1In their natural environment, some monkey species select particular plants to rub their pelage vigorously. This phenomenon, known as ‘fur rubbing’, has been reported in capuchin monkeys (Cebus spp.) (Baker, 1996, 1997, 2000; DeJoseph et al., 2002; Leca et al., 2007; Meunier, 2007; Meunier et al., 2007, 2008; Valderrama et al., 2000), in owl monkeys
(Aotus spp.) (Zito et al., 2003) and in black-handed spider monkeys (Ateles geoffroyi) (Campbell, 2000; Dare, 1974; Richard, 1970), but it has been particularly well studied in the white-faced capuchin monkeys (Cebus capucinus) (Baker, 1996, 1997, 2000; DeJoseph et al., 2002; Leca et al., 2007; Meunier, 2007; Meunier et al., 2007, 2008). Although it may also be involved in scent marking (Campbell, 2000), the items used in fur rubbing are purportedly functional in repelling insects or removing ectoparasites (Baker, 1996), and this behaviour has thus been mainly considered for its medicinal benefits (Baker, 1996; DeJoseph et al., 2002; Ludes and Anderson, 1995; Valderrama et al., 2000). Surprisingly, even though the self-medication function of fur rubbing has attracted considerable attention, only a few studies have addressed the underlying mechanisms of this behaviour. Some authors (Baker, 1997, 2000; Huffman, 2007) have studied the identification and selection of fur rubbing items by white-faced capuchins and examined the cognitive components of plant selection. Meunier et al. (2007) investigated the influence of the resource used to perform fur rubbing on the form of the collective behaviour (i.e. alone or in subgroups) and Leca et al. (2007) investigated the link between the form of fur rubbing and a number of social variables in two capuchin species (C. capucinus and C. apella). We also know that fur rubbing can be performed solitary, or collectively in subgroups of variable size and composition. When several individuals rub their fur simultaneously, they may be or not in physical contact (Baker, 1996; Quinn, 2004) but a real collective synchronization appears, including social facilitation, whatever the spatial distribution (body contact or not) of individuals rubbing their fur simultaneously (Meunier et al., 2008).
2When capuchins rub their fur in body contact of conspecifics, i.e. social fur rubbing (Baker, 1996; Quinn, 2004), they sometimes coil their semi-prehensile tail around the body of the capuchin(s) in contact (Leca et al., 2007) (Figure 1). The event of tail coiling behaviour was reported to be correlated to an increase in the duration of body contact (Leca et al., 2007). However, the underlying mechanisms of the tail coiling emission have never been studied previously. In this study, we examined the impact of socio-demographic factors on this behaviour. In particular, we focused on which individual demonstrates and receives tail coiling, when and in which context this behaviour is performed and, finally, we check for its recruitment role (i.e. does tail coiling induce a change in the duration of fur rubbing episodes for the receiver of the behaviour?). If the duration of fur rubbing episodes (during and after body contact) with emission of tail coiling is longer than fur rubbing episodes without tail coiling, then tail coiling could be considered as a recruitment behaviour. We finally discuss other hypotheses that could explain the role of tail coiling behaviour in fur rubbing.
3The group of white-faced capuchins was established in 1989 at the Primate Centre of Strasbourg University, France. The group contains 15 individuals from three separate lineages. Our experimental subjects were six adult females (ages 5, 6, 6, 11, 13 years-old, and one individual aged at least 20 years), two adult males (ages 6 and 12), two young males (ages 1 and 2), and two young females (ages 1 and 2); three individuals of less than one year old were not studied.
4The group was kept in a wooded outdoor area of 2332 m2 with natural vegetation and had free access to an indoor shelter. Commercial primate pellets and water were available ad libitum. Fresh fruits and vegetables were provided once a week, but not during testing.
Figure 1
Alpha male (on the right side in the picture) coiling its tail around the head of a juvenile (on the left). Both capuchins are rubbing their fur with oranges.
Mâle dominant (à droite sur la photo) enroulant sa queue autour de la tête d’un individu juvénile (à gauche sur la photo). Les deux capucins se frottent la fourrure avec des oranges.
5Observations took place between 14:00 and 17:30 from July to October 2005. Two observers collected data with two video cameras. The group was supplied with two types of plant materials: fruit of Citrus sp. (oranges) and bulbs of Allium sp. (onions). These two materials are regularly included in the diet provided once a week by the Primate Centre staff (previous to this experiment) and capuchins often eat or rub their fur with them. Eleven items were supplied per experiment and each of these items was cut up into four pieces. Sixty experiments were conducted: 30 with onions and 30 with oranges. The two resources were presented randomly and only one experiment was conducted per day.
6The resource was deposited in a 1 m² area on the park ground. Latencies for fur rubbing behaviour (including tail coiling) were calculated from resource deposition time (T0).
7We quantified affiliation by the frequency of body contacts among all identified dyads in the group. Body contacts were recorded from the inside of the enclosure using instantaneous sampling (scan) every 5 minutes (Altmann, 1974). We collected 292 scans out of all the testing periods. The affiliation score for each dyad was assessed using the number of scans during which the two partners were in body contact.
8To rank the individuals of the group in a dominance hierarchy, the avoidance behaviour and the unidirectional spontaneous aggressions were recorded. The ‘MatMan’ program (de Vries et al., 1993; de Vries, 1995, 1998) was used to calculate the improved index of linearity (Landau’s h’ index). Only the adults were considered as being hierarchically ordered. The analysis of the occurrence of 121 avoidance behaviours and of 31 unidirectional conflicts allowed adult subjects to be ranked (Landau’s linearity index h’ corrected for unknown relationships = 0.88, p < 0.004) (De Marco, personal communication). The dominance scores ranged from one, for the most dominant individual, to eight for the most subordinate.
9We carried out three measurements from the video recordings: a) focal animal sampling (Altmann, 1974) of fur rubbing, which allowed us to determine the duration of each fur rubbing behaviour; b) instantaneous scan sampling every 10 seconds (Altmann, 1974) recording the number and identity of subjects in the subgroup of focal animals performing social fur rubbing. We defined a fur rubbing subgroup as a group of individuals which rub their fur simultaneously and are separated by <1 m; and c) all occurrences of behaviour sampling (Altmann, 1974) of tail coiling emitted while fur rubbing (Figure 1). We noted their emission latency and the identities of the emitter and the receiver. Individuals were identified individually according to their morphological characteristics.
10Due to the potential effect of the type of resource (orange vs onion) on fur rubbing (Meunier et al., 2007), we analysed separately the results obtained with these two items.
11An episode of fur rubbing was defined as the moment when an individual began fur rubbing, until its cessation. For each individual, we obtained a total number of tail coiling behaviours emitted and received. Several matrices were built to analyse the effect of socio-demographic variables on tail coiling distribution between group members. We firstly reported, in an asymmetrical matrix, when one animal coils its tail around another monkey, i.e. the frequency of tail coiling for each dyad ‘emitter-receiver’. In two symmetrical matrices, we reported the time spent performing fur rubbing in body contact for each dyad and their respective affiliation score. Finally, we checked for the role of kinship in tail coiling by implementing, in a matrix, the degree of kinship using three types of dyads: non-kin, far-kin (siblings, half siblings, grandmother-grandchildren, aunt-nephew/niece), and close-kin (mother-offspring) dyads.
12A fur rubbing episode could be performed following several patterns, i.e. solitary, social or collective (see Baker, 1996; Quinn, 2004). In the event of tail coiling involving body contact between two partners, we focused on social fur rubbing (several individuals rubbing their fur and grouped spatially) and examined all the body contacts, recording their durations and the identities of individuals involved. To explore further the role of tail coiling, we compared body contact durations during fur rubbing with tail coiling to those without tail coiling. On the 443 and 389 events of tail coiling recorded with onions and oranges respectively, we only considered dyadic body contact in which only one tail coiling was observed (to eliminate the bias linked to the potential accumulative effect of several successive tail coiling events), i.e. 221 events with onions and 144 events with oranges. To examine the impact of tail coiling on fur rubbing duration, among these events, we took into account only those in which both partners fur rubbed. From these restrictions, 96 events with onions and 57 events with oranges were analysed (presented in the below results section). The number of body contacts during fur rubbing without tail coiling (control events) was 59 with onions and 9 with oranges.
13Matrix correlations were tested by using Matman (de Vries et al., 1993). We compared the asymmetrical matrix of emitter/receiver of tail coiling with its transposition receiver/emitter to test if this behaviour could be symmetrical (even a ‘response’ could be postponed or delayed). In Matman, we set the number of automatic permutations of matrices at 10000 and used Pearson’s correlation coefficient. Besides matrix correlations, we used the Spearman rank correlation test to investigate the impact of hierarchy on the number of tail coiling events emitted and received, and the Mann-Whitney U test was used to examine the effect of age on the number of tail coiling events emitted and received (Siegel and Castellan, 1988). Sex was not tested due to the low number of adult males in the group. Linear regression was used to investigate the link between the number of tail coiling events and the duration of body contacts during fur rubbing. A positive correlation between those two variables could mean that the higher the number of tail coilings, the longer the body contact lasts but could also be interpreted in a converse way, i.e. a long duration of body contact allows the emission of a high number of tail coilings, which therefore would not support the hypothesis of the recruitment function of tail coiling. To test the causal link between this two parameters and more particularly if tail coiling maintains or prolongs body contact, we compared the duration of body contact without tail coiling with the duration between a tail coiling emission and the end of body contact using the Mann-Whitney U test. In a same way, to test if tail coiling could maintain or prolong fur rubbing activity, we compared the duration of fur rubbing episodes without tail coiling (from the beginning of body contact) with the duration between a tail coiling event and the end of fur rubbing activity using the Mann-Whitney U test. We studied separately the emitter’s and receiver’s data. All tests were two-tailed and the significance level was fixed at 0.05.
14As illustrated in Figure 2a and 2b, even if all the individuals emited and received tail coiling events, differences in frequencies of emission and reception of tail coiling appeared between individuals. We tested if these differences could be explained by socio-demographic factors. Whatever the resource, we found no effect of hierarchy, age and kinship (Table 1) on the frequency of a tail coiling event. However, we found an effect of affiliation on tail coiling behaviour: the more the individuals are affiliated, the more they exchanged tail coiling. This result is significant when fur rubbing was performed with oranges (Pearson’s correlation r = 0.269; p = 0.037), while only a trend was observed when fur was rubbed with onions (Pearson’s correlation r = 0.195; p = 0.065).
Figure 2
Number of tail coiling emitted (2a) and received (2b) by each individual with onions (white bars) and oranges (black bars).
Nombre d’enroulements de la queue émis (2a) et reçus (2b) par chaque individu du groupe lors du comportement de frottement avec des oignons (en blanc) et des oranges (en noir).
Table I
Impact of socio-demographic factors on the number of tail coiling emitted and received.
Impact des facteurs sociodémographiques sur le nombre d’enroulements de la queue émis et reçus.
15Whatever the resource used, we found symmetry between emission and reception of tail coiling (onions: Pearson’s correlation r = 0.604; p = 0.002; oranges: Pearson’s correlation r = 0.701; p = 0.001).
16During the 39 hours of fur rubbing activity recorded with onions and 12 hours of fur rubbing activity recorded with oranges, we observed 443 tail coiling behaviours with onions and 389 with oranges, which corresponds to 11 and 32 tail coilings per hour of fur rubbing activity for onions and oranges respectively. With oranges, capuchins presented near three times more tail coiling behaviours per hour of fur rubbing than when they rubbed their fur with onions.
17Among all the body contacts observed during fur rubbing behaviour, only 16% with onions and 4% with oranges were without tail coiling. Even if several tail coiling events could be observed in a single body contact, there was nearly only one tail coiling per physical contact between two monkeys during a fur rubbing episode (62% with onions and 65% with oranges, Figure 3).
Figure 3
Distribution of the number of tail coiling events per fur rubbing episode, emitted in dyads in body contact.
Distribution du nombre d’enroulements de la queue par épisode de fur rubbing, émis par dyade en contact corporel.
18A significant linear regression was found between the duration of body contacts during fur rubbing and the number of tail coiling events when onions were used (F=13.76; r2 = 0.03682; p = 0.0002, Figure 4a) as well as for oranges (F=71.93; r2 = 0.2406; p < 0.0001, Figure 4b).
Figure 4
Body contact duration when fur rubbing according to the number of tail coiling events (4a: when using onions for fur rubbing; 4b: when using oranges for fur rubbing).
Durée des contacts corporels lors des comportements de frottement de la fourrure selon le nombre d’enroulements de la queue émis (4a : en utilisant des oignons pour se frotter la fourrure ; 4b : en utilisant des oranges pour se frotter la fourrure).
19For onions, we found that the duration of body contact without tail coiling is higher than the duration between tail coiling emission and the end of body contact (Mann-Whitney U test: U = 4502; N1 = 59; N2 = 221; p < 0.001). We did not find difference between durations for oranges (Mann-Whitney U test: U = 498; N1 = 9; N2 = 144; p = 0.245) (Table 2).
Table 2
Differences between durations of body contact during fur rubbing episodes without tail coiling and durations of body contact when fur rubbing since a tail coiling.
Différences entre les durées de contact corporel lors des épisodes de frottement de la fourrure sans enroulement de queue et les durées de contact corporel en se frottant la fourrure suite à un enroulement de queue.
20For the emitter, we found that durations of fur rubbing without tail coiling are higher than durations between tail coiling emission and the end of fur rubbing (Mann-Whitney U test: U = 4100.5; N1 = 118; N2 = 96; p = 0.001) for onions. With oranges we did not find any difference between these two durations (Mann-Whitney U test: U = 511; N1 = 18; N2 = 57; p = 0.980).
21For the receiver, we did not find any difference, neither for onions (Mann-Whitney U test: U = 4838.5; N1 = 118; N2 = 96; p = 0.067) nor for oranges (Mann-Whitney U test: U = 480.5; N1 = 18; N2 = 57; p = 0.687) (Table 2).
22To our knowledge, tail coiling has only been observed in white-faced capuchins during body contact between two individuals performing fur rubbing and until now, has been poorly documented. In this paper, we quantitatively described the behaviour of tail coiling and examined its function.
23Fur rubbing and tail coiling were reported for both onions and oranges. Nevertheless, some differences were found regarding the two food items. The patterns of social fur rubbing were previously observed to diverge strongly in relation to the resource used (Meunier et al., 2007). Indeed, capuchins performed more social fur rubbing with oranges than with onions (Meunier et al., 2007). We showed here that body contact does not necessarily involve tail coiling but this behaviour appears frequently and sometimes several times during a single body contact between two individuals. However, we found that most of the time, a body contact engages only one tail coiling event. But the time spent performing fur rubbing when in body contact was positively correlated with the number of tail coiling emitted. This tail coiling is, as a result, more frequent when capuchins rub their fur with oranges than with onions.
24We showed that tail coiling is emitted and received by all group members performing fur rubbing, including all social and demographic classes. We highlighted that affiliation could influence a tail coiling event between two individuals. Even if fur rubbing was mainly assessed for its medicinal benefits (Baker, 1996; DeJoseph et al., 2002; Ludes and Anderson, 1995; Valderrama et al., 2000), several authors enunciate the hypothesis that this behaviour may serve to reinforce social bonds (Baker, 1996; Leca et al., 2007). The affiliation influence we found on the tail coiling dyads are thereby concordant with this hypothesis. However, in our group, affiliated individuals spent more time performing fur rubbing when in body contact with other individuals, than non affiliated capuchins (Brotcorne, 2006) and the number of tail coiling events is correlated with the time spent fur rubbing in body contact. Thereby, it must be pointed out that both variables are positively correlated with ‘affiliation’ and could therefore influence each other.
25We found that the number of tail coiling events emitted during a single body contact was correlated with the duration of body contacts during fur rubbing, which is in accordance with the results of Leca et al. (2007). Indeed, it means that the higher the number of tail coilings, the longer the body contact lasts. However, as mentioned in the methods, such data could be interpreted in a converse way: a long duration of body contact allows the emission of a high number of tail coilings, which therefore would not support the hypothesis of the recruitment function of tail coiling. Moreover, our analyses reveal that the time elapsed since a tail coiling and the end of the body contact or the fur rubbing episode is not longer than the duration of a body contact or a fur rubbing episode without tail coiling. The fact that we found equalities between these durations could signify that tail coiling could maintain the body contact or the fur rubbing episode but does not support any recruitment hypothesis, as proposed in the Introduction. Even if recruitment of conspecifics in contact during fur rubbing could appear useful to optimize the efficiency of the collective behaviour (e.g. against ectoparasites), contacts between conspecifics does not seem necessary if individuals are synchronized. Indeed, synchronisation between group members could be sufficient and efficient by acting as a collective barrier to ectoparasite propagation in the group (Meunier et al., 2008).
26Two other functional explanations could be proposed for tail coiling. First, capuchins coil their tail around conspecifics but also around their own body, so using their tail and/or conspecifics’ tail may help capuchins to rub their distant body parts unattainable with their hands and feet. Capuchins coiling their tail around their own neck or head were sometimes observed (H.M., personal observation) and Baker (1996) described also capuchins using their tail, besides their hands and feet, to apply the plant material over various body parts or over their entire body. Second, the semi-prehensile tail of capuchins would be used to perform allo-rubbing of conspecifics in body contact, and could thus, such as grooming or manual/body contact, reinforce social bonds. Such allo-rubbing could thus occur more frequently among strongly affiliated dyads (or partners). This assumption is supported by the symmetry in emission and reception we found for tail coiling. This symmetry could be linked to the social organisation of the white-faced capuchin. Indeed, this species is described as highly social and presents strong affiliative bonds. Individuals show thereby a high rate of bidirectional aggression and conciliatory behaviour (Leca et al., 2002; Perry, 1996, 1998; Rose, 1994). This symmetry in tail coiling could reinforce the hypothesis of an allo-rubbing function of this behaviour, even if complementary studies are necessary to validate this functional explanation. Tail coiling being positively correlated with affiliation, it appears to further illustrate the reciprocity of affiliative bonds in this species.
27However, we are conscious that the size of our sample is limited to one social group and supplementary data from other groups are required to allow us to make strong generalizations.
28H.M. is thankful to ‘Institut de France’, ‘David & Alice van Buuren’ and ‘de Meurs-François’ foundations for supporting this work. J.L.D. is research associate from the Belgian national Funds for Scientific Research. H.M. would like to sincerely thank C. Schweitzer, R. Terramorsi and F. Brotcorne for their assistance during testing and F. Brotcorne for her assistance in data analysis. H.M. appreciates helpful comments and discussion with Pau Molina-Vila. Thanks are extended to Ruth Knowles for language editing. Lastly, authors thank the reviewers and the editor for their useful comments and suggestions. Observations and experiments with the animals were made in accordance with the CNRS guidelines.
29The authors have no conflict of interest to declare.