The manuscript reported on a study of friendships among captive chimpanzees and found that personality traits (e.g. sociability, boldness) affected the quality of friendship. The study investigated theoretically interesting, yet not fully explored questions. The study potentially contributes to the understanding of human friendship and its evolutionary roots. The data collection and statistical analysis seem reasonable. However, I would suggest some clarifications, specifically on important definitions, before it’s to be published.

1. Definition of friendship In literatures of friendship (both human and non-human animals), the definition of friendship varied across different studies. For instance, some scholars considered friendship among non-human primates as “close enduring social bonds that are not directly related to mating” (Silk, 2005) and focused on close social bonds among same sex individuals (Newton-Fisher, 2002; Langergraber, et al., 2007). Other scholars did not rule out mates from the category of friendship (Barrett & Henzi, 2002). Furthermore, in literatures of human friendship, the boundary between kinship and friendship was not clear either (Hruschka, 2010). Overall, there is no consistent way of defining friendship, which depended on the particular theoretical focus of the study.

For reasons above, I think it is important to provide a clear operational definition of friendship among chimpanzees upfront. Specifically, the definition should clarify whether friendship considered in the study included both kin and non-kin (which I think so), and whether it included mates. For instance, in Line 175-179, the authors specifically noted “UQ non-kin animals” as friends. While in the following paragraph (Line 181-182), the authors had maternal relatedness as a predictor of friendship quality in their LMM analysis. It might generate confusions. Moreover, the family tree (Table S1) provided information of genetic relatedness, but there was no information about whether they were mates or who were mating pairs.

Besides, the authors provided evolutionary predictions of both kin and non-kin friendship (from Line 280, in discussion). I would suggest moving the theory earlier to introduction, and providing theoretical reasons of differentiating kin friends from non-kin friends in later statistic analysis. For instance, in the Wilcoxon signed rank tests, the authors only included non-kin dyads (Line 176-177). In Line 205-208, the authors combined both kin and non-kin dyads and testing the effect of sociability on contact sitting. What were contact sitting scores of kin dyads? Was the distribution of the scores skewed? It was not quite clear why the authors included kin pairs in some analysis but not others. Furthermore, it is important to differentiate kin and non-kin friend pairs because the evolutionary roots of these two types of friendship could be different.

1. Define behavioral terms The authors might have provided detailed description of behavioral terms somewhere else, nevertheless, I think it’d be very helpful to provide a clear description of behavioral coding in this study (for both focal and scan data). For instance, did it require a specific duration of body contact to be coded as contact sitting? Was sitting in contact immediately preceding grooming coded independently? Particularly, was there a clear cut between contact sitting and sitting in proximity? If so, why? In the study, sociability was defined in terms of grooming and proximity (average # of individuals in 2m proximity) (Table 1). Friendship quality was measured in terms of contact sitting. To examine the effect of sociability on friendship quality, behaviors coded for these two variables should be carefully differentiated. Otherwise, the correlation between personality and friendship would be an artifact.

2. Additional comments Line 58-59: The authors argued that, “similarity in characteristics may promote friendship in non-human animals”. However, the authors did not discuss much about alternative strategies. In literature of human friendships, some scholars suggested that individuals tend to build friendships with those who are different from them or have different resources (see review of Hruschka, 2010; Granovetter, 1982). Forming friendship is thus considered to be a strategy to expand one’s niche and enhance fitness value. If homophily in friendship was an adaptive trait, it’d be helpful to discuss why is it, what kind of advantages homophily had over hetergenity.

The study did not seem particularly consider the direction of behavior, e.g. who initiated approaching and sitting in contact. However, in literatures of human friendship, some scholar differentiated reciprocal and non-reciprocal friendship, and found different behavioral patterns between these two types of relationship (Leider, et al., 2009). In the current study, if it was always one individual approached another (without the other actively responding), would it change the measure of friendship quality or the interpretation of the result?

The study investigated friendships among captive chimpanzees. In the section of literature review, the authors discussed much about field studies. Was the sample of the current study able to represent behavioral patterns in field settings? Whether or how did the captive setting affect friendship and sociability of individual chimpanzees?