UC Davis

Using a sample of 40 sources reporting milpa and mucuna-intercropped maize yields in Mesoamerica, we test Boserup’s (1965) prediction that fallow is reduced as a result of growing population density. We further examine direct and indirect effects of population density on yield. We find only mixed support for Boserupian intensification. Fallow periods decrease slightly with increasing population density in this sample, but the relationship is weak. Controlling for other covariates, fallow-unadjusted maize yields first rise then fall with population density. Fallow-adjusted maize yields peak at 390 kg/ha/yr for low population densities (8 persons / km2) and decline to around 280 kg/ha/yr for the highest population densities observed in our dataset. Fallow practices do not appear to mediate the relationship between population density and yield. The multi-level modeling methods we adopt allow for data clustering, accurate estimates of group-level variation, and they generate conditional predictions, all features essential to the comparative study of prehistoric and contemporary agricultural yields.