1Department of Plant Pathology, University of Georgia, Athens, GA 30602, USA; 2Scottish Crop Research Institute, lnvergowrie, Dundee DD2 5DA, UK; 3Department of Entomology and Nematology, Rothamsted Experimental Station, Harpenden, Hertfordshire AL5 2JQ, UK

Global change involves changes in atmospheric composition, average climate and climate variability, as well as land cover and land use. Because of the worldwide scale of these changes and their complex interactions with biological systems, an effective global change research and impact assessment programme can be achieved only with a multidisciplinary strategy that brings together relevant national and international research. With this consideration in mind, several international research networks, formal or informal, with focus on global change have been established in the biological and medical sciences, including the Global Change and Terrestrial Ecosystems (GCTE) project of the International Geosphere-Biosphere Programme (IGBP). In plant pathology, the network approach is being realised through a GCTE-IGBP project on 'Global Change Impact on Pests, Diseases and Weeds', the overall objective of which is two-fold: (i) to provide the foundation for an analysis of the impacts of global change on plant diseases; and (ii) to develop collaborative, global research networks to facilitate comparative studies of modelling approaches and the sharing of data. Specific objectives, contributing research projects, initial results and future challenges of the networking approach will be discussed using the GCTE-IGBP project as a case study.

Since there are far more possible global change scenarios than can be tested experimentally, global change research programmes in plant pathology, including the GCTE-IGBP project, have relied largely on modelling for biological impact assessment. Process-based approaches for modelling plant disease in relation to global change include the integration of global climate model (GCM) output scenarios with regional crop-growth models, disease models and crop loss models. This approach presents at least three major difficulties that must be overcome before credible impact assessments are possible: (i) GCM outputs need to be scaled down from continental to regional scales; (ii) crop and disease models need to be scaled up from field to regional scales, while at the same time taking account of the non-linearities in the growth responses of plants and pathogens; and (iii) measures of uncertainty need to be included in the assessment. A fourth concern with process-based simulations is the difficulty of accurately parameterizing all important interactions between host, pathogen and the environment. With these considerations in mind, we argue that empirical and semi-empirical modelling approaches may be more suitable for impact assessment than process-based models, at least in the near term. Examples of plant pathosystems in which empirical models were used for impact assessment will be discussed.

Possibly the most important driving forces for changes in plant disease in a changing environment will be shifts in cropping patterns. These will more probably result from socio-economic considerations than from climatic factors alone. However, we are currently severely limited in our ability to develop impact assessments with socio-economic components.