Genotype X Environment Interaction and Stability Analysis for Grain Yield of Bread Wheat (Triticum aestivum) Genotypes Under Low Moisture Stress Areas of Ethiopia

A multi-locations’ experiments were carried out from 2018 to 2019 main cropping seasons in moisture stress areas of Ethiopia to estimate the genotype x environment interaction and to select stable and adaptable variety/ies for grain yield of bread wheat. The genotypes consisted of 23 genotypes and two standard checks arranged in alpha lattice design replicated three times. Data were taken for agronomic traits and diseases. Analysis of variances and stability analysis were carried out for grain yield using R software. Combined analysis of variance showed a highly significant (p≤0.01) difference among the genotypes, locations, and GEI for grain yield suggesting a differential response of genotypes across testing environments. The grand mean yield over nine environments was 5251.90 kg ha-1 and the mean yield of genotypes across nine environments ranged from 1539.29 kg ha-1 in 2018 at Dhera to 7621.87 kg ha-1 in 2018 at Kulumsa, respectively. The recorded mean yield of the standard check Deka (5066.543 kg ha-1) and Ogolcho (4018.39 kg ha-1) was below the grand mean yield of genotypes across environments. The Genotypes ETBW 9136 (5731.79 kg ha-1), ETBW 9139 (5844.87 kg ha-1), ETBW 9646 (5754.01 kg ha-1), ETBW9172 (5634.01 kg ha-1), ETBW9641 (5545.03 kg ha-1), ETBW 9080 (5545.31 kg ha-1) and ETBW9396 (5467.04 kg ha-1) gave the highest mean grain yield across environments, whereas the standard check Ogolcho recorded lowest mean grain yield across environments. The first four principal components of the GEI explained 85.6% of the variation. Additive main effects and multiplicative interaction (AMMI) stability parameters revealed that the genotypes ETBW 9080 (G11), ETBW 9172 (G12), ETBW 9646 (G19), ETBW 9396 (G13), ETBW 9452 (G14), ETBW 9136 (G5) and ETBW 9139 (G6) were high yielder and more stable inferring little interaction of genotypes with the environment whereas Ogolcho (G25), ETBW 9119 (G3), ETBW 9647 (G20) and ETBW 9065 (G8) was low yielder and unstable suggesting high interaction with the environments. Based on stability parameters and other agronomic traits, the genotypes viz. ETBW 9396 (G13) and ETBW 9080 (G11), were proposed for variety verification and possible release in 2021.


Introduction
Wheat (Triticum spp.) is one of the most important cereal crops cultivated in Ethiopia. The production and productivity of wheat have increased by 14% due to the use of full packages for the last decade in Ethiopia [9]. The current wheat productivity is 2.7 tons per hectare but the potential yield could be more than 5.0 tons per hectare [3,13]. However, due to a higher increment in demands, about 37% (1.7 million tons) wheat grain deficit was observed during the 2018 budget year in Ethiopia (https://www.world-grain.com/articles/11880-ethiopiaswheat-production-to-increase/accessed in Nov. 2019). The yield gap observed could be attributed to different factors of which is lack of high yielder varieties sustainably tolerant to diseases and pests, tolerant to low pH and waterlogged areas and lack of use of full packages and knowledge of production in the farming communities. The Varietal differences in relation to Al tolerance exist in wheat; and the integrated use of all the available resources including acid tolerant and crop species, which improve and sustain soil and agricultural productivity, is of great practical significance [12]. Biotic factors viz. fungal diseases, rusts (Puccinia. spp.) are the devastating rust diseases which highly affecting wheat production of wheat in Ethiopia [18].
Multi-environment trials are important in plant breeding for evaluating genotypes for their overall stability and adaptability in the presence of genotype by environment (GE) interaction. An understanding of GE interaction is important at all stages of plant breeding, including ideotype design, parent selection, selection based on traits, including grain yield [19]. The G × E study is important in presence various agro-ecologies. Significant G × E interaction is a consequence of variations in the extent of differences among genotypes in diverse environments (called as a qualitative or rank changes) or variations in the comparative ranking of the genotypes (called as a quantitative or absolute differences between genotypes) [5,6,8].
Low moisture stress is becoming important in bread wheat producing areas of Ethiopia. In view of the problems, two product concepts (low moisture areas and high moisture areas) were designed in wheat breeding program at Kulumsa Agricultural Research Center and genotypes were selected based on the different agronomic traits viz. low sensitivity to low moisture stresses, early maturity and other agronomic traits. Therefore, a multilocations trials were designed with the objectives to estimate the magnitude of genotype by environment interactions; and to select best genotypes with stable and adaptable to the tested environments.

Experimental Sites' Descriptions
The experiments were conducted at nine environments viz. Dhera, Melkasa, Asasa, Kulumsa and Atsela during 2018 and at Dhera, Melkasa, Kulumsa and Asasa during 2019 cropping season. The descriptions of the locations are listed in Table 1.

Experimental Design and Data Management
Twenty-five genotypes including two standard checks were grown in Alpha-Lattice Design with three replications (Table 2). Each experimental unit consisted of six rows of 2.5 m length with 20 cm spacing between rows; 1 m and 1.5 m spaces were left between adjacent plots and replications, respectively. Data were collected from the entire plot for the parameters days to heading, days to maturity, grain yield, 1000 kernel weight, hectoliter weight, and from randomly sampled plants for the character; plant height. A seed rate of 125 kgha -1 was used at all locations. Fertilizers were applied based on the specific site recommendations and weeding were done uniformly to all plots in each location as necessary. Yellow rust (Puccinia striformisf. sp. tritici.) and stem rust (P. graminis) data, were recorded using the modified Cobb scale [15]. Disease severity was estimated which was used to determine the percentage of possible tissue infected and the maximum disease severity and reaction record were taken at each testing location for each genotype.

Statistical Analyses
Additive main effects and multiplicative interaction (AMMI) method integrates analysis of variance and principal components into a unified approach [2; 11]. AMMI method first fits the additive main effects of genotypes and environments by the usual analysis of variance and then describes the non-additive part, genotype by environment interaction, by principal component analysis. Data were subjected to analysis after checking for required assumptions of normality, homogeneity of variance using respective tests. The AMMI analysis was performed using the model suggested by Gauch and Zobel [11] for genotypes and environments is: Where, Yij is the mean yield of the i th genotype in the j th environment; µ is the general mean; gi is the i th genotypic effect; ej is the j th location effect; λn is the eigenvalue of the PCA axis n; and are the i th genotype j th environment PCA scores for the PCA axis n; is the residual; n' is the number of PCA axis retained in the model. Therefore, the interaction effect can be calculated as; AMMI Stability Value (ASV): ASV indicates the distance from zero in the two-dimensional plot of IPCA1 score against IPCA2 score in the AMMI model [16]. The ASV was calculated using the formula suggested by purchase [16] The larger the ASV value, either negative or positive, the more specifically adapted a genotype is to certain environments. Smaller ASV values indicate more stable genotypes across environments [16].
Where; SS IPCA1 is sum of square of interaction principal component 1 and SS IPCA2 IS sum of square of interaction principal component 2.
Yield Selection Index (YSI): Stability is not the only parameter for selection as most stable genotypes and would not necessarily give the best yield performance. Therefore, based on the rank of mean grain yield of genotypes (rYi) across environments and rank of AMMI stability value (rASVi). A genotype with the least YSI is considered as the most stable [7]. Yield selection index (YSI) was calculated for each genotype as: Eberhart and Russell regression model: The regression coefficient (bi) and deviation from regression (S 2 d) of genotype mean across environments index were computed as suggested by [4]. The analysis was performed using R statistical software.

Additive Main Effects and Multiple Interaction (AMMI) Model for Grain Yield
Combined analysis of variance showed that highly significant (p≤0.01) differences among the genotypes, environments, GEI and the first four Principal Components (PCs) for grain yield suggesting differences among environments and differential response of genotypes across testing environments ( Table 3). The first four interaction principal components (IPCA) of the GEI explained 82.1% of the total variation and 40.5% is explained by IPCA1 followed by 16.7%, 15.2%, and 9.7% for IPCA2, IPCA3, and IPCA4, respectively. The most accurate model for AMMI can be predicted by using the first two PCAs [10]. In this, the first two IPCAs were used to show genotype by environment interaction and placement on the biplots (Figure 1). The greater the IPCA scores (positive or negative) as it is a relative value, the more specifically adapted a genotype is to certain environments. The more IPCA scores approximate to zero, the more stable the genotype is across environments sampled [16; 1]. Accordingly, the genotypes ETBW 9080 (G11), ETBW 9172 (G12), ETBW 9646 (G19), ETBW 9396 (G13), ETBW 9452 (G14), ETBW 9136 (G5) and ETBW 9139 (G6) were relatively located near to the origin of the biplot and hence better stable (little interaction of genotypes with the environment) and widely adaptable genotypes across environments. Whereas the genotypes Ogolcho (G25), ETBW 9119 (G3), ETBW 9647 (G20) and ETBW 9065 (G8) far from the origin of biplot suggesting less stability (high interaction of genotypes with the environments) and have specific adaptability to certain environments.

Mean Grain Yield Performances of the Genotypes Across the Tested Environments
The mean grain yield of the 25 genotypes showed a wide range of variation in the different environments. The grand mean yield over nine environments was 5252 kg ha -1 and the mean yield of genotypes across nine environments ranged from 1539 kg ha -1 at Dhera in 2018 to 7622 kg ha -1 at Kulumsa in 2018, respectively. The recorded mean yield of the standard checks; Daka (5067 kg ha -1 ) and Ogolcho (4018 kg ha -1 ) were below the grand mean yield of genotypes across environments implying the possibility to select superior candidate genotypes better than the st. check varieties. The Genotypes ETBW 9136 (5732 kgha -1 ), ETBW 9139 (5845 kg ha -1 ), ETBW 9646 (5754 kg ha -1 ), ETBW9172 (5634 kgha -1 ), ETBW9641 (5545 kgha -1 ), ETBW 9080 (5545 kgha -1 ) and ETBW9396 (5467 kgha -1 ) gave the highest mean grain yield across environments, whereas the standard check Ogolcho recorded lowest mean grain yield across environments (Table 4).

Mean Performance of the Genotypes for Other Important Agronomic Traits
A mean for days to heading of genotypes was ranged from 57 to 65 days with an average value of 60.11 days which indicates almost all genotype had the narrow range of heading dates. Similarly, there was little difference between genotypes for days to maturity confirming that the tested genotypes can be categorized under similar maturity groups. Plant height varied from 82 to 93 cm with minimum values in genotypes ETBW 9651 and the maximum was for ETBW 9646. The mean 1000 kernel weight ranged from 28 g (ETBW 9396 and OGOLCHO) to 38 g (ETBW 9545) with an average value of 32.75 g. The genotype ETBW 9545 (22.5 % and 35%), ETBW 9647 (19% and 32%) ETBW 9080 (19% and 32%) had TKW advantage than the st. checks DEKA and OGOLCHO, respectively. Hectoliter weight provides a rough estimate of flour yield potential in wheat and is important to millers just as grain yield is important to wheat producers. The value the trait ranged from 64 kg/hl (OGOLCHO) to 69 hg/hl (ETBW 9136, ETBW 9080, ETBW 9646 and ETBW 9651) ( Table 6).

Severity and Response of Genotypes to Yellow and Stem Rust
The level of severity values and response were slightly different at each location and years indicating that the level of the rust's disease severity is dependent on the suitability of the environments. As was indicated by the Yield selection index the genotype, ETBW 9646, ETBW 9396, and ETBW 9172 were stable and top-yielding genotypes. But except ETBW 9396 the top-ranking and stable genotype have weak resistance to yellow and stem rust when the environment is appropriate for the occurrence of the disease. The degree of wheat susceptibility to yellow rust varied across locations due to variation in virulence spectra of the pathogen and climatic conditions for the pressure [17]. The highest yellow and stem rust disease score observed in the 2019 cropping season at kulumsa indicating this environment was highly suitable for buildup of both yellow and stem rust disease (Table 7). Among the studied genotypes relatively showed the best stability and score high mean grain yield; the genotype ETBW 9080 and ETBW 9396 were relatively recorded low yellow rust and stem rust score over location and years. These two genotypes were selected and proposed for variety verification and possible release in 2020.  1  DEKA  1MRMS  5MSMR  5MSMR  15MSMR  15SMS  30MSS  50S  15MSMR  2  ETBW 9116  0  1MS  5MSMR  5MSMR  10SMS  5MSMR  40S  1MR  3  ETBW 9119  5MS  10MSS  10MSS  10MR  15S  60S  50S  10MRMS  4  ETBW 9128  5MSMR  5S  5MSMR  60S  20S  50S  50S  5MS  5  ETBW 9136  1MR  5S  5MSMR  40MSS  15SMS  5MSMR  50S  5MRMS  6  ETBW 9139  5MR  10MSS  10MSMR  20MSS  154SMS  60S  50S  15S  7  ETBW 9149  5MR  1MS  1MSMR  5MR  5MSMR  5MSMR  40S  5MS  8  ETBW 9065  1MR  1MS  5MR  30MS  15SMS  60S  30S  1MR  9 ETBW

Conclusion
Due to changes in climatic factors and/or weather variabilities, most of the wheat producing regions in Ethiopia became low moisture stress areas. Hence, there is a need to develop varieties which could adapt to this short growing cycles. In view of this 25 bread wheat genotypes including two st. checks were evaluated in 9 environments from 2018 to 2019 cropping seasons with the objectives to estimate the magnitude of genotype by environment interaction and to select stable and adaptable varieties across the tested environments. The combined analysis of variance showed a highly significant (p≤0.01) differences among the genotypes, environments, and GEI for grain yield suggesting a differential response of genotypes across testing environments. The grand mean yield over nine environments was 5252 kg ha -1 and the mean yield of genotypes across nine environments ranged from 1539 kg ha -1 in 2018 at Dhera to 7622 kg ha -1 in 2018 at Kulumsa. The genotypes ETBW 9136 (5732 kg ha -1 ), ETBW 9139 (5845 kg ha -1 ), ETBW 9646 (5754 kg ha -1 ), ETBW9172 (5634 kg ha -1 ), ETBW9641 (5545 kg ha -1 ), ETBW 9080 (5545 kg ha -1 ) and ETBW9396 (5467 kg ha -1 ) gave the highest mean grain yield across environments, whereas the standard check Ogolcho recorded lowest mean grain yield across environments. The first four principal components of the GEI explained 85.6% of the total variations. Additive main effects and multiplicative interaction (AMMI) stability parameters revealed that the genotypes ETBW 9080 (G11), ETBW 9172 (G12), ETBW 9646 (G19), ETBW 9396 (G13), ETBW 9452 (G14), ETBW 9136 (G5) and ETBW 9139 (G6) were high yielders and more stable inferring little interaction of genotypes with the environment whereas Ogolcho (G25), ETBW 9119 (G3), ETBW 9647 (G20) and ETBW 9065 (G8) was low yielder and unstable suggesting high interaction with the environments. Based on the results of grain yield stability parameters and other agronomic traits including diseases resistances, the genotypes viz. ETBW 9396 (G13) and ETBW 9080 (G11), were selected for candidate varieties for variety verification and possible release in 2021 for low moisture stress areas of Ethiopia.