Strong social bonds can make an important contribution to individual fitness, but we still have only a limited understanding of the temporal period relevant to the adjustment of social relationships. While there is growing recognition of the importance of strong bonds that persist for years, social relationships can also vary over weeks and months, suggesting that social strategies may be optimized over shorter timescales. Using biological market theory as a framework, we explore whether temporal variation in the benefits of social relationships might be sufficient to generate daily adjustments of social strategies in wild baboons. Data on grooming, one measure of social relationships, were collected from 60 chacma baboons (Papio ursinus) across two troops over a six month period. Our analyses suggest that social strategies can show diurnal variation, with subordinates preferentially grooming more dominant individuals earlier in the day compared with later in the day. These findings indicate that group-living animals may optimize certain elements of their social strategies over relatively short time periods.
A fundamental characteristic of social animals is their propensity for social interaction. Such interactions give rise to social relationships between individuals, on a continuum of interaction from rare and fleeting to frequent and lengthy. Affiliative relationships at these two extremes are often described as weak and strong social bonds, respectively. The nature and balance of social relationships that an individual maintains reflects its social strategies, the patterns of investment in social interactions that maximizes its fitness with respect to particular strategic goals [1,2]. Because strong social bonds over several years can make an important contribution to the fitness of individuals in animal groups [3–5], we might expect such individuals to invest primarily in such long-term relationships. Yet, social interactions can also change dynamically in line with biological market theory [6,7], according to which a varying supply–demand ratio of a given commodity can explain changing partner preferences over shorter periods of weeks or months [8,9]. Here, we investigate whether changes in the need for one commodity, tolerance at feeding sites, might lead to corresponding changes in social partner preferences across the day. Such diurnal patterns would suggest that social strategies may be more flexible, and involve elements that are optimized over even shorter periods, than previously appreciated.
Most of our understanding of the dynamics and benefits of social relationships comes from primates, where affiliative social interactions are largely mediated by grooming. Grooming underpins strong social bonds that can last several years , but in the short-term, it is also traded as a commodity in a biological market in return for grooming or other services. Such services can include tolerance from dominant animals that allow subordinates to share their feeding sites [10,11]. Research in our study population of wild chacma baboons (Papio ursinus) similarly indicates an exchange of grooming for tolerance at feeding sites, because grooming and co-feeding are strongly correlated between individuals, dominant animals play a central role in co-feeding , and subordinates prefer food patches containing individuals with whom they have a positive grooming balance . In such populations, where the ‘purchase’ of tolerance from dominant individuals should be most valuable earlier in the day, when most of the day's foraging activities lie ahead and the need for foraging tolerance is greatest, we might expect a corresponding diurnal pattern in the social strategies of subordinate animals.
We test this hypothesis with three predictions. First, we predict that subordinates should prefer to groom more dominant animals earlier in the day (prediction P1a). Second, when subordinate groomers are unable (or choose not) to express this preference for more dominant partners, and instead groom similarly ranked animals, we predict that they should prefer more closely related partners earlier in the day (P2a), because close kin are more likely to provide support and bestow relatively more tolerance following grooming than less closely related individuals. Dominant groomers are not expected to show such preferences (P1b, P2b). Third, because the amount of tolerance received may be proportional to grooming given, and/or there will be more competition to groom dominant partners earlier in the day, we predict that subordinates’ relative contributions to grooming sessions should be greater earlier in the day, especially with high-ranking partners (P3).
2. Material and methods
Data were collected from two habituated troops of chacma baboons, across all age–sex classes (n = 24, 36; excluding infants), in our study population at Tsaobis Leopard Park in central Namibia (22°23′ S, 15°45′ E) between June and November 2009 (for further details of the study site and population, see ). Each troop was followed daily from dawn to dusk (ca 06.00–18.00 h). We conducted 1 h long focal follows on all identifiable individuals (1579 h in total with individual mean ± s.d.: 26 ± 12), recording all grooming and dominance interactions (the latter also recorded ad libitum), including the partner's identity, the direction of interaction and, for grooming, the duration. Individuals were sampled no more than once per day, with focal sampling stratified across four time blocks (less than 09.00, 09.00–12.00, 12.00–15.00, more than 15.00 h). Our unit of analysis was the grooming session, which was defined as ending when no grooming occurred for more than 1 min. We excluded sessions that were not observed in their entirety and those involving sexually receptive (swollen) females and mothers with young infants (less than three months old), where mating behaviour and the presence of infants may complicate grooming partner patterns [14,15]. Our final sample comprised 1674 grooming sessions, of which 74% were unidirectional, from 659 focal follows.
We conducted three analyses of grooming patterns between partners: model 1 (M1) explored rank differences (to test prediction P1), model 2 (M2) investigated relatedness (P2) and model 3 (M3) assessed the symmetry of contributions (P3). For M1 and M2, separate analyses were conducted for sessions involving a subordinate groomer (P1a, P2a) and dominant groomer (P1b, P2b). Because these separate analyses included both unreciprocated and reciprocated sessions, the latter sessions (n = 437, or 26% of the full sample) were incorporated in both analyses. In M3, symmetry was a binary term, where 1 indicated that the subordinate groomed more than 50% of the session's full duration. The fixed effects comprised time (hours after sunrise), and, as main effects and in interaction with time, the absolute value of the dyadic difference in relative ranks (hereafter ‘rank difference’; M2 and M3) and dyadic relatedness (M1 and M3). For further details of all variables, see the electronic supplementary material (section S1). The random intercepts comprised focal observation nested in focal identity nested in group identity. Grooming partner identity was also included, crossed with focal identity. Focal- and group-identity-specific random slopes for time were initially incorporated into each model. However, likelihood-ratio tests of models with and without focal- and group-specific random slopes showed that they did not explain any significant variation in any of the models (either alone or in combination), and so they were removed from further analyses. All of our models assume that subordinates have access to dominant animals and close kin as grooming partners throughout the day. This assumption is supported by the fact that grooming activity and high partner availability were observed at all times of the day in both groups: see the electronic supplementary material, section S2.
We used (generalized) linear mixed-effects models, (G)LMMs: LMMs for M1 and M2, and a GLMM with a binomial error structure for M3. Each term's significance was tested in the full model with Markov chain Monte Carlo analysis with 10 000 simulations (M1 and M2), or Wald test (M3). Full models were also compared against null models, i.e. with only the intercept and random (intercept) effects included, using a likelihood-ratio test to ensure avoidance of type 1 errors (false-positives), following . We used R v. 2.13.1 , packages ‘lme4’  and ‘languageR’ .
In those sessions involving a subordinate groomer, there was a larger rank difference between partners earlier in the day (table 1 and figure 1), suggesting that subordinates preferred to groom higher-ranked animals at this time. By contrast, there was no significant time-of-day effect for those sessions involving dominant groomers (table 1 and figure 1). The full model also provided a better fit to the data than the null model for subordinate groomers (χ2 = 20.6, d.f. = 3, p < 0.001), but not dominant groomers (χ2 = 3, d.f. = 3, p = 0.39). These results support predictions P1a and P1b, respectively. However, we found only mixed support for the prediction that subordinates grooming similarly ranked partners will prefer those that are more closely related earlier in the day (P2a). The model containing this interaction gave a better fit to the data than the null model (χ2 = 15.3, d.f. = 3, p = 0.002), but the interaction term did not achieve statistical significance (table 1). There was no evidence of this pattern for dominant groomers, either from a significant interaction (table 1) or better model fit (χ2 = 2.2, d.f. = 3, p = 0.54), in support of prediction P2b. Finally, there was no evidence of subordinates contributing more to grooming sessions earlier in the day (table 2), contrary to P3. Nevertheless, regardless of time, subordinates did contribute more to those sessions involving higher-ranked animals (table 2), leading to a better model fit (χ2 = 134.4, d.f. = 4, p < 0.001).
Our findings suggest that the social strategies of baboons can show diurnal variation. The observed patterns are consistent with the changing needs of subordinates for tolerance at shared food patches over the course of the day. Specifically, subordinates have a stronger preference for grooming more dominant animals earlier in the day (P1). We found only mixed support for the prediction that, among similarly ranked partners, subordinates prefer to groom closer kin earlier in the day (P2). However, we believe that the improved model fit over the null model, in the absence of a significant interaction, might indicate that this relationship is present but nonlinear. This interpretation is consistent with the presence of a significant interaction exhibiting this pattern when rank difference is treated as a dichotomous (large/small) rather than as a continuous variable (see the electronic supplementary material, section S3). Finally, although subordinates contributed more to grooming sessions with high-ranking partners, we found no support for the prediction that the contribution is relatively higher earlier in the day (P3). This may reflect the fact that grooming can be exchanged for a variety of services, not just tolerance at feeding sites [9,20], that subordinates seek from their grooming partners later in the day. We do not currently have sufficient information to assess the timescale over which an exchange of grooming and foraging tolerance occurred, but we anticipate it would be in the order of several hours. In captive chimpanzees (Pan troglodytes), for instance, de Waal  reported that the probability of animal A sharing food with B was influenced by whether B had groomed A in the preceding 2 h.
Our findings have several implications for the understanding of social relationships in group-living species. First, they suggest that social strategies can involve elements that are optimized across the day. Such a pattern would be in addition to those elements that are optimized over longer timescales, which in this species would include strong social bonds among females that can emerge seasonally  or remain stable across years  depending on the population. Second, because of the way partner preferences change across the day, the development of long-term strong social bonds from short-term social interactions may be complex. Here, our focus has been on changes in partner preference and the balance of grooming (within dyads) across the day, but the average partner preference and total grooming time may be more likely to drive the development and maintenance of long-term strong social bonds. In conclusion, our findings emphasize the short-term flexibility of social behaviour in group-living animals, and the importance of understanding the full range of time periods over which social strategies may be optimized.
Data uploaded to Dryad: http://dx.doi.org/10.5061/dryad.n4k6p.
The Fenner School of Environment and Society, Leakey Foundation, Animal Behaviour Society, International Primatological Society, Explorer's Club (A.J.C.), NERC and ZSL (H.H.M.) provided financial support.
We thank Katherine Forsythe, Rebecca Bodenham, Will Symes, Jenie Iles, Will Birkin and Hannah Peck for assistance with data collection; and Louise Barrett, Ronald Noë and four anonymous reviewers for very helpful comments. We also thank the Ministry of Environment and Tourism, Ministry of Lands and Resettlement, Snyman and Wittreich families, and Gobabeb Research and Training Centre for research permission and affiliation.
- Received March 22, 2014.
- Accepted June 13, 2014.
- © 2014 The Author(s) Published by the Royal Society. All rights reserved.