Behavioral Ecology and Sociobiology

, Volume 67, Issue 11, pp 1767–1780

Intersexual dominance relationships and the influence of leverage on the outcome of conflicts in wild bonobos (Pan paniscus)

Authors

    • Max Planck Institute for Evolutionary Anthropology
  • Gottfried Hohmann
    • Max Planck Institute for Evolutionary Anthropology
Original Paper

DOI: 10.1007/s00265-013-1584-8

Cite this article as:
Surbeck, M. & Hohmann, G. Behav Ecol Sociobiol (2013) 67: 1767. doi:10.1007/s00265-013-1584-8

Abstract

Dominance relationships between females and males are characteristic traits of species and are usually associated with sexual dimorphism. Exploring the social and contextual circumstances in which females win conflicts against males allows one to study the conditions triggering shifting power asymmetries between the sexes. This study investigates dominance relationships in bonobos (Pan paniscus), a species in which females are thought to display social dominance despite male-biased sexual dimorphism. To identify dominance relationships among females and males, we first explored how intrasexual dominance status affects the outcome of intersexual conflicts. Second, by incorporating social and behavioral information about the context of intersexual conflicts, we tested to which extent different components of power are relevant to the observed asymmetries in the relationships. Post-hoc analyses indicate a sex-independent dominance hierarchy with several females occupying the top ranks. Our results also reveal that two factors—female leverage and motivation to help offspring—had a significant influence on the outcome of intersexual conflicts. The results of our study do not indicate an overall reduction in male aggression against females but do show lower levels of male aggression in the mating context, and an absence of male aggression toward those females displaying visual signs of elevated fecundity. This indicates that both female sexuality and male mating strategies are involved in the shifting dominance relationships between the sexes.

Keywords

Female coalition formationFemale feeding priorityWinner–loser effectSelf-organization hypothesisSexual coercionApes

Introduction

Dominance relationships between males and females are usually a characteristic trait of a species' social system and have implications for differentiated access to resources, mating strategies of both sexes, and life history patterns (Yanca and Low 2004; Parker 2006). Female-biased asymmetries in intersexual dominance relationships are rare among social mammals and there are ongoing debates about the proximate and ultimate mechanisms involved in shifting dominance relationships between the sexes (Goymann et al. 2001; Kappeler and Schäffler 2008; Watts et al. 2009).

The concept of dominance relationships is based on the imbalance in the outcome of conflicts and the discrepancy in behaviors related to these conflicts (Rowell 1974; Bernstein 1981). In its original definition, dominance was considered a relative measure reflecting differences in the outcome of conflicts on a dyadic level (Schjelderup-Ebbe 1922). Although non-aggressive conflict resolution strategies exist (Hand 1986), the expression of dominance is often linked to aggressive behavior and the response it elicits (Drews 1993). At the same time, dominance is related to attributes that are relatively stable in expression and are regarded as intrinsic and individualistic traits, such as body mass or strength (McElligott et al. 2001). If such stable “prior attributes” are decisive for the outcome of conflicts, these outcomes become highly predictable (Archie et al. 2006). However, there is ample evidence to indicate that outcomes of conflicts that occur within the same dyad can fluctuate (Chase et al. 2002; Hewitt et al. 2009). Therefore, it may be more instructive to specify if there are sex differences in winning conflicts and if the outcomes vary across different contexts than to label societies as male- or female-dominant (Watts 2010). The “concept of power” describes asymmetries within dyads and accounts for this variability as it predicts that the outcome of conflicts depends on multiple parameters (Lewis 2002). First, dominance is regarded as the combined effect of intrinsic and derived dominance. Intrinsic dominance is based on an individual's own ability to use or threaten to use force while derived dominance, a component that in many species does not exist, comes from agonistic support from another group member (Chapais 1995). In addition, the concept of power includes a leverage component, a term that refers to the possession of a resource that is attractive to other individuals but cannot be taken by force, such as kinship or fertilizable eggs. Another component of the concept of power is motivation, a term that refers to the relative value a given resource has to different individuals (De Waal 1989a); this component may decide the outcome of a conflict in a way that differs from predictions based on intrinsic dominance. In sum, asymmetries within dyads and the outcome of conflict can be described by the combined effect of dominance, leverage, and motivation (Lewis 2002). Hemelrijk's “self-organization model” (Hemelrijk 1999; Chase et al. 2002) offers another rationale to explain the dynamics of dominance relationships. This model suggests that dominance is a consequence of the outcomes of previous conflicts such that the loser of a previous conflict is more likely to lose again and the winner of a previous conflict is more likely to win again, independent of the identity of the current contestant (Rutte et al. 2006). Therefore dominance relationships between individuals are a consequence of the self-reinforcing effects of winning and losing fights.

The two concepts are not mutually exclusive, as winner–loser effects can influence power asymmetries, and vice versa.

Intra- and intersexual selection and dominance relationships

Because of their physical superiority and propensity to engage in aggressive behaviors, males often have higher intrinsic dominance than females (Anderson 1994; Hunt et al. 2009). This results from intrasexual selection of traits such as weaponry, large body size, and aggressiveness, which benefits in the context of male mate competition and, as more aggressive fathers may sire more aggressive sons, may be propagated by female mate choice (Clutton-Brock et al. 1977; Emlen and Oring 1977; Hunt et al. 2009). While this suite of male characteristics is often related to physical fitness and resource holding potential, it is particularly pronounced in species in which access to females can be monopolized (van Hooff 2000). Consequently, male-biased skew in intrinsic dominance decreases when males cannot monopolize access to fertile females or as a result of intrasexual competition among females, which promotes increases in female body size and/or aggressiveness (Abbott et al. 2003; Schulte-Hostedde et al. 2004; Gowaty and Hubbell 2005; Clutton-Brock 2007). Other elements of power that can lead to a decrease in male power and egalitarian relationships between the sexes are high female derived dominance, the result of a high propensity of mutual agonistic support, or high female leverage, the result of high levels of sexual attractiveness or the inherently high value of an unfertilized egg. The latter might be particularly relevant if the female monopolization potential by males is low.

Using information from several primate species, Hemelrijk et al. (2008) found that female dominance increases with the number of male group members. According to the self-organization model, which is based on the reinforcing effects of winning and losing, the reduced dominance status of males over females is explained by the increase in male–male conflicts in multi-male groups. In the model, the resulting high number of lost conflicts for males increases their likelihood of losing conflicts against females (Hemelrijk et al. 2008).

Female dominance and co-dominance—proximate and ultimate explanations

In mammalian species, female dominance over males is rare and has been described only for some species of primates and carnivores (Ralls 1976; Hamilton et al. 1986; Richard 1987; Kappeler 1993). In some sexually monomorphic lemur species, adult males are consistently dominated by adult females. The causes of female dominance in lemurs are still unclear as neither energetic costs of reproduction, relaxation of male contest competition, nor female coalition formation seem to account for the dominant status of females (Kappeler 1996; von Engelhardt et al. 2000; Kappeler and Schäffler 2008, but see Pochron et al. 2003). In spotted hyenas, most female clan members dominate males (Smale et al. 1993) and different explanations for the evolution of female dominance in this species have been proposed, including intense feeding competition, prolonged periods of maternal care, and reduction of male contest competition (Goymann et al. 2001; Watts et al. 2009). On the proximate level, coalitions between related females and reduction in male aggressiveness are likely causes of female dominance in spotted hyenas (East and Hofer 2001; Goymann et al. 2001).

More common than the strict female dominance reported in lemurs and hyenas are dominance relationships between the sexes that either vary in relation to individual traits or depend on the context. For example, in rock hyraxes, African wild dogs, and wolves, individual females dominate some, but not all, male group members and the differentiated dominance relationships between females and males observed in these species have been associated with cooperative breeding (Frame et al. 1979; Mech 1999; Koren et al. 2006). In studies of other species, shifts in power asymmetries between males and females have been related to the following: (1) the influence of derived dominance such as coalition formation (Smuts 1987; chimpanzees: Newton-Fisher 2006), (2) differences in motivation (chimpanzees: Noë et al. 1980; Wittig and Boesch 2003; tufted capuchins: Janson and Vogel 2006), and (3) changes in leverage due to elevated fecundity (chimpanzees: Yerkes 1940). By exploring the social and contextual conditions in which females win conflicts against males, we may learn how females overcome male dominance and what conditions may trigger shifting dominance relationships between the sexes.

In this study, we test the impact of the components of power, including leverage and motivation, and the winner–loser effect on the outcome of intersexual conflicts in bonobos (Pan paniscus), a species in which females are known to win many conflicts with males. To do this, we first identify dominance relationships among females and males and explore how intrasexual dominance status affects the outcome of conflicts between the sexes. Second, by incorporating social and behavioral information about the context of intersexual conflicts, we test the extent to which different components of power are relevant for the observed asymmetries in the relationships between female and male bonobos.

Bonobos are hominoid primates that exhibit a moderately male-biased sexual dimorphism of around 1.17 in canine size and 1.21 in body mass (sexual dimorphism range in primates: 0.99 to 2.04 for canine dimorphism and 0.94 to 2.23 for body mass; Plavcan 1990). Previous studies have found that intrasexual dominance relationships in both sexes are more or less hierarchical and fairly stable over time (Furuichi 1989; Vervaecke et al. 2000; Surbeck et al. 2011). However, since the pioneering work by de Waal (1989b), the question of dominance relationships between male and female bonobos has been addressed in many studies remains a subject to much debate (e.g., Stanford 1998; White and Wood 2007; Parish and de Waal 2000). To date, the spectrum of interpretation regarding intersexual dominance relations in bonobos ranges from male dominance (Furuichi 1997; White and Wood 2007) to co-dominance (Paoli et al. 2006) to female dominance (Parish 1996). In societies in which dominance is biased towards one sex, the outcome of conflicts between males and females is likely to show a clear bias to one sex, independent of the intrasexual rank of individuals. However, in some cases, intrasexual rank positions are likely to affect the outcome of conflicts, with high-ranking individuals winning more often, independent of their sex. The discrepancy in defining intersexual power relationships in bonobos may reflect differences in ecology (captive versus field studies), demography (e.g., number of adult males/females), or methodology (e.g., definition of dominance). Alternatively, intersexual power relationships in bonobos may be flexible and change in relation to context, partner constellation, or resource value. For example, when individuals compete for access to food, the outcome may reflect state dependent differences in energy status or nutrition rather than dominance relationships (tufted capuchins: Janson and Vogel 2006). While rank attribution based on approach–retreat interactions in the feeding context have been shown to reflect dominance relations among bonobo females (Vervaecke et al. 1999), it may not always accurately represent dominance between the sexes (Kano 1992; White and Wood 2007).

Various hypotheses have been proposed to explain how [some] female bonobos achieve dominance over males. Several studies relate high dominance rank in female bonobos to female social bonds (Parish 1996; Vervaecke et al. 2000; Stevens et al. 2006). Although female bonobos typically disperse and therefore live in groups with unrelated females (Gerloff et al. 1999), they sometimes form coalitions with other females to control male aggression (Parish 1994; White and Wood 2007). Unlike females, bonobo males rarely form coalitions (Furuichi and Ihobe 1994).

The “self-organization hypothesis” attributes the high dominance rank of female bonobos to their gregariousness, which is high in comparison to their chimpanzee counterparts and influences the frequency of interactions between individuals, especially between males (Hemelrijk 2002). Another hypothesis suggests that female dominance over males reflects male deference, which results in females gaining priority of access to resources that have a stronger effect on female, rather than male, fitness (Furuichi et al. 1998). While these hypotheses differ in terms of the mechanisms that allow females to achieve a relatively high dominance status, they all imply that males have dyadic dominance over females but are unable to exert this dominance if females join forces against them or have higher motivation to win contests, or if they have experienced strong loser effects.

The “docile male hypothesis” (Stanford 1998), proposes that “male aggression towards females … hurts fitness” (Hare et al. 2012), and that male bonobos actually derive selective advantages from reduced aggression because females choose to mate with non-aggressive males (Wrangham and Peterson 1996).

The present study aims to test predictions that follow from each of these hypotheses using data on dominance relationships within and between the sexes and data on context and outcome of intersexual conflicts in wild bonobos. Specifically as follows:
  1. (A)

    The coalition hypothesis predicts that males win most dyadic intersexual conflicts but that the presence of female coalitionary partners influences a female's chance of winning an intersexual conflict.

     
  2. (B)

    The feeding priority hypothesis predicts that females are more likely to win food-related contexts and that the potential to monopolize food positively influences the likelihood of females winning against males.

     
  3. (C)

    The self-organization hypothesis predicts that a female's chance of winning a conflict is related to the outcome of her male opponent's preceding conflict. Rank can affect both the decision of other individuals to support third parties in agonistic interactions and the motivation to tolerate others at food patches (Cheney and Seyfarth 2007; Tiddi et al. 2011). Therefore in, testing predictions (A)–(C), we consider intrasexual rank of females as a potential influence on the outcome of intersexual conflicts. Female bonobos exhibit non-conceptive swelling cycles during large parts of their interbirth interval (Kano 1992; Furuichi 2011). These swellings provide an honest signal on the probability of ovulation, but not its exact timing (Reichert et al. 2002). Thus, we also include information on the state of a female's sexual swelling to account for the possibility that females derive leverage from visual cues of estrous and to look at the possible influence of this trait on the outcome of conflicts.

     
  4. D.

    The docile male hypothesis assumes that males derive selective advantages from not being aggressive toward females. Therefore, males should refrain from using aggression against females independent of the female's dominance status; that is, they should be equally docile to all female group members. Moreover, assuming that intersexual selection diminishes the benefits of aggression, it appears unlikely that the outcome of conflicts between the sexes varies with state dependent variables such as female reproductive status.

     

Methods

Study site and subjects

Field work was conducted at the LuiKotale field site in Salonga National Park, Democratic Republic of Congo, between December 2007 and July 2009 (Hohmann and Fruth 2003b). Members of the Bompusa bonobo community were habituated to human presence from the start of the study period. During the data collection period, the community consisted of 33–35 individuals, which included 5 adult and 4 sub-adult males, 11 parous females, and up to 5 nulliparous immigrant females. Age estimates were based on physical features such as body size, dentition, and (in females) genital swellings (Furuichi et al. 1998). All community members were identified at the start of the project reported here. The nine males were at least 10 years old, they engaged in competition over access to estrous females and actively participated in agonistic interactions with other males (Surbeck et al. 2011). Genetic analyses conducted for another project revealed that six of the nine males' mothers were in the group (Schubert, unpublished data; Surbeck et al. 2011). Occasionally, strange females visited the study community, but all except one disappeared after a few weeks or months. In this study, only data from females that were present throughout the period of this project were analyzed, including 11 parous females (3 of them gave birth at the beginning of the study period) and 1 nulliparous female. Immature individuals were categorized into juveniles (older than 5 years) and infants (younger than 5 years).

Behavioral observation

Parties preferentially containing males were followed from the time subjects left the nest in the morning until the time they constructed night nests in the evening. Party composition was recorded on the hour (N = 2,112 scores of party composition). All instances of observed aggressive and non-aggressive interactions, as well as events of food monopolization, were recorded during party follows and male focal follows (2,112 h of party follows and 470 h of focal follows; Altmann 1974). Focal follows lasted for 10 min and were separated for each individual by at least 1 h. Focal individuals were randomly chosen from the males traveling in the same party.

Aggression refers to directed agonistic behaviors and includes both contact and non-contact aggression. Contact aggression involves encounters that include physical aggression, such as hitting, pulling, and biting. Non-contact aggression involves directed and undirected charging displays. Branch dragging displays were included if they were clearly directed at another individual. Submission refers to different forms of retreat, such as fleeing, jumping aside, retreating and moving away. Counter-aggression refers to the aggressive behavior by the target of an aggressive approach.

Dominance: Assessment of dominance relationships among male community members were based on dyadic interactions, with individuals showing submissive behaviors in response to aggression or non-aggressive approaches by another male being classified as inferior. The results of this analysis revealed a linear dominance hierarchy and consistent dominance relationships among male community members throughout this study (Surbeck et al. 2011). We used the same criteria as we did with the males to assess dominance relationships among females. In some female dyads, agonistic interactions were rare or absent and aggression alone did not allow us to quantify dyadic dominance relationships. Studies on captive bonobos revealed a strong correlation between competitive feeding ranks and rank assignments to the outcome of agonistic interactions within one sex (Vervaecke et al. 1999). Therefore, we also considered interactions related to food competition to assess dyadic dominance relationships among females. This dataset included food items such as large fruit (e.g., Treculia africana, Annonidium manni) and vertebrate prey (e.g., forest antelopes, monkeys) that are highly preferred by all community members and can be monopolized by single individuals or a small group (Hohmann and Fruth 1996; Fruth and Hohmann 2002). Field protocols on communal feeding and food sharing were used to score interactions that appear to indicate dominance relationships, such as “taking food away,” “monopolizing food against another individual,” and “defending access to food”. We included 27 food-related interactions between males and 66 between females. In males, 85 % (N = 23) were found to agree with the linear dominance hierarchy based on agonistic interactions. In females, for the three dyads in which both measures of dominance relationship were available, the outcome of competitive feeding events was in accordance with the outcomes of agonistic interactions in five out of six cases.

Post-hoc assessment of intersexual dominance relationships was based on the outcome of dyadic agonistic interactions and outcomes of competitive feeding on monopolizable food.

Reversals refer to agonistic interactions which are won by the lower ranking individual.

Sexual swellings

In species in which females exhibit genital swellings, the term “estrous” refers to the period when female genital swellings are maximally tumescent (Dixson 1998). We scored genital swellings daily and distinguished between four swelling stages ranging from minimal (stage 1) to maximal (stage 4). Scores were based on firmness of swelling (tumescence) and skin surface structure (Hohmann and Fruth 2000).

Intersexual conflict

Intersexual conflicts are agonistic interactions between individuals of the opposite sex. The individual who displayed submissive behavior in response to aggression in an intersexual conflict was classified as the loser. When the intersexual conflict included several interactions within 10 min, the outcome of the last observed interaction was used to identify the inferior individual. For cases in which none of the individuals showed signs of submission, the interaction was categorized as undecided.

Context of interaction

Conflicts were assigned to different contexts, including mate competition, feeding, agonistic support of relatives, or social challenge. Mate competition was scored when conflicts between males and females were related to mating or mating attempts. This includes sexual harassment by males and females, females intervening against males attempting to access other females, and male aggression toward females who try to hinder their access to and monopolization of an estrous female. Conflicts over access to food resources were scored when conflicts took place within food patches and when the outcome of such conflicts appeared to affect food intake. Within the food context, we distinguished different size categories to account for variation in the potential to monopolize the food source using a scale ranging from 1 to 3, with 1 referring to trees with large crowns that cannot be monopolized (crown-diameter > 10 m, e.g., Dialium sp.), 2 referring to trees with small crowns (crown-diameter < 10 m, e.g., Polyalthia sp.), and 3 referring to single fruit or prey items that can be easily monopolized. Feeding events with only one monopolizable food item were also used to analyze feeding priority. If females have priority of access to food we would expect the females to be more likely to win in situations where food was monopolizable. Agonistic support refers to interventions by one individual on behalf of another individual who had become a target of aggression, but the conflict takes place only between helper and aggressor (cf., protective dominance (Noë et al. 1980)). In all observed cases, it was the mother who intervened on behalf of her offspring. The social challenge context was scored when conflicts represented challenges or reinforcements of dominance status (e.g., directed displays), when conflicts arose from competition for access to social partners, and when none of the other three contexts accounted for the interaction (Wittig and Boesch 2003).

Close female associates

To explore the potential impact of differentiated relationships among resident females on the dominance relationships between males and females, we considered patterns of spatial associations and information on agonistic aid among females involved in conflicts with males.

Assuming that females with close social bonds travel together more often than those without, we used party attendance to determine spatial association. We first calculated pairwise affinity indices. For each dyad, we then related the observed value of dyadic association calculated by the simple ratio index (SRIobs) to its expected value, controlled for individual differences in gregariousness and observation time (SRIexp) derived by randomization (see below). The dyad was classified as close associates if the SRIobs turned out to be significantly higher than the SRIexp.
$$ {\mathrm{SRI}}_{\mathrm{obs}}=\mathrm{Pa}(ab)/\ \left(\mathrm{Pa}(a)+\mathrm{Pa}(b)-\mathrm{Pa}(ab)\right) $$

Pa(ab) = number of parties containing both A and B.

Pa(a) = number of parties containing A.

Pa(b) = number of parties containing B.

To calculate the dyadic SRIexp, for each observation of a given individual, we assigned the observed number of party members by randomly drawing individuals with a probability corresponding to their overall frequency of appearance in the whole dataset. Dyadic SRIexp was derived by running 1,000 such randomizations, each time determining the SRI per dyad and finally averaging the SRI-values per dyad across all randomizations. If the deviation of the SRI in each randomization from the SRIobs in the dyad was smaller than zero in more than 95 % of the 1,000 cases, the dyadic association was categorized as a close association.

As female close associates may change over time (Hohmann and Fruth 2002), we also determined the close associates for each half of the study period separately.

To determine the extent to which conflict outcomes depend on the presence or absence of a coalitionary partner, we used information from behavioral observations on third party interventions by females in intersexual conflicts. Individuals who engaged in conflicts against a third individual were defined as coalitionary partners.

Data analysis

Calculation of intrasexual dominance rank

Using information on dyadic agonistic interactions and competitive feeding events (see above), we carried out hierarchical rank order analyses with MatMan (version MfW 1.1; earlier version described in de Vries et al. 1993) for all male and all female dyads separately. The resulting rank orders were used to assign nominal ranks to females and males independently. As we had only a small amount of agonistic interactions and competitive feeding events (N = 61) for the females, we tested the robustness of our MatMan-calculated female dominance hierarchy by simulating 10 dominance matrices of 12 individuals with a predefined dominance order and with the same amount of agonistic interactions and competitive feeding events as in our observations. We set the likelihood of winning a conflict proportional to the absolute rank difference between the contestants (which revealed a comparable amount of reversals as in the original data). Because the results of the simulations revealed a strong correlation between the resulting dominance hierarchies (mean correlation coefficient 0.84, range 0.69 to 0.95) and small deviations in the attributed ranks among the 10 simulations (mean deviation 1.3 ranks, range 0.5 to 2.2), we think that the derived female dominance ranks for our females were appropriate. Due to several unknown relationships, our female data does not allow us to draw conclusions about hierarchy linearity and steepness (Klass and Cords 2011), but this is inconsequential as these were beyond the scope of this study. To calculate sex-independent dominance ranks we used the same procedure including all intra- and intersexual agonistic interactions and competitive feeding events. To determine the degree of female dominance over males and to compare this to other primate species, we measured the relative position of females over males in the mixed-sex dominance hierarchy following the method described in Hemelrijk et al. (2008). Female dominance can range from 0 (no female dominant over any male) to 1 (all females dominant over all males).

Determinants of outcome of intersexual conflict

To analyze what predicts the outcome of an intersexual conflict, we used a generalized linear mixed model (GLMM) (Baayen 2009). We included the following variables as test variables in the model: the context of the conflict, the intrasexual dominance rank of the male and female, the number of adult females in the party, the presence of close associates of the female, and the size of the female's sexual swelling. As control variables we included the presence of the male's mother, the identity of the individual who initiated the aggression, and the number of males in the party. We also included male and female identity as well as the identity of the dyad as random effects. To test if the presence of a coalitionary partner influenced the female, we included this variable instead of the presence of a close associate in a second model. Bonobo females exhibit a form of socio-sexual behavior in which two females rub their genital region against each other (GG-rubbing). While it has been speculated that this behavior helps to form bonds between females (Furuichi 1989), GG-rubbing is more likely linked to reconciliation and tension regulation (Hohmann and Fruth 2000). Although we lack consistent data for the study period on GG-rubbing, the opportunistically collected observations (N = 48) were used to identify dyads in which this behavior occurred (N = 12). To test whether these dyads were more likely to support each other we compared the observed likelihood of choosing a GG-partner as coalition partner with the expected likelihood, given a random choice of partner among party members.

To test for the influence of food monopolization potential on conflict outcome we included only the interactions that took place in a feeding context and ran a reduced model with only food resource monopolization potential, male and female intrasexual rank, and the random effects. To test whether a male's previous interaction influenced the likelihood of a female winning an intersexual conflict, we restricted our interaction to the ones in which the included males were involved in a decided agonistic interaction on the same day. We included the outcome of the preceding agonistic interaction of the male as a predictor in a model that contained only intrasexual rank of the individuals as fixed effects as well as both individuals as random effects. This was due to the small sample size of the reduced dataset and the resulting problem of convergence of the model.

Calculation of likelihood of coalition formation

To test whether coalitions composed of different combinations of the sexes occurred more often than expected by chance, we performed a mantel-like permutation where the test statistic was defined as:
$$ \mathrm{TS}={{\displaystyle \sum_{i=1}^3\left(\overline{x_i}-\overline{\overline{x}}\right)}}^2 $$

With: \( \overline{x_i} \) = mean data in cell (with the three cells being female–female, male–male, mixed sex)

\( \overline{\overline{x}} \) = mean data over all cells.

We conducted 10,000 permutations into which we included the original data as one permutation.

Results

Intersexual conflict

A total of 297 conflicts were observed between females and males. Females won 56 % (N = 167, 108 wins against adult males) and males won 35 % (N = 103, 86 wins against parous females) of these interactions and the remaining 9 % (N = 27) had undecided outcomes (Fig. 1). The female initiated the aggression in 54 % (N = 161, 103 times against adult males) and the male initiated the aggression in 46 % (N = 136, 99 times against parous females) of the intersexual conflicts. Females won 96 % (N = 155) of the conflicts that they initiated while males won 73 % (N = 99) of those they initiated. Thus, whether or not an individual initiates a conflict is a good predictor of whether he or she will win or lose the conflict (85 % of all interactions). Out of 28 adult dyads with more than 2 agonistic interactions, 18 (65 %) showed reversals in the outcome of the conflict and the mean percentage of reversals was 24 % across these 28 dyads. Mean hourly rates of intersexual aggression within intersexual dyads did not differ between males and females (Mann–Whitney U: Nmales = 9, Nfemales = 14, U = 61.5, p = 0.94) (Fig. 1). There is a trend towards counter-aggression occurring more if males direct aggression toward females than if females direct aggression toward males (3 % of cases when males directed aggression toward a female and 6 % of the cases when a female directed aggression toward a male; chi-square = 3.477, df = 1, p = 0.062).
https://static-content.springer.com/image/art%3A10.1007%2Fs00265-013-1584-8/MediaObjects/265_2013_1584_Fig1_HTML.gif
Fig. 1

Individual rates of aggression within male–female dyads (displayed are means, 50 % quartiles and 95 % quartiles)

Intrasexual dominance rank and intersexual dominance relationships

Based on the outcome of agonistic interactions and competitive feeding events, we could reliably attribute dominance ranks to the females (p = 0.018, DC = 0.78, Table 1; for further information, see “Methods”; for dominance matrix, see Table A1 in Electronic Supplementary Material).
Table 1

Intrasexual dominance relationships among females (lower case letters indicate nulliparous female) and males (lower case letters indicate the sub-adult males)

Individual

Rank

Rank class

Females

  

 HA

1

High

 RI

2

High

 MA

3

High

 EV

4

High

 IR

5

Middle

 PA

6

Middle

 ZO

7

Middle

 OL

8

Middle

 GWa

9

Low

 SUa

10

Low

 lu

11

Low

 UMa

12

Low

Males

  

 CAb

1

 

 TIb

2

 

 JA

3

 

 DA

4

 

 BEb

5

 

 apb

6

 

 pnb

7

 

 emb

8

 

 mx

9

 

aPrimiparous females

bMales with a mother in the community

Intrasexual rank in both females (including sub-adult immigrants who stayed in the community) and males positively influenced the likelihood of winning a conflict with an individual of the opposite sex (females: GLMM; estimate  ±  SE, 1.78  ±  0.55, p < 0.001; males: GLMM; estimate  ±  SE, −2.23  ±  0.60, p < 0.001. Both based only on intrasexual interactions; Table 2).
Table 2

Results from the GLMM on the outcome of intersexual conflict (helping context excluded)

 

Estimate

SE

p value

Intercept

0.75

0.96

 

Female rank

1.78

0.55

<0.001

Male rank

−2.23

0.60

<0.001

Female swelling size

0.54

0.27

0.049

Context

  

0.42

Number of adult females

−0.37

0.34

0.27

Number of adult males

0.15

0.31

0.63

Presence of male mother

−0.35

0.74

0.63

Presence of female close associates

1.35

0.84

0.11

Intersexual ranks calculated on competitive feeding and agonistic interaction correlated significantly (Spearman's rank correlation: rs = 0.70, N = 21, p < 0.001). In 25 of 31 male–female dyads (81 %) with both, agonistic and competitive feeding interactions, there was agreement on the direction of dominance between the two measures of dominance among the individuals. Males lost more agonistic conflicts but won more food interactions in three of the remaining six dyads and females lost more agonistic interactions but won more food-related interactions in the other three dyads. Post-hoc analysis of dominance relationships among all individuals revealed a linear hierarchy (p < 0.001, DC = 0.82, Table 3, for dominance matrix see Table A2 in Electronic Supplementary Material). Calculating the degree of female dominance based on this hierarchy reveals a value of 0.68 on a scale from 0 (no female dominant over any male) to 1 (all females dominant over all males).
Table 3

Intersexual dominance hierarchy based on all intra- and intersexual dominance interactions. Same numbers beside the individuals indicate mother–son dyads

Individual

Rank

Sex

Age

HA3

1

F

Adult

RI

2

F

Adult

MA1,2

3

F

Adult

EV4

4

F

Adult

IR

5

F

Adult

PA5

6

F

Adult

CA2

7

M

Adult

Ti1

8

M

Adult

ZO6

9

F

Adult

JA

10

M

Adult

DA

11

M

Adult

BE6

12

M

Adult

OL

13

F

Adult

GW

14

F

Adult

SU

15

F

Sub-adult

LU

16

F

Sub-adult

AP3

17

M

Sub-adult

PN5

18

M

Sub-adult

EM4

19

M

Sub-adult

UM

20

F

Adult

Mx

21

M

Sub-adult

Context

Of the intersexual conflicts, 40 % (N = 120) occurred in a feeding context, 9 % (N = 27) occurred in a mating context, 13 % (N = 40) occurred in an agonistic support context, and 37 % (N = 110) occurred in a social challenge context. Females never lost an intersexual conflict in an agonistic support context. Agonistic support was given to sub-adult males (N = 22), juvenile non-dependent offspring of both sexes (N = 11) and dependent infants (N = 2).

All other contexts of intersexual conflicts had no significant influence on the outcome (GLMM, p = 0.42; Fig. 2, Table 2). Furthermore, the occurrence of conflict during a feeding context was not influenced by the monopolization potential of the food (monopolization of food: GLMM; estimate ± SE, 0.54 ± 1.14, p = 0.63). Females were in possession in 15 of the 22 cases in which they were feeding on a single monopolizable food items (seven meat eating events and 15 single fruit resource eating events). Given the party composition of these events, there is a trend towards females possessing food more often than would be expected by chance (binomial test: N = 22, p = 0.059).
https://static-content.springer.com/image/art%3A10.1007%2Fs00265-013-1584-8/MediaObjects/265_2013_1584_Fig2_HTML.gif
Fig. 2

Intersexual conflicts and their outcome in different contexts (ma mating, fe feeding, he agonistic support of a relative, so social challenge)

Female swelling size

Females with higher swelling scores were more likely to win conflicts with males (GLMM; estimate ± SE, 0.54 ± 0.37, p = 0.049; Table 2, Fig. 3). As their swelling stage increases, females do not become more aggressive (binomial test on the trend line slopes of female initiating aggression against males depending on swelling stage; Nfemales = 10, p = 0.38) but males initiate less aggression towards them (binomial test on the trend line slopes of males initiating aggression towards females depending on swelling stage; Nmales = 9, p = 0.02).
https://static-content.springer.com/image/art%3A10.1007%2Fs00265-013-1584-8/MediaObjects/265_2013_1584_Fig3_HTML.gif
Fig. 3

Percentage of conflicts won by females in relation to swelling state, ranging from 1 (minimal swelling size) to 4 (maximal swelling size)

Female associations and coalition

Of the 66 female–female dyads, 20 (30 %) were classified as close associates for the whole study period. This pattern remained the same during the two 10-month data collection periods (23 and 26 %, respectively, of dyads were close associates).

During this study we observed 26 female–female coalitions during conflicts. Female–female coalitions were significantly more likely to occur than male–female coalitions (N = 25) and there was a trend towards them also being more likely to occur than male–male coalitions (N = 7) (mantel-like permutation test: ff vs. fm, test statistic, TS = 1.22, p = 0.01; ff vs. mm, TS = 1.41, p = 0.06; mm vs. fm, TS = 0.20, p = 0.69). Of the coalitions between females, 50 % were between close associates (19 and 33 %, respectively, based on close associates during the two 10-month data collection periods). For the eight females that formed coalitions in parties containing a close associate (N = 21), the likelihood of choosing a close associate was not significantly higher than expected by chance (Wilcoxon signed rank test: V = 16, N = 8, p = 0.84). For the seven females forming coalitions in the presence of a GG-rubbing partner (N = 21), there was a trend of choosing such a partner for coalitionary support less frequently than expected by chance (Wilcoxon signed rank test: V = 25, N = 7, p = 0.08).

Neither the number of female party members nor the presence of a close associate or coalitionary partner had an influence on the outcome of an intersexual conflict (number of females: GLMM; estimate ± SE, −0.37 ± 0.34, p = 0.27; presence female close associate: GLMM; estimate ± SE, 1.35 ± 0.84, p = 0.11 (over 10 months: estimate ± SE, −0.38 ± 0.70, p = 0.59); presence female coalitionary partner: GLMM; estimate ± SE, −0.79 ± 0.90, p = 0.38; Table 2).

Male aggression and female coalitions

Male aggression against females was not randomly distributed (chi-square = 44.7427, df = 3, p value = 1.049e−09) but males did not direct more aggression than expected against immigrating females (given a random distribution of male–female aggression across all the females in relation to the time male and female individuals were observed together; binomial test N = 136, p = 0.17). Males directed aggression towards low-ranking females more often than expected and towards mid- and high-ranking females less often than expected (low-ranking females: binomial test N = 136, p < 0.001; mid-ranking females: binomial test N = 136, p = 0.012; high-ranking females: binomial test N = 136, p = 0.019) (Fig. 4). Only 2 % of male aggression towards females (N = 136) was followed by female coalitionary aggression towards that male.
https://static-content.springer.com/image/art%3A10.1007%2Fs00265-013-1584-8/MediaObjects/265_2013_1584_Fig4_HTML.gif
Fig. 4

Victims of male aggression (given a random distribution of male–female aggression across all the females in relation to the time male and female individuals were observed together; immigrant females: binomial test, N = 136, p = 0.17; low-ranking females: binomial test, N = 136, p < 0.001; mid-ranking females: binomial test, N = 136, p = 0.012; high-ranking females: binomial test, N = 136, p = 0.019)

The 26 observed female coalitions against males were preceded within 30 min by: male aggression towards a female (N = 3), male aggression toward offspring of a female (adult offspring: N = 1, sub-adult offspring: N = 5, juvenile non-dependent offspring: N = 2), male mating attempt with a female (N = 1), male undirected display (N = 2), no aggression by mail or no other apparent cause (N = 12). While most of the female coalitions with no preceding directed male aggression were formed between females who ranked higher than the attacked male (13 of 14 with all individuals identified), in over half the cases, the female coalitions that followed male aggression towards a female or her offspring were between females where at least one was lower ranking than the attacked male (six often with all individuals identified) (see Table 4).
Table 4

Causes and form of female coalitions

Cause of female coalition formation

Number of occurrences

Number with all individuals identified

Number of top down coalitions

Number of bridging coalitions

Male aggression against female

3

3

0

3

Male aggression against offspring

8

7

4

3

Male mating attempt

1

1

1

0

Male undirected display

2

2

2

0

No apparent cause

12

11

10

1

Winner–loser effect on observed intersexual conflict

For the male–female conflicts for which we could determine whether the involved male won or lost his previous interaction with another or the same individual on the same day (N = 91), we found no influence of this former outcome on the likelihood of losing a conflict against a female (GLMM; estimate ± SE, −1.00 ± 1.08, p = 0.35).

Discussion

This is the first study to look at aggressive conflicts between male and female bonobos across different contexts. Our results indicate that individuals higher in intrasexual rank and who initiated conflicts were more likely to win these conflicts. Post-hoc analysis of two correlated measures of dominance—outcome of agonistic interactions and food monopolization—revealed that adult bonobos from LuiKotale form a mixed-sex dominance. Adult females held the highest positions, adult males and some adult females held intermediate ranks, and sub-adult males as well as primiparous and nulliparous females held the lowest ranks.

The rank positions in the mixed-sex hierarchy were highly correlated with individual positions in the intrasexual dominance hierarchy and there was a strong correlation between the outcome of agonistic conflicts and competitive feeding situations. The assessment of female intrasexual rank was based on a small number of agonistic interactions and competitive feeding events, which reduces the power of the linearity test. However, given the results of post-hoc simulations (see “Methods”) and the fact that the attributed ranks are relevant in intersexual conflicts, the assignment of dominance ranks to females is justified and indicates their social relevance.

Unlike the unidirectional outcomes of dyadic conflicts between the males in this community (Surbeck et al. 2011), conflicts among females and between the sexes included reversals, suggesting that factors other than intrasexual rank also affect their outcome.

The coalition hypothesis predicts that female alliances control male aggression and agonistic aid among females facilitates the observed reversals in the outcome of intersexual conflicts. Some of the coalitions that we observed led to reversals, but they do not seem to be a prerequisite for bonobo females to win each conflict against a male, a finding that contrasts the situation of low-ranking individuals in some other species (de Villiers et al. 2003; Higham and Maestripieri 2010; Schülke et al. 2010; Berghänel et al. 2010). It has been suggested that even the presence of a close associate has the potential to alter the outcome of a conflict, either through vocal support or physical presence (Wittig et al. 2007). In this study, neither the presence of a female coalitionary partner or close associate nor the number of females in the party had a detectable impact on the outcome of intersexual conflicts. While more detailed work must focus on the nature of female bonds, overall, female alliances were rare and were almost never provoked by male aggression toward females. They were more often caused by males charging at immatures, and in all such cases of “protective dominance” (Noë et al. 1980), females stopped male aggression. Infanticide by males is considered an inherent threat to the reproductive success of most primate females (Palomit 2012) and female chimpanzees and females of other species have been observed to form coalitions against infanticidal males (Smuts and Smuts 1993; Ebensperger 1998; Sakamaki et al. 2001). Although infanticide has not yet been observed in bonobos, the behavioral response of females against aggression from male group members and strange males hints a potential risk of infanticide (de Waal 1997; Hohmann 2001). In this study, the aggressor males are long-term residents and the immature targets of male aggression were mostly independent offspring (>5 years). Although within-group infanticide in species with male residency has been observed (Hamai et al. 1992), it deviates from general predictions of the sexual selection hypothesis (Hrdy 1979). As interbirth intervals in bonobos average around 5 years (Takahata et al. 1996, MS and GH, unpublished data), it appears unlikely that male infanticide of immatures older than 5 years would shorten a female's interbirth interval and thereby enhance a male's possible reproductive success. While the threat of male infanticide may still influence the outcome of conflicts between the sexes, more detailed information is required to examine the cause and effect of male aggression against immature group members.

Feeding priority of females has been described for several species (e.g., Hurd 2006) and has been proposed as the ultimate cause of female dominance in sifakas (Pochron et al. 2003). In our study, females were not more likely to win conflicts in a feeding context and the potential for monopolizing access to a dense food patch did not affect the outcome of conflicts. In line with previous studies, females possessed monopolizable food items (e.g., large fruit or animal prey, Kuroda 1980; Vervaecke et al. 2000; Fruth and Hohmann 2002) but intersexual resource distribution was never counter to the intersexual dominance hierarchy.

Furthermore, there was no indication that shifting asymmetries in intersexual dominance relationships were related to the outcome of previous conflicts as is predicted by the self-organization hypothesis. This finding is in line with previous studies on captive bonobos that did not find the physiological effects that are associated with a winner–loser effect in a contest situation (Wobber et al. 2010). Given the small sample size, we cannot exclude the existence of such an effect in bonobos, but any such effect could not have been responsible for the observed reversals in the outcome of intersexual conflicts.

Leverage of estrous females

In primates, male aggression against females is particularly prominent in relation to mating (Smuts and Smuts 1993; Muller et al. 2007) and male aggression towards fertile females is often considered a form of coercion (Smuts and Smuts 1993; Muller et al. 2009). In chimpanzees, aggression outside a female's fertile phase can influence her mating decision during the times when the probability of fertilization is highest (Muller et al. 2011). Due to the negative effects of coercion on female reproductive success, strong selective pressure on females to develop counterstrategies exist. While some of these strategies, such as retaliating alone or together with other individuals, increase the costs to the aggressor, females can also use strategies such as estrous advertisement to manipulate male behavior indirectly (Smuts and Smuts 1993). In some primate species, females have genital swellings that show cyclic changes in size, color, and tumescence (Dixson 1998). Such signals of estrous advertisement coincide with physiological changes and are thought to indicate changing probabilities of fecundity or female quality to males (Beach 1976; for a review, see Nunn 1999). Consequently, sexual swellings induce male–male competition and, as a result, male aggression toward females. However, the potential payoff of focusing mating effort on those females showing visual displays of estrous is contingent on the precision of the signal and the extent to which the displays of multiple females overlap. When cycles are long and when females experience a larger number of cycles per interbirth interval, dominant males are less likely to monopolize access to fertile females (Ostner et al. 2008).

Female bonobos have genital swellings that change in size and color during the estrous cycle and are thought to provide a visual signal of fecundity (Dahl 1986). Nevertheless, these swellings are exhibited during large parts of the interbirth interval, including times of low fecundity such as pregnancy and lactation (Kano 1992; Furuichi 2011). The phase of maximal tumescence tends to be longer than in the closely related chimpanzees (Dahl 1986) and while ovulation is more likely to occur during maximal tumescence, the actual timing of ovulation does not always coincide with the visual signal (Reichert et al. 2002). In combination, these traits may cause an inflation of a signal that evolved to attract the attention of males.

Results of the present study indicate that female bonobos derive a source of leverage from the reduced predictability of ovulation, possibly because male reproductive success becomes contingent on proceptive behaviors by estrous females. Bonobo females at LuiKotale were more likely to win a conflict when they were maximally tumescent. This result was neither caused by an increase in female aggressiveness at higher swelling states as described elsewhere (Carpenter 1942), nor by an increase in male aggression toward estrous females. On the contrary, male aggression toward females was lower when they showed signs of elevated fecundity, which is in contrast with patterns seen in wild chimpanzees, where male aggression towards females increases when fecundity is higher (Muller et al. 2007). Although data from two bonobo populations show that rates of male aggression increase at times of high mating activity, this aggression rarely involves the female mating partner (Hohmann and Fruth 2003a; Surbeck et al. 2012a).

Though the idea that females derive leverage over males from the possession of fertilizable eggs has long been acknowledged, empirical support is scarce (Hand 1986; Lewis 2002). For example, it has been shown that female house finches win the majority of conflicts against males during the time of pair-formation, but males win conflicts during incubation and rearing of the young (Thompson 1960). In apes, indications of increased female leverage due to elevated fecundity are seen in orangutans, where females can break up intersexual associations relevant to male reproductive success (van Noordwijk and van Schaik 2009), mountain gorillas, where male aggression in multi-male groups increases at times of elevated mating activity (although the aggression is rarely directed against estrous females) (Sicotte 2002), and in a study on captive chimpanzees, where a female had better access to food during estrous (Yerkes 1940).

Our results indicate that changes in male behavior associated with mating competition are linked to changes in power asymmetries between the sexes and extend to other contexts. Whether bonobo males profit from being less aggressive towards estrous females—in exchange for sex, for example—will be addressed in future studies.

Conclusion

A recently established framework for bonobo evolution proposed selection against male aggression as a key parameter responsible for the physiology and cognition of this species (Hare et al. 2012). While the actual mechanism of this selection pressure was not addressed in the study by Hare et al., the data presented here plus results of a recent work on male competition in bonobos (Surbeck et al. 2012a, b) indicate that female reproductive biology and the mode of male mate competition play important roles. Instead of suppressing male aggression in general, female coalitions seem to suppress male aggression toward offspring and are not necessary for females to win conflicts against males.

The results of this study also indicate that changes in male behavior associated with mating competition are linked to changes in power asymmetries between the sexes. While the causal links among swelling patterns, variation in fecundity, and male mating strategies in bonobos remain to be explored, our results support the idea that the dominance relationships between male and female bonobos are related to female sexuality.

Acknowledgments

We thank the Institut Congolaise pour la Conservation de la Nature (ICCN) for granting permission to conduct research at Salonga National Park. Fieldwork at LuiKotale is supported by the Max-Planck-Society, the L.S.B. Leakey Foundation, the National Geographic Society, the Volkswagen Foundation, the Basler Stiftung für biologische Forschung, and private donors. MS was funded during parts of this study by the Förderkreis des Deutschen Primatenzentrums and by a SNSF grant. We thank Christophe Boesch for support during various stages of the project; Barbara Fruth for stimulating discussions and help in conducting fieldwork; Roger Mundry for lending a hand with data analysis; Lambert Booto, Isaac Schamberg and Wilson Schersten for assistance in the field; Tobias Deschner, Oliver Schuelke, David Watts, Carolyn Rowney, Mimi Arandjelovic and two anonymous reviewers for helpful comments on earlier versions of the manuscript.

Ethical standards

The methods used to collect observational data in the field are in compliance with the requirements and guidelines of the ICCN and adhere to the legal requirements of the host country, the Democratic Republic of Congo.

Supplementary material

265_2013_1584_MOESM1_ESM.docx (21 kb)
ESM 1(DOCX 21 kb)

Copyright information

© Springer-Verlag Berlin Heidelberg 2013