Animal Behaviour 81 (2011) 1231e1237

Contents lists available at ScienceDirect

Animal Behaviour
journal homepage: www.elsevier.com/locate/anbehav

Personality traits predict hierarchy rank in male rainbowfish social groups

a r t i c l e i n f o
Personality traits are becoming increasingly important in explaining adaptive individual differences in
Article history: animal behaviour and probably represent a leading edge of the evolutionary process. Despite the new-found
Received 15 June 2010 interest in animal personality among behavioural ecologists, few studies have investigated the link between
Initial acceptance 23 August 2010 personality traits and fitness measures. We examined this link using male rainbowfish, Melanotaenia
Final acceptance 10 March 2011 duboulayi, as a model species and found that a range of personality traits (aggression, activity and boldness)
Available online 13 April 2011 covaried with a male’s position in a hierarchy, which is directly related to reproductive success in this and
MS. number: 10-00430R many other species. Dominant fish were more aggressive, active, bold and also significantly larger than
subordinate fish. Moreover, we found strong correlations between activity levels and boldness suggesting
Keywords: that selection may act on a suite of traits in concert (sensu behavioural syndromes). When taken together
behavioural syndrome
with previous research, our results suggest that the activityeboldness syndrome is likely to be domain
dominance
specific. We suggest that multiple trait correlations may be generated by high levels of competition (e.g.
hierarchy
Melanotaenia duboulayi
sexual selection) in addition to predation pressure as identified by previous studies.
personality Ó 2011 The Association for the Study of Animal Behaviour. Published by Elsevier Ltd. All rights reserved.
rainbowfish
sexual selection

It is now well recognized that individuals differ consistently in several taxa including fish, birds, mammals, lizards and invertebrates
their behaviour. One of the major axes describing differences in concerning a range of different behaviours (Riechert & Hedrick 1993;
individual behaviour is the boldeshy continuum (Brown et al. 2005; Budaev 1997; Wilson 1998; Dingemanse et al. 2003; Bell & Stamps
Réale et al. 2007; Ioannou et al. 2008). Boldness is considered a major 2004; Bell 2005).
personality trait and has been linked to a number of fitness measures Interest in behavioural syndromes is growing because of its
including condition factor, diet, exposure to predators and parasite ecological origins and potential evolutionary implications. The
load (Wilson et al. 1993; Brown & Braithwaite 2004; Brown et al. existence of behavioural syndromes implies that behaviour is con-
2007). Differences in the degree of risk taking between individuals strained and thus individuals may not be capable of displaying the
have been observed both within and between populations (Brown degree of phenotypic plasticity expected under an adaptationist
et al. 2007; Harris et al. 2010). But this divergence in individual paradigm. In fact, syndromes could seem maladaptive in certain
behaviour also occurs in a number of other behavioural axes including contexts and the correlation between behaviours may be indicative
parental care (Budaev et al. 1999), mating behaviour (Magellan & of underlying physiological constraints, pleiotropy or genetic
Magurran 2007), activity and aggressiveness (Huntingford 1976; linkage. In the psychology literature two alternative views regarding
reviewed in Budaev & Brown, in press). Furthermore, research into the development of individual behaviour have been suggested, the
animal personalities has identified correlated behavioural axes, first of which is fairly compatible with the behavioural syndromes
labelled syndromes. A syndrome is defined as a suite of correlated approach. The domain-general view suggests that personality traits
behaviours reflecting consistent individual differences in behaviour are generally constrained and suites of traits are often correlated. In
across multiple situations (Sih et al. 2004). A population or species can this case, populations of animals occupying different environments
exhibit a behavioural syndrome, but individuals display a ‘behavioural would show similar correlations between traits. Second, the
type’ (Sih et al. 2004). Examples of syndromes have been found in domain-specific view, which is adaptationist in nature, suggests
that individual traits are largely under independent selection; thus
traits can become uncoupled and thereby permit a larger degree of
* Correspondence: C. Brown, Department of Biological Sciences, Macquarie phenotypic plasticity. Thus, domain-specific traits tend to be
University, Sydney 2109, Australia. correlated only under certain contexts. In great tits, Parus major, for
E-mail address: [email protected] (C. Brown).
1
M. Colléter is now at the Université Européenne de Bretagne, UMR 985
example, the link between exploratory behaviour and survival
Agrocampus OUEST, INRA ‘Ecologie et Santé des Ecosystèmes’, Ecologie halieutique, differs between the sexes and also with contemporary environ-
Agrocampus OUEST, 65 rue de Saint Brieuc, CS 84215, 35042 Rennes cedex, France. mental conditions (Dingemanse et al. 2004).

0003-3472/$38.00 Ó 2011 The Association for the Study of Animal Behaviour. Published by Elsevier Ltd. All rights reserved.
doi:10.1016/j.anbehav.2011.03.011
1232 M. Colléter, C. Brown / Animal Behaviour 81 (2011) 1231e1237

In a wide range of species, individuals compete with one another to Gladstone (Brown & Warburton 1999; Allen et al. 2002). In the
for access to resources and competition often results in the forma- wild, these fish form small shoals commonly consisting of one to six
tion of a hierarchy. Dominance has significant costs and benefits fish, but larger shoals also exist (Brown 2000). Their standard
associated with it, but dominant individuals tend to have higher length reaches 80 mm after maturation and they seldom live
reproductive success. Some of the benefits of dominance include beyond 2 years of age. This species is sexually dimorphic. Males are
preferential access to sexual partners, shelter or food resources. larger, more colourful and more aggressive than females. Both
Chase et al. (2002) suggested that an individual’s position in the males and females form hierarchies within the shoal, but malee
hierarchy is the product of both individual characteristics (physical male competition is generally more intense than that among
and behavioural) and dynamic social interactions. Although most females. We used mature male fish that were 18 months old that
research tends to focus on the influence of physical traits, suites of had been laboratory reared for multiple generations (for details see
correlated behavioural traits may also play a key role in determining Kydd & Brown 2009).
an individual’s position in the hierarchy. For example, less explor-
atory mountain chickadees, Poecile gambeli, were more dominant Experimental Design and Procedures
when paired with more exploratory conspecifics in pairwise
encounters, but no link existed between exploration and boldness We studied the aggressiveness, boldness and activity levels of 28
(Fox et al. 2009). In fishes, the opposite trend seems to exist, with males. The males were housed in four aquaria (92
36 cm and 50 cm
highly active individuals being more aggressive (Huntingford 1976). deep) in social groups consisting of seven males and four females.
Moreover, fish that show low stress responses while exploring a new Each aquarium was filled with tap water aged for 1 week prior to use
environment are more likely to become dominant (Schjolden et al. and was furnished with river gravel, plastic plants and spawning nets.
2005). To appreciate fully the importance of personality traits from These elements created a seminatural environment, which encour-
an evolutionary perspective, it is imperative that they are directly aged relatively natural behaviour. The presence of male and female
linked with individual fitness measures (Smith & Blumstein 2008). conspecifics facilitated shoaling behaviour and other social interac-
For example, bold geckos, Lepidodactylus lugubris, are less active but tions including hierarchy formation. The water temperature was
have higher foraging rates (Short & Petren 2008) and female guppies, maintained at 23  C, light was maintained on a 12:12 h light:dark
Poecilia reticulata, prefer to associate with bold males (Godin & ratio and the fish were fed every day with commercial flake food.
Dugatkin 1996). Links between personality traits and male repro- The 28 males allowed us to form seven groups of four unfamiliar
ductive success suggest that personality traits may be an important males, each selected from different aquaria. All fish were tagged
component of life history strategies (Reale et al. 2009). using Fluorescent Visible Implant Elastomer tags (Northwest
A boldnesseaggressivenesseactivity syndrome has been identi- Marine Technology Inc., Shaw Island, WA, U.S.A.). Two colours were
fied in some populations of three-spined sticklebacks, Gasterosteus used in four combinations of tag placement allowing us to distin-
aculeatus (Huntingford 1976; Bell 2005; Dingemanse et al. 2007). For guish between individuals. During the tagging procedure, each fish
example, Bell (2005) found a correlation between the three traits in was lightly anaesthetized by placing it in 1 litre of tank water,
one population of sticklebacks but not in another. When data for containing 0.6 g of NaHCO3 and 0.3 g of MS222, for 30 s. After
both populations were combined, however, bold sticklebacks tended tagging, each fish was weighed, measured and then placed back in
to be less aggressive than shy sticklebacks. In contrast, Huntingford the housing aquarium. No mortalities occurred as a result of
(1976) found strong correlations between boldness and aggression tagging and all fish recovered quickly. At the end of all the exper-
in sticklebacks collected from a range of habitats that varied over the iments all fish were rehomed with aquarium enthusiasts in accor-
course of the breeding season but no population-level analysis was dance with our ethics licence. This research was conducted under
performed. Dingemanse et al. (2007) similarly investigated the an animal ethics licence from Macquarie University.
relationship between activity, boldness and aggression in 12 pop-
ulations of sticklebacks that varied in the level of predation pressure. Experiment 1: Hierarchy Establishment
A positive correlation was found only in those populations subject to
high predation pressure, suggesting that key environmental vari- In this experiment, we aimed to establish new hierarchies in each
ables may drive the coevolution of a number of traits and that this group of four males, which also provided information regarding their
syndrome may be domain specific. We extended this work by level of aggression. In addition, we were interested to learn whether
examining the potential influence of this syndrome on an important a female audience would alter the nature of the aggressive interac-
fitness component, that of male hierarchy formation. tions between individuals (i.e. a mild change of context) as observed
In this study, we aimed to examine the link between aggressive- in previous studies (Dzieweczynski et al. 2005). Groups of familiar
ness, boldness and activity with individual rank in the hierarchy of males have well-established hierarchies; thus agonistic interactions
male crimson-spotted rainbowfish, Melanotaenia duboulayi. Previous occur relatively infrequently and are difficult to assess. Therefore,
studies examining female mate choice in this species have shown individuals were taken from different aquaria to ensure they were
that females exercise choice in a dichotomous chamber based on not familiar with one another, so as to maximize the interactions
male size and display rate when males are held behind barriers. between the fish during hierarchy formation. Moreover, hierarchy
When males and the choosing female interact freely, however, the position in small groups tends to be maintained even after
dominant male prevents subordinate males from approaching the subgroups are merged (Earley & Dugatkin 2006). Thus staging dyadic
female (Collins 1999). Thus male dominance in this species leads to interactions allowed us to maximize the extraction of behavioural
a form of mate guarding and enhances male reproductive success data while providing useful information about the relative hierarchy
(Young et al. 2010). position of males within each group.
Six trials were run in every group of fish allowing for dyadic
METHODS interactions between all individuals (i.e. 1 versus 2, 1 versus 3, 1
versus 4, 2 versus 3, 2 versus 4, 3 versus 4). The order of the
Study Organism interactions was randomized and a single contest within each
group was staged one after the other. The minimum separation
The crimson-spotted rainbowfish is a freshwater fish occupying between trials using the same individuals was 7 days. This
lakes and streams on the east coast of Australia from Coffs Harbour approach helped minimize potential winner/loser effects (Hsu et al.
 M. Colléter, C. Brown / Animal Behaviour 81 (2011) 1231e1237 1233

2006). The experimental arena was furnished in a fashion similar to Experiment 3: Measuring Activity Level
the home aquaria, including the provision of aquatic plants to
provide refuge so fish could escape and hide from aggressors. In Two weeks after the conclusion of experiment 2, the activity
addition, a clear divide could be added and removed as necessary. level of each male was determined. Activity is best examined in
For half of the trials, two males were placed directly in the exper- a familiar environment in which the animal does not feel threat-
imental arena without the divide and their behaviour was observed ened and thus displays its natural behavioural patterns. In this last
for 30 min. Subsequently, the males were removed for 20 min and experiment, we established an activity classification of each male
a transparent divide was added to create a small compartment fish in its home aquarium in the presence of its fellow shoalmates.
(20 cm in length) and a larger compartment (70 cm in length). An The four home aquaria were divided into six equal parts, each
unfamiliar female was placed in the smaller of the two compart- 15 cm long, by drawing on the glass using a marker pen. We hung
ments and allowed to settle for 15 min. The males were placed back a sheet 65 cm from the aquaria behind which an observer could hide.
in the large compartment and their behaviour was observed for A rectangular hole (90
2 cm) was cut in the sheet to allow the
a further 30 min. For the other half of the trials, we repeated the observer to watch the fish. We allowed the fish 24 h to habituate to
procedure but in reverse order: the males were first observed with the presence of the sheet and then recorded the activity of each male
a female present and then in the absence of the female. During our for 10 min. The level of activity was defined as the number of tran-
observations the following behaviours were documented using sects crossed during the observation period.
Etholog 2.2 (Ottoni 2000): lateral displays, head displays, fins up,
bite, charge, chase/flee and hide (for descriptions see Table 1). Statistical Analysis
Rainbowfish are insectivorous and have small mouths, so bites
cause no serious harm to conspecifics. Moreover, fish had the We analysed the behavioural data from experiment 1 using
opportunity to flee and hide if necessary. If fights became overly a principal components analysis (PCA) to identify suites of traits that
aggressive we planned to intervene, but this step was never clustered together as identifiable behavioural dimensions. Principal
necessary. For every pair there was a clear winner by the end of the components with eigenvalues greater than 1 were considered for
contest. By examining the winners in all dyadic interactions, indi- further interpretation. The coordinates of the individuals in dimen-
vidual fish were assigned ranks within their respective groups sion 1 of the PCA were considered as the aggressiveness scores while
(dominant rank 1 to subordinate rank 4). PC2 represented sociality scores (see Results). These scores were then
analysed together with boldness scores from experiment 2 and
Experiment 2: Measuring Boldness activity levels from experiment 3 using hierarchy position as the
independent variable. The time spent hiding was log þ1 transformed
The boldness experiment was conducted 2 weeks after the and the time spent close to the novel object was log (1 þ x) trans-
conclusion of experiment 1. In this experiment, we established formed to normalize the distributions. Jonckheere’s trend test was
a boldnesseshyness classification for each of the 28 males. The employed to examine the data because we had a specific hypothesis
experimental apparatus consisted of an aquarium (33
90 cm and relating dominance rank to our dependent variables. Comparisons
38 cm deep) with the addition of a transparent divide placed 20 cm between ranks were further analysed using Fisher’s protected least
from one end of the tank. This divide was equipped with a rectan- significant difference (PLSD) test (ANOVA Statview, SAS Institute Inc.,
gular trapdoor (15
20 cm) which could be opened remotely by an Cary, NC, U.S.A.). The female observer effect on male aggressiveness
observer. A novel object (a coke can) was placed at the other was analysed using ANOVA. Although we attempted to match males
extremity of the tank. A grid was constructed over the top of the for size, we also analysed male standard length as a dependent
tank and the behaviour of the fish was recorded on a video camera variable.
mounted overhead. Each fish was placed in the start compartment Finally, a linear regression was conducted examining the associ-
for 5 min with the trapdoor closed. During this time the fish ations between the aggressiveness score, number of transects
recovered from handling and explored the start compartment. It crossed (activity) and the time spent near a novel object during the
could also view the novel object from afar. After 5 min, the trapdoor first 5 min (boldness) using R 2.8.0 (The R Foundation for Statistical
was opened allowing the fish to emerge from the compartment, Computing, Vienna, Austria).
explore the novel environment and approach the novel object. We
analysed the progress of the fish for 15 min using Etholog 2.2 while RESULTS
watching the video on a remote computer monitor. The test aquaria
and the coke can were rinsed with tap water between trials. The Experiment 1: Hierarchy Establishment
time to emerge from the small compartment and the time spent
close to the novel object during the first 5 min (both measures of Each of the trials resulted in a clear winner and each social group
boldness) were recorded. of males displayed a clear and simple linear hierarchy. The variables
factor map (Fig. 1) showed a strong relationship between the
variables ‘chase’, ‘bite’, ‘charge’ and ‘lateral display’, all of which are
Table 1 characteristic of highly aggressive individuals in a range of fish
A brief explanation of the primary behaviours displayed by male rainbowfish during species. Similarly, the variables ‘fins up’ and ‘head display’ were
agonistic interactions
grouped together and probably represent general social displays.
Behaviour Description ‘Hide’, in contrast, was isolated by itself and diametrically opposed
Lateral display Fish swims laterally alongside the other fish to the aggressive behaviours.
with fins extended The first principal component explained 50% of the total variation
Head display Fish puts its head up, tail down, forms an S shape
(eigenvalue ¼ 3.5) and from this analysis it was apparent that the first
with the body while extending the first dorsal fin
Bite Fish bites the other fish dimension exclusively expressed aggressive behaviours; therefore,
Charge Fish swims rapidly in direction of the other fish’s flank the component of each male in this dimension was used as the
Chase Fish rapidly follows the other fish which tries to escape aggressiveness score for further analysis. The second principal
Fins up Fish opens mouth and extends dorsal fins component explained a further 24% of the variation
Hide Fish hides in the refuge out of view
(eigenvalue ¼ 1.5) and comprised the variables ‘hide’ and ‘head
1234 M. Colléter, C. Brown / Animal Behaviour 81 (2011) 1231e1237

Hide

Dimension 2 (21.19%) 0.5

Chase
Bite
Charge

0
Lateral display

−0.5
Head display
Fins up

−1

−1 −0.5 0 0.5 1
Dimension 1 (50.24%)

Figure 1. Variables factor map (principal components analysis) displaying the relationship between behaviours performed during agonistic interactions with respect to the first two
principal components.

display’. These two variables were strongly negatively correlated Our observations during the experiment suggest that the domi-
(with N ¼ 28, the minimum significant correlation r ¼ 0.37; Table 2) nant fish used different degrees of aggression in relation to the
and we interpreted this factor as a sociality score, ‘hide’ being an behaviour of the opponent fish. If the opponent was relatively meek,
asocial response and ‘head display’ being a social signal. The third dominant fish had little need to be aggressive and brief displays were
principal component explained 14% of the variance and comprised enough to settle the contest very quickly. If the opponent was very
solely the variable ‘fins up’. This factor had an eigenvalue just under aggressive, on the other hand, the dominant male employed a series
1 (0.97) and was therefore not considered in further analyses. of aggressive behaviours to exert his dominance over the opponent
As expected, analysis of the aggressiveness score revealed that and the length of the contest was extended. Moreover, subordinate
fish ranked higher in the hierarchy were significantly more aggres- fish sometimes initiated aggressive interactions with a series of
sive than those ranked lower in the hierarchy (Jonckheere’s trend displays but were eventually met with a single strong response from
test: S ¼ 174, C ¼ 4, N ¼ 7, P < 0.01; Fig. 2a). Post hoc analysis their dominant opponents, which ended the contest immediately.
(Fisher’s PLSD) revealed a significant difference between the These general observations may explain the absence of significant
aggressiveness of the dominant fish and those ranked 3rd and 4th differences between dominant and subordinates in some measures
(P ¼ 0.002 and P < 0.001, respectively). There were also significant (see below).
differences between the second-ranked males and those ranked 3rd Surprisingly, the presence or absence of the female had limited
and 4th (P ¼ 0.03 and 0.016, respectively). In general, the same was impact on male behaviour. Our analysis showed some trends but no
true for the sociality score. Highly ranked males were more social significant relationships (ANOVA: P > 0.05 in all cases).
than lowly ranked males (Jonckheere’s trend test: S ¼ 106, C ¼ 4,
N ¼ 7, P < 0.05; Fig. 2b). Specifically, fish from ranks 3 and 4 were Experiment 2: Boldness
less social than the fish from rank 1 (Fisher’s PLSD: P ¼ 0.007 and
P ¼ 0.034, respectively) and fish ranked 3 were less social than fish Every male emerged from the start compartment within the
ranked 2 (Fisher’s PLSD: P ¼ 0.031; Fig. 2b). allotted 15 min. Our analysis showed that male rank had an effect
on boldness scores (Jonckheere’s trend test: S ¼ 124, C ¼ 4, N ¼ 7,
Table 2
P < 0.01; Fig. 2c). In general, the dominant fish emerged from the
Pearson coefficients correlation matrix showing the relationship between
behavioural variables during agonistic interactions in male rainbowfish
start compartment the most quickly followed by males ranked 2, 3
and 4. Fish ranked 1 emerged more quickly than those ranked 4
Bite Charge Chase Fins up Head Hide Lateral
(Fishers PLSD: P ¼ 0.032). Dominant fish (rank 1) also spent more
display display
time near the novel object during the first 5 min than all the other
Bite 1.000 0.949 0.911 0.063 0.228 0.051 0.682
fish (Fisher’s PLSD: 1 versus 2: P ¼ 0.038; 1 versus 3: P ¼ 0.026;
Charge 1.000 0.898 0.180 0.254 0.158 0.724
Chase 1.000 0.070 0.282 0.122 0.458 1 versus 4: P ¼ 0.030). The trend test, however, was not significant
Fins up 1.000 0.226 0.241 0.351 because fish ranked 2e4 showed similar inspection levels (Jonck-
Head display 1.000 0.424 0.194 heere’s trend test: S ¼ 8, C ¼ 4, N ¼ 7, NS). When taken together,
Hide 1.000 0.170
these two measures of boldness suggest that higher-ranked males
Lateral display 1.000
had higher levels of boldness.
 M. Colléter, C. Brown / Animal Behaviour 81 (2011) 1231e1237 1235

100 30
90 (a) (b)
25
80
Aggressiveness score

Sociality score
70
20
60
50 15
40
10
30
20 5
10
0 0

400 120
(c) (d)
Boldness (latency to emerge)

350
100
300

Activity score
80
250
200 60
150
40
100
20
50
0 0
1 2 3 4 1 2 3 4
Male rank

Figure 2. Mean  SE (a) aggressiveness, (b) sociality, (c) boldness and (d) activity score for males ranked 1e4 in the hierarchy. All scores shown are raw data and statistics were
performed on transformed data where appropriate. The aggressiveness and sociality scores are composite scores derived from a principal components analysis, the boldness score is
the time to emerge from the start compartment and the activity score refers to the number of transects crossed.

Experiment 3: Activity Level DISCUSSION

Measurement of activity levels in the home tank showed a clear Personality traits are becoming increasingly important in
pattern. In general, the higher the rank of the fish in the hierarchy explaining individual differences in the behaviour of animals (Wilson
the higher its level of activity (Jonckheere’s trend test: S ¼ 102, 1998; Gosling 2001). Almost by definition, personality traits repre-
C ¼ 4, N ¼ 7, P < 0.05; Fig. 2d). Fish from ranks 1 and 2 were sent a leading edge of the evolutionary process since natural selection
significantly more active than fish ranked 4 (Fisher’s PLSD: acts directly on variation of individual phenotypes within pop-
P < 0.025 in each case). ulations. Surprisingly few studies have investigated the link between
personality traits and fitness-related variables (great tits:
Morphology Measures Dingemanse et al. 2003; fish, Brachyrhaphis episcopi: Brown et al.
2007; sheep, Ovis aries: Reale et al. 2009). Here, we have shown
The analysis of the morphological measures showed a strong that a range of personality traits (aggression, activity and boldness)
correlation between each of the variables (length, weight and body all contribute to a male’s position in a social hierarchy. Highly ranked
depth). Therefore we chose to conduct an ANOVA examining stan- individuals have the highest reproductive output because of their
dard length and its relationship with fish rank. Despite trying to ability to dominate access to resources in social contexts (Ellis 1995;
match the fish as much as possible in each group, size was still Young et al. 2010). Moreover, we have shown that activity levels in
significantly related to dominance rank (ANOVA: F3,24 ¼ 3.792, the home tank and boldness were significantly correlated suggesting
P ¼ 0.021). Dominant fish were about 4 mm larger than subordinates that an activityeboldnessehierarchy rank syndrome exists in male
(Fisher’s PLSD: P ¼ 0.045, 0.047 and 0.004 for fish ranked 2, 3 and 4, rainbowfish. Our results suggest that personality traits predict the
respectively), although no differences were observed in the standard outcome of maleemale competition that determines access to
length of fish ranked 2, 3 or 4. No significant correlations were found females and thereby male reproductive success. The correlations
between the morphology traits and all the other behavioural traits. between these traits, therefore, may be the product of sexual selec-
tion (Wolf et al. 2007; Biro & Stamps 2008).
Regression Analysis In male rainbowfish, we found a correlation between activity
levels in the home aquaria and boldness as measured as the tendency
Despite the link between activity, boldness, aggression and the to emerge into a novel arena and approach a novel object. Such
position in the hierarchy, not all of these behaviours were corre- correlations have been revealed previously in some populations of
lated with each other. In fact, we only found a significant correla- sticklebacks and zebrafish, Danio rerio, but not others (Huntingford
tion between activity in the home aquarium and boldness scores 1976; Bell 2005; Moretz et al. 2007; Dingemanse et al. 2007). Both
(linear regression: t ¼ 2.278, P ¼ 0.031). Bolder fish were more Huntingford (1976) and Dingemanse et al. (2007) suggested that
active than shy fish. There were no significant relationships variation in predation regime was driving the coevolution of certain
between activity and aggressiveness or boldness and aggressive- traits, because correlations between activity and boldness only exis-
ness (t ¼ 1.121, P ¼ 0.136 and t ¼ 0.937, P ¼ 0.179, respectively). ted in populations subject to high levels of predation. Here, we have
1236 M. Colléter, C. Brown / Animal Behaviour 81 (2011) 1231e1237

shown that coevolution of these behavioural traits may also emerge in examining the number of displays to a mirror image or responses to
a completely different context, that of interindividual conflict. model invaders (e.g. Peeke et al. 1969; Clotfelter & Kuperberg 2007),
Competition for limited resources such as food and mates is often but these methodologies, while testing individuals in isolation, still
fierce among individuals. Although obtaining a high position in the have problems with feedback loops. In addition, the composition of
hierarchy has multiple benefits in male fishes, competition is the shoal in the home tank from which the males were drawn could
primarily over access to females, since males tend to be motivated by have had a strong impact on male aggressive behaviour owing to
sex rather than by hunger (Griffiths & Magurran 1998). Thus, we prolonged winning and losing effects (Hsu et al. 2006).
suggest that correlations across personality traits can be driven by To conclude, once aggressiveness scores were assigned to indi-
competition between conspecifics, potentially by sexual selection, viduals, we were able to illustrate how aggressiveness, boldness and
and are therefore domain specific. Similar observations have been activity levels covaried with a male’s position in the hierarchy.
made in three strains of captive-reared zebrafish (Moretz et al. 2007) Dominant males were bolder, more active, more aggressive and
which suggests that such trait correlations can be maintained even slightly larger than subordinate males. While female rainbowfish are
under captive conditions. Further work is clearly required to choosy if given the opportunity, males compete heavily for females
isolate the relative importance of natural versus sexual selection in and dominant males tend to monopolize access to females (Collins
shaping the evolution and development of correlated behavioural 1999; Young et al. 2010). In this way, male dominance leads to
traits. a form of mate guarding which increases their reproductive success
The widespread appearance of the boldnesseaggressiveness (Birkhead 1987; Warner et al. 1995; Reavis 1997). Like predation
syndrome across a range of species has encouraged the search for pressure, high levels of competition can have dramatic impacts on the
a single underlying hormonal or genetic mechanism (Veenema et al. evolution of behavioural traits (Hamilton 1964; West-Eberhard
2003). Boldness scores in males are typically higher than those in 1979). In the case of male rainbowfish it appears that intrasexual
females across a range of species including humans (Wilson et al. selection is the most likely driving force behind the coevolution of
1994) and are related to testosterone levels (Daisley et al. 2005). behavioural traits that comprise the boldnesseactivity syndrome.
Experimental enhancement of testosterone in dark-eyed juncos, Clearly, further work needs to be conducted to illuminate the links
Junco hyemalis, increased spatial activity and aggression relative to between life history strategies and personality traits, and in partic-
controls (Chandler 1994). Bolder male guppies are also more ular, how they are shaped by natural and sexual selection.
aggressive than shy males (Huntingford 1976; Bell 2005; Dingemanse
et al. 2007). Not surprisingly, such traits are also linked to dominance
status in social groups. Male rainbowfish are both larger and more Acknowledgments
colourful than females. During the breeding season, males are highly
active, chase off rivals, pursue and display to females. Moreover, mate We thank Erin Kydd, Jennie Morgan, Anne-Laurence Bibost,
choice experiments show that male display rate, which is charac- Suzanne Artiss and Hervé Le Bris for their support in carrying out this
terized by high levels of activity, is a good predictor of female pref- project. This research was funded by an internship to M.C. from
erence and male quality in this and other fish species (Nicoletto 1993; Agrocampus Ouest and Macquarie University. C.B. was supported by
Matthews et al. 1997; Collins 1999). Females also prefer to mate with an Australian Research Fellowship from the Australian Research
larger males and larger males are often the victor in maleemale Council.
encounters (Collins 1999; Evans et al. 2010; Young et al. 2010). Thus it
makes intuitive sense that activity, boldness and aggressiveness
levels are a good indicator of a male’s position in the hierarchy and References
elevated activity is probably indicative of underlying androgen levels.
It is likely, therefore, that correlations between boldness and activity Allen, G. R., Midgley, S. H. & Allen, M. 2002. Field Guide to the Freshwater Fishes of
Australia. Perth: Western Australian Museum.
are driven by a common underlying hormonal mechanism. Thus
Bell, A. M. 2005. Behavioural differences between individuals and two populations
there is emerging evidence that boldness, aggression and activity of stickleback (Gasterosteus aculeatus). Journal of Evolutionary Biology, 18,
may all be linked to androgen levels across a range of taxa. 464e473.
Bell, A. M. & Stamps, J. A. 2004. Development of behavioural differences between
Intriguingly, aggression was not significantly correlated with
individuals and populations of sticklebacks, Gasterosteus aculeatus. Animal
activity levels or boldness despite the fact that all three factors Behaviour, 68, 1339e1348.
explain some of the variation in male hierarchy rank. There are Birkhead, T. 1987. Sperm competition in birds. Trends in Ecology & Evolution, 2,
several possible reasons to explain this result. First, aggressiveness 268e272.
Biro, P. A. & Stamps, J. A. 2008. Are animal personality traits linked to life-history
and activity were scored in the presence of conspecifics whereas the productivity? Trends in Ecology & Evolution, 23, 361e368.
boldness measures were recorded in isolated fish. There is emerging Brown, C. 2000 Behavioural ecology of predator avoidance in rainbowfish (Mela-
evidence that the expression of personality traits varies with social notaenia spp). Ph.D. thesis, University of Queensland.
Brown, C. & Braithwaite, V. A. 2004. Size matters: a test of boldness in eight
context (Ward et al. 2004; van Oers et al. 2005; Webster et al. 2007). populations of bishop, Brachyraphis episcopi. Animal Behaviour, 68, 1325e1329.
For example, pairs of bold fish might show signs of positive feedback Brown, C. & Laland, K. 2002. Social enhancement and social inhibition of foraging
thus causing them to behave more boldly when tested together, behaviour in hatchery-reared Atlantic salmon (Salmo salar). Journal of Fish
Biology, 61, 987e998.
whereas pairs of shy fish show the opposite effect (Brown & Laland Brown, C. & Warburton, K. 1999. Social mechanisms enhance escape responses
2002). During our staged dyadic interactions, a relatively aggressive in the rainbowfish Melanotaenia duboulayi. Environmental Biology of Fishes,
fish could face another aggressive individual, so both fish might 56, 455e459.
Brown, C., Jones, F. & Braithwaite, V. A. 2005. In situ examination of boldnesseshyness
escalate the conflict and express very high levels of aggression.
traits in the tropical poeciliid, Brachyraphis episcopi. Animal Behaviour, 70,
Obviously, the opposite would occur if an aggressive individual 1003e1009.
confronted a nonaggressive individual. To counteract this potential Brown, C., Jones, F. C. & Braithwaite, V. A. 2007. Correlation between boldness and
body mass in natural populations of the poeciliid Brachyrhaphis episcopi. Journal
problem to some extent, we recorded aggression across three
of Fish Biology, 71, 1590e1601.
interactions with different males. Nevertheless, one of our seven Budaev, S. V. 1997. ‘Personality’ in the guppy (Poecilia reticulata): a correlational
social groups showed very low levels of aggression across the board. study of exploratory behavior and social tendency. Journal of Comparative
The most aggressive male in this group displayed aggressive Psychology, 111, 399e411.
Budaev, S. V. & Brown C. In press. Personality traits and behaviour. In: Fish
behaviours that were more symptomatic of fish ranked 3 and 4 in Cognition and Behaviour. 2nd edn (Ed. by C. Brown, K. Laland & J. Krause).
the other social groups. Previous authors have tested aggression by Oxford: J. Wiley.
 M. Colléter, C. Brown / Animal Behaviour 81 (2011) 1231e1237 1237

Budaev, S. V., Zworykin, D. D. & Mochek, A. D. 1999. Individual differences in Matthews, I. M., Evans, J. P. & Magurran, A. E. 1997. Male display rate reveals
parental care and behaviour profile in the convict cichlid: a correlation study. ejaculate characteristics in the Trinidadian guppy, Poecilia reticulata. Proceedings
Animal Behaviour, 58, 195e202. of the Royal Society B, 264, 695e700.
Chandler, C. R. 1994. Effects of testosterone on spatial activity in free-ranging male Moretz, J. A., Martins, E. P. & Robison, B. D. 2007. Behavioral syndromes and the
dark-eyed juncos, Junco hyemalis. Animal Behaviour, 47, 1445e1455. evolution of correlated behavior in zebrafish. Behavioral Ecology, 18, 556e562.
Chase, I. D., Tovey, C., Spangler-Martin, D. & Manfredonia, M. 2002. Individual Nicoletto, P. F. 1993. Female sexual response to condition dependent ornaments in
differences versus social dynamics in the formation of animal dominance hier- the guppy, Poecilia reticulata. Animal Behaviour, 46, 441e450.
archies. Proceedings of the National Academy of Sciences U.S.A., 99, 5744e5749. van Oers, K., Klunder, M. & Drent, P. J. 2005. Context dependence of personal-
Clotfelter, E. D. & Kuperberg, E. S. 2007. Cerebral lateralisation and its relationship ities: risk-taking behavior in a social and a nonsocial situation. Behavioral
to phylogeny and aggression in Anabantoid fishes. Brain Behavior and Evolution, Ecology, 16, 716e723.
69, 169e275. Ottoni, E. B. 2000. EthoLog 2.2: a tool for the transcription and timing of
Collins, K. T. 1999. The influence of predator exposure on the reproductive behavior observation sessions. Behavior Research Methods, Instruments, &
behaviour of the crimson-spotted rainbowfish (Melanotaenia duboulayi). Computers, 32, 446e449.
Honours thesis, University of Queensland. Peeke, H. V. S., Wyers, E. J. & Herz, M. J. 1969. Waning of the aggressive response to
Daisley, J. N., Bromundt, V., Mostl, E. & Kotrschal, K. 2005. Enhanced yolk male models in the three-spined stickleback (Gasterosteus aculeatus L.). Animal
testosterone influences behavioral phenotype independent of sex in Japanese Behaviour, 17, 224e228.
quail chicks Coturnix japonica. Hormones and Behavior, 47, 185e194. Reale, D., Martin, J., Coltman, D. W., Poissant, J. & Festa-Bianchet, M. 2009. Male
Dingemanse, N. J., Both, C., van Noordwijk, A. J., Rutten, A. L. & Drent, P. J. 2003. personality, life-history strategies and reproductive success in a promiscuous
Natal dispersal and personalities in great tits (Parus major). Proceedings of the mammal. Journal of Evolutionary Biology, 22, 1599e1607.
Royal Society B, 270, 741e747. Réale, D., Reader, S. M., Sol, D., McDougall, P. T. & Dingemanse, N. J. 2007. Inte-
Dingemanse, N. J., Both, C., Drent, P. J. & Tinbergen, J. M. 2004. Fitness conse- grating animal temperament within ecology and evolution. Biological Reviews,
quences of avian personalities in a fluctuating environment. Proceedings of the 82, 291e318.
Royal Society B, 271, 847e852. Reavis, R. H. 1997. The natural history of a monogamous coral-reef fish, Valen-
Dzieweczynski, T. L., Earley, R. L., Green, T. M. & Rowland, W. J. 2005. Audience ciennea strigata (Gobiidae): 2. Behavior, mate fidelity and reproductive success.
effect is context dependent in Siamese fighting fish, Betta splendens. Behavioral Environmental Biology of Fishes, 49, 247e257.
Ecology, 16, 1025e1030. Riechert, S. E. & Hedrick, A. V. 1993. A test of correlations among fitness-related
Dingemanse, N. J., Wright, J., Kazem, A. J. N., Thomas, D. K., Hickling, R. & behavioral traits in the spider, Agelenopsis aperta (Araneae, Agelinadae). Animal
Dawnay, N. 2007. Behavioural syndromes differ predictably between 12 pop- Behaviour, 46, 669e675.
ulations of three-spined stickleback. Journal of Animal Ecology, 76, 1128e1138. Schjolden, J., Stoskhus, A. & Winberg, S. 2005. Does individual variation in
Earley, R. L. & Dugatkin, L. A. 2006. Merging social hierarchies: effects on domi- stress responses and agonistic behavior reflect divergent stress coping
nance rank in male green swordtail fish (Xiphophorus helleri). Behavioural strategies in juvenile rainbow trout? Physiological and Biochemical Zoology,
Processes, 73, 290e298. 78, 715e723.
Ellis, L. 1995. Dominance and reproductive success among nonhuman animals: Short, K. H. & Petren, K. 2008. Boldness underlies foraging success of invasive
a cross-species comparison. Ethology and Sociobiology, 16, 257e333. Lepidodactylus lugubris geckos in the human landscape. Animal Behaviour,
Evans, J. P., Box, T. M., Brooshooft, P., Tatler, J. R. & Fitzpatrick, J. L. 2010. Females 76, 429e437.
increase egg deposition in favor of large males in the rainbowfish, Melanotaenia Sih, A., Bell, A. & Johnson, J. C. 2004. Behavioral syndromes: an ecological and
australis. Behavioral Ecology, 21, 465e469. evolutionary overview. Trends in Ecology & Evolution, 19, 372e378.
Fox, R. A., Ladage, L. S., Rothe, T. C. & Pravosudov, V. V. 2009. Behavioural profile Smith, B. R. & Blumstein, D. T. 2008. Fitness consequences of personality: a meta-
predicts dominance status in mountain chickadees, Poecile gambeli. Animal analysis. Behavioral Ecology, 19, 448e455.
Behaviour, 77, 1441e1448. Veenema, A. H., Meijer, O. C., de Kloet, E. R. & Koolhaas, J. M. 2003. Genetic
Godin, J. G. J. & Dugatkin, L. A. 1996. Female mating preference for bold males in selection for coping style predicts stressor susceptibility. Journal of Neuroen-
the guppy, Poecilia reticulata. Proceedings of the National Academy of Sciences docrinology, 15, 256e267.
U.S.A., 93, 10262e10267. Ward, A. J. W., Thomas, P., Hart, P. J. B. & Krause, J. 2004. Correlates of boldness in
Gosling, S. D. 2001. From mice to men: what can we learn about personality from three-spined sticklebacks (Gasterosteus aculeatus). Behavioral Ecology and
animal research? Psychological Bulletin, 127, 45e86. Sociobiology, 55, 561e568.
Griffiths, S. W. & Magurran, A. E. 1998. Sex and schooling behaviour in the Warner, R. R., Shapiro, D. Y., Marcanato, A. & Petersen, C. W. 1995. Sexual conflict:
Trinidadian guppy. Animal Behaviour, 56, 689e693. males with highest mating success convey the lowest fertilization benefits to
Hamilton, W. D. 1964. The genetical evolution of social behaviour. I. Journal of females. Proceedings of the Royal Society B, 262, 135e139.
Theoretical Biology, 7, 1e16. Webster, M. M., Ward, A. J. W. & Hart, P. J. B. 2007. Boldness is influenced by social
Harris, S., Ramnarine, I. W., Smith, H. G. & Pettersson, L. B. 2010. Picking context in threespine sticklebacks (Gasterosteus aculeatus). Behaviour, 144,
personalities apart: estimating the influence of predation, sex and body size on 351e371.
boldness in the guppy Poecilia reticulata. Oikos, 119, 1711e1718. West-Eberhard, M. J. 1979. Sexual selection, social competition and evolution.
Hsu, Y., Earley, R. L. & Wolf, L. L. 2006. Modulating aggression through experi- Proceedings of the American Philosophical Society, 123, 222e234.
ence. In: Fish Cognition and Behaviour (Ed. by C. Brown, K. N. Laland & Wilson, D. S. 1998. Adaptive individual differences within single populations.
J. Krause), pp. 96e118. Oxford: Blackwell Science. Philosophical Transactions of the Royal Society B, 353, 199e205.
Huntingford, F. A. 1976. The relationship between anti-predator behaviour and Wilson, D. S., Coleman, K., Clark, A. B. & Biederman, L. 1993. Shy bold continuum
aggression among conspecifics in the three-spined stickleback, Gasterosteus in pumpkinseed sunfish (Lepomis gibbosus): an ecological study of a psycho-
aculeatus. Animal Behaviour, 24, 245e260. logical trait. Journal of Comparative Psychology, 107, 250e260.
Ioannou, C., Payne, M. & Krause, J. 2008. Ecological consequences of the boldeshy Wilson, D. S., Clark, A. B., Coleman, K. & Dearstyne, T. 1994. Shyness and boldness
continuum: the effect of predator boldness on prey risk. Oecologia, 157, 177e182. in humans and other animals. Trends in Ecology & Evolution, 9, 442e446.
Kydd, E. & Brown, C. 2009. Loss of shoaling preference for familiar individuals in Wolf, M., van Doorn, G. S., Leimar, O. & Weissing, F. J. 2007. Life-history trade-offs
captive-reared crimson spotted rainbowfish Melanotaenia duboulayi. Journal of favour the evolution of animal personalities. Nature, 447, 581e585.
Fish Biology, 74, 2187e2195. Young, M., Simmons, L. & Evans, J. 2010. Pre- and post-mating sexual selection
Magellan, K. & Magurran, A. E. 2007. Behavioural profiles: individual consistency in both favor large males in a rainbowfish. Behavioral Ecology and Sociobiology,
male mating behaviour under varying sex ratios. Animal Behaviour, 74, 1545e1550. 64, 915e925.