & Arnott, 2005), which is not the case as many animals, including fishes (Ali et al., 2003), display growth rates in nature that are kept below the physiological maximum (Arendt & Wilson, 1997, 1999; Gotthard, 2000). This indicates that the fitness advantage of rapid growth is naturally balanced against other life-history trait costs (Arendt, 1997) such as decreased escape performance (Sibly et al., 1985; Lankford et al., 2001; Munch & Conover, 2003), less vigilance and more willingness to risk predation exposure when foraging (Johnsson et al., 1996; Abrahams & Pratt, 2000; Tymchuk & Devlin, 2005a, b; Nyqvist et al., 2012) and greater predation mortality (Biro et al., 2004). Until now, theoreticians have overlooked the intrinsic underlying physiological cost of rapid growth (Stephen & Arnott, 2005) as most studies have only looked at the increased risk of predation incurred by individuals when foraging (Lima & Dill, 1990). Here, the intrinsic effect of feeding status on survival and predation-induced mortality was experimentally tested in GM C. carpio and wild specimens. Understanding variables such as the cost of rapid growth on antipredator success would prove to be pivotal in setting up sound risk assessments (RA) for GM fishes (Devlin et al., 2006) and in fully assessing the environmental risk associated with GM fish escapees.
MATERIALS AND METHODS SOURCE OF C. CARPIO The all-fish gene of growth hormone transgenic carp construct pCAgcGH contains the grass carp Ctenopharyngodon idellus (Valenciennes 1844) GH gene (gcGH) driven by the 𝛽-actin gene promoter in C. carpio. P0 , F1 , F2 , F3 , F4 and F5 all-fish GH-transgenic C. carpio were produced as reported by Guan et al. (2008). In May 2010 the F5 GH-transgenic C. carpio (T) were produced by cross-fertilization between F4 transgenic males and normal non-transgenic C. carpio (NT) wild-type females captured in the floodplain of the Yangtze River. Non-transgenic C. carpio were produced by cross-fertilization of males and females in F4 sibling fishes to minimize the effects of genetic differences and maternal and paternal effects. Both NT and T groups of C. carpio were reared in the contained ponds in Wuhan Duofu High-Tech Company Ltd. At the start of July, at least 1 month before the start of the experimental trials, both groups of C. carpio were transferred to an indoor water-recycling fish-rearing facility and reared in circular fibreglass tanks (diameter 150 cm and volume 1000 l). Cyprinus carpio were acclimated under a 12L:12D photoperiod regime and were fed twice daily at 0900 and 1600 hours with frozen Chironomus sp. larvae. Water temperature ranged between 27⋅9 and 31⋅7∘ C, dissolved oxygen was 6⋅75–7⋅35 mg l−1 and pH ranged between 8⋅27 and 8⋅45. In addition, the Mandarin fish Siniperca chuatsi (Basilewsky 1855) species was selected as a natural predator of C. carpio. It is a demersal piscivorous predator, feeding exclusively on live fishes and widely distributed in many rivers and lakes in China (Liang et al., 1998). All predators were captured from the wild in the Yangtze floodplain to maximize natural predatory behaviour and taken back to the indoor facility at the same time in July and reared in the same fibreglass test tanks as the C. carpio. Siniperca chuatsi were given at least 1 month before the start of the experiment to acclimate to the experimental environment and were fed once a day during that period with live juvenile C. carpio. E X P E R I M E N TA L P R O T O C O L A paired-contrast design was performed to test the effect on population of genotype (T v. NT), and food consumption (fed v. unfed) on vulnerability to predation. Thus, 24 h before
© 2014 The Fisheries Society of the British Isles, Journal of Fish Biology 2014, 84, 1527–1538
1530
L . Z H A N G E T A L.
any treatment all test individuals (predator and prey) were deprived of food. Thirty minutes before some trials, groups of C. carpio were allowed an unrestricted meal and thus labelled as fed C. carpio in these respective trials. Each trial consisted of a set of three predators of similar sizes per tank [mean ± s.e. mass = 283 ± 33⋅57 g; mean ± s.e. total length (LT ) = 26⋅41 ± 0⋅90 cm] and one of the four groups of prey. The four groups of tested C. carpio were set as follows: (1) NT fed (n = 30) and T fed (n = 30), (2) NT unfed (n = 30) and T unfed (n = 30), (3) NT unfed (n = 30) and NT fed (n = 30) and (4) T unfed (n = 30) and T fed (n = 30). Each trial was replicated at least four times and a maximum of six times. All C. carpio used in the trials were selected to be of similar mass and LT (NT C. carpio mean ± s.e. LT = 4⋅99 ± 0⋅35 cm; mean ± s.e. mass 2⋅02 ± 0⋅35 g and T C. carpio mean ± s.e. LT = 4⋅82 ± 0⋅42 cm; mean ± s.e. mass 1⋅81 ± 0⋅43 g). Schools of C. carpio were introduced into the net cage within each trial arena and allowed 15 min acclimation. Net cages were then removed and S. chuatsi were allowed to prey. All fish were observed and recorded visually during each trial and the experiment was stopped when c. 50% of C. carpio were captured. The time of each trial was ranged between 1⋅5 and 2 h. For each trial, each specific group was marked by fin clipping of either right or left ventral fins. Ventral fin side for each marked group was randomly selected between replicated trials to avoid confusion. All C. carpio were anaesthetized with Eugenol at the concentration of 20 mg l−1 before marking (Velisek et al., 2005), weighing and measuring. Fin clipping was performed as it generates extremely limited pain when compared with external tagging in the fish muscle and it is particularly suitable to the small size of C. carpio. In addition, clipped fins re-grow fairly rapidly and the effect of clipping the ventral fin on swimming ability is limited as ventral fins in C. carpio are not used for propulsion and escape behaviour. No other source of mortality apart from predation was noticed during the whole period of experimentation. All ethical considerations for the welfare of C. carpio during the regulated procedure were taken into account according to the institutional ethical regulations in place. In each experiment, a cut-off point of 50% mortality was used to minimize C. carpio suffering; a lower cut-off point would have represented an arbitrary end point rather than true predator preference. GM C. carpio have a strong desire to feed so they experience an increased risk of predation during foraging. A bare environment was used in each experiment to represent the natural situation in most Chinese lakes that, as a result of extensive fisheries exploitation, are fairly shallow, flat and characterized by the absence of macrophytes or overhanging vegetation on the banks. Thus, the lack of shelter was environmentally realistic and ensured that the experiments tested susceptibility to predation of C. carpio, not the antipredator ability (Munch & Conover, 2003).
S TAT I S T I C A L A N A LY S E S The relative vulnerability of C. carpio from contrasting treatments was estimated using frequencies of captured and non-captured individuals during the various trials. The frequency of captured individuals into mortality estimates was then translated. Mortalities from contrasting treatments were then analysed as 2 × 2 contingency tables (𝜒 2 -test in cross table) to test treatment effects. A significance level of P ≤ 0⋅05 was used for all tests. Statistical analyses were performed using SPSS 16.0 (www-01.ibm.com/software/analytics/spss/downloads/) and graphics were made with Excel and origin75 software. The number of C. carpio in each trial was based on the natural density of C. carpio in Chinese lakes. The number of replicates (between four and six) was the minimum number of replicates to statistically characterize significant differences based on the intrinsic variability (i.e. noise) in the data whilst minimizing the ethical impact of the study (i.e. reduction). D I F F E R E N T I A L V U L N E R A B I L I T Y O R P R E D AT O R PREFERENCE? As prey selection from predators could have resulted more from a preference for fed C. carpio that are visibly fatter than the unfed ones, rather than from the ability of prey to avoid predation, in a separate no-choice predation trial experiment, C. carpio evasiveness was measured independently from predator selectivity. In this evasiveness experiment the predators were
© 2014 The Fisheries Society of the British Isles, Journal of Fish Biology 2014, 84, 1527–1538
P R E D AT I O N R I S K I N G M C Y P R I N U S C A R P I O
1531
allowed to forage exclusively on either NT or T C. carpio (NT mean ± s.e. LT = 6⋅89 ± 0⋅35 cm; mass = 4⋅78 ± 0⋅92 g and T mean ± s.e. LT = 6⋅86 ± 0⋅33 cm; mass = 4⋅85 ± 0⋅80 g). Trials were initiated after the paired-contrast experiment by stocking each replicate tank with three new S. chuatsi that were deprived of food for 24 h. A school of each type of C. carpio (n = 20) was introduced into a net cage within each tank and allowed 15 min to acclimate. Net cages were then removed and S. chuatsi were allowed to prey for 100 min. In order to minimize interference with the running of the experiment, every 5–10 min, each tank was photographed to count the number of test C. carpio left in each trial. Thus, survival curves were established to characterize the mortality trade-off for each C. carpio type. Kaplan–Meier (K-M) survival probabilities were used to construct survivorship functions for each C. carpio treatment. Statistical analyses were performed with the SURVIVAL programme (Steinberg & Colla, 1988; Li et al., 2008). Non-parametric Mantel–Cox log-rank tests were then used to test whether survivorship curves obtained by the K-M method differed between the four groups of carp (Li et al., 2007). Median lethal time (tL50 ) and maximum-likelihood estimates of survival probability were calculated for each treatment using the non-parametric K-M product-limit method (Kaplan & Meier, 1958). Z-tests were used to test whether √ significant differences existed in tL50 of each treatment C. carpio: z = [S1 (t) − S2 (t)] ( {x2 [S1 (t)] + x2 [S2 (t)]})−1 , where x = s.e. and z is the value that could define whether there is a significant difference as the function of P-value. There is a significant difference when z < 0⋅05, whereas there is no significant difference when z ≥ 0⋅05; Si (t) is the estimated time of the tL50 of the test C. carpio in group i through the SURVIVAL analysis. Statistical analyses were performed again using SPSS 16.0 and graphics were made with Excel and origin75 software.
RESULTS P R E D AT I O N M O RTA L I T I E S
Size-matched groups of T and NT C. carpio genotypes differed significantly in their ability to avoid predators under both fed and unfed conditions, with fast-growing T genotypes being highly vulnerable to predation relative to slow-growing NT genotypes. In the trial with the unfed C. carpio, mortality rate of T C. carpio was significantly higher than that of the NT C. carpio [𝜒 2 = 5⋅842, d.f. = 1, P < 0⋅05; Fig. 1(a)]. Similar results were found in trials with fed T and NT C. carpio, where T C. carpio were more likely to be captured by predators than the NT C. carpio [𝜒 2 = 7⋅065, d.f. = 1, P < 0⋅01; Fig. 1(b)]. T and NT C. carpio that consumed meals suffered significantly higher predation mortality than their unfed counterparts. In T C. carpio, fed individuals were more likely to be captured by the predators than their unfed T C. carpio [𝜒 2 = 6⋅559, d.f. = 1, P < 0⋅05; Fig. 1(c)]; feeding also significantly reduced escape performance in NT C. carpio attacked by predators [𝜒 2 = 5⋅376, d.f. = 1, P < 0⋅05; Fig. 1(d)]. D I F F E R E N T I A L V U L N E R A B I L I T Y O R P R E D AT O R PREFERENCE?
At the population level, unfed T C. carpio were captured faster by predators than unfed NT C. carpio [𝜒 2 = 5⋅281, d.f. = 1, P < 0⋅05; Fig. 2(b)]. Similar results were found when C. carpio were allowed to feed before the trial [𝜒 2 = 6⋅237, d.f. = 1, P < 0⋅05; Fig. 2(a)], suggesting that T C. carpio are less able to escape predation than NT individuals (Fig. 2). Thus, survival times for fed T individuals were significantly lower than those for the unfed individuals [𝜒 2 = 7⋅225, d.f. = 1, P < 0⋅01; Fig. 2(d)]
© 2014 The Fisheries Society of the British Isles, Journal of Fish Biology 2014, 84, 1527–1538
1532 1·0
L . Z H A N G E T A L.
(a)
1·0
Predation mortality
0·9 0·8
(b)
1·0
0·9
*
0·8
(c)
1·0 0·9
0·9
**
*
0·8
0·7
0·7
0·7
0·6
0·6
0·6
0·6
0·5
0·5
0·5
0·5
0·4
0·4
0·4
0·4
0·3
0·3
0·3
0·3
0·2
0·2
0·2
0·2
0·1
0·1
0·1
0·1
0
0
0 T and fed NT and fed
*
0·8
0·7
T and fed NT and fed
(d)
0 T and fed NT and fed
T and fed NT and fed
Fig. 1. Effect of Cyprinus carpio genotype (i.e. transgenic, T or natural, NT) on predation vulnerability when either unfed for 24 h before the experiment or with unlimited feeding 30 min before the experiment. Experimental set-up included three Siniperca chuatsi of similar sizes, and either (a), (b) mixed schools (n = 60) of transgenic and non-transgenic individuals with different feeding status [(a) n = 4; (b): n = 5] or (c), (d) single genotyped schools (n = 60) also of different feeding status [(c) n = 5; (d) n = 4]. Mortalities from contrasting treatments were then analysed as 2 × 2 contingency tables (𝜒 2 -test in cross table) to test treatment effects. Significant differences in mortalities between treatments were tested using a 𝜒 2 -test (*P < 0⋅05 and **P < 0⋅01). Values are mean ± s.e.
and fed NT C. carpio [𝜒 2 = 19⋅546, d.f. = 1, P < 0⋅001; Fig. 2(c)]. Survival times of fed and unfed C. carpio indicated that consumption of meals reduced evasiveness in both genotypes (Fig. 2) and unfed T C. carpio were captured faster by the predators than fed NT C. carpio even though the difference between them was not significant [𝜒 2 = 1⋅821, d.f. = 1, P > 0⋅05; Fig. 2(e)]. The tL50 of the four treatment groups NT fed, T fed, NT unfed and T unfed was 20, 10, 50 and 30 min, respectively (Table I), with the tL50 of NT fed individuals being significantly lower than that of the unfed individuals (P < 0⋅001); the same for the T fed C. carpio (P < 0⋅001). Finally, the tL50 of T fed C. carpio was also significantly lower than that of the NT fed (P < 0⋅001) and the same for T unfed v. NT unfed C. carpio (P < 0⋅01). Analysis of failure times indicates that selection against C. carpio that display genetic potential for rapid growth could be at least partially explained by lower escape potential.
DISCUSSION The results clearly indicate a knock-on effect of the GH gene (gcGH) introduction into the C. carpio genome on their vulnerability to predation. The experiments unequivocally show that it is the genetic nature of the C. carpio rather than its size that affects the risk of predation. In addition, the satiety level of the C. carpio also increased the predation risk regardless of any specific genetic makeup. Thus, the study characterizes the existence of a trade-off between somatic growth and predator avoidance performance (Lankford et al., 2001; Li et al., 2007). Similar results were obtained in brown trout Salmo trutta L. 1758, in which simple injection of growth hormone influenced antipredator behaviour (Johnsson et al., 1996; Jönsson et al., 1998), making fast-growing S. trutta more risk-prone, significantly decreasing their critical swimming speeds and thus making them more vulnerable to predation (Farrell et al., 1997). Here,
© 2014 The Fisheries Society of the British Isles, Journal of Fish Biology 2014, 84, 1527–1538
1533
P R E D AT I O N R I S K I N G M C Y P R I N U S C A R P I O
(a)
(b)
(c)
(d)
1·0 0·8 0·6
Survival probability
0·4 0·2 0
1·0 0·8 0·6 0·4 0·2 0 0
20
40
60
80
100
10
20
30
40
50
60
70
80
(e) Survival probability
1·0 0·8 0·6 0·4 0·2 0 0
20
40
60
80
100
Survival time (min) Fig. 2. Survival curves of Cyprinus carpio genotype (i.e. transgenic, T or natural, i.e. non-transgenic, NT) on predation vulnerability when either unfed for 24 h before the experiment or with unlimited feeding 30 min before the experiment. In each of the four experiments, three Siniperca chuatsi of similar sizes were allowed to prey on a monotyped school (n = 20) of either transgenic fed (n = 4 replicates) or transgenic unfed (n = 4) or non-transgenic fed (n = 6) or non-transgenic unfed (n = 5) C. carpio. Comparison of the survival times between (a) T and fed ( ) and NT and fed ( ); (b) T and unfed ( ) and NT and unfed ( ); (c) NT and fed ( ) and NT and unfed ( ); (d) T and fed ( and T and unfed ( ) and (e) T and unfed ( ) and NT and fed ( ).
© 2014 The Fisheries Society of the British Isles, Journal of Fish Biology 2014, 84, 1527–1538
1534
L . Z H A N G E T A L.
the fed–unfed trials suggested that large meals represented a hidden additional cost of growth for the C. carpio. Previous research has argued that high rates of food consumption and rapid growth in fishes drive metabolism to its aerobic maximum, thereby producing a conflict with allocation of oxygen to other functions such as swimming (Priede, 1985; Wieser & Medgyesy, 1990; Wieser, 1991). Other studies on Salmo salar L. 1758 and Ictalurus punctatus (Rafinesque 1818) show that the overexpression in growth hormone-stimulated fishes has enhanced forage motivation, lowered resolution of the food selection, increased activity level and reduced clustering trends (Abrahams & Sutterlin, 1999; Dunham et al., 1999; Sundström et al., 2003, 2004a, b; Tymchuk & Devlin, 2005a, b). It is key to differentiate between farmed fishes and wild populations of C. carpio as both environments have separate selection pressures. Most research on growth-enhanced fishes, either through the use of growth hormones or the use of a modified genome, has an economic applied endpoint, which is to produce marketable fishes at a faster pace, thus lowering production cost. In such farm environments, fishes are not subjected to predation and thus, as seen in rainbow trout, Oncorhynchus mykiss (Walbaum 1792), for example, growth rates in domestic strains can reach three times those observed in wild strains (Johnsson et al., 1996). This differential in growth is achieved by a greater motivation to feed with elevated foraging rates and also increased aggression (Fleming et al., 2002). On the contrary, in natural environments, foraging competition is an important evolutionary pressure, so is predation. Thus, in the wild, the selection for rapid growth is counter selected by individuals with better predator avoidance capabilities (Arendt, 1997; Post & Parkinson, 2001; Munch & Conover, 2003; Stoks et al., 2005) and this trade-off explains the absence of individuals with maximized growth rates in nature (Werner & Gilliam, 1984; Werner & Anholt, 1993; Anholt & Werner, 1995; Gotthard, 2000; Biro et al., 2003a, b, 2004, 2006; Stoks et al., 2005). In light of the results of this study, the key environmental risk lies with the potential for transgenic fishes to escape aquaculture farms and contaminate wild natural stocks. Because of the relatively low economic value of C. carpio compared with salmonid species, their culture facilities tend to be less biosecure and often consist of earthen ponds located on the floodplain or fish cages in reservoirs, as these limit infrastructure costs. Such farming infrastructures are not closed systems and the use of transgenic Table I. Mean survival time and median lethal time (tL50 ) expressed as the survival time of 50% of a given school of Cyprinus carpio genotype (i.e. transgenic, T or natural, i.e. non-transgenic, NT) either unfed for 24 h before the experiment or with unlimited feeding 30 min before the experiment. The experimental set-up included three Siniperca chuatsi of similar size. The superscript lowercase letters in the median estimate column represent significant differences between the two values Mean Identification A B C D
Type
Median(tL50 )
Estimate s.e.
NT and fed NT and unfed T and unfed T and fed
27⋅40 49⋅44 35⋅83 20⋅47
2⋅98 4⋅67 3⋅85 2⋅33
95% c.l. 21⋅56–33⋅25 40⋅27–58⋅61 28⋅28–43⋅37 15⋅89–35⋅88
Estimate s.e. 20⋅0a 50⋅0b 30⋅0c 10⋅0d
95% c.l.
Z-value Pair
2⋅92 14⋅26–25⋅73
