Multi-scale system approaches in agronomic research at the landscape level

Spatial multi-scale analyses of actual land use system performance as deter mined by spatial yield variability reveals the need for landscape research in agronomy. Main 'drivers' of spatial yield variability for five different crops in Honduras, Costa Rica and Ecuador were identified. It is demonstra ted how they vary with spatial scales and that landscape-related factors of ten play a large and significant role in when the variability in yield is d etermined. These results indicate that landscape experiments in agronomy ar e relevant. Apart from empirical analysis, spatial-temporal explicit modeli ng of landscape process dynamics such as water and soil redistribution with in a landscape can give insight in the performance of agronomic systems wit hin a dynamic landscape context. For a case study in the South of Spain it is demonstrated how within a landscape this type of research can determine the on- and off-site effects of water and soil redistribution in a,oro-ecos ystems. Only after a spatially explicit multi-scale system analysis and exp lorative landscape process modeling is completed, relevant agronomic landsc ape experiments can be designed. (C) 2001 Elsevier Science B.V. All rights reserved.