PARAMETRIC STABILITY ANALYSIS OF DURUM WHEAT YIELD (TRITICUM DURUM DESF) ENVIRONMENTAL STABILITY AND PLASTICITY OF SPRING BARLEY CULTIVARS

The study purpose was to assess the phenotypic stability of 23 durum wheat genotypes using different stability parameters to identify both high-yielding and stable genotypes. Materials and methods . The study was carried out at 4 trial sites differing in soil and hydrothermal conditions in 2008/09 - 2009/10. To quantify the yield stability, 7 statistical parameters of the stability (bi, Pi, ASVi, CVi, S2di, S2i, and W2i) were calculated. Results and discussion . The grain yields of all the genotypes were significantly affected by growing conditions, which accounted for 88.2% of the total variance of the yield, while the contributions of the genotype and genotype-environment interactions only amounted to 2.9% and 8.9%, respectively. There were significant positive correlations between the average yield of the genotypes under investigation and the regression coefficient (bi) and between the average yield of the genotypes and the environment variance (S2i). Correlation analysis also sepa-rated Pi, bi, and S2i approaches that correlated with the average yield and ASV, W2i, and S2di approaches that evaluate the phenotypic stability of the genotypes regardless of the yield. Conclusions . The results show that the genotypes Bel, Amg, Miki, Bss and Msb were the most stable by the majority of the statistical models used. Miki, Amg and Msb were distinguished as the best genotypes combining high yield and high stability under various conditions. Avhur, and Balzam gave the most stable yields. The yields of cultivars Vziretz, Grin and awnless Modern were the most variable, i.e. these cultivars are plastic. Thus, the Plant Production Institute named after V.Ya. Yuriev of NAAS created barley cultivars for different growing conditions: both for regions with optimal conditions and for arid and risky farming regions. This is relevant, given possible climate changes towards warming. Purpose and onjectives. To distinguish stable and plastic varieties under contrast weather conditions of the eastern forest-steppe of Ukraine, to evaluate the weather conditions of the study years as environments with informative and differentiative capacity. cultivar test cultivar research crop rotation fields of the Plant Production Institute NAAS, in the competitive trial nurseries in four replicates, with of 10 m 2 The yield statistically ANOVA. The informative and differentiating capacity of the environments (years), as well as the breeding value of the cultivars, determined by GGE biplot. are intense varieties. The most stable yields were given by varieties Alehro, Parnas, Avhur, and Balzam. The decrease in the yields of these varieties under unfavorable weather conditions was 55–65%. The yields of varieties Vzitets, Hrin and awnless variety Modern were the most variable, with a decrease in their yields of 66–72%, i.e. these varieties are plastic. At the same time, Vzirets can reach its potential even in arid conditions. Thus, the Plant Production after created barley cultivars for different growing conditions: both for regions with optimal conditions and for arid and risky farming regions. This is relevant, given possible climate changes towards

For many years, Myronivka Institute of Wheat has been studying the yields and stability of local and foreign cultivars. They identified cultivars with high homeostatsity, breeding value and yields. Hudzenko V.M. et al. conducted a systemic evaluation using a number of parametric and non-parametric statistical tests as well as AMMI and GGE biplot and found that cultivars differed in their responses to the contrast conditions of the study years, and, accordingly, they will complement one another in production, provided a proper selection of cultivars [28,29,30].
The Plant Production Institute named after V.Ya. Yuriev of NAAS obtained positive results in the determination of stability by testing cultivars in different zones. Solonechnyi P.M. et al. [31,32,33,34,35,36,37,38] investigated domestic cultivars in the forest-steppe and steppe. The used GGE biplot and AMMI to analyze variations in the yield capacity, identified "ideal genotypes" as well as stable and plastic cultivars, and determined the degrees of influence of the genotype and the environment on fulfillment of the potential. The test material was studied in the research crop rotation fields of the Plant Production Institute named after V.Ya. Yuriev of NAAS, in the competitive trial nurseries in four replicates, with the plot area of 10 m 2 . The cultivation technology was typical for the zone.
The yield data were statistically processed by ANOVA. Informative and differentiating ability of the environment (years) as well as the breeding value of cultivars were determined by GGE-biplot [39,40].
The GGE-biplot graphs were constructed using the first two principal components (PC1 and PC2) obtained from data processing via singular value decomposition. The model only maintains two principal components, since it is better for detecting patterns and allows easy display of PC1 and PC2 on a two-dimensional biplot so that the interaction between each genotype and each environment can be visualized.
The basic model for the GGE biplot is as follows: Y ij -μ -β j = λ 1 i1 η j1 + λ 2 i2 η j2 + ε ij (1), whereY ij -mean yield of genotype i in environment j, -grand mean, jmean yield of all the genotypes in environment j, l and 2singular values (SVs) of the 1 st and 2 nd principal components (PC1 and PC2), i1 and i2eigenvectors of genotype i for PCl and PC2, respectively, 1j and 2jeigenvectors of environment j for PCl and PC2, respectively, ilresidual associated with genotype i in environment j. To generate a biplot, the aforecited model was transformed as (2): Y ij -μ -β j = g i1 e 1j + g i2 e 2j + ε ij (2), whereg i1 e 1j andg i2 e 2j -PCl and PC2 scores for genotype i and environment j, respectively. In a biplot, genotype i is displayed as a point defined by all g i values, and environment j is displayed as a point defined by all e j values.
To construct a GGE biplot, we used Genstat12 software. Results and discussion. Analysis of the weather in 2008-2015, in particular the average daily temperature, the sum of effective temperatures and the precipitation amount at the study location, gave the results summarized in Table 1.  We observed in a decline in the yields in all the genotypes under the unfavorable conditions (drought) ( Table 2).
In experiment 1, the most drastic decrease in the yields was recorded in standard Vzirets (66%), while Ahrarii, Schedryi and Alehro showed a relatively small decline in the yields (54-55%). In experiment 2, cultivar Hrin was the most responsive (60%) to the growing conditions (see Table 2), while Modern was the least responsive (47%). Basing on these data, it is difficult to assess the stability of the cultivar yields; therefore, we used the GGE biplot analysis. 0,08 *the yield is significantly higher than the standard The GGE biplot analysis allows visual assessment of the discriminating and representative capacity of the environment as a tester for assessing genotypes. The principal components PC1 and RS2 (for the genotype and year conditions) account for 87.02% and 96.10% of the total variability caused by the genotype-environment interaction in experiments 1 and 2, respectively ( Fig. 1-4).
The environment eigenvectors are proportional to the standard deviation of the genotype yields in the corresponding environment. Accordingly, the environments with long eigenvector are highly discriminatory, but if the marker of a tester environment is close to the biplot center, i.e. the environment's eigenvector is short, all the genotypes in the biplot are close to one another, thus, this environment is non-informative. In experiment 1, arid 2010 was non-informative: the genotypes were unable to exploit their potentials under such conditions. The favorable 2008, on the contrary, had a very high discriminating capacity (Fig. 1a). In experiment 2, all the years were equally informative (Fig. 1b). The smaller the angle between the environments' eigenvectors is, the greater the correlation between them is. Accordingly, in experiment 1 there is a close correlation between 2009 and 2011, while in experiment 2 the correlation is weak (see Fig. 1).
The "what-won-where" polygon view of the GGE biplot (showing where which of the genotypes wins) visualizes the patterns of the genotype-environment interaction (Figure 2). The polygon vertices are the markers of the genotypes that are as maximally removed from the biplot center, while the markers of the other genotypes are inside the polygon. The lines dividing the bipolt into sectors are a set of hypothetical environments. The genotype that is at the polygon vertex is most productive in the environments inside this sector. Another important feature of biplots is a possibility of grouping environments into a mega-environment. Thus, in experiment 1, the mega-environment is formed by E2 (2009) and E4 (2011), between which there is a close correlation. Cultivar Ahrarii (G3) fell in this meg-environment's sector (Fig. 2a). This means that the conditions of these years were optimal for this cultivar. Parnas (G2) exploited its potential under the favorable conditions in 2008 (E1). For cultivar Vzirets (G1), 2010 was the best (E3), despite the aridity of this year, and for awnless cultivar Schedryi (G7), the conditions in 2008-2011 were not favorable (see Fig. 2a).
In experiment 2, the mega-environment consists of E2 (2014) and E3 (2015), which had favorable for barley conditions. Cultivar Avhur (G5) was in this mega-environment's sector (Fig.  2b). Cultivar Modern (G2) fulfilled its potential under the dry conditions of 2013 (E1), and for cultivar Grin (G3), the conditions in 2013-2015 were not favorable (see Fig. 2b). GGE biplotting allows one to rank genotypes by average yields and stability in different environments (Fig. 3). The average environmental coordinate (AEC) (axis X) or the yield line is drawn through the origin of coordinates as an arrow indicating its positive end and ranks the genotypes by yield. The AEC Y axis, or the stability axis, is drawn through the origin of coordinates perpendicularly to the AEC X axis. The average yields of the genotypes are estimated by projecting their markers on the AEC X axis. In experiment 1, cultivars Ahrarii (G3), Parnas (G2) and Alehro (G5) had the highest average yields, and cultivar Schedryi (G7)the lowest. The yield of cultivar Vzirets (G1) was the most variable, while cultivars Alehro (G5), Parnas (G2) and Modern (G4) were highly stable (Fig. 3a).
In experiment 2, cultivars Avhur (G5) and Balzam (G4) combined the highest yields with the greatest stability, while Grin (G3) and Modern (G2) were the least yielding, and moreover, the least stable ones (Fig. 3b). GGE biplotting allows one to rank genotypes by -breeding value.‖ The center of concentric circles is the "ideal" genotype's position. The closer genotype to the "ideal" is, the more valuable it is (Fig. 4).