Deep learning for automatic facial detection and recognition in Japanese macaques: illuminating social networks.

Individual identification plays a pivotal role in ecology and ethology, notably as a tool for complex social structures understanding. However, traditional identification methods often involve invasive physical tags and can prove both disruptive for animals and time-intensive for researchers. In recent years, the integration of deep learning in research has offered new methodological perspectives through the automatisation of complex tasks. Harnessing object detection and recognition technologies is increasingly used by researchers to achieve identification on video footage. This study represents a preliminary exploration into the development of a non-invasive tool for face detection and individual identification of Japanese macaques (Macaca fuscata) through deep learning. The ultimate goal of this research is, using identification done on the dataset, to automatically generate a social network representation of the studied population. The current main results are promising: (i) the creation of a Japanese macaques' face detector (Faster-RCNN model), reaching an accuracy of 82.2% and (ii) the creation of an individual recogniser for the Kojima Island macaque population (YOLOv8n model), reaching an accuracy of 83%. We also created a Kojima population social network by traditional methods, based on co-occurrences on videos. Thus, we provide a benchmark against which the automatically generated network will be assessed for reliability. These preliminary results are a testament to the potential of this approach to provide the scientific community with a tool for tracking individuals and social network studies in Japanese macaques.

Social pressure drives "conversational rules" in great apes.

In the last decade, two hypotheses, one on the evolution of animal vocal communication in general and the other on the origins of human language, have gained ground. The first hypothesis argues that the complexity of communication co-evolved with the complexity of sociality. Species forming larger groups with complex social networks have more elaborate vocal repertoires. The second hypothesis posits that the core of communication is represented not only by what can be expressed by an isolated caller, but also by the way that vocal interactions are structured, language being above all a social act. Primitive forms of conversational rules based on a vocal turn-taking principle are thought to exist in primates. To support and bring together these hypotheses, more comparative studies of socially diverse species at different levels of the primate phylogeny are needed. However, the majority of available studies focus on monkeys, primates that are distant from the human lineage. Great apes represent excellent candidates for such comparative studies because of their phylogenetic proximity to humans and their varied social lives. We propose that studying vocal turn-taking in apes could address several major gaps regarding the social relevance of vocal turn-taking and the evolutionary trajectory of this behaviour among anthropoids. Indeed, how the social structure of a species may influence the vocal interaction patterns observed among group members remains an open question. We gathered data from the literature as well as original unpublished data (where absent in the literature) on four great ape species: chimpanzees Pan troglodytes, bonobos Pan paniscus, western lowland gorillas Gorilla gorilla gorilla and Bornean orang-utans Pongo pygmaeus. We found no clear-cut relationship between classical social complexity metrics (e.g. number of group members, interaction rates) and vocal complexity parameters (e.g. repertoire size, call rates). Nevertheless, the nature of the society (i.e. group composition, diversity and valence of social bonds) and the type of vocal interaction patterns (isolated calling, call overlap, turn-taking-based vocal exchanges) do appear to be related. Isolated calling is the main vocal pattern found in the species with the smallest social networks (orang-utan), while the other species show vocal interactions that are structured according to temporal rules. A high proportion of overlapping vocalisations is found in the most competitive species (chimpanzee), while vocal turn-taking predominates in more tolerant bonobos and gorillas. Also, preferentially interacting individuals and call types used to interact are not randomly distributed. Vocal overlap ('chorusing') and vocal exchange ('conversing') appear as possible social strategies used to advertise/strengthen social bonds. Our analyses highlight that: (i) vocal turn-taking is also observed in non-human great apes, revealing universal rules for conversing that may be deeply rooted in the primate lineage; (ii) vocal interaction patterns match the species' social lifestyle; (iii) although limited to four species here, adopting a targeted comparative approach could help to identify the multiple and subtle factors underlying social and vocal complexity. We believe that vocal interaction patterns form the basis of a promising field of investigation that may ultimately improve our understanding of the socially driven evolution of communication.