Based on the AMMI analysis of variance, genotypic main effects were the primary source of variation in yield, accounting for a substantial 92.31% of the total variance across the 27 genotypes evaluated in three environments. Environmental main effects explained a smaller portion (5.38%), while the genotype × environment (ENV x GEN) interaction contributed only 1.12% to the total variance (Table 2). In accordance with the findings of
Misra et al., 2009 and
Fentie et al., 2013, the study highlights that the variances due to environmental and ENV × GEN interactions were statistically significant.
The large sum of squares attributed to genotypes confirms significant inherent differences in yield potential among them. The AMMI model effectively captured the ENV × GEN interaction, as evidenced by the significance of the first three interaction principal component axes (Table 3). The first principal component (PC1) alone explained a considerable 75.8% of the interaction sum of squares, with the second principal component (PC2) accounting for an additional 24.2%. This indicates that the interaction between genotypes and the three environments can be largely predicted by the first two principal components of both genotypes and environments
(Pandey et al., 2020). In essence, while the overall contribution of the G x E interaction to the total variance was relatively small, its structure was well-defined and captured by the initial principal components of the AMMI analysis.
AMMI biplots
Biplot analysis is probably the most effective interpretative technique for AMMI models. The AMMI 1 biplot illustrates the primary effects and PC1 scores for both genotypes and environments in relation to each other, along with two additional basic AMMI biplots. The AMMI 2 biplot displays the scores for PC1 and PC2. Based on the average yield values of the genotypes across various environments, G13 exhibited the highest yield, while G19 showed the lowest yield.
AMMI 1 biplot
Biplots act as diagrams that illustrate the associations between genotypes and environments by plotting elements of both on the same axis. The AMMI 1 biplot is typically analyzed by observing that the movements along the vertical axis indicate differences in interaction effects, whereas deviations along the horizontal axis reflect variations in main (additive) effects. Environments that are grouped together exhibit a comparable influence on the genotypes, while genotypes that are combined demonstrate similar adaptations. The genotype exhibiting optimal adaptation is capable of strategizing for the future in relation to its environment. A genotype or environment is considered stable when the IPCA1 score is closer zero, signifying minimal interaction effects. An affirmative interaction takes place when both the genotype and environment align with the same sign on the PCA axis, while a negative interaction arises when they exhibit opposing signs. Utilizing the standard techniques outlined by
Gauch and Zobel in 1997, the AMMI 1 biplot allows for the calculation of predicted yield for each genotype and environment as illustrated in (Fig 1).
The AMMI 1 biplot shows that genotypes and environments on the same line or coordinate have similar yields. Those on the right side of the center exhibit higher yields. Genotypes G12, G9, G7, G13, G6, G14 and G11 displayed high yield and positive PC1 scores, with G11 identified as the best overall and specifically adapted to certain environments. Since no environment had a PC1 score near zero, no habitat was universally suitable. Consequently, the high-yielding genotypes don’t have universally favorable environments. Genotypes G2, G3, G24 and G16 had near-zero PC1 scores, indicating reliability across environments. Despite showing G × E interaction, the high-yielding genotypes are expected to perform well in various environments.
AMMI 2 biplot
The environmental values are connected to the origin through lateral arrows in the AMMI 2 biplot. Strong interaction forces are exerted by sites with lengthy spokes (Fig 2), where the lines denoting environments 1, 2 and 3 are joined to the origin, provides an illustration of it. While genotypes that occur out of contour may have a varied pattern of response across the environments, genotypes that occur in the contour tend to generate identical responses in every environment. Because they were farther from the origin, G9, G16 and G21 were more sensitive in the current investigation. All genotypes, with the exception of G9, G16 and G21, were located near the origin, suggesting their stability and lack of susceptibility to the impacts of environmental interactions.
Eigen values were utilised as a key criterion to establish the proper amount of elements to keep. Six factors with eigenvalues larger than one were chosen from the group (Table 4). The attributes’ cumulative variance was 83.30% explained by these six variables. As a result, only the data with strong explanatory power may be kept when the dimensionality of the data is reduced by about 17%. Sorting the 19 characteristics into these six variables was the following stage.
The best four genotypes-G11, G13, G4 and G5-were determined by ranking each genotype according to the required characteristic values using the MGIDI analysis (Fig 3). The MGIDI study effectively estimated 19 features and predicted selection differentials (SD) ranging from 3.36% to 39.10% for various traits in selected genotypes, which all exhibited good trait values. Genotype G6’s proximity to the selection boundary suggests it might possess unique characteristics. High heritability was observed for traits X1, X2, X3, X8, X10, X16, X17, X18 and X19, indicating strong potential for improvement through selective breeding. Traits X8, X10 and X19 showed both high heritability and high SD, highlighting them as key targets for breeding programs aimed at developing superior genotypes.
Strength and weakness of genotypes
Strengths and Weaknesses of Genotypes Based on MGIDI Factors: The strengths and weaknesses of several genotypes are shown in detail in (Fig 4), which is based on how each component contributes to the Multi-Trait Genotype-Ideotype Distance Index (MGIDI). G11 and G13, two genotypes linked to Factor 1 (FA1), exhibit distinct strengths in qualities including X4, X5, X7, X8, X10, X16, X18 and X19. Conversely, FA2-associated genotypes G5 and G4 exhibit strength in characteristics such as X1, X2, X3 and X13. Furthermore, features like X12 and X14 are stronger in genotypes G4 and G5, which are linked to FA3. Additionally, FA4-associated genotypes G5 and G11 show strength in characteristics like X6 and X11. Finally, both FA5 and FA6 were linked to genotypes G11 and G13, where FA6’s X9, X15 related qualities and FA5’s X6, X11 related characteristics (Table 5).
These understandings of genotype strengths and weaknesses can be very helpful in guiding the choice of parents for next breeding initiatives (
Vanitha and Mahendran., 2018). The study conducted by
Mamun et al., (2022) highlighted the importance of the optimum genotype found using MGIDI, highlighting its potential for better quantitative characteristics. The discovered genotypes-including some that are the focus of this study-prove to be the best for next breeding initiatives and are essential for improving crop quality in general. High-performing and strong varieties are developed as a result of this deliberate use of characteristics and variables
(Saidaiah et al., 2011).
More important influencing elements are located close to the middle of the strength and weakness plot, whereas smaller contributors are more distant out. This helps determine appropriate parentage for hybridisation projects. The radar graphic displays the characteristic influence hierarchy for each selected genotype by underlining the MGIDI participants. This helps to match characteristics to the ideotype, as seen by elements closer to the plot’s edge
(Ponsiva et al., 2024). After a comprehensive evaluation of several features, genotype ranking revealed that G11 and G13 were the best-performing genotypes. This suggests that they have the ability to improve characteristics associated with total yield. It is expected that the use of the MGIDI index for agricultural research would expand quickly. This index was used by
Singamsetti et al., 2023; Azrai et al., 2024; and
Zendrato et al., 2024 to determine the optimal genotypes based on their strengths and weaknesses. Genotype ranking was made easier by the incorporation of yield in additional target criteria, underscoring the need of evaluating several features at once.
