RESEARCH ARTICLE

Reconciliation in Wild White-Faced Capuchins(Cebus capucinus)

JOSEPH H. MANSON1–3*, SUSAN PERRY1–3, and DANIEL STAHL1

1Max Planck Institute for Evolutionary Anthropology, Leipzig, Germany2Department of Anthropology, University of California–Los Angeles, Los Angeles,California3Center for Behavior, Evolution, and Culture, University of California–Los Angeles,Los Angeles, California

The likelihood of reconciliation (defined as preferential peaceful contactamong former opponents following conflicts) has been predicted to varypositively with relationship value and compatibility, and negatively withrelationship security. Long-term data on wild white-faced capuchins(Cebus capucinus) indicate that dyads consisting of an adult female andan alpha male have high value and compatibility, but low security. Twostudies of C. capucinus postconflict (PC) behavior were carried out atLomas Barbudal Biological Reserve, Costa Rica. One study consisted of30-min PC and matched control (MC) follows. The second study extractedPC and MC periods from long follows, yielding PC/MC periods averaging105 min. In study 2, but not study 1, significantly more PC/MC pairs wereattracted (former opponents affiliated with each other sooner in the PCperiod than in the MC period) than were dispersed (former opponentsaffiliated with each other sooner in the MC period than in the PC period).Reconciliation in study 2 could not be explained as a by-product of formeropponents’ tendency to seek affiliative contact with conspecifics gener-ally, or of the spatial proximity of opponents following conflicts.Attempted reconciliation was less likely to be followed by renewedaggression when reconciliation attempts were delayed following conflicts.The data were insufficient for a formal test of differences in conciliatorytendency (the difference between the number of attracted and dispersedPC/MC pairs, divided by the total number of pairs) among dyad types tobe conducted. Am. J. Primatol. 65:205–219, 2005. r 2005 Wiley-Liss, Inc.

Contract grant sponsor: L.S.B. Leakey Foundation; Contract grant sponsor: National ScienceFoundation; Contract grant sponsor: National Geographic Society; Contract grant sponsor: SigmaXi; Contract grant sponsor: Rackham Graduate School; Contract grant sponsor: Evolution andHuman Behavior Program; Contract grant sponsor: University of Michigan Alumnae Society;Contract grant sponsor: Wenner-Gren Foundation; Contract grant sponsor: UCLA AcademicSenate.

*Correspondence to: Joseph H. Manson, Max Planck Institute for Evolutionary Anthropology,Deutscher Platz 6, D-04103 Leipzig, Germany. E-mail: manson@eva.mpg.de

Received 24 November 2003; revised 13 May 2004; revision accepted 17 September 2004

DOI: 10.1002/ajp.20110Published online in Wiley InterScience (www.interscience.wiley.com).

r 2005 Wiley-Liss, Inc.

American Journal of Primatology 65:205–219 (2005)

INTRODUCTION

In group-living animals, conflicts among companions may threaten beneficialrelationships, and it is therefore expected that psychological and behavioralmechanisms will evolve that function to repair relationships that have beendamaged by conflict [de Waal, 1989, 2000] (but see Silk [1996]). Reconciliation,which is defined as preferential peaceful contact among former opponents [deWaal & van Roosmalen, 1979; de Waal & Yoshihara, 1983], has now beendocumented in over 25 primate species and three other mammalian species[Aureli et al., 2002].

Cords and Aureli [1993, 2000] developed a set of predictions concerning thequalities of dyadic relationships that are likely to influence reconciliation rates.Relationships vary in 1) ‘‘value,’’ or fitness benefits to the partners (which maydiffer between the two partners); 2) ‘‘security,’’ or the perceived probability that arelationship will change; and 3) ‘‘compatibility,’’ or the general tenor of thepartners’ social interactions. The relationship value is predicted to be positivelyassociated with the likelihood of reconciliation, because animals have more to loseby allowing valuable relationships to be damaged by conflict. Indeed, reconcilia-tion is expected to occur only in valuable relationships. Security is predicted to benegatively associated with the likelihood of reconciliation, because insecurerelationships are more likely than secure ones to be damaged by unreconciledconflicts, and therefore insecure partners have more to lose by failing to reconcile.Compatibility is predicted to be positively associated with the likelihood ofreconciliation, because animals that associate frequently are more likely to beable to engage in friendly interactions, even in tense postconflict (PC) contexts. Alarge number of studies [reviewed by Aureli et al., 2002] have found evidence thatreconciliation is most common in dyads with valuable relationships, as measuredby kinship and/or the relative occurrence of friendly vs. aggressive interactionsunder baseline conditions, or (in one case) experimental manipulation ofrelationship value [Cords & Thurnheer, 1993]. However, little is known aboutthe effects of interdyadic variation in security or compatibility on the likelihood ofreconciliation.

In this paper, we examine whether wild white-faced capuchins (Cebuscapucinus) reconcile following escalated conflicts, and we test predictions fromCords and Aureli’s [1993, 2000] model regarding which dyad type is most likely toreconcile. Although three studies have examined whether captive or semi-free-ranging capuchins reconcile [Leca et al., 2002; Verbeek & de Waal, 1997; Weaver& de Waal, 2000], there have been no published studies of PC behavior in any wildplatyrrhine.

White-faced capuchins live in multimale, multifemale groups. Females arephilopatric. No females have immigrated into three study groups at LomasBarbudal, Costa Rica, during 24 group-years of observation (unpublished data; animmigration reported by Manson et al. [1999] was proven via genetic testing to bea case of mistaken identity). In contrast, males disperse from their natal groups,and periodically disperse later in life (unpublished data) [Fedigan, 1993; Jack &Fedigan, 2004a, b], often as part of a migration alliance. Replacement of breedingmales by invading individuals or alliances is associated reliably with an increasein infant mortality [Fedigan, 2003] (see also Manson et al. [2004]). Some adultfemale–female dyads, particularly those of similar dominance rank, form strongand often long-lasting affiliative relationships, as measured by frequencies ofallogrooming, proximity, and benign infant handling [Manson, 1999; Mansonet al., 1999; Perry, 1996a]. Female–female aggressive coalitions sometimes appear

206 / Manson et al.

to maintain female dominance hierarchies [Manson et al., 1999], and to increasefemales’ power relative to that of adult males, who are usually individuallydominant to females [Perry, 1997]. On two occasions, circumstantial evidence hassuggested that adult females played a key role in deposing (and in one case fatallywounding) an alpha male [Gros-Louis et al., 2003; Perry, 1998a]. Alpha males’grooming and proximity frequencies with adult females are much higher thanthose of subordinate males [Perry, 1997]. Male–female aggressive coalitionsmostly include the alpha male, and are usually directed against subordinatemales. Although male–male aggressive coalitions are important for bothintergroup [Jack & Fedigan, 2004b; Perry, 1996b] and intragroup [Perry,1998b] competition, adult males generally spend very little time grooming or inclose proximity to other adult males.

From these findings and from Cords and Aureli’s [1993, 2000] model, wederive the prediction that adult dyads consisting of a female and the alpha maleshould be more likely to reconcile following conflicts than other dyad types.Female–alpha male relationships have high value to both parties: females benefitfrom alpha males’ protection against subordinate males’ aggression and theprospect of takeovers by invading alliances of potentially infanticidal males, whilealpha males need female support to maintain their position. These relationshipsalso have low security: adult females sometimes turn against alpha males andparticipate in their overthrow, and alpha males sometimes emigrate voluntarilywhen more fertile females are available elsewhere. Based on demographic datacollected from six C. capucinus groups for up to 14 years at the Santa Rosa site,Jack and Fedigan [2004b] reported that alpha males voluntarily emigrated after amedian tenure of 54 months, whereas alpha males that were evicted from theirgroup experienced a median tenure of 33 months. The alpha male tenures at oursite are similar to those reported by Jack and Fedigan [2004b] (unpublished data).Because life expectancy in this species is comparable to that of great apes (50+years in captivity [Hakeem et al., 1996]), a typical female experiences relation-ships with multiple alpha males during her lifetime. Finally, female–alpha malerelationships have high compatibility, as indicated by their high grooming andproximity rates. Although female–female relationships have high value and highcompatibility, they are more secure than male–female relationships. Male–malerelationships have high value and low security, but low compatibility. Finally,relationships between females and subordinate adult males have low value. Wetherefore predict that female–subordinate male dyads will show the lowestreconciliation rate, whereas female–female and male–male dyads will showreconciliation rates intermediate between female–alpha male and female–subordinate male dyads.

MATERIALS AND METHODS

We collected data from two social groups of habituated white-faced capuchinsin Lomas Barbudal Biological Reserve, Costa Rica, and on nearby public andprivately owned land (101 29–320 N, 851 21–240 W). We habituated Abby’s group in1990, and J. Gros-Louis habituated Rambo’s group in 1996. Group compositionsare shown in Table I. The area consists largely of highly seasonal, tropical, drydeciduous forest traversed by several permanent streams [Frankie et al., 1988].We followed social groups from dawn to dusk every day for up to 25 sequentialdays per month. Adults and large juveniles were easily identified by scars, haircolor patterns, and facial contours. Small juveniles were marked with ClairolBorn Blondet hair dye squirted at their backs from a distance of 1–2 m by means

Reconciliation in Cebus capucinus / 207

of a 100-cc syringe from which the needle had been removed. This procedurenever produced noticeable distress in the subjects.

Data Collection: Study 1

Between May 1991 and May 1993, we conducted 10-min focal individualfollows on all individually recognized members of Abby’s group, rotating throughthe group to ensure unbiased sampling of individuals (see Perry [1996a, 1997] fordetails). Four observers collected data, and interobserver reliability wasestablished as described by Perry [1995]. All adults were recognized from thebeginning of the study, and juveniles were gradually recognized throughout late1991 and early 1992. Immediately following conflicts in which one participantscreamed, we discontinued any ongoing focal follow and began a 30-min PC followin which one of the participants was observed continuously and all of its socialinteractions were recorded. The beginnings of all conflicts used for PC followswere seen as well as heard, and all of these conflicts included other aggressive orfearful behavior, such as lunges, chases, and/or fleeing between clearly identifiedopponents. We usually conducted the PC follow on the victim (defined as therecipient of the first aggressive behavior between the opponents, even if thevictim later responded with aggressive behavior). When two or more monkeysformed a coalition against a single victim, we designated one individual as thevictim’s primary opponent, based on the level of aggression or the order in whichthe attackers began the aggression. At 2.5-min intervals during PC follows, werecorded the focal animal’s activity and its proximity to all conspecifics in view,using four categories: contact, within one monkey body length (see below), withinone to five lengths, within five to 10 lengths, and within 410 lengths but visible.PC follows were discarded if the focal animal disappeared from view before 30min had elapsed. On the next observation day, we conducted a 30-min matchedcontrol (MC) follow, using an identical protocol, of one of the conflict participantsat the same time of day (71 hr). If either participant had been involved in aconflict during the 1 hr preceding the planned MC follow, the MC follow waspostponed for 1 day. The MC follow was discarded and then attempted again onthe following observation day, if the original conflict participants were not in viewof one another at some point during the MC follow.

Data Collection: Study 2

Between February and June 2001, we attempted to conduct all-day focalindividual follows of the parous females and the adult and large subadult males of

TABLE I. Compositions of Study Groups

Study 1 Study 2 Study 2(Abby’s) (Abby’s) (Rambo’s)

Adult malesa 3–4 3–4 4Adult femalesb 5 10 8Subadult malesc 0 1 3Immaturesd 11–13 16–19 18

a10+years old.bParous females (usually 6+ years old).c7–9 years old.dNulliparous femals plus males 0–6 years old.

208 / Manson et al.

Abby’s group and Rambo’s group. We rotated through the available pool ofsubjects. Daily observations began as soon as an eligible focal animal was sightedand lighting conditions were adequate for the reliable identification ofindividuals, and continued until low light conditions prevailed late in theafternoon or the focal animal was lost for an extended period any time after 1500hr. As in study 1, we recorded all social interactions of the focal subjects, andrecorded the subjects’ activities and proximities to conspecifics at 2.5-minintervals. However, we recorded proximities only of conspecifics within 10monkey body lengths (see below) of the focal subject. As in study 1, four observerscollected data, and interobserver reliability was established as described by Perry[1995].

Behavioral Definitions

Aggressive behaviors included open-mouth threats, vocal threats, lunging,chases, hitting, wrestling, and biting. An intense vocal threat [Perry, 1995](possibly equivalent to the chortle [Oppenheimer, 1973]) consisted of a sequenceof loud bursts of sound accompanying a threat face. Screams were very loud, high-pitched vocalizations given exclusively during agonistic interactions. At leastthree types of C. capucinus screams can be discriminated by the human ear[Perry, 1995], but we do not yet have evidence that these are functionally distinct.As in previous work on this population [e.g., Perry, 1996a], we measuredproximity using as a metric the length from the nose to the tail base of an adultmale capuchin (approximately 40 cm (K. Glander, personal communication)).Affiliative behaviors included the following activities:

1. Approaches to within one body length in which the approacher produces noaggressive vocalizations or postures or facial threats.

2. Grooming or presenting a body part for grooming.3. Affiliative vocalizations (trilling, gargling, and peeping [Gros-Louis, 2001,

2002; Perry, 1995, 1998b] (unpublished data)).4. Nongrooming affiliative contact, including inspecting, nuzzling, touching,

fur-rubbing [Baker, 1996] in contact, playing, displaying interest in food, andallowing a conspecific to handle one’s infant.

5. Dancing, a behavioral pattern that includes pirouettes (a monkey fixes itsgaze on another monkey, twists its head from side to side, spins its body around,and looks over its shoulder or under its arms or legs at the other monkey) andpacing (a monkey runs back and forth while staring at another monkey). Dancesoccur as a component of courtship, and in both same- and opposite-sex dyadsduring socially tense situations [Manson et al., 1997; Perry, 1995]. Differentvocalizations characterize the two contexts. Because dances often culminate inmounting and thrusting [Manson et al., 1997; Perry, 1995], we regard them as anattempt to reduce interindividual distance, and therefore as a form ofreconciliation when observed in PC contexts.

6. Sexual behavior, i.e., copulatory mounting and thrusting, as well aspseudocopulatory mounting and thrusting between same-sex individuals.

7. Forming a coalition against a third party using the overlord [Oppenheimer& Oppenheimer, 1973] or ‘‘double threat’’ [Fedigan, 1993] posture (one monkeylies on another monkey’s back and clasps its hands around the bottom monkey’schest so that their heads are stacked one on top of the other embrace, cheek-to-cheek posture, and/or mutual headflagging (see Perry [1995, 1996a] for moredetailed descriptions). Because coalitions usually involve physical contact

Reconciliation in Cebus capucinus / 209

between partners, we regarded coalition-formation behaviors, like dances, asattempts to reduce interindividual distance.

Data Analysis: Study 1

Each PC/MC pair was classified as attracted, dispersed, or neutral [de Waal &Ren, 1988], depending on the relative timing of the first affiliation between theopponents during the PC and MC follows. Attracted pairs were those in whichlatency to first affiliative interaction in the PC was shorter than latency to firstaffiliative interaction in the MC, or in which opponents affiliated during the PCfollow but not during the MC follow. Dispersed pairs were those in which latencyto first affiliative interaction in the MC was shorter than latency to first affiliativeinteraction in the PC, or in which opponents affiliated during the MC follow butnot during the PC follow. Neutral pairs were those in which opponents did notaffiliate in either the PC or the MC follow, or latency to first affiliation was equalin the PC and MC follows.

Data Analysis: Study 2

To ensure that the results would not be biased by interactions that occurredwhile the focal subjects were out of view, we divided the follows into segmentsusing the following criteria: Any break in continuous observation of 45 min wasregarded as a break between segments. Breaks of r5 min were regarded asbreaks between segments only if 1) any conspecific was within five body lengths ofthe focal subject at the moment when the subject disappeared from view orreappeared, or 2) any scream or vocal threat was heard during the break thatcould possibly have been emitted by or directed toward the focal subject.

For our data set of conflicts, we extracted from the focal individual follows allconflicts involving screaming or intense vocal threats (21% of study 2 conflictsinvolved intense vocal threats but no screams). Conflicts were excluded from thesample if they occurred o30 min before the end of a segment. We initiallydesignated as PC periods the time from the occurrence of the conflict until theend of the current segment. MC periods were culled from other follows ofthe opponents [Aureli, 1992; Matsumura, 1996], and defined a posteriori by thefollowing procedure: The designated MC period was drawn from the focal follow(on a different day) of one of the conflict opponents on the day closest to theconflict (either before or after the conflict) in which 1) neither opponent wasabsent from the group, 2) continuous observations (i.e., uninterrupted by breaksbetween segments) of at least 30 min duration were carried out at a time of daywithin 1 hr of the time of the conflict, and 3) no aggression occurred between theopponents on the day of the MC period before or during the MC period. MCperiods began at a time of day as close as possible to the time of the conflict. Weexcluded conflicts from the sample if we could find no MC periods in the entiredata set that met criteria 1–3. MC periods potentially lasted until the end of thesegment in which they began. Finally, when the PC and MC periods (asestablished by this procedure) were of unequal lengths, we truncated the longer ofthe two periods so that the PC and MC periods of each matched pair were alwaysof equal length.

We scored coalitionary conflict incidents as multiple conflicts between thevictim and each of its opponents.

We designated PC/MC pairs as attracted, dispersed, or neutral, using thesame procedure as in study 1.

210 / Manson et al.

To test for the occurrence of selective PC affiliation, we used each victim as adata point and scored whether he or she engaged in more attracted than dispersedPC/MC pairs or vice-versa. We then compared these scores using a sign test. Webased our procedure on that used by Matsumura [1996], with the modificationthat we used a sign test rather than a Wilcoxon signed-rank test because theformer is more appropriate when the data set contains small numbers of PC/MCpairs per victim (see below). As a complementary technique, we also carried out aform of survival analysis (Kaplan-Meier product-limit method, using Cox’s F-testto compare two samples), in which each period (PC and MC) constituted a datapoint, and the test determined whether the distribution of latencies to theopponents’ first affiliation was shorter in one type of period than the other. Asurvival analysis uses data from ‘‘censored’’ intervals (i.e., PC and MC periodsthat ended before any affiliation occurred between the two opponents). FollowingVeenema et al. [1994], we calculated the corrected conciliatory tendency (CCT)for each dyad as:

ðAttracted paris � dispersedpairsÞTotal PC=MS pairs

RESULTS

Characteristics of the Data Sets

In study 1, we collected 70 PC/MC pairs involving 45 distinct dyads and 19different victims (mean 7 SD PC/MC pairs per victim = 3.7 7 1.9, range = 1–7).In study 2, we carried out follows on 72 days, sampled 30 individuals, andcollected a total of 525.2 hr of focal individual data (mean 7 SD hr/individual =17.5 7 5.1). Using the criteria described above, we designated 67 PC/MC pairs.Twenty of the study 2 conflicts were parts of polyadic aggressive interactions (fourtriadic conflicts and one conflict each involving four, five, and six individuals). Thematched PC/MC periods had a mean (7SD) duration of 105 7 85 min (range =32–463 min). The mean (7SD) absolute value of the difference between PC andMC dates within each pair was 34.4 7 20.9 days (range = 1–68 days). The sampleof PC/MC pairs involved 61 distinct dyads and 33 different victims (mean 7 SDPC/MC pairs per victim = 2.0 7 1.4, range = 1–7). During 10 of 67 PC follows(14.9%), and 16 of 67 MC follows (23.9%), at least one break in observationoccurred that did not meet the criteria (see above) for a break between segments.PC and MC follows did not differ in the frequency of such breaks (mean 7 SDbreaks per PC period = 0.30 7 0.87, mean 7 SD breaks per MC period = 0.31 70.61, Wilcoxon test, n = 67, T = 135, P = 0.67). Table II shows the frequencies ofdifferent behavior patterns as the first affiliative interaction during PC and MCperiods.

TABLE II. Frequency With Which Classes of Behavior Occurred as the First Affliative

Behavior Between Oponents in PC and MC Periods

Context Approach Coalition Dance Groomingor invite

Nongroomcontract

Vocalization(nondance)

PC 17 6 7 9 10 13MC 10 3 0 10 6 7

Reconciliation in Cebus capucinus / 211

Evidence for Selective PC Affiliation

In the study 1 data, we found no significant differences between theproportions of attracted and dispersed pairs (n = 19 victims, mean 7 SDattracted =0.27 7 0.24, mean 7 SDdispersed = 0.18 7 0.27, number of victims with attracted4 dispersed = 10, number of victims with dispersed 4 attracted = 5, sign test, P= 0.30). In the survival analysis, a Cox’s F-test also failed to yield a significanttendency for first affiliative interactions to occur earlier in PC periods than in MCperiods (n = 70 PC/MC pairs, F [38, 48] = 1.37, P = 0.15). The mean CCT(averaged across victims) in study 1 was 0.09. However, in the study 2 data, wefound significantly more attracted pairs than dispersed pairs (n = 33 victims,mean 7 SDattracted = 0.44 7 0.42, mean 7 SDdispersed = 0.17 7 0.34, number ofvictims with attracted 4 dispersed = 18, number of victims with dispersed 4attracted = 5, sign test, P = 0.012), and we found that first affiliative interactionsoccurred significantly earlier in PC periods than in MC periods (Cox’s F-test, n =67 PC/MC pairs, F [36, 60] = 2.06, P = 0.007). The mean conciliatory tendency(averaged across victims) in study 2 was 0.29. Because in study 2 (in contrast tostudy 1) we scored each agonistic dyad in polyadic conflicts as a separate conflict,we checked the effects of this methodological difference on our results by recodingthe study 2 data, designating a primary attacker for each victim and discarding allother dyads (n = 13) in polyadic conflicts. We still found significantly moreattracted than dispersed pairs in study 2 (number of victims with attracted 4dispersed = 17, number of victims with dispersed 4 attracted = 5, sign test, P =0.02). We examined the study 2 data to determine whether our subjects affiliatedpreferentially with their former opponents following conflicts, or merely showed ageneralized tendency to affiliate with conspecifics following conflicts. Each

0

2

4

6

8

10

12

14

16

1 4 7 10 13 16 19 22 25 28 31

Minutes after conflict or MC start

Nu

mb

er o

f fir

st a

ffili

ativ

e co

nfl

icts

PC

MC

or more

Fig. 1. Distribution of the first affiliative interaction between former opponents during PC and MCfollows. Note that the PC and MC data points on the far right collapse all latencies 4 30 min.

212 / Manson et al.

individual that was a focal subject in at least one PC period and its correspondingMC period contributed one (averaged) data point to this analysis. Conflictparticipants did not affiliate with more partners in PC periods than in MCperiods, either during the first 10 min of the periods (n = 14, meanPC = 1.8,meanMC = 1.7, T = 39, P = 0.65) or throughout the periods (n = 13, meanPC =6.9, meanMC = 5.5, T = 18, P = 0.18). We also examined whether PC/MC pairswere more likely to be attracted when the former opponents were within 10 bodylengths of each other at the first point sample after the conflict than when formeropponents were farther apart at the first point sample after the conflict. Theavailable sample size (individuals that were victims in at least one conflict in eachof the two proximity conditions) was small, but we found no significant effect ofPC proximity on the likelihood of a PC/MC pair being attracted (n = 9, meanin prox

= 0.55, meannot in prox = 0.37, T = 2, P = 0.27).Figure 1 shows, for studies 1 and 2 combined, the distribution of first

affiliative interactions between former opponents in PC and MC periods. Thirty-one percent of first PC affiliative interactions occurred in the first PC minute, and45% occurred within the first 3 min. To examine the effects on the results of usingequal PC/MC period durations (30 min) in the two studies, we recalculated thecomparison between the numbers of attracted and dispersed pairs in study 2 afterremoving from the data set all pairs (n = 10: four attracted and six dispersedpairs) whose status (attracted or dispersed) was determined by a first affiliativeinteraction that occurred 430 min after the start of a PC or MC follow. We stillfound more attracted than dispersed pairs (n = 17 victims, number of victimswith attracted 4 dispersed = 15, number of victims with dispersed 4 attracted= 2, sign test, P = 0.004).

Effects of PC Affiliation on the Probability of Renewed Conflict

Having demonstrated the occurrence of selective PC affiliation in study 2, weexamined this data set to determine whether PC affiliation decreased the risk ofrenewed aggression between the opponents. In fact, however, PC affiliation mayhave actually increased the probability of renewed aggression. Renewedaggression between opponents occurred in 14 of the 67 PC periods. In five ofthese cases, either renewed aggression between the opponents occurred but PCaffiliation did not occur, or renewed aggression occurred before PC affiliation. Inthe other nine cases, renewed aggression occurred after the first PC affiliation. Infive of these nine cases, the first PC affiliative interaction included the use ofvocalizations or physical contact, while in the other four cases it consisted entirelyof an approach to within one body length. Furthermore, the probability ofrenewed aggression decreased as the first PC affiliation was delayed after theinitial conflict. Combining data from studies 1 and 2, latency to first PC affiliationwas significantly negatively associated with the probability of post-affiliationrenewed aggression (n = 62 PC periods in which PC affiliation occurred, logisticregression, Wald w2 = 4.06, df = 1, odds ratio = 0.949, P = 0.044; Fig. 2).

Responsibility for Initiation of PC Affiliation

Combining data from studies 1 and 2 yielded 56 cases of PC affiliativeinteraction in which it was possible to determine whether the victim or theaggressor initiated the PC affiliative interaction. The victim initiated theaffiliative interaction in 29 cases, as compared to 27 initiated by the aggressor.

Reconciliation in Cebus capucinus / 213

Differences Among Dyad Types in Conciliatory Tendency

For each adult female, we calculated her mean CCT with three classes ofopponents: adult females, subordinate males, and alpha males. For each adultmale, we calculated his mean CCT with adult males. In calculating the female–female and male–male CCTs, we assigned each PC/MC pair to the individual thatwas the focal subject during the PC period. Because this procedure produced asmall number of individuals (three to five) with CCTs with multiple classes ofopponents, and because of the small numbers of PC/MC pairs per subject (one, inmost cases), we were unable to test statistically for differences in CCT across dyadtypes. Table III shows pooled CCTs for the four different adult dyad types.

0 20 40 60 80 100 120 140 160 1800.0

0.2

0.4

0.6

0.8

1.0Pr

obab

ility

of

rene

wed

agg

ress

ion

betw

een

form

er o

ppon

ents

Min to first post-conflict affiliation

Fig. 2. Probability of renewed aggression between opponents following PC affiliation as a function ofPC latency to first affiliative interaction.

TABLE III. Pooled Corrected Conciliatory Tendency by Adult Dyad Type

Dyad Attracteda Disperseda Totala Pooled CCTb Victims Dyads

Female–alpha male 10 3 16 0.44 8 8Female–subordinate male 8 5 32 0.09 15 24Female–female 3 2 11 0.09 8 11Male–male 3 0 11 0.27 6 8

aNumber of PC/MC pairs.b(Attracted pairs–dispersed pairs)/total pairs.

214 / Manson et al.

DISCUSSION

By restricting our sample to cases of escalated aggression (conflicts in whichscreaming or intense vocal threats occurred), we reduced our sample size to anumber below that which is typical in studies of PC behavior, although it fellwithin the range of other studies of this topic in wild primates [Aureli, 1992;Castles & Whiten, 1998; Matsumura, 1996] (Table IV). Nevertheless, we wereable to demonstrate reconciliation in wild white-faced capuchins in one of our twostudies. It is not clear why we failed to find significant reconciliation in study 1.The inclusion of each dyad in polyadic conflicts as a separate data point in study 2,but not study 1, apparently did not affect the results. The sample size of victimswas smaller in study 1 than in study 2, so the possibility remains that study 1simply lacked the statistical power to reveal the relationship shown by study 2.However, the survival analysis used each PC/MC pair rather than each victim as adata point, and it also revealed significant reconciliation in study 2 but not instudy 1, even though the number of PC/MC pairs was essentially the same in thetwo studies. A more interesting possibility is that the longer PC/MC periods instudy 2 allowed our subjects more time to reconcile, and that white-facedcapuchins, because of their unusually ‘‘pugnacious’’ temperament [Rose et al.,2003], require more time to ‘‘cool down’’ after conflicts than most other primates(see Rolland and Roeder [2000] for a possible example of delayed reconciliation inringtailed lemurs]. Our finding that prompt reconciliation attempts were morelikely than delayed attempts to result in renewed aggression is consistent withthis hypothesis. However, the study 2 PC/MC pairs whose classification (attractedor dispersed) was determined by affiliative interactions occuring 430 min afterthe beginning of the sample were actually more likely to be dispersed thanattracted, and we still found significantly more attracted than dispersed pairsafter these pairs were removed from analysis. This result suggests that it is notnecessary to conduct long PC/MC follows to discover reconciliation in white-facedcapuchins. Although the distribution of first PC affiliative contacts over time (Fig.1) may show a smaller skew toward early PC affiliation than that found incomparable studies of other wild primate populations (Table IV), this differencemay simply be a product of the shorter PC follows in the other studies comparedto our study. When we included in the analysis only those first PC affiliativeinteractions that occurred during the first 10 PC minutes (as did the authors ofthe comparable studies), we found that white-faced capuchins tended to engage inPC affiliation as quickly as longtailed macaques (Macaca fascicularis), moormacaques (M. maurus), and olive baboons (Papio anubis). Furthermore, in astudy of PC behavior in semi-free-ranging C. capucinus, Leca et al. [2002] foundreconciliation in male–female dyads using 10-min follows, although they did notfind reconciliation when they considered their data set as a whole.

A continuing area of debate [Cords & Aureli, 1996; Silk, 1996] is whetherreconciliation functions to repair damaged social relationships, or merely tofacilitate former opponents’ access to short-term benefits (e.g., cessation ofaggression, stress reduction, and contact with young infants). Cords and Aureli[1996], and Aureli et al. [2002] argued that even if individuals were motivatedonly by a desire for short-term benefits, reconciliation could still have the evolvedfunction of maintaining valuable long-term social relationships. We found thatattempts at reconciliation did not reduce the likelihood of renewed aggression; ifanything, they increased it. Furthermore, there was no tendency for victimsto initiate reconciliations more frequently than aggressors. Several studies[e.g., Aureli & van Schaik, 1991; Castles & Whiten, 1998] have used rates of

Reconciliation in Cebus capucinus / 215

TA

BL

EIV

.C

om

pa

riso

no

fT

ime

Dis

trib

uti

on

of

Fir

stP

CA

ffili

ati

ve

Inte

racti

on

sA

mo

ng

Stu

die

so

fW

ild

Pri

ma

tes

%of

firs

tP

Caffi

liati

ve

inte

ract

ion

socc

urr

ing

du

rin

ga

Sp

ecie

sP

Cp

erio

dd

ura

tion

N(P

C/M

Cp

air

s)F

irst

1m

inF

irst

3m

inR

efer

ence

Ma

caca

fasc

icu

lari

sM

ean

=1

2.9

min

15

65

37

7A

ure

li[1

99

2]

Ma

caca

ma

uru

s1

5m

in4

55

58

0M

ats

um

ura

[19

96

]P

ap

ioa

nu

bis

5–1

5m

in4

54

36

67

Cast

les

an

dW

hit

en[1

99

8]

Ceb

us

cap

uci

nu

s3

0–4

63

min

13

73

1a

(45

)b4

5a

(70

)bT

his

stu

dy

aA

mon

gall

firs

tP

Caffi

liati

ve

inte

ract

ion

s.bA

mon

gth

ose

firs

tP

Caffi

liati

ve

inte

ract

ion

socc

urr

ing

du

rin

gth

efirs

t1

0P

Cm

inon

ly.

216 / Manson et al.

self-directed behavior (SDB), such as scratching or self-grooming, as measures ofanxiety, and have examined whether conflicts elevate SDB rates and whetherreconciliation reduces SDB rates to baseline levels. Unfortunately, SDB ratescannot be used as measures of PC anxiety in white-faced capuchins because theydo not rise significantly following fights [Manson & Perry, 2000]. Although we didnot find that reconciliation in C. capucinus increased former opponents’ access toshort-term benefits, we could not directly test whether reconciliation in thisspecies repairs social relationships over long time periods. Our study 2 protocol(all-day follows of the same individual separated by several weeks) precluded anexamination of whether reconciled conflicts are followed by higher rates ofaffiliation or proximity between former opponents over multi-day periods,compared to unreconciled conflicts [Koyama, 2001; Silk et al., 1996]. We werealso unable to conduct such a test with our study 1 data set, because it includedtoo few dyads that engaged in both reconciled and unreconciled conflicts.

Based on Cords and Aureli’s [1993, 2000] relationship-qualities model, andprevious findings about white-faced capuchin social relationships [Fedigan, 1993;Jack & Fedigan, 2004b; Manson et al., 1999, 2004; Perry, 1995, 1996a, b, 1997,1998a, b], we expected to find that reconciliation was most frequent in dyadsconsisting of an adult female and an alpha male, and least frequent in dyadsconsisting of an adult female and a subordinate male. Because of the small samplesizes, however, we were unable to test this prediction, and the issue remains to beresolved by additional data.

ACKNOWLEDGMENTS

J. Anderson, H. Gilkenson, J. Gros-Louis, and L. Sirot assisted in collectingthe data. Our fieldwork was funded by the L.S.B. Leakey Foundation (twogrants), National Science Foundation (graduate fellowship to S.P., and grants toB. Smuts and J. Mitani), National Geographic Society, Sigma Xi, RackhamGraduate School (Research Partnership grant with B. Smuts, a predoctoralfellowship, and a dissertation grant), Evolution and Human Behavior Program,University of Michigan Alumnae Society, Wenner-Gren Foundation, and UCLAAcademic Senate. The Max Planck Institute for Evolutionary Anthropologyprovided support during the data analysis and write-up. We thank the CostaRican Servicio de Parques Nacionales and the ACT for permission to work inLomas Barbudal Biological Reserve, and the community of San Ramon deBagaces, Rancho Jojoba, and Hacienda Pelon de la Bajura for permission to workon private lands near the reserve. L. Sirot, M. Tomaszycki, M. Landys, and S.Newman assisted in data transcription and data entry. We thank E. Wikberg andF. Campos for help with the data analysis. G. and J. Frankie, the Rosales family,and M. Cedillos provided logistical assistance in the field. We thank F. Aureli andC. Shaffner for discussions, and two anonymous reviewers for comments.

REFERENCES

Aureli F, van Schaik CP. 1991. Post-conflictbehaviour in long-tailed macaques (Macacafascicularis): II. Coping with the uncer-tainty. Ethology 89:101–114.

Aureli F. 1992. Post-conflict behaviour amongwild long-tailed macaques (Macaca fascicu-laris). Behav Ecol Sociobiol 31:329–337.

Aureli F, Cords M, van Schaik CP. 2002.Conflict resolution following aggression ingregarious animals: a predictive frame-work. Anim Behav 64:325–343.

Baker M. 1996. Fur rubbing: use of medicinalplants by capuchin monkeys (Cebus capu-cinus). Am J Primatol 38:263–270.

Reconciliation in Cebus capucinus / 217

Castles DL, Whiten A. 1998. Post-conflictbehaviour of wild olive baboons. I. Reconci-liation, redirection and consolation. Ethol-ogy 104:126–147.

Cords M, Aureli F. 1993. Patterns of reconci-liation among juvenile long-tailed maca-ques. In: Pereira ME, Fairbanks LA,editors. Juvenile primates: life history,development, and behavior. Oxford: OxfordUniversity Press. p 271–284.

Cords M, Thurnheer S. 1993. Reconciliationwith valuable partners by long-tail maca-ques. Ethology 93:315–325.

Cords M, Aureli F. 1996. Reasons for reconcil-ing. Evol Anthropol 5:42–45.

Cords M, Aureli F. 2000. Reconciliation andrelationship qualities. In: Aureli F, de WaalFBM, editors. Natural conflict resolution.Berkeley, CA: University of CaliforniaPress. p 177–198.

de Waal FBM, van Roosmalen A. 1979.Reconciliation and consolation amongchimpanzees. Behav Ecol Sociobiol 5:55–66.

de Waal FBM, Yoshihara D. 1983. Reconcilia-tion and redirected affection in rhesusmonkeys. Behaviour 85:224–241.

de Waal FBM, Ren R. 1988. Comparison of thereconciliation behavior of stumptail andrhesus macaques. Ethology 78:129–142.

de Waal FBM. 1989. Peacemaking amongprimates. Cambridge, MA: Harvard Uni-versity Press. 310 p.

de Waal FBM. 2000. The first kiss: founda-tions of confict resolution research inanimals. In: Aureli F, de Waal FBM,editors. Natural conflict resolution. Berke-ley, CA: University of California Press.p 15–33.

Fedigan LM. 1993. Sex differences and inter-sexual relations in adult white-faced capu-chins (Cebus capucinus). Int J Primatol14:853–877.

Fedigan LM. 2003. The impact of male take-overs on infant deaths, births and concep-tions in Cebus capucinus at Santa Rosa,Costa Rica. Int J Primatol 24:723–741.

Frankie GW, Vinston SB, Newstrom LE,Barthell JF. 1988. Nest site and habitatpreferences of Centris bees in the CostaRican dry forest. Biotropica 20:301–310.

Gros-Louis J. 2001. Food-associated calls inwhite-faced capuchin monkeys (Cebus ca-pucinus): different functions from the per-spective of the signaler and the recipient[dissertation]. Philadelphia, PA: Universityof Pennsylvania. 103 p. Available from:University Microfilms, Ann Arbor, MI;AAI3015315.

Gros-Louis J. 2002. Contexts and behavioralcorrelates of trill vocalizations in wildwhite-faced capuchin monkeys (Cebuscapucinus). Am J Primatol 57:189–202.

Gros-Louis J, Manson JH, Perry S. 2003.Violent coalitionary attacks and intraspeci-fic killing in wild white-faced capuchinmonkeys (Cebus capucinus). Primates44:341–346.

Hakeem A, Sandoval RG, Jones M, Allman J.1996. Brain and life span in primates. In:Birren JE, Schaie KW, editors. Handbook ofthe psychology of aging. 4th ed. San Diego:Academic Press. p 78–104.

Jack KM, Fedigan L. 2004a. Male dispersalpatterns in white-faced capuchins, Cebuscapucinus. Part I: Patterns and causes ofnatal emigration. Anim Behav 67:761–769.

Jack KM, Fedigan L. 2004b. Male dispersalpatterns in white-faced capuchins, Cebuscapucinus. Part II: Patterns and causes ofsecondary dispersal. Anim Behav 67:771–782.

Koyama NF. 2001. The long-term effects ofreconciliation in Japanese macaques (Ma-caca fuscata). Ethology 107:975–987.

Leca J-B, Fornasieri I, Petit O. 2002. Aggres-sion and reconciliation in Cebus capucinus.Int J Primatol 23:979–998.

Manson JH, Perry S, Parish AR. 1997.Nonconceptive sexual behavior in bonobosand capuchins. Int J Primatol 18:767–786.

Manson JH. 1999. Infant handling in wildCebus capucinus: testing bonds betweenfemales? Anim Behav 57:911–921.

Manson JH, Rose LM, Perry S, Gros-Louis J.1999. Dynamics of female–female relation-ships in wild Cebus capucinus: data fromtwo Costa Rican sites. Int J Primatol20:679–706.

Manson JH, Perry S. 2000. Correlates of self-directed behaviour in wild white-facedcapuchins. Ethology 106:301–317.

Manson JH, Gros-Louis J, Perry S. 2004.Three apparent cases of infanticide bymales in wild white-faced capuchins (Cebuscapucinus). Folia Primatol 75:104–106.

Matsumura S. 1996. Postconflict affiliativeconflicts between former opponents amongwild moor macaques (Macaca maurus). AmJ Primatol 38:211–219.

Oppenheimer JG. 1973. Social and commu-nicatory behavior in the Cebus monkey. In:Carpenter CR, editor. Behavioral regulatorsof behavior in primates. Cranbury, NJ:Associated University Presses. p 251–271.

Oppenheimer JG, Oppenheimer EC. 1973.Preliminary observations of Cebus nigrivit-tatus (Primates: Cebidae) on the Venezue-lan Llanos. Folia Primatol 19:409–436.

Perry S. 1995. Social relationships in wildwhite-faced capuchin monkeys, Cebus ca-pucinus [dissertation]. Ann Arbor, MI:University of Michigan. 326 p. Availablefrom: University Microfilms, Ann Arbor,MI; AAI9542933.

218 / Manson et al.

Perry S. 1996a. Female–female relationshipsin wild white-faced capuchin monkeys,Cebus capucinus. Am J Primatol 40:167–182.

Perry S. 1996b. Intergroup encounters in wildwhite-faced capuchins, Cebus capucinus.Int J Primatol 17:309–330.

Perry S. 1997. Male–female social relation-ships in wild white-faced capuchin mon-keys, Cebus capucinus. Behaviour 134:477–510.

Perry S. 1998a. Male–male social relation-ships in wild white-faced capuchins, Cebuscapucinus. Behaviour 135:1–34.

Perry S. 1998b. A case report of a male rankreversal in a group of wild white-facedcapuchins (Cebus capucinus). Primates39:51–69.

Rolland N, Roeder JJ. 2000. Do ringtailedlemurs (Lemur catta) reconcile in the hourpost-conflict? Primates 41:223–227.

Rose L, Perry S, Panger M, Jack K, MansonJH, Gros-Louis J, MacKinnon K, Vogel E.

2003. Interspecific interactions betweenwhite-faced capuchins (Cebus capucinus)and other species: preliminary data fromthree Costa Rican sites. Int J Primatol24:759–796.

Silk JB. 1996. Why do primates reconcile?Evol Anthropol 5:39–42.

Silk JB, Cheney DL, Seyfarth RM. 1996. Theform and function of post-conflict interac-tions between female baboons. Anim Behav52:259–268.

Veenema HC, Das M, Aureli F. 1994. Metho-dological improvements for the study ofreconciliation. Behav Proc 31:29–38.

Verbeek P, de Waal FBM. 1997. Post-conflictbehavior of captive brown capuchins in thepresence and absence of attractive food. IntJ Primatol 18:703–725.

Weaver AC, de Waal FBM. 2000. The devel-opment of reconciliation in brown capu-chins. In: Aureli F, de Waal FBM, editors.Natural conflict resolution. Berkeley, CA:University of California Press. p 216–218.

Reconciliation in Cebus capucinus / 219