Introduction

Predation exerts one of the greatest selective pressures on prey organisms particularly during the vulnerable early juvenile growth phases [1]. In the presence of this selective force, it should not be surprising to discover that many organisms display innate anti-predator responses to visual or olfactory predator cues. In many circumstances, such innate responses are then finely honed following exposure to predators either directly (individual learning; [2]) or via the observation of attacks on conspecifics (social learning; [3]). In aquatic ecosystems the presence of predators is often signalled by chemosensory cues that may take a number of forms. In the simplest form, prey may be able to detect odours emanating directly from the predator. Some chemical cues, however, may indicate the threat of predation indirectly. Alarm substances released from damaged conspecifics, for example, can also signal that a predation event has taken place. Numerous papers have shown that the presence of such cues and their relative concentration, signal that a predator is in the vicinity, and the cues can be used to predict future predator attack [4], [5]. Consequently prey show anti-predator responses such as hiding or schooling when they detect these cues.

Innate responses to predator cues have been shown in a number of organisms. Even after 15 generations of isolation from predators, steelhead trout, Oncorhychus mykiss, still responded to the odour emanating from their natural predators [6]. Predator recognition may also be indicated by subtle observation of fish behaviour and or numerous neurophysiological variables associated with the flight or fight response. Both naïve Atlantic salmon, Salmo salar, and Nile tilapia, Oreochromis niloticus, increase opercular beat rates in the presence of predator cues [7], [8]. Similarly, heart rate is significantly elevated following predator detection [9]. Appropriate physiological responses to predatory cues that differ from exposure to control cues suggest that animals can differentiate between these cues and thus recognize them. Whether the recognition system is cognitive or an innate reflex is often difficult to determine. Where graded responses are illicited to cues that vary in threat content, however, it is likely that cognitive processes are involved as the animal refers to innate or learned templates during the recognition process [1].

It has been suggested that animals may be able to detect and respond to chemical cues during early embryonic stages. Salmonids, for example, may begin to imprint on the chemical signature of their home stream in the final stages of embryogenesis [10]. Chickens exposed to certain odours whilst still in the egg, later show preferences for such odours post-hatching [11]. Moreover, both salamander and frogs exposed to predator cues as embryos show appropriate anti-predator responses as tadpoles upon encountering the cues again [12].

Detection of predator cues by embryos can also effect the timing of hatching. Detection of potential egg predators speeds up development and causes early hatching in amphibians [13], while detection of potential larvae predators causes a delay in hatching [14], [15]. To date, however, few studies have examined predator detection by fish embryos despite the fact that the egg membrane is highly permeable [16] and early detection of predators may significantly enhance survival. Here we exposed four-day-old rainbowfish embryos (Melanotaenia duboulayi) to a host of predator cues and examined their response by observing changes in heart rate and hatch time.

Methods

Results

Analysis of the heart-rate data showed a highly significant effect of cue (ANCOVA: F_{3, 75} = 14.989, P<0.001; Fig. 1). Post-hoc analysis revealed significant differences between all treatments (Fisher's PLSD: P<0.03 in all cases) with the exception of conspecific extract and goldfish odour (Fisher's PLSD: P>0.05). All odours elicited a faster embryonic heart rate relative to the control (synthetic water), with the native predator (silver perch) odour producing the greatest increase in heart rate. Heart rate significantly increased with petri-dish number indicating that the eggs became increasingly stressed as we repeatedly sampled different eggs from the mops over the course of the day (ANCOVA: F_{1, 75} = 20.548, P<0.001).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 1. Mean (±SE) heart rate (beats per minute) of rainbowfish embryos exposed to a range of chemical cues.

All cues induced a significant increase in heart rate relative to the control (distilled water).

https://doi.org/10.1371/journal.pone.0076061.g001

Examination of the hatch time data showed no differences between cues (ANOVA: F_{3, 84} = 0.592, P = 0.622) nor did handling influence hatching time (ANOVA: F_{1, 84} = 0.084, P = 0.773). In general, however, eggs exposed to predator cues tended to hatch slightly later than controls. Post-hoc analysis using a one-tailed t-tests based on the assumption that embryos should delay hatching when detecting larval predators suggested that the hatch date of eggs exposed to conspecific extract showed a marginally delayed hatching relative to the control eggs (t = 1.556, P = 0.063) and while both the other treatments showed similar trends they were not significant (goldfish: t = 1.302, P = 0.099; perch: t = 0.85, P = 0.199). Hatching success did not differ between treatments (ANOVA: F_{3, 24} = 0.929, P = 0.442; average 42.5%), however those eggs that were handled were less likely to hatch than those that were not (ANOVA: F_{1, 24} = 13.636, P = 0.001).

Discussion

Rainbowfish embryos can distinguish between chemical cues emanating from various potential predators and from alarms substances released from damage inflicted on conspecifics. While a substantial increase in heart rate was observed in response to a novel predator (goldfish) relative to control levels, the greatest response was to the native predator (spangled perch). The response to the conspecific extract was indistinguishable to that of the goldfish. Quite clearly these embryos have had no prior exposure to predators given that they were raised in isolated aquaria thus the recognition system must be entirely innate. What is more surprising is the fish that the eggs were derived from have been in captivity for multiple generations [20]. Similar observations have been made in juvenile steelhead trout that have been isolated from predators for 15 generations [6]. While previous comparative studies of the anti-predator behaviour of rainbowfish have shown that isolation from predators over geological time scales can result in naiveté [21], [22], evidently innate predator recognition systems can be relatively long lived even in the absence of direct selection.

While we observed no significant shift in hatch day in response to predator cues, there was a tendency for all treatments to delay hatching relative to controls. The difficulty we face, however, is that hatching success is relatively low (ca 42%) and the embryonic stage is very short (just 7 days at 26°C), thus in order to detect significant delays in hatch date we undoubtedly require more power. Such low hatching success is typical of rainbowfish where individual females spawn hundreds of eggs per week, but this was further exacerbated by handling the eggs during the experiment. Most of them succumb to fungal infections in the lab but the hatch rate in the wild is likely to be significantly lower. It is interesting to note, however, that the eggs tended to delay hatching after exposure to the predator cues which is what would be expected if the cues were emanating from larval rather than egg predators [13], [14], [15]. Larval rainbowfish are only 4–5 mm long when they hatch and they undoubtedly fall prey to a wide range of predators including small fish and invertebrates. Juvenile spangled perch and goldfish are both well known for their broad dietary niche and both attack and consume larval fishes when they encounter them.

While we have clearly shown that rainbowfish larvae can detect and differentiate between predator cues, it remains unknown what the longer-term effects of this early exposure might be. Research in amphibians suggest that exposure to predator cues during embryogenesis can lead to appropriate avoidance behaviour as larvae [12]. No such response was observed in Atlantic salmon fry when exposed to pike odours between 27 and 1 day pre-hatch [23]. However, there are bound to be a number of other physiological and behavioural costs associated with accelerating or decelerating development. Not least of which is the potential for developmental instability. Clearly heart rate increases in fish embryos during exposure to predator cues, perhaps an indication of underlying stress, which may affect hatching success and larval behaviour. Studies on Atalantic salmon have shown that maternal stress has great impact on key larval characteristics including reduced body size, yolk sac volume, and an increase in morphological malformations [24]. Future studies will need to pay close attention to these potential long-term impacts of predator exposure during embryogenesis.

Supporting Information

Acknowledgments

We’d like to thank the Department of Biological Sciences for their continued support for student projects.