Driving factors in the colonization of oceania: Developing Island-Level statistical models to test competing hypotheses

Abstract

Migration is a key driver of human cultural and genetic evolution, with recent theoretical advances calling for work to accurately identify factors behind early colonization patterns. However, inferring prehistoric migration strategies is a controversial field of inquiry that largely relies on interpreting settlement chronologies and constructing plausible narratives around environmental factors. Model selection approaches, along with new statistical models that match the dynamic nature of colonization, offers a more rigorous framework to test competing theories. We demonstrate the utility of this approach by developing an Island-Level Model of Colonization adapted from epidemiology in a Bayesian model-selection framework. Using model selection techniques, we assess competing colonization theories of Near and Remote Oceania, showing that models of exploration angles and risk performed considerably better than models using inter-island distance, suggesting early seafarers were already adept at long-distance travel. These results are robust after artificially increasing the uncertainty around settlement times. We show how decades of thinking on colonization strategies can be brought together and assessed in one statistical framework, providing us with greater interpretive power to understand a fundamental feature of our past.