Sequence of quorums during collective decision making in macaques

ORIGINAL PAPER
Sequence of quorums during collective decision making
in macaques
Cédric Sueur & Jean-Louis Deneubourg & Odile Petit
Received: 7 March 2010 /Revised: 2 May 2010 /Accepted: 25 May 2010
# Springer-Verlag 2010
Abstract Synchronization of activity is one of the major
challenges of any society, and to what extent social animals
reach a consensus still remains to be established. In the case
of group movements, recent studies have underlined the
importance of the pre-departure period and suggested that
some individuals in a group express their motivation to
move by showing a preference for a specific direction.
However, how do other group members really choose the
time and direction of movement? This study shows that in
two semi-free ranging Tonkean macaque (Macaca tonkeana)
groups, several individuals propose different directions for
movement by displaying unique behavior. The whole
group eventually moves in the choice of direction
supported by the majority of individuals according to a
sequence of three quorum rules. Moreover, when the
number of individuals choosing another direction is higher
than their own group, individuals that proposed alternative
directions eventually renounce and follow the majority.
Despite conflict of interests, group members reach a
consensus before the actual start of group movement. This
demonstrates that processes of this type, which can be
considered to be voting processes, are not exclusive to
human societies and may be explained by a complex
sequence of simple rules.
Keywords Consensus . Voting process . Vote . Threshold .
Primates . Group movement . Self-organization
Introduction
Animal group decision making for group movement (Buhl
et al. 2006; Conradt and Roper 2005; Couzin 2007;
Danchin et al. 2004; Parrish 1999; Sumpter 2006; Ward et
al. 2008) has mainly been studied using the concept of
leadership (Couzin et al. 2005; Biro et al. 2006; Norton
1986). Several authors have suggested that one or a few
individuals decide about the time and the direction of
movement and subsequently “lead” the group (Biro et al.
2006; Couzin et al. 2005; Sueur and Petit 2008a, b).
However, other studies have shown that even if some
individuals consistently lead the group, the period preced-
ing the decision can be crucial with regards to making the
leader depart: other group members may display specific
Communicated by: M. Beekman
Electronic supplementary material The online version of this article
(doi:10.1007/s00265-010-0999-8) contains supplementary material,
which is available to authorized users.
C. Sueur : J.-L. Deneubourg : O. Petit
Unit of Social Ecology, Université Libre de Bruxelles,
Campus Plaine, Bd du Triomphe,
1050 Brussels, Belgium
C. Sueur : O. Petit
Université de Strasbourg, IPHC,
23 rue Becquerel 67087 Strasbourg, France
C. Sueur : J.-L. Deneubourg : O. Petit
CNRS,
UMR7178,
67037 Strasbourg, France
C. Sueur
Department of Ecology and Evolutionary Biology,
Princeton University,
Princeton 08542 NJ, USA
C. Sueur (*)
DEPE, IPHC,
23, rue Becquerel,
67087 Strasbourg Cedex, France
e-mail: cedric.sueur@c-strasbourg.fr
Behav Ecol Sociobiol
DOI 10.1007/s00265-010-0999-8

behaviors (in one or several directions) that consequently
influence its departure probability (Kummer 1968; Prins
1996; Sueur and Petit 2008a, b). It has often been claimed
that a decision depends on a quorum threshold as in the
case of fish, ants or bees (Pratt et al. 2002; Seeley and
Visscher 2004; Sumpter 2006; Ward et al. 2008). A
response to a quorum (or to a threshold) is observed when
the probability of animals exhibiting a particular behavior is
dependent on the number of individuals already performing
the behavior, whatever the process underlying the estima-
tion of this number (Pratt et al. 2002; Seeley and Visscher
2004; Sumpter 2006; Ward et al. 2008). In Temnothorax
ants, the probability that an ant will immigrate to a specific
nest chosen from two or several new nests depends on the
number of ants already present in this nest (Pratt et al.
2002). Similarly, a quorum threshold was described in
honey bees (Apis mellifera) when they selected a new nest
site (Seeley and Visscher 2004). An experimental study on
three-spine sticklebacks (Gasterosteus aculeatus) showed
that a quorum response could also explain group movement
decisions in fish and how they choose between two
proposed directions (Ward et al. 2008).
When mammals proposed different directions for move-
ment using specific behaviors (Kummer 1968; Prins 1996),
more complex processes called “voting behaviors” are
evident (Conradt and Roper 2005; Kummer 1968; Prins
1996; Sueur and Petit 2008a, b). These processes seem to
imply a more global type of communication (Conradt and
Roper 2005). A well-known study suggesting the existence
of concurrent proposals by group members was carried out
on hamadryas baboons (Papio hamadryas hamadryas)
during movement from the sleeping area to waterholes in
the morning (Kummer 1968). In that species, males,
supported by their harem, repeatedly tried to influence
other group members to follow their direction using a
specific posture called “notifying behavior” (“an individual
soothingly presents his hindquarter in the fleeting manner
customary between males,” Kummer 1968). At the end of
the decision-making process, the entire troop went in the
direction taken by the majority of group members. In a
study on African buffalo (Syncerus cafer), individuals were
usually seen to be resting several minutes prior to the
group's departure towards a new grazing location. Adult
cows intermittently stood-up and oriented their heads and
bodies in a particular direction. After such reorientation,
animals often resumed resting or grazing locally until
departure. Thus, the reorientation was clearly not a sign of
immediately impending behavior. However, the direction
taken for the eventual grazing location was the most
frequently observed direction in all notifying behaviors
before movement (Prins 1996). While the study on baboons
or that carried out on African buffalo provided descriptive
evidence for voting behavior, the quantitative data were
insufficient to establish whether group members truly
counted votes and decided according to the majority
(Conradt and Roper 2005; Kummer 1968; Prins 1996).
Our study provides quantitative evidence that similar
processes exist in Tonkean macaques.
Tonkean macaques are considered to be a tolerant
species (Anderson 2007; Thierry 2007) in which individ-
uals of every status can express their motivations before,
during, and after the start of group movements (Sueur and
Petit 2008a). All group members therefore take part in the
decision-making process, which reflects an equally shared
consensus (Conradt and Roper 2005; Sueur and Petit
2008a). We have previously shown that “preliminary
behaviors” (Sueur and Petit 2008a, 2009) displayed before
departure favor the recruitment of group members to the
group movement. The more preliminary behaviors oc-
curred, the more individuals joined the movement (Sueur
and Petit 2008a). The preliminary behaviors looked like
preparation for the future group movement and on rare
occasions occurred in two different directions. However, we
still do not know how this decision is taken prior to
departure. Given these previous findings on collective
decision making and the egalitarian basis of their society,
Tonkean macaques seem to be an ideal choice for the study
of voting processes. Here, we address this question in two
semi-free ranging groups of Tonkean macaques using
observations of behavior when two concurrent directions
for movement were proposed.
Methods
Subjects and environment
The two study groups of Tonkean macaques were bred in
semi-natural conditions at the Strasbourg University Centre
of Primatology. They ranged in a 0.5 ha park (fenced field),
containing different patches of vegetation between which
individuals could collectively switch. The first Tonkean
macaque group studied (November 2005 to March 2006)
consisted of ten individuals: one adult male (over 5 years),
five adult females (over 4 years), and four juveniles (from 1
to 3 years). The second group (study carried out from
December 2003 to April 2004) consisted of 22 individuals:
seven adult males (27, 15, 11, 9, 7, 6, and 5 years old),
eight adult females (33, 25, 24, 21, 21, 14, 12, and 11 years
old), one subadult male (4 years old), and six juveniles
(1 year old). The composition of both groups was similar to
that found in wild groups (Supriatna et al. 1992; Whitten et
al. 1987). Animals had free access to an inside shelter with
commercial pellets and water ad libitum. Fruit and
vegetables were distributed once a week, outside the
observation sessions.
Behav Ecol Sociobiol

Observational protocol
The beginning of a group movement was defined by the
first departure of an individual who walked more than 10 m
in less than 40 s. This criterion was the same as that used by
Leca et al. (2003), Jacobs et al. (2008), and Sueur and Petit
(2008a, b), and it allowed us to discriminate first departures
(i.e., initiations) from other movements such as foraging
movements or preliminary behaviors (Sueur and Petit
2008a). The departure of the first individual over a distance
of more than 10 m was an obvious signal for other group
members (Jacobs et al. 2008; Leca et al. 2003; Sueur and
Petit 2008a, b). A “joiner” was defined as any individual
walking for more than 5 m in a direction that formed an
angle a less than 45° with the direction of the first
individual to depart. We considered a group movement to
be finished when no individual joined the movement within
5 min of the departure of the first individual or the last
individual to join (Sueur and Petit 2008a, b; Sueur et al.
2009). The park of each group was marked with reference
points, and the position (±1 m) of each animal as well as the
distance it walked was recorded. Group movements occurring
in agonistic or sexual contexts were discarded from the
analysis. Conflicts or consorts could induce non-spontaneous
movements. Events were only taken into account if more than
2/3 of group members were present in or around the starting
zone when they occurred. We defined the starting zone to be
the area ≤10 m from the starting point of the first departed
individual. With this criterion, both groups were clumped in
the majority of cases; the diameter of the group was less than
10 m in both cases, whatever the study group (Sueur and Petit
2008a). Using video scoring, C.S. recorded the identity as
well as the frequency of any behavior displayed by each
group member. Measurements were taken using the all-
occurrence sampling method, both during the twenty
minutes prior to a group movement, and after the start of
the group movement (Altmann 1974). A back glance is
defined as an individual turning its head and looking in the
direction of other group members. In the cases in which the
eyes of animals could not be observed, we used the direction
of the head—with an angle greater than 135° from the
direction of the movement—to determine a back glance (see
Sueur and Petit 2009 for more details). An intention
movement is defined as the walking of an individual, for
between 1 to 5 m, in a specific direction. A stop of more
than 2 s after an advance was considered to show the end of
the behavior. This definition corresponds to that of Kummer
(1968) and Prins (1996). Indeed, the direction of the body
axis of individuals indicated the direction of the future group
movement (Prins 1996). We considered the back glances as
preliminary behaviors when an individual had previously
made some intention movements (at least one) and moved
away from the group. Back glances and intention move-
ments were recorded and named as “preliminary behaviors”
when exhibited before departure (Sueur and Petit 2008a).
We called individuals displaying preliminary behaviors
notifying individuals. We considered a direction to exist
when there were at least two preliminary behaviors (two
intention movements or one intention movement coupled
with one back glance) for that direction, performed by the
same individual or different ones. If the directions of at least
two individuals displaying preliminary behaviors formed an
angle greater than 45°, we considered these directions to be
different (direction 1 and direction 2). We considered
direction 1 as the direction eventually chosen by the entire
group (i.e., the direction of the future group movement) and
direction 2 as the direction that was not chosen by the group
at departure. A switch in direction was considered to have
occurred when an individual displaying preliminary behav-
iors in one direction changed its direction towards another
one. The pre-departure period was defined as the time
between the first preliminary behavior and the departure of
the initiator.
Assessing dominance rank
To establish the dominance hierarchy of each group, we
recorded data in two contexts: during spontaneous aggres-
sive interactions, and during drinking competition around a
single source of orange juice. We ranked individuals over
the age of one in a matrix according to the avoidance and
unidirectional aggression observed. We carried out an
analysis of hierarchical rank order using MatMan and
verified the hierarchy linearity (de Vries et al. 1993) for
both species (Group 1: h’=0.79, p<0.0006; Group 2: h’=
0.81, p<0.0001).
Statistical analysis
Even though many routes were available and taken by the
group in the enclosure, no event was recorded in which
more than two directions were displayed before the
departure of a group. We conducted two-tailed sign tests
to investigate whether the frequencies of behaviors, number
of notifying individuals, the mean hierarchical rank, and the
mean age of notifying individuals differed between direc-
tion 1 and direction 2. A positive difference given by the
sign test is the number of cases in which the value of
direction 1 exceeds that of direction 2. A null difference is
the number of cases in which the value of direction 1 is
equal to that of direction 2. A negative difference is the
number of cases in which the value of direction 1 is less
than that of direction 2. Differences in the rate of
notification between individuals were tested using a Chi-
square test. The same procedure was used to test whether
the initiator was more highly ranked or older than other
Behav Ecol Sociobiol

individuals in the movement. Departing in one direction
involves two different decisions: the time decision (when) and
the direction decision (where). We decided to study these two
decisions separately in order to understand the rules underly-
ing the initiator's choice, i.e., whether the initiator first
decided on the departure time, and then on the direction to
follow, or the converse. Analyses of first departure proba-
bility distributions (probability of being the first to depart
and probability of choosing one direction according to the
number of notifying individuals and behaviors) and switches
in direction were carried out using exponential, logarithmic
and sigmoid curve estimation tests (R²) and Spearman's rank
correlation tests (rs). The study of probability distributions
reveals how the rate of observations—the probability—
evolves according to a certain variable (here, we tested the
duration of the pre-departure period, the number of
preliminary behaviors, and the number of notifying individ-
uals for each direction).
Survival analysis, or analysis of probability distribution,
is the best statistical framework for quantifying time-
structured behaviors (Cox and Oakes 1984). This method
has already been successfully applied in numerous etho-
logical studies (Amé et al. 2006; Dussutour et al. 2005;
Meunier et al. 2006; Nicolis et al. 2003; Sueur et al. 2009;
Theraulaz et al. 2002). We used this methodology to assess
how the probability to switch direction or to depart in a
specific direction is influenced by different variables in the
pre-departure period: the duration of the pre-departure
period, the number of preliminary behaviors, and the number
of notifying individuals for each direction. The sigmoid and
logarithmic curves show a threshold that may correspond to
a quorum. This threshold is the value of the independent
variable—here, the number of preliminary behaviors or of
notifying individuals—for which the probability that indi-
viduals exhibit a behavior equals 0.5. On the other hand, the
exponential curve shows that the probability of displaying a
specific behavior is not dependent on the number of
notifying individuals and/or number of preliminary behav-
iors (Meunier et al. 2006; Sueur et al. 2009; Sumpter and
Pratt 2009). Only the values of the model fitting best with
the observed data are indicated in the results. Tests were
performed using SPSS 10 (SPSS inc., Chicago, USA). α
was set at 0.05. Means were±S.E.M.
Results
Among the 146 group movements recorded in group 1 and
the 119 recorded in group 2, 44 and 36, respectively, were
preceded by preliminary behaviors before departure.
Among these events we found 13 group movements in
group 1 and 14 in group 2 where several individuals
displayed preliminary behaviors in two different directions
before departure. A total of 9.6±0.3 and 18.0±4.4
individuals in groups 1 and 2, respectively, participated in
these movements with two proposed directions. With
regards to other movements for which no preliminary
behavior was displayed, the process of decision making
was only based on initiations, and fewer individuals
participated (Sueur and Petit 2008a, b; Sueur and Petit
2009; Sueur et al. 2009). Displaying preliminary behaviors
in two different directions reflects a conflict of interest
between group members and is a good opportunity to study
the decision-making process. In such cases, while the group
was stationary, at least two animals displayed preliminary
behaviors in two different directions (Fig. 1a, b). At a given
time, one individual (the initiator) initiated a movement in
one of the two proposed directions and was followed by all
group members (Fig. 1c). We called direction 1 the
direction taken by the initiator (the direction of the future
group movement) and direction 2 the other direction. We
conducted different analyses in order to investigate how the
initiator decided on its choice of direction for movement,
and how cohesion was maintained within the group.
Deciding on direction
Which direction was chosen, and how?
Before the initiator's departure, 66.1±5.4% of members of
group 1 and 35.4±3.3% of members of group 2 had already
displayed preliminary behaviors. More precisely, 38.5±
5.9% of individuals in group 1 and 24.3±3.1% in group 2
notified direction 1 (i.e., displayed intention movements in
direction 1) while 27.7±3.6% in group 1 and 11.1±1.2% in
group 2 notified direction 2. The duration of the pre-
departure period, i.e., the time elapsed between the first
preliminary behavior and the departure of the initiator, was
242±51 s for group 1 and 131±22 s for group 2. For group
1, the number of preliminary behaviors (N=13, P=0.0063,
sign tests, Table 1) and the number of notifying individuals
(N=13, P=0.038, sign tests, Table 1) were higher for
direction 1 than for direction 2 on departure of the initiator.
We obtained similar results for group 2, both for the
number of preliminary behaviors (N=14, P=0.0009, sign
tests, Table 1) and for the number of notifying individuals
(N=14, P=0.0063, sign tests, Table 1). In only one case,
group 1 moved in the direction notified by the less common
behavior. It never happened in group 2. There was no case
in which the majority of individuals notified one direction
but more preliminary behaviors occurred in the other
direction; the number of preliminary behaviors was always
the highest for the direction in which the highest number of
notifying individuals was. Moreover, only on one occasion
did groups 1 and 2 move in the direction notified by the
lower number of individuals. This suggests that the initiator
Behav Ecol Sociobiol

took the number of preliminary behaviors and/or number of
notifying individuals into account when choosing the
direction in which to move (Fig. 1c, analyses of the role of
the initiator in the pre-departure process are shown below).
The initiator chose the direction for which there were the most
preliminary behaviors and notifying individuals.
The probability (number of observations for each
independent variable divided by the total number of
observations) of the initiator choosing direction 1 according
to the difference in the number of preliminary behaviors
between direction 1 and direction 2 is a sigmoid function
(y ¼ 11þs�lx with λ=5, where y is the probability to choose
direction 1 and x is the difference in the number of
preliminary behaviors between direction 1 and direction 2)
for group 1 (N=22; rs=0.93, P<0.00001) and for group 2
(N=22; rs=0.89, P<0.00001). We obtained the same result
for the number of notifying individuals (y ¼ 11þs�lx with λ=
17, where y is the probability to choose direction 1 and x is
the difference in the number of notifying individuals
between direction 1 and direction 2; N=26; rs=0.87, P<
0.00001, Fig. 2a for group 1; N=28; rs=0.87, P<0.00001,
Fig. 2b for group 2). For both groups, the threshold was 0 for
both the number of preliminary behaviors and the number of
notifying individuals. The difference in the number of
preliminary behaviors and/or notifying individuals between
direction 1 and direction 2 only needs to equal one for the
initiator to choose the direction with the majority of
preliminary behaviors and/or notifying individuals. However,
the value of λ—representing the power factor influencing the
sigmoid curve slope and showing the non-linearity of the
phenomenon—was higher for the number of notifying
individuals, indicating a likely higher discrimination than
for the number of preliminary behaviors. However, this
result should be treated with caution, as only a minimal
difference can be seen in the influence of the number of
preliminary behaviors and notifying individuals. λ affects the
slope of the sigmoidal curves that have been fitted to the
data. Higher values of λ correspond to the fitted curves
having a steeper slope, particularly around the threshold
value of x (zero in this case). In essence a higher value of λ
direction 1 direction 1 direction 1direction 2 direction 2 direction 2
1m 2m
α ≥ 45
resting
individualsa
individuals displaying
preliminary behaviours
b c
Switch of direction departure: mouvement
in direction 1
Ndir1 = 2
Ndir2 = 1
Ndir1 > Ndir2
Ndir1 = 4+1
Ndir2 = 3-1
Ndir1 > Ndir2
Ndir1 = 10
Ndir2 = 0
cohesion
Fig. 1 Illustration of the decision-making process in Tonkean
macaques: a one sub-group of macaques proposes a direction
(direction 1) using preliminary behaviors, another individual proposes
an alternative direction (direction 2); b some individuals decide to
switch their direction, and the majority of these switches are from
direction 2 to direction 1; c the initiator departs and the movement
starts in direction 1, the choice of the majority of notifying
individuals; the remaining individuals notifying direction 2 join the
movement in direction 1 and then maintain cohesion with other group
members
Table 1 Number of cases in which each group of Tonkean macaques moved in direction 1 (direction finally chosen by the group) when the value
of the independent variable in direction 1 is less than that in direction 2 (D1<D2), when the value in direction 1 is equal to that in direction 2 (D1
=D2) and when the value in direction 1 is greater than that in direction 2 (D1>D2)
Number of cases for which the direction 1 was chosen
Group 1 Group 2
D1<D2 D1=D2 D1>D2 D1<D2 D1=D2 D1>D2
Number of preliminary behaviors 1 1 11 0 3 11
Number of notifying individuals 1 2 10 1 2 11
Mean hierarchial rank 7 0 6 7 0 7
Mean age 8 1 4 8 0 6
D1 direction 1, D2 direction 2
cases in which each group of Tonkean macaques
mov d in direction 1 (direction finally chosen by the group) when the
value of the indep ndent variable in direction 1 is less tha that in
direction 2 (D1<D2), when the value i direction 1 is equal o th t in
direction 2 (D1=D2) and when the value in direction 1 s greater than
that in direction 2 (D1>D2)
Behav Ecol Sociobiol