A multivariate approach for discriminating homegardens in different agro-ecological regions

Though the importance, functions and values of homegardens are well described, the interactive environment and functional features of individual components which vary widely due to agro-ecology, socio-economy and administrative structures are poorly identified. Therefore, the common interventions for improvement often do not result the expected outcome. To fill this knowledge gap, we followed a multidisciplinary approach employing multivariate techniques to identify variables critical in distinguishing homegardens among ecological and administrative units. A homegarden survey was conducted in four districts in Sri Lanka. Present status of the homegardens in different districts were compared and the characteristics important in distinguishing different classifications were identified by discriminant function analysis. Accordingly, most of the variables were important in distinguishing selected districts, climatic zones and Agro-Ecological Regions in the study. When distinguishing Agro-Ecological Regions within districts and climatic zones, fewer number of variables were contributed. The derived variables were utilized in comparing characteristics of homegardens and predicting homegardens. Correspondingly, discrimination of homegardens using climatic zones is recommended at global scale as any region can use this information and required fewer variables than districts. Further, predictions can be accomplished based on discrimination of homegardens using Agro-Ecological Regions within climatic zones at local scale as the spatial-scale specific interventions are possible.