Sterling B. Blanche

GGE Biplots and Traditional Stability Measures for Interpreting Genotype by Environment Interactions

Abstract Cotton (Gossypium hirsutum L.) breeders conduct multienvironment trials to determine the performance of genotypes in relation to environmental changes and to determine their area of adaptation. The objective of this study was to compare within-model and within-scaling GGE Biplot stability values (GE distance) with those generated by some of the “traditional” stability analytical methods. Correlation coefficients of GE distance of GGE Biplot (stability evaluation) with Cultivar Superiority Measure, Shuklas Stability Variance, Eberhart-Russell regression model, Kangs yield stability statistic, and AMMI were 0.54, 0.91, 0.86, 0.63, and 0.55, respectively. Correlation coefficients between GGE distance of GGE Biplot (mean performance + stability evaluation) and the Cultivar Superiority Measure, the Eberhart-Russell regression model, Kangs yield stability statistic, and AMMI were 0.95, 0.60, 0.85, and −33, respectively. Some of the “traditional” methods focus heavily on yield, while others focus on stability; GGE Biplot allows for a more versatile and easily comprehensible presentation of the data and variety selection based on both yield and stability. Based on the results of this study and our experience using GGE Biplot, Model 3 (uses replicated and standard error-standardized data) with an entry-focused scaling is the most valuable analysis for breeders to select widely adapted genotypes.

Determining Selection Gains and Discriminating Environments via GGE Biplots

Abstract Genetic variation is the basis for meaningful selection; thus, the use of locations that discriminate or result in greater variation among genotypes, for a trait or trait package should promote accurate selection of superior genotypes. The objectives of this study were to quantify the gains by selection in discriminating locations using superior parents for single or multiple-trait populations. GGEbiplot software was used to identify two levels (high and low) of discriminating locations for each of three distinct populations of cotton (Gossypium hirsutum L.). Populations were obtained by crossing parents recommended by a least squares means analysis for each population criteria, which included parents/populations selected for: (a) lint yield; (b) fiber micronaire, length, strength, uniformity, and elongation; and (c) lint yield, lint percent, fiber micronaire, length, and strength. F2 plants in 2003 and F2:3 plants in 2004 were planted in the high and low discriminating locations for each selection criteria. Heritability estimates (h2) were calculated by regressing the F2:3 plants on their F2 parents. Genotypic variance was greater among F2:3 progeny in discriminating environments compared with non-discriminating environments, regardless of population criteria. Heritability was greater in the population containing fiber traits compared with yield. The results of this study suggest that using discriminating locations during the selection phase of a breeding program can increase genotypic variance and enhance selection accuracy.

Visual Estimation of Rice (Oryza sativa L.) Grain Yield in Multiple Environments in Louisiana

Development of high-yielding rice (Oryza sativa L.) cultivars is a primary objective of most rice-breeding programs, and breeding progress is generally measured through increases in grain yield attributable to new cultivars. The pedigree-breeding method, commonly employed in rice-breeding programs, usually results in a large number of lines being discarded in early generations on the basis of visual estimates of grain yield potential. The objective of this research, conducted at five locations in 2007 and 2008 in Louisiana, was to evaluate the effectiveness of visual grain-yield predictions by two rice breeders. A second objective was to determine the effect of disease and lodging on the accuracy of visual selection. Across all years and locations, the grain yield estimates of both breeders were positively correlated with observed grain yield, indicating that visual selection for grain yield can be effective. Differences existed in the predictive ability of the breeders to estimate grain yield at different environments. Generally, breeders were more accurate in visually estimating grain yield when the standard deviation of the estimates was high, indicating that perceived phenotypic expression of grain yield is an important factor affecting the visual estimation of yield potential. Locations with high disease pressure and lodging resulted in a greater standard deviation of the yield estimates and improved the accuracy of visual estimation of grain yield.