Shared or unshared consensus decision in macaques?

Available online at www.sciencedirect.com
Behavioural Processes 78 (2008) 84–92
Shared or unshared consensus dec
a,b t a,b
hysio
ouis P
cemb
Abstract
Members onise
members ca one
the decision -ma
investigate t sed
and the rhes cision
individuals i oup
individuals kean
move, where sus d
might be lin
© 2008 Else
Keywords: C
1. Introduction
Animals
of resource
dent upon
decisions
nise their a
maintainin
living—pro
tion exchan
species, ind
nity throug
Consensus
the mechan
of the vari
(Conradt a
the most d
but such pr
∗ Correspon
23 rue Becqu
fax: +33 3881
E-mail ad
remain limited to a few pioneering studies (Boinski and Garber,
2000; Black, 1988; Kummer, 1968; Leca et al., 2003; Prins,
0376-6357/$
doi:10.1016/jroutinely face decisions regarding the exploitation
s. In social species, group cohesiveness is depen-
the ability of group members to take “collective”
(Conradt and Roper, 2005). Individuals synchro-
ctivities, forage collectively and move together. By
g cohesion they preserve the advantages of group
tection from predators, defence of food, informa-
ge (Alexander, 1974; Danchin et al., 2005). In many
ividuals reach decisions involving the entire commu-
h a “consensus” process (Conradt and Roper, 2005).
decision-making raises crucial questions regarding
isms of communication, the relative social influence
ous group members and their capacity to cooperate
nd Roper, 2005). Studying groups on the move is
irect way to investigate consensus decision-making,
ocesses remain largely unexplored and current data
ding author at: Ethologie des Primates, DEPE, IPHC, UMR7178,
erel, 67087 Strasbourg, France. Tel.: +33 388107457;
07456.
dress: odile.petit@c-strasbourg.fr (O. Petit).
1996; Reinhardt, 1983; Scott, 1945).
In the literature, the concept of a single group leader is
widespread (social carnivores: Holekamp et al., 2000; horses:
Feist and Mc Cullough, 1976; Tyler, 1972; primates: Boinski
and Garber, 2000; Watanabe and Brotoisworo, 1982). In such
a “consistent leadership” (Conradt and Roper, 2005) or “indi-
vidual leadership” (Leca et al., 2003), the leader is frequently
the most socially dominant animal in the group. This is typi-
cally the case in mountain gorillas (Gorilla beringei beringei),
where the silverback male heads in the direction of the future
movement at the time of departure, walking quickly, followed
by other members of the group (Watts, 2000). A similar process
occurs in wolves (Canis lupus, Mech, 1970) and mongooses
(Helogale parvula, Rasa, 1983). However, this is not the only
type of leadership encountered in animals. The traditional exam-
ple of the hamadryas baboons (Papio hamadryas) illustrates
how the departure of a troop can be preceded by a complex
consensus or process of negotiation (Kummer, 1968). Such “vot-
ing” behaviour has also been described in bar-headed geese
(Anser indicus, leaving calls; Lamprecht, 1992), African ele-
phants (Loxodonta Africana, vocalisations; Poole et al., 1988)
– see front matter © 2008 Elsevier B.V. All rights reserved.
.beproc.2008.01.004C. Sueur , O. Peti
a Ethologie des Primates, De´partement d’Ecologie, P
b Centre de Primatologie de l’Universite´ L
Received 18 September 2007; received in revised form 13 De
of a social group have to make collective decisions in order to synchr
n take part in the decision whereas in an unshared consensus decision,
for the rest of the group. It has been suggested that the type of decision
his further, we studied collective movements in two species with oppo
us macaque (Macaca mulatta). From our results, it appears that the de
n both groups. However, this consensus decision involved nearly all gr
took a prominent role in rhesus macaques. Thus, we suggest that Ton
as in the same context rhesus macaque exhibit partially shared consen
ked to the different social systems of the two studied species.
vier B.V. All rights reserved.
ollective movement; Decision-making; Leadership; Social styleision in macaques?
,∗
logie et Ethologie, France
asteur, France
er 2007; accepted 7 January 2008
their activities. In a shared consensus decision, all group
individual, usually a dominant member of the group, takes
king of a species could be influenced by its social style. To
social systems, the Tonkean macaque (Macaca tonkeana)
to move is the result of the choices and actions of several
members in Tonkean macaques whereas dominant and old
macaques display equally shared consensus decisions to
ecisions. Such a difference in making a collective decision

C. Sueur, O. Petit / Behavioural Processes 78 (2008) 84–92 85
and African buffalos (Syncerus caffer, postures; Prins, 1996).
Movement is undertaken in the direction chosen by the great-
est number
leadership”
on differen
a recent ex
capuchins
group mem
2003).
These tw
cesses of de
all member
individual,
and Roper,
versely, a
members o
leadership
it is unclea
sus. From
capuchins,
nas (Crocu
taking a co
to have an
flicts (Hofe
and/or coo
and Marten
al., 1995).
of the grou
process (H
to a shared
The pre
played by
macaques
movement.
system wo
decisions. T
metry of do
and are con
In this con
a collective
process, sim
sus macaqu
system wh
food (Petit
submissive
Thus, we ex
tred on the
would take
2. Materia
2.1. Subjec
The gro
Primatolog
semi-natur
matrilines a
oldest female of the Tonkean macaque group. At the time of the
study on Tonkean macaques (December 2003 to April 2004), the
was
, 9, 7
14,
juv
ues (
ndiv
s (16
ult fe
mpo
(Su
Eac
assy
and
ere d
s.
bser
h gr
twe
ril 2
o 21
e co
at a p
peri
1; P
17)
y at
two
perio
= Nf
= Nf
h co
ed su
ark
h an
con
than
g zon
nks
beh
rece
ng m
depa
0 s. T
his
03) i
in m
2005
ure o
is at
indo
hesio
70 s;
o folof individuals. These scenarii refer to a “variable
, where different group members lead group actions
t occasions (Conradt and Roper, 2005). Similarly,
perimental study on semi-free ranging white-faced
(Cebus capucinus) showed that the majority of the
bers can initiate a collective movement (Leca et al.,
o kinds of leadership are related to two different pro-
cision-making. In an “unshared consensus” process,
s of the group abide by the decision made by a single
usually the dominant or “individual leader” (Conradt
2005) which refers to a consistent leadership. Con-
“shared consensus” decision is taken by all group
r a particular sub-group and refers to the variable
proposed by Conradt and Roper (2005). However,
r why a species displays a certain type of consen-
these definitions, we can consider that white-faced
Cape hunting dogs (Lycaon pictus) and spotted hye-
ta crocuta) use a process of shared consensus when
llective decision. Moreover, these species are known
“open social system”, as they negotiate during con-
r and East, 2000; Leca et al., 2002; Perry et al., 2003a)
perate in several activities (Ku¨hme, 1965; Malcolm
, 1982; Perry et al., 2003b; Schaller, 1972; Smale et
Their social styles may probably allow each member
p to express its choice during the decision-making
olekamp et al., 2000; Leca et al., 2003) and thus lead
consensus.
sent study investigates the type of consensus dis-
Tonkean macaques (Macaca tonkeana) and rhesus
(Macaca mulatta) when they make decisions about
We investigated whether in both species, the social
uld predict the type of consensus behind collective
onkean macaques are characterised by a low asym-
minance, a great tolerance between group members
sidered as a highly tolerant species (Thierry, 2004).
text, we might expect that Tonkean macaques reach
decision about moving through a shared consensus
ilarly to capuchin monkeys. On the other hand, rhe-
es have a strictly hierarchical and nepotistic social
ereby dominant individuals monopolise access to
et al., 1992) and subordinate individuals display
behaviours towards dominant ones (Thierry, 2004).
pect that the decision-making process would be cen-
dominant male, i.e. the intolerant rhesus macaques
decisions through an unshared consensus.
l and methods
ts and environment
ups under investigation were bred in the Centre of
y at the University Louis Pasteur, Strasbourg, in
al conditions. The two groups were composed of two
nd all members were born in captivity apart from the
group
15, 11
21, 21,
and six
macaq
of 22 i
female
sub-ad
The co
groups
1978).
and gr
pellets
bles w
session
2.2. O
Bot
day be
13 Ap
2006 t
that th
occur
16:00,
d.f. = 1
P = 0.9
activit
on the
texts,
Nsocial
Nsocial
Eac
analys
each p
of eac
sexual
more
startin
Tha
as any
5 min p
sampli
as the
than 4
ual”. T
al. (20
capuch
et al. (
depart
and th
time w
the ad
max =
starts tcomposed of 22 individuals: seven adult males (27,
, 6 and 5 years old), eight adult females (33, 25, 24,
12 and 11 years old), one sub-adult male (4 years old)
eniles (1-year-old). At the time of the study on rhesus
May 2006 to August 2006), the group was composed
iduals: two adult males (17 and 8 years old), 11 adult
, 14, 12, 11, 11, 11, 8, 7, 7, 7 and 6 years old), two
males (both 4 years old) and 7 juveniles (1-year-old).
sition of both groups is comparable to those of wild
priatna et al., 1992; Whitten et al., 1987; Makwana,
h group lived in a 0.5-ha park including trees, bushes
areas. They had an inside shelter where commercial
water were provided ad libitum. Fruits and vegeta-
istributed once per week, outside of the observation
vational protocol
oups were observed and filmed for four hours per
en 10:00 and 16:00, from 29 December 2003 to
004 for the Tonkean macaques and from 27 May
August 2006 for the rhesus macaques. We checked
llective movements of the observed groups did not
articular period during the day (between 10:00 and
od of one half-hour; Tonkean macaques: Chi2 = 6.06,
= 0.863; rhesus macaques: Chi2 = 5.47, d.f. = 11;
and that the group did not engage in a particular
a specific time each day (Kolmogorov–Smirnov test
observed distributions of social and foraging con-
d of one half-hour; Tonkean macaques: z = 4.17,
oraging = 12, P = 0.249; rhesus macaques: z = 3.22,
oraging = 12, P = 0.487).
llective movement was recorded on video tapes and
ch that all related events were logged. We mapped
with reference marks and we recorded the position
imal (±1 m). Movements occurring in agonistic or
texts were discarded. Only start attempts for which
10 group members were present at or around the
e were taken into account.
to the video scoring, we recorded the identity as well
aviour displayed by every group member over the
ding a collective movement, using the all occurrence
ethod (Altmann, 1974). A start attempt was defined
rture of an individual walking more than 10 m in less
his individual was called the “first departed individ-
criterion was the same as the one used by Leca et
n their study on collective movements in a group of
onkeys and corresponded to the definition of Dumont
). If no other individual left within 5 min after the
f this first individual, the observation was stopped
tempt was then classified as a failed attempt (this
w was determined by the mean latency separating
n of two direct participants; mean = 21.81 ± 1.14 s;
where adhesion is the moment at which an individual
low). When the first departed individual was followed

86 C. Sueur, O. Petit / Behavioural Processes 78 (2008) 84–92
by at least one individual for more than 5 m, the progress of
the collective movement was observed using the focal-animal
sampling m
any individ
rior to 45◦
The positio
interaction
(pause, bac
of each par
ers) were re
observed o
2.3. Assess
To esta
used data fr
competitio
individuals
tion of avo
was found
al., 1993; P
2.4. Spatia
The foll
ture of the
(1) Group
when
less th
uals w
as “sub
distant
than 5
(2) Positio
positio
calcula
of the g
middle
departe
was eit
when i
2.5. Behav
The fol
before dep
movement.
observation
(1) Speed
its dep
(2) Back g
directi
of occu
(3) Pause:
The fr
recorded. A pause was qualified as a distinct event when
it was separated by more than 2 s from a preceding pause.
van
n of
fore
red.
tatist
lyse
s par
and
mac
nden
ete v
Poi
nden
hase
artur
ed b
ot c
onke
ues:
orrel
f ind
or th
est f
art a
e te
As.
rank
an h
as u
an ag
d as
le su
nal
ues
les w
d hie
ase
oux
SPSS
are
ults
hara
m
iffer
U =
onkea
rly,
elap
the
endethod (Altmann, 1974). A follower was defined as
ual walking in a direction that formed an angle infe-
with the direction of the first departed individual.
n, activity (foraging; resting, i.e. asleep, not in social
; socialising, i.e. grooming or playing) and behaviours
k glance, speed and advance; see definitions below)
ticipant (the first departed individual and all follow-
corded throughout the movement. Participants were
ne by one thanks to video scoring.
ing dominance rank
blish the dominance hierarchy of each group, we
om spontaneous aggressive interactions and drinking
n around a single source of orange juice. We ranked
over 1 year of age in a matrix according to the direc-
idances and unidirectional aggressions. Hierarchy
to be linear in both species (Matman; de Vries et
< 0.0001 for both species).
l patterns
owing patterns were recorded at the time of the depar-
first individual:
dispersion: the group was qualified as “grouped”
all individuals were separated from each other by
an 5 m, as “dispersed” when more than two individ-
ere more than 5 m apart from other individuals, and
-grouped” when group members formed sub-groups
of more than 5 m, in which individuals were not more
m apart.
n of the first departed individual: to determine the
n of the first departed individual prior to departure, we
ted the distance between the centre and the periphery
roup and divided this distance in three parts (centre,
and periphery). We defined the position of the first
d individual as “in-group” when the latter individual
her in the centre or in the middle and as “borderline”
t was in the periphery.
ioural units
lowing behaviours were recorded when exhibited
arture, at departure or during the progress of the
They were selected after a preliminary period of
s:
of each individual: in m s−1, during the first 10 m after
arture.
lance: the individual turns its head and looks in the
on of other group members, measured as frequency
rrence throughout the movement.
the individual stops moving during at least 2 s.
equency of pauses throughout the movement was
(4) Ad
tio
be
su
2.6. S
Ana
vidual
group
rhesus
indepe
a discr
with a
indepe
each p
at dep
acteris
were n
tion; T
macaq
rank c
rank o
tests f
of-fit t
new st
test. W
ANOV
chical
the me
dure w
the me
groupe
a sing
sible. A
macaq
juveni
sex, an
R (rele
Giraud
using
Means
3. Res
3.1. C
The
not d
test;
meanT
Simila
(time
from
at thece: the individual walks less than 5 m in the direc-
the collective movement. The frequency of advances
and after the departure of each individual was mea-
Any advance ended after a stop of 2 s.
ical analysis
s were based on 119 start attempts with 1–19 indi-
ticipating to the movement for the Tonkean macaque
on 131 start attempts with 1–22 participants for the
aque group. We considered each start attempt as an
t event. As the number of participants is a count, i.e.
ariable, we used a generalised linear model (GLM)
sson law. This dependent variable was tested with
t variables (spatial and behavioural variables) for
of a collective movement (i.e. before departure,
e and after departure). Each individual was char-
y its age, sex and hierarchical rank. Age and rank
orrelated in either species (Spearman rank correla-
an macaques: r = −0.05, N = 16, P = 0.849; rhesus
r = −0.07, N = 15, P = 0.807). We used Spearman
ations to test for the effect of age and hierarchical
ividuals on the dependent variable, Mann–Whitney
at of species and sex, and a Chi-square goodness-
or a specific analysis of age. Probabilities to make a
ttempt were compared using a Wilcoxon signed-rank
sted differences between individuals using one-way
We conducted sign-tests to compare the mean hierar-
of individuals displaying preliminary behaviours to
ierarchical rank of the whole group. The same proce-
sed to compare the mean age of these individuals to
e of the whole group. One-year-old individuals were
“juveniles” because they always moved together (as
b-group) and individual discrimination was impos-
yses were then based on 17 individuals for Tonkean
and 16 individuals for rhesus macaques. Therefore,
ere not included in analyses testing the influence of
rarchical rank. GLM analyses were performed using
2.5.0, R Development Core Team, 2007; FSC, 2008;
, 2007; Hothorn et al., 2006) and all other analyses
10 (SPSS Inc., Chicago, USA). α was set at 0.05.
presented ±S.E.
cteristics of collective movements
ean distance of a collective movement did
between the two species (Mann–Whitney
7320, P = 0.405, NTonkean = 119, Nrhesus = 131,
n = 29 ± 1.29 m, meanrhesus = 26.7 ± 0.73 m).
the mean duration of a collective movement
sed between the departure of the first individual
start point and the arrival of the last individual
point) was not different between the Tonkean

C. Sueur, O. Petit / Behavioural Processes 78 (2008) 84–92 87
macaques and the rhesus macaques (Mann–Whitney
test; U = 1916, P = 0.094, NTonkean = 119, Nrhesus = 131,
meanTonkean = 251 ± 21 s, meanrhesus = 351 ± 29 s). In Tonkean
macaques, 7.97 ± 0.49 individuals participated to a movement
whereas only 4.38 ± 0.42 did in rhesus macaques. Indeed,
the percentage of adhesion of group members, i.e. number of
participants per number of group members differed between
the two species (Mann–Whitney test; U = 4024, P < 0.00001,
NTonkean = 119, Nrhesus = 131, meanTonkean = 58.3 ± 2.9%,
meanrhesus = 39.5 ± 2.7%). In Tonkean macaques, 91.6%
(7.3 ± 0.46 individuals) of the group members participating
in a collective movement arrived at the same end point, and
this percentage was 97.8% (4.15 ± 0.41 individuals) in rhesus
macaques.
3.2. Before
3.2.1. Beh
A GLM
before a d
the movem
taken into
ing attemp
individuals
ber of ind
movement
inary behav
departed in
In Tonk
preceding a
liminary be
participant
ing attempt
displaying
number of p
individual;
vidual was
probability
than these
test, z = −2
ments, 3.09
behaviours
no effect (G
Fig. 1. Influen
on the numb
macaques (wh
only once in b
macaques, but the number of individuals displaying preliminary
behaviours and the number of preliminary behaviours exhibited
by the futu
the numbe
ber of indiv
P = 0.02 fo
the first dep
movements
liminary be
0.92 ± 0.3
first depart
seven cases
There w
nary behav
These indi
or
k: P
(sign
= 1)
Pred
dis
) an
ed i
part
598
sitio
caq
n-gr
mber
dispe
ed in
t dep
Stat
ong
ues,
art a
rhes
de at
ann
male
orrel
pea
infl
ues.
the
nk c
les an
t (C
01).
n the
.10,
mac
fou
diffthe departure
aviours preceding the departure
was carried out to test if the behaviours displayed
eparture predicted the number of participants in
ent (dependent variable). The independent variables
account for this analysis were: the failed preced-
t at departure (presence/absence), the number of
exhibiting preliminary behaviours (i.e. the num-
ividuals advancing in the future direction of the
and/or back glance), as well as the number of prelim-
iours (advance, back glance) displayed by the first
dividual.
ean macaques, the results showed that a failed
ttempt and the number of individuals displaying pre-
haviours both positively influenced the number of
s (GLM; z = 4.48, P < 0.00001 for the failed preced-
; z = 7.05, P < 0.00001 for the number of individuals
preliminary behaviours; z = 0.61, P = 0.504 for the
reliminary behaviours displayed by the first departed
Fig. 1). In 12 cases out of 17, the first departed indi-
the same as in the failed preceding attempt. The
that this individual made a new departure was higher
of any other group member (Wilcoxon signed-rank
.92, P = 0.0035). In 29.4% of the 119 collective move-
± 0.32 individuals emitted 5.46 ± 0.76 preliminary
. In rhesus macaques, a failed preceding attempt had
LM; z = −1.393, P = 0.164), in contrast to Tonkean
ce of the number of individuals displaying preliminary behaviours
er of participants in Tonkean macaques (black) and in rhesus
ite). For six individuals, preliminary behaviours were displayed
oth groups.
nor old
cal ran
group
age: P
3.2.2.
The
persed
depart
ber of
P = 0.0
the po
sus ma
more i
the nu
group
depart
3.3. A
3.3.1.
Am
macaq
two st
first in
13 ma
sex (M
meanfe
rank c
rank (S
had an
macaq
age on
man ra
juveni
attemp
P < 0.0
ence o
r = −0
rhesus
We
icantlyre first departed individual both positively influenced
r of participants (z = 7.32, P < 0.00001 for the num-
iduals displaying preliminary behaviours; z = 2.323,
r the number of preliminary behaviours exhibited by
arted individual; Fig. 1). In 9.2% of the 131 collective
, 3.45 ± 0.42 individuals performed 5.15 ± 0.75 pre-
haviours and the first departed individual displayed
preliminary behaviours per collective movement (the
ed individual displayed preliminary behaviours in
out of the 131 collective movements).
as no evidence that individuals displaying prelimi-
iours belonged to a specific class of dominance or age.
viduals were not specifically high- or low-ranking,
young in the Tonkean group (sign-test for hierarchi-
= 1; sign-test for age: P = 1) neither in the rhesus
-test for hierarchical rank: P = 0.581; sign-test for
.
isposition of the group
persion of the group (grouped, sub-grouped, dis-
d the position (in-group or borderline) of the first
ndividual had no significant effect on the num-
icipants in Tonkean macaques (GLM; z = −1.882,
for the group dispersion; z = −1.727, P = 0.0843 for
n of the first departed individual). However, in rhe-
ues, the more grouped individuals were and/or the
oup the first departed individual was, the higher was
of participants (GLM; z = −3.00, P = 0.0027 for the
rsion; z = 2.263, P = 0.023 for the position of the first
dividual).
arture
us of the first departed individual
the 17 individuals that might start in first in Tonkean
three never made a start attempt and 14 made at least
ttempts. Among the 16 potential that might start in
us macaques, three never made a start attempt and
least three start attempts (Fig. 2). However, neither
–Whitney, U = 30.5, Nfemale = Nmale = 8, P = 0.871,
= 7 ± 1.95, meanmale = 6.5 ± 1.96), age (Spearman
ation, r = −0.28, N = 17, P = 0.278) nor hierarchical
rman rank correlation, r = −0.29, N = 16, P = 0.258)
uence on the number of start attempts in Tonkean
Conversely, we found a tendency for the influence of
number of start attempts in rhesus macaques (Spear-
orrelation, r = 0.47, N = 16, P = 0.068). Indeed, the
d the sub-adult individuals never tried to make a start
hi-square goodness-of-fit test, Chi2 = 29.4, d.f. = 2,
In the latter species, hierarchical rank had no influ-
number of start attempts (Spearman rank correlation,
N = 15, P = 0.711). We did not test the effect of sex in
aques because of the low number of males.
nd that the number of participants did not signif-
er between the different first departed individuals

88 C. Sueur, O. Petit / Behavioural Processes 78 (2008) 84–92
Fig. 2. Numb ber o
in rhesus mac hatev
to their hierar
in Tonkean
P = 0.615,
to the ide
macaques (
Fig. 2). In
number of
Whitney,
meanfemale
man rank
archical ra
P = 0.446).
participant
first depart
N = 131, P
lation, r = 0
departed in
3.3.2. The
individual
We veri
behaviours
the speed
pauses and
individuals
way ANOV
similar res
way ANO
found sign
(one-way A
r of
30,
er-in
gateer of start attempts (white bars), failed start attempts (grey bars) and mean num
aques (b). The mean number of participants is considered for all start attempts w
chical rank, from dominant to subordinate.
macaques (one-way ANOVA, F = 0.87, d.f. = 118,
Fig. 2), while this variable did vary according
ntity of the first departed individual in rhesus
one-way ANOVA, F = 3.56, d.f. = 130, P = 0.00015,
numbe
d.f. = 1
the int
investiTonkean macaques, no significant effect on the
participants was found for either sex (Mann–
U = 1292.5, Nfemale = 55, Nmale = 53, P = 0.389,
= 8.34 ± 0.71, meanmale = 7.4 ± 0.81), age (Spear-
correlation, r = 0.12, N = 119, P = 0.675) or hier-
nk (Spearman rank correlation, r = 0.22, N = 108,
Conversely, in rhesus macaques the number of
s was influenced by the hierarchical rank of the
ed individual (Spearman rank correlation, r = −0.31,
= 0.0002), but not by its age (Spearman rank corre-
.22, N = 131, P = 0.464), with more dominant first
dividuals having a greater number of followers.
behaviours displayed by the first departed
fied if some first departed individuals exhibited more
than others. The variables taken into account were
of the first departed individual and its number of
back glances. There was no difference between
in Tonkean macaques for these three variables (one-
A, F < 1.59, d.f. = 118, P > 0.098). We obtained a
ult for speed in the rhesus macaque group (one-
VA, F = 0.834, d.f. = 130, P = 0.615) whereas we
ificant differences both in the number of pauses
NOVA, F = 3.296, d.f. = 130, P = 0.0004) and the
vidual on t
effect of ag
tion, r = 0.0
rank corre
trary, the h
positively
rank correl
of back gla
P = 0.015)
was, the m
attempting
3.3.3. Did
participant
In Tonk
played by t
of participa
The speed,
by the firs
on the num
z = 5.54, P
the number
back glanc
and/or freq
of individuf participants (curve) per individual in Tonkean macaques (a) and
er the result. Individuals are presented from left to right according
back glances emitted (one-way ANOVA, F = 2.002,
P = 0.03) between first departed individuals. Given
dividual heterogeneity found in rhesus macaques, we
d the influence of the status of the first departed indi-
he number of pauses and back glances. We found no
e (number of back glances: Spearman rank correla-
6, N = 131, P = 0.511, number of pauses: Spearman
lation, r = −0.05, N = 131, P = 0.577). On the con-
ierarchical rank of the first departed individual was
correlated with the number of pauses (Spearman
ation, r = 0.32, N = 131, P = 0.0002) and the number
nces (Spearman rank correlation, r = 0.21, N = 131,
it exhibited. The more subordinate a rhesus macaque
ore it tended to exhibit behaviours associated with
a movement.
these behaviours help the recruitment of
s?
ean macaques, we investigated if the behaviours dis-
he first departed individual could increase the number
nts (dependent variable) irrespective of its identity.
and the number of pauses and back glances displayed
t departed individual both had a significant effect
ber of participants (GLM, AIC = 848.42, d.f. = 115;
< 0.00001 for the speed; z = 3.93, P < 0.00001 for
of pauses; z = −3.44, P = 0.0006 for the number of
es). When the first departed individual walked fast
uently paused, it significantly increased the number
als which adhered to the movement. Conversely, the

C. Sueur, O. Petit / Behavioural Processes 78 (2008) 84–92 89
number of back glances was negatively correlated to the num-
ber of participants, which suggests a monitoring function of this
behaviour.
had no effe
d.f. = 129,
the influen
and the num
participant
3.4. After d
3.4.1. Influ
The ide
important
tigated wh
follower w
ther the se
Nfemale = 43
meanmale =
r = −0.15,
man rank
influence o
In rhesus m
(Spearman
the hierarch
tion, r = −0
with the sa
by the firs
number of
of pauses o
pants in To
P = 0.41 fo
pauses; z =
None of the
ber of parti
z = 0.054, P
number of
glances).
3.4.2. Wer
Finally,
group mem
first follow
lowers. Tw
behave like
10 m in less
the first ind
1990). The
participant
ments, GL
Fig. 3) and
on the 131
and P < 0.0
nant was no
rank correl
old individ
P = 0.922),
nflue
articip
.5, N
= 0.8
t was
66, N
mem
17).
ondl
ours
ck g
ced
, AIC
cases
uals
rly,
ts in
05, P
1 col
.27
cuss
pre
s liv
had
e co
cial
rst s
the p
prov
ses.
this
ould make a start attempt in Tonkean macaque and be
ed. Thus, the leadership was not restricted to a specific
ual within the group. This first result describes a type
up decision making that is similar to the phenomenon
ributed leadership (Leca et al., 2003) and the variable
hip proposed by Conradt and Roper (2005). In rhesus
ues, we found that if all adults could make a start attempt,
and sub-adult individuals never tried to move in first.
parison, young Tonkean macaques could make a start
t and were always followed. This major difference could
lained by the fact that in Tonkean macaques, mothers are
ermissive. Indeed, juvenile can establish strong relation-In rhesus macaques, the analysis showed that speed
ct on the number of participants (GLM, AIC = 928.73,
z = 0.85, P = 0.395). A partial correlation, retrieving
ce of dominance, showed that the number of pauses
ber of back glances had no effect on the number of
s (r < 0.08, d.f. = 129, P > 0.374).
eparture
ence of the first follower
ntity or the status of the first follower might be as
as that of the first departed individual. We inves-
ether sex, age or dominance status of the first
ere a predictor of the number of participants. Nei-
x of the first follower (Mann–Whitney, U = 548.5,
, Nmale = 48, P = 0.364, meanfemale = 10.3 ± 0.87,
11.3 ± 0.68), its age (Spearman rank correlation,
N = 102, P = 0.207), nor its hierarchical rank (Spear-
correlation, r = −0.04, N = 91, P = 0.704) had an
n the number of participants in Tonkean macaques.
acaques, this number was not influenced by the age
rank correlation, r = −0.17, N = 99, P = 0.103) nor by
ical rank of the first follower (Spearman rank correla-
.031, N = 99, P = 0.768). A GLM was then carried out
me dependent variable and the behaviours displayed
t follower as the independent variables (i.e. speed,
pauses and number of back glances). The number
f the first follower affected the number of partici-
nkean macaques (AIC = 618.76, d.f. = 99; z = 0.233,
r the speed; z = 2.069, P = 0.038 for the number of
−0.475, P = 0.634 for the number of back glances).
se behaviours significantly influenced the total num-
cipants in rhesus macaques (AIC = 543.62, d.f. = 84;
= 0.957 for the speed; z = −0781, P = 0.435 for the
pauses, z = −0.041, P = 0.298 for the number of back
e there other ’key’ individuals?
we investigated if some behaviours displayed by
bers other than the first departed individual and the
er could have an influence on the number of fol-
o cases were observed. First, any individual could
the first departed individual by walking more than
than 40 s in the same direction after the departure of
ividual (thus called a determinant; see Byrne et al.,
presence of a determinant increased the number of
s in Tonkean macaques (20 cases on the 119 move-
M, AIC = 593.74, d.f. = 117, z = 5.942, P < 0.00001,
to a similar degree to rhesus macaques (21 cases
movements, GLM, AIC = 646.23, d.f. = 129, z = 9.93
0001, Fig. 3). In Tonkean macaques, this determi-
t particularly dominant within the group (Spearman
ation, r = −0.305, N = 16, P = 0.250), nor was it an
ual (Spearman rank correlation, r = −0.026, N = 17,
nor consistently of a particular sex (Mann–Whitney,
Fig. 3. I
ber of p
U = 27
mmale
minan
r = 0.2
group
P = 0.0
Sec
behavi
and ba
influen
(GLM
In 16
individ
Simila
ticipan
z = 2.0
the 13
3.9 ± 0
4. Dis
The
specie
groups
the sam
than so
are a fi
affect
ology
proces
In
bers c
follow
individ
of gro
of dist
leaders
macaq
young
In com
attemp
be exp
very pnce of the determinant and post-departure behaviours on the num-
ants in Tonkean macaques (black) and rhesus macaques (white).
female = Nmale = 8, P = 0.645, mfemale = 1.25 ± 045,
75 ± 0.35). Similarly, in rhesus macaques the deter-
not an old individual (Spearman rank Correlation,
= 16, P = 0.337). However, it was often a dominant
ber (Spearman rank Correlation, r = −0.604, N = 15,
y, some individuals could exhibit post-departure
, i.e. they advance in the direction of the movement
lance after the start attempt. Such individuals also
the number of participants in Tonkean macaques
= 865.58, d.f. = 117, z = 5.55, P < 0.00001, Fig. 3).
out of the 119 collective movements, 3.12 ± 0.44
exhibited 4.81 ± 0.91 post-departure behaviours.
these individuals did influence the number of par-
rhesus macaques (GLM, AIC = 713.42, d.f. = 129,
= 0.045, Fig. 3) but less strongly. In 20 cases out of
lective movements, 2.5 ± 0.26 individuals exhibited
post-departure behaviours.
ion
sent study includes only one group of each of two
ing in semi-free ranging conditions. Still, the two
the same number of individuals and were bred in
nditions, which limits the influence of factors other
constraints on our results. In this context, our results
tep towards an understanding of how social style can
rocess of decision-making. In addition, our method-
ed adequate to identify consensus decision-making
study, we found that the majority of group mem-

90 C. Sueur, O. Petit / Behavioural Processes 78 (2008) 84–92
ships early in life with all group members and not just with its
kin (Thierry, 1985). In addition, a great tolerance exists towards
juveniles (
entire grou
uals. Conc
success of
carnivores
(Phoenicul
individuals
likelihood o
whereas th
cial compo
macaques.
macaques
first depart
the first dep
other mem
exhibition
several spe
initiate a m
weakly con
2000 for c
keys (Saim
(Macaca sy
(Pan troglo
sistent lead
the same in
because oth
nant indivi
in the grou
first depart
subordinate
need to sig
interestingl
for the succ
subordinate
pants. Thes
macaques,
collective m
One ind
event incre
macaques
decision in
rhesus mac
the current
considered
macaques
were alread
In both T
vidual was
success of
before dep
liminary be
participant
less widesp
This proce
departure i
phants (Poole et al., 1988) and African buffalo (Prins, 1996) and
suggests the sharing of the decision between group members.
epar
ute
inan
nna
ial f
ivid
as it
ues.
tial
strat
decis
pred
an m
he d
ute
nrad
erg
ins
socia
oper
ina
to di
com
ed t
he co
an th
qual
oper
t the
ayin
us m
dt an
s, an
e ex
need
, it h
verba
mem
rilla
ony”
ectio
ill u
to a
, 199
e tha
art o
e.
s larg
uum
ry, 2
may
ur re
r th
to e
f toThierry, 1986). Such tolerance may explain why an
p is willing to follow young first departed individ-
erning dominance, hierarchical rank influenced the
a movement in rhesus macaques only, like in social
(Holekamp et al., 2000) and green woodhoopoes
us purpureus; Radford, 2004). In other words, all
making a start attempt had approximately the same
f being followed by the group in Tonkean macaques,
e status of the first departed individual was a cru-
nent for the success of a start attempt in rhesus
Another discrepancy between Tonkean and rhesus
was found regarding the signals displayed by the
ed individual. In Tonkean macaques, the behaviour of
arted individual, through speed and pauses, favours
bers’ adhesion to the movement. Interestingly, the
of signals at the time of departure was reported in
cies in which more than one single individual could
ovement and in which social relationships were
strained by dominance (Leca et al., 2003 and Boinski,
apuchin monkeys; Boinski, 1988 for squirrel mon-
iri oerstedi); Melhman, 1996 for Barbary macaques
lvanus); Boesch and Boesch, 1989 for chimpanzees
dytes)). In contrast, in species characterised by a con-
ership in which the first departed individual is usually
dividual, such signals seemed to be absent, probably
er group members frequently monitored this domi-
dual (Boinski, 2000; Holekamp et al., 2000). Indeed,
p of rhesus macaques under investigation, dominant
ed individuals rarely used signals in comparison to
ones. It seems that dominant individuals did not
nal their departure to make the group follow, but
y, their in-group position before moving is critical
ess of the movement. Moreover, signals exhibited by
animals had no influence on the number of partici-
e results support again the hypothesis that in rhesus
dominance has a big influence on the success of a
ovement.
ividual could make a start attempt that failed. This
ased the efficiency of a new start attempt in Tonkean
whereas it did not seem to influence a subsequent
rhesus macaques. We may reasonably conclude that
aques mostly focused on the current event and on
first departed individual whereas Tonkean macaques
the general context of a movement. Indeed, rhesus
were more prone to follow an individual when they
y all grouped around it.
onkean and rhesus macaques, the first departed indi-
not the only individual playing a key role in the
a collective movement (see Norton, 1986). Indeed,
arture the number of individuals that displayed pre-
haviours had a significant impact on the number of
s in the movement. Nevertheless the phenomenon was
read in rhesus macaques than in Tonkean macaques.
ss seems similar to the phenomena observed before
n hamadryas baboons (Kummer, 1968), African ele-
After d
contrib
determ
In sava
as cruc
this ind
where
macaq
influen
demon
in the
As
Tonke
trate t
contrib
tus (Co
may em
capuch
els of
and R
of dom
seem
not in
propos
ever, t
Tonke
to an e
and R
ing tha
and pl
in rhes
Conra
specie
produc
would
Indeed
the sil
group
tain go
cerem
the dir
it is st
spond
(Byrne
assum
were p
outcom
It i
contin
(Thier
ences
From o
uum fo
shared
level oture, other individuals could intervene again and thus,
to the decision. In our study, another individual, the
t, particularly influenced the number of participants.
h baboons, Byrne et al. (1990) considered this event
or the subsequent process of adhesion. Nevertheless,
ual could be any group member in Tonkean macaques
more often involved dominant animals in rhesus
In Tonkean macaques, the first follower was also
and when considering the above results, we finally
ed that many different group members participated
ion-making of moving in this group.
icted, we found a shared consensus decision in
acaques. The results presented in this study illus-
emocratic way in which several individuals may
to a collective decision with no regard to their sta-
t and Roper, 2003). As hypothesised, such processes
e in species like Tonkean macaques and white-faced
because of their open social system and high lev-
l tolerance (Leca et al., 2003). Referring to Conradt
’s classification (2005), and despite the constraints
nce on their collective decisions, rhesus macaques
splay a partially shared consensus. This result is
plete accordance with our initial prediction, which
wo extreme processes for these two groups. How-
llective decision appears to be more consensual in
an in rhesus macaques and, thus corresponds more
ly shared consensus in Tonkean macaques (Conradt
, 2005). This conclusion is supported by the find-
number of individuals participating in a movement
g a key role is generally greater in Tonkean than
acaques. Moreover, our results are consistent with
d Roper’s suggestion (2005) that even in despotic
unshared consensus cannot exist because it tends to
cessive decisions, and because dominant individuals
to invest too much energy to impose their choices.
as been reported that besides the prominent role of
ck in the initiation of a collective movement, other
bers emit many grunts before a departure in moun-
s (Stewart and Harcourt, 1994). In wolves, a “group
has been described before the alpha-male shows
n of the future movement (Mech, 1970). However,
nclear whether such pre-departure activities corre-
response facilitation process or a negotiation one
4; Holekamp et al., 2000). In the latter case, we may
t the individuals displaying preliminary behaviours
f the quorum which will contribute to the decision
ely accepted that macaques can be classified on a
from strictly hierarchical to highly tolerant societies
004) and that many inter-specific behavioural differ-
be easily liable to the social style of each species.
sults, we propose the possibility of a parallel contin-
e mechanisms of collective decisions, from partially
qually shared consensus processes according to the
lerance displayed by the species. However, further

C. Sueur, O. Petit / Behavioural Processes 78 (2008) 84–92 91
comparative studies between other species of different domi-
nance styles are clearly required to corroborate this hypothesis.
Acknowled
The auth
analyses, P
English edi
Processes f
manuscript
Ministry (A
Reference
Alexander, R.
5, 326–38
Altmann, J.,
Behaviour
Black, J.M., 1
and flock
Boesch, C.,
in the T
doi:10.100
Boinski, S.,
monkeys
doi:10.100
Boinski, S., 2
primate gr
University
Boinski, S., G
Chicago,
Byrne, R.W., 1
(Eds.), Be
pp. 223–2
Byrne, R.W.,
baboons:
Primatol.
Conradt, L., R
155–158,
Conradt, L., R
Ecol. Evo
Danchin, E.,
the conce
doi:10.112
de Vries, H.,
the analys
Behaviour
Dumont, B.,
Consisten
measurem
Sci. 95, 55
Feist, J.D., M
in feral ho
Foundation fo
URL http:
Giraudoux, F
1.3.3. http
Hofer, H., Eas
hyenas. In
University
Holekamp, K
nivores. I
Chicago P
Hothorn, T., H
conditiona
Ku¨hme, W., 1965. Communal food distribution and division of labour in African
hunting dogs. Nature 205, 443–444.
Kummer, H., 1968. Social Organization of Hamadryas Baboons. University of
ago P
cht, J.
nalysi
B., Fo
cinus
B., G
rangi
0.100
a, S.
caca
, and
0.100
, J.R
ups in
, doi:
.D., 1
n, P.T
aques
aque
G.W
w ba
servat
., Bak
son, J
white
ate. C
., Man
erate
111.
, Desp
in tw
07–1
.H., P
of so
obiol
.H.T.
Hall,
, A.N
oes
9-03
.A.E.,
ya. Be
dt, V.,
aviour
, G.B
ersity
P., 19
omest
L., H
. Co
a (Cr
(95)8
K.J.,
als of
a, J.,
tat an
ids. T
B., 1
tta, M
ks of
B., 1
aques
B., 2
.), Ma
bridggments
ors are grateful to M. Pele´, J. Dubosq for the video
. Uhlrich, for technical assistance and to L. Halsey for
ting. Thanks are extended to the editor of Behavioural
or his helpful comments on a previous version of the
. This work was supported by the French Research
CI cognitique).
s
D., 1974. The evolution of social behavior. Annu. Rev. Ecol. Syst.
3, doi:10.1146/annurev.es.05.110174.001545.
1974. Observational study of behaviour: sampling methods.
49, 227–265.
988. Preflight signaling in swans: a mechanism for group cohesion
formation. Ethology 79, 143–157.
Boesch, H., 1989. Hunting behavior of wild chimpanzees
aı¨ National Park. Am. J. Phys. Anthropol. 78, 547–573,
2/ajpa.1330780410.
1988. Sex differences in the foraging behavior of squirrel
in a seasonal habitat. Behav. Ecol. Sociobiol. 23, 177–186,
7/BF00300352.
000. Social manipulation within and between troops mediates
oup movement. In: Boinski, S., Garber, P.A. (Eds.), On the Move.
of Chicago Press, Chicago, pp. 421–470.
arber, P.A., 2000. On the Move. University of Chicago Press,
822 pp.
994. The evolution of intelligence. In: Slater, P.J.B., Halliday, T.R.
haviour and Evolution. Cambridge University Press, Cambridge,
65.
Whiten, A., Henzi, S.P., 1990. Social relationships of mountain
leadership and affiliation in a non-female-bonded monkey. Am. J.
20, 313–329, doi:10.1002/ajp.1350200409.
oper, T.J., 2003. Group decision-making in animals. Nature 421,
doi:10.1038/nature01294.
oper, T.J., 2005. Consensus decision making in animals. Trends
l. 20, 449–456, doi:10.1016/j.tree.2005.05.008.
Giraldeau, L.A., Walone, T.J., Wagner, R.H., 2005. Defining
pt of public information—Response. Science 308, 355–356,
6/science.308.5720.353c.
Netto, W.J., Hanegraaf, P.L.H., 1993. Matman: a program for
is of sociometric matrices and behavioural transition matrices.
125, 157–175.
Boissy, A., Achard, C., Sibbald, A.M., Erhard, H.W., 2005.
cy of animal order in spontaneous group movements allows the
ent of leadership in a group of grazing heifers. Appl. Anim. Behav.
–66, doi:10.1016/j.applanim.2005.04.005.
c Cullough, D.R., 1976. Behaviour patterns and communication
rses. Z. Tierpsychol. 41, 337–371.
r Statistical Computing, Vienna, Austria. ISBN 3-900051-07-0,
//www.R-project.org.
., 2007. Pgirmess: data analysis in ecology. R package version
://perso.orange.fr/giraudoux/SiteGiraudoux.html.
t, M.L., 2000. Conflict management in female-dominated spotted
: Aureli, F., de Waal, F.B.M. (Eds.), Natural Conflict Resolution.
of California Press, Berkeley, pp. 232–234.
.E., Boydston, E.E., Smale, L., 2000. Group travel in social car-
n: Boinski, S., Garber, P.A. (Eds.), On the Move. University of
ress, Chicago, pp. 587–627.
ornik, K., van de Wiel, M.A., Zeileis, A., 2006. A lego system for
l inference. Am. Stat. 60, 257–263.
Chic
Lampre
an a
Leca, J.-
capu
Leca, J.-
free-
doi:1
Makwan
(Ma
sites
doi:1
Malcolm
of p
1–13
Mech, L
Melhma
mac
Mac
Norton,
yello
Con
Perry, S
Man
wild
prim
Perry, S
coop
109–
Petit, O.
tion’
90, 1
Poole, J
texts
Soci
Prins, H
and
Radford
hoop
j.143
Rasa, O
Ken
Reinhar
Beh
Schaller
Univ
Scott, J.
of d
Smale,
1995
hyen
3472
Stewart,
sign
Supriatn
habi
hybr
Thierry,
mula
wee
Thierry,
mac
Thierry,
(Eds
Camress, Chicago, 189 pp.
, 1992. Variable leadership in bar-headed geese (Anser indicus):
s of pair and family departures. Behaviour 122, 105–120.
rnasieri, I., Petit, O., 2002. Aggression and reconciliation in Cebus
. Int. J. Primatol. 23, 979–998, doi:10.1023/A:1019641830918.
unst, N., Thierry, B., Petit, O., 2003. Distributed leadership in semi-
ng white-faced capuchin monkeys. Anim. Behav. 66, 1045–1052,
6/anbe:2003.2276.
C., 1978. Field ecology and behaviour of the rhesus macaque
mulatta). Part I. Group composition, home range, roosting
foraging routes in the Asarori forest. Primates 10, 483–492,
7/BF02373310.
., Marten, K., 1982. Natural selection and the communal rearing
African wild dogs (Lycaon pictus). Behav. Ecol. Sociobiol. 10,
10.1007/BF00296390.
970. The wolf. Natural History Press, New York, 384 pp.
., 1996. Branch shaking and related displays in wild Barbary
. In: Fa, J.E., Lingburg, D.G. (Eds.), Evolution and Ecology of
Societies. Cambridge University Press, Cambridge, pp. 503–526.
., 1986. Leadership decision processes of group movement in
boons. In: Else, J.D., Lee, P.C. (Eds.), In Primate Ecology and
ion. Cambridge University Press, Cambridge, pp. 145–156.
er, M., Fedigan, L., Gros-Louis, J., Jack, K., MacKinnon, K.,
.H., Panger, M., Pyle, K., Rose, L., 2003a. Social conventions in
-faced capuchin monkeys: evidence for traditions in a neotropical
urr. Anthropol. 44, 241–268, doi:10.1086/345825.
son, J.H., Dower, G., Wikberg, E., 2003b. White-faced capuchins
to rescue a groupmate from a Boa constrictor. Folia Primatol. 74,
ortes, C., Thierry, B., 1992. Differential probability of ‘coproduc-
o species of macaque (Macaca tonkeana, M. mulatta). Ethology
20.
ayne, K., Langbauer Jr., W.R., Moss, C.J., 1988. The social con-
me very low frequency calls of African elephants. Behav. Ecol.
. 22, 385–392, doi:10.1007/BF00294975.
, 1996. Ecology and Behaviour of the African Buffalo. Chapman
London, 320 pp.
., 2004. Vocal coordination of group movement by green wood-
(Phoeniculus purpureus). Ethology 110, 11–20, doi:10.1046/
10.2003.00943.x.
1983. Dwarf Mongoose and Hornbill Mutualism in the Taru desert.
hav. Ecol. Sociobiol. 12, 181–190, doi:10.1007/BF00290770.
1983. Movement orders and leadership in a semi-wild cattle herd.
83, 251–264.
., 1972. The Serengeti Lion: A Study of Predator–Prey Relations.
of Chicago Press, Chicago, 480 pp.
45. Social behavior, organization and leadership in a small flock
ic sheep. Comp. Psychol. Monogr. 18, 1–29.
olekamp, K.E., Weldele, M., Frank, L.G., Glickman, S.E.,
mpetition and cooperation between littermates in the spotted
ocuta crocuta). Anim. Behav. 50, 671–682, doi:10.1016/0003-
0128-6.
Harcourt, A.H., 1994. Gorilla’s vocalisations during rest periods:
impending departure? Behaviour 130, 29–40.
Froehlich, J.W., Erwin, J.M., Southwick, C.H., 1992. Population
d conservation status of M. maurus, M. tonkeana and their putative
rop. Biodivers. 1, 31–48.
985. Social development in three species of macaque (Macaca
. fascicularis M. tonkeana): a preliminary report on the first 10
life. Behav. Process. 11, 89–95.
986. Affiliative interference in mounts in a group of Tonkean
(Macaca tonkeana). Am. J. Primatol. 11, 89–97.
004. Social epigenesis. In: Thierry, B., Singh, M., Kaumanns, W.
caques Societies: A Model for the Studies of Social Organization.
e University Press, Cambridge, pp. 267–284.

92 C. Sueur, O. Petit / Behavioural Processes 78 (2008) 84–92
Tyler, S.J., 1972. The behaviour and social organization of the new forest ponies.
Anim. Behav. Monogr. 5, 85–196.
Watanabe, K., Brotoisworo, E., 1982. Field observation of Sulawesi macaques.
Kyoto U. Overseas Res. Rep. Stud. Asian Non-Hum. Primat. 2,
3–9.
Watts, D.P., 2000. Mountain gorilla habitat use strategies and group movements.
In: Boinski, S., Garber, P.A. (Eds.), On the Move. University of Chicago
Press, Chicago, pp. 351–374.
Whitten, A.J., Mustafa, M., Henderson, G.S., 1987. The Ecology of Sulawesi.
Gadjah Mada University Press, Yogyakarta, 777pp.