Food sharing is linked to urinary oxytocin levels and bonding in related and unrelated wild chimpanzees

Humans excel in cooperative exchanges between unrelated individuals. Although this trait is fundamental to the success of our species, its evolution and mechanisms are poorly understood. Other social mammals also build long-term cooperative relationships between non-kin, and recent evidence shows that oxytocin, a hormone involved in parent-offspring bonding, is likely to facilitate non-kin as well as kin bonds. In a population of wild chimpanzees, we measured urinary oxytocin levels following a rare cooperative event-food sharing. Subjects showed higher urinary oxytocin levels after single food-sharing events compared with other types of social feeding, irrespective of previous social bond levels. Also, urinary oxytocin levels following food sharing were higher than following grooming, another cooperative behaviour. Therefore, food sharing in chimpanzees may play a key role in social bonding under the influence of oxytocin. We propose that food-sharing events co-opt neurobiological mechanisms evolved to support mother-infant bonding during lactation bouts, and may act as facilitators of bonding and cooperation between unrelated individuals via the oxytocinergic system across social mammals. © 2014 The Author(s) Published by the Royal Society. All rights reserved.

We analysed 79 urine samples from 26 chimpanzees (females: 10 adults, i.e. more than 15 years of age; three subadults, i.e. between 10 and 15 years of age; males: six adults, seven subadults; mean sample per chimpanzee ± s.d. = 3.0 ± 1.97) of the Sonso community in Budongo Forest, Uganda [59], between January 2009 and July 2010. We collected urine samples if subjects had engaged in no affiliative behaviours (e.g. grooming or copulation) other than the target behaviour in the hour prior to urination. This was determined through focal sampling [60] of subjects or all-occurrence sampling [60] of chimpanzee subgroups (or ‘parties’). The Composite Relationship Index (CRI) and dominance hierarchies were also determined from focal or all-occurrence sampling conducted between October 2009 and July 2010 by C.C., R.M.W. and seven experienced field assistants. Feeding time, food source and party composition were recorded in 15 min scan samples [60]. Faecal samples for genetic analysis of kinship were collected throughout the study period. Food sharing occurred when one individual was allowed access to food in possession of another, in the absence of aggression [47] (see electronic supplementary material, text S1). This could happen in one of two ways. Food was passively shared, such that the possessor allowed another to take the food, or to take over access to the food supply, in the absence of overt coercion in the form of aggression or screaming (see electronic supplementary material, video S1). Alternatively, food was actively shared, such that the possessor extended the food towards the receiver and released it in the absence of aggression (see electronic supplementary material, video S2). Food-sharing events could thus be single momentary events, multiple momentary events or protracted events (see electronic supplementary material, table S1). Food-sharing events could occur with begging behaviour. Begging definitions were taken from Gilby [55], with our additions shown in brackets. Begging was either (i) sitting and staring at the food item (or possessor), (ii) reaching towards but not touching the food item or possessor (with or without whimpering), (iii) touching the food item or possessor, or (iv) placing a hand directly over the possessor’s mouth. Begging behaviours (iii) and (iv) were considered to be low and high harassment, respectively [47,55,61]. We assessed the quality of relationships by calculating all-occurrence rates of the following behaviours over the current and preceding annual quarters: coalitionary support, food sharing, grooming, staying in (less than 1 m) proximity and aggression [62,63]. For all behaviours, each occurrence was recorded as a single event. From the resulting rates, we calculated the CRI, a measure of social bond strength [15,64]. The CRI is calculated over a period of three months and gives socio-positive (given or received food sharing, coalitionary support, allo-grooming and resting in less than 1 m proximity) and socio-negative (aggression given or received) behaviours equal weight: where SP1 = rate of grooming bouts plus rate of resting in 1 m proximity, SP2 = rate of food sharing plus rate of coalitionary support, NP = rate of aggression, i = individual and j = dyad partner. The index is positive when each individual within a dyad initiates on average more socio-positive than socio-negative interactions. ‘Bond partners’ were defined as dyads having a net socio-positive relationship lasting more than or equal to six months (at least two consecutive blocks of three months). This can occur through either a mutual socio-positive relationship (CRI > 0) during the annual quarter of the experiment and the preceding quarter, or a large mutual socio-positive relationship (CRI > 10) during one of the quarters and a socio-neutral or positive relationship (CRI ≥ 0) during the other quarter. According to this, 1.9% of kin dyads and 1.6% of non-kin dyads reached bond-partner status. Our target behaviours were single food-sharing events or 1 h of feeding in the presence of chimpanzees without sharing food. We collected urine samples as described by Crockford et al. [15]. Specifically, urine was collected 15–60 min after the target behaviour (time window of urinary clearance of oxytocin for primates [65]). Occasionally, subjects were sampled after engaging in more than one food-sharing event within the required time window. In both conditions, samples were not collected if grooming or copulation also occurred within 60 min prior to urination, as both of these behaviours are likely to independently increase urinary oxytocin levels [39]. A volume of 1.1 ml of the collected urine was pipetted into a cryovial containing 100 µl of 0.5 N phosphoric acid and stored on ice in a thermo flask. Solid-phase extraction was conducted later the same day [40]. All samples were then frozen until transported for assaying in the Assay Services Unit at the NPRC, Madison, WI, using an enzyme immunoassay kit (Assay Designs, Ann Arbor, MI; catalogue no. 901-153). To compensate for variation in urine concentration, we measured creatinine (crea) levels in each sample [66] and expressed all oxytocin values as pg mg−1 crea. We validated the measurement of urinary chimpanzee oxytocin levels through parallelism and accuracy tests, as described in a previous paper [15]. We collected pant-grunt vocalizations as a unidirectional indicator of dominance relationships, given by the subordinate to the dominant [53]. We calculated a linear dominance hierarchy for Sonso chimpanzees on the basis of the pant-grunts using Mat Man v. 1.1 (see electronic supplementary material, text S2). Donors and receivers of shared food were assigned a relative dominance relationship according to the hierarchy matrix. We collected fresh faecal samples, stored and extracted DNA following protocol of [15]. Dyads were classified as kin (n = 11) or non-kin (n = 10) according to a combination of (i) parentage analyses based on autosomal microsatellites and (ii) mitochondrial DNA and Y-chromosome microsatellite haplotype sharing information [13,67]. We were able to show that that all kin partners were either mother–offspring (n = 10) or maternal siblings (n = 1), and none of the non-kin partners were such close maternal relatives (see electronic supplementary material, text S3, S4 and table S5). Urinary oxytocin (OT) concentrations were log10-transformed to fit a normal distribution. Five generalized linear mixed models (GLMM) were run with maximum-likelihood estimates in SPSS v. 20, testing the effect of the predictor variables shown in table 1 on the response variable of log10-transformed urinary oxytocin levels. In model 1, we compared food-sharing events with social feeding without food sharing. We tested the effect of whether food is shared or not and its monopolizability on the urinary oxytocin concentration after controlling for subjects’ sex and age. Subjects’ identity was included as a random factor. Model 2 was divided into two separate tests due to small sample size. Model 2a investigated the variation within the food-sharing samples with regard to the sharers’ relationship. We tested the effect of close kinship and bond quality on the urinary oxytocin concentration after controlling for whether the subject received the food. Subjects’, partners’ (interaction partner) and dyads’ identity were included as random factors. Model 2b examined the variation within the food-sharing samples with regard to the possible function of meat sharing. We tested the effect of whether the shared food was meat, and the sex combination of the sharers, on the urinary oxytocin concentration. Subjects’, partners’ and dyads’ identity were included as random factors. Finally, in model 3, we compared urinary oxytocin levels after food sharing with those after another cooperative behaviour, grooming (taken from [15]). In model 3a, we tested the effect of food sharing compared with grooming on the urinary oxytocin concentrations after controlling for subjects’ sex. Subjects’ identity, partners’ identity and dyads’ identity were all included as random factors. In model 3b, we tested the effect of five different behavioural contexts (food sharing with bond partner, food sharing with non-bond partner, grooming with a bond partner, grooming with a non-bond partner and control situations) on the urinary oxytocin concentrations, while controlling for subjects’ sex. Subjects’ identity was included as a random factor. Predictor variables tested in the GLMMs. Model 1 refers to the GLMM shown in table 3, which tests sharing and non-sharing samples together (n = 79). Model 2 refers to the GLMM presented in table 4, using only food-sharing samples (n = 33). Model 3 refers to GLMM presented in table 5, contrasting food-sharing and grooming samples (n = 182). aIn a control run of model 2a the continuous variable Composite Relationship Index (CRI) replaced categorical variable ‘bond type’ in model 2a (see Material and methods). Five outliers (more than 2 s.d.) were excluded from the food-sharing dataset (three non-sharing and two sharing samples) to be sure that they were not driving any main effects. Their distribution relative to the main dataset is shown in the electronic supplementary material (figure S1). We excluded 13 outliers (more than 2 s.d.) from the grooming dataset for the comparison between grooming and food sharing. Nonetheless, when we ran the GLMMs with the full dataset, including the outliers, the results remained remarkably similar to the GLMM results excluding the outliers (see electronic supplementary material, tables S2–S4). Variables did not exhibit problems of collinearity [68] (Kendall’s τ and Spearman’s r < 0.7 in all cases). As a check of the overall significance of all predictor variables, we ran likelihood ratio tests comparing the full model with the respective null model (comprising only the random effects). We only considered significant effects of the individual predictors if the full model explained the variance significantly better than the null model. As models 2a and 2b investigated many predictor variables relative to the number of cases (n = 33, d.f. = 3 or d.f. = 5), reduced power may have led to false negatives (i.e. erroneously non-significant effects), as well as some risk of instability in the derived estimates. Hence, we ran an additional set of univariate GLMMs with each including only one of the predictor variables at a time (and the same random effects as in the full model). None of the predictor variables tested in the univariate models reached significance (bond type: F1,33 = 0.074, p = 0.787; CRI: F1,33 = 0.277, p = 0.602; kin relationship: F1,33 = 0.020, p = 0.888; sharing direction: F1,33 = 0.008, p = 0.931; food category: F1,33 = 1.289, p = 0.264; sex combination: F2,33 = 1.200, p = 0.314), which showed that the lack of significance in models 2a and 2b was unlikely to be due to power issues.

N/A

• wittig, roman m.
• crockford, catherine
• deschner, tobias
• langergraber, kevin e.
• ziegler, t. e.
• zuberbühler, klaus