Evaluation of oilseed sunflower genotypes based on the multivariate selection index FAI-BLUP under normal and limited irrigation conditions

Objective
The production and introduction of the ideal genotype require multi-site experiments and examination of the effects of genotype × environment interaction. For this purpose, several multi-attribute selection indices have been introduced. One of the advantages of multi-trait selection indices in the selection and introduction of desired genotypes is the consideration of trait relationships and the elimination of limitations caused by them. In this study, the efficiency of the FAI-BLUP selection index in evaluating oilseed sunflower genotypes was compared with two other indices: the Smith-Hazel and multi-trait genotype-ideotype distance.
Materials and Methods
One hundred oilseed sunflower genotypes were evaluated in 2012 and 2013 in terms of yield and several morphophysiological traits under two irrigation conditions: standard and limited, using a simple 10 × 10 lattice design. SAS software was used for variance analysis, and the calculation of three indices (MGIDI, FAI-BLUP, and Smith-Hazel) was done using the metan package in RStudio software (R 9.4).
Results
Based on the analysis of variance, significant differences were observed among genotypes in all studied traits except for head diameter and relative water content, indicating the presence of variation among genotypes in these traits. In the principal component analysis, four components with eigenvalues greater than one were identified, which explained 70.37% of the total variance of traits. The Smith-Hazel index was not successful in the selection based on 13 traits due to collinearity among them. By removing the RWC trait and using the remaining 12 traits, genotypes 13 (A-FLPOPA), 22 (AS3232), 57 (SDR19), 81 (CAY), and 94 (SF092) were selected based on this index. Both the FAI-BLUP and MGIDI methods are multivariate approaches based on factor analysis; however, the MGIDI method, in addition to factor analysis, also uses the distance from the ideal genotype to identify desirable genotypes. In this study, the results obtained using the MGIDI and FAI-BLUP methods were identical, indicating the efficiency and reliability of the FAI-BLUP method.
Conclusion
The use of multi-trait indices is a valuable method for accelerating breeding programs and identifying desirable genotypes. In this context, achieving optimal selection requires considering the relationships among traits in the selection process and utilizing an efficient index. As observed in this study, the Smith-Hazel index does not perform well in selecting genotypes when there are correlations between traits.