Bet-hedging strategies, agricultural change, and unpredictable environments: historical development of dryland agriculture in Kona, Hawaii Melinda S. Allen * Department of Anthropology, University of Auckland, Private Bag 92019, Auckland, New Zealand Received 17 September 2003; revised 12 February 2004 Available online 16 April 2004 Abstract It is argued that current models of agricultural change are overly focused on productive increases. Risk management strategies, practices that can be critical to the long-term survival of a given agronomic system, warrant a more prominent role. Neo-Darwinian theory, and the bet-hedging model in particular, offers a way to evaluate both kinds of agronomic change within a single theoretical paradigm, as well as a means of assessing the long-term outcomes of variant agronomic strategies. The bet-hedging model is used herein to assess agricultural change in an unpredictable Hawaiian environment, the Kona District of central West Hawaii. A fine-grained record of agronomic change from one of the regionÕs more productive areas is compared with patterns from the region at large. The analysis shows that variance minimizing (or risk management) strategies were initiated by 1450 AD, if not earlier, and occurred in even the most productive localities. Innovations in gardening architecture, novel cultivation practices and changes in the scale of agronomic integration, are argued to have been effective in reducing the impact of environmental variance. Later in time (after 1650 AD) there was a shift in emphasis to productive maximizing strategies, with implications for the regionÕs economic and socio-political stability. Ó 2004 Elsevier Inc. All rights reserved. Keywords: Agricultural intensification; Risk management; Bet-hedging; Neo-Darwinian theory; Hawaiian Islands; Kona field system The concept of intensification has played a pivotal role in archaeological studies of agronomic change. Recent discussions have focused on both elaboration of the core concept and empirical recognition of the in- tensification process on the ground (e.g., Kirch, 1994; Leach, 1999; Morrison, 1994). Current models of in- tensification, however, are strongly focused on produc- tive increases and often give insufficient attention to risk management, strategies that may be critical to the long- term survival of agricultural systems especially in un- predictable environments. Recent work in an extensive dryland field system on the central western slopes of Hawaii Island provides a new empirical context in which to discuss, in both substantive and theoretical terms, agricultural change. Particular attention is given to a case study from one of the most productive localities in the region, the historically important (see Beaglehole, 1967; Obeyesekere, 1992; Sahlins, 1995) land unit of Kealakekua AhupuaÔa in the Kona District. Ethnohis- toric and archaeological evidence from the region at large is also reviewed. In addressing the regionÕs agri- cultural history, issues of spatio-temporal variability in productive capacity, the degree and scale of agronomic integration, and the relative importance of productive maximizing versus risk minimizing strategies are explored. * Fax: +64-9-373-7441. E-mail address: Ms.allen@auckland.ac.nz. 0278-4165/$ - see front matter Ó 2004 Elsevier Inc. All rights reserved. doi:10.1016/j.jaa.2004.02.001 Journal of Anthropological Archaeology 23 (2004) 196–224 www.elsevier.com/locate/jaa