Combined analysis of fixed effect and mixed effect models aids in selection of stable and productive maize (Zea mays L.) inbreds

Abstract

A successful commercialization of a maize hybrid does not solely depend on the heterotic vigor and yield potential but also on the seed yielding ability of the parental lines which dictates the cost of hybrid seed production. Assessing the inbreds’ stability and yield performance similarly across environments could serve as an important pre-breeding step. In this study, seventy inbred were evaluated for grain yield and other traits. Joint analysis of variance showed significant differences across environments. Genotypes contributed to the highest variation followed by genotype x environment interaction effect and environments. AMMI analysis divided the variation into two PCs, with PC₁ and PC₂ explaining 56% and 44% of the total variation respectively. AMMI, GGE biplots and BLUP based analyses were employed to study the stability of genotypes for grain yield across environments. Which-won-where biplot revealed that UMI 1260 and UMI 1266 are suitable for Coimbatore (E₁) and Bhavanisagar (E­₂), while IMR 108200 and UMI 1257 are suited for Vagarai (E₃). Discriminativeness vs representativeness biplot showed that Coimbatore and Vagarai are discriminative in nature while Bhavanisagar is an ideal environment for selecting the genotypes. Multiple stability indices such as YSI, HMRPGV, WAASBY are employed to assess the stability of inbreds for grain yield. MTSI was employed to select inbreds based on multiple traits. Positive selection gains were observed for yield attributing traits, while negative gains were observed for plant height and ear height. A selection intensity of 15% is employed to select ten best inbreds though MTSI. Coincidence index at 20% selection intensity revealed high level of coincidence between WAASBY and HMRPGV, indicating the selection efficiency in using either of the indices. Genotypes UMI 1286, UMI 1276, UMI 1266 ranked consistently across all the stability indices and hence can serve well as potential parental lines for hybrid development.