Prehistoric agriculture and its relation to environment, population size, and social complexity is of longstanding interest in archaeology and allied disciplines. The research reported here takes up this issue in a dry land agricultural context from the northwestern portion of the island of Hawaii where there is both a rainfall elevation gradient and two major substrates of differing ages. A large dry land agricultural system (ca. 60 km² in size) was developed here between 500-175 yrs bp. Our research explores several related questions: 1. How did variation in rainfall and soil quality affect the expansion and intensification of dry land agriculture? 2. How did agricultural productivity vary in time and space and what were its effects on population dynamics? And 3. How was surplus agricultural production distributed over this area and what mechanisms were employed to redistribute it to others and/or to convert it, e.g., to livestock production.

Our methods employed a combination of archaeological and paleo-environmental fieldwork, archival documentation, and modeling to develop and test hypotheses related to these questions. Soil biochemistry analysis and dates for the construction of fixed dry land agricultural fields developed evidence for Question 1. Modeling potential agricultural productivity (based on soils, rainfall, temperature) was employed along with archival and archaeologically based chronologies for the development of more than 20 separate communities to answer Question 2. Agricultural surplus was also estimated from this modeling and we are now generating archaeological and ecological information for Question 3.

Results/Conclusions

Initial agricultural expansion was focused on localities that are optimal in terms of key parameters (rain and elevation), and that it later expanded to locations with greater variability in rainfall and at higher elevations. Intensification of dry land farms, measured by in-filling of the original fixed fields by smaller subdivisions, is distributed throughout the field system but occurred earlier in more optimal locations and was less important in environmentally more marginal areas. Agricultural production was initially evenly distributed, but as population increased, there were greater disparities across the area, with the more optimal locations maintaining or increasing their productivity relative to less optimal communities. This created substantial differences in agricultural production, and hence surplus. Control mechanisms lodged in Hawaiian ritual and political organization re-distributed this surplus among communities as insurance but to political elites to support warriors so as to be competitive with other political units in the islands.