Learned Recognition of Heterospecific Alarm Signals: The Importance ...

Ethology 107, 1007Ð1018 (2001) Ó 2001 Blackwell Wissenschafts-Verlag, Berlin ISSN 0179±1613

Department of Biology, University of Saskatchewan, Saskatoon

Learned Recognition of Heterospeci®c Alarm Signals: The Importance of a Mixed Predator Diet Reehan S. Mirza & Douglas P. Chivers Mirza, R. S. & Chivers, D. P. 2001: Learned recognition of heterospeci®c alarm signals: the importance of a mixed predator diet. Ethology 107, 1007Ð1018.

Abstract A wide diversity of aquatic organisms release alarm signals upon being attacked by a predator. Alarm signals may `warn' nearby individuals of danger. Moreover, the signals may be important in facilitating learned recognition of unknown stimuli. It is common for dierent prey species to respond to each other's chemical alarm signals. In many cases, the responses are learned but no learning mechanisms have been identi®ed to date. In this study we tested whether prey ®sh can learn the identity of an unknown alarm signal when they detect it in association with conspeci®c alarm cues in the diet of a predator. Chemical alarm cues are known to be conserved in the diet of predators. We conditioned fathead minnows (Pimephales promelas) with chemical stimuli from predatory yellow perch (Perca ¯avescens) fed a mixed diet of minnows and brook stickleback (Culaea inconstans), perch fed a mixed diet of swordtails (Xiphophorus helleri) and stickleback or distilled water. Minnows were subsequently exposed to chemical alarm cues of injured stickleback alone. Those minnows previously conditioned with perch fed a mixed diet of minnows and stickleback increased their use of shelter and `froze' signi®cantly more than minnows previously conditioned with perch fed a diet of swordtails and stickleback or those exposed to distilled water. These data demonstrate a mechanism by which minnows can learn the identity of a heterospeci®c alarm signal. Corresponding author: Reehan S. Mirza, Department of Biology, 112 Science Place, University of Saskatchewan, Saskatoon, SK, Canada S7N 5E2, E-mail: [email protected] Introduction Predation is a strong selective force that shapes many behavioural, life historical and morphological traits in prey animals (Sih 1987; Lima & Dill 1990; U. S. Copyright Clearance Center Code Statement: 0179-1613/2001/10711±1007$15.00/0 www.blackwell.de/synergy

1008

R. S. Mirza & D. P. Chivers

Chivers & Smith 1998). Prey animals are able to assess predation risk by using cues available to them in their environments. These cues may be visual, chemical, electrical or mechanical in nature. Chemosensory assessment of predation risk is of prime importance when visual cues are limited (Smith 1992). A wide diversity of aquatic animals release chemical cues when captured by a predator (reviewed in Smith 1992; Chivers & Smith 1998). Although controversy exists concerning the function of these chemical cues (Magurran et al. 1996; Smith 1997; Chivers & Smith 1998; Brown & Godin 1999), several studies report that prey animals respond to them with antipredator behaviour. The chemical cues may serve as alarm signals to `warn' nearby conspeci®cs of potential danger and may provide a survival bene®t for receivers (Mathis & Smith 1993a; Wisenden et al. 1999; Mirza & Chivers, in press). Prey animals often respond to chemical alarm cues released from other species (Mathis & Smith 1993b; Chivers et al. 1995a, b; Brown & Godin 1997). The ability of receivers to respond to chemical alarm cues from conspeci®cs and heterospeci®cs allows ¯exibility in their antipredator responses. Recognizing dierent alarm cues from both conspeci®cs and heterospeci®cs gives the prey animal more information to assess predation risk. Cross-species reactions to heterospeci®c alarm cues may result from partial conservation of chemical cues within the same taxonomic group. Alternatively, these responses may be learned since many of these prey animals represent members of the same prey guild, that is those species that co-occur together and share the same predators (reviewed in Smith 1999). Chivers & Smith (1994a) and Chivers et al. (1995b) have demonstrated that some cross-species responses must be learned, yet no mechanism has been identi®ed. The present study addresses a mechanism by which prey ®sh may learn the identity of a heterospeci®c alarm signal. Chemical alarm signals have been shown to be important in facilitating learning of predators by prey animals. For example, GoÈz (1941) reported that predator-naõÈ ve European minnows (Phoxinus phoxinus) could be conditioned to recognize chemical stimuli from northern pike (Esox lucius) by detecting pike cues associated with minnow alarm pheromone. This mechanism has been termed releaser-induced learning (Suboski 1990). Since GoÈz's (1941) study, many researchers have tested this mechanism of learned recognition of stimuli. For example, Chivers & Smith (1994a, b) documented that fathead minnows (Pimephales promelas) learn the sight or odour of pike cues when they were associated with conspeci®c alarm cues. Similarly, Yunker et al. (1999) demonstrated that fathead minnows could be conditioned to recognize a red light when it was associated with conspeci®c alarm pheromones. Chivers & Smith (1995) also documented that minnows learn to recognize the odour of dangerous habitats when they were associated with alarm cues. We hypothesized that prey animals may also be able to learn to recognize other stimuli, including heterospeci®c alarm signals that are associated with conspeci®c alarm cues. When may prey have the opportunity to associate conspeci®c and heterospeci®c alarm cues together? One obvious answer is that they may detect both cues simultaneously within dietary cues released from a

Learned Recognition of Heterospeci®c Alarm Signals

1009

predator. Previous studies have established that alarm cues can be conserved in the diet of predators (reviewed in Chivers & Smith 1998; Kats & Dill 1998; Chivers & Mirza 2001; Chivers & Mirza, in press). For example, Mathis & Smith (1993c) showed that predator-naive minnows respond to chemical stimuli from pike fed minnows, but not cues from pike fed another ®sh diet. Similarly, larval may¯ies (Siphlonurus spp. and Siphlonisca spp.) respond with antipredator behaviour to chemical stimuli from brook charr (Salvelinus fontinalis) fed conspeci®cs, but not to charr fed a diet of brine shrimp (Artemia spp.; Huryn & Chivers 1999). Fathead minnows and brook stickleback (Culaea inconstans) are small prey ®shes that often co-occur and share common predators (Mathis & Smith 1993b). The two species are phylogenetically distant. Minnows from populations with stickleback are known to recognize stickleback alarm cues whereas those that do not occur with stickleback do not respond to stickleback cues (Pollock, M. S., Mirza, R. S., Chivers, D. P. & Wisenden, B. D., unpubl. data). We conducted two experiments to attempt to address how minnows learn to recognize stickleback alarm cues. The ®rst experiment examined whether fathead minnows can learn the identity of brook stickleback alarm cues when they are exposed to a predatory perch (Perca ¯avescens) fed a mixed diet of minnows and stickleback. The second experiment was conducted to con®rm the results of the ®rst experiment and to ensure that we could attribute our observed response to learned recognition of the stickleback cues and not recognition of a dangerous stimulus being added to the tank. Study Animals Fathead minnows were collected from Briarwood Pond (approximately 4 ha in area), located in Saskatoon, Saskatchewan, Canada in Nov. 2000. Previous sampling revealed that fathead minnows were the only ®sh species present in the pond. Minnows were transported to our laboratory at the University of Saskatchewan and held in a 400-l aquarium. Brook stickleback were captured by minnow traps from a 1-ha pond on the University of Saskatchewan campus and held in 37-l aquaria. Yellow perch were collected by seine net from Blackstrap Lake in central Saskatchewan and held in 114-l aquaria in our laboratory. Swordtails (Xiphophorus helleri) were purchased from a commercial supplier and held in a 152-l aquarium. All ®shes were kept under a 16-h L : 8 h D light regime with a water temperature of 23°C. Minnows and swordtails were fed ad libitum on commercial ®sh ¯akes. Perch were maintained on a diet of minnows until used in the experiment. Experiment 1 The purpose of this experiment was to determine if fathead minnows could learn the identity of a heterospeci®c alarm signal through the diet of an unknown predator. In the initial phase of this study we conditioned individual minnows to chemical stimuli from (i) perch fed a mixed diet of stickleback and minnows

1010


(ii) perch fed a mixed diet of stickleback and swordtail or (iii) a control of distilled water. In subsequent test trials, we exposed all of the minnows to chemical alarm cues from injured stickleback. If minnows can learn the identity of stickleback alarm cues based on their association with minnow alarm cues in the diet of a predator, then minnows previously exposed to perch fed minnows and stickleback should subsequently respond to stickleback alarm cues alone. In contrast, minnows exposed to chemical cues from perch fed stickleback and swordtails should not subsequently respond to stickleback alarm cues alone. Likewise, minnows exposed to distilled water should not respond upon subsequent exposure to stickleback alarm cues alone. Methods Preparation of stimuli Perch stimulus was prepared from perch fed on a mixed diet of either (i) stickleback and minnow, or (ii) stickleback and swordtail. We fed two yellow perch (12 and 14 cm standard length) a diet of one fathead minnow plus one stickleback and fed two perch (13 and 14 cm) one swordtail plus one stickleback every 3 d for three consecutive feedings. Each perch was given 3±4 ml of ®sh (determined by volumetric displacement). One hour after the ®nal feeding, perch were removed from their holding tanks and placed into 37-l aquaria with fresh dechlorinated water. After 24 h, perch were removed and the remaining water (perch stimuli) was pipetted into 50-ml aliquots and frozen at )20°C. Distilled water was also frozen in 50-ml aliquots at this time. Stickleback skin extract was prepared from ®ve brook stickleback (mean SD 4.60 0.24 cm standard length). Fish were humanely killed with a single blow to the head in accordance with guidelines set by the Canadian Council on Animal Care. A ®llet of skin was removed from each side of each ®sh and placed in 50 ml of ice-chilled distilled water. A total of 13.4 cm2 of skin was used for stimulus preparation. Skin was homogenized and then ®ltered through ®lter ¯oss to remove large particles. The supernatant was then diluted with distilled water to make a ®nal volume of 280 ml. Stimulus was pipetted into 40-ml aliquots and frozen at )20°C. Conditioning trials Conditioning trials were conducted in 37-l aquaria that were visually occluded on three sides with black plastic. This prevented test ®sh from viewing ®sh in adjacent tanks. Each tank contained 3 cm of granite chips as substrate and a single airstone situated at the back of the tank. A single plastic tube was axed to the airstone to allow introduction of chemical stimuli. Each tank contained a single, centrally located shelter object consisting of a 15 ´ 15-cm ceramic tile on three 6-cm cylindrical glass legs.


1011

Individual minnows were allowed to acclimate to test aquaria for 24 h before the start of conditioning trials. All conditioning trials were conducted between 09:00 and 13:00 h. Each trial was 15 min in length and consisted of a 7-min preand 7-min poststimulus period, with a 1-min stimulus introduction period between the pre- and poststimulus periods. At the beginning of each trial, 60 ml of water was removed from the tank through the stimulus injection tube with a syringe and discarded. This removed any stagnant water in the injection tube. A second 60-ml syringe of water was removed and retained. After the prestimulus period, 50 ml of stimulus was added to the tank. The stimuli used in the conditioning trials included chemical stimuli from (i) perch fed a diet of minnow and stickleback (ii) perch fed a diet of swordtail and stickleback or (iii) distilled water. Each stimulus was injected into the tank via the tube and was ¯ushed through with the 60-ml of tank water. This process was done slowly at a rate of approximately 1 ml s)1, so as to minimize disturbance of the test ®sh. Dye trials indicated that it took approximately 30 s for the stimulus to distribute to all parts of the tank. For each trial we recorded the amount of time the test ®sh spent under shelter, as well as instances of dashing (an apparently randomly orientated burst of swimming activity) and freezing (remaining motionless on the substrate for a minimum of 30 s) for both the pre- and poststimulus periods. The order of treatments was randomized. We calculated the change (poststimulus minus prestimulus) in shelter use. Dierences in shelter use among treatments were compared using a Kruskal Wallis one-way ANOVA, followed by non-parametric post-hoc multiple comparisons between all pairs of treatments (Siegel & Castellan 1988). We employed onetailed tests because we predicted that minnows would increase shelter use in response to alarm signals (Chivers & Smith 1998). The number of individuals dashing and freezing was compared with a series of Fisher Exact tests comparing all pairs of treatments (Siegel & Castellan 1988). All statistics were calculated with SPSS 9.0 (SPSS Inc., Chicago, IL). Test trials Test trials were identical to the conditioning trials except that all minnows were exposed to 5 ml of stickleback skin extract. Test trials were conducted between 17:00 and 21:00 h on the same day as the conditioning trials (carried out between 9:00 and 13:00 h). Results Conditioning trials In the conditioning trials there was a signi®cant dierence in change in shelter use among treatments (v2 6.65, df 2, p 0.018, one-tailed, Kruskal± Wallis ANOVA). Post-hoc multiple comparisons revealed that minnows exposed to

1012


Fig. 1: Mean (+ SE) for change in shelter use of fathead minnows exposed to chemical cues from either yellow perch fed a diet of minnows and stickleback (MIN + SB), perch fed a diet of swordtails and stickleback (SWT + SB), or distilled water (conditioning trials) and subsequently exposed to stickleback skin extract (test trials). Dierent letters above bars represent signi®cant dierences (p < 0.05; see text for details)

chemical stimuli from perch fed minnows and stickleback increased shelter use significantly more than minnows exposed to distilled water, but not minnows exposed to perch fed swordtails and stickleback (Fig. 1). The low intensity increase in shelter use of minnows to cues of perch fed stickleback and swordtail may indicate that minnows were responding to a new stimulus being added to the tank. Minnows exposed to chemical stimuli from perch fed minnows and stickleback were significantly more likely to dash and freeze than minnows exposed to distilled water or perch fed stickleback and swordtail (Table 1). Test trials In the test trials (exposure to stickleback alarm cues) there was a signi®cant dierence in change in shelter use among treatments (v2 17.36, df 2, p < 0.001, one-tailed). Post-hoc comparisons showed that minnows previously exposed to perch fed minnows and stickleback used shelter signi®cantly more than

1013


Table 1: Results of Fisher Exact tests comparing number of individuals freezing and dashing between pairs of treatments for both conditioning and test trials. All tests are onetailed Treatment

Dashes

Freezes

Conditioning Trials Perch fed SWT + SB Perch fed MIN + SB p value

1 7 0.022

4 10 0.048

Perch fed MIN + SB Distilled water p value

7 0 0.004

10 0 < 0.001

Perch fed SWT + SB Distilled water p value

1 0 0.500

4 0 0.053

Test Trials Perch fed SWT + SB Perch fed MIN + SB p value

0 4 0.053

2 8 0.032

Perch fed MIN + SB Distilled water p value

4 0 0.053

8 0 0.002

Perch fed SWT + SB Distilled water p value

0 0 0.500

2 0 0.244

minnows previously exposed to perch fed swordtails and stickleback or to those previously exposed to distilled water (Fig. 1). There was a trend for individuals to exhibit more dashing if they were previously exposed to perch fed minnows and stickleback compared to minnows previously exposed to perch fed swordtail and stickleback (p 0.053) or distilled water (p 0.053). Moreover, minnows conditioned with chemical stimuli from perch fed minnows and stickleback were more likely to freeze compared with minnows conditioned with perch fed swordtails and stickleback or those conditioned with distilled water. There were no signi®cant dierences in the number of minnows dashing or freezing when previously exposed to perch fed swordtails and stickleback compared to minnows previously exposed to distilled water (Table 1). Experiment 2 The results of expt 1 provide strong evidence that minnows learn to recognize stickleback alarm cues based on the association of stickleback alarm cues with minnow alarm cues in the diet of the predator. However, the results of Mathis & Smith (1993c) provide a potential confounding of our interpretation of learned recognition of stickleback alarm cues. They showed that prey ®sh can potentially

1014


learn to recognize the method by which a frightening stimulus is introduced to their tank during a conditioning trial. This means that the response of minnows to stickleback cues in the test trials of expt 1 may indicate that the minnows had learned to recognize any stimulus introduced via the injection tube as dangerous and not speci®cally the stickleback stimulus. In expt 2, to con®rm that the results of expt 1 represented a response to stickleback cues, we conditioned minnows with chemical stimuli from perch fed minnows and stickleback and then subsequently tested half for a response to stickleback cues and half for a response to distilled water. If minnows were learning to recognize the stimulus introduction technique, they should respond to introduction of the distilled water. A response to stickleback cues but not distilled water would con®rm learned recognition of stickleback cues. Methods Preparation of stimuli Chemical stimuli were collected from two perch (13 and 15 cm) fed a diet of minnows and stickleback on the same feeding regiment as in expt 1. Stickleback stimulus was prepared in the same manner as in expt 1 using four stickleback (6.08 0.22 cm). Approximately 23 cm2 of skin was collected and homogenized in 100 ml of distilled water. The supernatant was then diluted with distilled water to make up a ®nal volume of 400 ml. Stimulus was pipetted into 50-ml aliquots and stored at )20°C. Distilled water was also frozen in 50-ml aliquots. Conditioning and test trials The experimental set-up and testing protocol were nearly identical to those of expt 1. Thirty-two minnows were exposed to chemical stimuli from perch fed a diet of minnows and stickleback (conditioning trials). In subsequent test trials, half the minnows were tested for a response to stickleback skin extract and the other half were tested for a response to distilled water. We quanti®ed the same behavioural activities as in expt 1 and calculated the change in shelter use between post- and prestimulus periods in the same manner. We conducted separate Wilcoxon Mann±Whitney tests (Siegel & Castellan 1988) on the change in shelter use. The frequency of dashing and freezing was analysed separately using Fisher Exact tests. Results Conditioning trials In the conditioning trials there was no signi®cant dierence in the change in time spent under shelter between the two groups of minnows

1015


Fig. 2: Mean (+ SE) for change in shelter use of fathead minnows exposed to chemical cues from yellow perch fed a diet of minnows and stickleback (MIN + SB; conditioning trials) and subsequently exposed to stickleback skin extract or distilled water (test trials)

Table 2: Results of Fisher Exact test comparing number of individuals freezing and dashing between pairs of treatments for both conditioning and test trials. All tests are onetailed Treatment

Dashes

Freezes

Conditioning Trials Perch fed MIN + SB Perch fed MIN + SB p-value

3 5 0.343

7 10 0.240

Test Trials Distilled Water Stickleback Extract p-value

0 3 0.113

0 8 0.001

(Z )0.076; p 0.47, n 16; Fig. 2). Moreover, there was no signi®cant dierence in the number of ®sh dashing or freezing between the two groups (Table 2).

1016


Test trials In the subsequent test trials, minnows exposed to stickleback skin extract increased shelter use signi®cantly more than minnows exposed to distilled water (Z )2.26, p 0.012, n 16; Fig. 2). Moreover, minnows exposed to stickleback skin were also signi®cantly more likely to dash and freeze than minnows exposed to distilled water (Table 2). These results con®rm the ®ndings of expt 1, and moreover indicate that the response of minnows to cues of stickleback represented a response to stickleback cues and not to the introduction of any stimulus into the tank. Discussion Cross-species responses to alarm signals are widespread in ®shes (reviewed in Chivers & Smith 1998). Whether these responses result from conservation of the alarm cues among dierent species or whether they result from learning remains unknown for most systems. However, there are some systems, including the minnow/stickleback system, where learning is required for recognition of heterospeci®c cues (Pollock, M. S., Mirza, R. S., Chivers, D. P. & Wisenden, B. D., unpubl. data). In cases where learning is important, the prey species are members of the same prey guild. Our results clearly demonstrate that fathead minnows learn to recognize stickleback alarm cues when they are associated with conspeci®c alarm cues in the diet of the predator. This is the ®rst study to document a mechanism by which ®sh can learn to respond to heterospeci®c alarm cues. Being able to learn heterospeci®c alarm cues aords prey animals more information from which they are able to assess predation risk. Prey animals that are warned by heterospeci®c alarm cues should have higher survival during encounters with predators. Previous studies have reported that chemical alarm cues are important in facilitating learning of novel stimuli. Animals learn to respond to the novel stimuli based on a single association with conspeci®c alarm cues (Chivers & Smith 1994a, b; Chivers et al. 1995a; Yunker et al. 1999). The alarm cues may be released during a predation event (GoÈz 1941; Magurran 1989; Brown & Smith 1998; Mirza & Chivers 2000), or they may be detected in the predator's diet (Mathis & Smith 1993c; Wilson & Lefcort 1993; Brown et al. 1995; Petranka & Hayes 1998; Chivers & Mirza 2001). Previous studies have documented the importance of conspeci®c alarm signals in learned recognition of visual cues of predators, chemical cues of predators and odours of dangerous habitats. The present study demonstrates that conspeci®c alarm cues are also important in facilitating learned recognition of heterospeci®c alarm cues. Ours is the ®rst study to identify a mechanism of learned recognition of heterospeci®c alarm cues. Our results show that minnows learn the identity of heterospeci®c alarm cues through pairing of the heterospeci®c cue with conspeci®c cues in the dietary cues of perch. This indicates that both the conspeci®c and heterospeci®c chemical cues are conserved in the predator's diet. If predators commonly feed on multiple prey


1017

species and it is common for alarm cues to be conserved within the diet of those predators, then we may expect learned recognition of heterospeci®c alarm signals to be widespread. By preying on multiple species, generalist predators may act as vehicles that bring together the alarm cues of dierent species. Assessment of predation risk is paramount to the survival of prey animals. Chemical cues convey vital information to prey animals to make that assessment (reviewed in Chivers & Smith 1998; Kats & Dill 1998). Chemical alarm cues `warn' nearby conspeci®cs and facilitate learning of other chemical stimuli. Our study demonstrates a mechanism by which conspeci®c alarm signals facilitate learning of heterospeci®c alarm cues. There may be other ways in which prey learn to recognize heterospeci®c alarm cues. Future research should address other possible mechanisms. Acknowledgements We would like to thank Don Harbidge, Mike Pollock, Kyra Gazdewich and Carrie and Jahmy Hindman for assistance in collection of ®shes. Dr Jean-Guy Godin provided many helpful comments on an earlier version of this paper. This study was funded by grants to D. P. Chivers from the University of Saskatchewan and NSERC of Canada. All work herein was approved by the University of Saskatchewan Animal Care and Use Committee (research protocol # 19920077).

Literature Cited Brown, G. E. & Godin, J.-G. J. 1997: Anti-predator responses to conspeci®c and heterospeci®c skin extracts by threespine sticklebacks: alarm pheromones revisited. Behaviour 134, 1123Ð1134. Brown, G. E. & Godin, J.-G. J. 1999: Chemical alarm signals in wild Trinidadian guppies (Poecilia reticulata). Can. J. Zool. 77, 562Ð570. Brown, G. E. & Smith, R. J. F. 1998: Acquired predator recognition in juvenile rainbow trout (Oncoryhnchus mykiss): conditioning hatchery-reared ®sh to recognise chemical cues of a predator. Can. J. Fish. Aquat. Sci. 55, 611Ð617. Brown, G. E., Chivers, D. P. & Smith, R. J. F. 1995: Fathead minnows avoid conspeci®c and heterospeci®c alarm pheromones in the faeces of northern pike. J. Fish. Biol. 47, 387Ð393. Chivers, D. P. & Mirza, R. S. 2001: The importance of predator diet cues in the responses of larval wood frogs to ®sh and invertebrate predators. J. Chem. Ecol. 27, 45Ð51. Chivers, D. P. & Mirza, R. S., in press: Predator diet cues and the assessment of predation risk by aquatic vertebrates: a review and prospectus. In: Chemical Signals in Vertebrates, Vol. 9. (Marchlewska-Koj, A., Lepri, J. J. & MuÈller-Schwarze, D., eds). Plenum Press, New York, in press. Chivers, D. P. & Smith, R. J. F. 1994a: Intra and interspeci®c avoidance of areas marked with skin extract from brook stickleback (Culaea inconstans) in a natural habitat. J. Chem. Ecol. 20, 1517Ð1524. Chivers, D. P. & Smith, R. J. F. 1994b: Fathead minnows (Pimephales promelas) acquire predator recognition when alarm substance is associated with the sight of an unfamiliar ®sh. Anim. Behav. 48, 597Ð605. Chivers, D. P. & Smith, R. J. F. 1995: Fathead minnows learn to recognize chemical stimuli from highrisk habitats by the presence of alarm substance. Behav. Ecol. 6, 155Ð158. Chivers, D. P. & Smith, R. J. F. 1998: Chemical alarm signalling in aquatic predator-prey systems: a review and prospectus. EÂcoscience 5, 338Ð352. Chivers, D. P., Brown, G. E. & Smith, R. J. F. 1995a: Acquired recognition of chemical stimuli from pike, Esox lucius, by brook sticklebacks, Culaea inconstans (Osteichthyes, Gasterosteidae). Ethology 99, 234Ð242. Chivers, D. P., Wisenden, B. D. & Smith, R. J. F. 1995b: The role of experience in the response of fathead minnows (Pimephales promelas) to skin extract of Iowa darters (Etheostoma exile). Behaviour 132, 665Ð674.

1018


GoÈz, H. 1941: UÈber den Art- und Individualgeruch bei Fischen. Z. Vergl. Physiol. 29, 723Ð730. Huryn, A. D. & Chivers, D. P. 1999: Contrasting behavioral responses by detritivorous and predatory may¯ies to chemicals released by injured conspeci®cs and their predators. J. Chem. Ecol. 25, 2529Ð2540. Kats, L. B. & Dill, L. M. 1998: The scent of death: chemosensory assessment of predation risk by prey animals. EÂcoscience 5, 361Ð394. Lima, S. L. & Dill, L. M. 1990: Behavioural decisions made under the risk of predation: a review and prospectus. Can. J. Zool. 68, 619Ð640. Magurran, A. E. 1989: Acquired recognition of predator odour in the European minnow (Phoxinus phoxinus). Ethology 82, 216Ð233. Magurran, A. E., Irving, P. W. & Henderson, P. A. 1996: Is there a ®sh alarm pheromone? A wild study and critique. Proc. R. Soc. Lond. B. 263, 1551Ð1556. Mathis, A. & Smith, R. J. F. 1993a: Chemical alarm signals increase the survival time of fathead minnows (Pimephales promelas) during encounters with northern pike (Esox lucius). Behav. Ecol. 4, 260Ð265. Mathis, A. & Smith, R. J. F. 1993b: Intraspeci®c and cross-superorder responses to chemical alarm signals by brook stickleback. Ecology 74, 2395Ð2404. Mathis, A. & Smith, R. J. F. 1993c: Fathead minnows (Pimephales promelas) learn to recognize pike (Esox lucius) as predators on the basis of chemical stimuli from minnows in the pike's diet. Anim. Behav. 46, 645Ð656. Mirza, R. S. & Chivers, D. P. 2000: Predator-recognition training enhances survival of brook trout: evidence from laboratory and ®eld enclosure studies. Can. J. Zool. 78, 2198Ð2208. Mirza, R. S. & Chivers, D. P. in press: Do chemical alarm signals enhance survival in aquatic vertebrates? An analysis of the current research paradigm. In: Chemical Signals in Vertebrates, Vol. 9. (Marchlewska-Koj, A., Lepri J. J., & MuÈller-Schwarze, D. eds). Plenum Press, New York, in press. Petranka, J. W. & Hayes, L. 1998: Chemically mediated avoidance of a predatory odonate (Anax junius) by American toad (Bufo americanus) and woodfrog (Rana sylvatica) tadpoles. Behav. Ecol. Sociobiol. 42, 263Ð271. Siegel, S. & Castellan, N. J. 1988: Nonparametric Statistics for the Behavioral Sciences, 2nd edn. McGraw-Hill. New York. Sih, A. 1987: Predator and prey lifestyles: an evolutionary and ecological overview. In: Predation: Direct and Indirect Impacts on Aquatic Communities (Kerfoot, W. C. & Sih, A., eds). Univ. Press of New England, Hanover, pp. 203Ð224. Smith, R. J. F. 1992: Alarm signals in ®shes. Rev. Fish. Biol. Fish. 2, 3Ð63. Smith, R. J. F. 1997: Does one result trump all others? A response to Magurran, Irving and Henderson. Proc. R. Soc. Lond. B. 264, 445Ð450. Smith, R. J. F. 1999: What good is smelly stu in the skin? Cross function and cross taxa eects in ®sh `alarm substances'. In: Advances in Chemical Signals in Vertebrates (Johnston, R. E., MuÈllerSchwarze, D. & Sorensen, P., eds). Kluwer Academic/Plenum Press, New York, pp. 475Ð487. Suboski, M. D. 1990: Releaser-induced recognition learning. Psych. Rev. 97, 271Ð284. Wilson, D. J. & Lefcort, H. 1993: The eect of predator diet on the alarm response of red-legged frog, Rana aurora, tadpoles. Anim. Behav. 46, 1017Ð1019. Wisenden, B. D., Cline, A. & Sparkes, T. C. 1999: Survival bene®t to antipredator behaviour in the amphipod Gammarus minus (Crustacea: amphipoda) in response to injury-released chemical cues from conspeci®cs and heterospeci®cs. Ethology 105, 407Ð414. Yunker, W. K., Wein, D. E. & Wisenden, B. D. 1999: Conditioned alarm behavior in fathead minnows (Pimephales promelas) resulting from association of chemical alarm pheromone with a nonbiological visual stimulus. J. Chem. Ecol. 25, 2677Ð2686. Received: February 26, 2001 Initial acceptance: April 22, 2001 Final acceptance: May 25, 2001 (J.-G. Godin)