The humid tropics are characterised by a high rainfall surplus, often acid soils with shallow crop root development and intensive leaching of soluble nutrients, especially NO³, below the crop root zone. In addition to the possible role of tree roots acting as nutrient pumps bringing nutrients from deep soil horizons otherwise inaccessible to the trees it has been proposed that tree roots may act as a safety net, catching nutrients which would otherwise be lost by leaching. This hypothetical role of tree root systems below the crop root zone is much debated but little hard data exists to prove/disprove it. Few attempts have been made to model and test the safety-net hypothesis which would help identify the conditions under which this role is most likely to contribute to the overall nutrient use efficiency of agroforestry systems. The agroforestry system in which trees and crops are grown simultaneously which has the best prospect for a quantitative understanding of the positive and negative tree-soil-crop interactions is alley-cropping or hedgerow intercropping system. Results with this system under farmers conditions are often disappointing, indicating that the technology has been taken out of its appropriate domain. Thus, there is a lack of understanding of the underlying principles governing success of hedgerow intercropping and failure in adapting the system to local conditions.

To estimate the degree of the safety-nets function of deep roots in agroforestry systems in the humid tropics and incorporate a proper description of this effect in existing (modular) agroforestry models.

• To test safety-net function of roots in agroforestry systems in the humid tropics i.e. the hypothesis that deep rooting trees are able to reduce leaching losses, and compete less with associated crops, thereby improving system productivity.
• To identify management options (pruning height, frequency etc.) which improve nutrient recycling and reduce competition between tree and crop components by altering the vertical root profiles of agroforestry trees.
• To identify simple indicators of below-ground tree architecture which allow a rapid evaluation of species.
• To incorporate the improves understanding of below-ground C and N and H²O dynamics into existing (modular) agroforestry models.
• To use the new understanding of below-ground competition between tree and crop components to identify locally adapted suitable tree-crop combinations and management options which will increase nutrient use efficiency in agroforestry systems and decrease risks of environmental pollution.