Multi-strata agroforestry systems are abundant in Asia, Africa and Latin America., often as small intensive land use units managed by women, to produce household goods and income, but in some areas they are more extensively used for income generation. There is an urgent pressure from resource-poor smallholder farmers in the forest margin to produce more from complex multi-strata agroforestry systems, to improve their food security and income generation. This is a widespread problem in the tropics, affecting a range of plant associations with broadly similar ecological structure (e.g.: forest gardens across the tropics, jungle rubber and damar in Indonesia, jungle cacao in Cameroon and Ghana, spice gardens in Sri Lanka and jungle tea in Thailand). Farmers' current rotational systems often involve clear-felling with slash and burn methods, which are environmentally destructive (including greenhouse gas emissions and severe reduction of the populations of long-lived plant species, and those animals dependent on the maintenance of forest cover) and carry risks of uncontrolled fire. Conventional solutions presently available involve conversion of complex agroforests to simple, intensively-managed, 'orchard'-type plantation systems, involving loss of biodiversity and consequent stability and sustainability problems. As an alternative, the productivity of these complex agroforests could be sustainably improved by incremental change to current practice, intensifying the multi-strata system. To enable this, there is a need to understand farmer decision making, particularly the extent to which their biological understanding is robust, and how it is integrated with socio-economic factors.

Strategies for the improved management of tree-based systems, integrated with animal and crop production systems developed and promoted.

• Local ecological knowledge, socio-economic factors and market information acquired.
• New generic method to formalise the socio-economic component of farmer decision-making, and integrate it with local knowledge and market information, developed and tested.
• Ecological model of agroforest stand development produced (funded by ICRAF).
• Generic ecological strategies for improvement of productivity and environmental protection identified.
• Improvement paths for multi-strata agroforestry systems in Indonesia, incorporating household and village level decisions, developed.
• Improved methods for formalising and incorporating farmer decision-making in research and development, and recommendations for improving productivity and environmental impact, promoted and disseminated.