Warren, A. L.,Bhat, U.,Sattenspiel, L.,Swedlund, A. C.,Gumerman, G. J.
Summary: Many fundamental questions about the prehistory of the American Southwest require accurate estimates of population size and other demographic parameters in order to be investigated properly. The Artificial Anasazi model and its successor, Artificial Long House Valley, are agent-based models that use archaeological and environmental data as well as empirically derived demographic estimates to study population dynamics in Long House Valley, Arizona from AD 800 to AD 1300. The Artificial Anasazi model focuses on household-level processes and uses constant demographic rates while the Artificial Long House Valley model focuses on individual-level processes and uses age-specific rates. In the present study, versions of the models described above have been adjusted to remove underlying environmental variation and other confounding variables, which allows in-depth analyses of the consequences of basing population size estimates on constant demographic rates, especially when age-specific rates are available. The original models used what appeared to be mathematically equivalent estimates of total fertility, but the simplified models reveal significant differences in population size estimates for the two fertility schemes (359.67 average individuals (constant fertility) versus 189.34 average individuals (age-specific fertility), over 1000 runs.) Higher fertility in the youngest age group when using a constant fertility rate resulted in increasing disparities in population size estimates over time. Although using constant demographic rates is more tractable when describing population trends for large areas over long spans of time, results such as those presented here highlight the sensitivity of population estimates to assumptions about the scale at which demographic processes operate.