Learning and context-specific exploration behaviour in hatchery and wild brown trout

Applied Animal Behaviour Science 132 (2011) 90–99

Contents lists available at ScienceDirect

Applied Animal Behaviour Science journal homepage: www.elsevier.com/locate/applanim

Learning and context-specific exploration behaviour in hatchery and wild brown trout Bart Adriaenssens a,b,∗ , Jörgen I. Johnsson a a b

Department of Zoology, Animal Ecology, University of Gothenburg, Box 463, SE-405 30 Göteborg, Sweden School of Biological, Earth and Environmental Sciences, University of New South Wales, Sydney 2052, Australia

a r t i c l e

i n f o

Article history: Accepted 2 March 2011 Available online 5 April 2011 Keywords: Hatchery Context Personality Exploration Learning Brown trout

a b s t r a c t In this study we investigate whether rearing environment (wild vs. hatchery) affects the ability of brown trout parr (Salmo trutta) to learn two foraging tasks. Hatchery- and wildreared brown trout were trained in two different foraging tasks: locating food hidden in a maze and finding a cryptic prey, and their performance within and across tasks was compared. Fish reduced their search time for cryptic prey, but not maze search time, by learning. In contrast to most previous studies hatchery-reared trout generally tended to be more successful feeders and showed faster learning than wild trout when foraging on cryptic prey. This appeared to be due to motivational effects rather than based on cognitive skills. In addition, we examine whether exploration behaviour in brown trout is repeatable across time and context (i.e. reflecting a behavioural syndrome). Individual exploration tendency was repeatable within tasks, suggesting the occurrence of personality in brown trout. However, individuals that were fast explorers in the cryptic prey task were not necessarily fast to explore the maze. Thus, a context-specific behavioural syndrome was found to best explain exploratory behaviour for both hatchery and wild trout. However, repeatability of exploration behaviour within tasks differed between hatchery and wild trout, where wild trout were found to be more consistent in their exploration strategy. © 2011 Elsevier B.V. All rights reserved.

1. Introduction Individual variation in behaviours such as learning or risk-taking have recently become of interest in fish farming and welfare (Brännäs and Johnsson, 2008; Huntingford and Adams, 2005). Releasing programs of captive-bred or translocated animals rely upon the assumption that animals can rapidly adjust to a wide diversity of novel challenges upon release. Unfortunately, however, many captive breeding programs fail to raise individuals with natural behaviour, with high mortalities upon release as a result (Araki et al., 2008). Recent experiments on fish show that

∗ Corresponding author. Tel.: +61 2 9385 3413; fax: +61 2 9385 1558. E-mail addresses: [email protected], [email protected] (B. Adriaenssens). 0168-1591/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.applanim.2011.03.005

the lack of stimulus variation in captive rearing conditions will influence the phenotype at many different levels, ranging from physiology, neurology to behaviour (reviewed by Brännäs and Johnsson, 2008; Huntingford, 2004; Olla et al., 1998). Sundström and Johnsson (2001), for instance, showed that hatchery reared brown trout require more time than wild reared conspecifics to learn to forage on a novel prey item. Life-skill training has been suggested as a method to counter poor post-release performance of hatchery release programs (Brown and Laland, 2001; Griffin et al., 2000). However, the sensitivity to life-skill training may differ between different personality types (Sih and Bell, 2008). Consistent personality traits have been found in many species (Gosling, 2001; Wilson et al., 1994), and are related to the concept of behavioural syndromes, describing a suite of correlated behaviours across multiple functional

B. Adriaenssens, J.I. Johnsson / Applied Animal Behaviour Science 132 (2011) 90–99

contexts (Sih and Bell, 2008; Sih et al., 2004). One major axis of personality, the shyness–boldness continuum, has been shown to differ between hatchery and wild fish, where hatchery selection often results in more bold personality traits across different situations than in wild trout (Huntingford, 2004; Sundström et al., 2004b). Personality traits can relate to learning skills in different ways. A higher tendency to explore new environmental stimuli may, for instance, result in faster learning in bold than in shy individuals (Sneddon, 2003). Yet, other studies suggest that bold individuals are only better in routine tasks, whereas shy individuals are more sensitive to changes in the environment (Benus et al., 1990; Verbeek et al., 1994). In addition, learning may also affect the repeatability of personality traits. For example, experience may create a positive feedback loop on previously performed behaviours, thereby increasing their consistency (Wolf et al., 2008) or, alternatively, learning can break up consistency when behavioural types are not affected equally by experience (Sih et al., 2004). When behaviour is consistent across individuals, personality traits may be context-general, i.e. consistent across different situations, or context-specific and change according to experience or context (Coleman and Wilson, 1998). One aim of personality research is to understand why individuals behave consistently in some situations and more flexibly in others (Dingemanse et al., 2009). This study investigates behaviour of hatchery and wild reared brown trout (Salmo trutta) in two different foraging tasks: (1) a spatial orientation task: locating food hidden in a maze and (2) a visual discrimination task: finding a cryptically coloured prey. In addition we study whether the tendency to explore a novel situation, a personality trait closely related to boldness (Réale et al., 2007), is consistent within and across tasks. Brown trout were used as a model species for this study. Since structural complexity is generally high in natural salmonid streams (Höjesjö et al., 2004; Orpwood et al., 2003), spatial orientation skills are expected to be critical for finding food. Moreover, young salmonids are opportunistic feeders which prey on a wide range of prey types (De Crespin De Billy and UsseglioPolatera, 2002), some of which adjust crypsis and activity levels to avoid predation (Feltmate and Williams, 1989; Hargeby et al., 2004). A recent study further showed that prey crypsis reduces the foraging efficiency of wild brown trout parr which, however, are able to increase their foraging success by learning (Johnsson and Kjällman-Eriksson, 2008). We addressed four related aspects of behaviour: first, we study the ability of brown trout to learn in both tasks. Second, we test two hypotheses concerning the foraging skills of trout: does development in a hatchery environment impair the (1) spatial orientation ability and (2) visual acuity/object discrimination ability of trout? These hypotheses predict that hatchery-reared fish should be less able (1) to find food hidden in a maze, and (2) to find cryptic prey, compared with wild conspecifics. Third, we investigate whether exploration behaviour is repeatable within and across tasks and forms an exploration syndrome. The structure of this repeatability was compared with existing syndrome hypotheses from the literature. Fourth, we test

91

whether hatchery rearing affected the average exploration tendency of brown trout in these tasks, and/or the structure of the exploration syndrome thereby causing repeatability patterns of exploration behaviour to differ between hatchery and wild trout. 2. Materials and methods During autumn 2004, eggs were collected from a batch of spawning brown trout from Norumsån, a small coastal stream in western Sweden (58◦ N 11◦ E), and transported to a nearby hatchery (E.ON Hatchery Laholm). After hatching (10–15 March 2005), trout were reared according to standard hatchery procedures (Pennell and Barton, 1996). On 24 August 2005 these fish, further called hatchery trout, were transported and transferred to a holding tank at the Department of Zoology (University of Gothenburg). On 7 September 2005 wild brown trout (age 0 yr) were caught in Norumsån by electrofishing and transported to another holding tank at the Department of Zoology. A semi-random sub-sample was taken from the total number of caught wild individuals to minimize effects of catching methods on behaviour (Wilson et al., 1993). Wild and hatchery-reared fish were kept at similar density (1.4 l/ind) and fed live maggots (‘bronze pinkies’, Fibe AB, Överkalix) and frozen chironomid larvae (commercial fish food supplier) at rates of 1% of their total wet mass. After at least 19 days of acclimatization to laboratory conditions, 20 wild trout and 20 hatchery trout of similar fork length (mean ± SE = 56.3 ± 1.3 mm; t-test: t38 = 0.03, P = 0.976) were subjected in random order to an experimental sequence to measure their ability to learn to solve two complex foraging tasks: finding a foraging route in a maze and feeding on a cryptic prey. The number of trials (n = 6) and procedures chosen were based on previous experiments successfully using the maze (Johnsson and Sundström, 2007) and cryptic prey set-up (Johnsson and Kjällman-Eriksson, 2008) to study learning. To enhance foraging motivation, no food was provided to the focal fish on the day prior to start of the experimental sequence. 2.1. Maze solving An opaque PVC maze structure (Fig. 1, Johnsson and Sundström, 2007) subdivided a flow-through tank into 4 rooms: a starting room (S), a “wrong side” room (W), a “right side” room (R) and a food room (F). From S, the fish could access the side rooms (W and R) through two circular entrances (30 mm diameter) with landmarks (Fig. 1). A third circular entrance (35 mm diameter) positioned 30 mm higher than the others connected the R room with the F room, which contained the water outlet and food. The W room led to a dead end. However, the PVC wall between the F room and the W room was perforated to enable chemical cues from the food to spread into both side rooms. The relative position of the W and R rooms was kept constant throughout trials on the same subject, but altered between subjects to account for any side biases affecting fish movement. Disturbance by movements from outside was minimized by covering the lateral sides of the tanks with black plastic. Observations were performed through

92

B. Adriaenssens, J.I. Johnsson / Applied Animal Behaviour Science 132 (2011) 90–99

Fig. 2. Schematic representation of cryptic prey foraging tank. (1) Experimental compartment, provided with bottom cover and gravel matched to prey size and colour. (2) Start compartment with water outlet. The area delimited by the dashed white line represents the prey foraging arena containing the prey. The opaque PVC divider (dark gray) between both compartments was removed at the start of each trial. Small black arrows represent the water flow. The water level was kept constant at 20 cm.

Fig. 1. Schematic representation of the maze, seen from above. (S) Start room, (W) wrong side room, (R) right side room and (F) food room (A, net with 3 maggots). All black lines represent irremovable opaque PVC dividers while the gray line represents the removable PVC divider shielding entry towards the maze between different trials. To provide landmark information, a dark blue square was painted above the left entrance and a triangle above the right entrance. The water level was kept constant at 20 cm. Thin black arrows represent water flow.

a mirror placed above the fish tank (Johnsson, 1993). Focal fish were only allowed to access the maze during observations. Between trials a movable PVC divider covered the entries to the W and R rooms. This divider could be operated from a distance using a wire system. Every fourth evening, four to six focal fish (balanced for origin) were each transferred to the S room of a separate maze set-up with maze entries closed. Each fish was allowed to settle down in the S room during the next day. A first trial was performed on the morning of the second day upon transfer. Two minutes before the start of each trial a small net containing three homogenized maggots was positioned 10 cm underneath the water outlet in the F room. This net provided chemical cues and a non-satiating food reward in return for finding the F room. The first of six 30 min trials started when the divider covering the maze was lifted. The following data were recorded: (a) maze search time: the time from the start of the trial until the fish entered the F room, (b) maze search error: the amount of erroneous room entrances before entering the F room, where every room change exceeding the ideal path of two room changes towards the F room (from S to R and from R to F) was considered as an error.

(c) Maze latency to activity: the time until the fish first entered one of side the rooms (W or R). (d) The maze search activity: the total number of room changes. At the end of the 30 min period, the fish was gently moved back to the S room (without netting), whereupon the food net and the entries towards the maze were closed. Five hours after the start of the first trial a second one was carried out following the same procedure. This process was repeated during the third (trials 3 and 4) and fourth (trials 5 and 6) day. On the evening of the 4th day all fish were moved from the maze tank to the cryptic prey foraging tank. 2.2. Cryptic prey foraging Cryptic prey foraging was measured in a flow-through tank (for dimensions see Fig. 2) where a removable opaque PVC divider separated the start compartment from the experimental compartment. To avoid disturbance of the fish, the lateral sides of the tank were covered with black plastic. At the short end of the tank an opening in the plastic enabled observation of the foraging behaviour without disturbance of the focal fish. The floor of the experimental compartment was covered with a dark brown adhesive film matching the colour of the prey. Gravel similar in form and colour to the prey was spread over the folio to further enhance prey crypsis during the experiment. Upon transfer from the maze to the cryptic prey setup, focal fish were given one day to explore their new environment with the divider lifted to enable access to all compartments. The evening of the first day after transfer each fish was gently moved to the start compartment (without netting), and the start compartment was closed by lowering the divider. The first trial was conducted the morning of the second day after transfer. Two minutes before the start of the trial, a living maggot (bronze pinkies,

B. Adriaenssens, J.I. Johnsson / Applied Animal Behaviour Science 132 (2011) 90–99

Calliphoriade spp., length 8–10 mm, Fibe AB, Överkalix) was positioned in the foraging arena. Each maggot was pseudorandomly placed in one of six equally sized squares to ensure that each of the six squares was used only once. The divider was gently lifted to start the observations. The following behaviours were recorded during 15 min: (a) latency to activity: the time until the fish crossed the border between the start compartment and the experimental compartment, (b) cryptic prey search time: the time until the fish first touched the prey with the mouth, and (c) search activity: the time spent actively moving around. At the end of each trial, the fish were gently moved back to the start compartment (without netting) which was then closed. A second trial was performed five hours after the start of the first trial and this procedure was repeated again on the third (trials 3 and 4) and fourth (trials 5 and 6) day. At the end of this series of trials all fish were anaesthetized using 2-phenoxyethanol (0.5 ml/l) and measured for LF whereupon they were transferred to a holding tank to recover from handling. 2.3. Data handling and statistics All behavioural data was recorded using The Observer 4.1 software package for behavioural observations (Noldus Information Technology bv, Wageningen, The Netherlands). Individuals which remained inactive during the 30 min observation time in the maze trials were given a maximum maze search time score of 1800 s as a dummy value. Similarly, individuals remaining inactive during the full 15 min after being offered a cryptic prey were allocated a maximum cryptic prey search time of 900 s. Principal component analysis (PCA, PASW statistic 18.0) was used to summarize correlated behavioural measures of latency to activity, search activity and search time for each task to reduce the amount of variables in further analyses. In order to give similar weight to different variables across trials and treatments, PCA was performed on measures across trials for both hatchery and wild trout. Separate PCA for hatchery and wild trout resulted however in very similar factory loadings (highest factor loading difference = 0.065). This procedure generated one principal component (PC1, Table 1) as a score with eigenvalue above one for each task (Kaiser criterion, Kaiser, 1961). Starting from a set of a priori hypotheses derived from existing literature (Fig. 3), we used methods outlined by Dingemanse et al. (2010), see also Dochtermann and Jenkins (2007) to determine the syndrome structure that best explained patterns of repeatability in exploration across trials. This three-step method (1) explores repeatability patterns of exploration behaviour in a correlation matrix before (2) comparing repeatability with a priori hypotheses about the structure of the exploration syndrome with a combination of structural equation modelling using latent variables for each syndrome (SEM, Amos 17.0) and AIC-based model comparison, and (3) uses SEM to compare the exploration syndrome between hatchery and wild fish (multi-group comparison in Amos 17.0). Because exploration in some trials was slightly skewed, AIC values were calculated using model discrepancies based on 1000 bootstraps (Dochtermann and Jenkins, 2007). Also, prior to

93

Table 1 Solution of principal component analysis on the behavioural traits during tasks. Loadings, eigenvalues and explained variance are given for the two emerging principal components (PC maze and PC cryptic prey). Only components with eigenvalue above one were retained (Kaiser–Guttman criterion). Principal components (a) PC maze Behaviour Latency to activity (s) Search activity (s) Maze search time (s) Eigenvalue % of total variance (b) PC cryptic prey Behaviour Latency to activity (s) Search activity (s) Prey search time (s) Eigenvalue % of total variance

PC

0.884 −0.879 0.896 2.36 78.57

0.931 −0.840 0.853 2.30 76.06

SEM, variables were z-transformed within each trial and treatment to equalize their mean and standard deviation while maintaining their covariance. Models with AIC values differing by more than two from the best supported model were considered not supported statistically (Anderson and Burnham, 2002; Richards, 2005). We then tested whether hatchery and wild trout have the same exploration syndrome with multigroup analysis in Amos 17.0. To identify specific paths in the SEM that differed, we performed ttests between factor loadings for hatchery and wild fish for all separate paths (Dingemanse et al., 2010; Zar, 1999).

Fig. 3. A priori hypotheses for the form of the behavioural syndrome affecting repeatability of exploration behaviour across trials. Arrows represent the causal relationships between a latent variable (S, syndrome, variance fixed to 1) and exploration behaviour across trials. Model 0, the null model, describes unrepeated (independence) of exploration behaviour across trials. Model 1 suggests a context-general behavioural syndrome (S1) resulting in repeatable exploration behaviour within and across all trials, independent of the context (Sih et al., 2004). Models 2–4 represent hypotheses of context-specific behaviour (Coleman and Wilson, 1998), differing in the amount of repeatability. In model 2 two behavioural syndromes (S1 and S2) cause behaviour to be repeatable within, but not across tasks. In model 3, exploration behaviour is only repeatable within the maze task, and in model 4 exploration behaviour is only repeatable within the cryptic prey task.

94

B. Adriaenssens, J.I. Johnsson / Applied Animal Behaviour Science 132 (2011) 90–99

Data that did not meet the requirements for parametric analysis was analysed using non-parametric methods (PASW statistic 18.0). A critical value of ˛ = 0.05 was applied throughout unless otherwise stated. In case of multiple comparisons, P-values were only stated significant if they remained significant after adjusting for false discovery rate across all trials (FDR, Benjamini and Hochberg, 1995). The abbreviation ‘NS after FDR’ denotes probabilities which were significant before, but not after this correction. 3. Results 3.1. Maze solving Neither wild (Friedman ranked test: 2 = 7.0, Nw = 20, df = 5, P = 0.22) nor hatchery trout (Friedman ranked test: 2 = 9.8, Nh = 20, df = 5, P = 0.08) reduced search time in the maze significantly over the six trials. Hatchery trout generally tended to be more successful in solving the maze. However, a significant effect was only found during trial 3, when hatchery trout were faster to find the food (Mantel–Cox log rank test: 2 = 12.3, Nh = Nw = 20, df = 1, P < 0.001; all other trials P > 0.08) and more hatchery individuals solved the maze than did wild fish (Fig. 4a, Pearson Chi-square test: 2 = 12.1, Nh = Nw = 20, df = 1, P < 0.001; all other trials P > 0.1). 3.2. Accuracy during first maze exploration Six of eight wild trout found the F room in the maze during the first trial without making any errors, compared to five of 13 hatchery trout. The proportion of wild fish entering the F room without errors was higher than under random expectations (binomial test: Nw = 8, P = 0.004) while hatchery trout performance corresponded with random expectations (Fig. 5, binomial test: Nh = 13, P = 0.20).

Fig. 4. (a) Maze solving task: absolute number of hatchery („ filled squares) and wild (, open squares) trout finding the F room in each trial. **P < 0.001 and significant after FDR. (b) Cryptic prey task: absolute number of hatchery () and wild () trout finding the cryptic prey in each trial. (*) P = 0.01 but NS after FDR.

3.3. Cryptic prey foraging Both hatchery and wild trout shortened prey search time over the course of the 6 trials (Fig. 6, Friedman ranked test: 2 = 42.1, Nh + Nw = 39, df = 5, P < 0.001). Again, hatchery trout tended to show more efficient foraging. Only during the third trial, however, hatchery trout found the prey significantly faster than wild trout (Fig. 6, Mantel–Cox log rank test: 2 = 8.08, Nh = Nw = 20, P = 0.004; all other trials P > 0.04, NS after FDR). No significant difference was found in the frequency of hatchery and wild trout that found the prey (Fig. 4b, Pearson Chi-square test: Nh = Nw = 20, all P > 0.01, NS after FDR). 3.4. Exploration behaviour Measures of latency to activity, search activity and search time were correlated during each trial (Spearman correlations P < 0.05 for each trial in both tasks). Exploration tendency, a principal component describing this covariance, accounted for a total of 79% of the variance in the separate behavioural measures in maze search behaviour and 76% for cryptic prey search behaviour

Fig. 5. Frequency of observed maze room changes before entering the F room for both hatchery (filled bars) as wild (open bars) trout during the first trial plotted against the expected frequencies predicted by random movement (striped bars). **P = 0.004.

***

*

0.702*** 0.623** 0.526*

0.163 0.539* 0.316 0.363 0.588** 0.612** 0.291

Original significance value indicated with P < 0.05, coefficients remaining significant after following the procedure for controlling false discovery rate are shown in bold (Benjamini and Hochberg, 1995). Original significance value indicated with P < 0.01, coefficients remaining significant after following the procedure for controlling false discovery rate are shown in bold (Benjamini and Hochberg, 1995). Original significance value indicated with P < 0.001, coefficients remaining significant after following the procedure for controlling false discovery rate are shown in bold (Benjamini and Hochberg, 1995). **

*

0.554* 0.642** 0.656** 0.769*** 0.424 0.414 0.623** 0.527* 0.575** 0.694*** 0.759*** 0.648** 0.810*** 0.735*** 0.769***

0.133 −0.046 −0.263 0.181 −0.260 −0.210 0.161 0.249 0.033 0.487* 0.147 0.141 0.155 0.263 0.273 0.441 0.196 0.354

0.125 −0.235 −0.330 −0.112 −0.486 −0.051

0.273 0.674** 0.615** 0.460 0.177 0.320 0.279

0.570* 0.772***

0.821***

0.098 0.012 −0.263 0.072 −0.361 −0.244

0.583** 0.264 0.509* 0.566* 0.611**

0.342 −0.040 0.090 0.022 0.280 0.261 0.120 0.334 0.239 −0.026 0.351 0.393 0.367 0.076 0.188 0.157 0.432 0.397 0.296 0.116 0.250 0.072 0.259 0.347 0.134 0.127 0.202 −0.039 0.159 0.252 0.370 0.220 0.445* 0.162 0.381 0.407 0.679*** 0.635*** 0.623** 0.735*** 0.820*** 0.557* 0.676** 0.668** 0.769*** 0.486* 0.695* 0.412 0.557* 0.502*

2 1 6 5 4 3 2

0.512*

5 3

4

Our results show that: (1) trout learned to reduce their search time in the cryptic prey task, but not in the maze, (2) hatchery trout generally tended to be more efficient foragers, (3) a context-specific syndrome best explained

Cryptic prey task

4. Discussion

Maze task

Pairwise Spearman correlations showed that exploration tendency was often consistent across trials within the same context, but rarely across tasks (Table 2). The exploration syndrome was, both for hatchery and wild trout, best explained by model 2, in which exploration behaviour is context-specific, but consistent across trials within each context (Table 3). The connections between exploration behaviours within each exploration syndrome (i.e. factor loadings) were positive across all trials for hatchery and wild fish (Table 4). High exploration tendency in one trial therefore corresponded with high exploration in the others. Multigroup analysis showed that the factor loadings of the context-specific exploration syndrome (model 2) differed between hatchery and wild trout (2 36 = 220.39, P < 0.001). Post hoc tests show that this was mainly due to differences in repeatability in the cryptic prey task (trials 1, 2 and 5, Table 4). In hatchery trout only 40% of variance in exploration was explained by the context specific syndrome, while this was 68% for wild trout. Together, this shows that the extent of repeatability of exploration behaviour differed between hatchery and wild trout, where hatchery fish generally were less consistent in their exploration strategy (Tables 3 and 4).

1

3.5. Covariance structure of exploration tendency

Trial

(Table 1). In both tasks, exploration tendency showed high negative loadings for search activity and high positive loadings for latency to activity and search time and as a result explains a behavioural axis with high scores representing individuals with a low tendency to explore. Average scores for exploration behaviour did not differ between hatchery and wild trout during any of the trials (Mann–Whitney U test: Nh = Nw = 20, all P > 0.03, NS after FDR).

Table 2 Spearman correlation coefficients across trials. Hatchery trout are shown in the lower left part of the matrix, and wild fish in the upper right part.

Fig. 6. Median total cryptic prey search time with inter quartile distance for hatchery (, filled squares) and wild (, open squares) trout. Trials with a #-sign represent trials where the median value is equal to the maximum observation time (900 s). *P = 0.004.

95

Maze task 1 2 0.604** 3 0.680** 4 0.211 5 0.528* 6 0.366 Cryptic prey task 1 0.413 2 0.356 3 0.257 4 0.772*** 5 0.278 6 0.322

6

B. Adriaenssens, J.I. Johnsson / Applied Animal Behaviour Science 132 (2011) 90–99

96

B. Adriaenssens, J.I. Johnsson / Applied Animal Behaviour Science 132 (2011) 90–99

Table 3 Results of model comparison using Akaike information criterion (AIC) between models outlined in Fig. 3. Smaller AIC values suggest a better fit of the model to the data, while penalizing model complexity (k).

Hatchery Model 2 Model 1 Model 4 Model 3 Model 0 Wild Model 2 Model 4 Model 1 Model 3 Model 0

AIC

k

Model discrepancy

AIC

Dx

24 24 18 18 12

141.26 162.45 186.30 191.93 236.97

189.26 210.45 222.3 227.93 260.97

0 21.19 33.04 38.67 71.71

0.40 0.31 0.21 0.19 0.00

24 18 24 18 12

69.40 136.09 127.32 152.62 219.30

117.40 172.09 175.32 188.62 243.3

0 54.69 57.92 71.22 125.9

0.68 0.38 0.42 0.30 0.00

k = the number of parameters in the model, Dx = the proportion of variance explained by the model relative to model 0.

repeatability of exploration behaviour in both tasks, causing exploration to be repeatable within but not across tasks, and (4) exploration tendency did not differ between hatchery and wild trout, but the shape of the exploration syndrome differed between them where hatchery trout tended to be less consistent in their behaviour within the cryptic prey task. Below we discuss these findings in more detail.

In contrast, fish reduced search time for cryptic prey and during the last trial 85% of the fish found the prey. As fish were also fed maggots in the holding tanks, these were not novel prey to them. The improved foraging success was therefore most likely due to increased ability to detect the prey on the matching background and confirms earlier results on brown trout (Johnsson and Kjällman-Eriksson, 2008).

4.1. Foraging and learning ability

4.2. Differences in foraging between wild and hatchery trout

Our study shows that six trials are enough for brown trout to reduce their search time for a cryptic prey through learning, but not enough to reduce the search time for food in a maze. Earlier experiments have shown that salmonids can adjust their movement to obtained information about simple spatial differences in food availability (e.g. Braithwaite et al., 1996; Gotceitas and Godin, 1992; Sneddon, 2003). The lack of such spatial learning in the current experiment may be explained by the higher overall complexity of the task and the limited number of training trials offered. Therefore, further trials would be necessary to exclude the possibility that brown trout are able to reduce search time in the maze.

Most previous studies suggest that hatchery rearing reduces foraging success on live prey in salmonids (reviewed in Brännäs and Johnsson, 2008; Huntingford, 2004). Hatchery trout in our experiment, however, were occasionally more successful feeders and learned faster than wild trout when foraging on cryptic prey. Many previous studies concerned foraging behaviour when exposed to a novel prey (Brown et al., 2003; Sundström and Johnsson, 2001), while our study focused on the search for a known prey in a novel environment. The relative importance of different foraging skills may have varied between the tasks, which may explain the discrepancies among these studies.

Table 4 Comparisons of the factor loadings between hatchery and wild trout using model 2. Path to

Maze Trial 1 Trial 2 Trial 3 Trial 4 Trial 5 Trial 6 Cryptic prey Trial 1 Trial 2 Trial 3 Trial 4 Trial 5 Trial 6

Standardized factor loadings

SE

t

df

P

−0.031 −0.302 0.010 −0.306 −0.021 −0.043

0.150 0.143 0.165 0.166 0.097 0.108

−0.207 −2.117 0.058 −1.848 −0.212 −0.400

35 35 35 35 35 35

0.837 0.041 0.954 0.073 0.833 0.692

−0.342 −0.254 −0.213 −0.291 0.441 0.037

0.100 0.088 0.116 0.105 0.134 0.132

−3.420 −2.875 −1.832 −2.772 3.284 0.283

35 35 35 35 35 35

0.002 0.007 0.076 0.009 0.002 0.779

Hatchery

Wild

Difference

0.594 0.417 0.738 0.488 0.867 0.812

0.625 0.719 0.728 0.794 0.888 0.855

0.560 0.659 0.645 0.544 1.001 0.748

0.902 0.913 0.858 0.835 0.560 0.710

Probabilities remaining significant after procedure for controlling for false discovery rate are shown in bold (Benjamini and Hochberg, 1995). SE = standard error for each path calculated from the pooled sample.

B. Adriaenssens, J.I. Johnsson / Applied Animal Behaviour Science 132 (2011) 90–99

The visual search for cryptic food items is expected to impose a difficult computation task to the animal brain, forcing a narrower visual attention angle and reduction of the search rate (Dukas, 2002; Winkler and KothbauerHellmann, 2001). Although debated (Lima and Bednekoff, 1999), such tasks may impair vigilance towards predators (Dukas, 2002; Jones et al., 2006). This notion is supported by the observation that sticklebacks (Gasterosteus aculeatus), foraging on dense swarms of water fleas Daphnia magna, detect approaching predators less frequently than when foraging in a low prey density swarm (Milinski, 1984). Similarly, the wild trout in our study may have been more experienced with predators and therefore less motivated than hatchery fish to perform foraging activities that impair vigilance. Consistently, poeciliid fish, Brachyrhapis episcopi from high-predation populations take more time to learn to solve a spatial task than fish from low predation populations (Brown et al., 2005). Despite the higher foraging rate, hatchery trout had a lower maze foraging accuracy than wild conspecifics. In order to be efficient, a predator has to balance prey search speed with search accuracy (Chittka et al., 2009). Foraging bumblebees (Bombus terrestris), for example, show consistent individual variation along this axis ranging from fast and error-prone to slow and accurate individuals (Chittka et al., 2003). Our data suggest that hatchery trout paid the cost of increased activity with poorer navigation skills. The hatchery trout generally solved both tasks faster, but the direction of their movement in the maze did not differ from random expectations whereas wild trout moved more directly towards the food source (Fig. 5). Observations on growth-enhanced transgenic coho salmon (Oncorhynchus kisutch) showed similar patterns (Sundström et al., 2004a): increased foraging motivation caused transgenic salmon to seize a novel prey (Acheta domesticus) faster than normal conspecifics, but they lost time and energy on attacking unprofitable prey (artificial angling lure flies). Increased foraging speed may be highly beneficial when offered a monotonous hatchery diet in a simple environment. However, reduced foraging accuracy can incur energetic costs in a more natural environment which offers more possibilities to make errors (Hutchinson, 2005), and may partly explain the higher capture rate of hatchery salmonids on artificial lures (Mezzera and Largiader, 2001; Nelson et al., 2005). 4.3. Exploration syndrome Exploration tendency was consistent across trials within the same context. This supports previous studies identifying personality traits (i.e. the shyness–boldness) in fish and other animals (Réale et al., 2007; Sih et al., 2004; Wilson et al., 1994). In the cryptic prey task, exploration behaviour was consistent over a strong gradient of learning, further confirming the concept that personality traits are not antagonistic to behavioural flexibility but instead constrain flexibility relative to other individuals (for further discussion see Dingemanse et al., 2009; Sih and Bell, 2008). Individuals exhibited different exploration personalities in the maze and in the cryptic prey task. Exploratory behaviour in our study was therefore best described as

97

a context-specific personality trait (also domain-specific, Sih et al., 2004). Evidence from other animal species suggests that selection can result in both context-specific (Coleman and Wilson, 1998; Sinn et al., 2008; van Oers et al., 2005; Wilson and Stevens, 2005) and context-general behavioural syndromes (Dochtermann and Jenkins, 2007; Huntingford, 1976). Further research is needed to explore which environmental stimuli caused individual differences in trout behaviour to reshuffle between different contexts. 4.4. Differences in exploration between wild and hatchery trout Even though several studies demonstrated less risksensitive behaviour in hatchery reared trout (Einum and Fleming, 2001; Huntingford, 2004), in our experiment they did not differ from wild trout in average exploration tendency. Nevertheless, they differed in the structure of their exploration syndromes, and wild trout tended to be more consistent in their exploration behaviour across trials within the cryptic prey task. Few studies have compared syndrome structure between different treatments (Dingemanse et al., 2010). In sticklebacks, the occurrence of a behavioural syndrome affecting several behaviours including activity, aggressiveness and boldness has been shown to vary between populations (Bell, 2005; Bell and Stamps, 2004; Brydges et al., 2008; Dingemanse et al., 2007). These results showed that syndromes were more common in predator-sympatric populations, suggesting a role for predation in shaping associations between behaviours (Bell, 2005; Dingemanse et al., 2007). In additional experiments, a combination of behavioural plasticity and natural selection in response to trout predation generated a previously absent correlation between boldness and aggression (Bell and Sih, 2007). These results suggest that natural selection penalizes inconsistency and support the adaptive hypothesis for behavioural syndromes (Bell, 2005). 5. Conclusions Ideally, captive rearing programs should aim at producing individuals which are as similar as possible to wild individuals (Brockmark and Johnsson, 2010; Brown and Laland, 2001; Brännäs and Johnsson, 2008; Huntingford, 2004). In our experiments, however, hatchery trout differed in several aspects from wild trout. Behavioural differences between hatchery and wild trout can be caused by the rearing environment per se or by natural selection removing more extreme individuals in the wild-reared population. For example, wild trout could either be more experienced with predators than hatchery fish or just the survivors after predator selection against more active individuals (Einum and Fleming, 2001). Further experiments are needed to understand how behavioural specialisation in animals is influenced by experience and selection in different environments. Captive environments provide useful model systems to study such changes. The outcome of such studies can help modifying rearing environments to balance raising ethical requirements with the various

98

B. Adriaenssens, J.I. Johnsson / Applied Animal Behaviour Science 132 (2011) 90–99

purposes of captive rearing (Brännäs and Johnsson, 2008; Huntingford, 2004). Acknowledgements The authors would like to thank Ned Dochtermann for valuable discussions and Alison M. Bell and two anonymous referees for comments on an earlier version of the manuscript. JIJ was financed by the Swedish Research Council for Agricultural Research and Spatial Planning. The experiments were approved by the Ethical Committee for Animal Research in Gothenburg (License 132/2005) and comply with current laws in Sweden. References Anderson, D.R., Burnham, K.P., 2002. Avoiding pitfalls when using information-theoretic methods. J. Wildl. Manage. 66, 912–918. Araki, H., Berejikian, B.A., Ford, M.J., Blouin, M.S., 2008. Fitness of hatcheryreared salmonids in the wild. Evol. Appl. 1, 342–355. Bell, A.M., 2005. Behavioural differences between individuals and two populations of stickleback (Gasterosteus aculeatus). J. Evol. Biol. 18, 464–473. Bell, A.M., Sih, A., 2007. Exposure to predation generates personality in threespined sticklebacks (Gasterosteus aculeatus). Ecol. Lett. 10, 828–834. Bell, A.M., Stamps, J.A., 2004. Development of behavioural differences between individuals and populations of sticklebacks, Gasterosteus aculeatus. Anim. Behav. 68, 1339–1348. Benjamini, Y., Hochberg, Y., 1995. Controlling the false discovery rate: a practical and powerful approach to multiple testing. J. R. Stat. Soc. Ser. B Stat. Methodol. 57, 289–300. Benus, R.F., Dendaas, S., Koolhaas, J.M., Vanoortmerssen, G.A., 1990. Routine formation and flexibility in social and nonsocial behavior of aggressive and nonaggressive male mice. Behaviour 112, 176–193. Braithwaite, V.A., Armstrong, J.D., Madam, H.M., Huntingford, F.A., 1996. Can juvenile Atlantic salmon use multiple cue systems in spatial learning? Anim. Behav. 51, 1409. Brännäs, E., Johnsson, J.I., 2008. Behaviour and welfare in farmed fish. In: Magnhagen, C., Braithwaite, V.A., Forsgren, E., Kapoor, B.G. (Eds.), Fish Behaviour. Science Publishers, Enfield, USA, pp. 593–627. Brockmark, S., Johnsson, J.I., 2010. Reduced hatchery rearing density increases social dominance, postrelease growth, and survival in brown trout (Salmo trutta). Can. J. Fish. Aquat. Sci. 67, 288–295. Brown, C., Laland, K., 2001. Social learning and life skills training for hatchery reared fish. J. Fish Biol. 59, 471–493. Brown, C., Davidson, T., Laland, K., 2003. Environmental enrichment and prior experience of live prey improve foraging behaviour in hatcheryreared Atlantic salmon. J. Fish Biol. 63 (Suppl. A), 187–196. Brown, C., Jones, F., Braithwaite, V., 2005. In situ examination of boldness–shyness traits in the tropical poeciliid, Brachyraphis episcopi. Anim. Behav. 70, 1003–1009. Brydges, N.M., Colegrave, N., Heathcote, R.J.P., Braithwaite, V.A., 2008. Habitat stability and predation pressure affect temperament behaviours in populations of three-spined sticklebacks. Anim. Behav. 77, 229–235. Chittka, L., Dyer, A.G., Bock, F., Dornhaus, A., 2003. Psychophysics: bees trade off foraging speed for accuracy. Nature 424, 388–1388. Chittka, L., Skorupski, P., Raine, N.E., 2009. Speed-accuracy tradeoffs in animal decision making. Trends Ecol. Evol. 24, 400–407. Coleman, K., Wilson, D.S., 1998. Shyness and boldness in pumpkinseed sunfish: individual differences are context-specific. Anim. Behav. 56, 927–936. De Crespin De Billy, V., Usseglio-Polatera, P., 2002. Traits of brown trout prey in relation to habitat characteristics and benthic invertebrate communities. J. Fish Biol. 60, 687–714. Dingemanse, N.J., Wright, J., Kazem, A.J.N., Thomas, D.K., Hickling, R., Dawnay, N., 2007. Behavioural syndromes differ predictably between 12 populations of three-spined stickleback. J. Anim. Ecol. 76, 1–11. Dingemanse, N.J., Kazem, A.J.N., Réale, D., Wright, J., 2009. Behavioural reaction norms: animal personality meets individual plasticity. Trends Ecol. Evol. 25, 81–89. Dingemanse, N.J., Dochtermann, N., Wright, J., 2010. A method for exploring the structure of behavioural syndromes to allow formal comparison within and between data sets. Anim. Behav. 79, 439–450.

Dochtermann, N.A., Jenkins, S.H., 2007. Behavioural syndromes in Merriam’s kangaroo rats (Dipodomys merriami): a test of competing hypotheses. Proc. R. Soc. Lond. B Biol. Sci. 274, 2343–2349. Dukas, R., 2002. Behavioural and ecological consequences of limited attention. Philos. Trans. R. Soc. Lond. B Biol. Sci. 357, 1539–1547. Einum, S., Fleming, I.A., 2001. Implications of stocking: ecological interactions between wild and released salmonids. Nord. J. Freshw. Res. 75, 56–70. Feltmate, B.W., Williams, D.D., 1989. A test of crypsis and predator avoidance in the stonefly Paragnetina media (Plecoptera, Perlidae). Anim. Behav. 37, 992–999. Gosling, S.D., 2001. From mice to men: what can we learn about personality from animal research? Psychol. Bull. 127, 45–86. Gotceitas, V., Godin, J.G.J., 1992. Effects of location of food delivery and social status on foraging site selection by juvenile Atlantic salmon. Environ. Biol. Fishes 35, 291–300. Griffin, A.S., Blumstein, D.T., Evans, C., 2000. Training captive-bred or translocated animals to avoid predators. Conserv. Biol. 14, 1317–1326. Hargeby, A., Johansson, J., Ahnesjo, J., 2004. Habitat-specific pigmentation in a freshwater isopod: adaptive evolution over a small spatiotemporal scale. Evolution 58, 81–94. Höjesjö, J., Johnsson, J., Bohlin, T., 2004. Habitat complexity reduces the growth of aggressive and dominant brown trout (Salmo trutta) relative to subordinates. Behav. Ecol. Sociobiol. 56, 286–289. Huntingford, F.A., 1976. The relationship between anti-predator behaviour and aggression among conspecifics in the three-spined stickleback, Gasterosteus aculeatus. Anim. Behav. 24, 245–260. Huntingford, F.A., 2004. Implications of domestication and rearing conditions for the behaviour of cultivated fishes. J. Fish Biol. 65, 122–142. Huntingford, F., Adams, C., 2005. Behavioural syndromes in farmed fish: implications for production and welfare. Behaviour 142, 1207–1221. Hutchinson, J.M.C., 2005. Is more choice always desirable? Evidence and arguments from leks, food selection, and environmental enrichment. Biol. Rev. 80, 73–92. Johnsson, J.I., 1993. Big and brave – size selection affects foraging under risk of predation in juvenile rainbow trout, Oncorhynchus mykiss. Anim. Behav. 45, 1219–1225. Johnsson, J.I., Kjällman-Eriksson, K., 2008. Cryptic prey colouration increases search time in brown trout (Salmo trutta): effects of learning and body size. Behav. Ecol. Sociobiol. 62, 1613–1620. Johnsson, J.I., Sundström, L.F., 2007. Social transfer of predation risk information reduces food locating ability in european minnows (Phoxinus phoxinus). Ethology 113, 166–173. Jones, K.A., Krebs, J.R., Whittingham, M.J., 2006. Interaction between seed crypsis and habitat structure influence patch choice in a granivorous bird, the chaffinch Fringilla coelebs. J. Avian Biol. 37, 413–418. Kaiser, H.F., 1961. A note on Guttman lower bound for the number of common factors. Br. J. Math. Stat. Psychol. 14, 1–2. Lima, S.L., Bednekoff, P.A., 1999. Temporal variation in danger drives antipredator behavior: the predation risk allocation hypothesis. Am. Nat. 153, 649–659. Mezzera, M., Largiader, C.R., 2001. Evidence for selective angling of introduced trout and their hybrids in a stocked brown trout population. J. Fish Biol. 59, 287–301. Milinski, M., 1984. A predator’s cost of overcoming the confusion-effect of swarming prey. Anim. Behav. 32, 1157–1162. Nelson, T.C., Rosenau, M.L., Johnston, N.T., 2005. Behavior and survival of wild and hatchery-origin winter steelhead spawners caught and released in a recreational fishery. North Am. J. Fisheries Manag. 25, 931–943. Olla, B.L., Davis, M.W., Ryer, C.H., 1998. Understanding how the hatchery environment represses or promotes the development of behavioral survival skills. Bull. Mar. Sci. 62, 531–550. Orpwood, J.E., Griffiths, S.W., Armstrong, J.D., 2003. Effects of body size on sympatric shelter use in over wintering juvenile salmonids. J. Fish Biol. 63, 166–173. Pennell, W., Barton, B.A., 1996. Principles of Salmonid Culture. Elsevier, Amsterdam. Réale, D., Reader, S.M., Sol, D., McDougall, P.T., Dingemanse, N.J., 2007. Integrating animal temperament within ecology and evolution. Biol. Rev. 82, 291–318. Richards, S.A., 2005. Testing ecological theory using the informationtheoretic approach: examples and cautionary results. Ecology 86, 2805–2814. Sih, A., Bell, A., 2008. Insights from behavioral syndromes for behavioral ecology. In: Advances in the Study of Behavior. Sih, A., Bell, A.M., Johnson, J.C., Ziemba, R.E., 2004. Behavioral syndromes: an integrative overview. Q. Rev. Biol. 79, 241–277.

B. Adriaenssens, J.I. Johnsson / Applied Animal Behaviour Science 132 (2011) 90–99 Sinn, D.L., Gosling, S.D., Moltschaniwskyj, N.A., 2008. Development of shy/bold behaviour in squid: context-specific phenotypes associated with developmental plasticity. Anim. Behav. 75, 433–442. Sneddon, L.U., 2003. The bold and the shy: individual differences in rainbow trout. J. Fish Biol. 62, 971–975. Sundström, L.F., Johnsson, J.I., 2001. Experience and social environment influence the ability of young brown trout to forage on live novel prey. Anim. Behav. 61, 249–255. Sundström, L.F., Lohmus, M., Devlin, R.H., Johnsson, J.I., Biagi, C.A., Bohlin, T., 2004a. Feeding on profitable and unprofitable prey: comparing behaviour of growth-enhanced transgenic and normal coho salmon (Oncorhynchus kisutch). Ethology 110, 381–396. Sundström, L.F., Petersson, E., Höjesjö, J., Johnsson, J.I., Jarvi, T., 2004b. Hatchery selection promotes boldness in newly hatched brown trout (Salmo trutta): implications for dominance. Behav. Ecol. 15, 192–198. van Oers, K., Klunder, M., Drent, P.J., 2005. Context dependence of personalities: risk-taking behavior in a social and a nonsocial situation. Behav. Ecol. 16, 716–723.

99

Verbeek, M.E.M., Drent, P.J., Wiepkema, P.R., 1994. Consistent individual differences in early exploratory behaviour of male great tits. Anim. Behav. 48, 1113. Wilson, A.D.M., Stevens, E.D., 2005. Consistency in context-specific measures of shyness and boldness in rainbow trout, Oncorhynchus mykiss. Ethology 111, 849–862. Wilson, D.S., Coleman, K., Clark, A.B., Biederman, L., 1993. Shy bold continuum in pumpkinseed sunfish (Lepomis gibbosus) – an ecological study of a psychological trait. J. Comp. Psychol. 107, 250–260. Wilson, D.S., Clark, A.B., Coleman, K., Dearstyne, T., 1994. Shyness and boldness in humans and other animals. Trends Ecol. Evol. 9, 442–446. Winkler, H., Kothbauer-Hellmann, R., 2001. The role of search area in the detection of cryptic prey by crested tits and coal tits. Behaviour 138, 873–883. Wolf, M., van Doorn, G.S., Weissing, F.J., 2008. Evolutionary emergence of responsive and unresponsive personalities. Proc. Natl. Acad. Sci. U.S.A. 105, 15825–15830. Zar, J.H., 1999. Biostatistical Analysis. Prentice-Hall, Upper Saddle River, New Jersey.