SEARCH

SEARCH BY CITATION

Keywords:

  • mandrill;
  • silent bared-teeth face;
  • reconciliation;
  • greeting signal;
  • signal of benign intent

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Method
  5. Results
  6. Discussion
  7. Acknowledgements
  8. Literature Cited

The distinction between signals that are friendly and those that are non-aggressive but motivationally neutral (signals of benign intent, SBIs) has not often been well elucidated in the literature. Although both signals occur in similar contexts, friendly signals should be exchanged more often between animals with good relationships whereas SBIs should be more commonly exchanged between animals with poor or unpredictable relationships. The importance of this distinction is particularly salient in the post-conflict context, because the two different types of signals may have disparate distributions and functions. This study examines the nature of the silent bared-teeth face (SBTF) in captive mandrills (Mandrillus sphinx) both during baseline interactions and following aggressive conflicts. We report that the SBTF is most commonly exchanged between mandrills with high rates of agonism. In addition, the SBTF is the most common post-conflict signal exchanged, and the mandrills exchanging this signal after fighting also have poor relationships. We conclude that, although one must be careful in generalizing from studies of captive populations, the mandrill SBTF observed in this study is most accurately classified as a signal of benign intent, but not a truly friendly signal and discuss problems with interpreting post-conflict data without distinguishing between these types of signals.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Method
  5. Results
  6. Discussion
  7. Acknowledgements
  8. Literature Cited

The distinction between various forms of non-aggressive communicative signals has historically been poorly understood and often neglected. Only sporadically have some scientists posited that there may be a motivational difference between the different types of signals that are broadly categorized as affiliative. One such example is the demarcation between signals that are truly friendly and those that are more motivationally neutral. One category of neutral signals is referred to as signals of benign intent (SBI) (Silk 2000; Silk et al. 2000), which serve to convey an animal's transient non-aggressive intentions. The specific signals referred to by Silk et al. (2000) are quiet vocalizations in rhesus macaques (girneys and grunts) and baboons (grunts) that indicated the signaler's reduced propensity for aggression and also increased their chances of handling an infant. Unlike truly affiliative signals, and despite their low cost, SBIs would be predicted to be more likely in cases with a high risk of aggression (Silk et al. 2000; Whitham et al. 2007); individuals with poor relationships have a greater need to signal their peaceful intentions prior to moving in proximity. A similarly functioning ‘greeting’ signal comprised of ear flattening, lip-smacking, and presenting behaviors is exchanged between male hamadryas baboons to reduce hostility between animals that approach one another (Kummer et al. 1974; Colmenares 1990; Colmenares & Silveira 2008). Although greeting signals in other species are often described as affiliative and more commonly exchanged between friends (Smuts & Watanabe 1990; Whitham & Maestripieri 2003; Dias et al. 2008; Smith et al. 2011), this signal in hamadryas is exchanged at a high rate during times of tension between males who do not spend much time in peaceful proximity and thus are unlikely to have good relationships (Colmenares 1990). The ‘Tension Reduction Hypothesis’ (Colmenares et al. 2000) proposes that instead of reaffirming social bonds or facilitating cooperation, some greeting signals serve to lessen the chances of an aggressive encounter between members of the same group. These types of greetings, also described in colobus monkeys (Kutsukake et al. 2006), could functionally be classified as SBIs. These SBIs must be distinguished from those that have classically been considered truly friendly.

One area of research in which the distinction between affiliative signals and SBIs is critical is in the field of reconciliation. Reconciliation is most typically defined as the exchange of ‘friendly’ signals between former opponents sooner after aggression than during baseline (Aureli et al. 2002). However, Colmenares & Silveira (2008) reported that the most common post-conflict (PC) signal exchanged between male hamadryas baboons was a non-contact greeting, a signal they showed was not aggressive but not friendly either. They demonstrated that combining this signal with affiliative signals exchanged during the PC context artificially inflated the extent of conciliatory behavior. Furthermore, PC anxiety levels were not reduced in former opponents after greeting one another, which provides some evidence that greetings do not reconcile conflicts. The silent bared-teeth face expression (SBTF) in mandrills is a visual gesture that consists of retracting the oral commissures in a figure-eight-like shape, which bares the canines, while the jaws remain in a closed position. It is almost always accompanied by a movement of the head in a lateral direction. The SBTF is a behavior that, as described in the literature, seems to share some characteristics with SBIs. Although in other primate species, bared-teeth expressions are often classified as submissive (de Waal & Luttrell 1985), the SBTF in mandrills has historically been described as aggressive (Andrew 1963; Jouventin 1975), sexual (Dixson 1998), emotionally ambivalent (van Hooff 1967), friendly (Bernstein 1970), and fearful (Redican 1975), with more recent evidence suggesting it has a peaceful function (Bout & Thierry 2005; Laidre & Yorzinski 2005). It is possible that the difficulty of appropriately classifying the SBTF is partially because it is a signal that is used to convey benign intent, and as such is often exchanged within dyads that do not have good relationships. Mandrills also exchange this signal after aggressive conflicts, suggesting that it might have a conciliatory function (Laidre & Yorzinski 2005; Schino & Marini 2011). However, if the SBTF is an SBI and not truly affiliative, the percentage of conflicts reported to be reconciled would be misleading.

This study aims to examine the function and distribution of the mandrill SBTF in PC and baseline conditions. We first aim to replicate previous reports (Schino & Marini 2011) that mandrills are faster to exchange the SBTF with one another after a conflict than during corresponding matched controls. We also hypothesize that the SBTF is properly classified as an SBI rather than a friendly behavior. We test the predictions of this hypothesis by analyzing the relationship between the SBTF and other behaviors, whether it is exchanged more or less in dyads with a high propensity for conflict, and the implications of including the SBTF as a conciliatory behavior.

Method

  1. Top of page
  2. Abstract
  3. Introduction
  4. Method
  5. Results
  6. Discussion
  7. Acknowledgements
  8. Literature Cited

Subjects

Behavioral data were collected from a captive group of mandrills (Mandrillus sphinx) housed at the Lowry Park Zoo in Tampa, FL. The group consisted of nine individuals: two adult females, one adolescent female, one juvenile female, two adult males, and three adolescent males. The outdoor enclosure measured approximately 12.2 × 7.6 m, and the animals inhabited it from 0900 to 1700. They were indoors during all other times. The animals were fed at 0900 and 1700 and had ad libitum access to water. At the time the study began, the group had been intact for approximately 6 yr, with the exception of the youngest female, who was born into the group 4 yr before the study began.

Data Collection

Behavioral data were collected in three contexts: baseline, during conflicts, and following conflicts. Data were recorded on a Dell Lattitude CPi – A-Series Model no. PPL laptop computer with Noldus Observer 5.0 software. Data recording methods were entirely observational.

Conflict Data Collection

When conflicts occurred during a focal animal sample, the focal sampling was discontinued and data were collected on the conflict. In this study, a conflict was defined by particular non-physical and physical forms of aggression. Non-physical aggression was operationalized as giving ground slaps, head bobs, and threat grunts (level 1 aggression) that were accompanied by either lunging toward or chasing another animal (level 2 aggression). Physical aggression was defined as biting, hitting, or grabbing (level 3 aggression). Level 1 aggression in isolation was considered to be part of conflict management strategies that prevent conflict of interest from escalating to physical aggression and thus was not included in this analysis, but instead was coded as agonism as suggested by Setchell & Wickings (2005). When the conflict began (which was considered the time point during which level 2 or level 3 aggression first transpired between animals), the identity of the aggressor, recipient, and type of aggression was recorded. The conflict was assumed to be finished immediately after the last aggressive exchange (including level 1 aggression) had terminated.

Post-Conflict Data Collection

After the conflict ended, the distance between the opponents was immediately recorded. In addition, one of the opponents was selected at random for 10 min of focal observation as above; this was the post-conflict (‘PC’) sample. If within 2 min of the PC's inception, further aggression between the individuals involved was observed, it was considered to be part of the same aggressive encounter and thus the PC observation was started over once the aggression had terminated. This 2 min interval was chosen because most post-conflict non-aggressive signal exchange transpires within 2 min of a preceding conflict (Aureli et al. 2002), and this was also the case in our study. No renewed aggression transpired between minutes 3 and 10 of a post-conflict observation. Frequency and timing of affiliative and peaceful behaviors were recorded, including reciprocal exchange of the SBTF, lip-smacking, grooming, head-shaking, playing, non-aggressive touching, and time spent in peaceful proximity. Peaceful proximity was included because it is very common to include implicit (i.e., peaceful proximity) in addition to explicit (i.e., physical contact) modes of peaceful post-conflict behavior (Call 2000). Furthermore, implicit strategies are often equally effective as explicit ones for repairing damage from conflict (Cords 1993).

Matched-Control Data Collection

For each PC sample collected, a corresponding 10-min matched-control (‘MC’) sample was collected from the same focal animal on the next day of observation and at approximately the same time as the previous PC observation. The researcher began the MC observation when the spatial distance between the former opponents was approximately equal to the distance between opponents at the inception of the corresponding PC observation. The aim of this criterion was to eliminate confounds due to initial proximity (Call 1999).

Baseline Data Collection

All occurrences of behavioral states and events were collected using 20-min focal animal samples (Altmann 1974). Focal individuals were randomly selected using a random number generator without replacement. Once a sample had been collected on each group member, the process of random sampling began again. A total of 400 h of data were collected between June of 2005 and June of 2006 during the hours of 0900–1700. Behaviors were scored as in Setchell (1999). In brief, agonistic behaviors included non-physical aggression, comprised of staring, head bobs, ground slaps, and stand-offs, as well as submissive behaviors, including presenting (to members of the same sex) and lip-smacking (see Setchell 1999 for a complete ethogram). Presenting to members of the same sex has been classified as a formalized indicator of submission in mandrills (Setchell 1999; Setchell & Wickings 2005), meaning that it is exclusively directed toward dominant individuals. Lip-smacking has been classified as an appeasement signal in several old world monkeys (Maestripieri 2005), and it also was primarily directed up the dominance hierarchy in our colony. Affiliative behaviors included playing, grooming, muzzle-to-muzzle contact, contact sitting, and touching. Peaceful proximity was defined as the amount of time sitting within 0.6 meters of another animal without either animal demonstrating any aggression or displacement behavior (comprised of yawning, scratching with hind limbs, body shaking, and teeth chattering). All behaviors were reported as rates per time observed.

Data Analysis

All data analyses were performed with Stata and R (R Core Team 2013) statistical software. All analyses were two-tailed.

Post-Conflict Behavior

Conciliatory tendency

Conciliatory tendencies were calculated to obtain a measure of the mandrills’ propensity to reconcile. The following formula was used to compute a corrected conciliatory tendency (Veenema et al. 1994):

Corrected conciliatory tendency (CCT) = 100 ×(No. attracted pairs – No. dispersed pairs)/(attracted + dispersed + neutral pairs). Attracted pairs are those that were faster to exchange friendly or peaceful signals after a conflict than during matched-control sessions, dispersed pairs are those that were faster to exchange friendly or peaceful signals during matched control sessions, and neutral pairs denote those that either did not exchange friendly/peaceful signals at all in either condition or those that exchanged such signals at the same time in each condition.

Latency of SBTF in post-conflict interactions

A survival analysis was performed to determine whether former opponents reciprocally exchanged the SBTF faster after an aggressive conflict than during matched-control sessions. We entered the subject identities (both aggressor and victim) as stratification variables to control for pseudoreplication while performing a stratified generalized Wilcoxon matched-pairs signed-ranks test to compare the latency until an SBTF exchange in post-conflict and matched-control sessions (Schino & Marini 2011).

Baseline Behavior

Covariance of baseline behaviors

Principal component analysis (PCA) represents a powerful tool for interpreting complex data sets with multiple dependent variables such as in behavioral observation (Rommeck et al. 2009) and can guide hypothesis testing by identifying variables that are likely interrelated. Perhaps, because PCA is unable to account for the problems presented by pseudoreplication, it has not been frequently utilized in analyzing dyadic behavioral data. However, the need to control for pseudoreplication in such studies remains a contentious issue; some have pointed out that an animal's behavior differs as a function of their interaction partner (Colmenares & Silveira 2008), while others have argued that pseudoreplication is itself a ‘testable hypothesis’ and should not be assumed to be confounding (Coss 2009; Schank & Koehnle 2009). We utilized a social relations analysis to demonstrate that none of our behaviors of interest varied as a function of either actor or recipient independently of dyads (Table 1). This analysis parses out the variability due to each individual within a dyad; this allows one to determine the effect of the unique interaction between two group members (Schönbrodt et al. 2012). With these results in mind, we conducted a principal component analysis (PCA) on all social behaviors at the dyadic level to evaluate the relationship between these different classes of behaviors. We used this PCA to generate predictions about which behaviors covary and thus formulate reasonable hypotheses that could be tested using methods that fully control for pseudoreplication.

Table 1. This table depicts the results from the social relations analysis to assess the independence of the data set. The model failed to detect significant variability explained by the actions of either individual within a dyad. Instead, all significant variability in social behavior is due to the unique interaction between each participant. The first value in each column indicates the t-value and the second is the p-value
BehaviorActorPartnerRelationship
SBTF0.907; 0.1950.906; 0.1964.688; 0.001
Agonism0.185; 0.4291.550; 0.0805.431; <0.001
Conflict rate0.234; 0.8210.462; 0.3285.238; <0.001
Affiliation1.346; 0.1080.797; 0.2244.842; 0.001
Proximity0.238; 0.4090.312; 0.3814.306; 0.001
Table 2. Results of mixed-model logistic regression using social behaviors and kinship to predict rates of the SBTF
Independent variable Coefficient z p
SBTF
Agonism0.4803.4600.001
Conflict Rate18.1262.0300.042
Affiliation−0.459−0.5800.564
Proximity0.0030.5400.588
Kinship−0.001−1.8800.060
Intercept0.0011.0600.290
Characteristics of dyadic interactions predicting SBTF rate

A mixed-model logistic regression was used to predict the impact of baseline exchange of ‘friendly’ (affiliation and percent time spent in peaceful proximity) or ‘unfriendly’ (agonism and conflicts) behaviors as well as kinship on SBTF exchange rates. While percent time spent in peaceful proximity might be expected to parallel affiliation, and conflicts would be expected to parallel agonism, we analyzed all four separately as independent variables both because this has been suggested by previous work (Call 1999; Setchell & Wickings 2005) and because our main behavior of interest, SBTF, may reasonably be expected to vary differentially as a function of any of these four behaviors. We entered the subject identities as crossed random-effect independent variables (one for ‘actor’, one for ‘recipient’) to control for error due to the inherent dependence of the data points and thus controlling for pseudoreplication as in van de Pol & Wright (2009). Kinship was categorized according to Charpentier et al. (2012), with categories for non-kin, maternal or paternal siblings, and mother– or father–offspring relationships. The category of non-kin was used as the reference value for statistical analysis.

Characteristics of dyadic interactions predicting conflict rate

We wished to address the possibility that the high proportion of reconciled conflicts could be explained by the fact that only dyads with valuable relationships (who would be expected to reconcile a greater percentage of their disputes) were engaging in physical aggression. Our low rate of conflicts within each dyad did not allow for proper analysis of variation in conciliatory tendency. Instead, we utilized a mixed-model regression with conflict rate as the dependent variable, selecting percentage time spent in peaceful proximity (based on the results of our PCA) and kinship as indicators of relationship quality. As discussed above with the analysis of SBTF rate, we entered the subject identities as crossed random-effect independent variables.

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Method
  5. Results
  6. Discussion
  7. Acknowledgements
  8. Literature Cited

Post-Conflict Behavior

Conciliatory tendency

Data were collected for a total of 51 dyadic conflicts. Conflicts were recorded from 19 of a possible 36 dyads. Thirty-five of the conflicts involved non-contact threats and chasing, whereas the remaining 16 conflicts involved physical aggression. 64.5% of first signals exchanged by former opponents after a conflict were an SBTF. The overall corrected conciliatory tendency (CCT) using conciliatory behaviors as defined in Schino & Marini (2011) was 45.10%. The CCT resulting from level 2 aggression alone was 36.1% and from level 3 (or level 2 and 3 in the same episode) was 64.7%. To test how the inclusion of SBTF as a conciliatory behavior affects the calculation of CCT, we calculated CCT using SBTF as the only behavior (52.94%) and using only non-SBTF affiliative behaviors, including peaceful proximity (23.53%).

Latency of SBTF in post-conflict interactions

Survival analysis demonstrated that former opponents exchanged the SBTF sooner during post-conflict than during corresponding matched-control observations (Wilcoxon test: χ= 13.92, p = 0.002; Fig. 1).

image

Figure 1. Survival analysis demonstrates latency of opponents to exchange SBTF was significantly shorter in the immediate post-conflict period (PC) than during corresponding matched-control periods (MC; Wilcoxon test: χ= 13.92, p = 0.002).

Download figure to PowerPoint

Baseline Behavior

Covariance of baseline behaviors

The Kaiser–Meyer–Olkin measure that verifies sampling adequacy was greater than the cutoff value of 0.5 (actual value was 0.6) (Kaiser 1974). Our factors also passed the Bartlett's test of sphericity (p < 0.001). The first two components were the only ones with Eigen values greater than one (comp 1 = 2.56, comp 2 = 1.56) and thus were the only ones included in the model. These two components explained 68% of the cumulative variance. Figure 2 depicts the distribution of social behaviors at the dyadic level revealed by the principal component analysis. It shows that the SBTF is collinear with agonism (submissive and non-physical aggressive behavior) and conflict rate (physical aggression or non-physical aggression plus chasing), as they load together on the same principal components. Percentage of time in peaceful proximity and affiliative behavior also loaded together on the same principal components.

image

Figure 2. Collinearity of behaviors using principle component analysis (PCA) depicting the first two principal components and the associated loading of the rates of exchange for each social behavior within dyads. Behaviors that are collinear are commonly exchanged within the same dyads.

Download figure to PowerPoint

Characteristics of dyadic interactions predicting SBTF rate

The results from the PCA suggested that the SBTF covaried with agonism (non-physical aggression and submissive behaviors) and conflict rate (i.e., physical aggression), but not with affiliative behaviors. Thus, the predictions that the SBTF was positively associated with agonism and conflict rate but not with affiliative behaviors were each tested (Table 1). A mixed-model logistic regression showed a positive association between SBTF rates and agonism rates (coefficient = 0.480, z = 3.460, p = 0.001) and between the SBTF and conflict rates (coefficient = 18.126, z = 2.030, p = 0.042), but no relationship between SBTF rates and affiliation rates (coefficient = −0.459, z = −0.580, p = 0.564) or peaceful proximity (coefficient = 0.003, z = 0.540, p = 0.588, Table 2). A negative relationship between the SBTF and kinship approached significance (coefficient = −0.001, z = −1.880, p = 0.060), meaning that there was a tendency for non-kin to exchange the SBTF more often than kin. SBTF was therefore predicted by a dyad's level of baseline agonism and conflict rate, but not by its level of ‘friendly’ behaviors.

Characteristics of dyadic interactions predicting conflict rate

A mixed-model logistic regression revealed a significant negative relationship between conflict rate and percentage time spent in peaceful proximity (coefficient = −0.0001, z = −2.09, p = 0.037; Fig. 3), and no significant relationship with kinship (coefficient = 1.4 * 10−5, z = 1.64, p = 0.102). This demonstrates that SBTF following conflicts is not well explained by a tendency for ‘friendly’ dyads to engage in such conflicts.

image

Figure 3. Line of best fit and 95% confidence interval depicting a significant negative correlation between conflict rate and time spent in peaceful proximity within dyads (coefficient = −0.0001, z = −2.09, p = 0.037).

Download figure to PowerPoint

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Method
  5. Results
  6. Discussion
  7. Acknowledgements
  8. Literature Cited

Considered together with the findings from other studies, the evidence from this research supports the notion that the mandrill SBTF is a motivationally non-aggressive but non-friendly signal. Previous evidence has demonstrated that the mandrill SBTF is more likely to precede non-aggressive interactions and that it does not often transpire during aggressive contexts (Bout & Thierry 2005; Laidre & Yorzinski 2005). However, association with non-aggressive contexts is not sufficient to conclude that a behavior is necessarily ‘friendly’. Analysis of the relationships between animals exchanging a signal can provide an important clue to the nature of that signal. Our study reveals that mandrills who frequently exchanged the SBTF had relationships that were characterized by high rates of agonism but not high levels of affiliative behavior or time spent in peaceful proximity. In contrast, mandrills that did spend lots of time in peaceful proximity were more likely to affiliate but were less likely to engage in physical aggression. If the SBTF behavior is truly affiliative, it would be expected to be exchanged most among friends. The SBTF also tended to be more frequently exchanged between unrelated animals, whereas affiliative signals are usually exchanged more often between kin. Thus, it is likely that mandrills are using the SBTF to signal their transient non-aggressive intentions to individuals with whom they have poor relationships. The distribution and function of this signal appear to be most similar to that of greeting signals used by male hamadryas baboons. This greeting is most often exchanged between animals that generally avoid one another and are likely to engage in agonistic encounters and thus have reason to signal their non-aggressive intentions when they get near each other (Colmenares 1990). However, it is important to note that, although the mandrill SBTF does share features of both signals of benign intent and greeting signals exchanged in some species (colobus monkeys: Kutsukake et al. 2006; hamadryas baboons: Colmenares 1990; male howler monkeys: Dias et al. 2008), it also lacks some key characteristics that are often found in those signals. For instance, greeting signals are thought to reduce tension (Kutsukake et al. 2006), but in many species, they are comprised of ritualized, risky behaviors that involve contact and exposure of vulnerable anatomic features to the partner (Whitham & Maestripieri 2003; Smith et al. 2011). Further, many argue that the function of these signals is to strengthen bonds between animals (Smuts & Watanabe 1990; Whitham & Maestripieri 2003; Smith et al. 2011). The mandrill SBTF, in contrast, is hardly a risky signal, and the animals most likely to exchange it do not affiliate or cooperate. Although it is highly stereotyped, it does not involve a series of steps like many other greetings do. On the other hand, the behaviors previously described as signals of benign intent facilitate affiliative interactions and appear to serve to gain access to desired resources, often infants. This includes embraces in spider monkeys (Slater et al. 2007), grunts in baboons (Silk 2000), grooming in capuchins (Tiddi et al. 2010), and grunts in red lemurs (Pfluger & Fichtel 2012). In contrast, as mandrills most likely to exchange the SBTF have poor relationships and generally avoid one another, they do not appear to be facilitating access to anything.

It is possible that the mandrill SBTF is indicative of an insecure or unpredictable, rather than a predictably poor, relationship. Thus, it would be important for animals with such relationships to signal their peaceful intentions before coming near one another as their intentions are not likely to be evident. In fact, Cords & Aureli (2000) proposed that the presence of appeasement or friendly gestures during an approach could be one way to measure relationship security, with these gestures being more common in less secure relationships. Further, Colmenares et al. (2000) postulated that the function of greetings in hyenas and baboons is to reduce the likelihood of agonism when there is uncertainty about one another's motivations. Although this is an interesting idea with a fair amount of support, our data cannot distinguish between the claim that the SBTF is exchanged because relationships are poor but predictable from the claim that the SBTF is exchanged within insecure or unpredictable relationships.

Because the SBTF was not often exchanged between close companions, its usefulness as a conciliatory behavior is questionable. Affiliative and agonistic behaviors have previously been utilized to draw inferences about relationship quality within dyads (Weaver & de Waal 2002). In our study, mandrills with high rates of aggressive conflicts had relationships characterized by high rates of baseline SBTF exchange and little time spent in peaceful proximity. Thus, dyads exchanging SBTF after fighting or at high baseline rates were not likely to have good or valuable relationships. At 45.10%, the overall conciliatory tendency here was much higher than previously reported in mandrills (12.1%, Schino & Marini 2011), but this is probably due both to the fact that we excluded non-physical aggression from our conflict data (which also reduced the number of conflicts we recorded relative to other studies) and that we included peaceful proximity in our analysis of conciliatory behavior. However, almost two-thirds of the first post-conflict signals exchanged between former opponents were an SBTF, and the CCT inflates to an astounding 52% if the SBTF is the only behavior included in the analysis. Higher CCT is one marker of egalitarian rather than despotic species (Dobson 2012). Mandrills exhibit mostly despotic features (e.g., high rates of unidirectional aggression, low tolerance for subordinates, and low levels of cooperation), and so we would not predict a high CCT for this species. Colmenares & Silveira (2008) reported a similar incongruence when they examined greeting signals vs. other affiliative behaviors in hamadryas baboons. They concluded that greeting signals preferentially exchanged after a conflict in hamadryas are not exchanged between friends, nor do they function to repair relationships. These authors enumerated several reasons why it is implausible that the baboons exchanging such signals after a conflict have valuable relationships, and a substantial number of studies have explained variation in conciliatory tendency by demonstrating that those with valuable relationships are most likely to reconcile (Aureli et al. 2002). Further, they demonstrated that CCT for hamadryas baboons, when calculated to include these greeting signals, was so high (57.56%) that it exceeds that of any others in the animal kingdom. However, when they calculated CCT without non-contact greetings, their CCT was 6.1%, much more in line with expectations for a society like that of hamadryas. Similarly, we demonstrated that our CCT was 23.5% when we excluded SBTF, a much more reasonable number given what we know of mandrill society.

This finding can most simply be interpreted in one of two ways: (1) Either the SBTF is not a classically defined conciliatory signal or ‘friendly reunion between former opponents’ (de Waal & van Roosmalen 1979) or (2) mandrill post-conflict behavior is not parsimoniously explained by the Valuable Relationship Hypothesis, which states that dyads that derive fitness benefits from one another should be most likely to reconcile (Aureli et al. 2002). Indeed, our own data suggest that, at least during baseline conditions, non-kin trend toward more frequent SBTF exchange, suggesting that the animals exchanging it have less valuable relationships than others. Schino & Marini (2011) found that mandrills that exchanged the SBTF were less likely to engage in renewed aggression. Renewed aggression is one index that has traditionally been considered one of the functional sequelae of post-conflict signals and is thus used to provide evidence that such a signal is conciliatory (Cords 1992). However, this operationalization of reconciliation fails to distinguish between signals that truly serve to repair relationships from those that simply indicate that the preceding conflict is temporarily over (see Silk 2000).

Perhaps the distribution or function of so-called conciliatory signals changes depending on whether such signals are benign or truly friendly in nature. Some have reported that hyenas use greeting displays to reconcile and that non-kin reconcile more frequently than kin (Wahaj et al. 2001). As with primates, hyena kin spend more time with one another, exchange more affiliative behavior, and are much more likely to form alliances than are non-kin. Moreover, conflicts between the hyenas were much more frequent between non-kin than between kin. Thus, it is unlikely that hyenas with the most valuable relationships are the ones reconciling (but see Smith et al. 2011 for a different interpretation). Schino & Marini (2011) found that mandrills were more likely to use affiliative contact after fights with kin but used non-contact to signal to opponents with whom the risk of renewed aggression was high. Moreover, McFarland & Majolo (2013) reported that being in proximity but not grooming effectively reduced post-conflict anxiety levels in Barbary macaques. In species that exchange SBIs after fighting, perhaps an explanation similar to Silk's Benign Intent Hypothesis (Silk 2000) is more valid, which argues that former opponents reconcile simply to convey that the conflict has terminated but not necessarily to repair damage to a relationship.

The results of this study, suggesting that SBTF in mandrills is an SBI, stand somewhat in contrast to previous discussions of the behavior, which have considered it to be affiliative, aggressive, and/or conciliatory in nature. It is possible that the social behaviors we examined were affected by the small sample size and composition of our group or the artificial nature of their enclosure, with higher density than would be seen in nature. However, previous studies of mandrill behavior have been conducted under qualitatively similar conditions (Bout & Thierry 2005; Laidre & Yorzinski 2005), and it is not necessarily a foregone conclusion that higher housing densities would alter the quality or nature of social interactions rather than simply their overall frequency (van Wolkenten et al. 2006).

To conclude, mandrill SBTFs are most accurately categorized as non-aggressive but not friendly signals, and it is important to distinguish truly friendly signals from those that are simply benign prior to interpreting behavioral data. Future studies should both include contextual analyses and examine the nature of the relationships between animals that exchange non-aggressive signals to gain insight into the function of those signals, analyzing the consequences and distribution of each signal type separately.

Acknowledgements

  1. Top of page
  2. Abstract
  3. Introduction
  4. Method
  5. Results
  6. Discussion
  7. Acknowledgements
  8. Literature Cited

We thank the Lowry Park Zoo staff, especially A. Belcher, T. Crist, J. Lefave, C. Massaro, and L. Rottman for their invaluable assistance and cooperation throughout the duration of the study. We are also grateful to A. Weaver for comments on earlier versions of the manuscript and assistance with data collection techniques. Finally, we are indebted to T. Patton, J. Rottenberg, and T. Shimizu for helpful suggestions on the manuscript. None of the authors have any conflicts of interest that would affect this manuscript.

Literature Cited

  1. Top of page
  2. Abstract
  3. Introduction
  4. Method
  5. Results
  6. Discussion
  7. Acknowledgements
  8. Literature Cited