Predicting Social Rankings in Captive Chimpanzees (Pan troglodytes) Through Communicative Interactions-Based Data-Driven Model.

Abstract

Primates demonstrate variability regarding the degree to which they display status hierarchies, which are influenced by a multitude of demographic and ecological factors. Additionally, primates must manage their interactions within these social hierarchies through the use of diverse communicative signals, including facial signals and manual gestures. Often times, these variables are assessed independently; however, it is probable that they collectively influence social rankings among primates. Our study investigates the application and accuracy of data-driven techniques, based on the genetic algorithm approach, in capturing social rankings among a group of captive chimpanzees, focusing on the analysis of communicative and demographic factors. We utilize observational data collected from a group of 18 chimpanzees residing at the Los Angeles Zoo from 2017 to 2019, derived from three previous studies carried out by the first author (BF). Our data-driven model exhibited a high degree of accuracy in capturing established social hierarchies in 2017, in addition to identifying notable fluctuations in rankings during periods of social instability from 2018 to 2019, especially in the aftermath of the passing of the highest-ranking female in the troop. Feature importance analysis revealed that social bond strength, measured via the dyadic composite sociality index (DCSI), was the most significant predictor of rank, highlighting the importance of social bonding in shaping status hierarchies. These models provide valuable insights for future research on primate behavior and social structures, as well as assist in making informed decisions for zoo management.