Detecting and Quantifying Networks of Biological Kinship via Exponential Family Random Graph Models

Genetic relatedness between ancient humans can help to identify close and distant connections between groups and populations, uncovering signatures of demographic histories such as identifying mating networks or long-range migration. Critical to researchers are the characteristics that connected individuals, or groups of individuals, share, and how these characteristics interact and are correlated. Here we use Exponential Random Graph models as a method to explore demographic and contextual parameters that may help to explain the significant drivers of the topology of mating networks, as well as to quantify their effects. We show through simulations that model selection and coefficient estimators facilitate the exploration of such networks, and apply the method to individuals from a collection of Avar-associated cemeteries from the Carpathian Basin dating to the 6 ^th to the 9 ^th centuries CE.