A multi-agent model of social and environmental complexity of deforestation was developed for the Caparo Forest Reserve, Venezuela. It includes three types of agents: settlers, government, and lumber concessionaires. Settlers represent people of limited economic resources that deforest and occupy reserve land to grow crops and eventually claim property rights of this land. Their agricultural practices generate unintended environmental problems. The concessionaires extract lumber using management plans approved and monitored by the government. The agent model links to a cellular automata simulation of the natural system. Representational tools include Galatea (multi-agents), Actilog (rule description), and SpaSim (cellular automata). Three scenarios were explored for government policies: hands-off, pro-forestry and agro-forestry. Results agree qualitatively well with history of land-use change in the area. Old-growth forest is replaced by logged and secondary forest but the rate at whi...