Title: 1

Behavioral syndromes and the evolution of correlated behavior in zebrafish 2

3

Authors: 4

Jason A. Moretz1, Emília P. Martins1* and Barrie D. Robison2567

1Department of Biology and Center for the Integrative Study of Animal Behavior, 8

Indiana University, Bloomington, IN 47405, USA9

10

2Department of Biological Sciences and Center for Reproductive Biology, University of 11

Idaho, Moscow, Idaho 443051, USA12

13

*Corresponding author: Emília P. Martins; Telephone (812)856-5840; e-mail: 14

emartins@indiana.edu15

16

Running Header:17

Behavior syndromes in zebrafish18

19

Acknowledgements 20

We thank Erin Churchill and Candice Clark for animal husbandry help. Padrick 21

Anderson and Jennifer Wright assisted in scoring videotaped trials. Erin Kelso, 22

Cuauhcihuatl Vital, Saúl Nava, Mayté Ruiz and Winnie Ho provided numerous 23

suggestions on earlier versions of this manuscript. We are grateful also to Anne Houde 24

and two anonymous reviewers for helpful comments. This research was supported in part 25

by a National Science Foundation award to EPM (IOB-0543491).and by a National 26

Science Foundation and Idaho University EPSCoR award (EPS-044789) to B. Robison. 27

All experiments comply with current laws of the United States and with the Animal Care 28

Guidelines of Indiana University (BIACUC Protocol Approval #: 05-096).29

30

ABSTRACT:31

Studies of “behavioral syndromes” in different populations and species of animals can be 32

used to shed light on the underlying mechanisms of evolution. For example, some 33

personality syndromes suggest the existence of an underlying hormonal link, whereas 34

other relationships between boldness and aggression appear to be the result of similar 35

selective pressures. Here, we used one wild-derived and two laboratory strains of 36

zebrafish (Danio rerio) to examine relationships among five behavioral measures: 37

Shoaling, Activity Level, Predator Approaches, Latency to Feed after a disturbance and 38

Biting to a mirror stimulus. We found evidence of an Activity Syndrome, as if 39

underlying metabolic costs influence variation in multiple forms of behavior. Evidence 40

for a relationship between Boldness and Aggression was also apparent, but depended 41

both on strain and which specific behavior patterns were identified as measures of 42

“boldness”. Although some comparisons of laboratory and wild-derived strains were 43

consistent with a Domestication Syndrome, others were not. Most observed relationships 44

were relatively weak and occasionally inconsistent, arguing against strong underlying 45

genetic linkages or pleiotropic effects relating any of the behavioral measures. Instead, it 46

may be more important to study the details of selective context or the long-term impact of 47

linkages between some, but not all, of a large set of genes influencing complex 48

behavioral traits. We found profound differences among strains in most behavior 49

patterns, but few sex differences. One strain (TM1) was consistently different from the 50

others being more social, more likely to approach predators, and taking less time to 51

recover from disturbance than were the other two strains.52

53

Keywords: 54

behavioral syndromes, aggression, boldness, domestication, zebrafish, phenotypic 55

correlation, genetic correlation, activity, evolution, metabolic syndrome56

57

INTRODUCTION57

Many behavior patterns occur in tightly-linked suites or “syndromes” (see reviews by Sih58

et al., 2004a, b). In some cases, studying complex behavior such as multimodal 59

communicative displays as examples of behavioral syndromes ensures that we consider 60

also emergent properties that appear only when multiple sensory elements are combined61

(e.g., Hebets 2005). In other cases, an observed link between behavioral phenotypes 62

encourages us to search for a single underlying genetic or physiological mechanism. For 63

example, studies of a boldness and aggression syndrome have encouraged further 64

exploration of underlying hormonal explanations (e.g., Veenema et al., 2003). Syndrome 65

studies can also shed light on phenotypic evolution. Both functional and mechanistic 66

relationships between behavioral traits can constrain the directions in which evolution 67

can occur and be crucial to the “evolvability” of complex phenotypes (e.g., Beldade et al., 68

2002; Hansen et al., 2003). Here, we estimate the magnitude of behavioral syndromes in 69

Zebrafish (Danio rerio), a popular model organism in genetic, developmental and 70

neurobiological research, thereby setting the stage for future research into the genetic 71

basis of behavioral syndromes.72

We focus on three syndromes that have been commonly-reported in the 73

behavioral literature: 1) boldness and aggression, 2) general metabolic response, and 3) 74

domestication. An aggression-boldness syndrome has been found in numerous animals 75

(see review by Koolhaas et al. 1999). In fish, it was classically documented in stickleback 76

(Huntingford, 1976), and is supported by more recent studies of cichlids (Brick and 77

Jakobsson, 2002) and trout (Schjolden et al. 2005). Bell (2005) suggests that in 78

stickleback, the syndrome exists only in populations with high predation levels, and is 79

thus likely due to selection acting on multiple aspects of the phenotype simultaneously 80

rather than a common underlying mechanism. Brown et al. (2005) showed that boldness 81

in a tropical poeciliid is related not only to predation but to growth rate and body size, 82

suggesting the importance also of metabolic factors. Similar complexes of behavioral 83

traits have been described in a wide variety of birds and mammals as a distinction 84

between “proactive” and “reactive” personalities, and has been linked to changes in the 85

hypothalamic-pituitary-adrenal (HPA) system which influences many behavior patterns 86

(e.g., Veenema et al., 2003). 87

The second syndrome we consider is a relationship between activity and other 88

forms of behavior which is commonly attributed to a general metabolic response. 89

Measures of activity may be related to measures of other behavior because of time budget 90

conflicts (e.g. time spent feeding vs. time spent defending a nesting site) or because 91

activity level may approximate metabolism (reviewed in Sih et al. 2004b). For example, 92

Budaev (1997a, b; Budaev and Zhuikov, 1998) found relationships between activity 93

level, exploratory behavior and shoaling in fishes. Activity level is also associated with 94

many other forms of behavior in flies (e.g., Meffert et al. 2002; Higgins et al. 2005). In 95

other cases, individuals that are more active in the absence of predators are often also 96

more active when predators are present (e.g. Sih et al., 2003; Brodin and Johansson, 97

2004).98

Third, we consider domestication, a complex of behavioral, morphological and 99

physiological changes due both to intentional selection for desirable traits as well as 100

inadvertent selection and a reduction of natural selection (e.g., Huntingford, 2004). For 101

example, in Dmitry Belyaev’s classic Farm-Fox experiment (summarized by Trut 1999), 102

silver foxes selected only for tameness evolved many additional behavioral 103

characteristics of the domestic dog such as barking, whining and licking of their handlers. 104

This finding suggests that there may be a common underlying mechanism such as a 105

pleiotropic gene or hormone with multiple effects that links these very different behavior 106

patterns. Decreases in fearfulness are common with domestication (e.g., jungle fowl: 107

Schutz et al., 2001; trout: Alvarez and Nicieza, 2003), as are changes in aggression, 108

although the variation in the direction of observed aggressive shifts (increased: Ruzzante, 109

1994; Einum and Fleming, 1997; Metcalfe et al., 2003; decreased: Fleming et al., 1996;110

Berejikian et al., 1997; Hedenskog et al., 2002; no difference: Johnsson et al. 1996;111

Yamamoto & Reinhardt 2003) suggests that domestication may also involve selection112

acting on multiple phenotypes simultaneously, rather than an underlying shared 113

mechanism. 114

Comparison of behavioral syndromes in different populations and across different 115

taxonomic levels offers a way to distinguish between phenotypic associations resulting 116

from underlying mechanisms and correlated selection. For example, Bell (2005) found 117

that an aggression-boldness syndrome observed in one population of stickleback fish was 118

unlikely to be the result of underlying genetic correlations because it was not also 119

observed in a second population of the same species. Baer and Lynch (2003) concluded 120

that relationships between body size and life history traits observed both within and 121

between populations offered evidence of strong pleiotropic constraints on body size 122

evolution. This comparative approach is limited given growing evidence that genetic 123

correlations also evolve and hence often differ among populations (see review by 124

Steppan, 2002). But similar phenotypic correlations observed both within and across 125

populations suggest an underlying genetic association or pleiotropic physiological 126

mechanism, and differences across populations argue for differential selective regimes 127

(e.g., Badyaev and Hill, 2000) or different genetic structures (e.g., Service, 2000), and 128

suggest that syndromes do not necessarily constrain the selective responses of behavior 129

because their underlying genetic or physiological basis can be uncoupled.130

Zebrafish are an ideal species for studying the genetic basis of behavioral 131

syndromes because of their extensive use as a model organism by developmental 132

biologists and the comprehensive toolkit of genetic techniques that are consequently 133

available, including a multitude of distinct, artificially-selected strains and a complete 134

genome sequence. While we know little about zebrafish behavior and natural history,135

recent studies have found that although fish from natural populations form shoals of two 136

to ten individuals (Pritchard et al., 2001), individuals are aggressive to one another and 137

able to monopolize resources (Hamilton and Dill, 2002; Spence and Smith, 2005). 138

Aggression is present in males as well as females (Pritchard, 2001) and females prefer to 139

associate with males rather than with other females (Delaney et al., 2002). Recent studies 140

comparing the behavior of laboratory and recently-derived wild-caught strains found that 141

laboratory strains exhibit characteristics commonly associated with domestication (e.g., 142

higher growth rates, decreased startle response), and that strain differences in “boldness” 143

have a genetic basis (Robison and Rowland, 2005; Wright et al., 2003, 2006). In contrast, 144

zebrafish aggression and shoaling behavior is more malleable, sometimes depending on 145

the social history of the tested individuals. Engeszer et al. (2004) showed that zebrafish’s146

preference for horizontal stripes depended on the social context during early 147

development, whereas Moretz et al. (2006) showed that tendency to shoal can be altered 148

even in adults after exposure to particular social contexts. 149

In this study, we measure the magnitude of phenotypic associations between five 150

behavior patterns: aggression, shoaling behavior, latency to feed after disturbance, 151

activity level, and predator response. We focus on associations indicative of the three 152

behavioral syndromes outlined above, relating measures of aggression and boldness, of 153

activity with other behavior patterns, and of behavior patterns that typically change with 154

domestication (including also recovery from disturbance and predator approaches). We 155

compare relationships within and across three strains of zebrafish to identify those which 156

are not only large but consistent. 157

158

METHODS159

Study species: Zebrafish are small cyprinids native to rivers of Pakistan, India and 160

Bangladesh (Eaton and Farley, 1974; Spence et al. 2006). Under laboratory conditions, 161

zebrafish are reproductively active throughout the year and are easily maintained 162

(Westerfield, 1995). We tested the behavior of individuals from three strains of 163

zebrafish, all of which were raised from eggs in similar environmental conditions. The 164

TM1 strain is derived from zebrafish acquired from a pet store in 1986 and is now 30 165

generations removed from that point (Robison and Rowland, 2005). The TM1 strain 166

(mean Standard Length [SL] = 31 mm, SE = 0.3) has a green fluorescent protein (GFP) 167

transgene which is detectable as a visible green emission when exposed to ultraviolet 168

light (Gibbs and Schmale, 2000). Individuals of the Nadia strain (mean SL = 26 mm, SE 169

= 0.3) tested in our experiment were only five generations removed from wild-caught fish 170

(see www.zfin.org for details), and exhibit several behavioral and morphological 171

differences from TM1 fish that suggest that Nadia have not yet evolved in response to the 172

laboratory environment (Robison and Rowland, 2005). The third strain, SH or Scientific 173

Hatcheries, (mean SL = 29 mm, SE = 0.4) is more similar to TM1 in that it was 174

developed in the early 1990s, and is available commercially from Scientific Hatcheries in 175

Huntington Beach, California. However, unlike the TM1 strain which is maintained by 176

breeding small numbers of individuals, SH fish are reared and bred in large numbers, 177

approximating conditions one would expect to find in commercial fish hatcheries (e.g. 178

trout hatcheries). The three strains have thus evolved under very different selective 179

regimes.180

We used mass breedings of adults from each strain to generate subjects for our 181

experiment, raising animals under standard conditions in pure-strain groups until they 182

were 5 months of age (adults). Hatchlings were kept in small groups in beakers and fed 183

Paramecium caudatum for the first 15 days. Juvenile zebrafish were then moved to 21 L 184

aquaria in groups of 10 where they were fed a diet of brine shrimp and flake food twice 185

daily. All aquaria were visually isolated from one another and were housed in a constant 186

temperature aquatic laboratory (28.5 C) on a 14 hour light / 10 hour dark cycle, as typical 187

to promote breeding and rapid development. The laboratory is serviced by a U. S. Filter 188

organic carbon filter water treatment system that delivers water to each tank separately 189

and continually.190

191

Behavioral Measures: We measured the behavior of 84 fish, including 15 males and 15 192

females from the TM1 strain, 22 males and 8 females from the Nadia strain and 11 males 193

and 13 females from the Scientific Hatcheries strain. As in Moretz et al. (2006), we used 194

a large testing arena (115 L aquarium) to measure three types of behavior: Leaving Shoal, 195

Activity level and Predator Approaches. Leaving Shoal was scored as the number of 196

times (in 5 min) that the subject fish moved more than 14 cm horizontally away from a 197

stimulus shoal (4 fish of the same strain as the subject), visible to the subject through a 198

solid Plexiglas barrier. Leaving Shoal might be interpreted as a measure of boldness with 199

bold individuals moving away from the shoal more often than wary individuals. Because 200

of the lack of acclimation period immediately before the trial began, Leaving Shoal might 201

also be interpreted as an inverse measure of the fish’s tendency to form groups when 202

recovering from a stressful situation. The stimulus shoal was then obscured from view by 203

an opaque Plexiglas barrier, and activity level was estimated as the number of lines 204

crossed on a horizontal grid during 5 min, including both forwards and backwards 205

motion. Finally, a second, opaque barrier was lifted, revealing a stationary epoxy model 206

(SL= 83.5mm, DV= 3.2mm), painted white with black bars to mimic the appearance of a 207

cichlid (Etroplus canarensis) found in the same geographic areas as zebrafish. We scored 208

Predator Approaches as the number of times (in 5 min) that the subject moved within 14 209

cm horizontal distance of the predator. Subjects with high Predator Approach scores 210

might be considered to be “bold”. Note that all three behavioral metrics are related to 211

each other in that they were measured in the same testing arena and all three involved 212

measurements of locomotion.213

214

Subjects were then moved to a small holding tank (21 L) to measure Latency to 215

Feed and Biting. Live brine shrimp were placed in the tank at the same time as the 216

subject, and we recorded Latency to Feed in seconds. Because the fish were offered the 217

opportunity to eat immediately after a major disturbance (being netted and placed in a 218

new tank), we interpret Latency to Feed as a measure of stress recovery or perhaps 219

boldness. Because Latency to Feed may also reflect individual variation in hunger, all 220

fish were tested 5-6 h after the most recent feeding. After, test subjects remained resting, 221

visually isolated from all other fish, until the next morning (at least 18h later) when we 222

scored Biting by placing a mirror at one end of an individual’s tank and recording the 223

number of bites directed at the image over a 5-min period. As explained further in Moretz 224

et al. (2006), biting to a mirror has been mostly interpreted as a measure of aggression 225

(Gerlai 2003, Marks et al. 2005), but may also indicate a more general measure of social 226

motivation, or the intent to interact with a social partner. Tests in the large arena were 227

videotaped for later scoring, whereas tests in the small holding tank were scored in real 228

time by a single observer (JAM). All tests took place early in the morning, within one 229

hour of lights being turned on and before daily feeding. 230

231Statistical analyses: Relationships between measures of boldness (Leaving Shoal, 232

Predator Approaches, and possibly Latency to Feed after a disturbance) and aggression 233

(Biting) would provide evidence for an Aggression-Boldness syndrome. Relationships 234

between our single measure of Activity level and each of the four other behavior patterns235

would provide evidence of an Activity Syndrome. Domestication is expected to influence 236

boldness, aggression, and also stress recovery (as reflected in our measure of Latency to 237

Feed after a disturbance). We thus began our analyses by estimating the magnitude of 238

pair-wise relationships between behavioral measures, using Pearson product-moment 239

correlations as our measure of effect size. To avoid multiple comparisons, and because of 240

the potential importance of sex and strain effects, we did not conduct hypothesis tests on 241

the correlation coefficients themselves. Instead, we combined data from all three strains 242

and used ANOVAs to estimate the same pair-wise relationships, but now including also 243

sex and strain as predictors. Note that F-values based on Type III sums of squares in 244

these analyses are identical regardless of which behavior pattern is set as predictor (X) 245

and which is identified as the response (Y). Because this approach still resulted in ten 246

tests of related data, we further applied a Bonferroni correction, considering only results 247

with p < 0.005 as being statistically significant, and limiting our attention to the single F-248

value that describes the trait relationship. We then used a MANOVA to test the overall 249

effects of sex and strain and a sex by strain interaction effect on behavior. We square-root 250

transformed Leaving Shoal and Predator Approaches and natural-log transformed 251

Latency to Feed to obtain normal and homoscedastic residuals in the ANOVAs and 252

MANOVA. All analyses were performed using SAS (2002).253

254

RESULTS255

Evidence of Syndromes256

Individual zebrafish that were more active also approached the predator dummy more 257

often (Fig. 1, Table 1), leading to a significant relationship between two behavioral 258

measures (F =9.7; df = 1, 73; P = 0.003, Table 2), suggestive of an activity syndrome. 259

This was the only pair-wise relationship that was both large and consistent across all 260

three strains (Table 2). A possible activity syndrome was also supported by an apparent261

relationship between activity levels and the tendency to leave the vicinity of the shoal (in 262

TM1 [r = 0.52] and SH [r = 0.33], but not Nadia [r = 0.02]) and between Activity levels 263

and Latency to Feed after a disturbance (only in SH fish [r = -0.49], Table 1). Because 264

these other relationships were larger in some strains than in others (Table 1), they were 265

not statistically significant when tested across all three strains (Table 2). 266

There was also a suggestion of a boldness/aggression syndrome in the267

relationships between biting and some measures of boldness (Leaving Shoal and Latency 268

to feed after disturbance), but not others (Predator Approaches, Table 1). But once again, 269

the details were inconsistent across strains and hence were not evident in the across-strain 270

tests. Both Nadia and TM1 fish which bit more often at a mirror stimulus were also more 271

likely to leave the immediate vicinity of a shoal (r = 0.32 for Nadia, 4 = 0.24 for TM1), 272

but SH fish showed a weaker pattern in the opposite direction (r = -0.11; Table 1), such 273

that the overall relationship was not statistically significant when subjected to the 274

conservative Bonferroni criterion (Table 2). Similarly, Biting was related to Latency to 275

Feed, but this pattern differed critically across strains (Table 1), and was hence not 276

significant overall (Table 2). TM1 fish that bit frequently took less time to feed after a 277

disturbance, SH that bit more often took more time to feed after a disturbance, and there 278

was no relationship between these two behavioral measures in the Nadia (Table 1). We 279

found no evidence of a relationship between Biting to the mirror stimulus and Predator 280

Approaches (-0.1 < r < 0.2; Table 1).281

Our results suggest a domestication syndrome in that individual SH which fed 282

quickly after a disturbance (a measure of stress recovery) were also more active (r = -283

0.49), less likely to bite at a mirror stimulus (r = 0.43), and more likely to approach a 284

predator (r = -0.42, Table 1). But these patterns were not strong in the other two strains or 285

showed conflicting patterns. For example, TM1 fish which fed rapidly after a disturbance 286

actually bit more often to the mirror stimulus (r = -0.28), and Nadia showed no strong 287

relationships between latency to feed after a disturbance and other measures. Thus, again, 288

these domestication relationships were not confirmed by the ANOVA tests because of 289

inconsistencies in the pattern across strains (Table 2). 290

291

Sex and Strain Differences292

Sex differences in behavior were small (Table 3). Male zebrafish were more 293

active, took longer to feed after a disturbance, approached the predator dummy and left 294

the vicinity of a shoal more often than did females. But when behavior patterns were 295

tested simultaneously in a MANOVA, neither sex differences (Wilks’ λ = 0.86; F = 1.87; 296

df = 5, 58; P = 0.1137) nor interactions between sex and strain (Wilks’ λ = 0.80; F = 1.39; 297

df = 10, 116; P = 0.1939) were statistically significant. 298

In contrast, we found robust and significant strain differences in behavior (Wilks’ 299

λ = 0.55; F = 4.04; df = 10, 116; P = 0.0001). Although these strain differences were 300

evident in all the behavior patterns (Fig. 2, Table 3), they were most pronounced in Biting 301

to a mirror (F = 10.0; df = 2, 62; P = 0.002) and Latency to Feed after a disturbance (F = 302

3.4; df = 2, 62; P = 0.04). TM1 individuals bit the mirror stimulus more often and fed 303

more quickly after a disturbance than did individuals of the other two strains (Table 3). 304

As might be expected for a non-domesticated strain, individual Nadia were less 305

active, less aggressive (bit the mirror less often), and less bold (feeding less soon after 306

disturbance and approaching the predator less) than were individuals of the TM1 307

laboratory strain (Fig. 2). Nadia also left the shoal slightly more often than did TM1 fish. 308

The domestication syndrome was not confirmed, however, by results from the SH 309

laboratory strain. Individual SH fish were often more similar to Nadia (wild-derived) than 310

to individuals of the TM1 laboratory strain (Fig. 2), and were less aggressive and more 311

likely to leave the vicinity of the shoal than were either of the two other strains. The two 312

laboratory strains (SH and TM1) were more similar to each other than to the more 313

recently domesticated Nadia strain only in Activity level (Table 2).314

315

DISCUSSION316

Our study finds relationships between behavior and activity level in three strains of 317

zebrafish, and weaker evidence of a boldness/aggression syndrome and a domestication 318

syndrome in some, but not other, strains. In the latter two cases, the direction of the 319

relationship in the syndrome also depended on the strain. In fact, strain differences in 320

nearly every behavior pattern were pronounced, whereas sex differences were generally 321

small.322

Specifically, we found relationships between activity level and 1) tendency to 323

approach a predator in all three strains, 2) tendency to move away from a shoal (TM1 and 324

SH, but not Nadia) and 3) latency to feed after a disturbance (SH only). These results are 325

consistent with the growing body of evidence that many forms of behavior are 326

constrained by underlying metabolic costs (reviewed in Sih et al., 2004b). As reviewed 327

by Meffert et al. (2002; see also Higgins et al., 2005), many forms of behavior in flies are 328

also tightly associated with activity level. Similar associations have also been found 329

between activity level, exploratory behavior and shoaling (Budaev, 1997a, b; Budaev and 330

Zhuikov, 1998). In part, these relationships in our own and other studies may be 331

explained by similarities in the underlying scoring procedure – we might expect predator 332

approaches or exploratory behavior to be associated with activity level when these 333

behavior patterns are scored in terms of locomotion about an arena. In our own study, the 334

lack of some relationships which involved measures of locomotion (e.g., between 335

Leaving Shoal and Predator Approaches) suggests that the observed Activity syndrome 336

goes beyond a relationship between scoring metrics.337

We found relationships between boldness and aggression, but the relationships 338

depended on both strain and the details of the behavior patterns being considered. Nadia 339

and TM1 that bit more frequently tended to be more willing to leave the vicinity of a 340

shoal, but the opposite pattern was seen in the SH strain. Similarly, the relationship seen341

in TM1 fish between biting and quick feeding after a disturbance was reversed in SH fish342

(SH that bit more often took longer to feed after a disturbance) and was not at all present 343

in Nadia fish. Bell (2005) suggests that the boldness and aggression syndrome is more 344

likely to be present in populations with high predation levels. Although the SH strain is 345

maintained in a predator-free environment, competition from conspecifics in high-346

density-rearing environments may have had a similar evolutionary effect to high 347

predation. Further studies comparing the selective impact of conspecific competition and 348

predation levels may be warranted. Similarly, strain differences in the direction of the 349

observed relationships suggest that although some fish strains exhibit “proactive” and 350

“reactive” personalities similar to those described for birds and mammals, further 351

research is needed to determine whether and how these personalities may be related to an 352

underlying HPA axis. 353

Although we found several major differences between two zebrafish strains (TM1 354

and Nadia) suggestive of a domestication syndrome (consistent with that proposed by 355

Robison & Rowland 2005), data for a third strain (SH) did not support this conclusion. 356

This is similar to previous results showing that fish domestication has had highly variable 357

effects on behavior. Domestication has lead to increases (e.g., Einum and Fleming, 1997; 358

Metcalfe et al., 2003), decreases (Fleming et al., 1996; Berejikian et al., 1997; Hedenskog 359

et al., 2002) and no change (Johnsson et al., 1996; Yamamoto and Reinhardt, 2003) in 360

fish aggression. As shown in our study, laboratory strains of zebrafish can vary 361

dramatically in their behavior. Wild-type zebrafish strains derived from different natural 362

populations can also vary markedly in behavior (Wright et al., 2003). Thus, the 363

wild/domesticated dichotomy may not be a sufficient description of the differences in 364

evolutionary history, genetic background and selective regime. 365

Although other behavioral syndrome studies have reported correlation values (r) 366

ranging from 0.4 to 0.8 (Huntingford, 1976; Hedrick, 2000; Sih et. al., 2003; Bell and 367

Stamps, 2004; Ward et. al., 2004; Bell, 2005), such strong correlations may have only a 368

short-term effect on behavioral evolution. Over time, strong phenotypic or genetic 369

correlations decrease differences among individuals in a population. Thus unless 370

correlational selection is acting on the two traits or the animals exist in environments 371

which fluctuate over time (Roff 1996), the originally strong phenotypic or genetic 372

correlations are likely to disappear as the two traits reach fixation. Alternatively, weak, 373

but persistent, genetic correlations may have a more profound evolutionary impact when 374

considered over long periods of time, especially for behavioral traits that are influenced 375

by tens or hundreds of genes (e.g., Anholt et al. 2003, Flint 2003). In these cases, strong 376

genetic linkages between subsets of these genes may produce only weak phenotypic 377

correlations which over long periods of time have profound effects on the direction of 378

evolutionary change. Comparative studies which infer selection from interspecific 379

variation (e.g., Hansen, 1997; Prum, 1997) are needed to determine when behavioral 380

syndromes measured in a single generation are sufficiently large to lead to long-term 381

behavioral evolution.382

383

Literature Cited383

Alvarez D, Nicieza AG. 2003. Predator avoidance behaviour in wild and hatchery-384reared brown trout: the role of experience and domestication. J Fish Biol 38563:1565-1577.386

387Anholt RRH, Dilda CL, Chang S, Fanara JJ, Kulkarni NH, Ganguly I, Rollman SM, 388

Kamdar KP, Mackay TFC. 2003. The genetic architecture of odor-guided 389behavior in Drosophila: epistasis and the transcriptome. Nature Gen 35:180–184.390

391Badyaev AV, Hill GE. 2000. The evolution of sexual dimorphism in the house finch. I. 392

Population divergence in morphological covariance structure. Evol 54:1784–3931794.394

395Baer CF, Lynch M. 2003. Correlated evolution of life-history with size at maturity in 396

Daphnia pulicaria: patterns within and between populations. Genet Res Comb 81: 397123-132.398

399Beldade P, Koops K, Brakefield PM. 2002. Modularity, individuality and evo-devo in 400

butterfly wings. Proc Natl Acad Sci USA 99:14262-14267.401402

Bell AM. 2005. Behavioural differences between individuals and two populations of 403stickleback (Gasterosteus aculeatus). J Evol Biol 18:464-473.404

405Bell AM, Stamps JA. 2004. Developmental differences between individuals and 406

populations of sticklebacks, Gasterosteous aculeatus. Anim Behav 68:1339-4071348.408

409Berejikian BA, Tezak EP, Schroder SL, Knudsen CM, Hard JJ. 1997. Reproductive 410

behavioral interactions between wild and captively reared coho salmon 411(Oncorhynchus kisutch). Ices J Mar Sci 54:1040-1050.412

413Brick O, Jakobsson S. 2002. Individual variation in risk taking: the effect of a predatory 414

threat on fighting behavior in Nannacara anomala. Behav Ecol 13:439-442.415416

Brodin T, Johansson F. 2004. Conflicting selection pressures on the growth/predation –417risk trade-off in the damselfly. Ecology 85:2927-2932.418

419Brown, C, Jones F, Braithwaite V. 2005. In situ examination of boldness-shyness traits in 420

the tropical poeciliid, Brachyraphis episcopi. Anim Behav 70: 1003-1009.421422

Budaev SV. 1997a. Personality in the guppy (Poecilia reticulata): a correlational study 423of exploratory behavior and social tendency. J Comp Psych 111:399-411.424

425Budaev SV. 1997b. Alternative styles in the European wrasse, Symphodus ocellatus: 426

boldness related schooling tendency. Env Biol Fishes 49:71-78.427

428

Budaev SV, Zhuikov AY. 1998. Avoidance learning and personality in the guppy 429(Poecilia reticulate). J Comp Psych 112:92-94.430

431

Delaney M, Follet C, Ryan N, Hanney N, Lusk-Yablick J, Gerlach G. 2002. Social 432interaction and distribution of female zebrafish (Danio rerio) in a large aquarium. 433Biol Bull 203:240-241.434

435Eaton RC, Farley RD. 1974. Spawning cycle and egg production of zebrafish, 436

Brachydanio rerio, in the laboratory. Copeia 1974:195-204.437438

Einum S, Fleming IA. 1997. Genetic divergence and interactions in the wild among 439native, farmed and hybrid Atlantic salmon. J Fish Biol 50:634-651.440

441Engeszer RE, Ryan MJ, Parichy DM. 2004. Learned social preference in zebrafish. 442

Current Biol 14:881-884. 443444

Erickson DL, Fenster CB, Stenoien HK, Price D. 2004. Quantitative trait locus analyses 445and the study of evolutionary process. Mol Ecol 13:2505–2522.446

447Fleming IA, Jonsson B, Gross MR, Lamberg A. 1996. An experimental study of the 448

reproductive behaviour and success of farmed and wild Atlantic salmon (Salmo 449salar). J Appl Ecol 33:893-905.450

451Flint J. 2003. Analysis of quantitative trait loci that influence animal behavior. J 452

Neurobiol 54:46-77.453454

Gerlai R. 2003. Zebrafish: An uncharted behavior genetic model. Behav Gen 33:461-455468.456

457Gibbs PDL, Schmale MC. 2000. GFP as a genetic marker scorable throughout the life 458

cycle of transgenic zebrafish. Marine Biotech 2:107-125.459460

Hamilton IM, Dill LM. 2002. Monopolization of food by zebrafish increases in risky 461habitats. Can J Zool 80:2164-2169.462

463Hansen TF. 1997. Stabilizing selection and the comparative analysis of adaptation. Evol464

51:1341-1351.465466

Hansen TF, Armbruster WS, Carlson ML, Pelabon C. 2003. Evolvability and genetic 467constraint in Dalechampia blossoms: Genetic correlations and conditional 468evolvability. J Exp Zool 296B:23-39.469

470Hebets E. 2005. Attention-altering signal interactions in the multimodal courtship display 471

of the wolf spider Schizocosa uetzi. Behavioral Ecology 16: 75-82.472

473Hedenskog M, Petersson E, Jarvi T. 2002. Agonistic behavior and growth in newly 474

emerged brown trout of sea-ranched and wild origin. Aggressive Behav 28:145-475153.476

477Hedrick AV. 2000. Crickets with extravagant mating songs compensate for predation 478

risk with extra caution. Proc Roy Soc Lond B 267:671-675.479480

Higgins LA, Jones KM, Wayne ML. 2005. Quantitative genetics of natural variation of 481behavior in Drosophila melanogaster: the possible role of social environment on 482creating persistent patterns of group activity. Evol 59:1529-1539.483

484Huntingford FA. 1976. The relationship between antipredator behavior and aggression 485

among conspecifics in the three-spined stickleback. Anim Behav 24:245-260.486487

Huntingford FA. 2004. Implications of domestication and rearing conditions for the 488behavior of cultivated fishes. J Fish Biol 65:122-142.489

490Johnsson JI, Petersson E, Jonsson E, Bjornsson BT, Jarvi T. 1996. Domestication and 491

growth hormone alter antipredator behaviour and growth patterns in juvenile 492brown trout, Salmo trutta. Can J Fish Aq Sci 53:1546-1554.493

494Koolhaas JM, Korte SM, De Boer SF, Van Der Vegt BJ, Van Reenen CG, Hopster H, De 495

Jong IC, Ruis MAW, Blokhuis HJ. 1999. Coping styles in animals: current status 496in behaviour and stress-physiology. Neurosci Biobehav Rev 23:925 935. 497

498Marks C, West TN, Bagatto B, Moore FBG 2005. Developmental environment alters 499

conditional aggression in zebrafish. Copeia 2005:901-908.500501

Meffert LM, Hicks SK, Reagan JL. 2002. Nonadditive genetic effects in animal 502behavior. Am Nat 160:S198-S213.503

504Metcalfe NB, Valdimarsson SK, Morgan IJ. 2003. The relative roles of domestication, 505

rearing environment, prior residence and body size in deciding territorial contests 506between hatchery and wild juvenile salmon. J Appl Ecol 40:535-544.507

508Moretz JA, Martins EP, Robison, BD. 2006. The effects of early and adult social 509

environment on zebrafish (Danio rerio) behavior. Env Biol Fishes, in press.510511

Pritchard VL. 2001. Behaviour and morphology of the zebrafish, Danio rerio. PhD 512dissertation. University of Leeds.513

514Pritchard VL, Lawrence J, Butlin RK, Krause J. 2001. Shoal choice in zebrafish, Danio 515

rerio: the influence of shoal size and activity. Anim Behav 62:1085-1088.516517

Prum RO. 1997. Phylogenetic tests of alternative intersexual selection mechanisms: 518Trait macroevolution in a polygynous clade (Aves: Pipridae). Am Nat 149:668-519692.520

521Riechert SE, Hedrick AV. 1993. A test for correlations among fitness-linked behavioural 522

traits in the spider Agelenopsis aperta (Araneae, Agelenidae). Anim Behav 52346:669-675.524

525Robison BD, Rowland W. 2005. Zebrafish (Danio rerio) as a model organism in 526

conservation biology: behavioral variation among wild and domesticated strains. 527Can J Fish Aqu Sci 62:2046-2054.528

529Roff, D. A. 1996. The evolution of genetic correlations, an analysis of patterns. Evolution 530

50: 1392-1403.531532

Ruzzante DE. 1994. Domestication effects on aggressive and schooling behavior in fish. 533Fish Aquacult 120:1-24.534

535SAS. 2002. SAS for Windows 9.1. NC: SAS Institute, Inc.536

537Schjolden J, Backstrom T, Pulman KG, Pottinger TG, Winberg S. 2005. Divergence in 538

behavioural responses to stress in two stains of rainbow trout (Oncorhynchus 539mykiss) with contrasting stress responsiveness.540

541Schutz KE, Forkman B, Jensen P. 2001. Domestication effects on foraging strategy, 542

social behaviour and different fear responses: a comparison between the red 543junglefowl (Gallus gallus) and a modern layer strain. Appl Anim Behav Sci 74:1-54414.545

546Schutz KE, Kerje S, Jacobsson L, Forkman B, Carlborg O, Andersson L, Jensen 547

P. 2004. Major growth QTLs in fowl are related to fearful behavior: possible 548genetic links between fear responses and production traits in a red junglefowl x 549white leghorn intercross. Behav Gen 34: 121-130.550

551Service PM. 2000. The genetic structure of female life history in D. melanogaster: 552

comparisons amoung populations. Genet Res 75:153–166.553554

Sih A, Katts LB, Mauer EF. 2003. Behavioural correlations across situations and the 555evolution of antipredator behavior in a sunfish-salamander system. Anim Behav55665:29-44.557

558Sih A, Bell A, Johnson JC. 2004a. Behavioral syndromes: an ecological and 559

evolutionary overview. TREE 19:372-378.560561

Sih A, Bell A, Johnson JC, Ziemba RE. 2004b. Behavioral Syndromes: An Integrative 562Overview. Quart Rev Biol 79:241-277.563

564Spence R, Smith C. 2005. Male territoriality mediates density and sex ratio effects on 565

oviposition in the zebrafish, Danio rerio. Anim Behav 69:1317–1323.566567

Spence R, Fatema MK, Reichard M, Huq KA, Wahab MA, Ahmed ZF, Smith C. 2006.568The distribution and habitat preferences of the zebrafish (Danio rerio) in 569Bangladesh. J Fish Biol 69:1435–1448.570

571Steppan SJ. 2002. Comparative quantitative genetics: evolution of the G matrix. TREE572

17:320–327.573574

Trut LN. 1999. Early canid domestication: The farm-fox experiment. Am Sci 87:160-575169.576

577Veenema AH, Meijer OC, de Kloet ER, Koolhaas JM. 2003. Genetic selection for coping 578

style predicts stressor susceptibility. J Neuroend 15: 256-267.579580

Ward AJW, Thomas P, Hart PJB. 2004. Correlates of boldness in three-spined581sticklebacks (Gasterosteus aculeatus). Behav Ecol Sociobiol 55:561-568.582

583Westerfield M. 1995. The Zebrafish Book. Institute of Neuroscience, University of 584

Oregon, USA.585586

Wright D, Rimmer LB, Pritchard VL, Krause J, Butlin RK. 2003. Inter and intra-587population variation in shoaling and boldness in the zebrafish (Danio rerio). 588Naturwissen 90:374-377.589

590Wright D, Nakamichi R, Krause J, Butlin RK. 2006. QTL analysis of behavioral and 591

morphological differentiation between wild and laboratory zebrafish. Behav Gen 59236:271-284.593

594Yamamoto T, Reinhardt UG. 2003. Dominance and predator avoidance in domesticated 595

and wild masu salmon Oncorhynchus masou. Fish Sci 69:88-94.596597

Figure 1. Relationship between activity level and predator approaches (0.28 < r < 0.45). 597There was a significant strain difference in both traits. Crosses = Nadia; Filled Circles = 598SH; Open triangles = TM1. 599

600

601

Figure 2. Behavioral means for each strain. Error bars indicate one standard error.602

603

604

Table 1. Effect size for behavioral syndromes as estimated using Pearson product-604

moment correlation coefficients (r) within each strain. Bold font highlights relationships 605

with |r| ≥ 0.25. Asterisks mark relationships that are statistically significant when tested 606

using ANOVAs (Table 2). 607

________________________________________________________________________608Biting Latency to Feed Predator Shoal609

________________________________________________________________________610Activity (number of line crosses in 5 min)611

Nadia 0.12 0.04 0.28* 0.02612

SH -0.15 -0.49 0.37* 0.33613

TM1 0.19 0.15 0.45* 0.52614

Bites to a mirror stimulus (number per 5 min) 615

Nadia – 0.04 0.19 0.32616

SH – 0.43 -0.10 -0.11617

TM1 – -0.28 0.18 0.24618

Latency to Feed after Disturbance (s) 619

Nadia – – -0.22 -0.04620

SH – – -0.42 -0.18621

TM1 – – -0.03 0.23622

Predator Approaches (number in 5 min) 623

Nadia – – – 0.31624

SH – – – -0.04625

TM1 – – – 0.15626

627

628

Table 2. F-values, degrees of freedom, and p-value (in parentheses) from ANOVAs 628

testing significance of behavioral relationships while including also sex, strain, and a sex 629

x strain interaction effect. *Relationship is statistically significant when tested using 630

Type III Sums of Squares to account first for sex and strain effects and employing a 631

Bonferroni correction for multiple tests (P < 0.005). 632

________________________________________________________________________633Biting Latency to Feed Predator Shoal634

________________________________________________________________________635Activity 0.6 1.5 9.7* 1.3636

(1, 72) (1, 66) (1, 73) (1, 71)637

0.461 0.228 0.003 0.254638

Bites 0.0 1.2 4.9639

(1, 70) (1, 75) (1, 75)640

0.956 0.267 0.030641

Latency to Feed after Disturbance 3.1 0.3642

(1, 69) (1, 69)643

0.084 0.609644

Predator Approaches 1.1645

(1, 74)646

0.292647

________________________________________________________________________648649

Table 3. Behavioral means (one standard error) for each sex and strain, as well as F, df 649

and p-values for MANOVA test of the overall significance of sex and strain effects. 650

________________________________________________________________________651

Nadia TM1 SH Overall652

________________________________________________________________________653

Activity (number of line crosses in 5 min)654

Females 63 (18.0) 84 (4.8) 92 (5.9) 82 (5.2)655

Males 74 (11.6) 102 (11.0) 111 (13.0) 90 (7.4)656

TOTAL *71 (9.6) 92 (5.8) 100 (9.6)657

Biting to mirror (in 5 min)658

Females 51 (10.9) 103 (11.7) 52 (10.9) 72 (7.8)659

Males 70 (9.3) 92 (14.6) 46 (9.0) 72 (7.0)660

TOTAL 65 (7.5) *97 (9.3) 50 (7.2)661

Latency to Feed after Disturbance (s)662

Females 21 (5.8) 14 (4.4) 50 (12.3) 27 (5.4)663

Males 68 (16.6) 26 (4.7) 31 (8.7) 46 (8.4)664

TOTAL 55 (12.9) *19 (3.4) 40 (7.6)665

Predator Approaches (number in 5 min)666

Females 8 (3.5) 11 (1.2) 10 (1.9) 10 (1.1)667

Males 10 (2.5) 17 (2.8) 10 (2.7) 12 (1.7)668

TOTAL 9 (2.0) *14 (1.6) 10 (1.6)669

Leaving Shoal (number in 5 min)670

Females 7 (2.7) 4 (1.0) 12 (3.5) 8 (1.6)671

Males 9 (1.5) 8 (2.6) 14 (2.0) 9 (1.1)672

TOTAL 9 (1.3) *6 (2.2) 13 (2.2)673

674

675

Combined Behavior Wilk’s λ F df p-value676

Sex 1.9 5, 58 0.114677

Strain 4.0 10, 116 < 0.001678

Sex x Strain 1.4 10, 116 0.194679

680

681682

Activity (line crosses in 5 min)

0 50 100 150 200

Pre

dato

r App

roac

hes

(in 5

min

)

0

10

20

30

40

50

Figure 1. Relationship between activity level and predator approaches (0.28 < r < 0.45). There was a significant strain difference in both traits. Crosses = Nadia; Filled circles = SH; Open triangles = TM1.

Leav

ing

Shoa

l

0

2

4

6

8

10

12

14

16

Act

ivity

Lev

el

0

20

40

60

80

100

120

Pre

dato

r App

roac

h

0

2

4

6

8

10

12

14

16

18

Late

ncy

to F

eed

0

20

40

60

80

Bite

s to

Mirr

or0

20

40

60

80

100

120

Nadia SH TM1

Nadia SH TM1

Nadia SH TM1Nadia SH TM1

Nadia SH TM1

Figure 2. Behavioral means for each strain. Error bars indicate one standard error.