Linear discriminant analysis of structure within African eggplant ‘Shum’

Abstract

African eggplant (Solanum aethiopicum) lags behind in terms of research attention; leading to, among other gaps, scantiness of information on intra-group genetic diversity. Hierarchical clustering was applied to reveal structure in a pioneer 20-entry set of S. aethiopicum Shum group accessions at Africa Solanaceae Research Network secretariat in Uganda, using 61 morpho-agronomic variables collected from a complete randomised design (CRD) with 12 plants per accession in screen-house. A MANOVA preceded linear discriminant analysis, to model each of 61 variables, as predicted by clusters and experiment to filter out non-significant traits. Four distinct clusters emerged, with a cophenetic relation coefficient of 0.87 (P<0.01). Canonical variates that best predicted the observed clusters include petiole length, sepal length (or seed colour), fruit calyx length, seeds per fruit, leaf fresh weight (or leaves per plant), fruit fresh yield, seedling vigour, fruits per plant, harvest index and plant growth habit. Results suggest prospects for variety discrimination at various stages; seedling, vegetative and reproductive. The observed diversity will boost chances of effective selection for desired traits; while the canonical discriminant traits are potential morphological markers, within S. aethiopicum ‘Shum’ for a low-cost germplasm characterisation approach. Key words: Hierarchical clustering, morphological markers, Solanum aethiopicum