Morphological traits of control-pollinated fruits in African plum (Dacryodes edulis (G.Don).Lam.) using multivariate statistical techniques

Abstract Phenotypic variation on 26 well-known accessions of African plum collected from four provenances established as genebanks was assessed under controlled-field conditions using a full nested mating design. Data were recorded for 12 agro-morphological fruit traits using multivariate statistical techniques. Descriptive statistics for each studied trait were calculated. In addition, patterns of morphological variation were assessed using principal component analysis (PCA). Studied accessions showed considerable variation in fruit length, fruit width, fruit and pulp weight, pulp thickness and fruit:kernel weight ratio. Clustering of accessions into similarity groups was performed using Ward's hierarchical algorithm based on squared Euclidean distances. The accessions based on studied traits were classified in 03 groups. Results showed that, fruits from accessions within Boumnyebel and Kekem provenances constitute cluster 1. Accessions in this cluster had better fruits traits and could be selected as raw material for breeding purposes or clonal multiplication. Principal component analysis (PCA) revealed that the first two principal components (fruit length, fruit width) accounted for 87.01% of the total variation. Among the studied traits, fruit length, fruit width, fruit and pulp weight, pulp thickness and fruit:kernel weight ratio showed strong and high positive link with the first component (PC1) whereas kernel weight and fruit length:width ratio showed positive link with the second component (PC2). These results suggest that fruit weight is a good predictor of pulp yield, although its predicting power differed among clusters.