A comprehensive comparison between Eberhart and Russell joint regression and GGE biplot analyses to identify stable and high yielding maize hybrids

Abstract

Multi-environment trials (MET) are commonly conducted in plant breeding programs to evaluate cultivars and hybrids. Identifying stable and high yielding cultivars/hybrids, using various analytical tools, forms the main crux of MET data analysis. The objective of this study was to compare the effectiveness of the Eberhart and Russell joint regression method and GGE biplot analysis, two stability analysis approaches that are widely used to identify high yielding and stable genotypes. The study included 24 maize hybrids evaluated at 24 locations across 7 Midwestern states in 2007. Environment (E) main effects accounted for >80% of all variation, compared to <20% for genotype (G) and genotype-by-environment interaction (GEI) effects together. Based on the variance deviation from error mean squares estimates, the Eberhart and Russell method identified HY24 > HY23 > HY22 as the most stable hybrids but could not reliably identify the high yielding hybrids. On the other hand, the GGE biplot reliably and graphically displayed the high yielding and stable hybrids (HY07 > HY20 > HY04) as well as low yielding (HY10 < HY19) hybrids consistent with their respective overall yield ranks. From the GGE biplot analysis, hybrids suited to specific locations/environments were also identified. Results from this study, even though based on one year MET data, indicated that GGE biplot analysis would serve as a better platform compared to Eberhart and Russell joint regression analysis to identify stable and high yielding hybrids.