The role of previous social experience on risk-taking and leadership in three-spined sticklebacks

Abstract

The emergence of leaders and followers is a key factor in facilitating group cohesion in animals. Individual group members
have been shown to respond strongly to each other’s behavior and thereby affect the emergence and maintenance of these social
roles. However, it is not known to what extent previous social experience might still affect individual’s leading and following
tendencies in later social interactions. Here, by pairing three-spined sticklebacks (Gasterosteus aculeatus) with 2 different consecutive partners, we show a carryover effect of a previous partner’s personality on the behavior of
focal individuals when paired with a new partner. This carryover effect depended on the relative boldness of the focal individual.
Relatively bold but not shy fish spent less time out of cover and led their current partner less if they had previously been
paired with a bolder partner. By contrast, following behavior was mainly influenced by the personality of the current partner.
Overall, the behavior of relatively bold fish was more consistent across the stages, whereas shy fish changed their behavior
more strongly depending on the current context. These findings emphasize how the history of previous social interactions can
play a role in the emergence and maintenance of social roles within groups, providing an additional route for individual differences
to affect collective behavior.

Full text

© The Author 2014. Published by Oxford University Press on behalf of
the International Society for Behavioral Ecology. All rights reserved. For
permissions, please e-mail: journals.permissions@oup.com
The ocial journal of the
ISBE
International Society for Behavioral Ecology
Behavioral
Ecology
Original Article
The role of previous social experience on
risk-taking and leadership in three-spined
sticklebacks
Jolle W.Jolles,a Adeline Fleetwood-Wilson,a ShinnosukeNakayama,a,b Martin C.Stumpe,c
Rufus A.Johnstone,a and AndreaManicaa
aDepartment of Zoology, University of Cambridge, Downing Street, Cambridge CB2 3EJ, UK,
bDepartment of Biology and Ecology of Fishes, Leibniz Institute of Freshwater Ecology and Inland
Fisheries, Müggelseedamm 310, Berlin 12587, Germany, and cAnTracks Computer Vision Systems,
Mountain View, CA, USA
Received 24 January 2014; revised 2 July 2014; accepted 18 July 2014.
The emergence of leaders and followers is a key factor in facilitating group cohesion in animals. Individual group members have
been shown to respond strongly to each other’s behavior and thereby affect the emergence and maintenance of these social roles.
However, it is not known to what extent previous social experience might still affect individual’s leading and following tendencies in
later social interactions. Here, by pairing three-spined sticklebacks (Gasterosteus aculeatus) with 2 different consecutive partners, we
show a carryover effect of a previous partner’s personality on the behavior of focal individuals when paired with a new partner. This
carryover effect depended on the relative boldness of the focal individual. Relatively bold but not shy fish spent less time out of cover
and led their current partner less if they had previously been paired with a bolder partner. By contrast, following behavior was mainly
influenced by the personality of the current partner. Overall, the behavior of relatively bold fish was more consistent across the stages,
whereas shy fish changed their behavior more strongly depending on the current context. These findings emphasize how the history
of previous social interactions can play a role in the emergence and maintenance of social roles within groups, providing an additional
route for individual differences to affect collective behavior.
Key words: boldness, collective decision-making, leadership, personality, responsiveness, shoaling.
INTRODUCTION
The emergence of leaders and followers plays a major role in pro-
moting group coordination and cohesion, with important conse-
quences for the social lives of humans as well as many nonhuman
animals (Krause and Ruxton 2002; Conradt and Roper 2009; Dyer
etal. 2009; King etal. 2009). There is a growing body of evidence
that individuals dier in their social roles, with some individuals
having a strong influence on group behavior while others mostly
follow (e.g., Reebs 2000; Harcourt etal. 2009; Nagy et al. 2010;
Flack etal. 2012; Nakayama etal. 2013). A key focus has been to
determine what factors predict which group members will become
leaders (Conradt and Roper 2003; Couzin et al. 2005; King etal.
2009). Many such factors have been identified in a large range of
species: body size (Krause et al. 1998; Reebs 2001), hunger level
(Krause et al. 1998; McClure etal. 2011; Nakayama, Johnstone,
et al. 2012), dominance (Peterson and Jacobs 2002; King et al.
2008; Jolles etal. 2013), social aliations (King et al. 2008; Jacobs
etal. 2011), sex (Peterson and Jacobs 2002; Barelli etal. 2008), age
(Réale and Festa-Bianchet 2003; Sueur and Petit 2008), boldness
(Beauchamp 2000; Ward etal. 2004; Harcourt etal. 2009; Kurvers
etal. 2009), sociability (Brown and Irving 2014), and knowledge or
experience (Reebs 2000; Couzin etal. 2005; Dyer etal. 2009; Flack
etal. 2012).
In recent years a few studies have started to go beyond the search
for such predictive factors and have shown that the actual dynam-
ics of interactions among individuals play an important role in
leading and following behavior (Harcourt etal. 2009; Nakayama,
Harcourt, et al. 2012; Nakayama et al. 2013; Pettit et al. 2013;
Ward et al. 2013). For example, although bold individuals typi-
cally lead and shy individuals mainly follow (Beauchamp 2000;
Harcourt etal. 2009; Kurvers etal. 2009; Nakayama et al. 2013),
these dierences in leading and following are strongly enhanced by
social feedback (Harcourt etal. 2009; Nakayama, Harcourt, etal.
Address correspondence to J.W. Jolles. E-mail: jj352@cam.ac.uk
Behavioral Ecology (2014), 00(00), 1–7. doi:10.1093/beheco/aru146
Behavioral Ecology Advance Access published August 21, 2014
at Cambridge University Library on August 22, 2014http://beheco.oxfordjournals.org/Downloaded from

Behavioral Ecology
2012). Furthermore, although bolder individuals are generally less
responsive to their partner’s behavior, both bolder and shyer indi-
viduals readily adjust to their partner when in the following role
(Nakayama, Harcourt, et al. 2012; Nakayama, Johnstone et al.
2012). These findings not only highlight the important modifying
role of social feedback, they also suggest the exciting possibility that
interactions with previous partners may play a role in later leading
and following behavior. As also highlighted in the human leadership
literature (Amit etal. 2009; Emery 2010; DuBrin 2013), addressing
this key outstanding issue may contribute to our understanding of
the emergence and maintenance of leadership and ultimately of
collective behavior and group decision making.
Most gregarious animals live in highly dynamic groups in which
they interact with multiple conspecifics (Krause and Ruxton 2002),
and a strong influence of previous social experience has already
been shown for neophobic and aggressive behavior (Hsu and Wolf
1999; Frost et al. 2007). In a previous study on leadership, fish
were shown to change their behavior based on a partner’s ability
to successfully locate food during joint trips, with experience over-
riding personality dierences in the tendency to follow but not to
lead (Nakayama etal. 2013). Here by pairing three-spined stickle-
backs (Gasterosteus aculeatus) with 2 dierent consecutive partners, we
investigated how previous social experience with other individuals
aected the propensity of fish to leave cover, to lead, and to follow
their current partner during joint trips. If individuals fine-tune their
behavior based on previous experiences, this potentially represents
a mechanism through which social roles can be reinforced. Since
bold individuals are known to be less responsive than shy individu-
als during social interactions (Pike etal. 2008; Nakayama, Harcourt,
et al. 2012; Nakayama, Johnstone, et al. 2012), we hypothesized
that bolder fish would be more consistent in their behavior across
dierent social and nonsocial environments and shyer fish to be
more responsive to the present context. We therefore predicted that
the behavior of bolder fish would be mainly explained by their own
personality and to a lesser extent by that of their current and pre-
vious partners, whereas for shyer fish the personality of their cur-
rent partner would be the main determinant of their behavior. This
approach provides a unique opportunity to describe important new
aspects of social feedback and personality that have thus far been
neglected in studies on group movements and leadership.
MATERIALS AND METHODS
Subjects and housing
We collected three-spined sticklebacks using a sweep net during
the summers of 2010–2012, from a small branch of the river Cam
(Cambridge, UK). Large groups of fish (~200 individuals) were
housed in a temperature-controlled laboratory (T=14 ± 1°C) with
a constant light regime (lights on from 09:00 to 19:00 h) and kept in
large glass holding aquaria (120 × 60 × 60 cm) that contained artifi-
cial plants, aeration, and undergravel filtration. Fish were fed frozen
bloodworm (Chironomidae) larvae ad libitum once a day before the
start of the experiment. During the experimental period, feeding
was rationed to one bloodworm a day to standardize hunger levels.
All fish used for the experiment were of similar length (50 ± 7 mm
from tip of snout to caudal peduncle) and were taken from a single
population to minimize population-specific genetic eects that may
influence personality (Bell 2005). Although the exact age of the fish
could not be determined, all caught individuals were juveniles and
are expected to only vary in age by a few weeks. Sex of the fish was
not identified as the temperature and photoperiod regime in the
lab prevented the fish from becoming sexually mature (Borg etal.
2004).
Experimentalsetup
During the experimental period, we housed fish individually in cus-
tom-holding tanks (60 × 30 × 40 cm) lined with gravel and divided
lengthwise into 6 compartments by transparent perspex partitions.
Five compartments were used to house a fish each and contained
an artificial plant at one end and a white perspex plate (2 × 2 cm) at
the other end where food was delivered. The remaining compart-
ment contained the undergravel filter and was not used to house
any fish. Partner fish were never housed in adjacent compartments.
Fish were allowed to acclimatize in their individual compartments
for 3days before the start of testing.
To investigate fish’s propensity to explore a risky area and lead
and follow conspecifics, we used a tank setup previously used in
our lab for similar experiments (Harcourt etal. 2009; Nakayama,
Harcourt, et al. 2012; Nakayama, Johnstone, et al. 2012). In
short, experiments took place in 4 identical experimental tanks
(70 × 30 × 30 cm), each divided lengthwise with either an opaque
white perspex partition or a transparent perspex partition to create
2 long lanes (see Supplementary Figure S1). Each lane was lined
with gravel in a slope ranging from a deep (15 × 15 cm; 14 cm depth)
“safe area” that contained an artificial plant to an increasingly shal-
low “exposed” area (4-cm depth at the other side). Only when fish
had fully emerged from this safe area we defined them to be “out
of cover.” No food was provided during the trials and fish were
thus not rewarded for leaving cover. This setup reflects the ecologi-
cally relevant problem where fish can either rest in a safe place or
explore a risky area in search of food (analogous to the exposed
area where food is delivered in their holding compartments). Fish
prefer to spend time under cover but, even in the absence of food
in the experimental tank, keep making regular trips out of cover to
explore the exposed area. Since fish have dierent preferences for
the number and length of trips out of cover they make yet prefer
to synchronize their activities and shoal together, there is a con-
flict on the timing of leaving and returning to cover. We have used
this ecologically relevant setup to look at the emergence of leaders
and followers in a number of previous papers (e.g., Harcourt etal.
2009, 2010). The walls of the tank were covered by white perspex
to minimize any disturbances from outside the tank. When not run-
ning experiments, the water of the experimental tanks was oxygen-
ated with an air stone. HD video cameras (Camileo X100; Toshiba
Corporation) were used to record fish movements from a fixed posi-
tion above each tank.
Experimental procedure
We tested 4 batches of fish (N = 136 in total), each over a 7-day
cycle (November to December 2011 and November to -December
2012)and randomly selected 44 fish as focals, 44 as partner for the
“first pairing,” and 44 as partner for the “second pairing,”. Fish
were tested across 3 stages. We started by testing fish in the experi-
mental tank in isolation to quantify their boldness (“isolation stage”).
On day 1 and 2, each fish was put in 1 of 2 lanes of the experi-
mental tank that were separated by an opaque partition so that fish
could not interact with each other. The behavior of each fish was
recorded for an hour each day. After a rest day, we randomly paired
each focal fish with a partner (“previous pairing stage”), and put the
2 fish in the same experimental tank, but this time with a transpar-
ent partition so that they could interact. Behavior was recorded for
an hour on each of 2 consecutive days. Finally, we paired each focal
Page 2 of 7
at Cambridge University Library on August 22, 2014http://beheco.oxfordjournals.org/Downloaded from

Jolles etal. • Previous social experience aects leadership behavior
with a new socially naive partner and observed their behavior for
another two 1-h sessions over 2 consecutive days (“current pairing
stage”). On each testing day, fish were transferred to the deep end of
the tank using a dip net and allowed to acclimatize to the tanks for
7 min before we tracked their movements. After each trial, fish were
moved back to their housing compartment. For each experimental
cycle, we randomized the daily testing order as well as the assign-
ment to tank and to the left and right lanes of a tank. Fish were
housed in their individual compartments for a week before their first
pairing to minimize any social experiences from being housed with
conspecifics in social housing tanks.
Data analysis
We tracked the exact movements of the fish at 10 frames/s using
automated motion tracking software (AnTracks, version 0.99). For
tracking we used a background subtraction acquisition method
that determined what pixels diered between the video and a
background image that was created from a random 5-min period
in each 1-h recording. As processing parameters we used gauss
subtraction, gauss blur, dilate and final thresholding for which we
adjusted the levels according to the specific light levels in each video
to ensure fish movements were tracked correctly. After tracking was
complete we checked all trajectories for each video. Any possible
noise tracked by the software was eliminated and discontinuous
trips where the software had lost track of the fish for a few frames
were joined.
Data were analyzed in R 3.0.2 (R Development Core Team,
2013). Based on the positional coordinates of both members in a
pair we calculated the relative time fish spent out of cover and their
number of trips out of cover. On average, fish spent 12.89% of
the time out of over (range 0–62.3%) and were consistent in this
proportion of time out of cover across the 2days of the isolation
stage (rs=0.55, N=136, P<0.001). Therefore, we used the aver-
age proportion of time individuals were out of cover across both
days as the boldness score for each fish, an approach commonly
adopted for examining the boldness personality trait (e.g., Harcourt
et al. 2009; Magnhagen and Bunnefeld 2009; King et al. 2013).
Ten fish that did not come out of cover during the isolation stage
were excluded.
The behavior of pairs of fish in a similar setup but without
previous experience has been described in detailed in previous
work (Harcourt et al. 2009; Nakayama, Harcourt, et al. 2012;
Nakayama, Johnstone, et al. 2012; Nakayama etal. 2013). In this
paper, we focus on the eect of previous experience (the first pair-
ing) on later interactions (the second paring). We focused on the
proportion of time spent out of cover by the focal fish, and on the
number of trips it made out of cover on its own, as a leader, and
as a follower. Leading was defined as a fish going out of cover and
being joined by its partner; following as a fish going out of cover
to join its partner that is already out. We considered the eects on
leading and following behavior separately as previous work has
shown that dierent factors (e.g., success of a partner in finding
food) may aect the tendencies to lead and follow in dierent ways
(e.g., Nakayama etal. 2013). For each of the 4 variables (time out
of cover, and the 3 types of trips), we used linear models with the
focal fish own boldness, the boldness of the previous partner, and
that of the current partner as predictors. We started with full mod-
els with all the predictors and obtained a minimal model by back-
ward stepwise elimination (i.e., sequentially dropping terms until
all terms retained in the model were significant). Statistics for non-
significant terms were obtained by fitting the minimal model with
each nonsignificant term added individually. As previous work has
shown that the relative personality between partners is a key predic-
tor of collective movements and leadership (Harcourt etal. 2009;
Nakayama et al. 2013), we ran separate models for focals bolder
than their second partner (bold focals) and focals relatively shyer
than their second partner (shy focals). Results based on the absolute
boldness scores were qualitatively similar and are documented in
the Supplementary Material. As our dataset consists of batches in
2 subsequent years, we additionally ran all models with year as an
extra fixed factor and found it had no significant eect in any of
the models. The residuals for all models were visually inspected to
ensure homogeneity of variance, normality of error, and linearity.
Finally, paired t-tests were used to investigate how the risk-taking
behavior of bold and shy focals changed across the isolation and
2 pairing stages. Repeatability across the 6 days of the experiment
was estimated using the intra-class correlation coecient (ICC),
following the method by Lessells and Boag (1987). All results with
0.10> P > 0.05 are reported as trends and P ≤ 0.05 as significant.
Means are quoted ± standard error (SE) throughout.
RESULTS
We focus on the data collected during the second pairing and inves-
tigate how the personalities of the previous and current partner
aect the behavior of focal fish bolder than their current partner
(bold focals) and focal fish shyer than their current partner (shy
focals). The relative boldness of focal fish ranged from −0.62 for
shy focals to +0.50 for bold focals (mean ± SE=−0.09 ± 0.03).
Time spent out ofcover
Bold focals spent more time out of cover the bolder they were
themselves (Figure 1A) but also the shyer their previous partner
had been (F2,7 = 18.77, P = 0.002; Table 1), together explaining
more than 80% of the variance (R2 = 0.84). The personality of
their current partner had no eect on the time bold focals were
out of cover (F1,7= 0.04, P= 0.84). By contrast, shy focals tended
to spend more time out of cover the bolder their current partner
(F1,22 = 4.11, P = 0.055; R2 = 0.16), while their own personality
(F1,22=0.33, P=0.571; Figure1B) and that of their previous part-
ner (F1,22=0.04, P=0.845) had no significant eect.
Number oftrips
The number of solo trips, when focal fish went out and returned
to cover without being followed by their partner, was relatively
higher in bold compared with shy focals (t = 2.56, P = 0.028;
13.8 ± 4.24 and 2.65 ± 1.02 trips, respectively). Bold focals went
on more solo trips the bolder they were themselves (F1,8= 6.60,
P= 0.033; R2= 0.45) while the personality of the current partner
and the previous partner had no eect on this behavior (F1,7=0.09,
P=0.777; F1,7=1.67, P=0.237 respectively; Table 1). The num-
ber of solo trips made by shy focals was not explained by either
their own personality (F1,22=0.17, P=0.687), that of their current
partner (F1,22=0.25, P=0.624) or that of their previous partner
(F1,22=0.11, P=0.743).
There was no significant dierence in the number of joint trips
led by bold and shy focal fish during the second pairing (t= 1.31,
P=0.211; 7.15 ± 2.14 and 3.96 ± 1.19 trips, respectively). Bold focals
led more trips the relatively bolder the focal individual (Figure 2A)
but also the shyer their previous partner (F2,7 = 12.98, P = 0.004;
Figure2B; Table1), together explaining 79% of the variance. The per-
sonality of the current partner did not aect the number of leadership
Page 3 of 7
at Cambridge University Library on August 22, 2014http://beheco.oxfordjournals.org/Downloaded from

Behavioral Ecology
trips for bold focals (F1,6=1.53, P=0.262). By contrast, shy focals led
more trips the bolder their current partner (F1,22= 5.75, P = 0.025;
R2= 0.21), while their own personality (F2,21 = 0.28, P = 0.600) and
that of their previous partner (F2,21=0.13, P=0.719) had no eect.
There was no dierence in the number of trips bold and
shy focals followed their current partner out of cover (t = 1.91,
P = 0.083; 7.15 ± 2.14 and 2.91 ± 2.45 trips, respectively). Bold
focals followed their partner more the bolder it was (Figure3) and
the shyer their previous partner had been (F2,7=41.74, P <0.001;
Table1), together explaining 92% of the variance. Bold focal’s own
personality did not play a role (F1,6= 2.28, P = 0.182). Shy focals
followed more the bolder their current partner was (F1,21= 7.78,
P = 0.011; Figure 3; R2 = 0.26), with no eect of their own per-
sonality (F1,21=0.60, P=0.448) and that of their previous partner
(F1,21=0.39, P=0.537).
Behavioral consistency across thestages
Bold focals were highly repeatable in the time they spent out of cover
on the 6days across the 3 stages (ICC=0.76, 95% confidence inter-
val [CI]: 0.55–0.92), whereas shy focals were not (ICC=0.17, 95%
CI: 0.05–0.35). On average, bold focals spent similar amounts of time
out of cover during the isolation stage and the first pairing (t9=1.81,
P=0.104) but tended to spend less time out of cover during the sec-
ond pairing compared with the isolation stage (t9=2.18, P=0.058).
By contrast, shy focals spent more time out of cover when they
could see their partner compared with when in isolation (first pair-
ing: t31=−2.29, P=0.029; second pairing: t31=−2.62, P= 0.013).
Additionally, looking at focals based on their absolute boldness cat-
egory (with bold fish spending more time out and shy fish less time
out than the average focal fish) we found bold fish (ICC=0.65, 95 %
CI: 0.47–0.82) were more consistent than shy fish (ICC=0.17, 95 %
CI: 0.04–0.39), as reflected by their nonoverlapping CIs.
DISCUSSION
In this study, we show for the first time that the eect of the per-
sonality of a previous social partner can carry over to later social
interactions, modulating the willingness of individuals to go out of
cover and lead their partner. By contrast, the tendency to follow
was mainly aected by the personality of an individual’s current
partner. Although bolder fish were more consistent than shyer fish
in the time they spent out of cover across the contexts, it was only
bold fish that were susceptible to social reinforcement by their pre-
vious social interactions. Shyer fish behaved much more flexibly
and responded most strongly to their current partner.
Previously, some studies have shown that previous social experi-
ence may aect neophobia and aggression (Hsu and Wolf 1999;
Frost etal. 2007) and that experience within the same pair may
override personality dierences in leadership tendencies (Nakayama
etal. 2013). Here we show for the first time how social experiences
with previous partners may aect later leadership behavior: the
bolder their previous partner, the relatively less time bold focals
spent out of cover, making them less successful in taking the lead.
These findings help answer the important question in the leader-
ship literature of what makes an initiator successful in triggering
collective movement (Petit and Bon 2010). Although bolder indi-
viduals are less sensitive to failure in recruiting a partner, they are
responsive to their partner’s behavior when it has taken on the
role of leader. This may be especially the case when bold focals
are paired with a relatively bold partner. In such a situation, bold
focals partner is relatively more likely to take the lead compared
with a shyer partner. Consequently, the focal individual may be less
likely to be followed, resulting in a reduction of positive feedback in
leadership and reduced performance in the pair (Nakayama etal.
2013). Such experience may then subsequently modulate focal fish’
willingness to go out of cover and lead their partner. Not only does
this finding highlight that bolder individuals may be more suscepti-
ble to social reinforcement than shy individuals, it indicates that for
leadership social experience is important. To be an eective leader,
an individual may need experience with good followers, provid-
ing positive social feedback and leading experience, and ultimately
more successful leadership. These findings may have potential for
our understanding of human leadership as a lack of knowledge of
the social dynamics underlying leadership has been highlighted in
the social sciences (Amit et al. 2009; Emery 2010; DuBrin 2013).
A
0.0
0.1
0.2
0.3
0.4
0.5
0.2 0.3 0.4 0.5 0.6 0.7 0.8
Boldness score bold focals
Proportion of time out of cover
B
0.0
0.1
0.2
0.3
0.4
0.5
0.10.2 0.30.4 0.5
Boldness score shy focals
Proportion of time out of cover
Figure1
The proportion of time focal fish spent out of cover during the current pairing was (A) positively correlated with the boldness scores of bold focals (N=10),
but (B) not significantly correlated with the boldness scores of shy focals (N=24). Boldness scores were square root transformed.
Page 4 of 7
at Cambridge University Library on August 22, 2014http://beheco.oxfordjournals.org/Downloaded from

Jolles etal. • Previous social experience aects leadership behavior
Future studies could look in more detail at the extent of the dier-
ence in personality scores between the partners and determine the
eect it may have on collective behavior.
The finding that bold but not shy focals were aected by a previ-
ous partner might be explained by the fact that shyer individuals
are in general more sociable (Ward et al. 2004; Pike etal. 2008)
and behaviorally less consistent (Nakayama, Johnstone, etal. 2012),
than bold individuals. Indeed, we found that the current partner’s
personality explains much more of shy focals’ behavior than that
of bold focals, which is in line with previous studies reporting that
shy individuals are more responsive to the actions of their (current)
group members (Pike etal. 2008; Nakayama, Harcourt, etal. 2012;
Nakayama, Johnstone, etal. 2012). This may also explain the more
general finding that shy but not bold focals spent considerably more
A
0
1
2
3
4
5
0.2 0.3 0.4 0.5 0.6 0.7
Boldness score bold focal
Nr of trips leading
B
−1
0
1
2
3
0.00.1 0.20.3 0.40.5
Boldness score previous partner
Residuals
Figure2
The number of trips bold focals (N=10) initiated and were joined by their partner during the current pairing was (A) positively related to their own boldness
score and (B) negatively related to the boldness score of their previous partner. The y axis of plot B shows residuals of the model on leading trips with focal
boldness score as the only factor. Scores above 0 indicate individuals were joined more than may be expected based on their own boldness score and scores
below 0 individuals were joined less than may be expected based on their boldness score. Boldness scores and number of leading trips were square root
transformed.
Table1
Linear models of proportion of time out, number of solo trips, number of led trips, and number of trips followed by bold and shy
focal fish
Bold focals Shy focals
Estimate SE F P Estimate SE F P
Proportion of time out
Constant 0.10 0.10 0.391 0.05 0.10 0.662
Personality focal 0.80 0.16 23.80 0.002 0.03 0.37 0.01 0.939
Personality current partner 0.07 0.32 0.04 0.844 0.45 0.22 4.11 0.055
Personality previous partner −0.46 0.19 6.01 0.044 −0.03 0.15 0.04 0.847
Number of solo trips
Constant −0.05 1.33 0.974 1.16 0.24 <0.001
Personality focal 6.61 2.57 6.60 0.033 −1.01 2.47 0.17 0.687
Personality current partner −1.48 5.04 0.08 0.777 0.78 1.57 0.25 0.624
Personality previous partner −3.72 2.88 1.67 0.237 0.35 1.05 0.11 0.743
Number of led trips
Constant 1.18 0.78 0.174 −0.09 0.72 0.899
Personality focal 4.28 1.23 12.13 0.010 −1.37 2.56 0.28 0.600
Personality current partner 2.62 2.12 1.53 0.262 3.71 1.55 5.75 0.025
Personality previous partner −3.88 1.39 7.79 0.027 0.38 1.05 0.13 0.719
Number of followed trips
Constant 1.05 0.47 0.062 −0.08 0.72 0.917
Personality focal 1.63 1.08 1.51 0.182 −1.96 2.53 0.60 0.447
Personality current partner 11.73 1.41 68.84 <0.001 4.30 1.54 7.78 0.010
Personality previous partner −2.93 1.06 7.51 0.029 0.65 1.04 0.39 0.537
These analyses looked at focals that were bolder than their final partner (N=10) and focals that were shyer than their final partner (N= 24). Statistics for sig-
nificant terms, shown in bold, were derived from the minimal model containing only significant terms, whereas statistics for nonsignificant terms were obtained
by running the minimal model with the term added individually. Coecient estimates represent the change in the dependent variable relative to the baseline
category and can therefore be interpreted as measures of eect size. All personality scores and response variables were square root transformed.
Page 5 of 7
at Cambridge University Library on August 22, 2014http://beheco.oxfordjournals.org/Downloaded from

Behavioral Ecology
time out of cover when there was a conspecific present compared
with when they were in isolation. Interestingly, in contrast to the
time spent out of cover and leading behavior, following behavior of
both bold and shy focals was primarily explained by the boldness of
their current partner. This result is in line with a number of recent
studies that have shown that both bold and shy individuals are
responsive when in the following position (Nakayama, Harcourt,
etal. 2012; Nakayama, Johnstone, etal. 2012) and that experience
may override personality dierences in the tendency to follow but
not in the tendency to lead (Nakayama etal. 2013). Together, these
findings thus suggest that regardless of an individual’s own person-
ality, its tendency to follow mainly depends on the behavior of its
current partner(s). Leadership, in contrast, is particularly dependent
on a bolder personality type, with a modifying eect of social feed-
back from previous experiences.
Overall, our findings demonstrate a general dierence in respon-
siveness between shy and bold individuals. Although both bold
and shy individuals adjusted their behavior, bold individuals were
more consistent in their behavior than shy individuals but adjusted
their behavior based on their previous partner, suggestive of social
reinforcement. In contrast, shy individuals mostly adjusted their
behavior based on their current partner. These results support 2
recent theoretical models that showed how a coevolutionary process
between responsiveness and consistency may eventually result in
populations that consist of highly responsive individuals that follow
and behaviorally consistent individuals that mainly lead (Johnstone
and Manica 2011; Wolf etal. 2011). Furthermore, these findings are
highly relevant in the light of the idea that individual dierences can
be seen as behavioral specializations (Dall etal. 2012). If individuals
dier in the extent that they change their behavior based on previ-
ous and current experiences, this represents a potential mechanism
through which social roles can be generated and reinforced to create
even longer lasting dierences between individuals. In other words,
personality dierences may be maintained in populations because
of their role in social coordination (see also King etal. 2009).
While the study of collective behavior, from pairs of individuals
to groups of thousands of individuals, was initially mostly focused
on homogeneous interaction rules (Couzin and Krause 2003; Petit
and Bon 2010; Vicsek and Zafeiris 2012), individual dierences
are increasingly taken into account when examining group behav-
ior (Conradt and List 2009; Herbert-Read etal. 2012; Jolles etal.
2013). Here we go one step further by showing that social dynam-
ics across time and social contexts may have a considerable eect
on individual and thereby group behavior. Our study is the first
to demonstrate that leadership roles are aected by social experi-
ences from previous partners and that this depends on an individ-
ual’s personality, with bold but not shy fish being aected by the
personality of a previous partner. These findings help understand
how leading and following behavior emerge and are maintained
and highlight the important influence current as well as previous
social experiences can have on individual and collective behavior.
SUPPLEMENTARY MATERIAL
Supplementary material can be found at http://www.beheco.
oxfordjournals.org/
FUNDING
This study was supported by a BBSRC scholarship to J.W.J.and a
fellowship from the Japan Society for the Promotion of Science to
S.N.
We thank 2 anonymous referees for their helpful feedback, N.Boogert for
her comments on a previous version of this paper, and B. Taylor for fish
husbandry. Animal care and experimental procedures were approved by the
Animal Users Management Committee of the University of Cambridge
under a nonregulated procedures regime.
Handling editor: Alison Bell
REFERENCES
Amit K, Popper M, Gal R, Mamane-Levy T, Lisak A. 2009. Leadership-
shaping experiences: a comparative study of leaders and non-leaders.
Leadersh Organ Dev J. 30:302–318.
Barelli C, Boesch C, Heistermann M, Reichard UH. 2008. Female white-
handed gibbons (Hylobates lar) lead group movements and have priority of
access to food resources. Behaviour. 145:965–981.
Beauchamp G. 2000. Individual dierences in activity and explora-
tion influence leadership in pairs of foraging zebra finches. Behaviour.
137:301–314.
Bell AM. 2005. Behavioural dierences between individuals and two popu-
lations of stickleback (Gasterosteus aculeatus). J Evol Biol. 18:464–473.
Borg B, Bornestaf C, Hellqvist A. 2004. Mechanisms in the photoperiodic
control of reproduction in the Stickleback. Behaviour. 141:1521–1530.
Brown C, Irving E. 2014. Individual personality traits influence group
exploration in a feral guppy population. Behav Ecol. 25:95–101.
Conradt L, List C. 2009. Group decisions in humans and animals: a survey.
Philos Trans R Soc B. 364:719–742.
Conradt L, Roper TJ. 2003. Group decision-making in animals. Nature.
421:155–158.
Conradt L, Roper TJ. 2009. Conflicts of interest and the evolution of deci-
sion sharing. Philos Trans R Soc B. 364:807–819.
Couzin ID, Krause J, Franks NR, Levin SA. 2005. Eective leadership and
decision-making in animal groups on the move. Nature. 433:513–516.
Couzin ID, Krause J. 2003. Self-organization and collective behavior in ver-
tebrates. Adv Study Behav. 32:1–75.
Dall SRX, Bell AM, Bolnick DI, Ratnieks FLW, Sih A. 2012. An evolution-
ary ecology of individual dierences. Ecol Lett. 15:1189–1198.
0
1
2
3
4
5
6
0.0 0.1 0.2 0.3 0.40.5 0.60.7 0.8
Boldness score current partner
Number of trips following
Figure3
The number of trips focals went out of cover and joined their partner
(following) during the current pairing is positively related to the boldness
score of the current partner, both for bold focals (N = 10, closed circles)
and shy focals (N= 24, closed triangles). Boldness scores and the number
of following trips were square root transformed. Two data points of the
bold focals overlap at the origin because they both never followed and had a
partner with a boldness score of 0.
Page 6 of 7
at Cambridge University Library on August 22, 2014http://beheco.oxfordjournals.org/Downloaded from

Jolles etal. • Previous social experience aects leadership behavior
DuBrin AJ. 2013. Leadership: research findings, practice, and skills. Mason:
South-Western, Cengage Learning.
Dyer JRG, Johansson A, Helbing D, Couzin ID, Krause J. 2009. Leadership,
consensus decision making and collective behaviour in humans. Philos.
Trans R Soc B. 364:781–789.
Emery C. 2010. Investigating leadership emergence using longitudinal lead-
ership networks. PhD Thesis, Università della Svizzera italiana.
Flack A, Pettit B, Freeman R, Guilford T, Biro D. 2012. What are leaders
made of ? The role of individual experience in determining leader–fol-
lower relations in homing pigeons. Anim Behav. 83:703–709.
Frost AJ, Winrow-Gien A, Ashley PJ, Sneddon LU. 2007. Plasticity in ani-
mal personality traits: does prior experience alter the degree of boldness?
Proc R Soc B. 274:333–339.
Harcourt JL, Ang TZ, Sweetman G, Johnstone RA, Manica A. 2009. Social
feedback and the emergence of leaders and followers. Curr Biol. 19:248–252.
Harcourt JL, Sweetman G, Manica A, Johnstone RA. 2010. Pairs of fish
resolve conflicts over coordinated movement by taking turns. Curr Biol.
20:156–160.
Herbert-Read JE, Krause S, Morrell LJ, Schaerf TM, Krause J, Ward AJW.
2012. The role of individuality in collective group movement. Proc R Soc
B. 280:20122564.
Hsu Y, Wolf L. 1999. The winner and loser eect: integrating multiple
experiences. Anim Behav. 57:903–910.
Jacobs A, Sueur C, Deneubourg JL, Petit O. 2011. Social network influ-
ences decision making during collective movements in Brown Lemurs
(Eulemur fulvus fulvus). Int J Primatol. 32:721–736.
Johnstone RA, Manica A. 2011. Evolution of personality dierences in
leadership. Proc Natl Acad Sci. 108:8373–8378.
Jolles JW, King AJ, Manica A, Thornton A. 2013. Heterogeneous structure
in mixed-species corvid flocks in flight. Anim Behav. 85:743–750.
Jolles JW, Ostojić L, Clayton NS. 2013. Dominance, pair bonds and bold-
ness determine social-foraging tactics in rooks, Corvus frugilegus. Anim
Behav. 85:1261–1269.
King AJ, Douglas CMS, Huchard E, Isaac NJB, Cowlishaw G. 2008.
Dominance and aliation mediate despotism in a social primate. Curr
Biol. 18:1833–1838.
King AJ, Fürtbauer I, Mamuneas D, James C, Manica A. 2013. Sex-
dierences and temporal consistency in stickleback fish boldness. PLoS
One. 8:e81116.
King AJ, Johnson DDP, Van Vugt M. 2009. The origins and evolution of
leadership. Curr Biol. 19:R911–916.
Krause J, Reeves P, Hoare D. 1998. Positioning behaviour in Roach shoals:
the role of body length and nutritional state. Behaviour. 135:1031–1039.
Krause J, Ruxton GD. 2002. Living in groups. Oxford: Oxford University Press.
Kurvers RHJM, Eijkelenkamp B, van Oers K, van Lith B, van Wieren SE,
Ydenberg RC, Prins HHT. 2009. Personality dierences explain leader-
ship in barnacle geese. Anim Behav. 78:447–453.
Lessells CM, Boag PT. 1987. Unrepeatable repeatabilities: a common mis-
take. Auk. 104:116–121.
Magnhagen C, Bunnefeld N. 2009. Express your personality or go along
with the group: what determines the behaviour of shoaling perch? Proc
R Soc B. 276:3369–3375.
McClure M, Ralph M, Despland E. 2011. Group leadership depends on
energetic state in a nomadic collective foraging caterpillar. Behav Ecol
Sociobiol. 65:1573–1579.
Nagy M, Akos Z, Biro D, Vicsek T. 2010. Hierarchical group dynamics in
pigeon flocks. Nature. 464:890–893.
Nakayama S, Harcourt JL, Johnstone RA, Manica A. 2012. Initiative,
personality and leadership in pairs of foraging fish. PLoS One.
7:e36606.
Nakayama S, Johnstone RA, Manica A. 2012. Temperament and hunger
interact to determine the emergence of leaders in pairs of foraging fish.
PLoS One. 7:e43747.
Nakayama S, Stumpe MC, Manica A, Johnstone RA. 2013. Experience
overrides personality dierences in the tendency to follow but not in the
tendency to lead. Proc R Soc B. 280:20131724.
Peterson R, Jacobs A. 2002. Leadership behavior in relation to domi-
nance and reproductive status in gray wolves, Canis lupus. Can J Zool.
1412:1405–1412.
Petit O, Bon R. 2010. Decision-making processes: the case of collective
movements. Behav Processes. 84:635–647.
Pettit B, Perna A, Biro D, Sumpter DJT. 2013. Interaction rules underlying
group decisions in homing pigeons. J R Soc Interface. 10:20130529.
Pike TW, Samanta M, Lindström J, Royle NJ. 2008. Behavioural phe-
notype aects social interactions in an animal network. Proc R Soc B
275:2515–2520.
R Development Core Team. 2013. R: A language and environ-
ment for statistical computing. Vienna: R Foundation for Statistical
Computing.
Réale D, Festa-Bianchet M. 2003. Predator-induced natural selection on
temperament in bighorn ewes. Anim Behav. 65:463–470.
Reebs S. 2001. Influence of body size on leadership in shoals of golden
shiners, Notemigonus crysoleucas. Behaviour. 138:797–809.
Reebs SG. 2000. Can a minority of informed leaders determine the forag-
ing movements of a fish shoal? Anim Behav. 59:403–409.
Sueur C, Petit O. 2008. Shared or unshared consensus decision in
macaques? Behav Processes. 78:84–92.
Vicsek T, Zafeiris A. 2012. Collective motion. Phys Rep. 517:71–140.
Ward AJW, Herbert-Read JE, Jordan LA, James R, Krause J, Ma Q ,
Rubenstein DI, Sumpter DJT, Morrell LJ. 2013. Initiators, leaders, and
recruitment mechanisms in the collective movements of damselfish. Am
Nat. 181:748–760.
Ward AJW, Thomas P, Hart PJB, Krause J. 2004. Correlates of boldness in
three-spined sticklebacks (Gasterosteus aculeatus). Behav Ecol Sociobiol.
55:561–568.
Wolf M, Van Doorn GS, Weissing FJ. 2011. On the coevolution of
social responsiveness and behavioural consistency. Proc R Soc B.
278:440–448.
Page 7 of 7
at Cambridge University Library on August 22, 2014http://beheco.oxfordjournals.org/Downloaded from