Use of correspondence analysis in agroecosystem health and sustainability assessment

Describing agroecosystems and assessing their sustainability and health has become of great interest to researchers, Development agents and communities. Such assessments are frequently carried out using a group of carefully chosen attributes, known as indicators. This was the approach adopted in a project whose objective was an integrated assessment of a tropical-highlands agroecosystem. Initially, information on a large numbers of variables was collected. However, it was felt that many of the variables initially measured were unlikely to provide important information and required great effort to collect. Thus, methods and approaches to develop useful subsets of indicators were investigated. In this paper we explore the use of multiple correspondence analysis (MCA) in two contexts: (1) as a basis for system description and analysis and (2) as a tool to choose indicator groupings that are relatively parsimonious and informative.