Multivariate classification of farming systems for use in integrated pest management studies

Abstract

On-farm studies provide a realistic setting to examine the impact of interactions of management practices on weed communities within various farming systems. Clustering farm units into groups that use similar management practices enables the replication of farm management systems within on-farm studies. The goal of this study is to objectively classify farm units into management systems on the basis of quantitative variables describing aspects of cropping history and chemical input levels using multivariate techniques. Twenty-eight Saskatchewan farmers provided details of their management practices from 1990 to 1997 through a series of questionnaires. Twelve variables derived from the questionnaires were used to describe cropping history, pesticide, tillage and fertilizer use on each farm unit. These variables were used to cluster farm units using minimum variance classification and NMS ordination. Both techniques identified seven farm management systems. The greatest differences were observed between organic and non-organic systems. Farm management systems that used annual fallow and continuous annual cropping histories were most similar. The consistent results obtained by use of the two unrelated methodologies indicate the utility of this approach for the classification of farm management systems.