1Anthropological Institute and Museum, University of Zurich, Zurich, Switzerland, 2Institute for Theoretical Physics, University of Zurich, Zurich, Switzerland, 3Seminar for Applied Mathematcis, Swiss Federal Institute of Technology, Zurich, Switzerland
Friday 10:00-10:15, Ballroom B
Morphological and genetic data gathered from hominin fossils always come from single individuals that live(d) during a short time span in a restricted spatial area. Such data, however, are implicitly taken as evidence for entire populations and/or species, and used to infer hominin paleopopulation dynamics and evolutionary scenarios. Here we ask how long-term large-scale patterns of change at the population/species level can be inferred from observed short-term/small-scale patterns of variation at the individual level. To tackle these questions, we capitalize on existing multibody modeling paradigms from physics and devise an agent-based modeling (ABM) framework running on a high-performance computing (HPC) platform. The ABM/HPC framework permits individual-based simulation of hominin population dynamics over extended spatial and temporal scales. We compare the outcome of ABM/HPC-based simulations with expectations and results of classical population dynamics theory, and with empirical genomic and phenomic data. These comparisons show that while population-level models are useful tools to identify general trends and standardized behaviors, they lack the spatiotemporal resolution to explore and identify non-standard, but probable, scenarios of hominin dispersals and evolution. We illlustrate our observations and findings with several examples from late hominin evolution.
Funded by the HP2C initiative of the Swiss National Supercomputing Centre, and by the Swiss National Science Foundation.