Analysis of mixed data to select bananas clones (musa spp.) to be included in a germplasm bank

Abstract

In an asexually reproducing hybrid such as banana (Musa spp.), the assessment of clones in the short term is limited because replications are frequently unavailable in the proper number. The aim of this work is to propose the Multiple Factor Analysis of Mixed Data (MFAmix) as a tool for establishing objective criteria to identify banana clones that preserve variability for qualitative and quantitative variables. In the long term, the aim is the development of a banana germplasm bank. MFAmix was applied on a population composed of 124 banana clones collected from different farmers? fields and four controls. Two groups of variables related to the agronomic aptitude of the clones were evaluated, one composed of nine quantitative variables, and the other, composed of three dichotomous qualitative variables. A Selection Index (SI) was built from the MFAmix coordinates in order to rank the clones and select a subset that allows to preserve the existing genetic variability. The first two axes of MFAmix explained a 49.47% of the total data variability. The set of the banana clones was successfully characterized based on quantitative and qualitative variables. In the long term, the creation of a banana germplasm bank should consider the height and diameter of the plant, the rachis bunch weight and the hands weight, and the qualitative variable plant leafiness.