Study on malting barley genotypes under diverse Agroecologies of north western Ethiopia

The study was conducted with the objectives to determine the magnitude effect of genotype, environment, and their interactions on economically important traits and identify stable malt barley (Hordeum distichon L.) genotypes. Combined analysis of variance indicated that the main effects due to environment, genotype and GxE interaction were highly significant for grain yield and economically important malting quality triats indicated that development of both specific and wide adaptable varieties are essential. The GxE interaction of grain yield was further partitioned using AMMI and it showed the first two IPCA axes explained most of the sum of squares. According to stability analysis measures genotype G1 was the most stable whereas G13 showed specific adaptation in low potential environments. Protein content and seed size variability measures revealed G9 and G11 in protein content and G1 and G11 in seed size, respectively as the least varying genotypes across environments.


INTRODUCTION
Barley (Hordeum distichon L.) is an important grain crop in Ethiopia.It has diverse ecologies being grown from 1800 to 3400 m altitude in different seasons and production systems.In the highest altitudes, barley is grown as a sole crop.The total area covered by the crop is about one million hectares with a total annual production of 1.3 million tons (CSA, 2005).Northwestern high and mid altitudes belong to the major barley producing areas however all of the produce is used for food purpose.On the other hand, breweries have been setup in the country, which require lots of malt annually.Majority of their requirement is obtained from import.Since malting barley is a new crop to the northwestern Ethiopia, information is unavailable about cultivar performance across diverse environments.The performance of malting barley grain yield and quality characteristics depends greatly on environmental conditions, which results in differential expression of grain quality from environment to environment (Verme and Nagaragam, 1996).The genotype by environment interaction is another important aspect in affecting performance of genotypes across diverse environments.The GxE interaction arises when there is differential response of genotypes in environmental changes.It reduces the correlation between the genotype and phenotype, hindering the genetic potential of the cultivar (Kang and Gorman, 1989).Selection of stable genotype is described as one of the strategies to reduce GxE interaction effect.High yield stability refers to a genotype's ability to perform consistently, whether at high or low yield levels, across a wide range of environments.According to Finlay and Wilkinson (1963), yield response and stability are heritable and can be measured.Lin et al. (1986) have described nine parametric stability statistics.Among them, are conventional coefficient of variation (CV i ), variance of genotypes across environments (Si 2 ), Finlay and Wilkinson (1963) regression coefficient (b i ), Eberhart and Russell (1966) deviation from regression (S di 2 ), Shukla's (1972) stability variance ( 2 i  ) and Wricke's (1962) ecovalence (W i 2 ) are parametric stability measures of genotypes.
The objectives of this paper were to: i) assess phenotypic performance of malting barley genotypes across northwestern Ethiopia, ii) determine the magnitude of effect of genotype, environment, and their interaction on yield and quality traits, and iii) to identify stable genotype in grain yield and consistence performance in protein content and seed size across environments.

Data collected
Harvesting and data collection were done from the four central rows.Yield data was recorded on clean, dried samples and plot yields were adjusted to 12.5% moisture level and converted to kilogram per hectare.Thousand kernels were counted by using electronic seed counter and weighted (g).Every sample was measured in standard hectoliter and then weighted to know their hectoliter weight.Germination energy in percent was determined from 100 seeds germinated in a petridish after 120 h.Two hundred seeds were soaked in a flask with 0.3M H2O2 (hydrogen peroxide) and counting after 24 h and converted to percentage to determine germination capacity.Seed size test was carried out using 2.2, 2.5, 2.8 mm size sieves and proportion of the seed trapped by each sieves were weighted and converted to percentage.Eight gram samples from each plot were grounded and 5.0 g flour was placed in moisture dishes and was oven dried for 1 h at 100°C.Percent moisture was determined from the mass of water lost on drying to the original milled sample and kernel protein content was also determined using Kjeldahl method.

Statistical analysis
Analyses of variance were performed on all traits.Thereafter, com-Bantayehu 549 bined analyses of variance were performed using fixed linear model where both genotypes and environments were fixed.Mean separation was carried out using least significant (LSD) at 5% level of significant.The G x E interaction was further partitioned using additive main effects and multiplicative interaction (AMMI) statistical model.The AMMI analysis of variance summarizes most of the magnitude of genotype x environment interaction into one or few interaction principal component analysis (Zobel et al., 1988;Guach, 1988).The larger the IPCA scores, either negative or positive, the more specifically adapted a genotype is to a certain environments; the smaller the IPCA scores, the more stable the genotype is over all environments studied.AMMI analysis was performed for grain yield.The method of Eberhart and Russell (1966) was used to calculate the regression coefficient (bi), deviation from regression (Sdi 2 ) and coefficient of determination (Ri 2 ).It was calculated by regressing mean grain yield of individual genotypes on environmental index.Shukula stability variance ( Ecovalence (Wi 2 ) suggested by Wricke (1962) were also computed, where stability values with minimum values are considered stable.Variability of genotypes for protein content and seed size across environments were also measured by coefficient of variation (CVi) (Francis and Kannenberg, 1978) and genotypic variance across environmental (Si 2 ).

Grain yield
Mean grain yield of locations averaged over genotypes was between 3376.0 and 831.0 kgha -1 at E5 and E11, respectively (Table 2).The mean grain yield over all the genotypes and locations was 2178.7 kgha -1 . Moreover, performances of genotypes were not consistent across locations.G1, G12 and G13 won all the genotypes at all most all environments except E4, E5 and E6.At E4, G12 with grain yield of 4427.0 kgha -1 and at E5, G5 with 4614.7 kgha -1 were the top performing genotypes.G17 produced the least when genotypes are ranked according to their grain yield performance.When environments were compared, the highest mean grain yield (3376.0 kgha -1 ) was obtained at E5, while E11 (831.0Kgha -1 ) and E12 (1791.50Kgha -1 ) were relatively low yielding environments.The low grain yield at these environments that represent Dabat could be because of the water logging of the trial site.Moreover, relatively small amount of annual rainfall coupled with high disease pressure resulted in poor stand and low grain yield (Table 1).

Malting quality traits
Highest mean thousand kernel weight was recorded at Adet (40.85 g) and lowest mean at Dabat (37.42 g).Genotypes G13 (46.1 g), G11(45.9g), G3 (43.37) and G1(42.49g) provided the highest thousand kernel weight but genotype G2, G16 and G17 had low mean thousand kernel weight over locations with corresponding values of , respectively.The mean thousand kernel weight of most genotypes, but G2 and G17 fulfill the requirement; however, hectoliter weight of genotypes averaged over all locations except G7, G10, G14 and G15 did not meet the standards.Kernel protein content of genotypes averaged from all locations was between 9.25% for G19 and G20 and 11.0% for G2 (Table 3).Protein content for G2, G4 and G5 was high with 11.0, 10.5 and 10.5%, respectively.All genotypes gave acceptable mean kernel protein and met the standards set by the National Standard Authority for malting barley that should be between 9 and 11.5.Kernel protein content that exceeds recommended levels is undesirable for malting because it increase steep times and cause uneven water uptake during steeping, uneven germination during malting, increased malt loss due to abnormal growth, excessive enzymatic activity, low extract yield, excessive nitrogenous compounds in the wort during brewing, and chill haze formation in beer (Burger et al., 1979).Table 4 reveals that Adet (11.49) and Debretabor (10.45) were the locations where high protein content was recorded while the mean kernel protein contents obtained from Dabat and Laygaint was 8.35 and 8.82, respectively and the standard set by Ethiopian National Standard Authority was not achieved at these locations.This might be due to low fertility and low temperature.
G1, G11 and G19 had the large mean percentage of kernels trapped by 2.8, 2.5 and 2.8 + 2.5 mm size sieves (Table 3).From the tested genotypes, only G1 and G11 fulfills the standard greater than 80% of the kernel that passed through 2.5 + 2.8 mm size sieve.Genotype that had high percent of screen loss was G2.These genotypes had small kernel size much below the standard as observed in the test using 2.2 and <2.2 mm size sieve.Other genotypes provided kernel sizes in between these two ranges.Highest percentage of acceptable kernel size (92.5%) was recorded at Laygaint.Adet was the locations where higher kernel sizes were recorded next to Laygaint (Table 4).From the result it can be deduced that genotypes grown at these location meet the standard.It can be seen from Table 3 that G4 obtained germination energy of 99.75% followed by G1 (99.42%).The least germination energy was obtained from G9 and G3 (97%).All genotypes showed germination energy that meets the standard (> 95% in 120 h).The highest germination energy (99.55%) was observed from seeds harvested at Dabat because of early harvesting (Table 4).All genotypes had germination capacity above 97% that is the minimum requirement set by National Standard Authority.G4 was the top in germination capacity with 100% seeds germinated.

Genotype by environment interaction components
The combined analysis of variance revealed a highly significant (p<0.01)variation for the genotype, environment and genotype by environment interaction effects for grain yield, thousand grain weight and hectoliter weight (Table 5).This indicated genotype, environment and their interaction are important in governing the expression of these traits.It agrees with the finding that yield and quality traits are influenced by genotype (G), environment factors (E) and their interaction (Uhlen et al., 1998).The presence of variation between environments point out the presence of dissimilarity and provide a baseline information to categorize potential, low potential and intermediate yielding agro ecologies.The genotype effect depicted clear variation among tested genotypes thus wide and specific adaptation of genotypes is crucial for vertical increase of production.The significant GxE interaction indicated the differential genotypic performance across environments.It reduces the association between phenotypic and genotypic values, and thus, genotypes that perform well in one environment may perform poorly in another (Fox et al., 1996).
Generally, larger interaction component cause difficulties in selection of widely adapted, high yielding genotypes under diverse environments.The presence of significant G x E interaction and environment effects on yield traits were reported by Finlay and Wilkinson (1963) and Tesfaye et al. (1998) in barley.

Additive main effects and multiplicative interaction analysis for grain yield
Additive main effects and multiplicative interaction     (AMMI) analysis proved significant (P<0.05)main effects and interaction effects for grain yield (Table 8).It showed that 75.24% of the total sum of squares was attributable to environmental effects, 9.32% of genotypic effects and 15.44% to GxE effects (Table 7).A large sum of squares for environments indicated that the environments were diverse; with large differences among environmental means causing most of the variation in grain yield.The magnitude of the G x E sum of squares was highly significant (p<0.01)indicating that there were large differences in genotypic response across locations.
Results from AMMI analysis also showed that the first principal component axis (IPCA1) of the interaction captures 35.80% of the interaction sum of squares at 29 of the degrees of freedom (Table 6).Similarly, Purchase et al. (2001) and Romagosa et al. (1996) reported 41 and 72% of the G x E interaction explained by the first IPCA in wheat and barley.The second interaction principal component axis explained a further 22.08% of the GxE sum of squares and only 11.6% by the third IPCA axis.The mean squares for the IPCA1, IPCA2 and IPCA3 were significant at P = 0.01 and cumulatively contributed to 69.5% of the total GxE.F-test at P = 0.01 revealed that the first four principal component axes of the interaction were significant for the model.However, the prediction assessment indicated that AMMI 2 with only two interaction principal component axes was the best predictive model (Zobel et al., 1988).Further interaction principal component axes captured mostly noise and therefore did not help to predict validation observations.Thus, the interaction of the twenty genotypes with twelve environments was best predicted by the first two principal components of genotypes and environments and genotypes and environments with similar signs of their IPCA scores interact positively for that trait.AMMI 2 biplot as shown in Figure 1 has four sections: genotype G12, G19 and G13 were good for locations Adet: genotype G6, G18 and G10 were good for location Debretabor G2 and was good for Laygaint; and for Dabat genotypes G16 was good.Genotypes G1, G3, G9 and G4 located near the plot origin have low GxE interaction than the vertex genotypes and thus stable.Genotypes G5, G17 and G7 located far from the vertex were unstable over locations.

Stability analysis
Stability in performance of genotypes across environments using Eberhart and Russell's regression coefficient, deviation from regression, Wricke ecovalence and Shukula stability variance were performed for grain yield.According to Eberhart and Russell's model, geno-types performance across environments is generally expressed in terms of three parameters, mean yield, the regression coefficient ( i ) and the deviations from regression (s 2 d i ).A stable genotype should have a high mean yield, unit regression coefficient and deviation from regression nearly equals to zero (Finlay and Wilkinson, 1963;Eberhart and Russell, 1966).Accordingly, genotype G1 was the most stable because it had high grain yield, its regression coefficient was almost equal to unity and it had none significantly the least deviation from regression.In contrast, G6, G10 and G11 had Sd i 2 values significantly different from zero and  i greater than one that is, and below average stability, so were regarded as sensitive to environmental changes.Genotypes G2 and G17 had regression coefficients less than unity (i <1.0) that is, above average stability and significantly different deviation from regression.These genotypes are therefore, insensitive to environmental changes and have specifically adapted to low potential environments. ) showed that G8, G14 and G19 were comparatively stable as their contribution to the GxE interaction sum of squares was least and with minimum stability variance ( 2 i  ).On the other hand, G17 and G2 were unstable in grain yield performance because these genotypes had relatively higher ecovalence (Wi) and stability variance ( 7).
Kernel protein content and seed size variability of genotypes across environments using Francis and Kannenberg (1978) conventional coefficient of variation (CV i ) and variance of genotypes across environments (Si 2 ) was performed as indicated in Table 8.The model identified G15, G19 and G12 in kernel protein content and G4, G1 and G3 in seed size respectively showing relatively consistent performance across locations.Kernel protein content and seed size of these genotypes might have a relatively small variation on varied locations while grain protein content of G3, G9 and G11 and seed size of G1, G11 and G19 show considerable variation across environments.

DISCUSSION
Despite its potential and market demand, production of barley for malt is not expanded in northwestern Ethiopia.These could be attributed to the lack of information on the effect of genotype, predictable and unpredictable environmental variations and their interaction on yield and quality attributes of the crop.Thus, twenty genotypes were tested at four locations under rain-fed conditions in 2004, 2005 and 2007 main cropping seasons to assess phenotypic performance, determine the effect of genotype, environment, and their interaction and to identify stable ones in yield, consistent performance in protein content and seed size.Data were collected on grain yield and quality traits and analysis of variance (ANOVA), additive main effects and multiplicative interactions analysis (AMMI) and stability analysis were undertaken.The combined analysis of variance over locations has shown highly significant difference among genotypes, environments and genotype by environment interaction on grain yield.The effect of environment on this trait was greater than genotypes and genotype by environment interaction.Even though genotypes' performance ranking was not consistent across environments, most genotypes performed well in grain yield at Debretabor and Adet locations.Kernel protein content, thousand kernel weight and hectoliter weight of genotypes were promising at Debretabor and Adet.
Additive main effects and multiplicative interactions (AMMI) model analysis was used to partition the GxE interaction of grain yield.The first two principal component axes for interaction alone captured most of the interaction sum of squares.The sign and magnitude of IPCA scores revealed the relative contribution of each genotype and environment for the genotype and environment interactions and the biplot graph of AMMI scattered genotypes and environments based on their interaction (Zobel et al., 1988;Guach, 1988).It helped to summarize the pattern and magnitude of GxE interaction and main effects that reveal clear insight into the adaptation of genotypes to environments and helped enhanced understanding of GxE interactions.Stability analysis using Eberharts and Russell's coefficient of regression ( i ) (1966), deviation from regression (s 2 d i ), Wricke's (1962) ecovalence (W i ) and Shukula's (1972) stability variance ( 2 i  ) on genotypes performance across environments have shown similar results in identifying the stable genotypes.Relatively, G1 had better grain yield stability.G13 and G12 showed specificity in environments; however, a potential high yielder averaged over locations.G3, G11 had better consistent performance across environments in protein content and G1, G11 in seed size.Therefore, G1 could be used for production and further malting barley improvement programs in the sub-region.

Table 1 .
Total annual rainfall, soil type and altitude of the locations.

Table 2 .
Mean grain yield (Kgha -1 ) of twenty malting barley genotypes evaluated at four testing sites in the 2004, 2005 and 2007 main rain cropping season.

Table 4 .
Mean values of malting quality traits of malting barley genotypes grown at four representative locations in the 2004 main cropping season.

Table 5 .
Combined analysis of variance for twenty genotypes for yield and quality traits grown in 2004, 2005 and 2007 main rain cropping season.

Table 6 .
Additive main effects and multiplicative interactions (AMMI) analysis of variance for grain yield of 20 genotypes of malting barley across 12 environments.
** P< 0.01, ** Highly significant at the 0.01 probability level, *Significant at the 0.05 probability level, IPCA-Principal component axis for interaction.

Table 7 .
Mean grain yield and genotypic stability parameters for 20 malting barley genotypes grown over 12 environments. 2i

Table 8 .
Estimates of variability in performance for protein content and seed size for twenty malting barley genotypes at four environments.