Mutual medication in capuchin monkeys – Social anointing improves coverage of topically applied anti-parasite medicines

Wild and captive capuchin monkeys will anoint themselves with a range of strong smelling substances including millipedes, ants, limes and onions. Hypotheses for the function of the behaviour range from medicinal to social. However, capuchin monkeys may anoint in contact with other individuals, as well as individually. The function of social anointing has also been explained as either medicinal or to enhance social bonding. By manipulating the abundance of an anointing resource given to two groups of tufted capuchins, we tested predictions derived from the main hypotheses for the functions of anointing and in particular, social anointing. Monkeys engaged in individual and social anointing in similar proportions when resources were rare or common, and monkeys holding resources continued to join anointing groups, indicating that social anointing has functions beyond that of gaining access to resources. The distribution of individual and social anointing actions on the monkeys’ bodies supports a medicinal function for both individual and social anointing, that requires no additional social bonding hypotheses. Individual anointing targets hard-to-see body parts that are harder to groom, whilst social anointing targets hard-to-reach body parts. Social anointing in capuchins is a form of mutual medication that improves coverage of topically applied anti-parasite medicines.

repellent for flying hematophagous insects 19 . These hypotheses are supported by the actions of substances that primates use to anoint in the wild and in captivity. Benzoquinone secretions from millipedes repel insects 19,20,21 and ticks 22 , formic acid from ants repels tick nymphs 23 , Piper plant leaves are traditionally used by people in Latin America to treat skin conditions 24 , onion (Allium cepa) oils kill cattle ticks (Boophilus annulatus) 25 and contain affective antimicrobial agents 26 , and compounds found in citrus fruit peel repel lone star ticks (Amblyomma americanum) 21 . Furthermore, wild Cebus capucinus in Costa Rica, wild Cebus olivaceus in Venezuela, and semi-free-ranging Sapajus sp. in Brazil, anoint more during the wet season when there are more flying insects 19,23,24 .
Despite evidence for the medical efficacy of such substances used for anointing, non-medicinal explanations for the behaviour have also been proposed. Neotropical primates have well-developed olfactory communication, and many species scent mark substrates (Heymann 2006 27 for a review). That scent marking can explain anointing in black-handed spider monkeys is evidenced by the observation that rubbing actions are limited to the chest, are more often performed by males, and are non-seasonal in their temporal distribution, despite seasonal differences in insect parasite abundance 5,6 . Similar hypotheses have been offered for anointing in capuchins, insofar as it may create a 'group scent' 24 .
An interesting observation relevant to these functional hypotheses is that Cebus 19,24,[28][29][30][31] , and owl monkeys Aotus 10 , often anoint 'socially' , when one monkey rubs against another during anointing. However, there is mixed evidence for social anointing in Sapajus. Leca et al. (2007) 32 and Paukner and Suomi (2008) 33 , found that their captive Sapajus rarely rubbed socially, and reported differences in the anointing behaviours of Sapajus and Cebus, hypothesizing that Cebus derives social benefits, such as the strengthening of social bonds, from fur rubbing, but that Sapajus does not. However, anting in Sapajus often involves numerous individuals 23 leading Alfaro et al. (2012) 14 to propose that the distribution and local abundance of anointing material determines the degree of sociality in capuchin anointing behaviours.
In a group of captive Sapajus sp. the frequency of aggression increased, and durations of affiliative behaviours decreased, following anointing with onion (Allium cepa) 33 . These findings are consistent with competition for resources. However, in a separate study, aggression towards animals that anointed in isolation also increased on their return to their group 34 , leading the authors to suggest a 'chemo-signalling' hypothesis in which there is a disruptive effect on olfactory communication. Since the effect was short-lived, we suggest it is also possible that dominant individuals might have been directing aggression at individuals that smelled of these resources as part of their normal despotic behaviour in controlling access to resources.
Regardless of the underlying functional reasons for anointing, there is undoubtedly a strong social element that has to be accounted for. Weeper capuchins (Cebus olivacious) anointed with millipedes 'without competitive friction' and age-sex classes that normally avoid each other came together to do so 19 . Individual Cebus might rub against other anointing monkeys simply to acquire the substance being applied, when resources are rare or at a low density in the wild (e.g. millipedes) 19 . Mutualism may also explain social anointing if it is a better way of covering hard-to-reach areas, such as between the shoulder blades, than individual anointing alone 31 , as has been shown for allogrooming in a range of primate species 35 .
Experiments on self-medicative behaviour in captive primates, outside of the context of disease, have been successfully employed to test hypotheses in greater detail than is possible in the wild 32,33,36,37 . Here we manipulate the abundance of an anointing resource given to two captive groups of Sapajus spp. to test key predictions derived from the main hypotheses for the functions of anointing and in particular, social anointing (Table 1).
Ticks and lice are partially controlled in primates by auto and social grooming 35,[38][39][40] . If anointing treats such skin parasites in capuchin monkeys, as in the 'medicinal hypothesis' (Table 1), we might expect individual (self) anointing to target areas that an individual has difficulty grooming, such as those not visible to itself 35,40 , and furthermore social anointing should target areas that are difficult for an individual to reach physically. On the other hand, the 'scent-marking hypothesis' predicts that different age sex classes will anoint at different rates, and the behaviour will be restricted to different body parts, as in Ateles 5 .
If the function of social anointing is to strengthen social bonds as in the 'social bonding' hypothesis, we predict no difference in the proportion of social anointing to individual anointing when resources are abundant or rare. This prediction is shared by the 'coordination of treatment' hypothesis 41 in which simultaneous medicinal treatment reduces re-infection of individuals, and by the 'mutual application' hypothesis 31 in which social anointing treats hard-to-reach areas, such as between the shoulder blades. Conversely, the rare resource hypothesis 19 predicts that because animals are socially anointing in order to obtain access to rare resources, social anointing will be rare when resources for anointing are abundant. Additionally, the 'social bonding hypothesis' predicts increased affiliative behaviour (e.g. grooming) following anointing sessions with more social anointing. The 'rare resource' hypothesis predicts that two monkeys that are both holding anointing resources will not form anointing dyads, whilst the 'mutual application' hypothesis predicts that monkeys holding anointing resources will continue to seek out other monkeys that hold resources, and that social anointing actions will target parts of the body that are inaccessible to a monkey rubbing individually.
We also address predictions generated by Paukner & Suomi's 33,34 observed changes in levels of aggression during and after anointing. The 'chemo-signalling' hypothesis, predicts changes in aggression during There will be no difference in the proportion of social anointing to individual anointing when resources are abundant or rare SUPPORTED There was no difference in the proportions of social anointing (in time or number of actions) between the rare and abundant resource conditions Monkeys will groom more immediately following sessions with more social anointing NOT SUPPORTED There was no significant difference between grooming rates immediately after rare resource (less social anointing) and abundant resource (more social anointing) conditions.
Rare resource hypothesis (Valderrama et al. 2000) Individuals without items are obtaining chemicals from the bodies of others because they do not have direct access to resources Social anointing will be much rarer, as a proportion of all anointing, when the anointing resource is abundant Treats hard-to-reach areas, such as between the shoulder blades, obtaining better coverage of topically applied medicines There will be no difference in the proportion of social anointing to individual anointing when resources are abundant or rare SUPPORTED There was no difference in the proportions of social anointing (in time or number of actions) between the rare and abundant resource conditions Animals holding anointing material will continue to seek out other anointing animals, and groups of monkeys in which more than one individual has an anointing resource will be common SUPPORTED Individuals holding onions socially anointed more often in groups of other monkeys that held onions Social anointing will target parts of the body that are inaccessible to an individual monkey and therefore achieve more complete coverage SUPPORTED Social rubbing actions on 'inaccessible' body parts were more frequent than on 'accessible' body parts. Aggression increases during and after rubbing through competition for access to resource pieces Lower-ranking individuals should anoint less than higher-ranking individuals to avoid aggression from higher-ranking individuals PARTIALLY SUPPORTED Subordinate adult males anointed infrequently in the rare resource condition, but frequently in the abundant resource condition Fewer pieces of resource should create more competition, and therefore more aggression, than more pieces NOT SUPPORTED There was no significant difference in the rates of aggression between the rare resource condition and the abundant resource condition

Dominance hypothesis
Increased aggression results from individuals re-affirming dominance relationships before and after the unusually close-proximity behaviour There will be more aggression when there is more social rubbing NOT SUPPORTED There was no significant difference in the rates of aggression between the rare resource (less social anointing) and abundant resource (more social anointing) conditions Table 1. Hypotheses, predictions and results for the functions of anointing and social anointing in Cebus and Sapajus, and for the social behaviour of Sapajus during anointing. It should be noted that the hypotheses are largely non-exclusive. and after rubbing because odours in the resource mask natural chemo-signalling, so levels of aggression will be different when resources, and therefore odour, are rare or abundant. However, aggression could conceivably increase or decrease. On the other hand, the 'dominance' hypothesis predicts higher levels of aggression after bouts when there is more social rubbing and the 'competition' hypothesis predicts that there will be more aggression during bouts when there are fewer anointing resources, and that lower-ranking individuals will anoint less than higher-ranking individuals to avoid aggression.

Study animals
The

Method
Between July and November 2011, we replaced onions and related foods (garlic & leeks) from the diet with alternative foods. We tested each capuchin group on dry days twice a week by introducing pieces of onion to a defined area (approx 5 × 5 m) within the outside enclosure. We presented two experimental conditions on different occasions; a 'rare-resource' condition with one half of a large onion for each group, and an 'abundant resource' condition with half a large onion for each non-infant monkey in the group. We tested the groups one after the other in a randomly counterbalanced order. Successive sessions were separated by at least 48 hours. In each session, we completed one 'focal individual follow' 46 , filming one focal individual with a video camera for 45 minutes from the introduction of the onion. We made one complete 45-minute focal follow for each non-infant monkey for each condition regardless of their anointing activity (a total of 28 sessions for the west group and 20 for the east group). Animals could move freely between the inside and outside enclosures and were filmed from the closest possible viewing points 45   not. Since anointing could involve frequent pauses, we defined an anointing bout as starting from the first rubbing action with onion, and finishing one minute after the last rubbing action.
We defined 'rubbing actions' as events where a body part or onion came into contact with and moved across the surface of another body part. We also recorded when carried babies were rubbed in this way. We discriminated eight regions of the capuchin body ( Fig. 1) of similar surface area (following Zamma 2002 40 ); the tail, stomach and groin, chest, hind limbs and fore limbs were defined as 'accessible' and 'visible' to the monkeys, the rump and lower back, and head were defined as 'accessible' and 'non-visible' and the upper back and shoulder blades was defined as 'inaccessible' and 'non-visible' , because they could not reach the area with their hands or see the area themselves. One research assistant recorded the number of rubs on each body part from the videos. Because rubbing actions were sometimes very fast and sustained, variable visibility reduced the coders' ability to accurately record the frequency of individual actions, and the monkeys frequently changed their position and body parts being rubbed, '1' was scored for a body part at the start of any continuous sequence of rubbing actions on that body part. Each rubbing action was classified as 'social' when the focal monkey's body part was rubbed by or on another monkey or onion held by it, or 'individual' when the body part was rubbed by the focal monkey's own body part or onion. Thus 'social rubs' are actions within the behaviour of 'social anointing' as defined above.

Aggression and affiliation. MB recorded all instances of aggression directed to or from the focal individuals for 45-minutes following the introduction of the onion, including:
Threaten. Open mouth, bared teeth, eyebrows raised and ears flattened, and direct staring towards another monkey, usually with rapid forward movements 47 , may include branch shaking or breaking and banging objects (pushed or pulled with hands, feet and/or tail). No physical contact is made.
Chase. Runs towards another monkey, displacing them with threats (see above) and/or aggressive vocalisations 47 , and without facial expressions associated with play 47 . No physical contact is made.
Attack. Contact, including biting, hitting, grabbing and pushing, accompanied by threats (see above) and/or aggressive vocalisations 47 , and without facial expressions associated with play 47 .
MB recorded the total time spent in social grooming 47 immediately following anointing bouts for each focal monkey as a measure of affiliation. This allowed us to look for short-term, but not long-term changes in affiliation.

Results
Monkeys anointed (including individual and social anointing) for longer in the abundant resource condition (mean + /− SD = 615 s + /− 366 s, N = 24) than in the rare resource condition (mean + /− SD = 243 s + /− 371 s, N = 24) (ANOVA, F = 12.26, p = 0.00104) (Fig. 2a). Of the 24 focal individuals, 15 anointed in both conditions, one in only the rare resource condition, seven in only the abundant resource condition and one, a juvenile female, did not rub in either condition (Fig. 3). Only 2 out of 6 subordinate (non-alpha) adult males anointed in the rare resource condition, and those only for short durations. However, in the abundant resource condition, all subordinate adult males anointed at similar In the rare resource condition, fewer monkeys anointed, but those that did often scavenged pieces of onion from other anointing monkeys, and anointed both socially and individually.
In the abundant resource condition, where monkeys had the opportunity to choose whether to anoint socially or individually (on no occasion did a monkey take two onion halves at the same time, so resources were available for all individuals), focal individuals holding onions socially anointed more often in groups of other monkeys that held onions (mean + /− SD = 285.5 s + /− 265.1 s, N = 24) than    with those that did not (mean + /− SD = 122.1 s + /− 166.9 s, N = 24) (ANOVA, F = 6.22, p = 0.0163) (Fig. 2b) indicating that the attraction was motivated by more than simply access to resources.
Areas of the body that are non-visible to an individual monkey were subject to more rubbing actions (mean + /− SD = 77.8 + /− 73.4, N = 22) than body parts that are visible (mean + /− SD = 26.6 + /− 23.9, N = 22) (matched pairs t-test, t = − 4.456, p = 0.00011) (Figs 4 & 5). Individual and social anointing rubbing actions focused on different body parts (Fig. 5). Individual anointing focused on the head and lower back, whilst social anointing actions focused on the arms, upper and lower back and the chest. A greater percentage of social anointing rubbing actions were made on 'inaccessible' body parts (mean + /− SD = 30% + /− 17, N = 14) than for individual anointing rubbing actions (mean + /− SD = 14% + /− 10, N = 14) (matched pairs t-test, t = − 4.247, p = 0.00095) (Fig. 6). Notably, social rubbing actions reached dorsally between the shoulder blades (coded within the upper back category), an area that capuchin monkeys cannot reach with their hands (feet were only rarely used in anointing and by only a few individuals), thus only social rubbing achieved complete coverage of the body.
There was also no significant difference in time spent grooming between the rare resource condition (

Discussion
Anointing is an energetically costly behaviour that can last over 20 minutes in captive and wild Cebus and Sapajus 14 , considerably longer than any other recorded scent marking behaviour in any species of neotropical primate 27 . Wild capuchins may engage in anointing every two days 14 . Since the behaviour must be costly, we would expect it to provide significant benefits to the subjects. Olfactory communication has been suggested as one explanation for anointing 14,33,34 , but a convincing argument for what the animals are communicating remains to be offered. The mechanism by which anointing with strong smelling substances could positively affect olfactory communication, and thus improve an individual's fitness, is not clear, unless rules as simple as 'stronger smelling is more attractive' dictate capuchin monkey olfactory communication systems. It has also been argued that capuchins might seek to mask their olfactory identity 33 , but in our study, all age-sex classes engaged in anointing, including dominant males, and occasionally young infants that were not subject to focal follows; both of these age-sex classes seem likely to benefit from a clear olfactory message of identity. Since olfactory signals evolved to benefit the signaller, it is hard to see how masking these could be advantageous in natural conditions. We saw no difference in time spent fur rubbing between age-sex classes, and monkeys' anointed all body parts. Medicinal explanations for anointing are better supported by our results (Table 1); monkeys anointed areas that are non-visible to them more than areas that are visible to them. These areas are harder for an individual monkey to groom, so if anointing is an alternative therapy for ticks and lice, we might expect these areas to be targeted more, as we see in our experiments.
The social anointing behaviours in capuchin monkeys may have lead to the formulation of the social bonding hypotheses for the function of anointing in the species. The behaviour is certainly in need of explanation, since the benefits of anointing socially must outweigh not only the costs of sharing resources, but also the costs of potentially being subject to aggression in the case of subordinate monkeys, or of curbing aggression for dominant individuals. The Sapajus groups in our study anointed socially in every experimental session, and proportions of social rubbing relative to individual rubbing were high (53% in the abundant resource condition and 66% in the rare resource condition) relative to reports for both wild and captive Cebus; Curú (C. capucinus, wild) 54.7% 24 19 . Bouts in our groups typically included salivating 48 and tail coiling 19 , behaviours previously only observed in anointing Cebus, whilst overall rates of aggression were low. These results, along with observations of wild tufted capuchins anting 14 , and frequent informal observations of social anointing with lime fruits in our study group, before and after our experiments (Supplementary Electronic Resource 1), are at odds with many of the reported differences in anointing between these genera. Larger quantities of onion (and therefore odour) did not lead to increased aggression, and lower-ranking subordinate adult male monkeys anointed as much as dominant male and adult female monkeys when they had access to resources, so the 'interference with olfactory communication' hypothesis 33,34 is not supported (Table 1).
Limiting the anointing resources did not lead to increased rates of aggression through competition. Competition for anointing materials, unlike competition for food, appears to be modulated by the benefits of sharing the resources. We found that anointing generally occurred with little aggression in our study groups. Our study groups were stable and had spacious enclosures, much like the captive Cebus capucinus in anointing studies 32 , which may have led to more natural anointing behaviour than in groups living in small enclosures 33 . We found no difference between the frequencies of social grooming immediately following bouts with more or less social anointing, which does not give us evidence for any short-term change in social behaviour. However longer-term changes could accumulate in groups that anoint frequently, and our tests do not exclude this possibility. Medium and long-term changes in affiliation after anointing could appropriately be measured using a social networking approach 49 .
The fact that there was no difference in the proportions of social anointing and individual anointing when resources were either abundant or rare, and the fact that individuals with materials continued to be attracted to other anointing individuals, suggest that there is more to social fur rubbing than individuals gaining access to rare resources 19 . Instead, the results showing that social anointing resulted in a more complete coverage of the body with anointing materials lends support to the mutual application hypothesis, in which social anointing leads to better coverage by medicinal substances (Table 1). Increased coverage on the upper back (and the upper arm or shoulder) may result from self-directed rubbing against other anointing individuals that may be a secondary source of material, but more frequent rubbing on the chest in social anointing results from individuals actively rubbing other individuals. The chest is very accessible to them, and is typically quickly saturated during anointing, since materials are held against the chest. Furthermore, young carried infants, that were not yet making anointing actions, were frequently anointed in this way. Thus, we report animals actively medicating other group members (see Supplementary Video S1), as observed for Cebus capucinus in Lomas Barbudal 31 .
Neither an anointer's awareness of the medicinal action of materials, nor an intention to medicate, are implied by a functional medicinal explanation for anointing and social anointing. They may well be explained in terms of innate behaviours 50 , but alternatively, Meunier et al. (2008) 41 showed that anointing in Cebus capucinus is a 'collective behaviour with a mimetic underlying mechanism' . These authors suggest that synchronised anointing within a group could be advantageous to reduce re-infection rates, as with many parasite treatments. Social anointing might additionally facilitate the transfer of preferences for particular anointing materials, through social learning 14 , leading to learned differences between groups in wild populations 31 . We additionally propose that social anointing physically treats other individuals with the substances (a la Perry 2008 31 ), with protective benefits to the self, through group hygiene and reduced re infection, and to likely kin in the group. We conclude that social anointing in capuchin monkeys is a form of mutual medication that improves coverage of topically applied anti-parasite medicines for both individuals and groups of capuchins.