Résumé
The need for new improved cultivars adapted to modern cultivation techniques and changing climatic environment requires genetic studies able to provide a better understanding of the phenotypic variance decomposition into genetic and environmental variances. In perennial crops, this decomposition requires evaluations carried out among years, trees within a genotype, samples within a tree and their interactions. The aim of this study was to perform such a detailed variance analysis of oil content components (fruit fresh weight-FFW, fruit moisture-FM, oil content on both fresh-OCFFW and dry-OCFDW basis) and fruit morphology traits (fruit maximum diameter-FMD and fruit circularity-FC) in an olive progeny issued from the cross ‘Oliviere’ × ‘Arbequina’. A total of 106 genotypes from this progeny, together with the parents cultivars, were evaluated over two consecutive years in two trees/genotype, and three samples/tree for oil content components and 30 samples/tree for fruit morphology traits. Two dataset were considered during analysis. The first one, including a limited number of genotypes without any missing data, was fully balanced and ANOVA was applied for variance components estimation from expected mean squares. The second dataset contained the full unbalanced information of the 106 genotypes and mixed linear models were built following to the same statistical model. Both analyses showed that variance among genotypes (σ G 2 ) was the main contributor to total variance for FFW, OCFFW and OCFDW and variance among years (σ Y 2 ) for FM. Residual error effect (σ ε 2 ) was the main contributor for morphology traits, whereas variance associated with the genotype × year interaction (σ GY 2 ) and tree × year interaction (σ TY 2 ) represented also important variance components for most traits, and variance among trees within a genotype (σ T 2 ) was negligible. Consistently, the mean broad sense heritabilities (H²) showed the lowest values for FM (0.40) and the highest values for OCFFW (0.94). An estimation of heritability and environmental variation under varying numbers of years and tree replications revealed that including more yearly replications could be more effective than adding trees replications. Finally, this study provides a global view of adequate spatial and temporal replications required to accurately develop criteria for both conventional and marker-assisted selection on fruit traits from olive crossing progenies, and more generally in fruit trees.