Integrating BLUP, AMMI, and GGE Models to Explore GE Interactions for Adaptability and Stability of Winter Lentils (Lens culinaris Medik.)

Abstract

Lentil yield is a complicated quantitative trait; it is significantly influenced by the environment. It is crucial for improving human health and nutritional security in the country as well as for a sustainable agricultural system. The study was laid out to determine the stable genotype through the collaboration of G × E by AMMI and GGE biplot and to identify the superior genotypes using 33 parametric and non-parametric stability statistics of 10 genotypes across four different conditions. The total G × E effect was divided into two primary components by the AMMI model. For days to flowering, days to maturity, plant height, pods per plant, and hundred seed weight, IPCA1 was significant and accounted for 83%, 75%, 100%, and 62%, respectively. Both IPCA1 and IPCA2 were non-significant for yield per plant and accounted for 62% of the overall G × E interaction. An estimated set of eight stability parameters showed strong positive correlations with mean seed yield, and these measurements can be utilized to choose stable genotypes. The productivity of lentils has varied greatly in the environment, ranging from 786 kg per ha in the MYM environment to 1658 kg per ha in the ISD environment, according to the AMMI biplot. Three genotypes (G8, G7, and G2) were shown to be the most stable based on non-parametric stability scores for grain yield. G8, G7, G2, and G5 were determined as the top lentil genotypes based on grain production using numerical stability metrics such as Francis's coefficient of variation, Shukla stability value (σi²), and Wrick's ecovalence (Wi). Genotypes G7, G10, and G4 were the most stable with the highest yield, according to BLUP-based simultaneous selection stability characteristics. The findings of graphic stability methods such as AMMI and GGE for identifying the high-yielding and stable lentil genotypes were very similar. While the GGE biplot indicated G2, G10, and G7 as the most stable and high-producing genotypes, AMMI analysis identified G2, G9, G10, and G7. These selected genotypes would be used to release a new variety. Considering all the stability models, such as Eberhart and Russell's regression and deviation from regression, additive main effects, multiplicative interactions (AMMI) analysis, and GGE, the genotypes G2, G9, and G7 could be used as well-adapted genotypes with moderate grain yield in all tested environments.

Keywords: AMMI; BLUP; GGE biplot; lentil; non-parametric and parametric stability indices; regression and deviation from regression.

Figure 1

The seed yield variation of the studied ten lentil genotypes across four environments (A,B). The AMMI1 and AMMI2 biplot (yield vs. principal component 1 (PC1)) for grain yield (kg/ha) of 10 lentil genotypes evaluated under four environments ((C,D), respectively).

Figure 2

Correlation among various stability parameters with yield data of 10 lentil genotypes evaluated under four test environments. AMGE—sum across environments of genotype × environment interaction (GEI) modeled by AMMI; ASI, AMMI Stability Index; ASV, AMMI Stability Value; ASTAB, AMMI-based stability parameter; AVAMGE—sum across environments of the absolute value of GEI modeled by AMMI; DA—Annicchiarico’s D parameter; DZ—Zhang’s D parameter; EV—averages of the squared eigenvector values; FA stability measure based on fitted AMMI model; MASI—Modified AMMI Stability Index; MASV—Modified AMMI Stability Value; SIPC—sums of the absolute value of the IPC scores; Za—absolute value of the relative contribution of IPCs to the interaction. *, **, and *** Significant at 0.05, 0.01, and 0.001 probability levels, respectively.

Figure 3

Wi²—Wricke’s (1962) ecovalence; b_i—regression coefficient; S²di—deviation from regression; σ²—Shukla’s (1972) stability variance; CV—Francis and Kannenberg’s (1978) coefficient of variance; Tai—Tai stability; R²—Pinthus’s (1973) coefficients of determination; S⁽ⁱ⁾—Huehn’s (1979) and Nassar and Huehn’s (1987); NP⁽ⁱ⁾—Thennarasu’s non-parametric (1995) measures *, ** and *** significant at 0.05, 0.01 and 0.001 probability levels, respectively.

Figure 4

The polygon (which-won-where) shows the (G + G × E) interaction effect of 10 lentil genotypes.

Figure 5

The mean vs. stability view shows the genotype main effects plus the genotypic environment interaction effect (G + G × E) of 10 lentil genotypes.

Figure 6

GGE biplot showing the ideal genotype and comparison of the genotypes with the ideal genotype.

Figure 7

GGE biplot shows the performance of environments relative to the ideal environment.

Figure 8

Discrimination ability vs. representativeness for four testing environments.

Figure 9

Study sites showing on the map. JAS = Jashore; ISD = Ishwardi and MYM = Mymensingh.