Signals use by leaders in Macaca tonkeana and Macaca mulatta: group-mate recruitment and behaviour monitoring
- First Online:
- Cite this article as:
- Sueur, C. & Petit, O. Anim Cogn (2010) 13: 239. doi:10.1007/s10071-009-0261-9
- 214 Views
Animals living in groups have to make consensus decisions and communicate with each other about the time, or the direction, in which to move. In some species, the process relies on the proposition of a single individual, i.e. a first individual suggests a movement and the other group members decide whether or not to join this individual. In Tonkean (Macaca tonkeana) and rhesus macaques (Macaca mulatta), it has been observed that this first individual displays specific signals at departure. In this paper, we aimed to explore the function of such behaviours, i.e. if these behaviours were recruitment signals or only cues about the motivation of the first departed individual. We carried out temporal analyses and studied the latencies of the first departed individual’s behaviours and of the joining of other group members. We also assessed whether the social style of a species in terms of dominance and kinship relationships influenced the patterns of signal emissions. We then analyzed how the first departed individual decided to make a pause or to stop it according to the identities of group members that joined the collective movement. Results showed that Tonkean macaques and rhesus macaques seemed to use back-glances to monitor the joining of other group members and pauses to recruit such individuals. This was especially the case for highly socially affiliated individuals in Tonkean macaques and kin-related individuals in rhesus macaques. Moreover, back-glances and pauses disappeared when such individuals joined the first departed individual. From these results, we suggested that such behaviour could be considered intentional. Such findings could not be highlighted without temporal analyses and accurate observations on primate groups in semi-free ranging conditions.
KeywordsDecision-makingCollective movementIntentionMacaqueSocial styleKinship
Most animals have to move to specific areas for drinking, resting or foraging. For group-living species, these movements need to be collective in order to keep the advantages of sociality for group members (Alexander 1974; Wrangham 1980). This synchronization requires communication and consensus between individuals so that collective decisions can be made about the time and the direction of movement (Conradt and Roper 2005; Krause and Ruxton 2002). The processes underlying these consensuses may vary across species. In Hamadryas baboons (Papio hamadryas hamadryas) and African buffalo (Syncerus caffer), voting behaviours seem to occur before the departure of the group; several individuals may suggest different choices and the group then decides to move in the direction for which the majority of individuals displayed such voting behaviour (Conradt and Roper 2005; Kummer 1968; Prins 1996). However, in some species the process relies on the proposition of a single individual, the leader or initiator (Leca et al. 2003; Couzin et al. 2005; Sueur and Petit 2008b; Stueckle and Zinner 2008). This first departed individual suggests the time or the direction to move (Hall and DeVore 1965; Kummer 1968; Sueur and Petit 2008a) and the other group members decide whether or not to join this individual (Leca et al. 2003; Stueckle and Zinner 2008). Signals used by this individual may be visual, acoustic and/or odorous and may depend on environmental conditions, such as vegetation density (Boinski and Garber 2000). In the Mountain gorilla (Gorilla gorilla berengei), the silverback walks quickly in the direction of the future movement (Watts 2000), whereas common chimpanzees (Pan troglodytes) seem to use drumming on buttressed trees (Boesch 1991) and Barbary macaques (Macaca sylvanus) seem to use branch shaking (Mehlman 1996) to signal the departure of a collective movement. Authors have observed that acoustic signals, such as a ‘deep hoarse cluck’ in the howler monkey (Alouatta palliata, Carpenter 1934) and ‘coo’ vocalization in Japanese macaques (Macaca fuscata, Itani 1963) and ‘trill’ in the white-faced capuchin monkey (Cebus capucinus, Boinski and Campbell 1995) are also given before groups depart. Similarly, a ‘loud call’ was reported in Sulawesi macaques in the context of collective movements (Riley 2005; Thierry et al. 2000). A more empirical study on semi-free ranging white-faced capuchin monkeys showed that the first departed individual displayed recruitment and monitoring behaviour during the onset of a collective movement (Meunier et al. 2007). Moreover in capuchin monkeys, the behaviour of other group members seemed to affect the behaviour of the first departed individual (as a feedback): when the numbers of joiners increased, the number of signals the first individual displayed decreased (Meunier et al. 2007). Similarly in Tonkean (Macaca tonkeana) and rhesus macaques (Macaca mulatta), the first departed individual displayed pauses and back-glances at departure (Sueur and Petit 2008a). However, we did not investigate the effect of such behaviour on other group members in this previous study. In this current study, we performed temporal analyses to explore the function of such behaviours and whether they differed between both species. We aimed to assess whether the behaviours displayed by the first departed individual were recruitment and/or monitoring’ signals. Moreover, we aimed to verify if the social style of a species (de Waal and Luttrell 1989), especially in terms of nepotism, may influence the emission of signals—as is the case for the organization of joining (Sueur and Petit 2008b). Rhesus macaques are more nepotistic than Tonkean macaques: many behaviours, such as grooming, reconciliation and social play are constrained by kinship in rhesus macaques but not in Tonkean macaques (Thierry 2004, 2007). Therefore, we assumed that rhesus macaques would stop displaying signals after kin-related individuals have joined. Conversely, Tonkean macaques should display signals to recruit strong affiliated individuals, whatever their degree of kinship might be.
Subjects and study area
The groups under investigation were bred in the Centre of Primatology at the Strasbourg University, in semi-natural conditions. All group members were born in captivity. The group of rhesus macaques composed of two matrilines. At the time of the study (May 2006 to August 2006) the group consisted of 22 individuals: two adult males (17 and 8-years-old), 11 adult females (16, 14, 12, 11, 11, 11, 8, 7, 7, 7 and 6-years-old), two sub-adult females (both 4-years-old) and 7 infants (<1-year-old). The group of Tonkean macaques composed of five matrilines. At the time of the study (November 2005 to March 2006), the group consisted of 10 individuals: one adult male (10-years-old), five adult females (10, 9, 7, 6 and 5-years-old), one subadult male (3-years-old) and three juveniles (2, 1 and 1-year-old). The composition of the two groups was comparable to wild groups (Makwana 1978; Supriatna et al. 1992; Whitten et al. 1987). Maternal kin relationships are known for both groups. We did not analyze the infants’ behaviour as their discrimination was impossible. The study was based on 15 rhesus macaque individuals and on 10 Tonkean macaque individuals. Each group lived in a park (fenced field of 0.5 ha) with trees, bushes and grassy areas. The area had an inside shelter (20 m2), where commercial pellets and water were provided ad libitum. Fruit and vegetables were distributed once a week, outside of observation sessions. For both species, groups were cohesive and moved collectively (as a whole group or in sub-groups) between areas devoted to specific activities (for details, see Sueur and Petit 2008a).
The beginning of a collective movement was defined by the departure of a single first individual (first departed individual) walking more than 10 m in <40 s (Leca et al. 2003; Sueur and Petit 2008a, b). Any individual walking for more than 5 m in a direction that formed an angle smaller than 45° with the direction of the first departed individual and within 5 min after the departure of the first departed individual was labelled a joiner (Sueur and Petit 2008a, b). This time window was determined by the mean latency separating the joining of two direct participants; mean = 27.9 ± 1.1 s for Tonkean macaques; mean = 64.1 ± 2.7 s for rhesus macaques; where joining is the moment at which an individual starts to follow (Sueur and Petit 2008a).
We defined the departure latency, of every joiner, as the time elapsed between its departure and the time of the first departed individual.
The duration of a collective movement is the time elapsed between the departure of the first individual from the start point and the arrival of the last individual at the end point (Sueur and Petit 2008a).
Speed of each individual: in m s−1, the time an individual needed to cover the first 10 m after its departure.
- Back-glance the individual looks in the direction of other group members, measured as a frequency throughout the movement (i.e. as long as the individual moves). In the cases, where eyes of animals could not be observed, we used the direction of the head (with an angle wider than 135° with the direction of the movement) to determine a back-glance (Fig. 1).
Pause the individual stops moving for at least 2 s. The frequency of pauses throughout the movement was recorded. A pause was qualified as a distinct event when it was separated by more than 2 s from a preceding one.
Loud call a high-pitched vocalization composed of phrases consisting of frequency modulated units (Thierry et al. 2000).
We considered only pauses that were longer than 2 s in order to discard pauses due to the display of other behaviours (back-glances or loud call). Nevertheless, when a back-glance was emitted simultaneously with a pause longer than 2 s, we considered the two events as independent. For analyses, we considered “a pause after a back-glance” as a pause occurring at least 2 s after the end of the preceding back-glance. We considered “a back-glance after a pause”, a back-glance occurring at least 2 s after the end of the preceding pause.
Groups were observed and filmed continuously by two observers (simultaneously in order that all individuals were always seen), 4 h per day between 10:00 and 16:00 hours. Each collective movement was recorded onto videotape and these videotapes were analyzed only by one person (C·S.). Participants (first departed individual and joiners) were observed one by one using video scoring. Movements for which data about behaviours were missing were not taken into account in the analyses. For example, we discarded cases in which an individual had not been continuously observed, since we could not determine whether behaviours, such as pause or back-glance might have been displayed. Movements occurring in context of conflict or sexual consort were discarded. Collective movements were taken into account only if more than 2/3 of group members were present in the starting zone (≤10 m from the point where the first departed individual started, Sueur and Petit 2008a, b). With this criterion, both groups were not dispersed in the majority of cases (the diameter was inferior to 10 m in both groups). The departure of the first departed individual over a distance of more than 10 m was an obvious signal for other group members (Leca et al. 2003; Sueur and Petit 2008a). We drew a plan of each park allowing us to measure the distance walked by group members (Sueur and Petit 2008a).
We observed 131 collective movements for the rhesus macaques and 146 movements for the Tonkean macaques.
We scored the identity and the behaviour (speed, back-glance, pause and loud call, see definitions above) of the first departed individual and of every joiner.
We scored the departure latency of every joiner. Using these latencies, we determined the order of individuals at each collective movement. At departure, we attributed rank 0 to the first departed individual and rank j to the individual that joined the movement when j individuals were already participating. Thus, jmax = n − 1, where n is the number of individuals of each group (except infants).
In the same way, we recorded the start time and the end time of each back-glance and of each pause. These data allowed us to study the dynamics of the collective movement and to assess whether behavioural changes implied changes in the joining of group members (or the converse).
To determine these times and latencies, we used the time indicated on the video-tape.
We considered two individuals as related when belonging to the same matriline regardless of their degree of relatedness and addressed a coefficient equalled 1 to the dyad. Then, in order to make results of both species comparable, we corrected the coefficient of kinship for each individual by the number of its kin-related individuals.
Affiliative relationships were quantified by calculating the number of events when two individuals were observed within body contact out of moving context. This was carried out, using instantaneous scan sampling every 5 min (Altmann 1974). We kept only scans, where all group members could be observed. We collected 88 scans for rhesus macaques and 111 for Tonkean macaques. For subsequent analyses, we used the “ratio of contacts” that was the number of scans for which two individuals were in contact per the total number of scans. Thus, affiliative relationships represented preferential relationships between group members (including kin and non-kin related partners).
The majority of group members, whatever the species, can be the first departed individual of a movement. This first departed individual was any group member in Tonkean macaques whereas only adults initiated a collective movement in rhesus macaques (Sueur and Petit 2008a). Rates of initiations were not different among initiators whatever the species (Sueur and Petit 2008a) was. Moreover, ranks are not tied to individuals (as found by Sueur and Petit 2008b). As a consequence, we analyzed the behaviour of the first departed individual and other ranks in the movement, regardless of their identities (Leca et al. 2003; Sueur and Petit 2008a; Stueckle and Zinner 2008).
We tested correlations between rank at departure and mean number of signals using the Spearman rank correlation test and analyzed the curves using curve estimation tests. We tested inter-rank differences using Kruskall-Wallis tests followed by post-hoc Dunn’s multiple comparisons test. We carried out these tests using SPPS 10.0 and GraphPad Prism 4.03. We then carried out matrices correlations using Socprog2.3 (Whitehead 1997, 2009) with Kr test taking into account missing values of a matrix and allowing to compare the values of each row with all other values in the row (Hemelrijk 1990; Whitehead 1997, 2009). This row wise matrix correlation method is suited for the evaluation of these types of covariance between behaviours and is appropriate for small-sized matrices (de Vries et al. 1993). We set the number of permutations at 10,000 for each correlation matrices test (Whitehead 1997, 2009). Permutations of the rows and columns of one of the two matrices were generated and for each permutation, statistical values were calculated. This method provided more accurate and stable p values (Hemelrijk 1990; de Vries et al. 1993; Whitehead 1997). Mean values are represented as ±SE. α = 0.05.
How did behavioural patterns vary according to ranks at departure?
The mean group stationary time before moving was 1,161 ± 210 s (19.35 ± 3.5 min) for Tonkean macaques and 868 ± 84 s (14.47 ± 1.4 min) for rhesus macaques. In both species, no loud calls occurred during collective movements. For each collective movement, at least one individual carried out pauses and back-glances. In Tonkean macaques, the first departed individual emitted at least one pause or back-glance in 109 cases out of 146 collective movements. In rhesus macaques, the first departed individual emitted at least one pause or back-glance in 92 cases out of 131 collective movements.
Pauses and back-glances are correlated for Tonkean macaques (Spearman rank correlation: N = 146, r = 0.25, p = 0.002) and for rhesus macaques (Spearman rank correlation: N = 131, r = 0.43, p < 0.0001). Indeed, it seems that the first departed individual emitted back-glances when it made pauses. However, even if these two variables are correlated, they are not collinear (collinearity diagnostics, VIF ≤ 1.272 for both species; collinearity is a high correlation between two variables: the variance inflation factor (VIF) has to equal at least 5 to consider two variables as collinear and thus as dependent (Pallant 2007)). We can therefore consider the number of pauses and the number of back-glances as independent variables.
Intensity of signals per collective movement (mean number of back-glances, mean number of pauses, mean speed in m/s; mean ± SE) according to the rank at departure for collective movements in the Tonkean macaques group and the rhesus macaques group
Mean no. of back-glances
Mean no. of pauses
0.53 ± 0.12
3.02 ± 0.28
0.62 ± 0.02
0.10 ± 0.04
2.91 ± 0.34
0.52 ± 0.03
0.03 ± 0.02
2.51 ± 0.33
0.55 ± 0.04
0.07 ± 0.04
2.48 ± 0.31
0.62 ± 0.06
0.07 ± 0.07
1.77 ± 0.21
0.70 ± 0.10
2.17 ± 0.28
0.59 ± 0.07
2.6 ± 0.37
0.63 ± 0 .09
2.40 ± 0.32
0.54 ± 0.04
0.03 ± 0.03
2.43 ± 0.47
0.48 ± 0.05
2.17 ± 0.50
0.66 ± 0.10
0.94 ± 0.12
1.19 ± 0.11
0.58 ± 0.03
0.41 ± 0.10
0.75 ± 0.13
0.58 ± 0.03
0.27 ± 0.08
0.72 ± 0.12
0.57 ± 0.04
0.15 ± 0.08
0.87 ± 0.14
0.58 ± 0.05
0.27 ± 0.09
0.81 ± 0.15
0.24 ± 0.04
0.05 ± 0.03
0.55 ± 0.13
0.49 ± 0.06
0.11 ± 0.08
0.59 ± 0.32
0.52 ± 0.07
0.21 ± 0.12
0.95 ± 0.35
0.40 ± 0.07
0.46 ± 0.27
0.52 ± 0.07
0.40 ± 0.30
0.8 ±0. 41
0.45 ± 0.09
0.28 ± 0.28
0.28 ± 0.19
0.62 ± 0.12
0.45 ± 0.05
0.50 ± 0.50
0.50 ± 0.50
The mean frequency of back-glances and rank were negatively correlated for Tonkean macaques (N = 10, rs = −0.75, p = 0.012) and rhesus macaques (N = 15, rs = −0.71, p = 0.04). For Tonkean macaques, we found that mean frequency of back-glances differed between ranks (Kruskall-Wallis test, H = 94.32, df = 9, p < 0.0001) with only rank 0 having a higher frequency than other ranks (Dunn’s multiple comparison test, p < 0.001). For rhesus macaques, mean frequency of back-glances also differed between ranks (H = 65.67, df = 14, p < 0.00001) with only rank 0 having a higher frequency than ranks 1 to 6 and rank 8 (Dunn’s multiple comparison test, p < 0.01). The first departed individual displayed 71% of all recorded back-glances in Tonkean macaques and 56% in rhesus macaques.
The mean frequency of pauses and rank were negatively correlated for Tonkean macaques (N = 10, rs = −0.65, p = 0.042) and rhesus macaques (N = 15, rs = −0.72, p = 0.002). In Tonkean macaques, ranks did not differ in their frequencies of pauses (Kruskall-Wallis test, H = 15.32, df = 9, p = 0.08). However, in rhesus macaques, mean frequency of pauses differed between ranks (H = 65.67, df = 9, p < 0.0001), with only rank 0 having a higher frequency than ranks 1 to 6 (Dunn’s multiple comparison test, p < 0.001). The first departed individual displayed 19% of all recorded pauses in Tonkean macaques and 34% in rhesus macaques.
If the speed is a recruitment signal, then the first departed individual and the first joiners should have a speed higher than the last joiners. However, the mean speed and rank were not correlated in Tonkean macaques (N = 10, rs = −0.08, p = 0.830) and in rhesus macaques (N = 15, rs = −0.40, p = 0.22). We found that mean speed did not differ between ranks in Tonkean macaques (Kruskall-Wallis test, H = 14, df = 14, p = 0.120) and in rhesus macaques (Kruskall-Wallis test, H = 14.52, df = 9, p = 0.153).
How did the number of joiners influence the behaviour of the first departed individual?
Based on the previous results, we analyzed further pauses and back-glances for the first departed individual only.
If pauses or back-glances are used by the first departed individual to recruit and/or monitor joiners as in white-faced capuchins (Meunier et al. 2007), their frequencies should decrease when the number of joiners increases. Frequencies of pauses should also decrease if pauses reflected some uncertainty. Indeed, the first departed individual would express its hesitation by waiting for other group members to join the movement. Thus when no individuals joined the first departed individual, it seems logical that this latter one tried to recruit group members by emitting signals and/or waited their joining. When the first individual is joined by a majority of individuals, it did not need to further use signals or wait.
The mean frequency of pauses and the number of joiners were not correlated in Tonkean macaques (N = 10, rs = −0.38, p = 0.248, Fig. 2b) whereas these were negatively correlated in rhesus macaques (N = 15, rs = −0.89, p = 0.0002, Fig. 2d). However, we found in Tonkean macaques that mean frequency of pauses differed according to the number of joiners (Kruskall-Wallis test, H = 72.2, df = 9, p < 0.00001). Indeed, the first departed individual displayed more pauses when no individual had joined the movement than when 2 to 8 individuals had joined it (Dunn’s multiple comparison test, p < 0.01, Fig. 2b). In rhesus macaques, mean frequency of pauses differed according to the number of joiners (Kruskall-Wallis test, H = 107.7, df = 14, p < 0.00001). The first departed individual displayed more pauses when no individual had joined the movement than when 1 to 13 individuals had joined it (Dunn’s multiple comparison test, p < 0.05, Fig. 2d). When it was not joined by any individuals, the first departed individual displayed 50% of pauses in Tonkean macaques and 90% in rhesus macaques. This result seems to show that the first departed individual might want to be joined and that pauses may be used to recruit and/or to wait for group-mates.
The link between the number of behaviours of the first departed individual and the number of joiners may be due to an influence of the time (the more elapsed time, the more the first departed individual can display behaviours and the more the other individuals can join the movement). However, the number of joiners was not correlated to the duration of a collective movement in Tonkean macaques (Spearman rank correlation, N = 10, rs = 0.333, p = 0.347) nor in rhesus macaques (N = 15, rs = 0.482, p = 0.058). This result suggests that the behaviours of the first departed individual probably have a direct influence on the number of joiners.
The relationship between the mean frequencies of pauses and back-glances displayed by the first departed individual and the number of joiners is not linear (Fig. 2a–d). Moreover, the first departed individual appeared to stop displaying these behaviours after 2–3 group members have joined. A curve estimation test shows that the best predictor of observed curve is an inverse curve for behaviours both in Tonkean macaques (back-glances: R2 = 0.90, df = 9, p < 0.00001; pauses: R2 = 0.89, df = 9, p < 0.00001) and rhesus macaques (back-glances: R2 = 0.92, df = 14, p < 0.00001; pauses: R2 = 0.73, df = 14, p < 0.0007). If the curve would be linear, this may suggest that the first departed individual wanted to recruit/to wait all group members. However, as the curve was not linear, it seems that the first departed individual did not recruit/wait for the whole group but only for some individuals.
Did the first departed individual modify its behaviour according to the number of joiners?
We checked whether the frequency of pauses decreased after a back-glance according to the number of joiners in the movement. Indeed, we believed that the first departed individual used back-glances to monitor conspecifics and thus estimate the number of joiners, whereas it used pauses as an encouragement for them to join, i.e. to recruit them. According to this assumption, the number of pauses after a back-glance would decrease with the number of joiners while the number of back-glances after a pause would not decrease with this number. An alternative explanation could be that these behaviours reflected internal motivational state of the first departed animal, like a hesitation or an uncertainty for going alone. According to this assumption, we would predict that both the number of pauses after a back-glance and the number of back-glances after a pause should decrease with the number of joiners.
Back-glances were not displayed always. As mentioned in the previous section, the first departed individual seemed to stop emitting signals after the joining of 2–3 joiners to the movement. They occurred only when there were 0, 1, 2, 4, 6, 7 and 8 joiners in Tonkean macaques (N = 7) and for 0, 1, 2, 4 and 7 joiners in rhesus macaques (N = 5). The following analysis will be conducted on these numbers. We checked whether the absence of signals for 3 and 5 joiners was due to an identity effect of the first departed individual, of the 3rd or of the 5th joiner. Results showed that the majority of individuals occupied each of these ranks and that there was no difference between individuals in both Tonkean macaques (Chi square test, df = 10, χ² ≤ 6.810, p ≥ 0.500) and rhesus macaques (Chi square test, df = 14, χ² ≤ 15.532, p ≥ 0.353).
Pauses after a back-glance
The frequency of pauses after a back-glance decreased with the number of joiners in Tonkean macaques (N = 7, rs = −0.78, p = 0.038) and in rhesus macaques (N = 5, rs = −0.89, p = 0.04).The mean number of pauses after a back-glance is reduced by half when one individual has joined the movement (whatever the species) and was nil when there were 6 joiners in Tonkean macaques and 7 joiners in rhesus macaques.
Back-glances after a pause
To ensure that the relationship between pauses and number of joiners was not an artefact due to a link with the duration of movement, we tested if the converse configuration i.e. frequency of back-glances after a pause and the number of joiners were also correlated. Such correlation was not found in Tonkean macaques (Spearman rank correlation, N = 10, rs = −0.43, p = 0.208, pauses were displayed for any number of joiners), and in rhesus macaques (N = 7, rs = −0.05, p = 0.806, pauses were only displayed for 0, 1, 2, 3, 4, 7 and 8 joiners). The mean number of back-glances after a pause was 0.13 ± 0.06 from 0 to 8 joiners in Tonkean macaques and 0.59 ± 0.19 from 0 to 7 joiners in rhesus macaques.
From these results, we suggested that the first departed individual used back-glances to monitor the number of joiners and then adapted its number of pauses according to this number or more particularly according to the presence of specific individuals (kin-related or affiliated) among these joiners instead of only waiting its conspecifics.
Did the first departed individual recruit specific group members?
When monitoring its group-mates, the first departed individual may not only monitor the number of joiners but also their identities. In the following section, the identities of participants are considered. We assessed if the relationships a first departed individual a had with a joiner b influenced it restarting after a pause. During a pause of a, if an individual b joins the movement, a can restart or stay in pause. For each a–b dyad, we scored the number of times a first departed individual restarted after a joining (n1) and the number of times it stayed in pause (n2). For both groups, the matrix of the ratios n1/(n1 + n2)ab was compared to the matrix of kinship and the matrix of affiliation using matrices correlation tests (Hemelrijk 1990).
In Tonkean macaques, the matrix of the ratios n1/(n1 + n2)ab correlated with the matrix of affiliation (Kr = 0.45, p = 0.0004) but not with the one of kinship (Kr = 0.21, p = 0.087). Conversely, in rhesus macaques, the matrix of the ratios n1/(n1 + n2)ab correlated with the matrix of kinship (Kr = 0.28, p = 0.004) but not with the matrix of affiliation (Kr = −0.15, p = 0.593).
These results confirm that the first departed individual specifically recruited and/or waited for some individuals, the highly affiliated ones in Tonkean macaques and the kin-related ones in rhesus macaques. Thus, it ceased its waiting and/or its recruitment as soon as these particular individuals had joined.
Our study was conducted in semi-free ranging conditions, where the visibility was higher and inter-individual distances lower than in natural conditions (Judge and de Waal 1997). Animals might thus rely only on visual signals (Gros-Louis 2004; Mitani and Nishida 1993). This may explain why we did not hear any loud calls in our studied group contrary to Riley (2005), who reported the use of a ‘loud call’ to initiate group movements and keep cohesion in Tonkean macaques. In both groups, results seemed to show that the first departed individual seemed to use pauses to recruit its conspecifics. Another explanation could be that pauses reflected internal motivational state of the first departed animals, such as hesitation, uncertainty or solely waiting for its conspecifics. On the other hand, first departed individuals seemed to use back-glances to monitor the number and/or the identity of joiners. The use of these behaviours is also similar for other species (chimpanzees, Menzel 1971; white-faced capuchins, Leca et al. 2003; Meunier et al. 2007; hyenas (Crocuta crocuta), Holekamp et al. 2000) and in another group of Tonkean macaques (Sueur and Petit 2008a).
The first departed individual seemed to modify its number of pauses according to the number of joiners evaluated by monitoring. The more the first departed individual displayed pauses, the more numerous were the joiners. The first departed animal might pause because it could feel reluctant to go alone and pauses could express animal’s uncertainty; when other individuals joined the movement, the uncertainty of the first departed individual probably decreased and it displayed fewer pauses. However, the duration of a collective movement did not depend on the number of joiners. This result suggests that the behaviours of the first departed individual may have a direct influence on the number of joiners and that making a pause does not mean merely waiting but is certainly a cue in itself for joiners. This result does not imply that the first departed individual should be conscious of the effect of its pause. In a second step, the first departed individual diminished its number of pauses. This feedback loop favoured group coordination. Sueur and Petit (2008a) suggested that the speed of the first departed individual could be a recruitment signal, as its intensity was positively correlated to the joiners’ number. In the same way, Leca et al. (2003) found that a low speed in white-faced capuchins favoured the joining of group members. However, in the present study, the speed of the first departed individual was not higher than the one of other group members, a result already found by Meunier et al. (2007) for white-faced capuchins. It has been suggested that a high speed of movement of an individual may rather be a cue of its goal-directed-movement or of its motivation (Altmann and Altmann 1974; Garber 1988; Noser and Byrne 2007; Pochron 2001; Sigg and Stolba 1981) than a real recruitment signal, even if speed seemed to favour the joining process. Indeed, Tonkean macaques are able to use cues conveyed by group-mates to localize food sources (Chauvin and Thierry 2005; Drapier et al. 2002; Ducoing and Thierry 2004) and may use speed as a simple cue.
We showed that pause seemed to be a cue for joiners but from the first departed individual’s point of view, a pause may be a recruitment signal or the expression of its uncertainty. Nevertheless, we found that the first departed individual ceased to display pauses when some particular individuals had joined. This finding seemed to go beyond the uncertainty hypothesis. If the first departed individual hesitated to go alone, then it should cease emitting pauses when joined whatever the identities of the joiners might be. However, it stopped displaying pause solely when some specific individuals joined it. According to this result, we suggested that pause would be more a recruitment signal than uncertainty index.
Recruitment may be a mechanism as simple as a threshold function, dependent on the number of joiners. However, we found that the recruitment process was not only dependent on the joiners’ number but also on their identities. Indeed, the first departed individual seemed willing to recruit mostly kin-related individuals in rhesus macaques and highly affiliated ones in Tonkean macaques and stopped displaying signals when this goal was reached. This difference between rhesus and Tonkean macaques reflects their different social styles (de Waal and Luttrell 1989; Thierry 2007). Rhesus macaques are more nepotistic than Tonkean macaques, as found in many of their daily interactions (Thierry 2004). As a consequence, rhesus macaques formed kin-related sub-groups during collective movements contrary to Tonkean macaques, for which associations during collective movements are based on affiliative relationships (Sueur and Petit 2008b). Such influence of social style is also found in the current study.
Tomasello and Call (1997) defined behaviour as intentional if it implies a goal and some flexibility in the means for attaining it. Ducoing and Thierry (2003) found such flexibility in Tonkean macaques, where informed subordinate individuals used tactical manoeuvres in order to reach a food source alone. Flack and de Waal (2007) found a similar result with context modulating dominance-related signal meaning in pig-tailed macaques (Macaca nemestrina). In the same way, loud calls are used to initiate movements in wild groups (Riley 2005) and when a pig-tailed macaque mother used the pucker behaviour to encourage its infant to join it, this behaviour has also been considered as intentional (Maestripieri 1996a, b). In this study, we suggested that the first departed individual used pauses to recruit specific group members. We previously reported that when not joined, the first departed individual went back to the group and started a new start attempt (Sueur and Petit 2008a). Moreover, Tonkean and rhesus macaques seemed to show their intention to move. Indeed, they indicated with their body position the direction in which they were willing to go before the departure of a collective movement (Sueur and Petit 2008a). So, macaques seemed to use different signals to recruit other group members: loud calls, pucker behaviour, intention movements, a new start attempt and maybe pauses. We may suggest that these behaviours are intentional (Tomasello and Call 1997). We are aware that our results came from two semi-free ranging groups and that further observations or experiments are needed to test the intentionality of this kind of behaviours in macaques. However, these conditions were essential to record the behaviour of each individual during observation sessions. Moreover, comparing two groups of two different species in the same environmental conditions allowed us to find similar rules explaining group cohesion during collective movements and gave robustness to our findings. This fact is corroborated by the interspecific differences we found in the identity of recruited congeners. Such differences are easily understandable by considering social styles of both species, as has been previously shown for conciliatory tendencies (Thierry et al. 2008), social play (Petit et al. 2008) and many other social behaviours (Thierry et al. 2004).
The authors are grateful to J. Dubosq, V. Wyss, H. Roger-Bérubet and A. Coulon for their help, P. Uhlrich, for technical assistance, N. Poulin for statistical advices and R. Knowles for English corrections. This work was supported by the French Research Ministry (EGIDE), the French Foreign Ministry (Lavoisier Excellence Scholarship) and the European Doctoral College of Strasbourg Universities. Thanks are extended to J. L. Deneubourg and B. Thierry for scientific discussions. These experiments comply with the current laws of the country in which they were performed.