Stability of some of chickpea (Cicer arietinum L.) genotypes by AMMI indices and biplots

Chickpea (Cicer arietinum L.) is one of the most important legumes in the world after pea and bean and is rich in protein (21.7-23.4%), minerals (iron, phosphorus, calcium, zinc, potassium and magnesium), carbohydrates (41.1-47.4%) and vitamins (B1, B2, B3, B5, B6, B9, C, E, K). The interaction of genotype by environment, as a response of genotypes to the environmental variation is a source of complexity for breeding programs and preparation of high yielding and stable genotypes. One of the most important ways to discover the nature of genotype by environment interaction is stability analysis, which identified the stable or compatible genotypes. Different methods for investigation of genotype by environment interaction and determination of stable genotypes have been reported, which generally include uni-variate and multivariate methods. One of the multivariate methods is AMMI analysis. The purpose of present study is the evaluation of the stability of chickpea genotypes using AMMI indices and biplots. 

Eighteen selective advanced genotypes of chickpea from ICARDA with two check verities (Arman and Azad) evaluated across four locations (Gachsaran, Ilam, Gonbad and Khoramabad) at three growing seasons, in a completely randomized block design with three replications. The data of 3rd year in Gonbad were lost and therefore, the analysis of data performed on 11 environments. Average seed yield of genotypes estimated at each environment (combination of location and growing season) and used for analysis. Statistical analyses including simple analysis of variance, combined analysis of variance and stability analysis carried out by metan (Multi environment trial analysis) R package. Five AMMI stability indices including ASV (AMMI stability value), SIPC (Sum of IPCs scores), EV (Eigenvalue stability parameter of AMMI), Za (Absolute value of the relative contribution of IPCs to the interaction), WASS (Weighted average of absolute scores) and simultaneous selection index (ssi) of these parameters was used for stability evaluation of genotypes. 

Combined analysis of variance indicated environment, genotype and genotype by environment interaction accounted 79.2, 2.4 and 10.0% of the phenotypic variation of seed yield, respectively. The significant effect of genotype indicated the wide genetic background of genotypes, while the significant effect of genotype by environment interaction is indicated the diversity of genotypes in test locations and growing years and exhibited the necessary of evaluation of genotypes in multiple environments. According to the significant effect of genotype by environment interaction, AMMI analysis was carried out by principal components analysis and the results indicated the significant effects of five principal components on seed yield. The first two principal components contributed to 42.5 and 19.4% of genotype by environment interaction. Genotype G3 (1663 kg ha-1) followed by G1, G17, G20 and G5 had the highest seed yield. According to the ASV and WAAS indices, G3, G4 and G13; EV and ZA indices, G3, G14, G16 and G6; and SIPC, G14, G3, G11, G4 and G16 were the most stable genotypes. Selection of these genotypes was done only on stability aspect of genotypes, therefore simultaneous selection index (ssi) based on any of these parameters was used to evaluate the simultaneously selection of genotypes based on seed yield stability and performance. Genotypes G3, G1, G4, G13 and G16 selected as superior genotypes by ssiASV, ssiZA and ssiWAAS indices; while based on ssiSIPC and ssiEV, G3, G16 and G20 were as the best genotypes. The other applications of AMMI stability analysis are selection of best genotypes in any of environments. According to this procedure, genotype G3 was placed in the first order in E1, E5 and E7; in the second order in E2, E3 and E9; in the third order in the E6 and E7; and in the fourth order in the E5. The AMMI1 biplot (IPCA1 vs grain yield) identified G1, G3, G17 and G13 as high yielding and stable genotypes with seed yield higher than total mean and lowest IPCA1 values. This biplot was also indicated environments E1, E9, E5, E2 and E3 had the lowest contribution in genotype by environment interaction interaction. The first principal components explained only 42.5% of genotype by environment interaction, and therefore, it seems that using the AMMI2 biplot is more efficient to identify the superior genotypes. In AMMI2 biplot (IPCA1 vs IPCA2), genotypes G4, G3, G13, G1 and G10 had high general stability. The environments E3, E4 and E10 with long vectors, had high discriminating ability and can estimate the relative efficiency of genotypes well.

In general, based on different indices, G3, G1 and G13 had high yield in most of environments, and in most methods had good stability and could be candidates for introduction of new cultivars.