Genotype x environment interaction and stability analysis for yield and yield related traits of Kabuli-type Chickpea ( Cicer arietinum L . ) in Ethiopia

Chickpea is the major pulse crop cultivated in Ethiopia. However, its production is constrained due to genotype instability and environmental variability. This research was carried out to examine the magnitude of environmental effect on yield of chickpea genotypes and to investigate the stability and adaptability of genotypes under different agro-ecologies. Seventeen (17) genotypes were evaluated in randomized complete block design (RCBD) with four replications in five locations. Various stability indices were used to assess stability and genotype by environment performances. Combined analysis of variance (ANOVA) for yield and yield components revealed highly significant (P≤0.01) differences for genotypes, environments and their interaction. The significant interaction showed genotypes respond differently across environments. At Akaki, Chefe Donsa, Debre Zeit, Dembia and Haramaya, top performing genotype were DZ-2012-CK-0001 (2933 kg/ha), Arerti (3219 kg/ha), Arerti (3560 kg/ha) DZ2012-CK-0013 (2675 kg/ha) and Arerti (2019 kg/ha), respectively. The first two PCs explained 74.45% of the variance. Based on ASV value, DZ-2012-CK-0002 were most stable genotypes. As per AMMI biplot, Arerti and DZ-10-4 were most widely adapted genotypes. Dembia and Haramaya were most discriminative environments for genotypes. Debre Zeit and Chefe Donsa were favorable environment for genotype. Genotypes DZ-2012-CK-0004, DZ-2012-CK-0010, DZ-2012-CK-0013, DZ-2012-CK-0007 and DZ10-4 are recommendable to Akaki, Chefe Donsa, Debre Zeit, Dembia and Haramya, respectively.

Kabuli type chickpeas are characterized by white-colored seed with ram's head shape, thin seed coat, smooth seed surface, white flowers, and lack of anthocyanin pigmentation on the stem.The plant is medium to tall in height, with large leaflets and white flowers.When compared with Desi types, the Kabuli types have higher levels of sucrose and lower levels of fiber.The Kabuli types generally have large sized seeds and receive higher market price than Desi types (Gaur et al., 2010).Chickpea seeds are eaten fresh as green vegetables, parched, fried, roasted and boiled as snack food, sweet and condiments (Dawar et al., 2007).Environmental factors such as soil moisture, sowing time, fertility and temperature and day length have strong influence during various stages of plant growth (Bull et al., 1992).The environment is changing day-by-day and this implies that it is necessary to evaluate crop genotypes at different locations to assess their performances.One approach to improve the chickpea yield is to identify stable genotypes that perform consistently better under diverse environments (Ghulam et al., 2012).The performance of a genotype is not always the same in different locations as it is influenced by environmental factors.To assess yield stability among varieties, multi-location trials with appropriate stability analysis method is required.Differences in genotype stability and adaptability to environment can be qualitatively assessed using the biplot graphical representation that scatters the genotypes according to their principal component values (Vita et al., 2010).In Ethiopia, there is no sufficient information on the genotype by environment interaction effects on yield and yield related traits of Kabuli-type chickpea.Therefore, the current research was undertaken to examine the magnitude of environmental effect on yield and yield related traits of Kabuli-type chickpea genotypes, to study the nature and extent of genotype by environment interaction on seed yield of Kabuli -chickpea genotypes and to investigate the stability and adaptability of the genotypes under different agro-ecological condition.

MATERIALS AND METHODS
The experiment was conducted during the 2012/13 main cropping season at five locations representing various chickpea growing agro-ecologies of Ethiopia.The environments were Akaki, Chefe Donsa, Debre Zeit, Dembia and Haramaya.Thirteen (13) pipelines and four released Kabuli-type chickpea varieties were included in the study (Tables 1 and 2).The plant materials were obtained from Debre Zeit Agricultural Research Center.Planting of the genotypes was done in early mid August up to first week of September using randomized complete block design with four replications at each site under rain fed conditions.Each genotype was planted in six rows of 4 m row length and at 1.2 m width.A spacing of 30 cm row to row distance and 10 cm plant to plants were used on a plot size of 4.8 m 2 .Fertilizer was not applied.Weeding and other management practice were done as required for each site.Data were recorded on days to 50% flowering, 90% physiological maturity, plant height, the number of pods per plant, the number of seeds per plant, 100-seed weight, biomass yield, grain yield, and harvest index.

Statistical analysis
Data were computed by using SAS 9.1.3for analysis of variance, Genstat13 th for biplot graph and Agrobase20 for stability analysis.

Performance of Kabuli-type chickpea genotypes for yield
Performance trials have to be conducted in multiple environments because of the presence of GE.For the same reason, the analysis of genotype by environment data must start with the examination of the magnitude and nature of genotype by environmental interaction (Ezatollah et al., 2011).Yield and its components are polygenic traits and are strongly influenced by environment in chickpea.Significant variation was observed for grain yield in Kabuli chickpea genotypes.Similar finding were reported by Khan et al. (1987Khan et al. ( , 1988).Bartlett's test showed homogenous error variance for the grain yield and allowed to proceed further pooled analysis across environments.
The combined analysis of variance (Table 3) for grain yield exhibited significant (P≤0.01)effects of locations, genotypes and genotype by environment interaction, indicating differences in environments and the presence of genetic variability among genotypes.The presence of significant genotype by environment interaction in chickpea was reported by various authors (Singh et al., 1990;Bozoglu and Gulumser, 2000).The overall mean yield of the location varied from 1469 to 2970 kg/ha (Table 4) and thus, the five environments showed wide variation in yield potential.The highest mean grain yield was obtained at Debre Zeit (2970 kg/ha) and the lowest was from Haramaya (1469 kg/ha).The possible reason was that late planting was done at Haramaya and due to this moisture stress occurred at vegetative and pod setting stage while relatively sufficient moisture was available at Debre Zeit.Genotypic means across the locations indicated that maximum mean grain yield across all the five locations in one year were obtained from DZ-2012-CK-0013 genotype (2635 kg/ha) and the minimum was from the local variety (1510 kg/ha).
Genotype by environment interaction causes differences in yield rank of genotypes in different locations; thus, it becomes important for the chickpea breeders in terms of selection efficiency and genotype suggestions for different locations.

Performance of Kabuli-type genotype for yield related traits
From the combined analysis of variance, the mean squares due to genotypes, environments and genotype by environment interaction were highly significant for the traits, days to flowering, days to maturity, plant height, number of pods per plant, hundred seed weight, above ground dry biomass and harvest index.However, there were no-significant effects of all these three source of variation on the number of seeds per pod (Table 3).The separate analysis of variance for all yield related traits, except for number of seed per pod at each location exhibited highly significant (P≤0.01)differences among Kabuli-type chickpea genotypes for the days to flowering, days to maturity, number of pods per plant, plant height, hundred seed weight, above ground dry biomass and harvest index at all locations.Similar results were reported by different researchers who worked on chickpea (Singh et al., 1990;Bozoglu andGulumser, 2000 andValimohammadi et al., 2007).The responses of genotypes in terms of all yield related traits were different both within and across locations.This indicated that the efficiency of a breeding program aimed at yield improvement is impaired due to genotype by environment interaction, which complicates the process of crop variety development especially when varieties are selected in one environment and used in others (Ahmad et al., 2011).

Days to flowering and maturity
The result reveals significant effects not only for genotypes but also for locations and genotype by environment interaction, variability in experimental material as well as difference in the environmental conditions (Table 3).Early flowering and early maturing genotypes were observed at Haramaya and Debre Zeit (47 and 108 days) and at the same time late flowering and mature genotypes were noted at Dembia and Chefe Donsa (64 and143 days), respectively (Table 5).The probable reason was due to high temperature and early cessation of rain at Haramaya and, relatively long rain season and low temperature at Chefe Donsa.Ejere and   6).

Number of pods per plant
Number of pods per plant is an important selection criterion for the development of high yielding genotypes and is strongly influenced by environment in chickpea (Malik et al., 1988).Marked variation was observed in the performance of genotypes over the five locations (Table 3).Number of pods per plant was highest at Haramaya (49) and least at Akaki (25) (Table 5).The genotypes mean values for number of pods per plant varied from 28 for DZ-2012-CK-0006 to 52 DZ-10-4.The highest mean number of pods per plant was recorded for genotypes Dz-10-4 (52) followed by Arerti, DZ-2012-CK-0010 (43), Habru (39) (Table 6).These results are consistent with the findings of Singh and Bains (1984) and Malik et al. (1988).These results indicate variability for number of pods per plant and its sensitiveness to environmental fluctuations.

Plant height (cm)
Significant effects were observed not only for genotypes but also for locations and genotype by environment interaction, reflecting genetic variability in experimental material as well as difference in the environmental conditions (Table 3).Averaged over all genotypes the highest plant height was recorded at Dembia (47 cm) and the shortest was Akaki (36 cm) (Table 5).Plant height was sensitive to environmental fluctuations and it indicated that the relative performance of genotypes was markedly inconsistent over the locations.Averaged over all locations the shortest genotype was Arerti (35 cm) and the longest genotype was DZ-2012-CK-0009 (49 cm) (Table 6).These results are consistent with the findings in chickpea of Malik et al. (1988) who also found high magnitude of genotype by environment interaction.

100-grain weight (g)
Statistically significant variance was observed for genotypes, location and genotype and environment interaction (Table 3).Over all genotypes hundred seed weight was highest for Debre Zeit (34.2 g) and lowest for Haramaya (28.3 g) (Table 5).In addition, the relative performance of genotypes is quite inconsistent across the environments.The genotype with the smallest 100-grain weight was DZ-10-4 (17 g) and the one with the highest was DZ-2012-CK-0006 (37 g) (Table 6).The significant pooled deviation for 100-grain weight suggested that these genotypes differ considerably with respect to their suitability for this character.The present results are in agreement with the findings of Singh and Singh (1974) and Sanghi and Kandakar (2001).

Above-ground dry biomass
Statistically highly significant variance was observed for genotypes, locations and genotype and locations interaction (Table 3).Averaged across all genotypes above ground dry biomass was highest for Dembia (2519 g) and lowest for Haramaya (834 g) (Table 5).

Wricke's ecovalence analysis
Wricke's ecovalence (Wi) was calculated for each of the 17 Kabuli-type chickpea genotypes evaluated at five diverse locations for one year in the major chickpea growing regions of Ethiopia (Table 7).The genotypes with the lowest ecovalence contributed the least to the genotype by environment interaction and are therefore more stable.

Eberhart and Russell's joint regression stability analysis
The mean square for genotype by environment significant was (p≤0.01) for grain yield (Table 3).This permitted the partitioning of genotype by environment effects in environment linear, G x E (linear) interaction effects (sum squares due to regression (bi) and unexplained deviation from linear regression (pooled deviation mean squares (S 2 di).The genotype by environment (linear) interaction was not significant indicating that the stability parameter 'bi' estimated by linear response to change in environment was the same for all genotypes or genotypes have the same slope (Table 8).Similar results were obtained in bean genotypes tested (Firew, 2003;Setegn and Habtu, 2003) in different part of Ethiopia and in Brazil (Ferreira et al., 2006).Our results reveal that the genotype by environment interaction was not a linear function of environment indices.The variation among the genotypes and for genotype by environment interaction were significant effects which means that genotypes exhibited different performances in different environments which is due to their different genetic makeup or the variation due to the environments or both.The mean sums of squares due to pooled deviation from regression were significant (p≤0.01) for grain yield indicating the importance of non linear genotype by environment.The most stable genotype with the lowest S th and 8 th , respectively.Therefore, these genotypes were best fit for specific adaptation in favorable environments where there were high levels inputs.If the mean yield, regression coefficient value (bi) and the deviation from the regression (S 2 di) are considered together simultaneously, there was no stable genotype.All genotypes had regression coefficients (bi) greater than one (that is, below average stability and significant deviation from regression).Therefore, these genotypes were specifically adapted to favorable environments (Table 9).

AMMI analysis of 17 Kabuli-type chickpea genotypes tested at five environments
The AMMI analysis of variance of grain yield of 17 Kabulitype chickpea genotypes tested in five environments is presented on Table 7.The analysis revealed that Kabulitype chickpea genotypes were significantly (P≤0.01)affected by environments (E), genotypes (G) and genotype by environment interaction.The main effects of E and G accounted for 53.1 and 11.9%, respectively, and G X E interaction accounted for 15.9% of the total variation of Genotype by environment data for grain yield (Table 10).The first two principal components (PC1 and PC2), which were used to create a two-dimensional biplot, explained 52.5 and 21.95% of AMMI sum of squares, respectively.According to the AMMI model, the genotypes which are characterized by means greater than grand mean and the IPCA score nearly zero are considered as generally adaptable to all environment (Ezatollah et al., 2013).However, the genotype with high mean performance and with large value of IPCA score are consider as having specific adaptability to the environments.The large sum of squares for environments showed that the environments were diverse, with large differences among environmental means causing most of the variation in grain yield.This is in synchronization with the findings of Singh et al. (1990), Yan (2002) and Yan and Tinker (2006) in chickpea production.This result also indicates the considerable influence of environments on the yield performance of Kabuli-type chickpea genotypes in Ethiopia.The magnitude of the genotype by environment sum of squares was more than two times that for genotypes, indicating that there were considerable differential genotype responses across environments.The AMMI I, biplot for grain yield of the 17 Kabuli-type chickpea genotypes at five environmental conditions is shown in Figure 1.The main effects (genotypes and environments) accounted for 65.02% of the total variation and IPCA 1 accounted for 52.5% of the total variation due to genotype by environment interaction alone.Environments showed high variation in both main effects and interactions (IPCA1) (Figure 1).Chefe Donsa and Debre Zeit are the most favorable environments; Haramaya and Dembia are the least favorable environments, while Akaki is the average environment.Environments are classified into three main groups based on their IPCA 1 scores Haramaya and Dembia are in quadrant I and have got large positive IPCA1 scores, which interact positively with genotypes that have positive      for its financial support.We are also indebted for the assistances provided by the research and technical staff of the highland Pulse Improvement Program of Debre Zeit Agricultural Research Center (DZARC) and Gonder Agricultural Research Center (GARC).

Figure 1 .
Figure1.AMMI biplot analysis of IPCA scores genotype and environment means for Kabuli-type genotypes.

Table 1 .
Geographic and environmental conditions of experimental area.

Table 2 .
List of Kabuli-type chickpea genotypes included in the experiment

Table 3 .
Mean sum of squares of yield and other traits from combined ANOVA of 17 Kabuli-type chickpea genotypes grown across five environments in Ethiopia.

Table 5 .
Mean values for yield related traits of Kabuli -type chickpea genotypes tested at five locations in Ethiopia (averaged over all genotypes).Grain yield, DF = days to flower, DM = days to mature, PPP= pod per plant, SPP= seed per pod, PHT=plant height, BM= biomass yield, HI= harvest index, HSW= hundred seed weight.

Table 6 .
Mean performance for yield related traits of 17 Kabuli-type chickpea genotypes grown at five environments.

Table 8 .
Sum of square and mean sum of squares from the analysis of variance for linear regressions of Kabulitype chickpea genotypes means on environmental index according toEberhart and Russell's joint  regression model (1966).

Table 9 .
Mean yield, regression coefficients (b), coefficients of determination and deviation from regression (S 2 di) of Kabuli genotype.
bi = Regression coefficients, r 2 i-coefficients of determination, S 2 dideviation from regression.SC = Standard check; LC = Local check.

Table 10 .
Additive Main effects and Multiplicative interaction (AMMI) analysis of variance for grain yield (kg/ha) of the 17 Kabuli-type genotypes tested across five locations.

Table 11 .
Yield and parametric stability statistics for grain yield on 17 Kabuli genotypes grown in five environments.Wricke´s ecovalence; S 2 di, deviation from regression; bi, regression coefficient; r 2 i, coefficient of determination; IPCA1 and IPCA2, interaction principal components axes 1 and 2, respectively; ASV, AMMI stability value; SC, standard check; LC, local check.