AMMI analysis
As shown in Table 1, a combined analysis of variance and AMMI analysis revealed that environment, genotype and G×E interaction had a highly significant difference. Significant variance was found for grain yield showing that yield was significantly affected by the environment (Table 1). Variation was seen among the environment as a high sum of the square was seen for the environment. Therefore, due to genotypic and environmental effects, plant grain yield differed.
AMMI stability value (ASV) ranks the genotype using the AMMI model
(Purchase et al., 2000). As shown in Table2, the ASV ranking value reveals that genotype 15 was found to be most stable, having the lowest ASV, whereas genotype 3 and 9 were found unstable with higher ASV.
AMMI model shows that G×E interaction is present and this interaction is spitted among the first and second IPCA (interaction principal component axis) Fig 1. Through the biplot graph, genotype 15 lying near the center was found stable in all four environments.
Genotype and genotype-environment analysis (GGE analysis)
GGE biplot based on environment focused scaling and is used for estimating the pattern of environments. Environment PC1 score had negative and positive scores, which shows the difference in the genotypes yield over different environments resulting in cross-over G ×E interactions.
The vector view of the GGE biplot (Fig 2) represents a summarized view of interrelationships among environments. Two groups of the environment were formed having E1, E2 environment in one group and E3, E4 environments in other groups. A positive correlation was found between E1and E2, having an angle less than 90. The correlation was seen within the environment group (E1, E2) and group (E3, E4), revealing that indirect selection can be made across these environments. For example, a genotype having adaptation in E1 may also show adaptation in E2.
That won where the GGE biplot pattern shows which genotype was found best in each environment and forms environment groups (Fig 3). It also represents a polygon view of GGE biplot. Genotypes scores present at the furthest point from the origin are joined to form a polygon and within it, remaining genotypes are present. Genotype won is found based on genotypes relationship with site scores
(Yan et al. 2000). At polygon vertices, eight genotypes
viz, G9, G12, G3, G7, G8, G15, G16 and G5 are present. Genotypes present at the polygon vertex are the best desirable genotypes for the environment present in the same sector where vertex genotype is present
(Yan 2002; Yan and Tinker 2006). Equity lines form eight sectors in a polygon and all four environments are present in three sectors, hence forming three mega-environment. G9 genotype won in first mega-environment E1, G12 genotype performs best in second mega-environment E2 and G3 was found winning in third mega-environment E3, E4 (Fig 3).
The concentric circles on the biplot picture the length of the environment vectors, which measures the environments’ discriminating ability. Therefore, the most discriminating (informative) environments where E3, E1 and E2 and E4 least discriminating (Fig 4). The average environment (represented by the small circle at the end of the arrow) has the average coordinates of all test environments. AEA is the line that passes through the average environment and the biplot origin. E2 was the most representative having a smaller angle with the AEA (Fig 4), whereas E4 was the least representative. E2 environment was both discriminating and representative; thus, E2 was found to be good test environments for selecting generally adapted genotypes.
(Yan and Tinker, 2006).
Genotypes should be evaluated on both mean performance and stability across environments in a single mega-environment. In Fig 5, the GGE biplot shows the average-environment coordination (AEC) view. AEC abscissa (or AEA) points to higher mean yield across environments. Thus, G3 had the highest mean yield, followed by G18, G11, G10, G12 and G13,
etc. G1 had a mean yield similar to the grand mean and G15 had the lowest mean yield. Another line is the AEC ordinate it points to greater variability (poorer stability) in either direction. Thus, G9 was highly unstable, whereas G1, G13, G15 were highly stable. G9 was highly unstable because it had lower than expected yield in environments E3 and E4 but higher than expected yield in E1.
(Yan and Tinker, 2006).
The ideal genotype graph shows genotype ranking for grain yield and stable performance (Fig 6).
A high stable genotype with high performance over the different environments is said to be the ideal genotype
(Yan and Tinker 2006). A Pointing arrow in a graph represents an ideal genotype (Fig 6). Taking the ideal genotype at the center, concentric circles are drawn. These concentric circles locate the genotype’s distance from an ideal genotype
(Yan and Tinker 2006). A desirable genotype is present near the ideal genotype. Genotype G12 next to it G18, G13, G11, G2 and G10 were present near-ideal genotypes; therefore, these genotypes ranked up for high mean yield and stable performance. These genotypes were found most desirable over four environments
(Karimizadeh et al., 2013).
In this study, GGE biplot and AMMI was used to evaluate yield stability and different environment condition representativeness for wheat genotypes. G12, G18, G13, G11, G2, G10 and G15 showed higher grain yields and stability compared with other genotypes for partial irrigation and late sowing condition. On the other hand, two groups of the environment were formed having E1, E2 environment in one group and E3, E4 environments in other groups. A positive correlation was found between E1and E2, having an angle less than 90. The correlation was seen within the environment group (E1, E2) and group (E3, E4), revealing that indirect selection can be made across these environments. E2 environment was both discriminating and representative; thus, E2 was found to be good test environments for selecting generally adapted genotypes.
