Behavioral Data Collection

A team of two to three observers conducted behavioral observations throughout the breeding season for each introduction. All observers achieved a reliability score of 100% agreement in animal identification and 95% agreement on all behaviors, which were calculated using Krippendorff’s alpha (Krippendorff, 2011). Observations occurred from 8am – 3pm Monday through Thursday during the breeding season, barring days when the introduction was paused (e.g., inclement weather, or injury to a high-ranked male that required removal for treatment). During visual/ protected contact, observers noted the males’ behavior (i.e., interactions with other males and with females through the caging); data collection on female-female interactions within the compound was not conducted during this phase. Once males were released into the female compound, behavioral interactions were recorded among all group members 3 years of age and older.

Using an event sampling design, one observer recorded aggressive (e.g., threats, chases) interactions and a second observer recorded status signaling (e.g., displacements, silent-bared-teeth signals) interactions. Observers recorded the participants’ identity and their behaviors/ responses. Complex, polyadic events such as interventions by a third party or redirected aggression were recorded as an ordered series of pairwise interactions connected via a common label in the data. These observation methods have been successfully applied in past work on social stability in large captive groups of rhesus macaques (Beisner et al., 2011; McCowan et al., 2011).

Affiliative interactions, including social grooming, huddling/ social contact, and proximity (sitting within arm’s reach), were recorded using scan sampling. During observation hours, an affiliation scan was performed every 20 minutes.