1Department of Anthropology, Indiana University, Bloomington, IN, 2Department of Anthropology, Central Michigan University, MI, 3Department of Bioengineering, Rice University, Houston, TX
Thursday All day, Clinch Concourse
We use the agent-based, spatial simulation model HOMINIDS to evaluate hypotheses relating the foraging behavior of Ardipithecus ramidus to habitat structure. The hypothesized forest-grassland mosaic paleohabitat of Ardipithecus shares many characteristics with the current habitat of the Mugiri chimpanzee (Pan troglodytes) community, a relatively dry, heterogeneous habitat in western Uganda within the Toro-Semliki Wildlife Reserve. Similarities in brain size, morphology, and shared phylogeny of chimpanzees and Ardipithecus make it possible to test our model’s accuracy against real life data and can be used to develop hypotheses for Ardipithecus behavior.
We use botanical records from Semliki to develop a virtual landscape suggested for Ardipithecus paleohabitat. Behavior of different agents in the simulations is determined by a synthesis of known dietary and locomotor attributes of Mugiri chimpanzees and published interpretations of Ardipithecus fossils. We first evaluate the utility of our model for simulating actual patterns of Mugiri chimpanzee food selection and ranging behavior in a virtual 80km2 landscape, which approximates the estimated Mugiri community home range of ~96km2 . A negative control investigates the influence of uniform vs. realistic habitat distribution on chimpanzee-agent ranging. Varied experimental conditions simulate the effects of published differences in locomotion and feeding efficiency between chimpanzee- and Ardipithecus-agents. Our model allows us to analyze the degree to which ranging and feeding behaviors of simulated Ardipithecus and Pan agents diverge within the same landscapes, and results suggest that further observations of the Mugiri chimpanzees will be useful for developing testable hypotheses for Ardipithecus behavior.
Research was supported by the National Science Foundation grant BCS 98-15991, by the Indiana University Foundation, and by Indiana University’s Faculty Research Support Program. This material is based upon work conducted on Indiana University’s Big Red supercomputer, which is supported by the National Science Foundation under Grant No. ACI-0338618l, OCI-0451237, OCI-0535258, and OCI-0504075, and supported in part by Shared University Research grants from IBM, Inc. to Indiana University, and supported in part by the Indiana METACyt Initiative. The Indiana METACyt Initiative of Indiana University is supported in part by Lilly Endowment, Inc.