Science of the Total Environment 439 (2012) 62–66

Contents lists available at SciVerse ScienceDirect

Science of the Total Environment journal homepage: www.elsevier.com/locate/scitotenv

Understanding the importance of episodic acidification on fish predator–prey interactions: Does weak acidification impair predator recognition? Grant E. Brown a,⁎, Chris K. Elvidge a, Maud C.O. Ferrari b, Douglas P. Chivers c a b c

Department of Biology, Concordia University, 7141 Sherbrooke Street West, Montreal, QC, Canada H4B 1R6 Department of Biomedical Sciences, Western College of Veterinary Medicine, University of Saskatchewan, 52 Campus Drive, Saskatoon, SK, Canada S7N 5B4 Department of Biology, University of Saskatchewan, 112 Science Place, Saskatoon, SK, Canada S7N 5E2

H I G H L I G H T S ► ► ► ► ►

Learned predator recognition increases survival during encounters with predators. At near-neutral pH, trout can learn to recognize novel predator at same or lower pH. Trout can generalize to closely related predator, but only at near-neutral pH. Reducing pH of predator cues to 6.0 eliminates generalized response to novel predators. These data suggest sublethal effects of episodic acidification to aquatic prey.

a r t i c l e

i n f o

Article history: Received 30 July 2012 Received in revised form 4 September 2012 Accepted 11 September 2012 Available online 11 October 2012 Keywords: Predator recognition Threat-sensitive trade-offs Predation risk Acid rain Salmonids

a b s t r a c t The ability of prey to recognize predators is a fundamental prerequisite to avoid being eaten. Indeed, many prey animals learn to distinguish species that pose a threat from those that do not. Once the prey has learned the identity of one predator, it may generalize this recognition to similar predators with which the prey has no experience. The ability to generalize reduces the costs associated with learning and further enhances the ability of the prey to avoid relevant threats. For many aquatic organisms, recognition of predators is based on odor signatures, consequently any anthropogenic alteration in water chemistry has the potential to impair recognition and learning of predators. Here we explored whether episodic acidification could influence the ability of juvenile rainbow trout to learn to recognize an unknown predator and then generalize this recognition to a closely related predator. Trout were conditioned to recognize the odor of pumpkinseed sunfish under circumneutral (~pH 7) conditions, and then tested for recognition of pumpkinseed or longear sunfish under both neutral or weakly acidic (~ pH 6) conditions. When tested for a response to pumpkinseed odor, we found no significant effect of predator odor pH: trout responded similarly regardless of pH. Moreover, under neutral conditions, trout were able to generalize their recognition to the odor of longear sunfish. However, the trout could not generalize their recognition of the longear sunfish under acidic conditions. Given the widespread occurrence of anthropogenic acidification, acid-mediated impairment of predator recognition and generalization may be a pervasive problem for freshwater salmonid populations and other aquatic organisms. © 2012 Elsevier B.V. All rights reserved.

1. Introduction Prey animals are under constant pressure to balance the conflicting demands of predator avoidance with the need to forage, defend territories and/or court potential mates (Lima and Dill, 1990). Learning to recognize potential threats allows prey to avoid realistic predation risks while reducing the costs associated with responding to ecologically irrelevant cues (Dall et al., 2005; Brown et al., 2011a). Learning, however, is a ⁎ Corresponding author. Tel.: +1 514 848 2424x4327; fax: +1 514 848 2881. E-mail addresses: [email protected] (G.E. Brown), [email protected] (C.K. Elvidge), [email protected] (M.C.O. Ferrari), [email protected] (D.P. Chivers). 0048-9697/$ – see front matter © 2012 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.scitotenv.2012.09.026

potentially costly process (Mery and Kawecki, 2003). Under natural conditions, prey may need to survive an initial high risk encounter with a novel predator in order to acquire recognition of the relevant visual and/or chemosensory risk cues (Ferrari et al., 2007). Taxonomically related members of predator guilds with similar foraging behaviors and habitat preferences may pose similar threats to prey (Popova, 1978). Consequently, if prey are able to generalize their recognition of one predator to similar predator types with which they have no experience, they may be able to effectively reduce the potential costs associated with learning (Ferrari et al., 2008). There are a handful of studies that have demonstrated generalization of predator recognition abilities. In the first study, Griffin et al. (2001) demonstrated that tammar wallabies (Macroptus eugenii)

G.E. Brown et al. / Science of the Total Environment 439 (2012) 62–66

conditioned to recognize a red fox (Vulpes vulpes) as a predation threat exhibited increased avoidance of novel feral cats (Felis catus), but not juvenile goats (Capra hircus). The similarity in appearance between the fox and the cat allowed for generalization by the wallabies. Ferrari et al. (2007) have shown that predator naïve fathead minnows (Pimephales promelas) trained to recognize the odor of lake trout (Salvelinus namaycush) respond to the chemosensory cues of the taxonomically related brook charr (Salvelinus fontinalis) and rainbow trout (Onchorhynchus mykiss) but not the distantly related northern pike (Esox lucius). Other examples of generalization have been documented in mammals (Stankowich and Coss, 2007), reptiles (Webb et al., 2009) and embryonic and larval amphibians (Ferrari and Chivers, 2009; Ferrari et al., 2009). Recently, Ferrari et al. (2010) have shown that the ability to generalize learned predator recognition can be impaired by anthropogenic changes in water clarity. Predatornaïve fathead minnows conditioned to recognize the sight of brown trout (Salmo trutta) were able to generalize their predator avoidance to the sight of rainbow trout under clear water conditions, but not under turbid water conditions. Consequently, increased turbidity as a result of habitat degradation may represent a significant constraint on the ability of prey to generalize acquired predator recognition information. Within aquatic prey systems, one widely demonstrated form of learning occurs when prey are exposed to the cues of a novel predator paired with the sight or odor of an injured conspecific (Brown and Chivers, 2005; Ferrari et al., 2010; Brown et al., 2011a). Prey animals subsequently respond to the learned predator leading to an increased probability of survival (Mirza and Chivers, 2001a; Darwish et al., 2005). However, Leduc et al. (2004, 2007) demonstrated that response to alarm cues and learning can be impaired by acidification of waterbodies. Under weakly acidic conditions (pH~ 6.0–6.4) the ability of juvenile Atlantic salmon (Salmo salar) to learn the predator is inhibited because the damage-released chemical alarm cues are rendered inactive. Episodic acidification resulting from high precipitation events or spring snowmelts (Van Sickle et al., 1996; Serrano et al., 2008) may lead to rapid change in ambient pH (Baker et al., 1996; Komai et al., 2002; Leduc et al., 2009). Depending upon the acid neutralizing capacity (ANC) of a catchment area, even relatively small amounts of rainfall and subsequent input from watershed runoff may significantly reduce ambient pH. For example, Komai et al. (2002) noted a change of ~ 0.3 pH units after as little as 10 mm of rainfall, while Laudon and Bishop (2002) recorded a reduction of 2.4 pH units following significant rainfalls in streams with low ANC. Episodic acidification may thus create instances in which prey are able to learn to recognize the odor of a novel predator at circumneutral pH and be subsequently exposed to the learned odor at a markedly different pH. One study has examined how episodic acidification could influence learning of predators. Smith et al. (2008) conditioned juvenile rainbow trout to recognize pumpkinseed (Lepomis gibbosus) odor that had been buffered to neutral or weakly acidic levels and tested for recognition of pumpkinseed odor at either pH level. When tested for recognition two days post-conditioning, they found strong antipredator responses if the pH of the predator odor matched the pH at the time of conditioning. However, they found a weaker response (significant reduction in only one of two behavioral measures) if the pH differed from that experienced during conditioning. When tested for recognition seven days post-conditioning, they found no significant response if the predator odor pH differed from that experienced during conditioning. These results suggest that episodic acidification can impair the strength and retention of predator recognition learning. It remains unknown, however, if changes in ambient pH will influence the generalization of acquired predator cues by freshwater fishes as do changes in turbidity (Ferrari et al., 2010). Here we test the potential impacts of weak acidification on the ability of juvenile rainbow trout conditioned to recognize the odor of pumpkinseed as predators to generalize this learned information and avoid the chemical cues of congeneric longear sunfish (Lepomis megalotis).

63

2. Methods 2.1. Test fish and chemical cues Juvenile rainbow trout (approximately 150) were obtained from a commercial hatchery (Pisciculture des Arpents Verts, Ste Edwidge-deClifton, Quebec) and acclimated to laboratory conditions for ~1 month prior to testing. Trout were held in 390 L recirculating holding tanks (~18 °C, ~pH 7.0) under a 14:10 L:D cycle and fed ad libitum daily with commercial trout pellets (Corey Mills). Adult pumpkinseed (n=2) and longear sunfish (n=2) used as donors of predator odors were seined from Lachine Canal, Montréal, Québec, and held under identical conditions to the trout. Pumpkinseeds and longear sunfish were fed ad libitum with brine shrimp and commercial cichlid pellets daily for at least two weeks prior to use. Damage-released chemical alarm cues used during the conditioning phase were obtained from skin filets removed from donor trout (N = 12, mean ± SD fork length = 5.22 ± 0.21 cm). We euthanized donors via cervical dislocation, removed skin filets from both lateral flanks of donors and immediately placed them into chilled distilled water. We homogenized and filtered the skin samples through polyester wool, and diluted the solution to the desired final volume. We collected a total of 72.64 cm 2 of skin (in a final volume of 720 mL of distilled water). We collected predator odors from two adult pumpkinseed (14.2 and 14.9 cm standard length) and two longear sunfish (15.5 and 16.1 cm standard length) donors. Predators were placed individually into aerated but unfiltered 37 L glass aquaria filled with 15 L of dechlorinated tap water and a gravel substrate for 3 days, during which they were unfed to preclude the collection of dietary cues. After the three days, predators were returned to holding tanks and water from the two pumpkinseed or longear sunfish donor tanks was combined and filtered to generate a single predator odor for each species. We froze all chemical cues in 60 mL aliquots at −20 °C until needed. 2.2. Experimental protocol Individual rainbow trout (4.06 ± 0.61 cm FL at time of testing) were placed into 37 L glass aquaria filled with 35 L of dechlorinated tap water (~ 18 °C, pH 6.9–7.2) and allowed to acclimate overnight. The tanks were unfiltered, with a gravel substrate and a single airstone at the back wall. Each tank was wrapped in black plastic to prevent visual communication between tanks. A 1.5 m length of airline tubing was attached near the airstone to allow the introduction of chemical cues without mechanical disturbance to the test fish. Immediately prior to conducting a trial, we withdrew and discarded 60 mL of water through the stimulus injection tube, then withdrew and retained an additional 60 mL of water. Behavioral observations consisted of a 5 min pre-stimulus period, the injection of chemical stimuli and then the 60 mL of retained tank water to ensure delivery of the stimuli, followed by a 5 min post-stimulus period. 2.3. Conditioning phase Acclimated trout were fed with commercial trout pellets one hour prior to conditioning. Trout were either conditioned (10 mL of conspecific alarm cue+ 10 mL of pumpkinseed odor) or pseudo-conditioned (10 mL of distilled water+ 10 mL of pumpkinseed odor) to recognize the chemical cues of predators. During the conditioning phase, the pumpkinseed odor was untreated (see below). We measured the pH of the pumpkinseed odor immediately prior to use with a Multiline P4 digital meter (WTW Wissenschaftlich-Technische Werkstätten GmbH, Weilheim, Germany) calibrated daily. The pH range of untreated pumpkinseed odor was 7.05–7.20. Approximately 1 h post-conditioning, trout were transferred to identical tanks and tested ~24 h later. We did not conduct behavioral observations during the conditioning phase.

64

G.E. Brown et al. / Science of the Total Environment 439 (2012) 62–66

2

Following an identical protocol to the conditioning trials, recognition trials consisted of 10 mL injections of 1) pumpkinseed odor at pH 7.0 (PS-N), 2) pumpkinseed odor titrated to pH 6.0 (PS-A), 3) longear sunfish at pH 7.0 (LE-N) or 4) longear sunfish odor titrated to pH 6.0 (LE-A). Predator odors were titrated with dilute H2SO4 after Leduc et al. (2004) and Smith et al. (2008). Given the small quantities of H2SO4 used to acidify the predator odors and the dilution factor in the test tanks (i.e., 10 mL in 35 L) there was no discernable effect of the acidic stimuli on the pH of tank water.

1

During both pre- and post-stimulus observations, we recorded the time spent moving and the number of foraging attempts. The pre-trial introductions of trout pellets ensured both the initiation of foraging activity and that sufficient food particles remained throughout the observation period to quantify foraging (Vavrek et al., 2008). A foraging attempt was defined as a pecking motion involving visible opercular expansion towards the substrate, or within the water column. Reductions in movement and foraging rates are consistent with an increased predator avoidance response under laboratory conditions (Martel and Dill, 1993; Brown and Smith, 1997). We conducted a total of 15 replicates for each of 8 treatment combinations (N = 120 trials) and each fish was tested only once. All observations were conducted blind to the treatment.

2.6. Statistical analysis For both time spent moving and the frequency of foraging attempts, we calculated the change between pre- and post-stimulus observation periods and used these difference scores as dependent variables. As the data did not meet the assumptions of normality and had heteroskedastic variances, we ranked the difference scores and conducted our analysis on the ranked values (Quinn and Keough, 2002). We tested for the effects of conditioning stimulus (alarm cue vs. distilled water), predator odor (pumpkinseed vs. longear sunfish) and predator odor pH (neutral or acidic) using a MANOVA. To further explore the observed main effects of our model, we conducted separate MANOVAS for trout exposed to pumpkinseed and longear sunfish odor.

A

0 -1 -2 -3 -4 -5 -6 40

Mean change in time moving (sec)

2.5. Behavioral measures

Mean change in foraging attempts

2.4. Recognition testing

B

20

0

-20

-40

-60

PS Neutral

PS Acidic

LE Neutral

LE Acidic

Fig. 1. Mean (±SE) differences in (A) number of foraging attempts and (B) time spent moving demonstrated by juvenile trout conditioned (closed bars) or pseudo-conditioned (open bars) to recognize pumpkinseed odor as a risky cue in response to neutral (pH~7) or acidified (pH~6) odors of pumpkinseed or longear sunfish. N=15 per treatment combination.

3. Results Our overall MANOVA revealed significant main effects for conditioning stimulus (F2, 111 = 16.10, P b 0.001), predator type (F2, 111 = 6.42, P = 0.002) and predator odor pH (F2, 111 = 3.17, P = 0.046), as well as a significant conditioning stimulus × predator odor interaction (F2, 111 = 4.45, P = 0.014; Fig. 1). All other interactions were not significant. When considering trout tested for the response to pumpkinseed odor only, we found a significant effect of conditioning stimulus (F2, 55 = 18.07, P b 0.001), but no effect of predator odor pH (F2, 55 = 2.12, P = 0.129) nor a conditioning stimulus × pH interaction (F2, 55 = 0.91, P = 0.41). However, for trout conditioned to pumpkinseed and tested for response to longear sunfish, we found a significant conditioning stimulus × predator odor pH interaction (F2, 55 = 3.34, P = 0.043). Trout conditioned to recognize pumpkinseed odor were capable of generalizing their response to the odor of the related longear sunfish, but only at the neutral pH (Mann–Whitney U test, foraging attempts: P = 0.041, Fig. 1A; time moving: P = 0.021, Fig. 1B). When the predator odors were titrated to pH 6.0, trout were no longer able to recognize the longear sunfish cue (foraging attempts, P = 0.35, Fig. 1A; time moving: P = 0.29, Fig. 1B).

4. Discussion Our results demonstrate that juvenile rainbow trout are able to generalize the acquired recognition of pumpkinseeds to the closely related longear sunfish, which is in agreement with previous work. Brown et al. (2011b) demonstrated that juvenile rainbow trout conditioned to recognize the chemical cues of pumpkinseed as a predation threat could generalize this acquired information to the closely related congener, longear sunfish, and the confamiliar rock bass (Ambloplites rupestris), but not to the more distantly related yellow perch (Perca flavescens). Similar generalized predator recognition has been demonstrated in other aquatic (Ferrari et al., 2007) and terrestrial (Griffin et al., 2001; Webb et al., 2009) prey species. The capacity to respond to learned information regarding the identity of local predation threats is essential to the ability of prey to balance the conflicting demands of predator avoidance and the need to engage in other fitness related activities (Dall et al., 2005; Ferrari et al., 2005; Brown et al., 2011a). Taxonomically related predators often look similar and have similar chemical ‘signatures’ (sensu Wyatt, 2010). Given that closely related predators also share

G.E. Brown et al. / Science of the Total Environment 439 (2012) 62–66

common foraging tactics, as is the case with pumpkinseed and longear sunfish (Popova, 1978), they likely pose similar levels of risk to prey. Thus, prey likely benefit from responding to both learned predator cues and generalizing their response to similar predators (Ferrari et al., 2007; Brown et al., 2011b). Our results convincingly demonstrate that the ability of trout to generalize recognition of predators is impaired under weakly acidic conditions. While trout exhibited significant antipredator responses towards pumpkinseed odor at lower pH, they did not exhibit any observable response towards longear sunfish odor in the reduced pH treatment. Indeed, the weak acidification did not cause a simple reduction in the intensity of response, it completely eliminated the response to the predator. Given the demonstrated survival benefits associated with learned predator recognition (Mirza and Chivers, 2001a,b; Darwish et al., 2005), acidification may lead to reduced survival and recruitment. Previous studies have demonstrated that chemically-mediated learning can be impaired due to the chemical degradation of the alarm cue under weakly acidic conditions (Leduc et al., 2004, 2007). The current results, and those of Smith et al. (2008), suggest that a change in pH may also lead to partial or complete degradation of predator odor cues as well. The loss of generalization under weakly acidic conditions, as tested here, is likely due to such effects. We would expect the chemical signatures of closely related predators to be similar, but not identical (Brown et al., 2011b). If so, any degradation of the predator odor, resulting in a lowered functional concentration or structural change in the predator odor itself, would result in reduced recognition of learned information (Smith et al., 2008). Recent studies suggest that reduced ambient pH may have additional effects on the learning ability of prey species. Ferrari et al. (2012) and Nilsson et al. (2012) demonstrate that elevated CO2 resulting in a reduction in pH leads to cognitive impairment of larval coral reef prey species. For example, predator naïve larval damselfish (Pomacentrus amboinesis) exposed to elevated CO2 levels and then conditioned were unable to learn to recognize a novel predator. Such a mechanism is unlikely to explain the current results, as both the alarm cue and the water within the holding and testing tanks were left untreated (pH ~ 7.0). However, under natural conditions, prey would be exposed to reduced ambient pH. As a result, the chemical degradation of the predator cue, coupled with the prey's exposure to weakly acidic conditions might be expected to have an additive (or synergistic) effect on the impairment of predator recognition learning. While a great deal is known regarding the effects of heavy (pH b 5.0) and moderate (pHb 6.0) levels of acidification on aquatic communities, less is known about the sublethal effects of weak acidification (pH~6.0–7.0). Below ~pH 5.0, decreases in fish abundance and recruitment and increased physiological stress and mortality have been reported (Environment Canada, 1997). Moderate pH levels (~5.0–6.0) have been shown to impair swimming ability, growth rates, immune function and sensory perceptions (Lorz and McPherson, 1976; Wilson et al., 1994). Some sublethal effects have also been demonstrated under weakly acidic conditions (~pH 6.0–6.5). For example, upstream migrations and spawning of hime (landlocked sockeye) salmon (Oncorhynchus nerka) is severely reduced within the weakly acidic range (Ikuta et al., 2001, 2003). Damage-released chemical alarm cues in cyprinids (Brown et al., 2002), centrarchids (Leduc et al., 2003) and salmonids (Leduc et al., 2004, 2006) are not longer detectable under weakly acidic conditions. This loss of function is due to the chemical degradation of the alarm cue itself rather than a loss of olfactory sensitivity of those detecting the cue (Brown et al., 2002; Leduc et al., 2010). Acidic inputs to natural waterways may result from natural and/or anthropogenic sources (Gorham et al., 1986). Regardless of the source, pulses of acidity resulting from episodic precipitation may result in dramatic fluctuations in ambient pH. These fluctuations may in turn generate conditions whereby prey may experience a novel predator cue under near neutral pH but subsequently encounter that cue at a very different pH. Recently, Smith et al. (2008) showed that episodic

65

acidification could influence the strength and retention of learning. Our current results add to the documented detrimental effects of episodic acidification, demonstrating that generalized predator recognition is also impaired. Given that periods of episodic acidification are often coupled with increased turbidity of stream habitats, prey fishes may be doubly impaired as a result of the lack of generalization of both chemical cues (current study) and visual cues (Ferrari et al., 2010). In recent decades, there has been a global reduction in the rates of anthropogenic acid deposition (Clair et al., 2011). Why then should we be concerned with potential effects of weakly acidic conditions on fish communities? The depletion of lake and stream-bed acid neutralizing capacity (ANC) due to decades of anthropogenic acidification means that despite reductions in acidic inputs, many watershed will remain weakly acidic. Moreover, as heavily impacted watersheds recover, they may persist at weak levels of ambient acidity. For example, Clair et al. (2004) demonstrated that despite a 50% reduction in acidic inputs, the ANC of many Eastern Canadian lakes has not improved. They suggest that even if acidic inputs are completely eliminated, many Eastern North American watersheds will not recover to pre-industrial pH levels for at least 100 years. As such, the combination of continued weakly acidic inputs and depleted ANC may lead to suboptimal conditions for prey populations to learn and avoid ecologically relevant predation cues. As such, we argue that more attention is required to further examine the potential sublethal effects of acidification on aquatic communities, especially as it pertains to management and restocking programs.

Acknowledgments We wish to thank Drs. James Grant and Robert Weladji for comments on earlier versions of this manuscript. Svetlana Dolgova, Dale McNaught, Marc-Andre Coutier and Patrick H. Malka provided assistance with fish maintenance and data collection. Financial support was provided by Concordia University and the Natural Science and Engineering Research Council of Canada (NSERC) to GEB and by the University of Saskatchewan and NSERC to DPC and MCOF. All work reported herein was conducted in accordance with Concordia University Animal Research Ethics protocol #AREC-2010-BROW.

References Baker JP, Van Sickle J, Gagen CJ, DeWalle DR, Sharpe WE, Carline RF, et al. Episodic acidification of small streams in the Northeastern United States: effects on fish populations. Ecol Appl 1996;6:422–37. Brown GE, Chivers DP. Learning as an adaptive response to predation. In: Barbosa P, Castellanos I, editors. Ecology of predator–prey interactions. Oxford: Oxford University Press; 2005. p. 34–54. Brown GE, Smith RJF. Conspecific skin extract elicits anti-predator behaviour in juvenile rainbow trout (Oncorhynchus mykiss). Can J Zool 1997;75:1916–22. Brown GE, Adrian Jr JC, Lewis MG, Tower JM. The effects of reduced pH on ostariophysan alarm signalling. Can J Fish Aquat Sci 2002;59:1331–8. Brown GE, Ferrari MCO, Chivers DP. Learning about danger: chemical alarm cues and threat-sensitive assessment of predation risk by fishes. In: Brown C, Laland K, Krause J, editors. Fish cognition and behaviour. West Sussex, United Kingdom: Blackwell; 2011a. p. 59–80. Brown GE, Ferrari MCO, Malka PH, Russo S, Tresidder M, Chivers DP. Generalization of predators and nonpredators by juvenile rainbow trout: learning what is and is not a threat. Anim Behav 2011b;81:1249–56. Clair TA, Dennis IF, Amiro PG, Cosby BJ. Past and future chemistry changes in acidified Nova Scotian Atlantic salmon (Salmo salar) rivers: a dynamic modeling approach. Can J Fish Aquat Sci 2004;64:1965–75. Clair TA, Dennis IF, Vet R. Water chemistry and dissolved organic carbon trends in lakes from Canada's Atlantic Provinces: no recovery from acidification measured after 25 years of lake monitoring. Can J Fish Aquat Sci 2011;68:663–74. Dall SRX, Giraldeau L-A, Olsson O, McNamara JM, Stephens DW. Information and its use by animals in evolutionary ecology. Trends Ecol Evol 2005;20:187–93. Darwish TL, Mirza RS, Leduc AOHC, Brown GE. Acquired recognition of novel predator odour cocktails by juvenile glowlight tetras. Anim Behav 2005;70:83–9. Environment Canada. Canadian acid rain assessment volume 3: the effects on Canada's lakes, rivers and wetlands. CA-Quebec: Ministry of the Environment; 1997. Ferrari MCO, Chivers DP. Sophisticated early life lessons: generalization of predator recognition by frog embryos. Behav Ecol 2009;20:1295–8.

66

G.E. Brown et al. / Science of the Total Environment 439 (2012) 62–66

Ferrari MCO, Trowell JJ, Brown GE, Chivers DP. The role of learning in the development of threat-sensitive predator avoidance in fathead minnows. Anim Behav 2005;70: 777–84. Ferrari MCO, Gonzalo A, Messier F, Chivers DP. Generalization of learned predator recognition: an experimental test and framework for future studies. Proc R Soc Lond B 2007;274:1853–9. Ferrari MCO, Vavrek MA, Elvidge CK, Fridman B, Chivers DP, Brown GE. Sensory complementation and the acquisition of predator recognition by salmonid fishes. Behav Ecol Sociobiol 2008;63:113–21. Ferrari MCO, Brown GE, Messier F, Chivers DP. Threat-sensitive generalization of predator recognition by amphibians. Behav Ecol Sociobiol 2009;63:1369–75. Ferrari MCO, Lysak KR, Chivers DP. Turbidity as an ecological constraint on learned predator recognition and generalization in a prey fish. Anim Behav 2010;79:515–9. Ferrari MCO, Manassa RP, Dixson DL, Munday PL, McCormick MI, Meekan MG, et al. Effects of ocean acidification on learning in coral reef fishes. PLoS One 2012;7: e31478. Gorham E, Underwood JK, Martin FB, Ogden JG. Natural and anthropogenic causes of lake acidification in Nova Scotia. Nature 1986;324:451–3. Griffin AS, Evans CS, Blumstein DT. Learning specificity in acquired predator recognition. Anim Behav 2001;62:577–89. Ikuta K, Munakata A, Aida K, Amano M, Kitamura S. Effects of low pH on upstream migratory behavior in land-locked sockeye salmon Oncorhynchus nerka. Water Air Soil Pollut 2001;130:99-106. Ikuta K, Suzuki Y, Kitamura S. Effects of low pH on the reproductive behavior of salmonid fishes. Fish Physiol Biochem 2003;28:407–10. Komai Y, Umemoto S, Inoue T. Application of an automatic sampling and measurement system to a mountainous stream invesitigation during rain events. Water Sci Technol 2002;45:213–8. Laudon H, Bishop K. Episodic stream water pH decline during autumn storms following a summer drought in northern Sweden. Hydrol Process 2002;19:1725–33. Leduc AOHC, Noseworthy MK, Adrian Jr JC, Brown GE. Detection of conspecific and heterospecific alarm signals by juvenile pumpkinseed under weak acidic conditions. J Fish Biol 2003;63:1331–6. Leduc AOHC, Ferrari MCO, Kelly JM, Brown GE. Learning to recognize novel predators under weakly acidic conditions: the effects of reduced pH on acquired predator recognition by juvenile rainbow trout. Chemoecology 2004;14:107–12. Leduc AOHC, Roh E, Harvey MC, Brown GE. Impaired detection of chemical alarm cues by juvenile wild Atlantic salmon (Salmo salar) in a weakly acidic environment. Can J Fish Aquat Sci 2006;63:2356–63. Leduc AOHC, Roh E, Breau C, Brown GE. Effects of ambient acidity on chemosensory learning: an example of an environmental constraint on acquired predator recognition in wild juvenile Atlantic salmon (Salmo solar). Ecol Freshw Fish 2007;16: 385–94. Leduc AOHC, Roh E, Brown GE. Effects of acid rainfall on juvenile Atlantic salmon (Salmo salar) antipredator behaviour: loss of chemical alarm function and potential survival consequences during predation. Mar Freshw Res 2009;60:1223–30. Leduc AOHC, Roh E, Macnaughton CJ, Benz F, Rosenfeld J, Brown GE. Ambient pH and the response to chemical alarm cues in juvenile Atlantic salmon (Salmo salar): mechanisms of reduced behavioural responses. Trans Am Fish Soc 2010;139: 117–28.

Lima SL, Dill LM. Behavioural decisions made under the risk of predation: a review and prospectus. Can J Zool 1990;68:619–40. http://dx.doi.org/10.1139/z90-092. Lorz HW, McPherson BP. Effects of copper or zinc in fresh water on the adaptation to sea water and ATPase activity, and the effects of copper on migratory disposition of coho salmon (Oncorhynchus kisutch). J Fish Res Board Can 1976;33:2023–30. Martel G, Dill LM. Feeding and aggressive behaviors in juvenile coho salmon (Oncorhynchus kisutch) under chemically-mediated risk of predation. Behav Ecol Sociobiol 1993;32: 365–70. Mery F, Kawecki TJ. A fitness cost of learning ability in Drosophila melanogaster. Proc R Soc Lond B 2003;270:2465–9. Mirza RS, Chivers DP. Chemical alarm signals enhance survival of brook charr (Salvelinus fontinalis) during encounters with predatory chain pickerel (Esox niger). Ethology 2001a;107:989-1005. Mirza RS, Chivers DP. Do chemical alarm signals enhance survival of aquatic vertebrates? An analysis of the current research paradigm. In: Marchlewska-Koj A, Lepri JJ, MullerSchwarze D, editors. Chemical signals in vertebrates IX. New York: Kluwer Academic; 2001b. p. 19–26. Nilsson GE, Dixon DI, Domenici P, McCormick MI, Sørensen C, Watson S-A, et al. Near-future carbon dioxide levels alter fish behaivour by interfering with neurotransmitter function. Nat Clim Change 2012;2:201–4. Popova OA. The role of predaceous fish in ecosystems. In: Gerking SD, editor. Ecology of freshwater fish production. Oxford: Blackwell Scientific Publications; 1978. p. 215–49. Quinn GP, Keough MJ. Experimental design and data analysis for biologists. Cambridge: Cambridge University Press; 2002. Serrano I, Buffam I, Palm D, Brannas E, Laudon H. Thresholds for survival of brown trout during the spring flood acid pulse in streams high in dissolved organic carbon. Trans Am Fish Soc 2008;137:1363–77. Smith JJ, Leduc AOHC, Brown GE. Chemically mediated learning in juvenile rainbow trout. Does predator odour pH influence intensity and retention of acquired predator recognition? J Fish Biol 2008;72:1750–60. Stankowich T, Coss RG. The re-emergence of felid camouflage with the decay of predator recognition in deer under relaxed selection. Proc R Soc Lond B 2007;274: 175–82. Van Sickle J, Baker JP, Simonin HA, Baldigo BP, Krester WA, Sharpe WE. Episodic acidification of small streams in the northwestern United States: fish mortality in field bioassays. Ecol Appl 1996;6:408–21. Vavrek MA, Elvidge CK, DeCaire R, Belland B, Jackson CD, Brown GE. Disturbance cues in freshwater prey fishes: do juvenile convict cichlids and rainbow trout respond to ammonium as an ‘early warning’ signal? Chemoecology 2008;18:255–61. Webb JK, Du WG, Pike DA, Shine R. Chemical cues from both dangerous and nondangerous snakes elicit antipredaotr behaivours from a nocturnal lizard. Anim Behav 2009;77: 1471–8. Wilson RW, Wright PM, Munger S, Wood CM. Ammonia excretion in freshwater rainbow trout (Oncorhynchus mykiss) and the importance of gill boundary layer acidification: lack of evidence for Na+/NH4+ exchange. J Exp Biol 1994;191:37–58. Wyatt TD. Pheromones and signature mixtures: defining species-wide signals and variable cues for identity in both invertebrates and vertebrates. J Comp Physiol A 2010;10:685–700.