- Email: [email protected]
The role of learning in the development of threat-sensitive predator avoidance by fathead minnows
ANIMAL BEHAVIOUR, 2005, 70, 777–784 doi:10.1016/j.anbehav.2005.01.009
The role of learning in the development of threat-sensitive predator avoidance by fathead minnows MAUD C. O. FERRARI *, J ENN IF ER J. TR OWELL*, GR AN T E. BROW N† & DOUG LAS P . CH IVERS * *Department
of Biology, University of Saskatchewan, Saskatoon, SK yDepartment of Biology, Concordia University, Montreal, QC
(Received 25 May 2004; initial acceptance 5 August 2004; final acceptance 4 January 2005; published online 9 August 2005; MS. number: A9899)
The ability to recognize a potential predator and display adaptive antipredator behaviour is crucial to the survival of prey animals. Prey should gain a fitness advantage by displaying antipredator responses with an intensity that matches their risk of predation. Understanding how such responses develop is the focus of our current study. Many prey fish do not show innate recognition of predators. Thus, learning is necessary and a strong selection pressure should exist to make learning as efficient as possible. In this study we investigated the ability of predator-naı¨ve fathead minnows, Pimephales promelas, to learn to recognize an unknown predator, brook charr, Salvelinus fontinalis. First, we conditioned minnows to recognize charr odour by exposing them to various concentrations of chemical alarm cues simultaneously with the odour of the charr and we subsequently tested them for recognition of this odour as a predation threat. Our objective was to test whether there was a match between the intensity of their antipredator responses during conditioning and recognition trials. Second, we tested whether minnows could learn to recognize charr through cultural transmission (i.e. by observing a conspecific responding to the odour of the predator) and we tested for a correlation between the intensity of response of the tutors during conditioning and the intensity of the learned responses by the observers during recognition trials. For both learning modes, the intensity of the response during the conditioning phase was retained during subsequent recognition trials. Our results suggest that minnows learn to respond more intensely to predation cues associated with high risk. Ó 2005 The Association for the Study of Animal Behaviour. Published by Elsevier Ltd. All rights reserved.
Predation represents a strong selection pressure acting on both morphology and behaviour of individuals. Prey that fail to respond appropriately to predators may lose their life. Consequently, it is not surprising that selection pressure acts to favour prey displaying adaptive responses. These behaviours can include local assessment of predation risk, by choosing low-predation habitats in which to forage and reproduce (Lima & Dill 1990). Indeed, prey should be selected if they are able to match the intensity of their antipredator response to the risk posed by their predators. Understanding how such threat-sensitive responses develop is the focus of our current study. Many prey species do not show an innate recognition of their potential predators (invertebrates: Rochette et al. 1998; birds: Curio et al. 1978; mammals: McLean et al. 1996; Griffin et al. 2001), including fish (Mathis et al. 1993;
Correspondence: M. C. O. Ferrari, Department of Biology, University of Saskatchewan, 112 Science Place, Saskatoon, SK S7N 5E2, Canada (email: [email protected]). 0003–3472/05/$30.00/0
Chivers & Smith 1994a; but see: Berejikian et al. 2003; Vilhunen & Hirvonen 2003), hence learning is necessary for them to acquire this information. By extension, learning should also be required to determine the degree of threat associated with different predators. Damage-released chemical alarm cues have been shown in a wide variety of organisms including protozoans, flatworms, annelids, arthropods, molluscs, fish and amphibians (Wisenden 2003). Among fish, members of the superorder Ostariophysi, which includes minnows, tetras and catfish, have received the most attention. However, alarm cues are also known in salmonids, gobies, poeciliids, sticklebacks, percids, sculpins, cottids, cichlids and centrarchids (review Chivers & Smith 1998; Brown 2003). These chemicals are located in the epidermis and are released in the water column through mechanical damage of the skin, typically following a predation attempt in which the prey is injured or captured. When detected by conspecifics (and some sympatric heterospecifics), these chemical alarm cues can elicit a dramatic and immediate increase in antipredator behaviours such as increased
777 Ó 2005 The Association for the Study of Animal Behaviour. Published by Elsevier Ltd. All rights reserved.
778
ANIMAL BEHAVIOUR, 70, 4
group cohesion, increased shelter use, decreased activity level and rapid escape to avoid areas where cues have been detected (review Chivers & Smith 1998). Two studies (Dupuch et al. 2004; Zhao & Chivers, in press) have shown that the intensity of antipredator response is proportional to the concentration of conspecific alarm cues. Chemical alarm cues are known to be important in facilitating learned recognition of predators in a variety of prey (flatworms: Wisenden & Millard 2001; snails: Rochette et al. 1998; amphibians: Woody & Mathis 1998). Fish acquire recognition of a novel predator based on the pairing of alarm cues with the visual and/or chemical cues of the predator (review Chivers & Smith 1998; Smith 1999) through a mechanism known as releaser-induced recognition learning. In this form of associative learning, alarm cues ‘release’ or facilitate the rapid acquisition of novel stimuli as future indicators of predation risk. For example, European minnows, Phoxinus phoxinus, and fathead minnows acquire the recognition of the chemical cues (odour) of a novel predator after a single exposure to the predator cue paired with conspecific alarm cues (Magurran 1989; Mathis & Smith 1993; Chivers & Smith 1994a). Chivers & Smith (1994b) showed that fathead minnows similarly learn to recognize the visual cues of a predator following a single conditioning experience. Learned predator recognition also occurs through cultural transmission. While much of the work has concentrated on birds (e.g. Curio et al. 1978; Curio 1988), there are few studies on fish. This mode of learning allows visual recognition (Magurran & Higham 1988; Kelley et al. 2003) as well as chemical recognition of a novel predator. Suboski et al. (1990) demonstrated that zebra danios, Brachydanio rerio, were able to recognize a synthetic chemical (morpholine) when they observed conspecifics showing a fright response in an adjacent tank. Mathis et al. (1996) demonstrated that pike-naı¨ve fathead minnows (observers) have the ability to learn to recognize the chemical cues of northern pike, Esox lucius, by observing the fright response of experienced conspecifics (tutors) in the same tank paired with pike odour. A single conditioning event was enough for the fish to learn the visual and/ or chemical identity of a previously novel predator. In this study, we investigated the ability of fathead minnows to learn the intensity of response associated with the risk of predation. In a first experiment, we conditioned predator-naı¨ve minnows to recognize brook charr, Salvelinus fontinalis, odour as a threat by exposing them to charr cues paired with various concentrations of conspecific alarm cues. We compared the intensity of the minnows’ response during conditioning to the intensity of their response when tested with charr odour alone in a subsequent recognition trial. We hypothesized that the intensity of minnows’ response during conditioning and recognition trials would vary according to the concentration of skin extract used for conditioning (i.e. that fish conditioned with higher concentrations of alarm cues would respond and learn to respond with a higher intensity to charr odour alone than fish conditioned with lower concentrations of alarm cues). The second experiment involved cultural transmission as a learning mode,
whereby we investigated whether minnows could learn to recognize charr by observing the fright response of conspecifics paired with charr odour. We tested for the existence of a correlation between the intensity of response displayed by the ‘tutors’ in the learning phase and the intensity of response displayed by the ‘observers’ when exposed to predator cues alone. We hypothesized that observers learning from tutors displaying a high intensity of antipredator response would respond with a higher intensity than observers learning from tutors displaying low intensities of response.
GENERAL METHODS
Test Fish Fathead minnows were captured from a local pond using minnow traps in October 2003. They were housed in a 6000-litre flow-through pool filled with dechlorinated tap water at 11 C and fed ad libitum once a day with commercial flakes. The photoperiod was adjusted to a 14:10 h light:dark cycle. Brook charr were obtained from the Fort Qu’Appelle fish hatchery, Saskatchewan, and were also housed in a 6000-litre flow-through pool filled with dechlorinated tap water at 11 C. Charr were fed ad libitum once a day with commercial charr pellets.
Stimulus Collection Minnow skin extract We collected skin extract from six fathead minnows (mean G SE fork length: 4.62 G 0.16 cm). Minnows were killed with a blow on the head (in accordance with the Canadian Council on Animal Care) and skin fillets were removed from either side of the body and placed in 100 ml of chilled distilled water. Skin fillets were then homogenized and filtered through glass wool to remove any remaining tissues. We collected a total of 21.96 cm2 of skin in a total of 439 ml of distilled water, which constituted our standardized stock solution containing w1 cm2 of skin per 20 ml of distilled water. This solution was diluted to make three experimental solutions: low (1 cm2 of skin per 240 litre), medium (1 cm2 of skin per 120 litre), and high (1 cm2 of skin per 40 litre) concentrations of alarm cues. Skin extracts were frozen in 20-ml aliquots at 20 C until required.
Charr odour Prey animals often show antipredator responses to chemical cues of predators fed conspecific of the prey, but not those fed another diet (Chivers & Mirza 2001). Thus, two arbitrarily chosen brook charr (20.6 and 24.5 cm fork length) were kept in a 115-litre tank and fed brine shrimp (Artemia spp.) for 5 days prior to stimulus collection. Both charr were then transferred to a 72-litre tank containing 60 litres of dechlorinated tap water and an air stone but no filter. The charr remained in the stimulus-collection tank for 24 h, at which time they were transferred back to their initial holding pond. Water
FERRARI ET AL.: LEARNED PREDATOR RECOGNITION
containing charr odour was frozen at 20 C in 60-ml aliquots until required.
Experiment 1: Conditioned Learning by Exposure to Alarm Cues This experiment consisted of two phases: conditioning trials followed by recognition trials. During conditioning trials, three minnows were exposed to charr odour paired with one of the three concentrations of conspecific alarm cues (low, medium or high) or a distilled water control. The fish were exposed to charr odour alone 24 h later during recognition trials and the intensities of their responses during recognition trials were compared to the intensity of response displayed during their conditioning.
Conditioning trials Groups of three minnows were placed in 37-litre tanks, filled with dechlorinated tap water. Each tank had a 3 ! 3grid pattern drawn on the side and contained a gravel substrate and an air stone, near to which we attached a 2m-long piece of plastic tube to inject test stimuli into the tanks. Prior to testing, minnows were acclimatized for a 24-h period in their testing tanks (water at 12–14 C, same photoperiod as mentioned above). We tested 180 minnows (mean G SE fork length: 4.73 G 0.04 cm) in total with N Z 15 per treatment. All trials were conducted between 1330 and 1600 hours. Observations consisted of an 8-min prestimulus and an 8min poststimulus injection period. Prior to the prestimulus period, we withdrew and discarded 60 ml of water from the injection tubes (to remove any stagnant water) and then withdrew and retained an additional 60 ml. Following the prestimulus period, we injected either 5 ml of one of the three concentrations of skin extract or distilled water as well as 20 ml of charr odour into the tank. We used the retained tank water to slowly flush the stimuli into the tank. Once the stimuli were fully injected, we began the poststimulus observation period. As measures of antipredator response, we measured the shoaling index of the three fish every 15 s (1: no fish within a body length of another; 2: two fish within a body length of each other; 3: all the fish within a body length of each other). During the first 8 s of the 15-s periods (a stopwatch was set to beep after 8 s), the number of line crosses (using the 3 ! 3-grid pattern drawn on the side of the tank) was also recorded for one of the three minnows (randomly chosen, the same fish was observed until the end of the conditioning period). An increase in shoaling index and a decrease in activity level are two typical antipredator responses in minnows (reviewed in Chivers & Smith 1998).
Recognition trials An hour after the end of the conditioning trials, the fish were moved to a similar 37-litre tank filled with clean dechlorinated tap water. After 24 h, recognition trials were performed. The protocol was the same as that used for the
conditioning trials except that only charr odour (20 ml) was injected in the tank after the prestimulus period.
Statistical analysis For both conditioning and recognition trials, we calculated the change in line crosses and in shoaling from the prestimulus baseline. A reduction in line crosses as well as an increase in shoaling index would indicate an increase in antipredator behaviour. Our data met the assumptions of the nonparametric tests (normal distribution of the data but nonhomogeneity of the variances between treatments); hence, we conducted two-tailed Mann–Whitney tests to compare the effect of different concentrations of alarm cues during the conditioning and recognition trials (Siegel & Castellan 1988). The family-wise error rate was assessed and controlled for using the modified Bonferroni test following Keppel (1991). The modified Bonferroni test specifies that corrections to the family-wise error rate are introduced only when the number of comparisons exceeds k 1, where k is the number of treatments (Keppel 1991). The modified alpha value is obtained by the formula: alpha (0.05) times the degrees of freedom associated with treatment conditions (4 1 Z 3), divided by the number of comparisons (6). Thus, the rejection probability was set at 0.025 for each comparison. Wilcoxon signed-rank tests were performed to compare the response between conditioning and recognition trials within treatments.
Experiment 2: Conditioned Learning by Cultural Transmission This experiment also consisted of conditioning trials followed by recognition trials. During the conditioning trials, a single observer fish was exposed simultaneously to charr odour and to the sight of three conspecific tutors in an adjacent tank displaying different intensities of antipredator response. The tutors used in this experiment were the fish that were exposed to the various concentrations of alarm cues paired with charr odour in experiment 1. The conditioning trials for both experiments were performed simultaneously. The experimental set-up consisted of paired 37-litre tanks put side by side and separated with a one-way mirror (Fig. 1). The observer tanks were identical to the tutor tanks; they were equipped with a gravel substrate, an air stone and an injection tube. A light source was placed above the tutor tanks, with a black plastic partition preventing the light from reaching the observer tank. The difference in luminosity between the two tanks made it easier for the observer fish to look at the tutors through the one-way mirror. Moreover, the tutors could not see the observer, so their behaviours were not influenced by the observer fish. We chose to set up a single observer to maximize the effect of the tutors on the observer.
Conditioning trials Prior to conditioning, observers were acclimatized for a 24-h period in their respective tanks (water at 12–14 C,
779
780
ANIMAL BEHAVIOUR, 70, 4
Light source
Predator odour
Alarm cue Predator odour
Air
Air Observer
Tutors
Air stone
One-way mirror Figure 1. Schematic diagram (side view) of test tanks used in experiments 1 and 2.
same photoperiod as mentioned before). We conditioned 60 observers (mean G SE fork length: 4.73 G 0.07 cm), each of which was paired with a group of three tutor fish from experiment 1. During the conditioning trials, we simultaneously introduced 20 ml of charr odour into the observer tanks and skin extract and charr odour into the tutor tanks. We did not quantify the behaviour of the observer fish during the conditioning trials. Indeed, any behavioural responses from the observer at that time would be due to social facilitation and not a result of learning. To ensure that the observers had acquired recognition of charr odour as a threat, we needed to test them alone with the predator odour in subsequent trials.
Recognition trials An hour after the end of the conditioning trials, the observers were transferred into other 37-litre tanks filled with clean dechlorinated tap water. The tanks contained a gravel substrate, an air stone, an injection tube and a shelter that consisted of a 10 ! 20-cm ceramic tile mounted on three 3.5-cm-long cylindrical glass legs. After 24 h, recognition trials were performed. The same experimental procedure as the conditioning trials was used. After the prestimulus period, 20 ml of charr odour was injected in the observer tank. The behavioural measures recorded were time spent moving and time spent under shelter.
Statistical analysis For both conditioning and recognition trials, we calculated the change in time spent moving and in time spent under shelter from the prestimulus baseline. A reduction in time spent moving as well as an increase in shelter use would indicate an increase in antipredator response. Treatments were compared using two-tailed Mann–Whitney tests. As with experiment 1, the rejection probability (P) was set at 0.025 for each comparison.
To examine whether the intensity of the tutors’ response was correlated with the intensity of the observers’ response during the recognition trials, we ranked tutor responses from 1 to 60 (1 Z lowest response; 60 Z highest response) for both change in line crosses and shoaling index, and observer responses from 1 to 60 for change in time spent moving and under shelter. Then, each of the experimental pairs was analysed with a Spearman rank correlation.
RESULTS
Experiment 1: Conditioned Learning by Exposure to Alarm Cues Conditioning trials The intensity of the behavioural responses of minnows to alarm cues varied with alarm cue concentrations. There was no difference in the response of minnows to distilled water and low concentration of skin extract (Mann– Whitney U test: line crosses: U Z 93.5, N1 Z N2 Z 15, P Z 0.436; Fig. 2a; shoaling index: U Z 112.5, N1 Z N2 Z 15, P Z 1.0; Fig. 2b). However, fish exposed to the medium concentration showed an increase in shoaling (U Z 1.5, N1 Z N2 Z 15, P ! 0.001) and a reduction in line crosses (U Z 40.5, N1 Z N2 Z 15, P Z 0.02) compared with the distilled water control. Similarly, fish exposed to high concentration of skin extract showed an increase in shoaling (U Z 3.0, N1 Z N2 Z 15, P ! 0.001) and a reduction in line crosses (U Z 20.5, N1 Z N2 Z 15, P ! 0.001). The low-concentration treatment was significantly different from the medium (U Z 27.5, N1 Z N2 Z 15, P ! 0.001; U Z 4.5, P ! 0.001) and high ones (U Z 2.5, N1 Z N2 Z 15, P ! 0.001; U Z 1.0, P ! 0.001) for line crosses and shoaling, respectively. For change in line crosses, the response to medium concentration of skin extract was significantly different from the high
FERRARI ET AL.: LEARNED PREDATOR RECOGNITION
(a)
Change in line crosses
10 0 –10 –20 –30 –40 –50 –60 –70 DW
Low
Medium
High
For change in shoaling index, the low-concentration treatment was significantly different from the medium(U Z 44.0, N1 Z N2 Z 15, P Z 0.004; Fig. 2b) and highconcentration treatments (U Z 35.5, N1 Z N2 Z 15, P Z 0.001). Responses of minnows conditioned with a high concentration were not significantly higher than those conditioned with the medium concentration (U Z 79.5, N1 Z N2 Z 15, P Z 0.174). For change in line crosses, low and medium, medium and high, low and high were not significantly different from each other (U Z 93.0, N1 Z N2 Z 15, P Z 0.436; U Z 78.0, P Z 0.161; U Z 63.5, P Z 0.041, respectively; Fig. 2a), however, a pattern is clearly observable on the graph.
(b) Change in shoaling index
1.4 1.2
Comparisons within treatments
1 0.8 0.6 0.4 0.2 0 DW
Low
Medium
High
Figure 2. Mean G SE change in (a) number of line crosses and (b) shoaling index for fathead minnows exposed to different concentrations of conspecific skin extract or distilled water (DW) paired with charr odour during conditioning trials (,), or exposed to charr odour alone during recognition trials (-) (N Z 15/treatment).
No difference between conditioning and recognition was found for distilled water (Wilcoxon signed-ranks test: line crosses: Z Z 1.93, N Z 15, P Z 0.233; shoaling index: Z Z 0.221, N Z 15, P Z 0.825), medium (line crosses: Z Z 1.051, N Z 15, P Z 0.293; shoaling index: Z Z 0.171, N Z 15, P Z 0.865) or high (line crosses: Z Z 1.079, N Z 15, P Z 0.281; shoaling index: Z Z 0.682, N Z 15, P Z 0.496) treatments. However, differences were found for the low treatment for line crosses (Z Z 2.215, N Z 15, P Z 0.027) and shoaling index (Z Z 3.352, N Z 15, P Z 0.001) (Fig. 2a, b).
Recognition trials The responses of minnows to charr odour in the recognition trials was influenced by the cues the fish were exposed to during the conditioning trials. Minnows initially exposed to the low concentration of skin extract paired with charr odour displayed a significant fright response to charr odour when compared to those initially exposed to distilled water combined with charr odour (line crosses: U Z 41.5, N1 Z N2 Z 15, P Z 0.002; Fig. 2a; shoaling index: U Z 16.0, N1 Z N2 Z 15, P ! 0.001; Fig. 2b). Moreover, responses to charr odour for fish in the control treatment were significantly different from those initially exposed to charr odour combined with medium and high concentrations for line crosses (U Z 17.0, N1 Z N2 Z 15, P ! 0.001; U Z 16.5, P ! 0.001) and shoaling (U Z 2.5, N1 Z N2 Z 15, P ! 0.001; U Z 0.0, P ! 0.001). These results show that minnows learned to recognize charr odour as a threat when they were conditioned with each of the three concentrations of skin extract.
0 –50 –100 –150 –200 –250 DW
Low
Medium
High
DW
Low
Medium
High
(b) Change in shelter use (s)
concentration (U Z 54.5, N1 Z N2 Z 15, P Z 0.015). However, no difference was found for shoaling index (U Z 64.5, N1 Z N2 Z 15, P Z 0.045) but a pattern is clearly observable on the graph.
Change in time moving (s)
(a)
200 150 100 50 0
Figure 3. Mean G SE change in time spent (a) moving and (b) under shelter for observer fathead minnows conditioned with tutors exposed to different concentrations of skin extract or a distilled water (DW) control (N Z 15/treatment).
781
ANIMAL BEHAVIOUR, 70, 4
Experiment 2: Conditioned Learning by Cultural Transmission
Spearman correlations We found significant correlations (all P ! 0.001) between the responses of the tutors during conditioning trials and the responses of the observers during recognition trials for each of the measured response variables. The correlation coefficient (rS) was 0.688 for change in shoaling index of the tutors and change in time moving of the observers (Fig. 4a). Similar correlation coefficients were observed for other pairs (0.546 for shoaling index versus shelter use, Fig. 4b; 0.708 for line crosses versus time moving, Fig. 4c; 0.567 for line crosses versus shelter use, Fig. 4d).
DISCUSSION The results of our experiments demonstrate that minnows can learn to recognize the identity of unknown predators through conditioning with alarm cues and through cultural transmission. More interestingly, minnows also learned predator recognition in a threat-sensitive manner. They matched the intensity of their behavioural response in the conditioning and recognition trials. Prey individuals capable of adjusting the intensity of their antipredator response to match their risk should be at a selective advantage (Chivers et al. 2001). In experiment 1, fish seemed to increase their intensity of response when exposed to increased concentrations of alarm cues. This result suggests the existence of a graded response for minnows exposed to a concentration gradient of natural alarm cues. We use the term ‘graded’ to express the existence of a correlation between the intensity of response and the concentrations used. Graded does not necessarily mean that the relationship between the two factors is linear. Similar graded responses were found in goldfish, Carassius auratus (Zhao & Chivers, in press) and redbelly dace, Phoxinus eos (Dupuch et al. 2004) exposed to a concentration gradient of conspecific alarm cues. This type of response appears to be adaptive because it allows fish to match the intensity of antipredator response to the predation risk that they are exposed
Observer rank for change in time moving
60
(a)
rS = 0.688
50 40 30 20 10 0 0
10 20 30 40 50 60 Tutor rank for change in shoaling index
70 Observer rank for change in shelter use
For both change in time spent moving (Fig. 3a) and under shelter (Fig. 3b), minnows that had the opportunity to learn from tutors that were exposed to low concentrations of alarm cues did not respond differently from fish that had the opportunity to learn from tutors that were exposed to distilled water (U Z 99.0, N1 Z N2 Z 15, P Z 0.595; U Z 100.0, P Z 0.624). However, response of minnows having the opportunity to learn from tutors exposed to medium concentrations (U Z 29.5, N1 Z N2 Z 15, P ! 0.001; U Z 52.0, P Z 0.011) and high concentrations of skin extract (U Z 3.0, N1 Z N2 Z 15, P ! 0.001; U Z 29.5, P ! 0.001) were significantly different from those having the opportunity to learn from tutors exposed to distilled water.
60
(b)
70
rS = 0.549
50 40 30 20 10 0 0
10 20 30 40 50 60 Tutor rank for change in shoaling index
70
70 Observer rank for change in time moving
Comparisons between treatments
70
60
(c)
rS = 0.708
50 40 30 20 10 0 0
10 20 30 40 50 60 Tutor rank for change in line crosses
70
70 Observer rank for change in shelter use
782
60
(d)
rS = 0.567
50 40 30 20 10 0 0
10 20 30 40 50 60 Tutor rank for change in line crosses
70
Figure 4. Correlation between tutor rank during conditioning and observer rank during recognition trials for change in shoaling versus change in time spent moving (a), change in shoaling versus change in shelter use (b), change in line crosses versus change in time spent moving (c) and change in line crosses versus change in shelter use (d). N Z 60 for both tutors and observers.
FERRARI ET AL.: LEARNED PREDATOR RECOGNITION
to, if the concentration indeed reflects predation risk. In contrast, Brown et al. (2001a) exposed fathead minnows to a concentration gradient of hypoxanthine-3-N-oxide (H3NO), the artificial ostariophysan alarm cue, and found a nongraded response. Differences in these results may be explained by differential response of minnows to natural and artificial alarm pheromone, or they may simply reflect differences between populations, or between the body condition or parasite load of the fish, for instance. During recognition trials, fish initially conditioned with higher concentrations of alarm cues displayed stronger fright responses to charr odour than did fish conditioned with lower concentrations. These data suggest that, in general, the graded response is conserved when fish are subsequently exposed to predator cues alone. That is, the fish not only acquired predator recognition during conditioning trials, but also learned to associate a given concentration of predator cues with a specific level of risk. Interestingly, fish exposed to the lowest alarm cue concentration in the conditioning trials did not show overt responses but still acquired the recognition of charr as a predator. A similar learning effect with a subthreshold concentration of H3NO was demonstrated by Brown et al. (2001b). Even if fish do not display overt antipredator behaviour, this subthreshold concentration increases their vigilance and reliance on secondary visual cues (Brown & Magnavacca 2003; Brown et al. 2004). In experiment 2, observer fish simultaneously exposed to charr odour and to tutors given distilled water or subthreshold concentrations of alarm cues did not show any fright response when subsequently exposed to charr odour alone. This suggests that observers may not learn from tutors that do not display overt antipredator responses. The correlations clearly showed that observers appeared to match their intensity of response to the intensity of response displayed by their respective tutors. This is the first experiment demonstrating that the intensity of a fright response of fish can be culturally transmitted. Similar to the previous experiment, minnows appeared to acquire the recognition of the predator as well as the level of risk associated with the predator cues. Although this phenomenon has not been investigated in predator avoidance learning in fish, a positive correlation between the acquired alarm behaviour of observers and that of tutors during training has been demonstrated in other taxonomic groups, including birds and mammals (reviewed in Griffin 2004). Prey animals often show threat-sensitive predator avoidance, adjusting the intensity of their antipredator behaviour to match their level of risk. Such threatsensitive responses are often based on the size of the predator (Chivers et al. 2001). Often, prey that are smaller have a reduced ability to escape and hence are more vulnerable to a predator than are larger prey. Consequently, prey need to be cautious to learn from appropriate tutors (i.e. those that perceive risk in the same manner). If prey learn from other individuals that do not share the same risk assessment, they may learn to overestimate or underestimate the risk associated with a given predator.
For prey, the ability to recognize predators using chemical cues offers the advantage that the cues can be detected from a distance. Chemical cues become less concentrated the further they travel in the medium. Consequently, the concentration of those chemicals can be a good indicator of the distance of the odour source. Minnows seem to innately adjust the intensity of their antipredator behaviours using the concentration of conspecific alarm cues present in the water column, responding less to lower concentrations. Thus, alarm cue concentration appears to give minnows an indication of the level of predation risk (Dupuch et al. 2004). Kusch et al. (2004) also demonstrated that experienced fathead minnows could determine the degree of threat posed by northern pike based on the concentration of chemical predator cues used. These results follow the threat-sensitive predator avoidance hypothesis (Helfman 1989). Prey should display appropriate responses to different degrees of predation threat, thereby optimizing their time and energy by avoiding costly antipredator behaviours. Although it is easy to see how different concentrations of alarm cues or predator odours might correspond to different levels of risk in the laboratory, where water flows are controlled and predictable, it would be interesting to investigate how unpredictable flows, typically found in streams for instance, affect the ability of fish to associate a specific concentration of alarm cue or predator odour to a specific level of threat. Exposure to alarm cues and cultural transmission are two conditioned-learning modes that can be used by prey to acquire recognition of a novel predator. However, predators do not often allow a second chance for their prey. Because learned predator recognition increases survival, learning mechanisms are subject to strong selection pressure, and acquiring as much information as possible in a short time becomes crucial. We already knew that a single exposure to conspecific alarm cue paired with predator odour is enough for the fish to learn (reviewed in: Chivers & Smith 1998; Smith 1999). However, the results of our study suggest that a single exposure is enough for threat-sensitive responses to be acquired through these two learning modes. Being able to display the ‘right’ behaviours without having been in direct contact with the predator seems a real advantage for prey species. Even though a single exposure is enough to learn the recognition of a novel predator as well as the intensity of response associated with the predation risk, learning should occur continuously, so that every experience sharpens the antipredator response. More work needs to be done to understand how multiple learning experiences affect antipredator response and survival of prey animals. Acknowledgments Natural Sciences and Engineering Research Council (Canada) and University of Saskatchewan provided financial support to D. P. Chivers. All work reported herein was in accordance with University of Saskatchewan Committee of Animal Care and Supply protocol no. 19920077. We thank the editor and the two referees for their useful comments.
783
784
ANIMAL BEHAVIOUR, 70, 4
References Berejikian, B. A., Tezaka, E. P. & LaRaeb, A. L. 2003. Innate and enhanced predator recognition in hatchery-reared chinook salmon. Environmental Biology of Fishes, 67, 241–251. Brown, G. E. 2003. Learning about danger: chemical alarm cues and local risk assessment. Fish and Fisheries, 4, 227–234. Brown, G. E. & Magnavacca, G. 2003. Predator inspection behaviour in a characin fish: an interaction between chemical and visual information? Ethology, 109, 739–750. Brown, G. E., Adrian, J. C., Jr & Shih, M. L. 2001a. Behavioural responses of fathead minnows to hypoxanthine-3-N-oxide at varying concentrations. Journal of Fish Biology, 58, 1465–1470. Brown, G. E., Adrian, J. C., Jr, Patton, T. & Chivers, D. P. 2001b. Fathead minnows learn to recognize predator odour when exposed to concentrations of artificial alarm pheromone below their behavioural-response threshold. Canadian Journal of Zoology, 79, 2239–2245. Brown, G. E., Poirier, J.-F. & Adrian, J. C., Jr. 2004. Assessment of local predation risk: the role of sub-threshold concentrations of chemical alarm cues. Behavioral Ecology, 15, 810–815. Chivers, D. P. & Mirza, R. S. 2001. Predator diet cues and the assessment of predation risk by aquatic vertebrates: a review and prospectus. In: Chemical Signals in Vertebrates. Vol. 9 (Ed. by D. A. Marchlewska-Koj, J. J. Lepri & D. Mu ¨ ller-Schwarze), pp. 277–284. New York: Plenum. Chivers, D. P. & Smith, R. J. F. 1994a. The role of experience and chemical alarm signalling in predator recognition by fathead minnows, Pimephales promelas. Journal of Fish Biology, 44, 273–285. Chivers, D. P. & Smith, R. J. F. 1994b. Fathead minnows, Pimephales promelas, acquired predator recognition when alarm substance is associated with the sight of unfamiliar fish. Animal Behaviour, 48, 597–605. Chivers, D. P. & Smith, R. J. F. 1998. Chemical alarm signalling in aquatic predator–prey systems: a review and prospectus. Ecoscience, 5, 338–352. Chivers, D. P., Mirza, R. S., Bryer, P. J. & Kiesecker, J. M. 2001. Threat-sensitive predator avoidance by slimy sculpins: understanding the importance of visual versus chemical information. Canadian Journal of Zoology, 79, 867–873. Curio, E. 1988. Cultural transmission of enemy recognition by birds. In: Social Learning: Psychological and Biological Perspectives (Ed. by T. R. Zentall & B. G. Galef, Jr), pp. 75–97. Hillsdale, New Jersey: L. Erlbaum. Curio, E., Ernst, U. & Vieth, W. 1978. Cultural transmission of enemy recognition: one function of mobbing. Science, 202, 899– 901. Dupuch, A., Magnan, P. & Dill, L. M. 2004. Sensitivity of northern redbelly dace, Phoxinus eos, to chemical alarm cues. Canadian Journal of Zoology, 82, 407–415. Griffin, A. S. 2004. Social learning about predators: a review and prospectus. Learning & Behavior, 32, 131–140. Griffin, A. S., Evans, C. S. & Blumstein, D. T. 2001. Learning specificity in acquired predator recognition. Animal Behaviour, 62, 577–589. Helfman, G. S. 1989. Threat-sensitive predator avoidance in damselfish–trumpetfish interactions. Behavioral Ecology and Sociobiology, 24, 47–58. Kelley, J. L., Evans, J. P., Ramnarine, I. W. & Magurran, A. E. 2003. Back to school: can antipredator behaviour in guppies be enhanced through social learning? Animal Behaviour, 65, 655–662.
Keppel, G. 1991. Design and Analysis: a Researcher’s Handbook. 3rd edn. Engelwood Cliffs, New Jersey: Prentice Hall. Kusch, R. C., Mirza, R. S. & Chivers, D. P. 2004. Making sense of the predator scents: investigating the sophistication of predator assessment abilities of fathead minnows. Behavioral Ecology and Sociobiology, 55, 551–555. Lima, S. L. & Dill, L. M. 1990. Behavioral decisions made under the risk of predation: a review and prospectus. Canadian Journal of Zoology, 68, 619–640. McLean, I. G., Lundie-Jenkins, G. & Jarman, P. J. 1996. Teaching an endangered mammal to recognise predators. Biological Conservation, 56, 51–62. Magurran, A. E. 1989. Acquired recognition of predator odour in the European minnows (Phoxinus phoxinus). Ethology, 82, 216– 233. Magurran, A. E. & Higham, A. 1988. Information transfer across fish shoals under threat. Ethology, 78, 153–158. Mathis, A. & Smith, R. J. F. 1993. Chemical alarm signals increase the survival time in fathead minnows (Pimephales promelas) during encounters with northern pike (Esox lucius). Behavioral Ecology, 4, 260–265. Mathis, A., Chivers, D. P. & Smith, R. J. F. 1993. Population differences in responses of fathead minnows (Pimephales promelas) to visual and chemical stimuli from predators. Ethology, 93, 31–40. Mathis, A., Chivers, D. P. & Smith, R. J. F. 1996. Cultural transmission of predator recognition in fish: intraspecific and interspecific learning. Animal Behaviour, 51, 185–201. Rochette, R., Arsenault, D. J., Justome, B. & Himmelman, J. H. 1998. Chemically-mediated predator recognition learning in a marine gastropod. Ecoscience, 5, 353–360. Siegel, S. & Castellan, N. J. 1988. Nonparametric Statistics for the Behavioural Sciences. 2nd edn. New York: McGraw-Hill. Smith, R. J. F. 1999. What good is smelly stuff in the skin? Cross function and cross taxa effects in fish ‘alarm substances’. In: Advances in Chemical Signals in Vertebrates (Ed. by R. E. Johnston, D. Mu ¨ ller-Schwarze & P. W. Sorensen), pp. 475–488. New York: Kluwer Academic. Suboski, M. D., Bain, S., Carty, A. E., McQuoid, L. M., Seelen, M. I. & Seifert, M. 1990. Alarm reaction in acquisition and social transmission of simulated predator recognition by zebra danio fish (Brachydanio rerio). Journal of Comparative Psychology, 104, 101– 112. Vilhunen, S. & Hirvonen, H. 2003. Innate antipredator responses of Arctic charr (Salvelinus alpinus) depend on predator species and their diet. Behavioral Ecology and Sociobiology, 55, 1–10. Wisenden, B. D. 2003. Chemically-mediated strategies to counter predation. In: Sensory Processing in the Aquatic Environment (Ed. by S. P. Collin & N. J. Marshall), pp. 236–251. New York: SpringerVerlag. Wisenden, B. D. & Millard, M. C. 2001. Aquatic flatworms use chemical cues from injured conspecifics to assess predation risk and to associate risk with novel cues. Animal Behaviour, 62, 761– 766. Woody, D. R. & Mathis, A. 1998. Acquired recognition of chemical stimuli from an unfamiliar predator: associative learning by adult newts, Notophthalmus viridescens. Copeia, 1998, 1027–1031. Zhao, X. & Chivers, D. P. In press. Response of juvenile goldfish (Carassius auratus) to chemical cues: relationship between response intensity, response duration and the level of predation risk. In: Chemical Signals in Vertebrates. Vol. 10 (Ed. by R. T. Mason, M. LeMaster & D. Mu ¨ ller-Scharze). New York: Plenum.
The role of learning in the development of threat-sensitive predator avoidance by fathead minnows MAUD C. O. FERRARI *, J ENN IF ER J. TR OWELL*, GR AN T E. BROW N† & DOUG LAS P . CH IVERS * *Department
of Biology, University of Saskatchewan, Saskatoon, SK yDepartment of Biology, Concordia University, Montreal, QC
(Received 25 May 2004; initial acceptance 5 August 2004; final acceptance 4 January 2005; published online 9 August 2005; MS. number: A9899)
The ability to recognize a potential predator and display adaptive antipredator behaviour is crucial to the survival of prey animals. Prey should gain a fitness advantage by displaying antipredator responses with an intensity that matches their risk of predation. Understanding how such responses develop is the focus of our current study. Many prey fish do not show innate recognition of predators. Thus, learning is necessary and a strong selection pressure should exist to make learning as efficient as possible. In this study we investigated the ability of predator-naı¨ve fathead minnows, Pimephales promelas, to learn to recognize an unknown predator, brook charr, Salvelinus fontinalis. First, we conditioned minnows to recognize charr odour by exposing them to various concentrations of chemical alarm cues simultaneously with the odour of the charr and we subsequently tested them for recognition of this odour as a predation threat. Our objective was to test whether there was a match between the intensity of their antipredator responses during conditioning and recognition trials. Second, we tested whether minnows could learn to recognize charr through cultural transmission (i.e. by observing a conspecific responding to the odour of the predator) and we tested for a correlation between the intensity of response of the tutors during conditioning and the intensity of the learned responses by the observers during recognition trials. For both learning modes, the intensity of the response during the conditioning phase was retained during subsequent recognition trials. Our results suggest that minnows learn to respond more intensely to predation cues associated with high risk. Ó 2005 The Association for the Study of Animal Behaviour. Published by Elsevier Ltd. All rights reserved.
Predation represents a strong selection pressure acting on both morphology and behaviour of individuals. Prey that fail to respond appropriately to predators may lose their life. Consequently, it is not surprising that selection pressure acts to favour prey displaying adaptive responses. These behaviours can include local assessment of predation risk, by choosing low-predation habitats in which to forage and reproduce (Lima & Dill 1990). Indeed, prey should be selected if they are able to match the intensity of their antipredator response to the risk posed by their predators. Understanding how such threat-sensitive responses develop is the focus of our current study. Many prey species do not show an innate recognition of their potential predators (invertebrates: Rochette et al. 1998; birds: Curio et al. 1978; mammals: McLean et al. 1996; Griffin et al. 2001), including fish (Mathis et al. 1993;
Correspondence: M. C. O. Ferrari, Department of Biology, University of Saskatchewan, 112 Science Place, Saskatoon, SK S7N 5E2, Canada (email: [email protected]). 0003–3472/05/$30.00/0
Chivers & Smith 1994a; but see: Berejikian et al. 2003; Vilhunen & Hirvonen 2003), hence learning is necessary for them to acquire this information. By extension, learning should also be required to determine the degree of threat associated with different predators. Damage-released chemical alarm cues have been shown in a wide variety of organisms including protozoans, flatworms, annelids, arthropods, molluscs, fish and amphibians (Wisenden 2003). Among fish, members of the superorder Ostariophysi, which includes minnows, tetras and catfish, have received the most attention. However, alarm cues are also known in salmonids, gobies, poeciliids, sticklebacks, percids, sculpins, cottids, cichlids and centrarchids (review Chivers & Smith 1998; Brown 2003). These chemicals are located in the epidermis and are released in the water column through mechanical damage of the skin, typically following a predation attempt in which the prey is injured or captured. When detected by conspecifics (and some sympatric heterospecifics), these chemical alarm cues can elicit a dramatic and immediate increase in antipredator behaviours such as increased
777 Ó 2005 The Association for the Study of Animal Behaviour. Published by Elsevier Ltd. All rights reserved.
778
ANIMAL BEHAVIOUR, 70, 4
group cohesion, increased shelter use, decreased activity level and rapid escape to avoid areas where cues have been detected (review Chivers & Smith 1998). Two studies (Dupuch et al. 2004; Zhao & Chivers, in press) have shown that the intensity of antipredator response is proportional to the concentration of conspecific alarm cues. Chemical alarm cues are known to be important in facilitating learned recognition of predators in a variety of prey (flatworms: Wisenden & Millard 2001; snails: Rochette et al. 1998; amphibians: Woody & Mathis 1998). Fish acquire recognition of a novel predator based on the pairing of alarm cues with the visual and/or chemical cues of the predator (review Chivers & Smith 1998; Smith 1999) through a mechanism known as releaser-induced recognition learning. In this form of associative learning, alarm cues ‘release’ or facilitate the rapid acquisition of novel stimuli as future indicators of predation risk. For example, European minnows, Phoxinus phoxinus, and fathead minnows acquire the recognition of the chemical cues (odour) of a novel predator after a single exposure to the predator cue paired with conspecific alarm cues (Magurran 1989; Mathis & Smith 1993; Chivers & Smith 1994a). Chivers & Smith (1994b) showed that fathead minnows similarly learn to recognize the visual cues of a predator following a single conditioning experience. Learned predator recognition also occurs through cultural transmission. While much of the work has concentrated on birds (e.g. Curio et al. 1978; Curio 1988), there are few studies on fish. This mode of learning allows visual recognition (Magurran & Higham 1988; Kelley et al. 2003) as well as chemical recognition of a novel predator. Suboski et al. (1990) demonstrated that zebra danios, Brachydanio rerio, were able to recognize a synthetic chemical (morpholine) when they observed conspecifics showing a fright response in an adjacent tank. Mathis et al. (1996) demonstrated that pike-naı¨ve fathead minnows (observers) have the ability to learn to recognize the chemical cues of northern pike, Esox lucius, by observing the fright response of experienced conspecifics (tutors) in the same tank paired with pike odour. A single conditioning event was enough for the fish to learn the visual and/ or chemical identity of a previously novel predator. In this study, we investigated the ability of fathead minnows to learn the intensity of response associated with the risk of predation. In a first experiment, we conditioned predator-naı¨ve minnows to recognize brook charr, Salvelinus fontinalis, odour as a threat by exposing them to charr cues paired with various concentrations of conspecific alarm cues. We compared the intensity of the minnows’ response during conditioning to the intensity of their response when tested with charr odour alone in a subsequent recognition trial. We hypothesized that the intensity of minnows’ response during conditioning and recognition trials would vary according to the concentration of skin extract used for conditioning (i.e. that fish conditioned with higher concentrations of alarm cues would respond and learn to respond with a higher intensity to charr odour alone than fish conditioned with lower concentrations of alarm cues). The second experiment involved cultural transmission as a learning mode,
whereby we investigated whether minnows could learn to recognize charr by observing the fright response of conspecifics paired with charr odour. We tested for the existence of a correlation between the intensity of response displayed by the ‘tutors’ in the learning phase and the intensity of response displayed by the ‘observers’ when exposed to predator cues alone. We hypothesized that observers learning from tutors displaying a high intensity of antipredator response would respond with a higher intensity than observers learning from tutors displaying low intensities of response.
GENERAL METHODS
Test Fish Fathead minnows were captured from a local pond using minnow traps in October 2003. They were housed in a 6000-litre flow-through pool filled with dechlorinated tap water at 11 C and fed ad libitum once a day with commercial flakes. The photoperiod was adjusted to a 14:10 h light:dark cycle. Brook charr were obtained from the Fort Qu’Appelle fish hatchery, Saskatchewan, and were also housed in a 6000-litre flow-through pool filled with dechlorinated tap water at 11 C. Charr were fed ad libitum once a day with commercial charr pellets.
Stimulus Collection Minnow skin extract We collected skin extract from six fathead minnows (mean G SE fork length: 4.62 G 0.16 cm). Minnows were killed with a blow on the head (in accordance with the Canadian Council on Animal Care) and skin fillets were removed from either side of the body and placed in 100 ml of chilled distilled water. Skin fillets were then homogenized and filtered through glass wool to remove any remaining tissues. We collected a total of 21.96 cm2 of skin in a total of 439 ml of distilled water, which constituted our standardized stock solution containing w1 cm2 of skin per 20 ml of distilled water. This solution was diluted to make three experimental solutions: low (1 cm2 of skin per 240 litre), medium (1 cm2 of skin per 120 litre), and high (1 cm2 of skin per 40 litre) concentrations of alarm cues. Skin extracts were frozen in 20-ml aliquots at 20 C until required.
Charr odour Prey animals often show antipredator responses to chemical cues of predators fed conspecific of the prey, but not those fed another diet (Chivers & Mirza 2001). Thus, two arbitrarily chosen brook charr (20.6 and 24.5 cm fork length) were kept in a 115-litre tank and fed brine shrimp (Artemia spp.) for 5 days prior to stimulus collection. Both charr were then transferred to a 72-litre tank containing 60 litres of dechlorinated tap water and an air stone but no filter. The charr remained in the stimulus-collection tank for 24 h, at which time they were transferred back to their initial holding pond. Water
FERRARI ET AL.: LEARNED PREDATOR RECOGNITION
containing charr odour was frozen at 20 C in 60-ml aliquots until required.
Experiment 1: Conditioned Learning by Exposure to Alarm Cues This experiment consisted of two phases: conditioning trials followed by recognition trials. During conditioning trials, three minnows were exposed to charr odour paired with one of the three concentrations of conspecific alarm cues (low, medium or high) or a distilled water control. The fish were exposed to charr odour alone 24 h later during recognition trials and the intensities of their responses during recognition trials were compared to the intensity of response displayed during their conditioning.
Conditioning trials Groups of three minnows were placed in 37-litre tanks, filled with dechlorinated tap water. Each tank had a 3 ! 3grid pattern drawn on the side and contained a gravel substrate and an air stone, near to which we attached a 2m-long piece of plastic tube to inject test stimuli into the tanks. Prior to testing, minnows were acclimatized for a 24-h period in their testing tanks (water at 12–14 C, same photoperiod as mentioned above). We tested 180 minnows (mean G SE fork length: 4.73 G 0.04 cm) in total with N Z 15 per treatment. All trials were conducted between 1330 and 1600 hours. Observations consisted of an 8-min prestimulus and an 8min poststimulus injection period. Prior to the prestimulus period, we withdrew and discarded 60 ml of water from the injection tubes (to remove any stagnant water) and then withdrew and retained an additional 60 ml. Following the prestimulus period, we injected either 5 ml of one of the three concentrations of skin extract or distilled water as well as 20 ml of charr odour into the tank. We used the retained tank water to slowly flush the stimuli into the tank. Once the stimuli were fully injected, we began the poststimulus observation period. As measures of antipredator response, we measured the shoaling index of the three fish every 15 s (1: no fish within a body length of another; 2: two fish within a body length of each other; 3: all the fish within a body length of each other). During the first 8 s of the 15-s periods (a stopwatch was set to beep after 8 s), the number of line crosses (using the 3 ! 3-grid pattern drawn on the side of the tank) was also recorded for one of the three minnows (randomly chosen, the same fish was observed until the end of the conditioning period). An increase in shoaling index and a decrease in activity level are two typical antipredator responses in minnows (reviewed in Chivers & Smith 1998).
Recognition trials An hour after the end of the conditioning trials, the fish were moved to a similar 37-litre tank filled with clean dechlorinated tap water. After 24 h, recognition trials were performed. The protocol was the same as that used for the
conditioning trials except that only charr odour (20 ml) was injected in the tank after the prestimulus period.
Statistical analysis For both conditioning and recognition trials, we calculated the change in line crosses and in shoaling from the prestimulus baseline. A reduction in line crosses as well as an increase in shoaling index would indicate an increase in antipredator behaviour. Our data met the assumptions of the nonparametric tests (normal distribution of the data but nonhomogeneity of the variances between treatments); hence, we conducted two-tailed Mann–Whitney tests to compare the effect of different concentrations of alarm cues during the conditioning and recognition trials (Siegel & Castellan 1988). The family-wise error rate was assessed and controlled for using the modified Bonferroni test following Keppel (1991). The modified Bonferroni test specifies that corrections to the family-wise error rate are introduced only when the number of comparisons exceeds k 1, where k is the number of treatments (Keppel 1991). The modified alpha value is obtained by the formula: alpha (0.05) times the degrees of freedom associated with treatment conditions (4 1 Z 3), divided by the number of comparisons (6). Thus, the rejection probability was set at 0.025 for each comparison. Wilcoxon signed-rank tests were performed to compare the response between conditioning and recognition trials within treatments.
Experiment 2: Conditioned Learning by Cultural Transmission This experiment also consisted of conditioning trials followed by recognition trials. During the conditioning trials, a single observer fish was exposed simultaneously to charr odour and to the sight of three conspecific tutors in an adjacent tank displaying different intensities of antipredator response. The tutors used in this experiment were the fish that were exposed to the various concentrations of alarm cues paired with charr odour in experiment 1. The conditioning trials for both experiments were performed simultaneously. The experimental set-up consisted of paired 37-litre tanks put side by side and separated with a one-way mirror (Fig. 1). The observer tanks were identical to the tutor tanks; they were equipped with a gravel substrate, an air stone and an injection tube. A light source was placed above the tutor tanks, with a black plastic partition preventing the light from reaching the observer tank. The difference in luminosity between the two tanks made it easier for the observer fish to look at the tutors through the one-way mirror. Moreover, the tutors could not see the observer, so their behaviours were not influenced by the observer fish. We chose to set up a single observer to maximize the effect of the tutors on the observer.
Conditioning trials Prior to conditioning, observers were acclimatized for a 24-h period in their respective tanks (water at 12–14 C,
779
780
ANIMAL BEHAVIOUR, 70, 4
Light source
Predator odour
Alarm cue Predator odour
Air
Air Observer
Tutors
Air stone
One-way mirror Figure 1. Schematic diagram (side view) of test tanks used in experiments 1 and 2.
same photoperiod as mentioned before). We conditioned 60 observers (mean G SE fork length: 4.73 G 0.07 cm), each of which was paired with a group of three tutor fish from experiment 1. During the conditioning trials, we simultaneously introduced 20 ml of charr odour into the observer tanks and skin extract and charr odour into the tutor tanks. We did not quantify the behaviour of the observer fish during the conditioning trials. Indeed, any behavioural responses from the observer at that time would be due to social facilitation and not a result of learning. To ensure that the observers had acquired recognition of charr odour as a threat, we needed to test them alone with the predator odour in subsequent trials.
Recognition trials An hour after the end of the conditioning trials, the observers were transferred into other 37-litre tanks filled with clean dechlorinated tap water. The tanks contained a gravel substrate, an air stone, an injection tube and a shelter that consisted of a 10 ! 20-cm ceramic tile mounted on three 3.5-cm-long cylindrical glass legs. After 24 h, recognition trials were performed. The same experimental procedure as the conditioning trials was used. After the prestimulus period, 20 ml of charr odour was injected in the observer tank. The behavioural measures recorded were time spent moving and time spent under shelter.
Statistical analysis For both conditioning and recognition trials, we calculated the change in time spent moving and in time spent under shelter from the prestimulus baseline. A reduction in time spent moving as well as an increase in shelter use would indicate an increase in antipredator response. Treatments were compared using two-tailed Mann–Whitney tests. As with experiment 1, the rejection probability (P) was set at 0.025 for each comparison.
To examine whether the intensity of the tutors’ response was correlated with the intensity of the observers’ response during the recognition trials, we ranked tutor responses from 1 to 60 (1 Z lowest response; 60 Z highest response) for both change in line crosses and shoaling index, and observer responses from 1 to 60 for change in time spent moving and under shelter. Then, each of the experimental pairs was analysed with a Spearman rank correlation.
RESULTS
Experiment 1: Conditioned Learning by Exposure to Alarm Cues Conditioning trials The intensity of the behavioural responses of minnows to alarm cues varied with alarm cue concentrations. There was no difference in the response of minnows to distilled water and low concentration of skin extract (Mann– Whitney U test: line crosses: U Z 93.5, N1 Z N2 Z 15, P Z 0.436; Fig. 2a; shoaling index: U Z 112.5, N1 Z N2 Z 15, P Z 1.0; Fig. 2b). However, fish exposed to the medium concentration showed an increase in shoaling (U Z 1.5, N1 Z N2 Z 15, P ! 0.001) and a reduction in line crosses (U Z 40.5, N1 Z N2 Z 15, P Z 0.02) compared with the distilled water control. Similarly, fish exposed to high concentration of skin extract showed an increase in shoaling (U Z 3.0, N1 Z N2 Z 15, P ! 0.001) and a reduction in line crosses (U Z 20.5, N1 Z N2 Z 15, P ! 0.001). The low-concentration treatment was significantly different from the medium (U Z 27.5, N1 Z N2 Z 15, P ! 0.001; U Z 4.5, P ! 0.001) and high ones (U Z 2.5, N1 Z N2 Z 15, P ! 0.001; U Z 1.0, P ! 0.001) for line crosses and shoaling, respectively. For change in line crosses, the response to medium concentration of skin extract was significantly different from the high
FERRARI ET AL.: LEARNED PREDATOR RECOGNITION
(a)
Change in line crosses
10 0 –10 –20 –30 –40 –50 –60 –70 DW
Low
Medium
High
For change in shoaling index, the low-concentration treatment was significantly different from the medium(U Z 44.0, N1 Z N2 Z 15, P Z 0.004; Fig. 2b) and highconcentration treatments (U Z 35.5, N1 Z N2 Z 15, P Z 0.001). Responses of minnows conditioned with a high concentration were not significantly higher than those conditioned with the medium concentration (U Z 79.5, N1 Z N2 Z 15, P Z 0.174). For change in line crosses, low and medium, medium and high, low and high were not significantly different from each other (U Z 93.0, N1 Z N2 Z 15, P Z 0.436; U Z 78.0, P Z 0.161; U Z 63.5, P Z 0.041, respectively; Fig. 2a), however, a pattern is clearly observable on the graph.
(b) Change in shoaling index
1.4 1.2
Comparisons within treatments
1 0.8 0.6 0.4 0.2 0 DW
Low
Medium
High
Figure 2. Mean G SE change in (a) number of line crosses and (b) shoaling index for fathead minnows exposed to different concentrations of conspecific skin extract or distilled water (DW) paired with charr odour during conditioning trials (,), or exposed to charr odour alone during recognition trials (-) (N Z 15/treatment).
No difference between conditioning and recognition was found for distilled water (Wilcoxon signed-ranks test: line crosses: Z Z 1.93, N Z 15, P Z 0.233; shoaling index: Z Z 0.221, N Z 15, P Z 0.825), medium (line crosses: Z Z 1.051, N Z 15, P Z 0.293; shoaling index: Z Z 0.171, N Z 15, P Z 0.865) or high (line crosses: Z Z 1.079, N Z 15, P Z 0.281; shoaling index: Z Z 0.682, N Z 15, P Z 0.496) treatments. However, differences were found for the low treatment for line crosses (Z Z 2.215, N Z 15, P Z 0.027) and shoaling index (Z Z 3.352, N Z 15, P Z 0.001) (Fig. 2a, b).
Recognition trials The responses of minnows to charr odour in the recognition trials was influenced by the cues the fish were exposed to during the conditioning trials. Minnows initially exposed to the low concentration of skin extract paired with charr odour displayed a significant fright response to charr odour when compared to those initially exposed to distilled water combined with charr odour (line crosses: U Z 41.5, N1 Z N2 Z 15, P Z 0.002; Fig. 2a; shoaling index: U Z 16.0, N1 Z N2 Z 15, P ! 0.001; Fig. 2b). Moreover, responses to charr odour for fish in the control treatment were significantly different from those initially exposed to charr odour combined with medium and high concentrations for line crosses (U Z 17.0, N1 Z N2 Z 15, P ! 0.001; U Z 16.5, P ! 0.001) and shoaling (U Z 2.5, N1 Z N2 Z 15, P ! 0.001; U Z 0.0, P ! 0.001). These results show that minnows learned to recognize charr odour as a threat when they were conditioned with each of the three concentrations of skin extract.
0 –50 –100 –150 –200 –250 DW
Low
Medium
High
DW
Low
Medium
High
(b) Change in shelter use (s)
concentration (U Z 54.5, N1 Z N2 Z 15, P Z 0.015). However, no difference was found for shoaling index (U Z 64.5, N1 Z N2 Z 15, P Z 0.045) but a pattern is clearly observable on the graph.
Change in time moving (s)
(a)
200 150 100 50 0
Figure 3. Mean G SE change in time spent (a) moving and (b) under shelter for observer fathead minnows conditioned with tutors exposed to different concentrations of skin extract or a distilled water (DW) control (N Z 15/treatment).
781
ANIMAL BEHAVIOUR, 70, 4
Experiment 2: Conditioned Learning by Cultural Transmission
Spearman correlations We found significant correlations (all P ! 0.001) between the responses of the tutors during conditioning trials and the responses of the observers during recognition trials for each of the measured response variables. The correlation coefficient (rS) was 0.688 for change in shoaling index of the tutors and change in time moving of the observers (Fig. 4a). Similar correlation coefficients were observed for other pairs (0.546 for shoaling index versus shelter use, Fig. 4b; 0.708 for line crosses versus time moving, Fig. 4c; 0.567 for line crosses versus shelter use, Fig. 4d).
DISCUSSION The results of our experiments demonstrate that minnows can learn to recognize the identity of unknown predators through conditioning with alarm cues and through cultural transmission. More interestingly, minnows also learned predator recognition in a threat-sensitive manner. They matched the intensity of their behavioural response in the conditioning and recognition trials. Prey individuals capable of adjusting the intensity of their antipredator response to match their risk should be at a selective advantage (Chivers et al. 2001). In experiment 1, fish seemed to increase their intensity of response when exposed to increased concentrations of alarm cues. This result suggests the existence of a graded response for minnows exposed to a concentration gradient of natural alarm cues. We use the term ‘graded’ to express the existence of a correlation between the intensity of response and the concentrations used. Graded does not necessarily mean that the relationship between the two factors is linear. Similar graded responses were found in goldfish, Carassius auratus (Zhao & Chivers, in press) and redbelly dace, Phoxinus eos (Dupuch et al. 2004) exposed to a concentration gradient of conspecific alarm cues. This type of response appears to be adaptive because it allows fish to match the intensity of antipredator response to the predation risk that they are exposed
Observer rank for change in time moving
60
(a)
rS = 0.688
50 40 30 20 10 0 0
10 20 30 40 50 60 Tutor rank for change in shoaling index
70 Observer rank for change in shelter use
For both change in time spent moving (Fig. 3a) and under shelter (Fig. 3b), minnows that had the opportunity to learn from tutors that were exposed to low concentrations of alarm cues did not respond differently from fish that had the opportunity to learn from tutors that were exposed to distilled water (U Z 99.0, N1 Z N2 Z 15, P Z 0.595; U Z 100.0, P Z 0.624). However, response of minnows having the opportunity to learn from tutors exposed to medium concentrations (U Z 29.5, N1 Z N2 Z 15, P ! 0.001; U Z 52.0, P Z 0.011) and high concentrations of skin extract (U Z 3.0, N1 Z N2 Z 15, P ! 0.001; U Z 29.5, P ! 0.001) were significantly different from those having the opportunity to learn from tutors exposed to distilled water.
60
(b)
70
rS = 0.549
50 40 30 20 10 0 0
10 20 30 40 50 60 Tutor rank for change in shoaling index
70
70 Observer rank for change in time moving
Comparisons between treatments
70
60
(c)
rS = 0.708
50 40 30 20 10 0 0
10 20 30 40 50 60 Tutor rank for change in line crosses
70
70 Observer rank for change in shelter use
782
60
(d)
rS = 0.567
50 40 30 20 10 0 0
10 20 30 40 50 60 Tutor rank for change in line crosses
70
Figure 4. Correlation between tutor rank during conditioning and observer rank during recognition trials for change in shoaling versus change in time spent moving (a), change in shoaling versus change in shelter use (b), change in line crosses versus change in time spent moving (c) and change in line crosses versus change in shelter use (d). N Z 60 for both tutors and observers.
FERRARI ET AL.: LEARNED PREDATOR RECOGNITION
to, if the concentration indeed reflects predation risk. In contrast, Brown et al. (2001a) exposed fathead minnows to a concentration gradient of hypoxanthine-3-N-oxide (H3NO), the artificial ostariophysan alarm cue, and found a nongraded response. Differences in these results may be explained by differential response of minnows to natural and artificial alarm pheromone, or they may simply reflect differences between populations, or between the body condition or parasite load of the fish, for instance. During recognition trials, fish initially conditioned with higher concentrations of alarm cues displayed stronger fright responses to charr odour than did fish conditioned with lower concentrations. These data suggest that, in general, the graded response is conserved when fish are subsequently exposed to predator cues alone. That is, the fish not only acquired predator recognition during conditioning trials, but also learned to associate a given concentration of predator cues with a specific level of risk. Interestingly, fish exposed to the lowest alarm cue concentration in the conditioning trials did not show overt responses but still acquired the recognition of charr as a predator. A similar learning effect with a subthreshold concentration of H3NO was demonstrated by Brown et al. (2001b). Even if fish do not display overt antipredator behaviour, this subthreshold concentration increases their vigilance and reliance on secondary visual cues (Brown & Magnavacca 2003; Brown et al. 2004). In experiment 2, observer fish simultaneously exposed to charr odour and to tutors given distilled water or subthreshold concentrations of alarm cues did not show any fright response when subsequently exposed to charr odour alone. This suggests that observers may not learn from tutors that do not display overt antipredator responses. The correlations clearly showed that observers appeared to match their intensity of response to the intensity of response displayed by their respective tutors. This is the first experiment demonstrating that the intensity of a fright response of fish can be culturally transmitted. Similar to the previous experiment, minnows appeared to acquire the recognition of the predator as well as the level of risk associated with the predator cues. Although this phenomenon has not been investigated in predator avoidance learning in fish, a positive correlation between the acquired alarm behaviour of observers and that of tutors during training has been demonstrated in other taxonomic groups, including birds and mammals (reviewed in Griffin 2004). Prey animals often show threat-sensitive predator avoidance, adjusting the intensity of their antipredator behaviour to match their level of risk. Such threatsensitive responses are often based on the size of the predator (Chivers et al. 2001). Often, prey that are smaller have a reduced ability to escape and hence are more vulnerable to a predator than are larger prey. Consequently, prey need to be cautious to learn from appropriate tutors (i.e. those that perceive risk in the same manner). If prey learn from other individuals that do not share the same risk assessment, they may learn to overestimate or underestimate the risk associated with a given predator.
For prey, the ability to recognize predators using chemical cues offers the advantage that the cues can be detected from a distance. Chemical cues become less concentrated the further they travel in the medium. Consequently, the concentration of those chemicals can be a good indicator of the distance of the odour source. Minnows seem to innately adjust the intensity of their antipredator behaviours using the concentration of conspecific alarm cues present in the water column, responding less to lower concentrations. Thus, alarm cue concentration appears to give minnows an indication of the level of predation risk (Dupuch et al. 2004). Kusch et al. (2004) also demonstrated that experienced fathead minnows could determine the degree of threat posed by northern pike based on the concentration of chemical predator cues used. These results follow the threat-sensitive predator avoidance hypothesis (Helfman 1989). Prey should display appropriate responses to different degrees of predation threat, thereby optimizing their time and energy by avoiding costly antipredator behaviours. Although it is easy to see how different concentrations of alarm cues or predator odours might correspond to different levels of risk in the laboratory, where water flows are controlled and predictable, it would be interesting to investigate how unpredictable flows, typically found in streams for instance, affect the ability of fish to associate a specific concentration of alarm cue or predator odour to a specific level of threat. Exposure to alarm cues and cultural transmission are two conditioned-learning modes that can be used by prey to acquire recognition of a novel predator. However, predators do not often allow a second chance for their prey. Because learned predator recognition increases survival, learning mechanisms are subject to strong selection pressure, and acquiring as much information as possible in a short time becomes crucial. We already knew that a single exposure to conspecific alarm cue paired with predator odour is enough for the fish to learn (reviewed in: Chivers & Smith 1998; Smith 1999). However, the results of our study suggest that a single exposure is enough for threat-sensitive responses to be acquired through these two learning modes. Being able to display the ‘right’ behaviours without having been in direct contact with the predator seems a real advantage for prey species. Even though a single exposure is enough to learn the recognition of a novel predator as well as the intensity of response associated with the predation risk, learning should occur continuously, so that every experience sharpens the antipredator response. More work needs to be done to understand how multiple learning experiences affect antipredator response and survival of prey animals. Acknowledgments Natural Sciences and Engineering Research Council (Canada) and University of Saskatchewan provided financial support to D. P. Chivers. All work reported herein was in accordance with University of Saskatchewan Committee of Animal Care and Supply protocol no. 19920077. We thank the editor and the two referees for their useful comments.
783
784
ANIMAL BEHAVIOUR, 70, 4
References Berejikian, B. A., Tezaka, E. P. & LaRaeb, A. L. 2003. Innate and enhanced predator recognition in hatchery-reared chinook salmon. Environmental Biology of Fishes, 67, 241–251. Brown, G. E. 2003. Learning about danger: chemical alarm cues and local risk assessment. Fish and Fisheries, 4, 227–234. Brown, G. E. & Magnavacca, G. 2003. Predator inspection behaviour in a characin fish: an interaction between chemical and visual information? Ethology, 109, 739–750. Brown, G. E., Adrian, J. C., Jr & Shih, M. L. 2001a. Behavioural responses of fathead minnows to hypoxanthine-3-N-oxide at varying concentrations. Journal of Fish Biology, 58, 1465–1470. Brown, G. E., Adrian, J. C., Jr, Patton, T. & Chivers, D. P. 2001b. Fathead minnows learn to recognize predator odour when exposed to concentrations of artificial alarm pheromone below their behavioural-response threshold. Canadian Journal of Zoology, 79, 2239–2245. Brown, G. E., Poirier, J.-F. & Adrian, J. C., Jr. 2004. Assessment of local predation risk: the role of sub-threshold concentrations of chemical alarm cues. Behavioral Ecology, 15, 810–815. Chivers, D. P. & Mirza, R. S. 2001. Predator diet cues and the assessment of predation risk by aquatic vertebrates: a review and prospectus. In: Chemical Signals in Vertebrates. Vol. 9 (Ed. by D. A. Marchlewska-Koj, J. J. Lepri & D. Mu ¨ ller-Schwarze), pp. 277–284. New York: Plenum. Chivers, D. P. & Smith, R. J. F. 1994a. The role of experience and chemical alarm signalling in predator recognition by fathead minnows, Pimephales promelas. Journal of Fish Biology, 44, 273–285. Chivers, D. P. & Smith, R. J. F. 1994b. Fathead minnows, Pimephales promelas, acquired predator recognition when alarm substance is associated with the sight of unfamiliar fish. Animal Behaviour, 48, 597–605. Chivers, D. P. & Smith, R. J. F. 1998. Chemical alarm signalling in aquatic predator–prey systems: a review and prospectus. Ecoscience, 5, 338–352. Chivers, D. P., Mirza, R. S., Bryer, P. J. & Kiesecker, J. M. 2001. Threat-sensitive predator avoidance by slimy sculpins: understanding the importance of visual versus chemical information. Canadian Journal of Zoology, 79, 867–873. Curio, E. 1988. Cultural transmission of enemy recognition by birds. In: Social Learning: Psychological and Biological Perspectives (Ed. by T. R. Zentall & B. G. Galef, Jr), pp. 75–97. Hillsdale, New Jersey: L. Erlbaum. Curio, E., Ernst, U. & Vieth, W. 1978. Cultural transmission of enemy recognition: one function of mobbing. Science, 202, 899– 901. Dupuch, A., Magnan, P. & Dill, L. M. 2004. Sensitivity of northern redbelly dace, Phoxinus eos, to chemical alarm cues. Canadian Journal of Zoology, 82, 407–415. Griffin, A. S. 2004. Social learning about predators: a review and prospectus. Learning & Behavior, 32, 131–140. Griffin, A. S., Evans, C. S. & Blumstein, D. T. 2001. Learning specificity in acquired predator recognition. Animal Behaviour, 62, 577–589. Helfman, G. S. 1989. Threat-sensitive predator avoidance in damselfish–trumpetfish interactions. Behavioral Ecology and Sociobiology, 24, 47–58. Kelley, J. L., Evans, J. P., Ramnarine, I. W. & Magurran, A. E. 2003. Back to school: can antipredator behaviour in guppies be enhanced through social learning? Animal Behaviour, 65, 655–662.
Keppel, G. 1991. Design and Analysis: a Researcher’s Handbook. 3rd edn. Engelwood Cliffs, New Jersey: Prentice Hall. Kusch, R. C., Mirza, R. S. & Chivers, D. P. 2004. Making sense of the predator scents: investigating the sophistication of predator assessment abilities of fathead minnows. Behavioral Ecology and Sociobiology, 55, 551–555. Lima, S. L. & Dill, L. M. 1990. Behavioral decisions made under the risk of predation: a review and prospectus. Canadian Journal of Zoology, 68, 619–640. McLean, I. G., Lundie-Jenkins, G. & Jarman, P. J. 1996. Teaching an endangered mammal to recognise predators. Biological Conservation, 56, 51–62. Magurran, A. E. 1989. Acquired recognition of predator odour in the European minnows (Phoxinus phoxinus). Ethology, 82, 216– 233. Magurran, A. E. & Higham, A. 1988. Information transfer across fish shoals under threat. Ethology, 78, 153–158. Mathis, A. & Smith, R. J. F. 1993. Chemical alarm signals increase the survival time in fathead minnows (Pimephales promelas) during encounters with northern pike (Esox lucius). Behavioral Ecology, 4, 260–265. Mathis, A., Chivers, D. P. & Smith, R. J. F. 1993. Population differences in responses of fathead minnows (Pimephales promelas) to visual and chemical stimuli from predators. Ethology, 93, 31–40. Mathis, A., Chivers, D. P. & Smith, R. J. F. 1996. Cultural transmission of predator recognition in fish: intraspecific and interspecific learning. Animal Behaviour, 51, 185–201. Rochette, R., Arsenault, D. J., Justome, B. & Himmelman, J. H. 1998. Chemically-mediated predator recognition learning in a marine gastropod. Ecoscience, 5, 353–360. Siegel, S. & Castellan, N. J. 1988. Nonparametric Statistics for the Behavioural Sciences. 2nd edn. New York: McGraw-Hill. Smith, R. J. F. 1999. What good is smelly stuff in the skin? Cross function and cross taxa effects in fish ‘alarm substances’. In: Advances in Chemical Signals in Vertebrates (Ed. by R. E. Johnston, D. Mu ¨ ller-Schwarze & P. W. Sorensen), pp. 475–488. New York: Kluwer Academic. Suboski, M. D., Bain, S., Carty, A. E., McQuoid, L. M., Seelen, M. I. & Seifert, M. 1990. Alarm reaction in acquisition and social transmission of simulated predator recognition by zebra danio fish (Brachydanio rerio). Journal of Comparative Psychology, 104, 101– 112. Vilhunen, S. & Hirvonen, H. 2003. Innate antipredator responses of Arctic charr (Salvelinus alpinus) depend on predator species and their diet. Behavioral Ecology and Sociobiology, 55, 1–10. Wisenden, B. D. 2003. Chemically-mediated strategies to counter predation. In: Sensory Processing in the Aquatic Environment (Ed. by S. P. Collin & N. J. Marshall), pp. 236–251. New York: SpringerVerlag. Wisenden, B. D. & Millard, M. C. 2001. Aquatic flatworms use chemical cues from injured conspecifics to assess predation risk and to associate risk with novel cues. Animal Behaviour, 62, 761– 766. Woody, D. R. & Mathis, A. 1998. Acquired recognition of chemical stimuli from an unfamiliar predator: associative learning by adult newts, Notophthalmus viridescens. Copeia, 1998, 1027–1031. Zhao, X. & Chivers, D. P. In press. Response of juvenile goldfish (Carassius auratus) to chemical cues: relationship between response intensity, response duration and the level of predation risk. In: Chemical Signals in Vertebrates. Vol. 10 (Ed. by R. T. Mason, M. LeMaster & D. Mu ¨ ller-Scharze). New York: Plenum.