Best linear unbiased prediction based trait association and multivariate analysis in maize (Zea mays L.) hybrids

Ninety-four CIMMYT maize (Zea mays L.) hybrids, along with 6 commercial checks, were evaluated at Kovilpatti, Tamil Nadu, India, during the rabi 2024 and kharif 2025 seasons using an alpha lattice design with two replications to identify major traits associated with grain yield. Significant variability was observed among the hybrids for grain yield and its related traits. Grain yield showed positive association with ear height, number of plants per plot, ear length and ear girth, indicating that ear related traits play a critical role in yield improvement. Path coefficient analysis revealed that number of plants per plot, ear girth and ear length exerted the highest positive direct effects on grain yield, suggesting that these traits can be effectively used as selection criteria in maize breeding programs. Phenotypic correlations followed similartrends but with relatively lower magnitudes, reflecting the influence of environmental factors on trait expression. In contrast, best linear unbiased predictors (BLUP) based correlations were more conservative and provided stable estimates across environments. However, the major trait associations remained consistent across seasons. Hierarchical cluster analysis grouped the hybrids into three distinct clusters, demonstrating substantial variability for grain yield and related traits. Principal component analysis (PCA) further confirmed that ear-related and plant architectural traits contributed significantly to the total phenotypic variation. Overall, integrating BLUP based correlation, path coefficient and multivariate analyses enhances selection efficiency and accelerates genetic gain in multi-environment maize breedingprograms.