Cultural transmission of predator recognition in fishes: intraspecific and interspecific learning

Anim. Behav., 1996, 51, 185–201

Cultural transmission of predator recognition in fishes: intraspecific and interspecific learning ALICIA MATHIS, DOUGLAS P. CHIVERS & R. JAN F. SMITH Department of Biology, University of Saskatchewan (Received 2 September 1994; initial acceptance 2 January 1995; final acceptance 19 June 1995; MS. number: 7087)

Abstract. Individuals that live in groups may have the opportunity to learn to recognize unfamiliar predators by observing the fright responses of experienced individuals in the group. In intraspecific trials, naive fathead minnows, Pimephales promelas, gave fright responses to chemical stimuli from predatory northern pike, Esox lucius, when paired with pike-experienced conspecifics but not when paired with pike-naive conspecifics. These pike-conditioned minnows retained the fright responses to pike odour when tested alone, indicating that learning had occurred, and transmitted their fright responses to pike-naive minnows in subsequent trials. Brook stickleback, Culaea inconstans, are found in mixed-species aggregations with fathead minnows and are also vulnerable to predation by northern pike. In a series of interspecific tests, pike-naive brook stickleback gave fright responses to chemical stimuli from northern pike when paired with pike-experienced minnows but not when paired with pike-naive minnows. Pike-conditioned stickleback also retained the fright responses when tested alone and subsequently also transmitted the fright responses to pike-naive minnows. Individuals may benefit from observations of the fright responses of conspecifics or heterospecifics by (1) being alerted to the immediate presence of unfamiliar predators and (2) learning to recognize unfamiliar predators as a ? 1996 The Association for the Study of Animal Behaviour potential threat. Individuals in either monospecific or mixedspecies groups may experience a number of advantages compared with solitary individuals. One advantage may be the opportunity for cultural transmission of beneficial behaviour patterns from experienced group members to younger or less experienced individuals (reviewed in Mainardi 1980; Curio 1988; Mineka & Cook 1988). Cultural transmission of information used in predator recognition has been reported for a number of species. For example, predator-naive individuals of several species of birds learn to recognize predators by observing the mobbing responses of conspecifics (jackdaws, Corvus monedula: Lorenz 1952; European blackbirds, Turdus merula: Curio et al. 1978; zebra finches, Taeniopygia guttata: Vieth et al. 1980). Cultural transmission of predator recognition has also been reported for ducks Correspondence: A. Mathis, Department of Biology, Southwest Missouri State University, Springfield, MO 65804-0095, U.S.A. D. P. Chivers and R. J. F. Smith are at the Department of Biology, University of Saskatchewan, 112 Science Place, Saskatoon, Saskatchewan S7N 5E2, Canada. 0003–3472/96/010185+17 $12.00/0

?

(Klopfer 1957) and for several species of primates (Herzog & Hopf 1984; review in Mineka & Cook 1988), and has been suggested for gastropod molluscs (Wood 1968; Morgan 1972). For some fishes, predator-naive individuals may be alerted to the presence of potential predators via their associations with either experienced conspecifics (Verheijen 1956; Magurran & Higham 1988) or heterospecifics (Krause 1993). The short-term consequence of such information transfer may be an increased probability of survival of the current encounter with an unfamiliar predator. There may also be long-term benefits if naive individuals become conditioned to recognize the predator as dangerous in future encounters. ‘Conditioned’ individuals may subsequently transmit the learned information to naive individuals. Only one study has reported such cultural transmission of learned fright responses for fishes. Zebra danios, Brachydanio rerio, in one tank transmitted the recognition of a synthetic chemical (morpholine) to naive conspecifics in an adjacent tank (Suboski et al. 1990). 1996 The Association for the Study of Animal Behaviour

185

Animal Behaviour, 51, 1

186 Control N1

A

N2

Paired

No fright response B

No fright response

N1

A

A

No fright response

B N3

Paired

E

Paired

Fright response

Alone

No fright response C

Experimental

A Fright response

Alone N2

Fright response C

C

No fright response D

Fright response Alone

No fright response

Paired

C Fright response D

Alone

Weak fright response

Figure 1. Flow chart illustrating the general experimental design of the intraspecific tests. Letters by arrows indicate the four different experiments. Results of each experiment are indicated in italics. Sizes of drawings illustrate relative size differences between minnows. N=pike-naive minnow, E=pike-experienced minnow. Minnows were tested either in pairs (experiments A and C) or alone (B and D).

The purpose of this study was two-fold. First we examined whether fathead minnows, Pimephales promelas, can culturally transmit the recognition of the odour of a natural predator, northern pike, Esox lucius, between conspecifics. The second goal was to determine whether cultural transmission of predator recognition also can occur between heterospecifics. Fathead minnows and brook stickleback, Culaea inconstans, are good candidates for tests of cultural transmission, because these two species are often sympatric (e.g. Robinson & Tonn 1989), occupy similar microhabitats (Whitaker 1968) and share common predators (Becker 1983). Northern pike often prey upon both fathead minnows and brook stickleback, and experienced individuals of both species can recognize pike as a potential predator based solely on chemical cues, whereas inexperienced individuals do not (Gelowitz et al. 1993; Mathis et al. 1993; Chivers & Smith 1994a). For both intraspecific and interspecific tests, we determined whether (1) pike-naive individuals can

receive and respond to information from pikeexperienced individuals, (2) the conditioned individuals retain the fright response to pike when subsequently tested alone, and (3) conditioned individuals can transmit the learned information to naive individuals. EXPERIMENTAL SERIES 1: INTRASPECIFIC CULTURAL TRANSMISSION Overview of Experimental Design The test for intraspecific cultural transmission included four experiments (Fig. 1). Experiment 1a documented the differential response of experienced and naive fathead minnows to pike odour and tested whether naive minnows were influenced by the behaviour of their experienced or naive partners. Differential responses by naive minnows could be explained by social facilitation and do not necessarily demonstrate learning by the naive minnows. Experiment 1b tested whether the

Mathis et al.: Cultural transmission by fishes

187

‘conditioned’ minnows retained their response to pike odour when tested alone. A positive result in this test would indicate cultural transmission of information for predator recognition between experienced and naive conspecifics. The final two experiments tested whether minnows can subsequently communicate information learned via cultural transmission to other naive conspecifics.

maintained the minnows in an outdoor pool (approximately 18 000 litre), where they were fed daily with Tetramin flakes. In the spring of 1993, we transferred the minnows to the laboratory where they were maintained at approximately 15)C on a 14:10 h L:D cycle. Minnows reared in Pike Lake show a fright reaction to pike odour (Mathis et al. 1993; Chivers & Smith 1993, 1994a).

Collection and Maintenance

Northern pike

Pike-naive fathead minnows

We collected pike from Pike Lake and housed them individually in separate compartments of a 300-litre stream tank at approximately 15)C on a 14:10 h L:D cycle. We fed them one or two fathead minnows every 5 days.

In May 1992, we placed artificial spawning surfaces (broken flower-pots) in Pike Lake, an oxbow lake of the South Saskatchewan River in south-central Saskatchewan. In June, we transferred the spawning surfaces containing fathead minnow eggs to separate aquaria in the laboratory, where they were incubated at approximately 20)C. The eggs hatched in June and July, and the minnows were fed daily with commercial fish food (Tetramin liquifry, Tetramin baby food for egg layers, and eventually Tetramin flakes). In June– August we supplemented their diet with live food (including brine shrimp, Artemia fransciscana, Daphnia and other zooplankton). We maintained the minnows in the laboratory on a 14:10 h light: dark cycle until April 1993 when the experiments began. Laboratory-reared Pike Lake minnows do not give an anti-predator response to chemical stimuli from pike even though they come from a population inhabiting a lake that contains pike (Chivers & Smith 1994a). In the autumn of 1992, we also collected pike-naive fathead minnows from Marshy Creek in south-central Saskatchewan. Marshy Creek is within an enclosed evaporation basin with no piscivorous fish species. Minnows from Marshy Creek do not show a fright response to chemical stimuli from pike (Mathis et al. 1993). We maintained the minnows in a large (approximately 18 000 litre) outdoor pool where they were fed daily with Tetramin flakes. In the spring of 1993, we transferred the minnows to the laboratory where they were maintained at approximately 15)C on a 14:10 h L:D cycle. We fed the fish daily as described before. Pike-experienced fathead minnows In the autumn of 1992, we collected pikeexperienced fathead minnows from Pike Lake. We

Pike Stimulus Preparation Our stimulus preparation methods were identical to those of Mathis et al. (1993). Prior to stimulus collection, we fed two pike (fork lengths=21 and 22 cm) approximately equal volumes (range 3–4 ml, measured by volumetric displacment in water) of swordtails, Xiphophorus helleri, for each of three feedings (i.e. once every 5 days). We used a swordtail diet to eliminate alarm pheromone cues associated with the pike’s diet of fathead minnows (Mathis & Smith 1993a, b). Approximately 17 h after feeding, we removed the two pike from their holding tanks, rinsed them with dechlorinated tap water to remove any swordtail residue from the pike’s skin and then transferred them to separate, clear plastic chambers (26#8#8 cm) that each contained 1200 ml of tap water that had been run through charcoal and sodium zeolite filters to remove chlorine and ammonia, which were present in the local municipal water supply. The chambers were well aerated but contained no filtration system to avoid removing active compounds from the water. After approximately 3 days, we removed the pike from the chambers, combined the stimulus water (individual pike identity does not affect predator recognition abilities of prey fishes; Mathis & Smith 1993a, b; Mathis et al. 1993; Gelowitz et al. 1993) and pippetted it into either 5-ml polypropylene containers or 20-ml plastic bags, and froze it at approximately "20)C. The stimulus water was thawed immediately prior to use.

188

Animal Behaviour, 51, 1

Experiment 1a: Responses of Pike-naive Minnows Paired with Pike-experienced Minnows Methods To test the responses of pairs of minnows to chemical stimuli from pike, we filled a series of 37-litre aquaria with dechlorinated tap water. The tanks were aerated with an airstone located at the back of each tank. The stimulus water was injected into the tank via clear plastic Tygon tubing tied to the airline so that the end of the tubing was approximately 1 cm from the surface of the airstone. The inflow of air from the airstone rapidly dispersed the stimulus water. Each aquarium contained a centrally located shelter that consisted of a ceramic tile (9.8#20.0 cm) mounted on three cylindrical glass legs (5.5 cm long). We placed a partial divider (1.5#9.8 cm) along the centre of the tile to reduce aggressive interactions when both fish were beneath the shelter. We randomly placed 30 pike-naive minnows (X& fork length=4.7&0.51 cm; six fish from each of five broods) into separate testing tanks. We paired each of 15 of these pike-naive minnows with a pike-experienced minnow (fork length= 5.6&0.22 cm; experimental treatment group), and paired each of the other 15 pike-naive minnows with another pike-naive minnnow (fork length= 5.3&0.49 cm; control treatment group). To enable the observer to tell the minnows apart, naive test fish within each trial were the smaller of the pair. The small differences in size between minnows in a pair does not influence the interpretation of the results of the experiment. Both control and experimental pike-naive subjects were ‘small’ and both control (other pike-naive minnows) and experimental (pike-experienced minnows) partners were ‘large’. Therefore, test subjects in both treatment groups were similarly sized and were treated identically except regarding the experience of their partners. We tested each pair of minnows once. We conducted the observations at a mean water temperature of 20)C on a 14:10 h L:D cycle, after the minnows had remained in the observation tanks undisturbed, except for daily feedings, for 3 days. Immediately prior to the beginning of each trial, we drew 60 ml of tank water through the stimulus-injection tube by a syringe and discarded it to remove any stagnant water that might have

collected in the tubing. We drew an additional 60 ml of tank water into the syringe and saved it. We filled a second syringe with 20 ml of the pike stimulus water. During each trial we first observed each pair undisturbed for 8 min and quantified shelter use every 15 s by recording whether each fish of the pair was beneath the shelter. For each trial, we calculated a pre-stimulus index of shelter use as the number of observations that the focal minnow was under the shelter divided by the total number of observations, with a possible range of 0.0 (no shelter use) to 1.0 (always under the shelter). At the end of 8 min, we slowly (approximately 1 ml/s) injected the 20 ml of pike stimulus water into the tank, and immediately thereafter injected 60 ml of tank water to flush the pike stimulus out of the stimulus-injection tube. Observations continued for an additional 8 min, and we calculated post-stimulus indices of shelter use for each trial. During the post-stimulus period we also recorded the occurrence of two additional behaviour patterns: (1) dashing (rapid, apparently disoriented swimming) and (2) freezing (the fish drops to the bottom of the tank and remains immobile for a minimum of 30 s). Shelter use, dashing and freezing are components of the fright reaction of fathead minnows (e.g. Lawrence & Smith 1989; Mathis & Smith 1993c; Chivers & Smith 1995a). We recorded a number of response variables (see above paragraph) to increase our power to detect a fright response. Fathead minnows exposed to stimuli associated with predation may show a number of fright responses (e.g. increased shelter use, dashing, freezing, decreased foraging), but not all individuals show all of these behavioural patterns. Furthermore, the different classes of responses are not mutually exclusive. For example, an individual may immediately respond to the stimulus by dashing for a few seconds and then freeze. We interpreted the appearance of any of these behavioural patterns as an indication of a fright response. It is not known whether the various response variables represent different levels of intensity of a fright response (e.g. is dashing a more intense response than freezing?). For each treatment group, we compared prestimulus versus post-stimulus indices of shelter use with two-tailed Wilcoxon matched-pairs signedranks tests (Zar 1984). Because dashing and freezing never occurred during pre-stimulus observations, pre-stimulus versus post-stimulus

Mathis et al.: Cultural transmission by fishes

1.2

(a)

Pre-stimulus Post-stimulus

189

(b)

Mean shelter use index

1

0.8 ** 0.6 NS

**

0.4 NS

0.2

0

Experienced

Naive

Experienced

Naive

Figure 2. Mean (+) shelter use index (a) for pike-experienced and pike-native fathead minnows and (b) for pike-naive minnows that were paired with the two types of minnows during 8 min before and after introduction of chemical stimuli from northern pike. Wilcoxon matched-pairs signed-ranks test: **P<0.001; , P>0.25.

comparisons of these behavioural patterns were not appropriate. Instead, we directly compared the treatment groups for differences in the proportion of post-stimulus trials in which dashing and freezing occurred using either the chi-squared test or the Fisher exact probability test (Siegel 1956). In addition to evaluating the responses of the control and experimental subjects, we also conducted the same statistical tests for the partners of the test subjects to document that the two treatment groups were exposed to different stimuli. We used a Spearman rank-order correlation to determine whether the changes in shelter use by paired minnows were positively correlated. Results Pike-experienced fathead minnows significantly increased their use of shelter following exposure to water from tanks that had held northern pike (Wilcoxon T=91, N*=13, P<0.001; Fig. 2). Pikenaive fathead minnows paired with experienced minnows also showed a significant increase in shelter use after exposure to the pike-stimulus (T=78, N*=12, P<0.001; Fig. 2). In contrast, there was no

significant change in shelter use by either pikenaive ‘trainer’ minnows (T=42, N*=12, P>0.80; Fig. 2) or the naive minnows that were paired with them (T=62, N*=13, P>0.25; Fig. 2). Changes in shelter use indices (post-stimulus minus pre-stimulus) by minnows of each pair were positively correlated for both the experimental treatment group (rS =0.627, N=15, P<0.01) and the control treatment group (rS =0.91, N=15, P<0.001). During the post-stimulus period, pikeexperienced minnows showed significantly more dashing than did pike-naive minnows, but there were no significant differences in the amount of freezing (Table I). Pike-naive minnows paired with pike-experienced minnows showed a significantly greater proportion of freezing responses than did naive minnows paired with other pike-naive minnows (Table I); the two treatment groups did not differ in proportion of dashes. In subsequent tests we refer to the fathead minnows tested with pike-experienced minnows in experiment 1a as ‘pike-conditioned’ and the minnows tested with pike-naive minnows as ‘pike-pseudo-conditioned’.

Animal Behaviour, 51, 1

190

Table I. Per cent of intraspecific trials in which minnows showed dashing and freezing behaviour following exposure to chemical stimuli from northern pike (experiment 1a) Behaviour of pike-experienced and pike-naive fathead minnows paired with pike-naive conspecifics

Dashing Freezing

Experienced (% of trials)

Naive (% of trials)

P

60 0

0 0

<0.001 —

Behaviour of pike-naive fathead minnows paired with either pike-experienced or pike-naive conspecifics Partner

Dashing Freezing

Experienced (% of trials)

Naive (% of trials)

P

27 33

7 0

0.165 0.021

Statistical results are of Fisher exact probability tests. Percentages are based on number of trials out of 15.

Experiment 1b: Retention of the Conditioned Fright Response by Fathead Minnows Methods The results of experiment 1a demonstrated that pike-naive minnows show an anti-predator response to chemical stimuli from pike while in the presence of an experienced minnow performing anti-predator behaviour patterns. The purpose of this experiment was to determine whether small pike-conditioned fathead minnows from experiment 1a retained the fright response to chemical stimuli from pike during subsequent tests in the absence of large experienced conspecifics. Mathis & Smith (1993a) demonstrated that, for fathead minnows, a fright response may be elicited by a neutral chemical stimulus (i.e. water) that is presented in the same manner as a fright stimulus (chemical stimuli from pike which had eaten fathead minnows). In that study, minnows moving to a different experimental situation eliminated the fright response to the neutral stimulus. To control for possible conditioning to test conditions in our study, we also changed the experimental set-up in which the minnows were tested. One day after being used in experiment 1a, the small pike-conditioned minnows and the small

pike-pseudo-conditioned minnows were placed individually into Plexiglas acclimation tanks (45#45#20 cm). Water was supplied to the acclimation tanks at a constant rate of approximately 250 ml/min, maintaining a water depth of 4–5 cm. We fed the minnows daily as before and kept them under a 14:10 h L:D cycle. After 2 days, we transferred the minnows to separate testing tanks that were identical to the acclimation tanks except that the flow rate was increased to approximately 450 ml/min. Testing occurred the following day; 4 days passed between the time that the minnows were tested in experiments 1a and 1b. We surrounded each testing tank with an OptoVarimex-Aqua tracking meter (Columbus Instruments) which established a grid of light beams in the tank. The activity tracking meter was interfaced to a computer that was preset to ‘scan’ the light beams for breaks at intervals of 0.125 s. The computer integrated and digitized information from the scans to quantify the fish’s level of activity. A detailed description of both the acclimation tanks and the testing apparatus is given by Lemly & Smith (1986, 1987). Our system differed from that of Lemly & Smith’s in that outflowing water from out tanks was discarded rather than recirculated. Two measures of activity that typically decrease during a fright reaction by fathead minnows (e.g. Lawrence & Smith 1989; Mathis et al. 1993) were quantified by the computer: (1) total distance travelled (cm), and (2) number of stereotypic movements (i.e. activity in which a fish breaks light beams without moving outside of one grid square); many stereotypic movements would indicate high levels of activity even though this activity may not translate into linear distance. The activity meter did not recognize dashes and freezes, so these data are not available for the solitary trials. We conducted the experiment from an observation room that was adjacent to the test room. The observer viewed the experimental tanks via a video monitor and injected the stimulus solution into the inflow water lines that ran through the observation room before entering the test room. The design of this experiment was similar to that of experiment 1a. Trials lasted 16 min. At the end of 8 min, the observer injected 5 ml of pike-stimulus water from the same source as that used in the previous experiment. Injections of vegetable dye into the inflow

Mathis et al.: Cultural transmission by fishes

1200

(a)

Pre-stimulus Post-stimulus

191

(b)

Mean activity levels

1000 **

NS

800

600 **

NS

Conditioned

Pseudo-conditioned

400

200

0

Conditioned

Pseudo-conditioned

Figure 3. Mean (+) activity levels (a: distance travelled (cm); b: number of stereotypic movements) for pike-conditioned and pike-pseudo-conditioned fathead minnows during 8 min before and after introduction of chemical stimuli from northern pike. Conditioned minnows had been previously paired with pike-experienced minnows; pseudo-conditioned minnows had been previously paired with pike-naive minnows. Wilcoxon matchedpairs signed-ranks test: **P<0.01; , P>0.45.

lines demonstrated that it took approximately 14 s for the dye to reach the experimental tanks following injection. We made statistical comparisons of pre- and post-stimulus activity using two-tailed Wilcoxon matched-pairs signed-ranks tests (Zar 1984). We interpreted a significant decrease in activity as a fright response; previous experiments in activity meters demonstrated that pike-experienced (but not pike-naive) fathead minnows responded to chemical stimuli from pike with decreased activity (Mathis et al. 1993). Results Pike-conditioned minnows continued to show a fright response to chemical stimuli from pike when tested alone (distance travelled: Wilcoxon T=112, N*=15, P<0.002; stereotypic moves: T=107, N*=15, P<0.01; Fig. 3). Pike-pseudoconditioned minnows showed no significant change in activity following exposure to the pikestimulus (distance travelled: T=61, N*=14,

P>0.60; stereotypic P>0.45; Fig. 3).

moves:

T=73,

N*=15,

Experiment 1c: Response of Pike-naive Minnows Paired with Pike-conditioned Minnows Methods Experiments 1a and 1b showed that naive fathead minnows can learn to recognize chemical cues from pike by associating with experienced conspecifics. The purpose of this experiment was to determine whether the pike-conditioned minnows can transmit this learned information to new pike-naive minnows. Experimental protocol and statistical analyses were identical to those used in experiment 1a. We paired 26 large pike-naive minnows (X& fork length=5.5&0.40 cm) from Marshy Creek with 14 small pike-conditioned minnows and 12 small pike-pseudo-conditioned minnows from experiment 1b (one pike-conditioned minnow and three pike-pseudo-conditioned minnows died in the 8-day interval between experiments 1b and 1c).

Animal Behaviour, 51, 1

192

1.2

(a)

Pre-stimulus Post-stimulus

(b)

Mean shelter use index

1

0.8 NS

**

0.6 ** NS

0.4

0.2

0

Conditioned

Pseudo-conditioned

Conditioned

Pseudo-conditioned

Figure 4. Mean (+) shelter use (a) for pike-conditioned and pike-pseudo-conditioned fathead minnows and (b) for the pike-naive conspecifics that were paired with them during 8 min before and after introduction of chemical stimuli from northern pike. Conditioned minnows had been previously paired with pike-experienced minnows; pseudoconditioned minnows had been previously paired with pike-naive minnows. Wilcoxon matched-pairs signed-ranks test: **P<0.005; , P>0.35.

We conducted tests between 1030 and 1530 hours at approximately 20)C. Results Both pike-conditioned minnows (Wilcoxon T=87.5, N*=13, P<0.005) and the minnows that were paired with them (T=90, N*=13, P<0.0005) showed a significant increase in shelter use following exposure to chemical stimuli from pike (Fig. 4). Neither the pseudo-conditioned minnows (T=15.5, N*=7, P>0.80) nor the minnows tested with them significantly changed their use of shelter following injection of the pike-stimulus (T=37, N*=10, P>0.35; Fig. 4). Changes in shelter use by the minnow pairs were positively correlated for both the experimental treatment group (rS =0.507, N=14, P<0.05) and the control treatment group (rS =0.80, N=14, P<0.003). Pikeconditioned minnows also showed significantly more dashing than pike-pseudo-conditioned minnows, but there was no difference between the proportion of freezing shown by minnows in these two treatment groups (Table II). There were also no differences in either the proportion of dashing

Table II. Per cent of intraspecific trials in which minnows showed dashing and freezing behaviour following exposure to chemical stimuli from northern pike (experiment 1c) Behaviour of pike-conditioned and pike-pseuoconditioned fathead minnows paired with pike-naive conspecifics

Dashing Freezing

Conditioned (% of trials)

Pseudo-conditioned (% of trials)

P

43 14

0 0

0.013 0.280

Behaviour of pike-naive fathead minnows paired with either pike-conditioned or pike-pseudo-conditioned conspecifics Partner

Dashing Freezing

Conditioned (% of trials)

Pseudo-conditioned (% of trials)

P

21 0

0 0

0.140 —

Statistical results are of Fisher exact probability tests. Percentages are based on number of trials out of either 12 (conditioned) or 14 (pseudo-conditioned).

Mathis et al.: Cultural transmission by fishes or freezing shown by minnows paired with either conditioned and pseudo-conditioned minnows (Table II).

193

EXPERIMENTAL SERIES 2: INTERSPECIFIC CULTURAL TRANSMISSION Overview of Experimental Design

Experiment 1d: Retention of the Conditioned Fright Response by Fathead Minnows Methods The results of experiment 1c indicated that the pike-naive fathead minnows responded to pike odour with a fright response (increased shelter use) when in the presence of pikeconditioned minnows. The purpose of this experiment was to determine whether the fathead minnows from experiment 1c retained the fright response to chemical stimuli from pike during subsequent tests in the absence of the pikeconditioned minnows. The experimental protocol and statistical analyses were identical to those used in experiment 1b. We conducted tests between 0730 and 1330 hours at approximately 20)C.

Tests for interspecific cultural transmission consisted of three experiments (Fig. 5) and generally followed the protocol established in experimental series 1. In experiment 2a, we documented differential responses of experienced and naive fathead minnows to pike odour. In addition, this experiment tested whether the behaviour of their experienced or naive fathead minnow partners influenced naive brook sticklebacks. Experiment 2b tested whether the conditioned sticklebacks retained their response to pike odour when fathead minnows were absent. Retention of fright responses would indicate learning by the stickleback and would indicate cultural transmission of information for predator recognition between experienced and naive heterospecifics. Experiment 2c determined whether ‘conditioned’ sticklebacks can subsequently communicate culturally transmitted information to naive fathead minnows.

Results Collection and Maintenance Fathead minnows that were previously tested with pike-conditioned fathead minnows showed no significant differences in activity in response to pike odour when tested alone (distance travelled: X decrease=19%; Wilcoxon T=39, N*=14, P>0.40; stereotypic moves: X decrease=9.9%; T=33, N*=14, P>0.20). In control tests, minnows previously tested with pike-pseudoconditioned minnows also showed no significant change in distance travelled (X increase=14.8%; T=21, N*=12, P>0.15), but showed a significant increase in the number of stereotypic moves (X increase=24.7%; T=6, N*=12, P<0.01). Because of the significant increase in stereotypic movements for minnows in the control condition, and the tendency for stereotypic movements to decrease for the minnows in the experimental condition, we also directly compared the change in stereotypic movements for minnows in the two treatment groups. The change in stereotypic movements was significantly different for minnows in control versus experimental treatments (Wilcoxon–Mann–Whitney tests; Siegel & Castellan 1988: Wx=150, m=14, N=12, P<0.05).

We collected pike-naive fathead minnows and brook sticklebacks from Marshy Creek; like fathead minnows, brook sticklebacks from Marshy Creek do not display fright responses to chemical stimuli from northern pike (Gelowitz et al. 1993). We collected ‘experienced’ fathead minnows from Pike Lake which, as previously noted, do respond to chemical stimuli from northern pike with a fright reaction. We maintained the minnows and sticklebacks in the laboratory in separate compartments of a 300-litre artificial stream tank at approximately 15)C on a 14:10 h L:D cycle. Minnows were fed daily with Tetramin flakes and sticklebacks were fed daily with frozen brine shrimp, Artemia franciscana. Northern pike were collected and maintained as described in the previous tests.

Pike Stimulus Preparation We prepared the pike water stimulus as described for the previous tests, except that we used two different pike (fork lengths=22.5 and 23.0 cm).

Animal Behaviour, 51, 1

194 Control N

Experimental N

N1

A

Paired

No fright response B

A

C No fright response

A

No fright response

B N2

Paired

Paired

Fright response

Alone

No fright response

E

A Fright response

Alone N

Fright response C

No fright response

C Fright response

Paired

C Fright response

Figure 5. Flow chart illustrating the general experimental design of the interspecific tests. Letters by arrows indicate the three different experiments. Results of each experiment are indicated in italics. N=pike-naive stickleback or fathead minnow, E=experienced minnow. Fishes were tested either in pairs (experiments A and C) or alone (experiment B).

Experiment 2a: Response of Pike-naive Sticklebacks Paired with Pike-experienced Fathead Minnows Methods We tested stickleback/minnow pairs in 37-litre aquaria identical to those described for experiment 1a. We randomly paired each stickleback (X& total length=4.9&0.43 cm) with either a pike-naive fathead minnow from Marshy Creek (fork length=4.9&0.34 cm) or a pike-experienced fathead minnow from Pike Lake (fork length= 4.8&0.50 cm) with 20 pairs in each treatment group. Pairs remained in the tanks for 3 days prior to testing on the same photoperiod and feeding schedule as before. We conducted trials between 0950 and 1730 hours at a mean water temperature of 22)C. We conducted trials as described for experiment 1a. In addition to the measures of activity recorded in experiment 1a, we also recorded the presence or absence of feeding responses (nipping at either the surface of the water or the substrate) during the post-stimulus period. We classified such behaviour as feeding responses even though no food was introduced into the tanks because it

appears identical to the behaviour of the fishes during feeding. We considered these responses to be potentially important because they may indicate that the fishes respond to the novel chemical as a feeding stimulant rather than as an indication of danger.

Results Pike-experienced fathead minnows from Pike Lake significantly increased their use of shelter following exposure to water from tanks that had held northern pike (Wilcoxon T=12, N*=14, P<0.01; Fig. 6). Pike-naive sticklebacks paired with pike-experienced fathead minnows also significantly increased shelter use after exposure to the pike-stimulus (T=1, N*=11, P<0.01; Fig. 6). In contrast, there was no significant change in shelter use by either pike-naive fathead minnows (T=36.5, N*=13, P>0.50) or sticklebacks paired with them (T=37.4, N*=17, P>0.05) following exposure to the pike-stimulus (Fig. 6). During the post-stimulus period, pikeexperienced fathead minnows showed significantly fewer feeding responses and significantly more

Mathis et al.: Cultural transmission by fishes

195

Figure 6. Mean (+) shelter use index (a) for pike-experienced and pike-naive fathead minnows and (b) for the pike-naive sticklebacks that were paired with them during 8 min before and after introduction of chemical stimuli from northern pike. Wilcoxon matched-pairs signed-ranks test: **P<0.01; , P>0.05.

incidences of freezing and dashing than did pikenaive minnows (Table III). Pike-naive sticklebacks paired with pike-experienced minnows also gave significantly fewer post-stimulus feeding responses than the stickleback paired with pikenaive minnows, but there was no significant difference between the two groups in proportions of trials with freezing responses (Table III). Sticklebacks made no dashes. Experiment 2b: Retention of Conditioned Fright Responses by Sticklebacks Methods The purpose of this experiment was to determine whether the sticklebacks from experiment 2a retained the fright response to chemical stimuli from northern pike during subsequent tests in the absence of fathead minnows. We conducted tests using the same protocol and testing chambers (i.e. the activity meters) as in experiment 1b. Statistical comparisons of pre- and poststimulus activity were similar to those used in experiment 1b. As before, we interpreted a significant decrease in activity as a fright response; previous experiments in these activity meters

showed that pike-experienced (but not pike-naive) brook sticklebacks responded to chemical stimuli from pike with decreased activity (Gelowitz et al. 1993). Results Pike-conditioned sticklebacks continued to give a fright response to chemical stimuli from pike when tested alone (distance travelled: Wilcoxon T=23.5, N*=20, P<0.005; Fig. 7). Pike-pseudoconditioned sticklebacks showed no significant change in activity following exposure to the pikestimulus (T=75.5, N*=20, P>0.20; Fig. 7). Experiment 2c: Response of Pike-naive Fathead Minnows Paired with Pike-conditioned Brook Sticklebacks Methods Experiments 2a and 2b demonstrated that sticklebacks can learn chemical recognition of pike from fathead minnows. The purpose of experiment 2c was to determine whether the pikeconditioned sticklebacks from experiment 2b can transmit this information to pike-naive fathead minnows from Marshy Creek.

Animal Behaviour, 51, 1

196

Table III. Per cent of interspecific trials in which brook sticklebacks and fathead minnows exhibited feeding responses, freezing behaviour or dashing behaviour following exposure to chemical stimuli from northern pike (experiment 2a) Behaviour of pike-experienced and pike-naive fathead minnows paired with pike-naive brook sticklebacks Experienced Naive (% of trials) (% of trials) ÷2 P Feeding responses Freezing Dashing

0 40 40

60 0 0

14.40 — —

<0.002 0.003 0.003

Behaviour of pike-naive brook sticklebacks paired with either pike-experienced or pike-naive fathead minnows Partner

Feeding responses Freezing Dashing

Experienced (% of trials)

Naive (% of trials)

÷2

P

5 40 0

60 10 0

11.40 3.33 —

<0.002 >0.050 —

Statistical results are based on either two-tailed chi-squared 2#2 contingency tables (df=1) or Fisher exact probability tests. Percentages are based on number of trials out of 20.

Results Both pike-conditioned sticklebacks (Wilcoxon T=0, N*=10, P<0.005) and minnows paired with them (T=0, N*=9, P=0.005) showed a significant increase in shelter use following exposure to chemical stimuli from pike (Fig. 8). Neither pseudo-conditioned sticklebacks (T=2, N*=5, P=0.20) nor minnows tested with them significantly changed their use of shelter following

injection of the pike-stimulus (T=4.5, N*=7, P>0.10; Fig. 8). Pike-conditioned sticklebacks and pseudoconditioned sticklebacks did not differ in the proportion of trials in which there were feeding Mean distance travelled (cm)

The testing protocol was identical to that of experiment 1a. We randomly paired fathead minnows from Marshy Creek (X& fork length=5.9&0.39 cm) with either pikeconditioned sticklebacks (those paired with pikeexperienced fathead minnows in experiment 2a) or pseudo-conditioned sticklebacks (those paired with pike-naive fathead minnows in experiment 2a). Owing to a labelling error, we were unable to identify previous treatment condition for half of the sticklebacks from experiment 2b. Because we could not test these individuals, the sample size for experiment 2c decreased to 10 minnow/stickleback pairs in each treatment group. We conducted tests between 1030 and 1450 hours at approximately 22)C.

Pre-stimulus Post-stimulus

250 200

NS

**

150 100 50 0

Conditioned stickleback

Pseudo-conditioned stickleback

Figure 7. Mean (+) activity levels (distance travelled) for pike-conditioned and pike-pseudo-conditioned sticklebacks during 8 min before and after introduction of chemical stimuli from northern pike. Conditioned sticklebacks had been previously paired with pikeexperienced fathead minnows; pseudo-conditioned sticklebacks had been previously paired with pike-naive minnows. Wilcoxon matched-pairs signed-ranks test: **P<0.005; , P>0.20.

Mathis et al.: Cultural transmission by fishes

1.2

(a)

Pre-stimulus Post-stimulus

197

(b)

Mean shelter use index

1

0.8

0.6

**

**

NS

NS

0.4

0.2

0

Conditioned

Pseudo-conditioned

Conditioned

Pseudo-conditioned

Figure 8. Mean (+) shelter use (a) for pike-conditioned and pike-pseudo-conditioned brook sticklebacks and (b) for pike-naive fathead minnows that were paired with them during 8 min before and after introduction of chemical stimuli from northern pike. Conditioned sticklebacks had been previously paired with pike-experienced minnows; pseudo-conditioned sticklebacks had been previously paired with pike-naive minnows. Wilcoxon matched-pairs signed-ranks test: **P<0.005; , P>0.10.

responses during the post-stimulus period (Table IV). Pike-conditioned sticklebacks did, however, show a significantly higher incidence of freezing following exposure to the pike-stimulus than did pseudo-conditioned sticklebacks (Table IV). For fathead minnows, there was also no significant difference between the two treatment groups in feeding responses (Table IV). None of the fathead minnows froze in response to the pike-stimulus and dashing was not observed for either species in this experiment. Two minnows died after experiment 2c, resulting in a further decrease in statistical power and an unacceptable increase in the risk of a type II error (failing to reject a false null hypothesis) if further tests were conducted. Therefore, we felt that we did not have enough remaining minnows to test for retention of the fright response.

DISCUSSION Intraspecific Cultural Transmission The results of this study demonstrated that pike-naive fathead minnows give a fright response

to pike odour when paired with pike-experienced conspecifics (experiment 1a). This type of social transmission of a fright response has been demonstrated for harlequin fish, Rasbora heteromorpha (Verheijen 1956) and two species of gobies, Asterropteryx semipunctatus and Gnatholepis anjerensis (Smith & Smith 1989). Naive minnows that were present during the response of experienced fish continued to respond to pike odour with an anti-predator response during subsequent exposures in the absence of pikeexperienced conspecifics (experiment 1b). We therefore considered such minnows to be pikeconditioned. Taken together, these results demonstrate cultural transmission of the recognition of pike odour between experienced and naive fathead minnows. We also demonstrated that pairing pike-conditioned minnows with pikenaive minnows results in anti-predator responses by naive minnows (experiment 1c) and, by at least one measure (stereotypic movements, experiment 1d) minnows appear to retain these learned responses. Therefore, cultural transmission of predator recognition may have far-ranging consequences.

Animal Behaviour, 51, 1

198

Table IV. Per cent of interspecific trials in which brook sticklebacks and fathead minnows showed feeding responses, freezing behaviour or dashing behaviour following exposure to chemical stimuli from northern pike (experiment 2c) Behaviour of pike-conditioned and pike-pseudoconditioned brook sticklebacks paired with pike-naive fathead minows PseudoConditioned conditioned (% of trials) (% of trials) P Feeding responses Freezing Dashing

30 80 0

40 20 0

>0.950 0.023 —

Behaviour of pike-naive fathead minnows paired with either pike-conditioned or pike-pseudo-conditioned brook sticklebacks Partner PseudoConditioned conditioned (% of trials) (% of trials) Feeding responses Freezing Dashing

10 0 0

50 0 0

P 0.140 — —

Statistical results are based on two-tailed Fisher exact probability tests. Percentages are based on number of trials out of 10.

The nature of the fright response may change as the sequence of testing progresses. In the first learning trials (experiment 1a), naive minnows paired with experienced minnows showed significantly more freezing, but not more dashing, than controls (Table I), but the same minnows showed the opposite responses in experiment 1c (Table II). We do not know whether this change in behaviour reflects differences in response intensity. Comparing the responses (activity levels) of solitary individuals, however, shows that the intensity of the response at the first transmission (experiment 1b) is stronger than that observed at the second transmission (experiment 1d). Although we suspect that the success of the transmission by our minnows should eventually be eliminated, Curio et al. (1978) demonstrated that a conditioned mobbing response by blackbirds was passed along a chain of six birds without any detectable decrease in response intensity. The apparent rapidity of response attenuation between the first and second transmission in our study may be

confounded by the difference in size between the larger pike-naive minnows and the smaller pike-conditioned minnows. Given the general correlation between body size and age, and therefore experience, large minnows may be somewhat less apt to learn from small minnows than from large minnows (Dugatkin & Godin 1993). Previous studies of ostariophysan fishes have shown that an alarm pheromone, or Schreckstoff, is given off when a predator mechanically damages the skin of a prey fish (reviewed in Smith 1992). Alarm pheromone released by an injured minnow can condition predator-naive minnows to respond to chemical (Göz 1941; Magurran 1989; Chivers & Smith 1994a) and visual (Chivers & Smith 1994b) stimuli from predators. In addition to this mechanism of learned predator recognition, Mathis & Smith (1993a, b) showed that alarm substance ingested by predators chemically labels them as predators that can then be recognized by naive prey. These mechanisms of predator recognition, together with the mechanism of cultural transmission demonstrated by this study, may have profound implications for animals in natural habitats. Chivers & Smith (1995b) introduced 10 northern pike into ponds containing approximately 20 000 naive minnows; the minnows learned to respond to pike odour with a fright reaction within 14 days. Such a fright response may result in an increased probability of escape (Mathis & Smith 1993c). The fright response may also warn other prey in the vicinity (Verheijen 1956; Smith & Smith 1989; this study) and may reduce predator success and thus discourage future foraging in the area (Trivers 1971). Other naive individuals may also learn to recognize predators by being present during the fright response. Finally, the fright response of the individual may act directly upon the predator to inform it that its presence has been detected, making it unlikely that an attack will be successful (Högstedt 1983). Interspecific Cultural Transmission The results of this study show that inexperienced brook sticklebacks and fathead minnows may benefit from mixed-species associations by gaining information from heterospecifics concerning recognition of potentially dangerous predators. For brook sticklebacks, the benefits may be two-fold, because individuals not only are alerted

Mathis et al.: Cultural transmission by fishes to an immediate threat (experiment 2a), but may also learn to recognize the predator as dangerous in subsequent encounters (experiment 2b). In our study, fright responses to conditioned stimuli were retained without reinforcement for at least two exposures after the initial learning trials, lasting for a period of approximately 1 month. Pike-naive fathead minnows showed fright responses to chemical stimuli from pike when in the presence of pike-conditioned sticklebacks, indicating that the information transfer from stickleback to minnow can also occur (experiment 2c). Predator-naive individuals in this study gained information concerning recognition of predators from heterospecifics. It is not known, however, whether the opportunity for predator-related cultural transmission is greater in mixed-species groups than in monospecific groups. Although brook sticklebacks and fathead minnows often occupy the same microhabitat, their niches do not completely overlap. Therefore, individuals in mixed-species aggregations may be able to capitalize on a wider range of experiences by group members than individuals in strictly monospecific groups. That is, there may be a higher probability that at least one individual will be familiar with a given predator in mixed-species shoals than in single-species shoals. Mixed-species aggregations may also offer other benefits over single-species groups, including increased foraging efficiency due to either social learning (Krebs 1973), flushing of prey by heterospecifics (Siegfried 1971) or shared vigilance (FitzGibbon 1990), decreased competition for food while still providing group-related protection from predation (Moynihan 1962), and early warning of danger (Thompson & Barnard 1983; Smith & Smith 1989; Krause 1993; Mathis & Smith 1993d). Less preferred prey may benefit by associating with prey that are more preferred or more easily captured (FitzGibbon 1990). This latter benefit may apply to brook stickleback/ fathead minnow shoals because the spines of brook sticklebacks make them less vulnerable to some fish predators (Reist 1980a), but more vulnerable to some invertebrate predators (Reist 1980b). Associating with heterospecifics instead of conspecifics may also carry costs. One benefit of group living is the confusion effect, whereby predators find it difficult to select an individual target; this effect is reduced if individuals in the

199

group differ in size and/or appearance (Magurran 1990). Furthermore, individuals that look different from other group members may be the most vulnerable target for predators (Mueller 1971). Single-species shoals may also be more effective at predator-evasion manoeuvres than are mixedspecies groups (Allan & Pitcher 1986). The dynamics of mixed-species aggregations are poorly understood for most taxa, and, for fishes, the costs and benefits associated with multispecies groups have only recently begun to be addressed (Pitcher & Parrish 1993). Our study suggests that interspecific cultural transmission of information concerning predator recognition may be an important variable in determining the adaptive significance of mixed-species aggregations.

ACKNOWLEDGMENTS We thank Cherie Gelowitz for assistance with collection of the fishes. We also thank Brian Wisenden, Grant Brown and Debra Lancaster for providing helpful suggestions concerning the manuscript. Financial support was provided by the Natural Sciences and Engineering Research Council of Canada and the University of Saskatchewan.

REFERENCES Allan, J. R. & Pitcher, T. J. 1986. Species segregation during predator evasion in cyprinid fish shoals. Freshwat. Biol., 16, 653–659. Becker, G. C. 1983. Fishes of Wisconsin. Madison, Wisconsin: University of Wisconsin Press. Chivers, D. P. & Smith, R. J. F. 1993. The role of olfaction in chemosensory-based predator recognition in the fathead minnow, Pimephales promelas. J. chem. Ecol., 19, 623–633. Chivers, D. P. & Smith, R. J. F. 1994a. The role of experience and chemical alarm signalling in predator recognition by fathead minnows (Pimephales promelas). J. Fish Biol., 44, 273–285. Chivers, D. P. & Smith, R. J. F. 1994b. Fathead minnows, Pimephales promelas, acquire predator recognition when alarm substance is associated with the sight of unfamiliar fish. Anim. Behav., 48, 597–605. Chivers, D. P. & Smith, R. J. F. 1995a. Fathead minnows, Pimephales promelas, learn to recognize chemical stimuli from high risk habitats by the presence of alarm substance. Behav. Ecol., 6, 155–158. Chivers, D. P. & Smith, R. J. F. 1995b. Free-living fathead minnows (Pimephales promelas) rapidly learn

200

Animal Behaviour, 51, 1

to recognize pike (Esox lucius) as predators. J. Fish Biol., 46, 949–954. Curio, E. 1988. Cultural transmission of enemy recognition in birds. In: Social Learning: Psychological and Biological Perspectives. (Ed. by T. R. Zentall & B. G. Galef, Jr), pp. 75–98. Hillsdale, New Jersey: Lawrence Erlbaum Associates. Curio, E., Ernst, U. & Vieth, W. 1978. The adaptive significance of avian mobbing. II. Cultural transmission of enemy recognition in blackbirds: effectiveness and some constraints. Z. Tierpsychol., 48, 184–202. Dugatkin, L. A. & Godin, J.-G. J. 1993. Female mate copying in the guppy (Poecilia reticulata): agedependent effects. Behav. Ecol., 4, 289–292. FitzGibbon, C. D. 1990. Mixed-species grouping in Thomson’s and Grant’s gazelles: the antipredator benefits. Anim. Behav., 39, 1116–1126. Gelowitz, C. M., Mathis, A. & Smith, R. J. F. 1993. Chemosensory recognition of northern pike (Esox lucius) by brook stickleback (Culaea inconstans): population differences and the influence of predator diet. Behaviour, 127, 105–118. Göz, H. 1941. Ûber den Art- und Individualgeruch bei Fischen. Z. vergl. Physiol., 29, 1–45. Herzog, M. & Hopf, S. 1984. Behavioural responses to species specific warning calls in infant squirrel monkeys reared in social isolation. Am. J. Primatol., 7, 99–106. Högstedt, G. 1983. Adaptation unto death: function of fear screams. Am. Nat., 121, 562–570. Klopfer, P. H. 1957. An experiment on emphatic learning in ducks. Am. Nat., 91, 61–63. Krause, J. 1993. Transmission of fright reaction between different species of fish. Behaviour, 127, 37–48. Krebs, J. R. 1973. Social learning and the significance of mixed-species flocks of chickadees (Parus spp.). Can. J. Zool., 51, 1275–1288. Lawrence, B. J. & Smith, R. J. F. 1989. Behavioral response of solitary fathead minnows, Pimephales promelas, to alarm substance. J. chem. Ecol., 15, 209–219. Lemly, A. D. & Smith, R. J. F. 1986. A behavioral assay for assessing effects of pollutants on fish chemoreception. Ecotoxicol. environ. Saf., 11, 210–218. Lemly, A. D. & Smith, R. J. F. 1987. Effects of chronic exposure to acidified water on chemoreception of feeding stimuli in fathead minnows (Pimephales promelas): mechanisms and ecological implications. Environ. toxicol. Chem., 6, 225–238. Lorenz, K. 1952. King Solomon’s Ring. New York: Crowell. Magurran, A. E. 1989. Acquired recognition of predator odour in the European minnow, Phoxinus phoxinus. Ethology, 82, 216–223. Magurran, A. E. 1990. The adaptive significance of schooling as an anti-predator defence in fish. Annls Zool. Fenn., 27, 51–66. Magurran, A. E. & Higham, A. 1988. Information transfer across fish shoals under predator threat. Ethology, 78, 153–158. Mainardi, D. 1980. Tradition and the social transmission of behaviour in animals. In: Sociobiology: Beyond

Nature/Nurture? (Ed. by G. W. Barlow & J. Silverberg), pp. 227–255. Boulder, Colorado: Westview Press. 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. & Smith, R. J. F. 1993a. 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, 645–656. Mathis, A. & Smith, R. J. F. 1993b. Chemical labelling of northern pike, Esox lucius, by the alarm pheromone of fathead minnows, Pimephales promelas. J. chem. Ecol., 19, 1967–1979. Mathis, A. & Smith, R. J. F. 1993c. Chemical alarm signals increase the survival time of fathead minnows, Pimephales promelas, during encounters with northern pike, Esox lucius. Behav. Ecol., 4, 260–265. Mathis, A. & Smith, R. J. F. 1993d. Intraspecific and cross-superorder responses to chemical alarm signals by brook stickleback. Ecology, 74, 2395–2404. Mineka, S. & Cook, M. 1988. Social Learning: Psychological and Biological Perspectives (Ed. by T. R. Zentall & B. G. Galef, Jr), pp. 51–74. Hillsdale, New Jersey: Lawrence Erlbaum Associates. Morgan, P. R. 1972. The influence of prey availability on the distribution and predatory behaviour of Nucella lapillus (L.). J. Anim. Ecol., 41, 257–274. Moynihan, M. 1962. The organisation and probable evolution of some mixed species flocks of neotropical birds. Smithson. misc. Collns, 143, 1–140. Mueller, H. C. 1971. Oddity and specific searching image more important than conspicuousness in prey selection. Nature, Lond., 233, 345–346. Pitcher, T. J. & Parrish, J. K. 1993. Functions of shoaling behaviour in teleosts. In: Behaviour of Teleost Fishes. 2nd edn (Ed. by T. J. Pitcher), pp. 363–439. London: Chapman & Hall. Reist, J. D. 1980a. Selective predation upon pelvic phenotypes of brook stickleback, Culaea inconstans, by northern pike, Esox lucius. Can. J. Zool., 58, 1245–1252. Reist, J. D. 1980b. Predation upon pelvic phenotypes of brook stickleback, Culea inconstans, by selected invertebrates. Can. J. Zool., 58, 1253–1258. Robinson, C. L. K. & Tonn, W. M. 1989. Influence of environmental factors and piscivory in structuring fish assemblages of small Alberta lakes. Can. J. Fish. Aquat. Sci., 46, 81–89. Siegel, S. 1956. Nonparametric Statistics for the Behavioural Sciences. New York: McGraw-Hill. Siegel, S. & Castellan, N. J. 1988. Nonparametric Statistics for the Behavioral Sciences. 2nd edn. New York: McGraw-Hill. Siegfried, W. R. 1971. Feeding association between Podiceps ruficallis and Anas smithii. Ibis, 113, 236– 238. Smith, R. J. F. 1992. Alarm signals in fishes. Rev. Fish Biol. Fish., 2, 33–63. Smith, R. J. F. & Smith, M. J. 1989. Predatorrecognition behaviour in two species of gobiid fishes,

Mathis et al.: Cultural transmission by fishes Asterropteryx semipunctatus and Gnatholepsis anjerensis. Ethology, 83, 19–30. 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 simulatedpredator recognition by zebra danio fish (Brachydanio rerio). J. comp. Psychol., 104, 101–112. Thompson, D. B. A. & Barnard, C. J. 1983. Antipredator responses in mixed-species associations of lapwings, golden plovers and black-headed gulls. Anim. Behav., 31, 581–593. Trivers, R. L. 1971. The evolution of reciprocal altruism. Q. Rev. Biol., 46, 35–57. Verheijen, F. J. 1956. Transmission of a fright reaction amongst a school of fish and the underlying sensory mechanisms. Experientia, 12, 202–204.

201

Vieth, W., Curio, E. & Ernst, U. 1980. The adaptive significance of avian mobbing. III. Cultural transmission of enemy recognition in blackbirds: cross-species tutoring and properties of learning. Anim. Behav., 28, 1217–1229. Whitaker, J. W. 1968. Habitat selection and competition in Saskatchewan stickleback (Gasterosteidae) during the ice-free season. M.Sc. thesis, University of Saskatchewan. Wood, L. 1968. Physiological and ecological aspects of prey selection by the marine gastropod Urosalpinx cinerea (Prosobranchia: Muricidae). Malacologia, 6, 267–320. Zar, J. H. 1984. Biostatistical Analysis. 2nd edn. Englewood Cliffs, New Jersey: Prentice-Hall.