An introduction to generative network models and their use in animal sociality research

Josefine Bohr Brask, Matthew Silk , Michael N. Weiss

Social networks constitute an important approach in the study of animal social behaviour. So far, the main focus has been on statistical analysis of animal social network structures. However, social networks can also be studied by generative network models - procedures that create simulated network structures. These models play a key role in wider network science, but while they have been used in the animal behaviour field, they have not yet been as well integrated as other approaches. We believe that generative network models have considerable unexploited potential as a tool for understanding animal social systems. Here: 1) we provide a general introduction to generative network models, including a description of questions they are used for investigating in wider network science, explanation of key model features, and an overview of common models; 2) we consider generative network models in relation to the study of animal social behaviour, including description of questions about animal systems they can be used to investigate (demonstrated by case studies), an overview of animal behaviour studies that have used generative network modelling, the relevance of the key model features for animal behaviour studies, and consideration of how to choose a suitable generative network model for studies of animal social systems. We hope that this can help to further integrate generative network models into the study of animal sociality.