Differential retention of predator recognition by juvenile rainbow trout

Maud C.O. Ferrari1,4) , Grant E. Brown2) , Christopher D. Jackson2) , Patrick H. Malka2) & Douglas P. Chivers3) (1 Department of Environmental Science and Policy, University of California, Davis, CA, 95616 USA; 2 Department of Biology, Concordia University, Montreal, QC, Canada H4B 1R6; 3 Department of Biology, University of Saskatchewan, Saskatoon, SK, Canada S7N 5E2) (Accepted: 22 September 2010)

Summary There is a wealth of studies that have examined the way in which prey animals acquire information about their predators, yet the literature on how long prey retain this information is almost non-existent. Here, we investigated if the memory window associated with learned recognition of predators by juvenile rainbow trout was fixed or variable. Specifically, we tested whether the retention of predator recognition was influenced by the risk level associated with the predator. We conditioned juvenile trout to recognize predatory pumpkinseed sunfish posing a high, low or no threat and tested their response to the predator after either 1 or 8 days, and found that trout responded to the odour of the pumpkinseed longer if the risk associated with the predator was higher. We discuss the way in which memory associated with predator risk information provides fundamentally different costs/benefits trade-offs than those associated with foraging. Keywords: predator recognition, memory, threat-sensitivity, learning, rainbow trout Oncorhynchus mykiss.

Introduction To be successful, prey animals have to decrease the possibility of being eaten by predators while maximizing foraging and reproduction (Lima & Dill, 4)

Corresponding author’s e-mail address: [email protected]

© Koninklijke Brill NV, Leiden, 2010 DOI:10.1163/000579510X535677

Behaviour 147, 1791-1802 Also available online - www.brill.nl/beh

1792

Ferrari, Brown, Jackson, Malka & Chivers

1990). To optimize this trade-off, prey have to maintain accurate and reliable information about the risk associated with potential predators, which allows them to display adaptive, threat-sensitive antipredator responses (Helfman, 1989; Chivers et al., 2001). One way that prey gather information about their predators is through learning. Whether the information is acquired through personal experience (direct learning) or through conspecifics (social learning), prey learn not only to recognize the identity of the predator but also the level of risk posed by the predator (e.g., Ferrari et al., 2005, 2007). However, the risk posed by predators may fluctuate through time (Sih, 1992). Hence, as time passes, the reliability of the learned information may decrease. In other words, the uncertainty associated with the risk of the predator increases if the prey cannot obtain updated information regarding the status of that predator. While hundreds of studies have investigated ways in which prey gather information about their predators, very little is known about the way in which prey species use information learned about their predators as time goes on and the uncertainty associated with the predator cues increases (Ferrari et al., 2010a). Contrasting with the sparse record of predation studies, a number of theoretical and empirical studies have looked at the concept of memory and forgetting in a foraging context (McNamara & Houston, 1987; Stephens, 1987; Hirvonen et al., 1999; Dall & Johnstone, 2002). McNamara and Houston (1987) provided an early model for which optimality of information use was dependent on the rate at which the environment was changing. If the environment was constant, then all pieces of information regarding the value of a foraging patch should be used in the decision process and they should be given equal weight. However, if the environment was rapidly changing, then recent, accurate information should weight more in the decision process, while older information should weigh less. It was assumed that the individual had a given memory window, and all information outside of this window were not used in the decision process, and were considered ‘forgotten’. Hirvonen et al. (1999) suggested that the size of this memory window should not be fixed, but flexible. They proposed that the size of the memory window should be dependent on the payoffs associated with the use of the information. In other words, a good payoff, that is, a good correlation between the information and the food reward, would result in the memory window being extended, and the information used for longer. Conversely, a poor payoff, that is, poor or no correlation between the information and the food reward,

Memory of predators

1793

would result in the memory window being reduced, and the information discarded earlier. While this concept seems adaptive for foragers, it might not be the case for prey. Using suboptimal information for a forager will likely result in missed foraging opportunities and a potential decrease in fitness. The cost of using suboptimal information regarding the risk posed by a predator results in wasted time and energy if the risk associated with the predator is over-estimated, and death if the risk is under-estimated. Thus, predation and foraging will likely lead to different ways in which individuals deal with increased uncertainty. A number of studies have reported prey species having different ‘memory windows’. Fathead minnows still recognized a predatory pike two months post-learning (Chivers & Smith, 1994), while juvenile salmonids only retained information about their predators for a period ranging from 10 days (Mirza & Chivers, 2000) to 21 days (Brown & Smith, 1998). Iberian green frog tadpoles stopped responding to predator cues after 9 days (Gonzalo et al., 2009). Different species may have different memory windows for a number of reasons relating to, for example, body or brain sizes, ecology, evolutionary or life history. An exciting line of work is to examine whether this memory window is fixed or plastic for any given individual, and if it is plastic, to investigate the factors influencing the size of the memory window. To date, two studies have investigated flexibility of memory windows in prey animals. Iberian green frog tadpoles (Pelophylax perezi), conditioned to recognize the odour of a fish predator had differential retention of the predator based on whether the tadpoles were conditioned to recognize the fish using chemical alarm cues from injured conspecifics or disturbance cues from disturbed conspecifics. Disturbance cues represent a lower level of certainty with regards to the identity of the predator risk than alarm cues released from damage conspecifics, and the memory associated with learning through disturbance cues is correspondingly shorter (Gonzalo et al., 2010). Similarly, woodfrog tadpoles (Rana sylvatica) have differential retention of potential salamander predators based on the uncertainty associated with the identity of salamander (Ambystoma tigrinum) (Ferrari et al., 2010a). As uncertainty increases, the length of the memory is shorter. In this study, we tested whether the risk level associated with the predator would influence the retention of the information learned by prey. We hypothesized that, because the cost of under-responding to a low-risk predator should be lower than the cost of under-responding to a high-risk predator,

1794

Ferrari, Brown, Jackson, Malka & Chivers

prey should increase the size of their memory window as the risk associated with the predator increases (i.e., retain recognition of an acquired predator cue for longer periods). To test this hypothesis, we conditioned predatornaïve juvenile rainbow trout to recognize a pumpkinseed sunfish (Lepomis cyanellus) as a high-risk or low-risk predator, and subsequently tested them after 1 and 8 days for their response to the predator odour. To teach the trout to recognize the predator, we exposed them to conspecific alarm cues paired with the odour of pumpkinseeds. This mode of learning is widespread in a variety of aquatic species, from flatworms to larval amphibians (reviewed by Ferrari et al., 2010b). In addition, the concentration of alarm cues used during the conditioning phase has been shown to mediate the risk level associated with the predator (Ferrari et al., 2005). Hence, trout conditioned with higher concentrations of alarm cues should label the predator as a higher threat than trout conditioned with a lower concentration of alarm cues.

Methods Test fish We obtained juvenile rainbow trout from a commercial hatchery (Pisciculture des Arpents Verts, Ste-Edwidge-de-Clifton, QC, Canada). Trout were housed in 390-l circular flow-through holding tanks, supplied with continuously filtered water at approx. 500 ml/min (18◦ C, pH approx. 7.2, 12 :12 light/dark cycle). Trout were fed ad libitum twice daily with commercial trout chow. Pumpkinseeds used as predator odour donors, were collected with beach seines from Canal Lachine, near Montreal. Pumpkinseeds were held as described above for trout and were fed daily with commercial trout pellets and brine shrimp (Artemia sp.). Stimulus collection We generated our stock rainbow trout alarm cue from 12 donors (standard length (SL): mean ± SD = 4.15 ± 0.32 cm). Donors were killed via cervical dislocation (in accordance with the Concordia University Animal Care Protocol AREC-2008-BROW). We removed skin fillets from either side of the donor and immediately placed them into 100 ml distilled water. Skin fillets were homogenized, filtered through polyester floss and diluted to final concentration (approx. 1 cm2 skin in 10 ml water) with the addition of distilled

Memory of predators

1795

water. This concentration reliably elicits antipredator responses in rainbow trout (Brown & Smith, 1998; Leduc et al., 2004). We collected a total of 67.3 cm2 skin, diluted in 670 ml water. We froze the alarm cue in 20 ml aliquots at −20◦ C until needed. As a control, we also froze 20 ml aliquots of distilled water. To generate the odour of a novel predator, four pumpkinseed (SL range 9.2–12.5 cm) were individually placed into 37-l glass aquaria, filled with 15 l dechlorinated tap water. The tank water was unfiltered and pumpkinseeds were not fed during this period. For a period of at least one week prior to odour collection pumpkinseed were only fed brine shrimp, to ensure the resulting predator odour did not contain any diet-related cues that may have been recognized by the juvenile trout. Pumpkinseed were left in the tanks for 48 h, removed and returned to the stock holding tank. The remaining tank water (predator odour) was combined, filtered and frozen in 50-ml aliquots at −20◦ C until needed. Experimental protocol The experiment consisted of 3 phases. Juvenile rainbow trout were first conditioned to recognize predatory pumpkinseed as high-threat, low-threat or no-threat predators, through simultaneous exposure to pumpkinseed odour paired with a high or a low concentration of alarm cues or distilled water, respectively. The trout were then tested 1 and 8 days post-conditioning for their responses to the odour of pumpkinseed alone. Conditioning phase Individual trout, selected haphazardly from our stock tanks, were placed into 37-l tanks and allowed a 24-h acclimation period prior to conditioning. Approximately 30 min prior to observations, we fed the trout in each test tank with sufficient commercial trout chow so that there was some food left in the test tank during the trial. Tanks contained dechlorinated tap water (approx. 18◦ C, pH 7.0–7.2), and were equipped with a gravel substrate, a single airstone affixed to the rear wall, but no filter. We attached an additional 1.5 m length of airline tubing, terminating adjacent to the airstone to allow for the introduction of test stimuli. Conditioning trials consisted of a 5-min pre-stimulus and a 5-min post-stimulus period, during which we recorded the time spent moving (in s) and total number of foraging attempts of the trout.

1796

Ferrari, Brown, Jackson, Malka & Chivers

A foraging attempt was defined as a biting movement towards food particles on the substrate or in the water column. Decreased time spent moving and foraging are well documented antipredator responses in this species (e.g., Leduc et al., 2004). Prior to the pre-stimulus observation, we withdrew and discarded 60 ml tank water through the stimulus injection tube. We then withdrew and retained an additional 60 ml to be used to flush the stimulus into the tank. Immediately following the pre-stimulus observation period, we introduced 10 ml predator odour paired with 10 ml of a high-concentration alarm cue, a low-concentration alarm cue or a control of distilled water. The high concentration was the one used for the alarm cue stock solution. The low concentration was obtained by taking 1 ml stock alarm cue and diluting it with 9 ml distilled water. Following the conditioning observations, we moved each focal fish to an identical tank filled with dechlorinated tap water. Recognition phase Recognition trials were conducted as described above for the conditioning trials, except that we exposed the trout to 10 ml predator odour only. We tested trout for recognition of the predator odour 24 h post-conditioning and 8 days post-conditioning. Each focal fish was placed into an identical tank filled with dechlorinated tap water between the two observation days. We tested 15 trout in each of the three treatments (SL at testing 3.69 ± 0.42 cm). The order of treatment was randomized. All behavioural observations were done blind with regards to test cues and conditioning treatments. Statistical analysis We calculated the change in time spent moving and number of foraging attempts between the pre- and post-stimulus observation periods, and used these difference scores as our dependant variables. Since time moving and foraging rates were highly correlated, we used a multivariate approach to account for the dependency of the two behaviours. We used a repeatedmeasures MANOVA, with change in time moving and foraging as our dependant variables, risk level of the predator (high vs. low vs. no risk) as the independent variable and time of the experiment (conditioning vs. 1-day recognition vs. 8-days recognition) as the repeated variable. Given the significant predator risk level × time interaction (see below), we tested for the

Memory of predators

1797

effects of predator risk level for each of the three times separately using a one-way MANOVA.

Results Our overall repeated-measures MANOVA revealed a significant interaction between the predator risk level and time (F 4,84 = 3, p = 0.021). During the conditioning phase, trout exposed to both the high-concentration and low-concentration alarm cue treatments exhibited significantly stronger antipredator responses (decreased time moving and frequency of foraging attempts) than did those exposed to the distilled water control (F 2,42 = 19, p < 0.001; Figure 1). We found a similar pattern 1-day post conditioning (F 2,42 = 8, p = 0.001; Figure 1). However, the pattern changed when trout were tested 8-day post-conditioning (F 2,42 = 9, p = 0.001; Figure 1). Only trout conditioned with the high concentration of alarm cue showed an antipredator response to the predator odour. Trout initially conditioned with a low concentration of alarm cues did not respond differently to the predator odour than the trout from the distilled water control.

Discussion Our results indicate that juvenile rainbow trout have a longer retention of information relating to high-risk predators than low-risk predators. After 8 days, the only trout displaying an antipredator response upon detecting pumpkinseed odour were those that were conditioned with a high concentration of alarm cues. Our study is the first to demonstrate a variable memory window for predation information in fishes. From a purely empirical point of view, it is impossible for us to distinguish whether the absence of an antipredator response in the low-risk trout was the result of forgetting per se, i.e., the information about the predator was forgotten, or whether the trout did recognize the predators but ‘decided’ not to respond to them. However, studies have shown that increases in stress during information acquisition increase the memory window for which this information is retrievable (e.g., Oitzl et al., 2001). For the purpose of this paper, we will not discuss the proximate explanations any further, and rather will focus on the adaptive value of the behavioural outcome.

1798

Ferrari, Brown, Jackson, Malka & Chivers

Figure 1. Mean (± SE) change from the pre-stimulus baseline in time spent moving (top panel) and number of foraging attempts (bottom panel) during conditioning (day 1) or at day 2 or 9. Juvenile trout were conditioned to recognize a predatory pumpkinseed as a high threat (solid bars), low threat (hatched bars) or no threat (open bars).

Memory of predators

1799

The only two studies to look at factors affecting the length of the memory window in a predation context manipulated the level of certainty associated with the identity of the potential predators. Gonzalo et al. (2010) showed that Iberian green frog tadpoles conditioned to recognize a predator from cues reflecting a higher certainty of risk (injured conspecific cues) rather than a more uncertain cue (cues released from disturbed conspecifics) responded to the learned predator for longer. Ferrari et al. (2010a) showed that larval woodfrogs respond to a known predator for a longer period than to a species whose identity was generalized from a known predator (i.e., whose identity has yet to be confirmed as an actual threat). Our study is the first to directly investigate the effects of risk level and not certainty, on the memory window of prey. In addition, the two aforementioned studies were performed on larval amphibians, whose life history strategy may fundamentally differ from that of freshwater fishes. For metamorphosing species, the need to maximize growth will result in a life history strategy targeted towards maximizing foraging at the cost of antipredator behaviour (growth maximizer). On the other hand, other species may put more emphasis onto predator avoidance at the cost of decreased foraging returns, since they do not face a temporal constraint linked to metamorphosis. Hence, larval amphibians may be more prone to forgetting than other species, like fishes, that do not face such life history constraints. It is, thus, important to investigate memory windows in taxa with varying life histories. Our results fit within the theoretical framework for information forgetting in prey species proposed by Ferrari et al. (2010a). The duration for which information about predators is remembered should be linked to the costs and benefits associated with maintaining or discarding the information. We will first assume that maintaining information is costly. This is a reasonable assumption, since some studies are showing memory itself may be costly (Dukas, 1999; Mery & Kawecki, 2005; Kuhl et al., 2007). Information should be discarded as soon as its informative value is lower than random. Rainbow trout are prey to a variety of predators. As they grow, they most certainly have the chance to outgrow some of their potential predators. Moreover, their burst speed should increase dramatically, rendering them less vulnerable to predators. Taken together, this would suggest that trout responding to previously dangerous but presently harmless species would be selected against, since they would incur additional costs due to missed foraging and reproductive opportunities. They should instead continually update the information about the threat level of potential predators.

1800

Ferrari, Brown, Jackson, Malka & Chivers

Studies considering the adaptive value of forgetting predators are at their infancy. While we have known for some time that there were species differences in the length of memory (e.g., Chivers & Smith, 1994; Brown & Smith, 1998; Gonzalo et al., 2009) only recently have we considered whether the memory window was fixed or plastic for any given individual, and if plastic, to determine which factors influence the size of the memory window. Our current paper, along with that Gonzalo et al. (2010) and Ferrari et al. (2010a) fit this category. All of the studies so far have investigated the memory window associated with the recognition of a single predator. The next step will be to consider how prey species deal with information about the myriad of predators they may encounter in their natural environment. The foraging literature provides potential insights into the limitations that prey species may be facing in such a case. Evidence from the insect and bird literature suggests that increasing the number of pieces of information that individuals need to memorize in their environment could interfere with their ability to memorize new pieces of information (Kintsch, 1977; McGregor & Avery, 1986). Such interference may limit the ability of species to memorize new food types and may explain interspecific differences in memory (Johnson et al., 1994). Presumably, the ability of prey to memorize the identity of additional potential predators is at least as, if not more important, than their ability to recognize additional food items. While individuals may only remember the most rewarding food items, a prey may be killed by the most dangerous predators, but may also be killed by less dangerous predators. Since predation leaves very little opportunities for mistakes, prey may in fact benefit from remembering all species that may represent a threat. Thus, the next line of research should focus on the differential ability of individuals to remember additional information in a foraging and a predation context. Can interference play a different role in foraging and predation, or is it a physical limitation of the cognitive ability of the species? What role does predator generalization play in the ability of species to identify and remember different species of predators? Generalization of predator recognition (Stankowich et al., 2007; Ferrari et al., 2007, 2008) is the mechanism by which prey can recognize the predatory nature of a novel species, based on the similarities between this novel species and known predators. Ferrari et al. (2008) showed that prey are more likely to generalize their predator recognition if the predator represents a high threat. If prey species have a cognitive limitation in the amount

Memory of predators

1801

of coding for individual predator species that they can do, one would predict that their memory would switch from identifying individual predator species to identifying ‘classes’ of predators. In other words, their memory capacity would be traded for their discriminability, and this may result in a much more general frame of predator recognition. Besides possible intraspecific limitations, one can predict interspecific differences due to differential evolutionary history. Species having evolved in environments of low predator diversity may be unable to cope with a rapid exposure to highly diverse predator communities, since this cognitive ability was never under selection. Adaptive forgetting of predators may be a key element of success for prey species and we encourage others to consider additional tests in other systems.

Acknowledgements We thank NSERC for Discovery Grants to G.E.B. and D.P.C., and a Post-Doctoral Fellowship to M.C.O.F. All work reported herein was conducted in accordance with Concordia University Animal Research Ethics Protocol AREC-2008-BROW.

References Brown, G.E. & Smith, R.J.F. (1998). Acquired predator recognition in juvenile rainbow trout (Oncorhynchus mykiss): conditioning hatchery-reared fish to recognize chemical cues of a predator. — Can. J. Fish. Aquat. Sci. 55: 611-617. Chivers, D.P. & Smith, R.J.F. (1994). 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., 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. — Can. J. Zool. 79: 867-873. Dall, S.R.X. & Johnstone, R.A. (2002). Managing uncertainty: information and insurance under the risk of starvation. — Phil. Trans. Roy. Soc. B 357: 1519-1526. Dukas, R. (1999). Costs of memory: ideas and predictions. — J. Theor. Biol. 197: 41-50. Ferrari, M.C.O., Trowell, J.J., Brown, G.E. & Chivers, D.P. (2005). The role of leaning in the development of threat-sensitive predator avoidance in fathead minnows. — Anim. Behav. 70: 777-784. Ferrari, M.C.O., Gonzalo, A., Messier, F. & Chivers, D.P. (2007). Generalization of learned predator recognition: an experimental test and framework for future studies. — Proc. Roy. Soc. Lond. B: Biol. 274: 1853-1859. Ferrari, M.C.O., Messier, F. & Chivers, D.P. (2008). Can prey exhibit threat-sensitive generalization of predator recognition? Extending the predator recognition continuum hypothesis. — Proc. Roy. Soc Lond. B: Biol. 275: 1811-1816.

1802

Ferrari, Brown, Jackson, Malka & Chivers

Ferrari, M.C.O., Brown, G.E., Bortolotti, G.R. & Chivers, D.P. (2010a). Linking predator risk and uncertainty to adaptive forgetting: a theoretical framework and empirical test using tadpoles. — Proc. Roy. Soc. Lond. B: Biol. 277: 2205-2210. Ferrari, M.C.O., Wisenden, B.D. & Chivers, D.P. (2010b). Chemical ecology of predatorprey interactions in aquatic ecosystems: a review and prospectus. — Can. J. Zool. 88: 698-724. Gonzalo, A., López, P. & Martín, J. (2009). Learning, memory and apparent forgetting of chemical cues from new predators by Iberian green frog tadpoles. — Anim. Cogn. 12: 745-750. Gonzalo, A., López, P. & Martín, J. (2010). Risk level of chemical cues determines retention of recognition of new predators in Iberian green frog tadpoles. — Behav. Ecol. Sociobiol. 64: 1117-1123. Helfman, G.S. (1989). Threat-sensitive predator avoidance in damselfish-trumpetfish interactions. — Behav. Ecol. Sociobiol. 24: 47-58. Hirvonen, H., Ranta, E., Rita, H. & Peuhkuri, N. (1999). Significance of memory properties in prey choice decisions. — Ecol. Model. 115: 177-189. Kintsch, W. (1977). Memory and cognition. — Wiley, New York, NY. Kuhl, B.A., Dudukovic, N.M., Kahn, I. & Wagner, A.D. (2007). Decreased demands on cognitive control reveal the neural processing benefits of forgetting. — Nature Neurosci. 10: 908-914. Johnson, R.A., Rissing, S.W. & Killeen, P.E. (1994). Differential learning and memory by co-occurring ant species. — Insectes Sociaux 41: 165-177. Leduc, A.O.H.C., Ferrari, M.C.O., Kelly, J.M. & Brown, G.E. (2004). Learning to recognize novel predators under weakly acidic conditions: the effects of reduced pH on acquired predator recognition by juvenile rainbow trout. — Chemoecology 14: 107-112. Lima, S.L. & Dill, L.M. (1990). Behavioral decisions made under the risk of predation: a review and prospectus. — Can. J. Zool. 68: 619-640. McGregor, P.K. & Avery, M.I. (1986). The unsung song of great tits (Parus major): learning neighbors’ songs for discrimination. — Behav. Ecol. Sociobiol. 18: 311-316. McNamara, J.M. & Houston, A.I. (1987). Memory and the efficient use of information. — J. Theor. Biol. 125: 385-395. Mery, F. & Kawecki, T.J. (2005). A cost of long-term memory in Drosophila. — Science 308: 1148. Mirza, R.S. & Chivers, D.P. (2000). Predator recognition training enhances survival of brook trout: evidence from laboratory and field enclosure studies. — Can. J. Zool. 78: 21982208. Oitzl, M.S., Fluttert, M., Sutanto, W. & de Kloet, E.R. (2001). Continuous blockade of brain glucocorticoid receptors facilitates spatial learning and memory in rats. — Eur. J. Neurosci. 10: 3759-3766. Sih, A. (1992). Prey uncertainty and the balancing of antipredator and feeding needs. — Am. Nat. 139: 1052-1069. Stankowich, T. & Coss, R.G. (2007). The re-emergence of felid camouflage with the decay of predator recognition in deer under relaxed selection. — Proc. Roy. Soc. Lond. B: Biol. 274: 175-182. Stephens, D.W. (1987). On economically tracking a variable environment. — Theor. Popul. Biol. 32: 15-25.