Implications of statistical analysis of initial agroforestry experiments

Data from the tree components of fifteen agroforestry experiments were examined for patterns of variability, normality and correlation. Variances of data from intercropped plots were generally similar to those from monocropped plots and similar patterns for both types of plot were demonstrated over time. Non-normality of distributions were frequently indicated for both types of plot. Some examination was made of correlations between tree and crop components. Recommendations are made for suitable statistical analysis of such data and indications of experimental needs are given.