Testing an ecologically based classification tool on fruit-based agroforestry in northern Thailand

Ecologists are increasingly using multivariate analytical approaches to reveal relationships between communities. These methods have promise in other fields as well. The use of multivariate methods to delineate relationships and classify an agroforestry system was tested among fruit-based agroforestry gardens in northern Thailand. Data on crop species composition, species abundance, perennial-crop age groups, and other physical and ecological factors from 82 gardens in three villages in a Highland watershed in northern Thailand were used in this study. Using Hierarchical Cluster Analysis the gardens were divided into clusters, each representing a different garden type (or fruit-based agroforestry subsystem). Non-metric multidimensional scaling (NMS) analysis was used to assist in the interpretation of classification groupings and analysis gradients. The NMS analysis shows overall crop diversity, herbaceous food crops, size and market potential of the fruit planting as important classifying factors. However, this analysis did not produce as clear distinctions as hoped among gardens in a continuum of gradually changing and overlapping characteristics.