Author Description

. Although these phenomena were observed as long ago as the beginning of this century, they are still causing concern. Until recently, analysis of variance was applied to any form of undesired variation in the experimental results. The statistical elimination of variation from the research coincided with attempts to uniformise agriculture in order to optimise production. For a long time, deviations from the standardised average were regarded as undesirable random effects. Recent research has shown that variations in yield are very common in agroecosystems. They may be large, especially under difficult climatic conditions, and may even be considered as an asset to farmers. There is evidence that variations are not random, but are the result of systematic interaction between environment, crop genotype and management. This agrodiversity has important relations with the higher-level heterogeneity of the environment and diversity of farm households. At present, no comprehensive approach to its analysis is available, largely because agrodiversity is basically multidimensional in nature and may comprise several levels of aggregation. The objective of this study was to contribute to the understanding of diversity in agroecosystems by focusing explicitly on variations of yield and their explanation at field and household level. Between 1991 and 1993, field work was done in one village in northern Cameroon. Yield variations of three crops, field properties, crop and management characteristics and household characteristics were assessed systematically. Various techniques for statistical analysis were employed to determine the magnitude of variation and to define the agrodiversity of Within the agroecosystern of the selected village, variations in yield were observed for rainfed sorghum, cotton and dry season sorghum. The magnitude of variation was more or less constant over the years of observation, but varied between the crops. Two rainfed sorghum cropping systems were distin