Mountain gorillas maintain strong affiliative biases for maternal siblings despite high male reproductive skew and extensive exposure to paternal kin

  1. Nicholas M Grebe  Is a corresponding author
  2. Jean Paul Hirwa
  3. Tara S Stoinski
  4. Linda Vigilant
  5. Stacy Rosenbaum
  1. Department of Anthropology, University of Michigan-Ann Arbor, United States
  2. The Dian Fossey Gorilla Fund International, United States
  3. Department of Primatology, Max Planck Institute for Evolutionary Anthropology, Germany

Abstract

Evolutionary theories predict that sibling relationships will reflect a complex balance of cooperative and competitive dynamics. In most mammals, dispersal and death patterns mean that sibling relationships occur in a relatively narrow window during development and/or only with same-sex individuals. Besides humans, one notable exception is mountain gorillas, in which non-sex-biased dispersal, relatively stable group composition, and the long reproductive tenures of alpha males mean that animals routinely reside with both maternally and paternally related siblings, of the same and opposite sex, throughout their lives. Using nearly 40,000 hr of behavioral data collected over 14 years on 699 sibling and 1235 non-sibling pairs of wild mountain gorillas, we demonstrate that individuals have strong affiliative preferences for full and maternal siblings over paternal siblings or unrelated animals, consistent with an inability to discriminate paternal kin. Intriguingly, however, aggression data imply the opposite. Aggression rates were statistically indistinguishable among all types of dyads except one: in mixed-sex dyads, non-siblings engaged in substantially more aggression than siblings of any type. This pattern suggests mountain gorillas may be capable of distinguishing paternal kin but nonetheless choose not to affiliate with them over non-kin. We observe a preference for maternal kin in a species with a high reproductive skew (i.e. high relatedness certainty), even though low reproductive skew (i.e. low relatedness certainty) is believed to underlie such biases in other non-human primates. Our results call into question reasons for strong maternal kin biases when paternal kin are identifiable, familiar, and similarly likely to be long-term groupmates, and they may also suggest behavioral mismatches at play during a transitional period in mountain gorilla society.

Editor's evaluation

This study fundamentally advances our understanding of the ways in which kin recognition might operate in complex social societies using a compelling longitudinal dataset of a wild mountain gorilla population. Relying on detailed behavioural and pedigree data for 157 individuals, and robust statistical analyses, the authors clearly show affiliative biases among maternal siblings with aggression being more likely for unrelated siblings. Given the structure of mountain gorilla society, this research calls into question the assumptions underlying maternal kin preferences, particularly when paternal kin may be distinguishable, and therefore provides a valuable contribution to discussions on mechanisms of kin selection and kin discrimination.

https://doi.org/10.7554/eLife.80820.sa0

Introduction

In humans and non-human animals alike, sibling social relationships are marked by continuous dynamics of conflict and cooperation that begin before birth (Trivers, 1974), and can persist throughout an entire lifespan, with important fitness consequences for the individuals involved (Lu, 2007; Hudson and Trillmich, 2008; Nitsch et al., 2013). While classical frameworks of sibling interactions emphasized competition among brood- or litter-mates for limited parental resources during times of dependency (e.g. Mock and Parker, 1997), subsequent developments across numerous academic disciplines demography: e.g., Sear and Mace, 2008; Nitsch et al., 2013; sociology: e.g., Steelman et al., 2002; Lu, 2007; behavioral ecology: e.g., Silk, 2002; Hudson and Trillmich, 2008; developmental psychology: e.g., Lamb and Sutton-Smith, 2014 have explored the full arc of sibling competition and cooperation across the lifespan and demonstrated the complexity and diversity inherent to sibling relationships. In understanding the evolution of human sibling relationships in particular, comparative studies of our primate cousins have identified a number of factors that predict when and how siblings cooperate and compete. Inconsistent results within and between species, along with the remarkable flexibility of human social systems; however, limits the translational value of many primate models. Here, we address these gaps by presenting an extensive longitudinal study of wild mountain gorillas (Gorilla beringei beringei), an endangered great ape whose unique, flexible social structure serves as a valuable comparative model for humans.

Classic models of kin selection predict that the social/mating structure of animal groups creates patterns of relatedness between group members, which then selects for kin recognition mechanisms that manifest in differences in cooperative, affiliative, competitive, and/or aggressive behavior (Hamilton, 1964; Grafen, 1990; Mateo and Hauber, 2015). This straightforward idea has spawned a large body of work on kin selection that spans animal groups (reviewed in Widdig, 2007; Smith, 2014). Within primates in particular, investigations of kin discrimination have yielded notably mixed results. Some studies report no bias towards related individuals (e.g. for food-sharing in long-tailed macaques: Schaub, 1996; for various affiliative and aggressive behaviors in yellow baboons: Erhart et al., 1997). Others support the existence of sophisticated kin discrimination based at least partially on phenotype matching, perhaps in interaction with familiarity, as evidenced by social partner preferences (Wu et al., 1980), rates of affiliative behavior (Smith et al., 2003; Streich et al., 2002; Smith et al., 2003), or visual gaze biases (Pfefferle et al., 2014). Still others find support for familiarity alone as the determinant of interaction patterns, with familiarity being assessed by time spent living together (in pig-tailed macaques: Fredrickson and Sackett, 1984), common clustering around mothers (in ursine colobus: Wikberg et al., 2014), or age differences (in bonnet macaques: Silk, 1994; in chimpanzees: Langergraber et al., 2007). There is also extensive debate over differences in patterns of discrimination between paternally and maternally related kin (e.g. Chapais, 2001). Some have suggested that non-monogamous primates evince matrilineal but not patrilineal sibling kin discrimination, perhaps due to restrictive dispersal patterns that may limit opportunities for long-term social relationships with paternal kin (Mitani et al., 2000; Langergraber et al., 2007) or polygynandrous mating systems with low reproductive skew that complicate efforts to accurately identify them (Galezo et al., 2022). Yet other perspectives challenge a clean distinction between maternal and paternal kin, suggesting that complex interactions between familiarity and other modes of kin discrimination structure social bonds across primates (see e.g. Silk, 2002; Streich et al., 2002; Widdig et al., 2006; Silk, 2009; Lynch et al., 2017).

As one of the main contributors to familiarity, age differences within sibling and non-sibling dyads might influence social dynamics (Widdig et al., 2001; Streich et al., 2002; Langergraber et al., 2007; Pfefferle et al., 2014; Wikberg et al., 2014). On one hand, siblings close in age might be more likely to compete for limited parental resources (Tung et al., 2016; Salmon and Hehman, 2021); on the other hand, as longer-lasting co-residents within the same family environment, they might also be expected to form stronger affiliative bonds than siblings distant in age (though, again, this may not apply equally to maternal and paternal sibships; Streich et al., 2002). It is unclear to what extent age proximity effects are restricted to genetic relatives. Female rhesus macaques appear to bias affiliation towards similarly aged peers, even when unrelated to them (Widdig et al., 2001). Among female baboons, even in individuals not related through the matriline, dyadic bond strength weakened with increasing age differences; however, when analyses are restricted to females unrelated through both the matriline and patriline, effects of age differences attenuated sharply (Smith et al., 2003; Silk et al., 2006). These results once again imply social familiarity (as indexed by age differences) and kin discrimination are both important for predicting sibling relationship qualities (Godoy et al., 2016), though their additive and/or interactive effects remain poorly defined.

Finally, the sex makeup of the dyad might influence interaction styles due to the differential benefits males and females receive from interactions with brothers, sisters, and unrelated partners (e.g. Lonsdorf et al., 2018). For males, especially in species who engage in aggressive intrasexual competition, other males, brothers included can represent important allies (e.g. Meikle and Vessey, 1981; Goodall, 1986) or rivals (Daly and Wilson, 1988; Snowdon and Pickhard, 1999; Chagnon et al., 2017) during status-striving efforts in adulthood. In either case, assessing physical capacities or formidability would aid in these efforts. Accordingly, rough-and-tumble play between males might serve as a rehearsal for intrasexual competition in adulthood (Gray, 2019), suggesting such a behavior should occur most often in male-male relationships—a prediction supported by research on male-dominant primates (e.g. Brown and Dixson, 2000; Maestripieri and Ross, 2004). While male-male interaction patterns might generally differ from those of other sex configurations, these differences may themselves partially depend on kinship. In chimpanzees, some evidence suggests that fraternal relationships among adolescents and adults are more affiliative and cooperative than relationships between unrelated males (e.g. Mitani, 2009; Sandel et al., 2020). From the female perspective, evidence for fraternal influences on fitness outcomes is mixed. While there are some reports of chimpanzee brothers ‘adopting’ immature sisters (Hobaiter et al., 2014; Reddy and Mitani, 2019), and one demographic study of humans reports benefits of older brothers on women’s lifetime fitness (Nitsch et al., 2013), other research on primate species with sex-biased dispersal suggests no lasting fitness effects (except perhaps when mothers die; Engh et al., 2009).

From the perspective of both males and females, sisters may represent important future alloparental helpers, either for the individual themselves (e.g. Hamilton et al., 1982; Gould, 2000; Hobaiter et al., 2014), or the individual’s offspring (e.g. Johnson et al., 1980; Nishida, 1983; but see Silk et al., 2006). Thus, cultivating relationships with sisters via affiliative interactions might be beneficial for both males and females. Last, for females in particular, sororal relationships may exert important influences on future rank and resource acquisition outcomes (Charpentier et al., 2008; Lea et al., 2014; cf. Engh et al., 2009). However, these kinds of sex-biased interactions might additionally depend on age differences between siblings (Lonsdorf et al., 2018), once again underscoring the complicated mix of demographic factors that may influence sibling relationships.

Understanding the nature and evolution of complex social relationships, such as those between siblings, requires long-term investigations that reveal how individuals respond behaviorally to socioecological variation (e.g. Alberts and Altmann, 2012). Additionally, comparative models for human sibling relationships in particular are most useful when the model species shares key social features. With these principles in mind, mountain gorillas in particular are a compelling model for studies of sibling dynamics. First, long-term monitoring of wild mountain gorillas by the Dian Fossey Gorilla Fund has revealed social structures marked by extensive diversity in relatedness, age proximity, and sex makeup infrequently observed in other non-human primate groups (Robbins et al., 2009b; Roy et al., 2014). Mountain gorillas regularly form multi-female, single-male groups, as well as multi-female, multi-male groups in which multiple males reproduce, though paternity data and unsophisticated paternal kin discrimination mechanisms are consistent with historically high reproductive skew (Bradley et al., 2005; Rosenbaum et al., 2015; Vigilant et al., 2015). As a result of their highly variable social structure, researchers regularly observe co-resident immatures that are full siblings, paternal half-siblings, maternal half-siblings, or unrelated to one another.

Second, like humans but unlike nearly all other primates, both male and female mountain gorillas may opt to disperse or remain in their natal groups upon reaching maturity (Robbins et al., 2009b; Stoinski et al., 2009). This permits fraternal, sororal, and mixed-sex relationships that can last for an entire lifespan, meaning that siblings can be an important source of support well into adulthood. For example, adult males benefit greatly from allies who help them fend off outside male challengers and prevent females from transferring out of their groups (Sicotte, 1993; Rosenbaum et al., 2016b; Mirville, 2018). Adult females with offspring benefit from male protection from infanticide (Harcourt and Stewart, 2007; Robbins et al., 2013) in multi-male groups, this protection could potentially come not only from mates, but from brothers as well, who receive indirect fitness benefits from their sister’s reproductive success. While the benefits of female-female relationships remain understudied in this species, higher-ranking females have better energy balance and shorter inter-birth intervals, and it is plausible that support from other females plays a role in achieving and maintaining dominance (Robbins et al., 2005; Wright et al., 2014; Wright et al., 2020). At the same time, sibling bonds in mountain gorillas are characterized by the potential for competition as well as cooperation. Even after co-resident mountain gorilla siblings cease competing over parental resources, they are likely to compete over other limited resources, including dominance positions, mating opportunities, and preferred foods (Harcourt and Stewart, 2007), which highlights the complexity of making straightforward predictions about affiliative and aggressive interactions and relatedness (e.g. Silk et al., 2010).

In the present study, we use nearly 40,000 hr of behavioral data spanning 14 years to describe patterns of interactions between siblings and demographically comparable non-sibling dyads in social groups of wild mountain gorillas. Using extensive maternity and genetic paternity data available for 157 identifiable individuals studied from late infancy through adulthood, we examine whether full siblings, maternal half-siblings (hereafter, ‘maternal siblings’), paternal half-siblings (‘paternal siblings’), and unrelated co-residents (‘non-siblings’) exhibit differing patterns of affiliation (playing, grooming, and time spent in close proximity) and agonism (contact and non-contact aggression) in line with models of kin selection, after adjusting for the potential mediating presence of mothers in these interactions. Currently, there are no data regarding kin discrimination patterns among mountain gorilla siblings. While research in other primate species has reported both greater cooperation and greater competition between siblings compared to non-siblings, on balance we expect that mountain gorillas should be more cooperative/affiliative with siblings than non-siblings due to the potential for inclusive fitness benefits. Regarding matrilineal versus patrilineal kin biases, it is unclear whether gorilla siblings should exhibit the same maternal sibling bias observed in other primate species. In species with meaningfully lower reproductive skew, paternal sibling discrimination may be challenging or of limited use due to dispersal patterns that limit opportunities to benefit from paternal kin. Gorillas’ high reproductive skew, in contrast, could conceivably facilitate paternal sibling discrimination—but it could also obviate the need to develop any such mechanisms. Given these competing considerations, we ask which scenario is more consistent with the dynamics of gorillas’ sibling relationships.

In addition to kinship, we also investigate the impacts of age differences and sex. We determine whether age differences, commonly used in kinship research as a proxy for familiarity between social partners, predict patterns of affiliative or agonistic behavior, and whether these patterns differ between paternal and maternal kin (as some evidence from cercopithecine monkeys suggests; e.g. Streich et al., 2002). Finally, we test whether male-male, female-female, and mixed-sex sibling relationships are characterized by differing rates and types of social interactions, and whether these sex category differences are restricted to kin. While developmental changes in social relationships are not the focus of this study, these comparisons do speak to questions regarding the kinds of benefits siblings might be expected to deliver later in life for one’s status-striving and/or resource acquisition efforts: e.g., among males, are fraternal relationships marked by higher rates of playing and fighting, and sororal relationships higher rates of grooming?; among females, are sororal relationships marked by the highest rates of grooming compared to any other dyad configuration?; are affiliative patterns unique to siblings, or are comparable trends found in unrelated dyads?

Results

Affiliative behaviors

In our full sample of 1934 unique dyads spanning 7832 dyad-years, full siblings (n=43 dyads) and maternal siblings (n=101 dyads) played and groomed each other significantly more than did paternal siblings (n=555 dyads) or non-siblings (n=1235 dyads; all comparisons p<0.05; Figure 1A and B). Age differences (in our sample, mean: 5.85 years; SD: 4.53 years; range: 0–23.5 years) interacted with relatedness in predicting grooming (p=0.017), but not play (p=0.427). Play consistently dropped for siblings and non-siblings alike as age differences increased (γ ranging from –0.31 to –0.37, all p<0.001; Figure 2A). By contrast, grooming rates were relatively unrelated to age differences between partners (γ ranging from –0.09 to 0.00, all p>0.05; Figure 2B).

Figure 1 with 2 supplements see all
Box and dot plots comparing relatedness categories (A, B) and sex categories (C, D) for play rates (left) and grooming rates (right).
Estimated rates of play (A) and grooming (B) across a range of age differences, separated by relatedness category.

Bars represent 95% CI for rates of behavior at a given age difference.

Male-male dyads (n=503) played more than either mixed-sex (n=977) or female-female dyads (n=454); conversely, female-female dyads groomed each other more than either mixed-sex or male-male dyads (all p<0.001; Figure 1C and D). These patterns for play and grooming were strongly moderated by age differences (ps<0.006) and were significantly, albeit more weakly, moderated by relatedness (p=0.012 and 0.045, respectively). Play dropped rapidly with increasing age differences (γ = –0.37 to –0.30) for all sex configurations (all p<0.001; Figure 3A). Grooming was steadily low in male-male and mixed-sex dyads (γ = –0.02 and –0.03, p>0.32), though it dropped with increasing age differences in female-female dyads (γ = –0.11, p<0.001), such that differences between sex categories became indistinguishable after approximately a 10-year age difference (Figure 3B). Play was consistently highest among male-male dyads within all relatedness categories, with the exception of maternal siblings, who exhibited more comparable rates of play among male-male and mixed-sex dyads (Figure 1—figure supplement 1). Grooming was also consistently highest among female-female dyads of all types, though the magnitude of this difference varied between relatedness categories (Figure 1—figure supplement 2).

Estimated rates of play (A) and grooming (B) across a range of age differences, separated by sex category.

Bars represent 95% CI for rates of behavior at a given age difference.

When restricting analyses to dyads who lived in the same social group during the first year of the younger partner’s life (n=6724 dyad-years), results for affiliative behavior remained qualitatively similar: full and maternal siblings played and groomed more than paternal siblings or non-siblings (see Appendix 1—tables 1 and 2). Interactions between relatedness categories and sex makeup weakened and were no longer significant (for play: p=0.387; for grooming: p=0.347); sex category × age difference interactions remained significant for grooming (p<0.001) and for play (p=0.031). Additionally, all reported results were robust to the inclusion of average age of the dyad as a covariate (Appendix 3).

Time spent in proximity

The time dyads spent in close proximity (<2 m) with each other also varied between relatedness categories (p<0.001), with maternal siblings and full siblings once again spending more time near each other than non-siblings, who themselves spent more time in close proximity than paternal siblings did (all comparisons p<0.001; Figure 4A). However, these patterns too were moderated by age differences (p<0.001). Proximity decreased with increasing age differences in maternal siblings and paternal siblings (γ = –0.08 and –0.09, p<0.001), but did not decrease significantly in full siblings or non-siblings (γ = –0.04 and 0.01, p>0.29). Thus, while all classes of siblings spent more time near each other than non-siblings when near in age, even when adjusting for their mother’s presence, this distinction was partially reversed at large age differences, when paternal siblings spent much less time near each other than any other dyad category (Figure 4B).

Box and dot plots (A) and estimated trends across a range of age differences (B) for the time gorilla dyads spent in close proximity, separated by relatedness category.

Bars in (B) represent 95% CI for rates of proximity at a given age difference.

When restricting analyses to dyads who lived in the same social group during the first year of the younger partner’s life, results for proximity remained consistent: full and maternal siblings spent more time together than paternal siblings or non-siblings, though this difference was reversed at large age differences in the same manner as our primary model (see Appendix 1—tables 1 and 2 for full results).

Competitive behaviors

Neither relatedness nor sex category on their own significantly predicted rates of aggressive behavior (p=0.876 and 0.838, respectively). However, our model did reveal a significant sex makeup × relatedness interaction term (p=0.025; Figure 5A). Decomposing this interaction, among female-female and male-male dyads, there were no statistically significant contrasts between relatedness categories. In mixed-sex dyads, non-siblings engaged in substantially more aggression than any sibling category (all p<0.050). While sex category also interacted with age differences in predicting aggression (p=0.009), marginal trends were consistently negative (γ = –0.14 to –0.08, p<0.001), such that, across sex categories, dyads more distant in age engaged in less aggression than dyads closer in age (Figure 5). All reported results were robust to the inclusion of average age of the dyad as a covariate (Appendix 3).

Figure 5 with 1 supplement see all
Box and dot plots (A) and estimated trends across a range of age differences (B) for aggression within gorilla dyads, separated by relatedness and sex category.

Bars in (B) represent 95% CI for rates of aggression at a given age difference.

Unlike results for affiliative behavior, when restricting analyses to dyads who lived in the same social group during the first year of the younger partner’s life, results for aggressive behavior qualitatively shifted. Relatedness and sex categories no longer significantly interacted (p=0.444), and the significant pairwise differences between siblings and non-siblings in mixed-sex dyads shrank markedly and were no longer significant (all p>0.60). Age difference remained a highly significant predictor of aggression (p<0.001). Inspection of our data confirmed that much aggression specifically occurred in the context of females transferring into groups–that is, when females first encountered unrelated, unfamiliar males–accounting for non-significant pairwise differences among male-female dyads with early-life familiarity (Figure 5—figure supplement 1). See Appendix 1—tables 1 and 2 for full results.

Discussion

In a comprehensive examination of dyadic mountain gorilla social relationships spanning 14 years and nearly 40,000 hr of observation, we find complex patterns of affiliation and competition within gorilla pairs that speak to sex-, age-, and relatedness-specific social biases. In general, siblings affiliated with each other more and spent more time together than non-siblings, even when accounting for the presence of mothers. But within siblings, affiliative patterns further varied: full and maternal siblings were in most cases much more affiliative than paternal siblings, whose behavior more closely resembled that of non-siblings. We consistently observed a trend for male-male dyads to play more, for female-female dyads to groom more, and for mixed-sex dyads to fall intermediate between these groups. Examining competitive behaviors, on the other hand, revealed a narrower sibling bias. Aggression was most common in mixed-sex non-sibling dyads, and larger age differences similarly predicted less aggression across all dyad types.

At the broadest level, our results support the existence of affiliative biases towards kin in mountain gorillas. While past research has been largely equivocal about the extent of kin discrimination that relies on mechanisms beyond familiarity (e.g. Wikberg et al., 2014; Godoy et al., 2016; Lynch et al., 2017)–and indeed, our results do support a role of familiarity (at least as indexed by age difference) in structuring social interactions–our findings are unlikely to be entirely explained by mere exposure for at least three reasons:

First, gorilla social groups are tight-knit and cohesive compared to their close ape relatives (Goodall, 1986; Remis, 1997; Morrison et al., 2021b; Schaller and Emlen, 1963; Doran and McNeilage, 1998), reflected by remarkable features such as the ability for these groups to buffer the fitness costs of maternal loss (Morrison et al., 2021a). Due to this highly cohesive structure, all individuals in a group, related or not, are very likely quite familiar with one another. Even within this tight-knit context, however, ubiquitous 'play groups' of infants and juveniles frequently form around adult males (Rosenbaum and Silk, 2022), providing similar-aged animals even more opportunities to interact and become familiar with one another. Second, prosocial biases towards siblings are not fully explained by the greater familiarity that closer-in-age animals likely have (Widdig et al., 2001; Smith et al., 2003; Lynch et al., 2017). We observed clear biases towards kin at all but the largest age differences–and some siblings in our data set were 20 or more years apart in age–even though sibling and non-sibling age-mates in the same social group would typically be expected to possess close familiarity. Finally, while mothers undoubtedly mediate social interactions of offspring, especially for immature individuals, biases towards siblings persist even when adjusting for the frequency of her presence during interaction periods. Jointly, these considerations suggest a sibling bias in mountain gorillas subject to influence, but not determination, by demographic factors, which we interpret as aiding in the development of sibling relationships that exist across timescales rarely observed in other non-human primates.

The observation that full and maternal siblings groomed, played, and spent more time near each other than paternal siblings or non-siblings, who tended to affiliate at comparable rates, further suggests that mountain gorillas, like several other primate species studied (Langergraber et al., 2007; Silk et al., 2006; Lynch et al., 2017), evince much stronger maternal than paternal kin bias (see also Rosenbaum et al., 2015, who found little evidence for paternal kin discrimination among fathers and offspring). Interestingly, this ‘asymmetric bias’ in affiliation seems to persist even though the high reproductive skew of mountain gorillas opposes the dynamic thought to underlie such biases in other primate species (i.e. low reproductive skew; Galezo et al., 2022). Thus, one question concerns why mountain gorillas do not appear to more strongly favor paternal siblings. Current evidence indicates that single-male gorilla groups across research sites are genetically polygynous (reviewed in Rosenbaum and Silk, 2022), and while there can be considerable temporal variation, reproductive skew is generally much higher in multi-male gorilla groups than in, for example, chimpanzee, savannah baboon, or rhesus macaque groups (Vigilant et al., 2015; Surbeck et al., 2017; Alberts et al., 2003; Widdig et al., 2004). We propose that, despite possessing a mating system quite unlike these other primate species, mountain gorillas still exhibit a comparable maternal sibling bias due to a mismatch between their historical mating structure–which we infer consisted of highly polygynous one-male units–and their contemporary social structure of tight-knit, often multi-male groups. In other words, while individuals in highly polygynandgrous groups might find it too difficult to detect and adjust affiliation toward paternal kin, perhaps mountain gorillas fail to do so because, until very recently, it was unnecessary. If co-residency was enough to identify paternal kin with reasonable accuracy, a more sophisticated recognition mechanism would be unlikely to evolve.

One important limitation of our analyses is that we have far less information about sibling relationships in single-male groups—where shared paternity is nearly or entirely certain—than we do about siblings in multi-male groups, where reproductive skew is high but not 100%. The structure of the research groups during our 14 years of data was primarily multi-male, with only 60 dyad-years of data (n=10 total dyads) from single-male groups. All of these dyads were paternal half-siblings (i.e. there were no unrelated or maternally related dyads available in single-male groups). This small sample size and lack of variability in relatedness precludes a test of whether natal group structure predicts features of sibling relationships. If all available partners are paternal kin, then discrimination becomes a question of full versus paternal siblings, rather than discrimination among four categories (full, half maternal, half paternal, and unrelated). Choice is constrained, perhaps limiting the utility of a discrimination mechanism. It would be interesting to know whether animals in single-male groups treat paternal versus full siblings differently, but the currently available data do not allow us to answer this question.

Notably, while we see little evidence that mountain gorillas show a prosocial bias towards paternal siblings, patterns of aggressive behavior suggest there may still be kin recognition mechanisms at play for all sibling types. Aggression remained low across most combinations of relatedness and sex configurations, with one exception: mixed-sex interactions among non-siblings. This pattern is consistent with males deploying aggression in the context of mate attraction or coercion. Our interpretation is further bolstered by the distinct statistical patterns that emerge for aggression when examining unrelated mixed-sex dyads with early-life familiarity, which suggest much of the mixed-sex aggression in our data took place among unrelated and unfamiliar male-female dyads. These encounters were most common when females transferred into new groups and were thus likely to be particularly attractive mates for males. Past research in gorillas suggests male aggression towards females may have a number of non-mutually exclusive functions: to police female-female aggression, to discourage female dispersal or mate choice, or to indicate protective ability or overall condition (Robbins, 2009a; Breuer et al., 2016). The fact that this kind of aggression was observed less frequently among related male-female pairs is another observation consistent with accurate kin discrimination. It also suggests active inbreeding avoidance, to the extent that aggression truly serves a mate attraction function. While death and dispersal have been suggested to obviate the need for sophisticated inbreeding avoidance mechanisms in some primates (e.g. baboons; Galezo et al., 2022), such an explanation is unlikely to apply to contemporary mountain gorillas. Living with opposite-sex relatives after sexual maturity is a routine occurrence in this species. Prior research confirms strong inbreeding avoidance between father-daughter dyads in this species (Vigilant et al., 2015), but further work is needed to investigate the extent to which male mate choice is manifested via female-directed aggression, and whether females, for their part, possess additional mechanisms to avoid mating with kin, including paternal siblings.

Together, these observations–prosocial biases towards kin that do not appear to be fully explained by familiarity; a stronger maternal than paternal sibling prosocial bias; and avoidance of intersexual aggression across all sibling types–both speak to key questions about the development of great ape sibling relationships and present two additional puzzles for interpretation. First, traditional mechanistic explanations for sibling biases that typically appeal to exposure during developmental periods appear largely inconsistent with our results and the nature of mountain gorilla sociality, in which siblings and non-siblings, and maternal and paternal siblings, are all likely to have significant exposure to one another during development. It is possible that maternally mediated early-life exposure effects that we could not measure–e.g., via repeatedly sharing night nests (Fossey, 1979) –function analogously to the manner in which co-residence duration serves as a key component of kin recognition in humans (Lieberman et al., 2007), or that preferential mother-father relationships post-birth might lead to social preferences among siblings (Rosenbaum et al., 2016a). Individuals may also possess some degree of phenotype matching ability (Widdig, 2007; Parr et al., 2010; Langergraber, 2012; Pfefferle et al., 2014).

Second, the lack of evidence for a prosocial bias towards paternal siblings is not readily reconciled with clear behavioral evidence of reduced aggression within these same dyads. This remarkable disjunct between apparent sibling recognition and sibling bias suggests that from a mountain gorilla’s perspective, paternal siblings are known entities that nevertheless are less attractive social partners than maternal siblings, despite each being equal relatives. There may be multiple, non-mutually exclusive explanations for this dynamic. Perhaps the presence of paternal siblings provides fewer benefits to an individual than do other sibling types–this possibility, while previously suggested (e.g. Cords et al., 2018), has not been systematically investigated and is an ideal target for future research. Relatedly, the strength of maternal versus paternal kin bias may to some degree depend on the costs of the behaviors in question (Widdig et al., 2006). A mismatch between historical and current social structure might also lead to inconsistent, weakened kin recognition among paternal siblings that manifests in the contrasting patterns we report. Ultimately, disentangling these potential explanations within a species that only exists in the wild may depend on the opportunity to study long-term mating patterns and the impacts of ‘natural experiments’ such as early maternal loss or adoption (most often carried out by adult males in this species; Fossey, 1979; Morrison et al., 2021a).

Conclusion

Our analyses of sibling relationships in mountain gorillas provide extensive, large-scale information on the dynamics of cooperation and competition in a primate society where, as in humans, potential social partners vary greatly in the genes, developmental stage, and biological sex they share with each other. We find a selective sibling bias for prosocial behaviors, in that siblings who share matrilineal kinship affiliate at greater rates than either paternal siblings or non-siblings, and that this bias weakens as individuals become more distant in age. While such a result is consistent with a wide range of previous research, none of the reasons proposed for this selective bias in primates appear to apply to our population: mountain gorillas gain regular exposure to siblings of all types, across life stages; furthermore, patterns of aggressive behavior, in contrast to affiliation, suggest that mountain gorillas can in fact recognize paternal siblings, though they evidently do not favor them as cooperative partners. Ultimately, our study underscores a diversity of means, some evidently yet to be revealed, through which individuals might perceive and engage in sibling relationships to achieve fitness outcomes.

Materials and methods

Our study subjects came from a population of habituated wild mountain gorillas living in Volcanoes National Park, Rwanda, that have been monitored nearly continuously for the last 54 years by the Dian Fossey Gorilla Fund. Using focal follow and scan data collected by researchers and staff, we compiled a dataset of all available dyadic gorilla behavior spanning the years of 2003–2017. We then supplemented this dataset with demographic and relatedness data (for maternal relatedness, via direct observation; for paternal relatedness, via genetic paternity determination–see e.g. Vigilant et al., 2015) on individuals pulled from long-term records. From this combined dataset, we excluded interactions with infants <1 year of age at time of observation, parent-offspring interactions, and interactions between dyads for which we could not calculate relatedness from available data. This yielded a final, curated dataset containing 157 unique individuals studied from late infancy to adulthood (75 F, 82 M; average age at time of observation = 9.75 years; age range: 1–38.5 years old) and 38,996 total hours of observation.

Composition of dyads

Request a detailed protocol

Our dataset of behavior from 157 individuals contained 1934 unique dyad pairs. Of these dyads, 1235 shared neither a mother nor father (‘non-siblings’), 555 shared a father but not a mother (‘paternal siblings’), and 43 shared both a mother and a father (‘full siblings’). In addition to dyads known to share a mother but not a father (n=50), there were a number of dyads with the same mother, but with paternity data missing for one or both individuals (n=51). To maximize sample size, we combined these two groups into the category of ‘maternal siblings’; due to this analytic choice, this category can be effectively conceived of as ‘at least maternal siblings’. See Appendix 2—Tables 1 and 2 for analyses using only confirmed maternal siblings, which were very similar to those reported below. Mixed-sex dyads were the most common sex category in our dataset (n=977), followed by male-male (n=503) and female-female (n=454). Dyads differed in age by an average of 5.85 years (SD: 4.53 years; range: 0–23.5 years); for reference, the average interbirth interval in mountain gorillas is 3.9 years (Eckardt et al., 2016). We used this continuous age difference variable as our primary index of familiarity between individuals, following a number of previous studies on primate kinship (e.g., Widdig et al., 2001; Pfefferle et al., 2014; Wikberg et al., 2014). While we had information on shared group membership in early life, which could also serve as a potential index of familiarity within dyads, we do not focus on this variable in our primary analyses, as it did not allow us to disambiguate between relatedness and familiarity–dyads of individuals who grew up in different natal groups were virtually never (n=3) siblings in our dataset. However, as a robustness check of our main findings, we also performed supplementary analyses on the subset of dyads who lived in the same social group during the first year of the younger partner’s life; i.e., those who had substantial early-life familiarity with one another (n=6,724 dyad-years). We report the correspondence between these analyses and our primary models in our main results section, and we provide full output of these secondary models in our appendix (see Appendix 1—Tables 1 and 2).

Behavioral measures

Request a detailed protocol

We evaluated five different categories of dyadic behaviors as outcome variables: grooming, playing, non-contact aggression, contact aggression, and time spent in close (2 m) proximity. We operationalized these behaviors from standardized definitions used in previous publications about this gorilla population (see e.g. Rosenbaum et al., 2015). Trained observers regularly undergo interobserver reliability tests. The former four behavioral categories were evaluated as counts (corrected for exposure time; see Data analysis) within the dyad during focal observations, regardless of directionality, while the latter category of time in close proximity was evaluated by counting the number of instantaneous scan samples in which a dyad was observed within 2 m of each other (also corrected for exposure time). Across primates, a substantial body of work has investigated how rates of these social behaviors might change with age (e.g. play: Fagen, 1993; aggression: Del Giudice et al., 2009; Kulik et al., 2015; Grebe et al., 2019; grooming: Almeling et al., 2016; Schino and Pinzaglia, 2018). General age-related trends in our dataset corroborate previously established patterns such as large play decreases, large aggression increases, and moderate grooming decreases as average age within the dyad increases (Appendix 3—figure 1). Importantly, our primary results pertaining to relatedness, sex configuration, and age differences are robust to the inclusion of age as a covariate; see Appendix 3—Tables 1 and 2.

Data analysis

Request a detailed protocol

We conducted all analyses in R (version 4.1.2). Our main statistical models for each behavioral outcome consisted of cross-classified generalized linear mixed models (conducted using the glmmTMB package; Brooks et al., 2017) these models included separate random intercept terms for each individual within the dyad and the dyad itself, in addition to random slope terms for relatedness, age difference, and sex makeup within dyads. Given low incidences of many behaviors, we aggregated behaviors into annual counts, making the dyad-year the fundamental unit of analysis (total n=7832). Even with annual aggregation, instances of aggression were uncommon. Therefore, counts of contact and non-contact aggression were summed into a single category for analysis (see Appendix 4—figures 1 and 2; Appendix 4—Tables 1 and 2 for results with individual aggression categories, which were qualitatively similar to those reported below).

In models predicting each behavioral outcome, we included terms for relatedness, age difference, and sex makeup, as well as two-way interactions between relatedness and sex makeup, relatedness and age difference, and sex makeup and age difference. As mothers plausibly mediate many of the social behaviors we examined, especially early in life, we also included the proportion of observations with mothers in close proximity (i.e. the average proportion between the two members of the dyad for that observation year), and this variable’s interaction with relatedness, as covariates in all models. We did not include the natal group structure of the dyad (i.e. single versus multi-male) due to the small number of dyads we had available in which both individuals grew up single-male groups (n=10 dyads, 60 dyad-years), and due to the lack of variability of relatedness–all dyads in these groups were paternal siblings.

In models containing significant main effect or interaction terms, we decompose omnibus comparisons and report targeted marginal effects and contrasts using the emmeans package (Lenth, 2022), with all reported p-values corrected for false discovery rate. We modeled our count outcomes as rates with a negative binomial family in glmmTMB and an offset term for exposure time (either logged hours of observation, or logged sum of scans for both individuals, per dyad-year). Regression coefficients (γ) are reported as changes in the log of the outcome variable with each unit increase in the predictor variable. For each model, we verified model fit by inspecting the deviation, dispersion, and outliers of residuals using the DHARMa package (Hartig, 2022). All data and code necessary to reproduce our results are available publicly at https://osf.io/6qgj5.

Appendix 1

Results for analyses restricted to dyads with “early-life familiarity” (see Materials and methods for details of categorization; total dyad-years: 6724). Full model results available from data and code posted publicly at https://osf.io/6qgj5.

Appendix 1—table 1
Omnibus statistics for target parameters.

Effects p<0.05 bolded.

Effect (early life familiarity only)
RelatednessSex categoryAge differencesRel×sex categoryRel×age diffSex ×age diff
PlayF(3, 6682) = 1.61,
p = 0.184
F(2, 6682) = 5.79,
p = 0.003
F(1, 6682) = 167.31,
p < 0.001
F(6, 6682) = 1.06,
p = 0.387
F(3, 6682) = 0.10,
p = 0.963
F(2, 6682) = 3.47,
p = 0.031
GroomingF(3, 6682) = 24.15,
p < 0.001
F(2, 6682) = 3.83,
p = 0.022
F(1, 6682) = 5.28,
p = 0.022
F(6, 6682) = 1.12,
p = 0.347
F(3, 6682) = 0.94,
p = 0.419
F(2, 6682) = 9.56,
p < 0.001
ProximityF(3, 6682) = 43.90,
p < 0.001
F(2, 6682) = 3.58,
p = 0.028
F(1, 6682) = 23.33,
p < 0.001
F(6, 6682) = 0.36,
p = 0.904
F(3, 6682) = 11.56,
p < 0.001
F(2, 6682) = 0.26,
p = 0.771
AggressionF(3, 6682) = 0.18,
p = 0.910
F(2, 6682) = 0.67,
p = 0.510
F(1, 6682) = 34.49,
p < 0.001
F(6, 6682) = 0.97,
p = 0.444
F(3, 6682) = 0.59,
p = 0.621
F(2, 6682) = 4.46,
p = 0.012
Appendix 1—table 2
Estimated marginal means and standard errors across relatedness and sex categories.
RelatednessSex Category
Full siblingsMaternal halfPaternal halfNon-siblingsFemale-femaleMale-maleMixed-sex
Play2.35 (0.40)1.72 (0.33)0.99 (0.09)1.33 (0.10)1.16 (0.16)2.23 (0.26)1.36 (0.15)
Grooming1.94 (0.43)1.44 (0.27)0.22 (0.03)0.27 (0.03)1.57 (0.19)0.30 (0.05)0.56 (0.07)
Proximity42.5 (3.6)41.4 (2.9)14.2 (0.7)15.5 (0.6)30.4 (1.7)22.8 (1.4)22.4 (1.2)
Aggression0.41 (0.07)0.44 (0.08)0.55 (0.03)0.59 (0.03)0.39 (0.05)0.62 (0.06)0.48 (0.04)

Appendix 2

Results for analyses using a ‘stricter’ categorization of maternal siblings (n = 50; see Materials and methods for details of categorization)

Appendix 2—table 1
Omnibus statistics for target parameters.

Effects p<0.05 bolded.

Effect (strict relatedness categories)
RelatednessSex categoryAge differencesRel×sex categoryRel×age diffSex ×age diff
PlayF(3, 7586) = 1.23,
p = 0.296
F(2, 7596) = 2.65,p = 0.071F(1, 7586) = 121.42,
p < 0.001
F(6, 7596) = 2.60,
p = 0.016
F(3, 7586) = 1.01,
p = 0.387
F(2, 7596) = 5.66,
p = 0.003
GroomingF(3, 7586) = 15.28,
p < 0.001
F(2, 7596) = 2.82,
p = 0.060
F(1, 7586) = 3.74,
p = 0.053
F(6, 7596) = 1.36,
p = 0.229
F(3, 7586) = 2.73,
p = 0.043
F(2, 7596) = 8.26,
p < 0.001
ProximityF(3, 7586) = 31.67,
p < 0.001
F(2, 7596) = 2.34,
p = 0.097
F(1, 7596) = 16.75,
p < 0.001
F(6, 7596) = 1.30,
p = 0.255
F(3, 7586) = 18.83,
p < 0.001
F(2, 7596) = 0.25,
p = 0.778
AggressionF(3, 7586) = 0.14,
p = 0.936
F(2, 7596) = 0.19,
p = 0.825
F(1, 7596) = 21.96,
p < 0.001
F(6, 7596) = 2.17,
p = 0.043
F(3, 7586) = 0.16,
p = 0.927
F(2, 7596) = 5.21,
p = 0.006
Appendix 2—table 2
Estimated marginal means and standard errors across relatedness and sex categories.
RelatednessSex category
Full siblingsMaternal halfPaternal halfNon-siblingsFemale-femaleMale-maleMixed-sex
Play2.35 (0.43)2.16 (0.43)1.08 (0.11)1.04 (0.08)1.23 (0.17)2.45 (0.31)1.22 (0.16)
Grooming2.19 (0.46)1.67 (0.41)0.26 (0.03)0.31 (0.03)1.55 (0.19)0.39 (0.08)0.68 (0.09)
Proximity43.4 (3.7)42.1 (4.0)15.1 (0.7)17.7 (0.7)31.4 (1.8)23.5 (1.6)25.0 (1.5)
Aggression0.47 (0.08)0.62 (0.14)0.61 (0.04)0.75 (0.04)0.53 (0.07)0.70 (0.09)0.60 (0.06)

Appendix 3

Results for analyses including average age as a covariate in analyses.

Appendix 3—table 1
Omnibus statistics for target parameters.

Effects p<0.05 bolded.

Effect (average age adjusted)
RelatednessSex categoryAge differencesAverage ageRel.×sex categoryRel.×age diff.Sex ×age diff.
PlayF(3, 7782) = 5.92,
p < 0.001
F(2, 7782) = 6.86,
p = 0.001
(1, 7782) = 105.95,
p < 0.001
F(1, 7782) = 859.52,
p < 0.001
F(6, 7782) = 1.22,
p = 0.294
F(3, 7782) = 1.94,
p = 0.121
F(2, 7782) = 8.99,
p < 0.001
GroomingF(3, 7782) = 23.14,
p < 0.001
F(2, 7782) = 5.74,
p = 0.003
F(1, 7782) = 3.20,
p = 0.074
F(1, 7782) = 11.83,
p < 0.001
F(6, 7782) = 2.16,
p = 0.044
F(3, 7782) = 3.31,
p = 0.019
F(2, 7782) = 6.68,
p = 0.001
ProximityF(3, 7782) = 39.59,
p < 0.001
F(2, 7782) = 3.79,
p = 0.023
F(1, 7782) = 31.58,
p < 0.001
F(1, 7782) = 23.46,
p < 0.001
F(6, 7782) = 1.14,
p = 0.339
F(3, 7782) = 25.41,
p < 0.001
F(2, 7782) = 0.10,
p = 0.907
AggressionF(3, 7782) = 0.24,
p = 0.871
F(2, 7782) = 0.35,
p = 0.703
F(1, 7782) = 40.38,
p < 0.001
F(1, 7782) = 103.42,
p < 0.001
F(6, 7782) = 1.55,
p = 0.158
F(3, 7782) = 0.11,
p = 0.957
F(2, 7782) = 4.73,
p = 0.009
Appendix 3—figure 1
Estimated trends of (A) play, (B) grooming, (C) close proximity, and (D) aggression across average age of gorilla dyads.

Trends adjusted for all parameters listed in Appendix 3—table 1. Bars represent 95% CI for rates of behavior at a given age .

Appendix 3—table 2
Estimated marginal means and standard errors across relatedness and sex categories, adjusting for average age.
RelatednessSex category
Full siblingsMaternal halfPaternal halfNon-siblingsFemale-femaleMale-maleMixed-sex
Play1.77 (0.29)1.61 (0.22)0.77 (0.06)0.87 (0.05)0.74 (0.08)2.20 (0.21)1.00 (0.10 )
Grooming2.08 (0.45)1.54 (0.29)0.26 (0.03)0.31 (0.03)1.61 (0.19)0.36 (0.06)0.62 (0.08)
Proximity44.6 (4.3)41.3 (2.9)15.1 (0.7)17.9 (0.7)31.6 (1.7)24.1 (1.5)24.2 (1.4)
Aggression0.46 (0.07)0.42 (0.08)0.62 (0.04)0.75 (0.04)0.45 (0.05)0.66 (0.07)0.55 (0.05)

Appendix 4

Results for analyses assessing contact aggression and non-contact aggression separately.

Appendix 4—table 1
Omnibus statistics for target parameters.

Effects p<0.05 bolded.

Effect (separate aggression categories)
RelatednessSex categoryAge differencesRel.×sex categoryRel.×age diff.Sex ×age diff.
Contact
aggression
F(3, 7783) = 2.72,
p = 0.042
F(2, 7596) = 1.02,
p = 0.359
F(1, 7596) = 14.29,
p < 0.001
F(6, 7596) = 2.11,
p = 0.049
F(3, 7586) = 0.72,
p = 0.541
F(2, 7596) = 2.19,
p = 0.113
Non-contact
aggression
F(3, 7783) = 0.28,
p = 0.834
F(2, 7596) = 2.71,
p = 0.067
F(1, 7596) = 7.11,
p = 0.008
F(6, 7596) = 2.09,
p = 0.052
F(3, 7586) = 0.11,
p = 0.955
F(2, 7596) = 1.33,
p = 0.264
Appendix 4—table 2
Estimated marginal means and standard errors across relatedness and sex categories.
RelatednessSex Category
Full siblingsMaternal halfPaternal halfNon-siblingsFemale - femaleMale - maleMixed -
sex
Contact
aggression
0.25 (0.04)0.19 (0.05)0.31 (0.02)0.35 (0.02)0.22 (0.03)0.29 (0.04)0.30 (0.03)
Non-contact
aggression
0.12 (0.04)0.18 (0.05)0.18 (0.02)0.23 (0.02)0.21 (0.03)0.18 (0.03)0.14 (0.02)
Appendix 4—figure 1
Box and dot plots showing estimated non-contact aggression within gorilla dyads, separated by relatedness and sex category.
Appendix 4—figure 2
Box and dot plots showing estimated contact aggression within gorilla dyads, separated by relatedness, and sex category.

Data availability

All data and code necessary to reproduce our results are available publicly at https://doi.org/10.17605/OSF.IO/6QGJ5.

The following data sets were generated
    1. Grebe NM
    2. Hirwa JP
    3. Stoinski TS
    4. Vigilant L
    5. Rosenbaum S
    (2022) Open Science Framework
    Dynamics of cooperation and competition in mountain gorilla siblings.
    https://doi.org/10.17605/OSF.IO/6QGJ5

References

    1. Alberts SC
    2. Altmann J
    (2012)
    In Long-Term Field Studies of Primates
    261–287, The amboseli baboon research project: 40 years of continuity and change, In Long-Term Field Studies of Primates, Berlin, Heidelberg, Springer, 10.1007/978-3-642-22514-7_12.
  1. Book
    1. Daly M
    2. Wilson M
    (1988)
    Homicide
    A. de Gruyter.
    1. Fagen R
    (1993)
    Juvenile Primates
    182–196, Primate juveniles and primate play, Juvenile Primates, Oxford University Press.
    1. Fossey D
    (1979)
    The Great Apes
    137–184, Development of the mountain gorilla (Gorilla gorilla beringei), through the first thirty-six months, The Great Apes, The Benjamin-Cummings Publishing Company.
  2. Book
    1. Goodall J
    (1986)
    The Chimpanzees of Gombe: Patterns of Behavior
    The Belknap Press of Harvard University.
    1. Gray P
    (2019)
    The Cambridge Handbook of Play: Developmental and Disciplinary Perspectives
    84–102, Evolutionary functions of play: practice, resilience, innovation, and cooperation, The Cambridge Handbook of Play: Developmental and Disciplinary Perspectives, Cambridge University Press.
  3. Book
    1. Harcourt AH
    2. Stewart KJ
    (2007)
    Gorilla Society: Conflict, Compromise, and Cooperation Between the Sexes
    Chicago IL: The University of Chicago Press.
  4. Book
    1. Lamb ME
    2. Sutton-Smith B
    (2014)
    Sibling Relationships: Their Nature and Significance across the Lifespan
    New York, NY: Psychology Press.
    1. Langergraber KA
    (2012)
    The Evolution of Primate Societies
    491–513, Cooperation among kin, The Evolution of Primate Societies, London, The University of Chicago Press.
  5. Thesis
    1. Mirville MO
    (2018)
    The causes and consequences of intergroup interactions in mountain gorillas (Gorilla beringei beringei). PhD Thesis
    University of Western Australia.
  6. Book
    1. Mock D
    2. Parker G
    (1997)
    The Evolution of Sibling Rivalry
    Oxford University Press.
    1. Robbins MM
    (2009)
    Sexual Coercion in Primates
    Male aggression toward females in mountain gorillas: courtship or coercion, Sexual Coercion in Primates, Cambridge, Harvard University Press.
  7. Book
    1. Schaller GB
    2. Emlen JT
    (1963) Observations on the ecology and social behavior of the mountain gorilla
    In: Howell FC, Bourliere F, editors. African Ecology and Human Evolution. Aldine, Chicago: Routledge. pp. 686–687.
    https://doi.org/10.1126/science.144.3619.686
    1. Silk JB
    (2009) Nepotistic cooperation in non-human primate groups
    Philosophical Transactions of the Royal Society of London. Series B, Biological Sciences 364:3243–3254.
    https://doi.org/10.1098/rstb.2009.0118

Decision letter

  1. Ammie K Kalan
    Reviewing Editor; University of Victoria, Canada
  2. George H Perry
    Senior Editor; Pennsylvania State University, United States

Our editorial process produces two outputs: (i) public reviews designed to be posted alongside the preprint for the benefit of readers; (ii) feedback on the manuscript for the authors, including requests for revisions, shown below. We also include an acceptance summary that explains what the editors found interesting or important about the work.

Decision letter after peer review:

Thank you for submitting your article "Neither kin selection nor familiarity explain affiliative biases towards maternal siblings in wild mountain gorillas" for consideration by eLife. Your article has been reviewed by 3 peer reviewers, and the evaluation has been overseen by a Reviewing Editor and George Perry as the Senior Editor. The reviewers have opted to remain anonymous.

The reviewers have discussed their reviews with one another, and the Reviewing Editor has drafted this to help you prepare a revised submission.

Essential revisions:

All three reviewers and I read this manuscript with great interest. This study has the potential to provide a strong and novel contribution to our understanding of kin selection in the primate lineage but only if the following major concerns can be addressed.

1. The operationalization of familiarity as 'age difference' is problematic for a number of reasons. I would suggest the authors avoid using the term familiarity (in the title and elsewhere) and instead simply use the term 'age difference' when appropriate. However, given the importance of familiarity with this topic, it will still need to be introduced and discussed as an alternative explanation for the results (see all three Reviewer comments for details).

2. Can the authors include random slopes in their analyses? Such a large dataset would be ideal, if not necessary, to fit random slopes in their models and it could have a critical influence on the results.

3. Given that there are both single-male and multi-male groups in this analysis could you please (a) test/control for the influence of natal group structure (single or multi-male) in your analysis and (b) add the natal group structure to your csv file for each individual gorilla? We think this is critical to the study, given that in single male groups (especially those with long silverback tenure), virtually everyone would be paternally related therefore the space for kin selection to operate would be significantly reduced (see also Review #2 for detailed comments).

4. The authors need to address how the social behaviours they examine vary with the age of individuals and how this might affect their results (see also Reviewers #1 and #3 for detailed comments).

5. The authors need to provide clear predictions as to why kin selection would even benefit mountain gorilla siblings, and for both sexes. How are mountain gorilla competition and cooperation possibly affected? Please go beyond the general theory and provide specific expectations based on the social system of these mountain gorillas (see also Reviewers #2 and #3 for detailed comments). Also please summarize the main results of critical studies cited in the introduction to ease readership for a non-specialist audience.

Reviewer #1 (Recommendations for the authors):

I think that this paper represents an impressive use of the long-term data at Karisoke. My main concern with the main arguments in the paper is that the discussion leaves little room for a simpler familiarity mechanism and, as the analysis stands, the argument that gorillas can recognize their paternal kin, specifically, has not been thoroughly tested. I understand the constraint that there were very few out-group related pairs, and that dominant males do sire a majority of offspring, but what about in-group unrelated dyads? If the sample is large enough, then this work stands the best chance of putting a finger on the mechanism. If not, that's fine too, but the limitation should be expanded upon in the discussion.

A few other analytical points: the first is one that I mention in the public review about age. I apologize that I don't have a specific idea of the absolute best way to deal with the problem. I would be satisfied with any evidence you can provide in the main body or the supplementary that age-related changes in social behavior weren't a major shaping force on affiliation/aggression between dyads.

Social behaviors were aggregated into annual counts – was there any variation in social behaviors between years either overall or within-dyads? My assumption would be that if there were lean years (which I know is relaxed for gorillas) or a group takeover or break-up that this might have a large effect on rates of affiliation and aggression in a given year. Usually, the year might be included as a main effect or a random term to account for any variation that was due to events like that, so I was hoping the authors could clarify here to explain the choice to exclude such a term.

L373-376: Could you please clarify the term for mother's proximity here? Is it the proportion of scans that the mother was in proximity to both partners in a dyad, either partner in a dyad, etc.?

L34-38: These last few sentences are a little clunky and I had to reread them a few times to sort out what the authors meant. Please consider rewording to make the points more clear.

Reviewer #2 (Recommendations for the authors):

My main recommendations would be to:

– Make clearer predictions about expected patterns of affiliation and aggression, e.g. replacing the questions in the paragraph on lines 144-152 with predictions based on the functions of the different behaviors and how they relate to kinship (e.g. aggression in mixed-sex dyads should be lower among kin because it reflects mating). i think this would be useful to go beyond a simplistic notion of 'higher r = more affiliation, less aggression', which then leads to the conclusion that 'kin selection does not explain affiliation'.

– Reconsider the operationalization of familiarity, especially for maternal kin. I just find it hard to believe statements like "sibling and non-sibling age-mates in the same social group would typically be expected to possess close familiarity" (lines 236-237). Surely, the mother is a main focal point for all young gorillas, and they would get more exposure to their maternal siblings than to others simply by virtue of sharing the mother's proximity? As mentioned before, the age difference measure is also somewhat problematic because there cannot be an age difference of 0 (or even <2-3y) for maternal sibs (I assume there weren't many twins), but there can be for paternal sibs; in fact, that's when paternal sibship is most likely.

Reviewer #3 (Recommendations for the authors):

In a revised version I suggest the authors provide more detail about the nature and function of social interactions (including cooperative ones) in mountain gorillas and how they change during development.

In explanations for why maternal but not paternal kin bias occurs in mountain gorillas, one important measure to consider is that a mother may be more permissive of one of her own elder offspring vs. other similarly-aged group members, interacting with her dependent offspring (which might influence bonding between gorillas and their younger group mates – even when mothers are not in close proximity).

This is not a hypothesis that requires testing here, but I think it could be helpful to speak to this – if relevant to the behavior of the study species – in the Discussion. (How often are female gorillas aggressive when others play roughly with their infants or juveniles, for example? If their adult daughter vs. another adult female carried their infant away to play with them would the mother's behavior be different?)

Last, I understand that the title is correct with respect to core kin selection theory, but I personally found it slightly misleading, because the results do find a kin bias, simply a maternal-kin bias and familiarity (though clearly defined as a predictor within the paper) has a vague broader meaning for the title. It refers particularly to familiarity within maternal siblings of varying age (age difference), but it is easily interpreted to mean familiarity on a broad level (e.g. any individuals in the same natal group – which could not be included in models because it was too correlated with other variables). As a suggestion, it might be clearer to make the title something like "Mountain gorillas demonstrate affiliative biases toward maternal but not paternal siblings of all ages."

[Editors’ note: further revisions were suggested prior to acceptance, as described below.]

Thank you for resubmitting your work entitled "Mountain gorillas maintain strong maternal affiliative biases despite high male reproductive skew and extensive exposure to paternal kin" for further consideration by eLife. Your revised article has been evaluated by George Perry (Senior Editor) and a Reviewing Editor.

The manuscript has been much improved but there are some remaining issues that need to be addressed, primarily concerning edits to the text to aid clarity of certain terms, the sampling, and results.

Please take into account the detailed recommendations outlined below by two reviewers when you revise your article:

Reviewer #2 (Recommendations for the authors):

I appreciate all the changes the authors made to the manuscript and find it much improved. The analyses and results are compelling as they are, and my comments are merely intended to further clarify the "packaging". Sorry that the comments turned out to be rather long, it's really not a huge issue but just took some time to explain.

Main comment: From kin selection to kin recognition to expressed kin preferences – I can't help but feel like the argument sometimes comes across as a bit backward, e.g. in the abstract where a preference for maternal kin (over paternal kin) occurs 'despite high relatedness certainty', whereas normally low skew underlies such 'biases' (lines 34-38). This sounds like gorillas lack something, or got something "wrong", though really it's the case that most other primates lack something (high skew) and therefore can't possibly recognize paternal kin, and are limited to maternal kin preferences – hence the bias is borne out of necessity, not because it is somehow preferable. To me at least, the flow of the argument generally goes more like this: From a theoretical perspective, we would predict individuals to preferentially cooperate with kin due to inclusive fitness benefits, which should select for kin recognition whenever possible, though in practice maternal kin recognition is much easier than paternal kin recognition (because one can use simple cues like associating with the same adult female, while paternity uncertainty in polygamous mating systems obviates the use of simple cues). Paternal kin recognition only becomes feasible (in the absence of some kind of phenotypic matching) when one can use similarly simple cues like associating with the same adult male, which becomes more reliable the fewer adult males there are and the higher the reproductive skew. So gorillas have all the ingredients for paternal kin recognition (at least historically), which sets up the expectation in the introduction of paternal kin preferences (over non-kin). yet the results show that they don't seem to express these preferences (with the discussion focusing on why they don't express these preferences, e.g. mismatches in group structure derailing paternal kinship cues, or paternal kin not typically offering cooperative benefits). I know the authors are aware of all this (and in some places the argument is very clear), and I'm sorry for rambling, but I just feel like the argument could be laid out a bit more clearly in some places. Take it or leave it!

Reviewer #3 (Recommendations for the authors):

This revision is much improved from the previous version. I appreciated all the changes the authors have made in responses to all editor and reviewer comments.

I still have several concerns regarding age changes/developmental perspectives in this study:

First I think the manuscript is strengthened by the supplemental analyses that incorporate the age of individuals in each dyad for each year.

However the authors still need to explicitly inform readers that their study includes both immature and mature individuals and provide context for how siblings relationships may change during age that are rooted in gorilla socioecology.

I appreciate that this is understudied:

Here is an additional resource that provides descriptive details on how social interactions between gorillas and their mothers compared to other group members manifest during different developmental stages.

Watts, David P., and Anne E. Pusey. "Behavior of juvenile and adolescent great apes." Juvenile primates: Life history, development, and behavior (2002): 148-167

I also realize that documenting behavioral shifts in development are not the focus of this study.

But not providing basic information about age ranges and what we know and do not know about how relationships change with age limits and even misguides how readers unfamiliar with the study system will interpret findings. For example they may even assume you studied only adult sibling pairs as is true in many of the key kin selection studies cited in the introduction.

In addition, it is not only that behaviors change with age (which the authors address well in this revision) but that dynamics of dyadic relationships do.

For example, as authors touch upon in their introduction, in baboons and chimpanzees sister-brother sibling relationships can be critical to survival and/or improve fitness while one sibling is immature, but not in adulthood based on dispersal and/or species-typical dominance/cooperation patterns (e.g. Engh et al., 2009; Hobaiter et al., 2014).

https://doi.org/10.7554/eLife.80820.sa1

Author response

Essential Revisions (for the authors):

All three reviewers and I read this manuscript with great interest. This study has the potential to provide a strong and novel contribution to our understanding of kin selection in the primate lineage but only if the following major concerns can be addressed.

1. The operationalization of familiarity as 'age difference' is problematic for a number of reasons. I would suggest the authors avoid using the term familiarity (in the title and elsewhere) and instead simply use the term 'age difference' when appropriate. However, given the importance of familiarity with this topic, it will still need to be introduced and discussed as an alternative explanation for the results (see all three Reviewer comments for details).

In response to this comment, and related ones from reviewers, we’ve made a number of changes. We summarize these changes here, but we also respond in detail to specific reviewer comments below.

First, we’ve changed the title to “Mountain gorillas maintain strong maternal affiliative biases despite high male reproductive skew and extensive exposure to paternal kin”, to avoid mentioning familiarity before readers have a chance to understand what we mean by the term.

Next, we recognize that age difference is an imperfect measure, but we feel that it is the best choice amongst the available options (the other being shared natal group). We now explain more about why we think age difference is a useful index of familiarity. First, there is precedent for its use in the literature; second, it is useful for disambiguating between shared natal group and relatedness; and third, mountain gorillas of similar ages spend considerable time in age-structured ‘play groups’ when they are young:

Lines 439-443: “We used this continuous age difference variable as our primary index of familiarity between individuals, following a number of previous studies on primate kinship (e.g., Widdig et al., 2001; Pfefferle et al., 2014; Wikberg et al., 2014). While we had information on shared group membership in early life, which could also serve as a potential index of familiarity within dyads, we do not focus on this variable in our primary analyses, as it did not allow us to disambiguate between relatedness and familiarity–dyads of individuals who grew up in different natal groups were virtually never (n = 3) siblings in our dataset.”

Lines 295-299: “Due to this highly cohesive structure, all individuals in a group, related or not, are very likely quite familiar with one another. Even within this tight-knit context, however, ubiquitous 'play groups' of infants and juveniles frequently form around adult males (Rosenbaum and Silk, 2022), providing similar-aged animals even more opportunities to interact and become familiar with one another.”

Third, we report robustness checks for our main analyses in which we only examine dyads who had “early life familiarity” (i.e., its members belonged to the same group when the younger member was born, and the dyad members were still in the same group at the younger member’s first birthday), in response to Reviewer 1’s comments (see our Response #6):

Lines 443-448: “However, as a robustness check of our main findings, we also performed supplementary analyses on the subset of dyads who lived in the same social group during the first year of the younger partner’s life; i.e. those who had substantial early-life familiarity with one another (n = 6724 dyad-years). We report the correspondence between these analyses and our primary models in our main Results section, and we provide full output of these secondary models in our supplementary materials (see Tables S1 – S2).”

Results for affiliation remain quite similar, while results for aggression change appreciably. We’re grateful for the reviewer’s suggestion for this additional analysis, as we feel it provides valuable context to our results–namely, that even when individuals have extensive exposure to each other, relatedness matters for patterns of affiliation. When it comes to aggression, on the other hand, much of it seems to be driven by males aggressing against unrelated females who transfer into a group (i.e., females who are both unfamiliar and unrelated). We now provide this additional context in two locations (lines 272-275, 351-356):

“Inspection of our data confirmed that much aggression specifically occurred in the context of females transferring into groups–that is, when females first encountered unrelated, unfamiliar males–accounting for non-significant pairwise differences among dyads with early-life familiarity.”

And

“Our interpretation is further bolstered by the distinct statistical patterns that emerge for aggression when examining unrelated mixed-sex dyads with early-life familiarity, which suggest much of the mixed-sex aggression in our data took place among unrelated and unfamiliar male-female dyads. These encounters were most common when females transferred into new groups, and were thus likely to be particularly attractive mates for males.”

2. Can the authors include random slopes in their analyses? Such a large dataset would be ideal, if not necessary, to fit random slopes in their models and it could have a critical influence on the results.

Thank you for raising this point. Indeed, we had originally included maximal random-slopes terms in our models, and though we opted to abandon them in our original submission because of consistent non-convergence issues, this comment led us to revisit the issue. In our revised analyses, we’ve followed an approach that balances the arguments made by Barr et al., (2013) and Matuschek et al., (2017). We begin with the ‘maximal’ random effects model for dyads, and as necessary, we remove individual terms to achieve convergence. The upshot is that all of our models now include random slopes and intercepts, with the correlation between these terms omitted to permit convergence. The qualitative pattern of results from these models is nearly identical to before, with some minor differences in pairwise comparisons. These changes are tracked and reported in the results, and the methods section was updated to clarify that we include both random slopes and intercepts.

Barr, D. J., Levy, R., Scheepers, C., and Tily, H. J. (2013). Random effects structure for confirmatory hypothesis testing: Keep it maximal. Journal of Memory and Language68(3), 255-278.

Matuschek, H., Kliegl, R., Vasishth, S., Baayen, H., and Bates, D. (2017). Balancing Type I error and power in linear mixed models. Journal of Memory and Language, 94, 305-315.

3. Given that there are both single-male and multi-male groups in this analysis could you please (a) test/control for the influence of natal group structure (single or multi-male) in your analysis and (b) add the natal group structure to your csv file for each individual gorilla? We think this is critical to the study, given that in single male groups (especially those with long silverback tenure), virtually everyone would be paternally related therefore the space for kin selection to operate would be significantly reduced (see also Review #2 for detailed comments).

Thanks for this comment. Indeed, as we state in the introduction, mountain gorillas regularly form single-male and multi-male social groups. Interestingly, for a combination of potential reasons—including (a) chance variation during our study period, and (b) the much smaller number of offspring that result from single-male groups in general—there is a comparatively tiny number of rows in our dataset where one or both members of a dyad grew up in a single-male group (210 and 60 rows, respectively, out of 7832; this variable is now included in the public dataset as requested). Additionally, for this n=60 subset (10 dyads in total, because most dyads are represented across multiple years) where both partners had a single-male natal group, all of the dyads were paternal siblings, meaning we have no relatedness variation. There were zero full sibling, maternal sibling, or unrelated dyads. Given these constraints of the data, we do not feel that we can make any meaningful claims about the effect of natal group structure on sibling relationships. Karisoke currently follows several single-male groups, so we hope that this is something we will be able to follow up on in the future when more data from these groups becomes available.

We have now included a paragraph about this limitation in the discussion (lines 333-345):

“One important limitation of our analyses is that we have far less information about sibling relationships in single-male groups—where shared paternity is nearly or entirely certain—than we do about siblings in multi-male groups, where reproductive skew is high but not 100%. The structure of the research groups during our 14 years of data was primarily multi-male, with only 60 dyad-years of data (n=10 total dyads) from single-male groups. All of these dyads were paternally related (i.e., there were no unrelated or maternally related dyads available in single-male groups). This small sample size and lack of variability in relatedness precludes a test of whether natal group structure predicts features of sibling relationships. If all available partners are paternal kin, then discrimination becomes a question of full versus paternal siblings, rather than discrimination among four categories (full, half maternal, half paternal, and unrelated). Choice is constrained, perhaps limiting the utility of a discrimination mechanism. It would be interesting to know whether animals in single-male groups treat paternal versus full siblings differently, but the currently available data do not allow us to answer this question.”

Additionally, we added a sentence to the methods section to clarify why this variable was not included in the models (lines 483-486):

“We did not include the natal group structure of the dyad (i.e., single versus multi-male) due to the small number of dyads we had available in which both individuals grew up single-male groups (n=10 dyads, 60 dyad-years), and due to the lack of variability of relatedness–all dyads in these groups were paternal siblings.”

4. The authors need to address how the social behaviours they examine vary with the age of individuals and how this might affect their results (see also Reviewers #1 and #3 for detailed comments).

Please see our Response #7, where we address this point in detail. Briefly, while the rates of these behaviors do indeed change as animals age, accounting for this effect does not meaningfully change the primary results we report. We now include supplementary analyses demonstrating this, and we reference them in our methods section.

5. The authors need to provide clear predictions as to why kin selection would even benefit mountain gorilla siblings, and for both sexes. How are mountain gorilla competition and cooperation possibly affected? Please go beyond the general theory and provide specific expectations based on the social system of these mountain gorillas (see also Reviewers #2 and #3 for detailed comments). Also please summarize the main results of critical studies cited in the introduction to ease readership for a non-specialist audience.

We’ve extensively addressed both the issue of predictions, and the issue of summarizing studies, in our response to individual reviewers (see e.g. our Response #11). To summarize our changes to the first issue, we’ve added two paragraphs to the introduction that clarifies the potential dynamics of kin selection opportunities in gorillas specifically (lines 140-162):

“This means that siblings can be an important source of support well into adulthood. For example, adult males benefit greatly from allies who help them fend off outsider challengers and prevent females from transferring (Sicotte, 1993, Rosenbaum et al., 2016, Mirville et al., 2018). Adult females with offspring benefit from male protection from infanticide (Harcourt and Stewart, 2007, Robbins et al., 2013); in multi-male groups, this protection could potentially come not only from mates, but from brothers as well, who receive indirect fitness benefits from their sister’s reproductive success. While the benefits of female-female relationships remain understudied in this species, higher-ranking females have better energy balance and shorter inter-birth intervals, and it is plausible that support from other females plays a role in achieving and maintaining dominance (Robbins et al., 2005, Wright et al., 2014, Wright et al., 2020).

As noted above, however, sibling relationships are also characterized by the potential for competition as well as cooperation, and mountain gorillas are no exception. Even after co-resident mountain gorilla siblings cease competing over parental resources, they are likely to compete over other limited resources, including dominance positions, mating opportunities, and preferential foods (Harcourt and Stewart, 2007). In some other primate species, rates of aggression are as high or higher among kin than non-kin, in part because kin simply spend more time together, highlighting the complexity of making straightforward predictions about affiliative and aggressive interactions and relatedness (e.g. Silk et al., 2010). On balance, however, we expect that mountain gorillas should be more cooperative/affiliative with siblings than non-siblings due to the substantial potential for inclusive fitness benefits.”

In response to the second issue, we now elaborate on the key results of studies cited in the introduction (see in particular p.4 of our revised manuscript).

Reviewer #1 (Recommendations for the authors):

I think that this paper represents an impressive use of the long-term data at Karisoke. My main concern with the main arguments in the paper is that the discussion leaves little room for a simpler familiarity mechanism and, as the analysis stands, the argument that gorillas can recognize their paternal kin, specifically, has not been thoroughly tested. I understand the constraint that there were very few out-group related pairs, and that dominant males do sire a majority of offspring, but what about in-group unrelated dyads? If the sample is large enough, then this work stands the best chance of putting a finger on the mechanism. If not, that's fine too, but the limitation should be expanded upon in the discussion.

Thanks for this comment. The issue of untangling familiarity and relatedness is one that our team spent considerable time discussing when writing the original draft. Familiarity has been assessed in a number of different ways in previous studies, including, as the reviewer suggests, by shared group membership in early life. When deciding on analyses, we did create a binary variable that assesses “early familiarity”—a dyad was coded as “yes” if its members belonged to the same group when the younger member was born, and if the members were still in the same group at the younger member’s first birthday. The vast majority of rows in the dataset (84%) consist of dyads with early familiarity. Ultimately, because of the confound between this variable and relatedness (which the reviewer notes), we decided to instead focus on age differences as our primary measure. Age differences are also frequently used to assess familiarity in other studies of primate kinship (see e.g. Widdig et al., 2002; Pfefferle et al., 2014; Wikberg et al., 2014; we now cite these studies where we further explain our use of age differences as an index of familiarity in the introduction).

Regardless, we are happy to oblige the reviewer’s request and only examine dyads who had early-life familiarity. When re-running our analyses on just these dyads, results are qualitatively very similar for affiliation: full and maternal siblings play, groom, and spend more time together than paternal siblings or non-siblings. Interestingly, patterns of aggression are qualitatively different: mixed-sex non-siblings no longer stand out as sources of high aggression. This suggests that much of this aggression occurs within “unfamiliar” mixed-sex dyads, i.e. when females transfer into new groups. We’re grateful for the reviewer suggesting this analysis, as we think it provides some very interesting context for interpreting our results. Thus, we report these new analyses in two locations of the revised manuscript: first, as a supplementary robustness check for our results; and second, as context for our observed patterns of aggressive behavior (e.g. lines 351-356).

A few other analytical points: the first is one that I mention in the public review about age. I apologize that I don't have a specific idea of the absolute best way to deal with the problem. I would be satisfied with any evidence you can provide in the main body or the supplementary that age-related changes in social behavior weren't a major shaping force on affiliation/aggression between dyads.

We’re happy to have the opportunity to address this point. Indeed, it is well-established from previous studies that the rates of social behaviors change with age. Play generally becomes much less frequent, aggression generally becomes more frequent, and patterns of grooming appear to reflect interactions between age and other factors. We find similar patterns in our dataset: average age within the dyad has a massive negative effect on play rates; a marginally significant negative effect on grooming; a positive effect on the frequency of close proximity to others; and a very large positive effect on aggression. Given that these results for age do not meaningfully deviate from previously established patterns, and that they were not the focus of our paper, we elected not to discuss them in the main text. But, to address the reviewer’s concern that these patterns might significantly alter our primary results–something readers might also wonder–we now report the results of models including average age in our supplementary materials (Tables S5 and S6, Figure S4). The upshot is that our results stay consistent, even when controlling for these age-related changes that are sometimes quite large. In other words, the differences we describe in our results are over and above general age trends, which we now note in our Methods section (lines 458-465):

“Across primates, a substantial body of work has investigated how rates of these social behaviors might change with age (e.g. play: Fagen, 1993; aggression: Del Giudice et al., 2009; Kulik et al., 2015; Grebe et al., 2019; grooming: Almeling et al., 2016; Schino and Pinzaglia, 2018). General age-related trends in our dataset corroborate previously established patterns such as large play decreases, large aggression increases, and moderate grooming decreases as average age within the dyad increases (Table S5, Figure S4). Importantly, our primary results pertaining to relatedness, sex configuration, and age differences are robust to the inclusion of age as a covariate; see Tables S5 – S6.”

Social behaviors were aggregated into annual counts – was there any variation in social behaviors between years either overall or within-dyads? My assumption would be that if there were lean years (which I know is relaxed for gorillas) or a group takeover or break-up that this might have a large effect on rates of affiliation and aggression in a given year. Usually, the year might be included as a main effect or a random term to account for any variation that was due to events like that, so I was hoping the authors could clarify here to explain the choice to exclude such a term.

We originally chose not to include a random effects term for year for a few reasons: first, because gorillas don’t share the same environmental constraints that lead to the ‘lean years’ that are, for example, seen during droughts with Amboseli baboons; second, while it’s certainly plausible that social upheaval could affect rates of social behavior, our impression from years in the field with these groups is that behavioral disruption is fairly transient; and finally, we believe year-to-year variation is already captured to a large extent by the repeated sampling of individuals and dyads across multiple years. Any residual variance accounted for by year would have to be ‘on top of’ individual- or dyad-specific variance, and we just didn’t expect there to be very much of that. That said, the reviewer may still wonder about the effect of including year as a random effect. We re-ran our primary analyses including this term and found it did not affect any substantive conclusions–these results are available as part of our code posted publicly on OSF.

L373-376: Could you please clarify the term for mother's proximity here? Is it the proportion of scans that the mother was in proximity to both partners in a dyad, either partner in a dyad, etc.?

Certainly. We now clarify that the “mom’s proximity” variable is coded as the proportion of scans in that year in which an individual’s mother was in close proximity, averaged between the two individuals in the dyad.

L34-38: These last few sentences are a little clunky and I had to reread them a few times to sort out what the authors meant. Please consider rewording to make the points more clear.

We’ve now reworded these lines in our abstract.

Reviewer #2 (Recommendations for the authors):

My main recommendations would be to:

– Make clearer predictions about expected patterns of affiliation and aggression, e.g. replacing the questions in the paragraph on lines 144-152 with predictions based on the functions of the different behaviors and how they relate to kinship (e.g. aggression in mixed-sex dyads should be lower among kin because it reflects mating). i think this would be useful to go beyond a simplistic notion of 'higher r = more affiliation, less aggression', which then leads to the conclusion that 'kin selection does not explain affiliation'.

During the writing of the introduction we debated about the extent to which we should advance specific predictions. For the most part, we did not feel that we could do so in a principled manner; given the well-documented complexity of sibling relationships in a variety of species, it is easy to find support for virtually any prediction one cares to make (as just one example: it would be quite easy to justify predicting that male-male, male-female, or female-female sibling dyads should be the closest type of sibling dyads, or alternatively, that any of those should engage in the most aggression; it simply depends on what assumptions one makes about the value of what they can offer one another relative to the strength of the competition they impose). Therefore, we elected to refrain from making specific predictions and instead (1) examine the interactions between our predictor variables of interest, and (2) phrase our inquiry as questions to be answered, rather than predictions to be supported/refuted. Please note that we also added some explanatory language to the introduction that we hope helps to clarify why we used this approach, as well as to address reviewer requests for additional details about the specifics of gorilla dynamics, specifically (lines 151-162):

“As noted above, however, sibling relationships are also characterized by the potential for competition as well as cooperation, and mountain gorillas are no exception. Even after co-resident mountain gorilla siblings cease competing over parental resources, they are likely to compete over other limited resources, including dominance positions, mating opportunities, and preferential foods (Harcourt and Stewart, 2007). In some other primate species, rates of aggression are as high or higher among kin than non-kin, in part because kin simply spend more time together, highlighting the complexity of making straightforward predictions about affiliative and aggressive interactions and relatedness (e.g. Silk et al., 2010). On balance, however, we expect that mountain gorillas should be more cooperative/affiliative with siblings than non-siblings due to the substantial potential for inclusive fitness benefits.”

As a final note, there was disagreement among the reviewers about the approach we took to this problem. Reviewer 3 notes that “A second major strength is the opportunity this dataset and study system provides to test predictions about proposed mechanisms for kin recognition in primates. The authors do a good job of making these details about their study system and their predictions clear.” Therefore, we hope that in the revision we have managed to strike a balance between not making predictions we do not feel are necessarily theoretically justified, while making it clear what questions we are proposing to answer and why.

– Reconsider the operationalization of familiarity, especially for maternal kin. I just find it hard to believe statements like “sibling and non-sibling age-mates in the same social group would typically be expected to possess close familiarity” (lines 236-237). Surely, the mother is a main focal point for all young gorillas, and they w”uld get more exposure to their maternal siblings than to others simply by virtue of sharing the mother’s proximity?

We’re happy to address this comment. Mothers undoubtedly mediate social interactions to some extent, which was our rationale for including the intensity of her presence, and its interaction with the relatedness of the target dyad, as covariates. In the public review, the reviewer expresses skepticism that this “fully accounts for the ways mothers structure one’s social network, increasing exposure among maternal sibs throughout one’s lifetime”. We agree that there may well be forms of early exposure that we’re unable to assess, such as the sharing of night nests or preferential mother-father relationships—both of which we noted on p. 18 of the original manuscript.

At the same time, we stand by our argument that mountain gorilla groups are remarkably close-knit, which should lead to close familiarity among siblings and non-siblings alike. It has been observed for decades that “[Mountain] gorilla groups are quite cohesive in that the members rarely drift far from each other. The diameter of feeding or resting groups is usually 200 feet or less” (Schaller and Emlen, 1963; p. 372). Often, the focal point for social interactions is an adult male (whether the father or not), rather than mothers (Stewart, 2001; Rosenbaum and Silk, 2022). While it is reasonable to believe that maternal siblings may receive somewhat more exposure to one another via common proximity to their mother, even when we attempt to adjust for that variable, we (a) acknowledge this possibility and (b) have good reason to believe that mountain gorillas routinely spend large amounts of time very close to, and interacting with, both related and unrelated age-mates—especially compared to other great apes like chimpanzees.

That said, we take the point that we should qualify the claim the reviewer challenges. We have now edited this paragraph in the discussion to explain that “First, gorilla social groups are tight-knit and cohesive compared to their close ape relatives (Goodall, 1986; Remis, 1997; Morrison et al., 2021b; Schaller and Emlen, 1963). Even within this tight-knit context, however, ubiquitous ‘play groups’ of infants and juveniles frequently form around adult males (Rosenbaum and Silk, 2022), providing similar-aged animals even more opportunities to interact and become familiar with one another.” We also have added to our discussion of sharing night nests to specify that this is a manner of maternally mediated exposure we could not measure.

Stewart, KJ (2001). Social relationships of immature gorillas and silverbacks. In Robbins MM, Sicotte P, Stewart KJ, editors. Mountain gorillas: three decades of research at Karisoke. Cambridge, UK: Cambridge University Press, pp. 184-213.

Schaller, GB, and Emlen, JT. (1963). Observations on the ecology and social behavior of the mountain gorilla. In African ecology and human evolution, Routledge, pp. 368-384.

As mentioned before, the age difference measure is also somewhat problematic because there cannot be an age difference of 0 (or even <2-3y) for maternal sibs (I assume there weren’t many twins), but there can be for paternal sibs; in fact, that’s when paternal sibship is most likely.

Here, we take the reviewer to be arguing that, because certain age differences (namely, those less than two years) do not exist among full and maternal siblings, our age difference variable is likely to lead to biased estimates. It’s of course true that interbirth intervals preclude certain age differences in certain relatedness categories, but it does not follow that this makes the variable problematic for estimation. Our models estimate the effect of relatedness across the full range of age differences in our dataset, not just from 0 – 2 years, and we find that in several instances this wide range provides important moderating context to our pairwise differences across relatedness and/or sex categories. (In the figures, we depict estimates between 0 and 15 years age difference, because relatively few observations after 15 years leads to very wide confidence intervals.) We do appreciate the reviewer bringing this point up, however, in part because it alerted us to an oversight in our age difference figures—the trend lines should not extrapolate to “impossible” values for full and maternal siblings. This has now been fixed.

Reviewer #3 (Recommendations for the authors):

In a revised version I suggest the authors provide more detail about the nature and function of social interactions (including cooperative ones) in mountain gorillas and how they change during development.

The first part of this suggestion is something we now address and that was also requested by other reviewers; please see our Responses #5 and #11, in which we paste the exact changes made. Regarding changes in social relationships in mountain gorillas over the course of development, this is a topic on which there is remarkably little information. In fact, the dearth of data on the ontogeny of social relationships was originally one of the motivations behind this analysis. There are some data on the ontogeny of relationships between adults and immatures, particularly parents and offspring (Fletcher, 2001; Rosenbaum et al., 2011; Rosenbaum et al., 2016), but there is very little outside of that specific context/dyad type. This is a hole in the literature that this paper is helping to fill.

Fletcher, A. W. (2001). Development of infant independence from the mother in wild mountain gorillas. Mountain gorillas: three decades of research at Karisoke.

Rosenbaum, S., Hirwa, J. P., Silk, J. B., and Stoinski, T. S. (2016). Relationships between adult male and maturing mountain gorillas (Gorilla beringei beringei) persist across developmental stages and social upheaval. Ethology, 122(2), 134-150.

Rosenbaum, S., Silk, J. B., and Stoinski, T. S. (2011). Male–immature relationships in multi‐male groups of mountain gorillas (Gorilla beringei beringei). American Journal of Primatology, 73(4), 356-365.

In explanations for why maternal but not paternal kin bias occurs in mountain gorillas, one important measure to consider is that a mother may be more permissive of one of her own elder offspring vs. other similarly-aged group members, interacting with her dependent offspring (which might influence bonding between gorillas and their younger group mates – even when mothers are not in close proximity).

This is not a hypothesis that requires testing here, but I think it could be helpful to speak to this – if relevant to the behavior of the study species – in the Discussion. (How often are female gorillas aggressive when others play roughly with their infants or juveniles, for example? If their adult daughter vs. another adult female carried their infant away to play with them would the mother’s behavior be different?)

This is an interesting point. In other primate species (e.g. baboons, chimpanzees), mothers are often quite intolerant of unrelated individuals handling, touching, or getting close to their infants. In mountain gorillas, our personal observation from years of experience in the field is that this is much less true. While we are not aware of any papers on this subject, unrelated individuals frequently interact even with young infants, and mothers usually only interfere if their baby shows signs of distress (e.g. whimpering or screaming), which is rare. In those cases, the offending individual typically stops what they are doing and moves away from the baby quickly, acquiescing to the mother’s authority. There is no real equivalent of the ‘kidnapping’ behavior that is common in baboons, and which likely motivates maternal suspicion of unrelated animals handling babies. While unfortunately we do not have empirical evidence to support this (yet–it would certainly be interesting to look at!), our sense is that mothers do not have strongly different reactions to relatives and non-relatives handling/interacting with their offspring. One of us (SR) also studies baboons, and she confirms that there is a very strong qualitative difference in the reaction of baboon mothers and mountain gorilla mothers to ‘outsiders’ interacting with infants.

Last, I understand that the title is correct with respect to core kin selection theory, but I personally found it slightly misleading, because the results do find a kin bias, simply a maternal-kin bias and familiarity (though clearly defined as a predictor within the paper) has a vague broader meaning for the title. It refers particularly to familiarity within maternal siblings of varying age (age difference), but it is easily interpreted to mean familiarity on a broad level (e.g. any individuals in the same natal group – which could not be included in models because it was too correlated with other variables). As a suggestion, it might be clearer to make the title something like “Mountain gorillas demonstrate affiliative biases toward maternal but not paternal siblings of all ages.”

In response to this comment and those of another reviewer and the editor (see our Response #1), we’ve changed the title of our manuscript to eliminate reference to familiarity. We still believe age difference to be an important measure of familiarity within mountain gorilla groups, but we don’t wish to mislead readers before they understand what we mean by the term. Our title is now “Mountain gorillas maintain strong maternal affiliative biases despite high male reproductive skew and extensive exposure to paternal kin”.

[Editors’ note: further revisions were suggested prior to acceptance, as described below.]

Reviewer #2 (Recommendations for the authors):

I appreciate all the changes the authors made to the manuscript and find it much improved. The analyses and results are compelling as they are, and my comments are merely intended to further clarify the “packaging”. Sorry that the comments turned out to be rather long, it’s really not a huge issue but just took some time to explain.

Main comment: From kin selection to kin recognition to expressed kin preferences – I can’t help but feel like the argument sometimes comes across as a bit backward, e.g. in the abstract where a preference for maternal kin (over paternal kin) occurs ‘despite high relatedness certainty’, whereas normally low skew underlies such ‘biases’ (lines 34-38). This sounds like gorillas lack something, or got something “wrong”, though really it’s the case that most other primates lack something (high skew) and therefore can’t possibly recognize paternal kin, and are limited to maternal kin preferences – hence the bias is borne out of necessity, not because it is somehow preferable. To me at least, the flow of the argument generally goes more like this: From a theoretical perspective, we would predict individuals to preferentially cooperate with kin due to inclusive fitness benefits, which should select for kin recognition whenever possible, though in practice maternal kin recognition is much easier than paternal kin recognition (because one can use simple cues like associating with the same adult female, while paternity uncertainty in polygamous mating systems obviates the use of simple cues). Paternal kin recognition only becomes feasible (in the absence of some kind of phenotypic matching) when one can use similarly simple cues like associating with the same adult male, which becomes more reliable the fewer adult males there are and the higher the reproductive skew. So gorillas have all the ingredients for paternal kin recognition (at least historically), which sets up the expectation in the introduction of paternal kin preferences (over non-kin). Yet the results show that they don’t seem to express these preferences (with the discussion focusing on why they don’t express these preferences, e.g. mismatches in group structure derailing paternal kinship cues, or paternal kin not typically offering cooperative benefits). I know the authors are awar’ of all this (and in some places the argument is very clear), and I'm sorry for rambling, but I just feel like the argument could be laid out a bit more clearly in some places. Take it or leave it!

We appreciate this thoughtful comment, and we agree that the difference in opinion here mostly comes down to framing. Regarding the term “bias”—we simply use it to mean “disproportionate weight in favor of something”, without intending to attach any sort of valence to it. In terms of gorillas ‘lacking’ something (which the reviewer mentions here and in a later comment), we once again did not intend this to highlight a deficiency—we were just attempting to draw a contrast between the social features of better-studied primate species, and mountain gorillas, who do not possess those same features.

However, we understand that the inadvertent side effect might have been to imply a deficiency. To remove any suggestion of this, we edited our discussion (lines 352-354) to simply state “Interestingly, this “asymmetric bias” in affiliation seems to persist even though the high reproductive skew of mountain gorillas opposes the dynamic thought to underlie such biases in other primate species.”

We elected to keep our language largely the same in the abstract (minus some small changes that we detail in responses below), for the reason we outline above: the established idea is that low reproductive skew leads to maternal kin biases only, and it is this idea that our results challenge, as we see (mostly) similar biases in a primate with high reproductive skew. The reviewer might ask, “why not flip the comparison, and say mountain gorillas are expected to have kin discrimination abilities that other primates lack?” Our response is that we don’t necessarily think that’s the case. To detail our thinking, we’ve added a paragraph to the introduction (line 180-196) explaining how one could make an argument that high skew makes paternal kin discrimination achievable, or that it makes it unnecessary. Our results are generally more consistent with the latter, though we also find some evidence for the former (regarding aggression). Going in, we weren’t confident which outcome was more likely. For us, the key takeaway is that our results in many ways resemble those from e.g. baboons, even though gorilla society is quite different from baboon society, and that’s something that calls for revisiting prominent theoretical accounts of kin selection.

Reviewer #3 (Recommendations for the authors):

This revision is much improved from the previous version. I appreciated all the changes the authors have made in responses to all editor and reviewer comments.

I still have several concerns regarding age changes/developmental perspectives in this study:

First I think the manuscript is strengthened by the supplemental analyses that incorporate the age of individuals in each dyad for each year.

However the authors still need to explicitly inform readers that their study includes both immature and mature individuals and provide context for how siblings relationships may change during age that are rooted in gorilla socioecology.

I appreciate that this is understudied:

Here is an additional resource that provides descriptive details on how social interactions between gorillas and their mothers compared to other group members manifest during different developmental stages.

Watts, David P., and Anne E. Pusey. “Behavior of juvenile and adolescent great apes.” Juvenile primates: Life history, development, and behavior (2002): 148-167

I also realize that documenting behavioral shifts in development are not the focus of this study.

But not providing basic information about age ranges and what we know and do not know about how relationships change with age limits and even misguides how readers unfamiliar with the study system will interpret findings. For example they may even assume you studied only adult sibling pairs as is true in many of the key kin selection studies cited in the introduction.

In addition, it is not only that behaviors change with age (which the authors address well in this revision) but that dynamics of dyadic relationships do.

For example, as authors touch upon in their introduction, in baboons and chimpanzees sister-brother sibling relationships can be critical to survival and/or improve fitness while one sibling is immature, but not in adulthood based on dispersal and/or species-typical dominance/cooperation patterns (e.g. Engh et al., 2009; Hobaiter et al., 2014).

We appreciate this critique, and we certainly wish to be responsive to it.

https://doi.org/10.7554/eLife.80820.sa2

Article and author information

Author details

  1. Nicholas M Grebe

    Department of Anthropology, University of Michigan-Ann Arbor, Ann Arbor, United States
    Contribution
    Conceptualization, Data curation, Formal analysis, Investigation, Visualization, Methodology, Writing - original draft, Writing - review and editing
    For correspondence
    ngrebe@umich.edu
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-1411-065X
  2. Jean Paul Hirwa

    The Dian Fossey Gorilla Fund International, Atlanta, United States
    Contribution
    Conceptualization, Data curation, Formal analysis, Investigation, Writing - review and editing
    Competing interests
    No competing interests declared
  3. Tara S Stoinski

    The Dian Fossey Gorilla Fund International, Atlanta, United States
    Contribution
    Resources, Supervision, Funding acquisition, Project administration, Writing - review and editing
    Competing interests
    No competing interests declared
  4. Linda Vigilant

    Department of Primatology, Max Planck Institute for Evolutionary Anthropology, Leipzig, Germany
    Contribution
    Funding acquisition, Investigation, Methodology, Writing - review and editing
    Competing interests
    No competing interests declared
  5. Stacy Rosenbaum

    Department of Anthropology, University of Michigan-Ann Arbor, Ann Arbor, United States
    Contribution
    Conceptualization, Data curation, Formal analysis, Supervision, Funding acquisition, Investigation, Visualization, Methodology, Writing - original draft, Writing - review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-7576-5302

Funding

National Science Foundation (1122321)

  • Stacy Rosenbaum

Dian Fossey Gorilla Fund

  • Nicholas M Grebe
  • Jean Paul Hirwa
  • Tara S Stoinski
  • Linda Vigilant
  • Stacy Rosenbaum

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

The authors are grateful to Winnie Eckardt and Robin Morrison for their expertise and assistance with data. We thank the Rwandan government and the Rwanda Development Board for their long-term support of the research, monitoring, and protection activities of The Dian Fossey Gorilla Fund’s Karisoke Research Center. We are deeply indebted to all Karisoke field staff for their tireless support in collecting long-term behavioral and demographic data. This study was supported by the University of Michigan, the NSF Graduate Research Fellowship Program and Doctoral Dissertation Improvement Grant No 1122321, and the donors who support The Dian Fossey Gorilla Fund.

Senior Editor

  1. George H Perry, Pennsylvania State University, United States

Reviewing Editor

  1. Ammie K Kalan, University of Victoria, Canada

Publication history

  1. Received: June 6, 2022
  2. Preprint posted: June 16, 2022 (view preprint)
  3. Accepted: September 21, 2022
  4. Accepted Manuscript published: September 22, 2022 (version 1)
  5. Version of Record published: October 3, 2022 (version 2)
  6. Version of Record updated: May 18, 2023 (version 3)

Copyright

© 2022, Grebe et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 693
    Page views
  • 123
    Downloads
  • 1
    Citations

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Open citations (links to open the citations from this article in various online reference manager services)

Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)

  1. Nicholas M Grebe
  2. Jean Paul Hirwa
  3. Tara S Stoinski
  4. Linda Vigilant
  5. Stacy Rosenbaum
(2022)
Mountain gorillas maintain strong affiliative biases for maternal siblings despite high male reproductive skew and extensive exposure to paternal kin
eLife 11:e80820.
https://doi.org/10.7554/eLife.80820

Further reading

    1. Ecology
    2. Evolutionary Biology
    Laure Olazcuaga, Raymonde Baltenweck ... Julien Foucaud
    Short Report

    Most phytophagous insect species exhibit a limited diet breadth and specialize on a few or a single host plant. In contrast, some species display a remarkably large diet breadth, with host plants spanning several families and many species. It is unclear, however, whether this phylogenetic generalism is supported by a generic metabolic use of common host chemical compounds (‘metabolic generalism’) or alternatively by distinct uses of diet-specific compounds (‘multi-host metabolic specialism’)? Here, we simultaneously investigated the metabolomes of fruit diets and of individuals of a generalist phytophagous species, Drosophila suzukii, that developed on them. The direct comparison of metabolomes of diets and consumers enabled us to disentangle the metabolic fate of common and rarer dietary compounds. We showed that the consumption of biochemically dissimilar diets resulted in a canalized, generic response from generalist individuals, consistent with the metabolic generalism hypothesis. We also showed that many diet-specific metabolites, such as those related to the particular color, odor, or taste of diets, were not metabolized, and rather accumulated in consumer individuals, even when probably detrimental to fitness. As a result, while individuals were mostly similar across diets, the detection of their particular diet was straightforward. Our study thus supports the view that dietary generalism may emerge from a passive, opportunistic use of various resources, contrary to more widespread views of an active role of adaptation in this process. Such a passive stance towards dietary chemicals, probably costly in the short term, might favor the later evolution of new diet specializations.