Clustering shrub and tree legumes grown in acid and non-acid soil conditions using rank performance data.

Abstract

Clustering of ten woody and shrub legumes (Acacia auriculiformis, Albizia lebbeck, Gliricidia sepium, Leucaena diversifolia, L. leucocephala var. K28 and var. K636, Lonchocarpus sericeus, Cajanus cajan, Crotalaria juncea and Tephrosia candida) was performed using the partitioning around medoid and the Fuzzy clustering methods to assess the appropriateness of their earlier performance rank scores for evaluating their dissimilarity within an Alfisol and an Ultisol. Differences among formed clusters were observed for the acid and non-acid soil conditions. Some species such as Albizia lebbeck and Leucaena leucocephala var. K636 isolated each in a single-species cluster, while others, depending on the soil type belonged to different closest hard clusters most likely in relation to their closeness in architecture and growth habit as well as to their genetic affinity which appear to be strong determinants of their performance similarity. Clustering analysis of species performance compared reliably well with the simple ranking in the Alfisol than it did in the Ultisol. In both soils, however, clustering analysis overcame the rigidity of the simple ranking procedure and thus led to more realistic ranking classes of species performance to the extent that, where possible, it should be recommended for discriminating species more efficiently.