Plant selection matrix for classifying elephant forage plants in Sri Lanka using metadata

A Cluster and Classification -and- Regression Tree Approach




Elephant conservation, elephant forage grass establishment, cluster analysis, Classification and Regression tree analysis


The elephant, the largest terrestrial herbivore in the world, requires a higher amount of plant material for survival. The degradation and shrinking of forests on the island resulted due to anthropogenic activities, drive to deplete the habitats and niches of elephants migrating towards new habitats and food sources. Identifying the preferences of plant species is imperative in enhancing habitat enrichment endeavours to mitigate Human-Elephant Conflict (HEC). There was a scarcity of in-depth knowledge available on the feeding behavior and dietary patterns of Asian elephants grazing on rangelands and secondary forests. The work carried out on the elephant forage, in general, indicated that most of the Sri Lankan elephants preferred grasses over the other plants, the preferences vary with the seasonal availability of forage plants during dry/wet seasons. The establishment of forages could be considered one of the remedial measures for wildlife conservation. The objective of the study was to develop a plant selection matrix using plants and ecological characters from metadata analyzed using Cluster analysis (CA) and, Classification and Regression Tree (CART). CA produced three clusters at 80% phenon level and characterized by plant height (p < 0.05), soil pH (p < 0.05) and habitat type (p < 0.05). CART recognized the type of lifecycle, soil type and method of pollination as decisive features.  Both cluster and classification and regression tree analyses identified the important character /character state that helped to group elephant forage. These characters were used to construct a forage plant section matrix for identifying proficient forge plants with minimum effort.

Author Biographies


In order to become a novel scientist with non-traditional thinking and technology enthusiastic with a genuine interest in studying, researching, hardworking, and committed to strive out smart solutions to accomplish the challengeable requirements of living beings. My goal is to achieve excellence in my career in diverse fields while fulfilling organizational goals with high in quality and rich in creativity to build up a unique professional and a researcher at the same time.

Rangika Umesh Halwatura, University of Moratuwa

Department of Civil Engineering, Faculty of Civil Engineering

Somaratne Seneviratne, Open University of Sri Lanka

Department of Botany, Faculty of Natural Sciences




How to Cite

UYANAHEWAGE, D. ., Halwatura, R. U., & Seneviratne, S. (2022). Plant selection matrix for classifying elephant forage plants in Sri Lanka using metadata: A Cluster and Classification -and- Regression Tree Approach. Journal of the Sri Lanka Association for the Advancement of Science, 4(1), 22–39.