Stability analysis for yield and yield contributing characters in hybrid maize ( Zea mays L . )

Twenty one hybrids along with two check varieties of maize were assessed for genotype environment interaction (GEI) and stability for selection of the best hybrid maize in three different Agro-Ecological Zones (AEZ) of Bangladesh during rabi season 2014/2015. The experiment was laid out in Randomized Complete Block Design with 3 replications. The additive main effects and multiplicative interaction (AMMI) model was used to analyze the genotype-environment interaction over three locations to select the hybrids having higher yield (yld) and other potential attributes. The mean sum of square for genotypes was highly significant for the studied characters. Similarly, environmental variances were also highly significant for all characters. Variances due to genotype × environment interaction were highly significant for cob length and thousand seed weight (TSW). AMMI Component 1 showed variation for TSW. But AMMI Component 2 and G×E (Linear) revealed insignificant variation for all the characters. The highly significant effects of environment indicated high differential genotypic response across the different environments. The environments of Gazipur (Ij=-1.42**) and Barisal (Ij=-0.068) were poor but Rangpur (Ij=1.49)) was a favourable environments due to positive environmental index for tested maize hybrids. Considering the mean, bi and S 2 di for all the parameters, it was evident that all the genotypes showed different response of adaptability under different environmental conditions. Among the hybrids , WL4×WL5, WL1×WL3, WL2×WL3, WL1×WL4 and the check NK 40 exhibited the higher grain yield; bi~1 and S 2 di~0 indicated that these hybrids were stable across the environment. All the hybrids showed insignificant values for regression co-efficient and also deviation from regression except WL1×WL5. The AMMI biplot showed four grouping of genotypes having none of them, low yielding and unstable; one hybrid was low yielding but moderately stable; eight were high yielding and stable hybrids, and fourteen were high yielder but highly unstable.


INTRODUCTION
Maize (Zea mays L.; 2n=20) is a monoecious, C4 plant which belongs to the tribe Maydeae of the family Poaceae.It is a tall, robust, annual, usually with a single dominant stem, although there may be few tillers in some genotypes and environments.Its leaves are districhous (two ranks of single leaves borne in alternate position) with overlapping sheaths and relatively long broad leaves.
Maize is a crop with versatile nature.It grows in a wide range of landscapes and agro-ecological settings across the world.This crop has brought an agricultural revolution in the United States (U.S.) and is the first crop in the world that had high-yielding varieties within a decade of the Mendelian discovery in 1900 (Dasgupta, 2014).
In Bangladesh, although maize can be grown in both kharif (monsoon) and rabi (winter) seasons but the potentiality of realizing very high yields is possible only during the rabi season.In kharif cultivation, farmers face various problems such as waterlogging, high infestation of diseases and pests, etc. Due to this reason it is now extensively being grown in rabi season under irrigated conditions.However, kharif cultivation is also possible in some suitable areas; the area of growth is now gradually increasing mainly due to T. Aman-potato-maize cropping pattern (Mondal et al., 2014) Maize in Bangladesh is becoming an important crop in the rice based cropping system.It is the third important cereal crop after rice and wheat.In recent year's maize is gaining popularity among the farmers mainly due to high yield, more economic return and versatile uses; it is the highest yielding grain crop having multiple uses.The need for demand of maize is increasing gradually.The area and production of maize is increasing day by day in Bangladesh and it continues to expand rapidly at an average rate of 20% year -1 (CIMMYT, 2008).Stability refers to the consistency of phenotype in varying environment.It is one of the most important properties of a genotype to be released as a variety for wide cultivation.For quantitative traits like yield the relative performance of different genotypes often varies from one environment to another.Thus, genotype×environment interaction does exist when phenotypic response made by a change in environment is not the same for all genotypes (Comstock and Moll, 1963).Stability analysis helps to select genotypes adaptable for wide and specific environments and divides large geographical areas into subareas.It provides effective allocation of resources for testing genotypes across location years.
A commercial variety must have stable performance and broad adaptation over a range of environments in addition to high yield potential.But its evaluation becomes complicated due to genetic heterogeneity, a complex biological basis, and genotype × environment interactions (Austin and Lee, 1988).Thus, genotype × environment interaction limits the effectiveness of selection when selection is done based only on mean yields (Dehganpour and Moghadam, 1999).The potential of genotypes and stability of their performance can be judged by multi environment testing (Mahajan and Khehra, 1992).It is more practical to develop and release varieties which are adapted to more than a single environment and can be successfully grown over a range of environments.
The analysis of G×E, therefore, becomes an important tool employed by breeders for evaluating varietal adaptation and also for selecting parents for base populations.The additive main effects and multiplicative interaction (AMMI) model was found suitable to handle both the main effects and G×E interactions in multilocational yield trials more effectively and efficiently than other statistical packages (Gauch, 1993).Considering these the experiment was laid out with the following objectives: (i) To estimate G×E interaction of single cross hybrids and (ii) To identify stable single cross hybrids for yield and yield related characters.
Seeds of each entry were sown in 2 rows, 4 m long plots with 60 and 20 cm spacing between rows and hills, respectively.Seeds *Corresponding author.E-mail: quamrul_islam76@yahoo.com.
Author(s) agree that this article remain permanently open access under the terms of the Creative Commons Attribution License 4.0 International License were sown at Gazipur on 24 November, Barisal on 27 November, Rangpur 1 December.One healthy seedling per hill was kept after thinning.Fertilizers were applied at the rate of 250, 55, 110, 40, 5 and 1.5 kg/ha of N, P2O5, K2O, S, Zn and B, respectively.Standard agronomic practices were followed (Quayyum, 1993) and plant protection measures were taken as required.Two border rows at both end of each replication were used to minimize the border effect.Data on cob length (CL), cob diameter (CD), thousand seed weight (TSW) and yield (t/ha) were recorded.All the plants in two rows were considered for plot yield and converted to t/ha.The analysis of variance (ANOVA) was used and the GE interaction was estimated by the AMMI model (Zobel et al., 1988).In this procedure, the contribution of each genotype and each environment to the GE interaction was assessed by use of the biplot graph display in which yield means were plotted against the scores of the first principal component of the interaction (IPCA1).The computational program for AMMI analyses was supplied by Duarte and Vencovsky (1999).The stability parameters, regression coefficient (bi) and deviation from regression (S2 di) were estimated according to Eberhart and Russel (1966).Significance of differences among bi value and unity was tested by t-test while between S 2 di and zero by F-test.All the data were processed and analyzed using Cropstat 7.2 program and PB Tools.

Analysis of variance
Twenty one single cross maize hybrids and two check hybrids [BARI Hybrid Maize7 (BHM 7) and NK 40] were evaluated in three AEZ for four characters.The combined analysis of variance for four characters was presented in Table 1.The mean square for genotypes was highly significant for the characters studied.Similarly, environmental variances were also highly significant for all characters.Variances due to genotype × environment interaction were highly significant for cob length and TSW.AMMI Component 1 showed variation for TSW.But AMMI Component 2 and G×E (Linear) revealed insignificant variation for all the characters.The pooled deviation (nonlinear portion of variance) which is unpredictable portion of G×E interaction was significant for only TSW.The highly significant effects of environment indicated high differential genotypic response across the different environments.Environment relative magnitude was much higher than the genotypic effect, suggesting that performance of each genotype was influenced more by environmental factors.Significant differences in Genotype (G), Environments (E), Interaction G×E and Pooled deviation of maize were also recorded in the studies of Shiri (2013), Patel and Kathiria (2016), Banik et al. (2010), Lata et al. (2010), Miah et al. (2011), Islam et al. (2011), Rahman et al. (2010), Fan et al. (2007), Akbar et al. (2009) and Kaundal and Sharma (2006).

Cob length (cm)
Cob length (cm) along with the value of phenotypic indices (Pi), regression coefficient (bi) and deviation from regression (S 2 di) were presented in Table 2.The environmental mean and genotypic mean ranged from 18.91 to 28.9 and 19.29 to 28.15 cm, respectively.
Thirteen hybrids including one check (BHM 7) showed positive phenotypic index while the other ten hybrids including the check (NK 40) had negative phenotypic index for CL.Thus, positive phenotypic index represented higher yielder and negative index represented lower yielder hybrids.
The regression coefficient (bi) and deviation from regression (S 2 di) values of these hybrids ranged from 0.67 to 1.36 and 0.00 to 0.15, respectively.

Cob diameter (cm)
Cob diameter (cm) along with the value of phenotypic indices (Pi), regression coefficient (bi) and deviation from regression (S 2 di) were presented in Table 3.The environmental mean and genotypic mean ranged from 4.10 to 4.95 and 3.83 to 4.75 cm.The highest overall mean for CD was recorded in WL3×WL4 and WL4×WL7 (4.75 cm) followed by WL2×WL3 (4.69 cm), WL4×WL6 (4.66 cm) and WL4×WL5 (4.56 cm) and WL1×WL3 (4.56 cm) higher yields of the hybrids.The lowest CD was found in WL1×WL5 (3.83 cm).Twelve hybrids including two checks showed positive phenotypic index while the other eleven hybrids had negative phenotypic index for CD.Thus, positive phenotypic index represented higher yield and negative index represented the lower yield hybrids.
Among the twelve, six hybrids such as WL4×WL7, WL3×WL4, WL2×WL3, WL3×WL6, WL1×WL3 and WL1×WL4 have positive phenotypic index, coupled with near unit regression co-efficient and non-significant deviation from regression recorded stable.Though the hybrid WL4×WL6 had positive phenotypic index, deviation from regression (S 2 di)=0, but the regression coefficient (bi) was significant, so it was the unstable one according to Eberhart and Russel (1966).
Again, positive and negative environmental index (Ij) reflected the favourable environment and unfavourable environment for this character, respectively.The environment Barisal (Ij = 0.47) was favourable and Gazipur (Ij =-3.80) and Rangpur (Ij =-0.86) were unfavourable environments for this character of hybrids.The regression coefficient (bi), deviation from regression (S 2 di) values of these hybrids ranged from 0.53 to 1.41 and 0 to 0.03, respectively.

Thousand seed weight (TSW)
The thousand seed weight (TSW) along with the value of phenotypic indices (Pi), regression coefficient (bi) and deviation from regression (S 2 di) were presented in Table 4.The environmental mean and genotypic mean ranged from 338.2 to 397.9 and 330.11 to 405.89 g.
Eleven hybrids including one check NK 40 showed positive phenotypic index while the other eleven genotypes had negative phenotypic index for yield.Thus, positive phenotypic index revealed the higher TSW and negative index represented the lower TSW among the genotypes.Again, positive and negative environmental index (Ij) reflected the rich or favourable and poor or unfavourable environments for this character, respectively.Thus the environment of Gazipur and Barisal were rich whereas Rangpur was negative environments for this character.Barisal was highly suitable for hybrid maize cultivation followed by Gazipur.
The regression coefficient (bi) deviation from regression (S 2 di) values of these genotypes ranged from -0.02 to 2.04 and 33.11 to 10457.82,respectively.These differences in bi values indicated that all the genotypes responded differently to different environments.Considering the mean, bi and S 2 di three parameters, it was evident that all the genotypes showed different response of adaptability under different environmental conditions.Among the hybrids , WL3×WL4, WL3×WL7, WL5×WL7, WL1×WL2, WL1×WL7 and check NK 40 exhibited the higher TSW, bi~1 and S 2 di~0 indicating that the hybrids were stable across the environment according to Eberhart and Russel (1966).

Grain yield (t/ha)
The grain yield along with the value of phenotypic indices (Pi), regression coefficient (bi) and deviation from regression (S 2 di) were presented in Table 5.The environmental mean and genotypic mean ranged from 7.58 to 10.48 t/ha and 3.94 to 10.49 t/ha, respectively.
Thirteen hybrids showed positive phenotypic index while the other eight genotypes had negative phenotypic index for yield.Thus, positive phenotypic index represented the higher yield and negative index represents the lower yield among the genotypes.Again, positive and negative environmental index (Ij) reflected the rich or favourable and poor or unfavourable environments for this character, respectively.Thus the environment of Gazipur and Barisal were poor whereas Rangpur was positive environments for hybrid maize production.Rangpur was highly suitable for hybrid maize cultivation followed by Barisal.
The regression coefficient (bi), deviation from regression (S 2 di) values of these genotypes ranged from 0.242 to 1.924 and 0.00 to 4.17, respectively.These differences in bi values indicated that all the genotypes responded differently to different environments.Considering the mean, bi and S 2 di three parameters, it was evident that all the genotypes showed different response of adaptability under different environmental conditions.Among the hybrids , WL4×WL5, WL1×WL3 , WL2×WL3, WL1×WL4 and check NK 40 exhibited the higher grain yield, bi~1 and S 2 di~0 indicated that the hybrids are stable across the environment.All the hybrids showed insignificant values for regression co-efficient except WL1×WL5 and also deviation from regression.
In the present study, it was found that the level of stability for a particular genotype was not similar for all characters.Grada and Ciulca (2012), Patel and Sanghi (1989) and Sharma and Hore (1991) reported that the levels of stability were not similar to any genotype for all characters, no uniform pattern of association between stability of yield and yield components and stress components could be established for individual genotype.
Figure 1 represents the Q-Q plot, histogram and scatter plot which revealed distribution pattern of the data.Figure 2 indicated yield range of hybrid maize.

AMMI biplot
The AMMI biplot provides a visual expression of the relationships between the first interaction principal component axis (IPCAI) and means of genotypes and environments (Figure 3) with the biplot accounting up to 95.3% of the treatment sum of squares.The IPCA1 was highly significant and explained the interaction pattern better than other interaction axes.The mean genotypes or environments in AMMI biplot located on the same parallel line, relative to the ordinate, had similar yield, while those located on the right side of the center of the axis had higher yields than those on the left hand side (Figure 1).

Conclusion
Among the hybrids , WL4×WL5, WL1×WL3, WL2×WL3, WL1×WL4 and check NK 40 exhibited higher grain yield; bi~1 and S 2 di~0 indicated that the hybrids found stable across the environment.All the hybrids showed insignificant values for regression co-efficient and also deviation from regression except WL1×WL5.The AMMI biplot showed four grouping of genotypes having none of them, low yielding and unstable; one hybrid was low yielding but moderately stable; eight hybrids including check NK 40 are high yielding and stable hybrids; 14 hybrids including BHM7-check were high yielder but highly unstable.Rangpur was highly suitable for hybrid maize cultivation followed by Barisal and Gazipur.

Figure 1 .
Figure 1.Normality test of the data for yield of maize.

Figure 2 .
Figure 2. Boxplot showing yield range of hybrid maize.

Figure 3 .
Figure 3. Biplot of the first AMMI interaction (IPCA1) score (Y-axis) plotted against mean yield (X-Axis) of twenty three maize hybrids and three environments.

Figure 5 .
Figure 5. Adaptation map of maize hybrids in three locations.

Table 1 .
Full joint analysis of variance including the portioning of the G×E interactions of maize hybrids.

Table 2 .
Stability analysis for cob length (cm) of maize hybrids over three environments.

Table 3 .
Stability analysis for cob diameter (cm) of maize hybrids over three environments.

Table 4 .
Stability analysis for TSW (g) of maize hybrids over three environments.

Table 5 .
Stability analysis for yield (t/ha) of maize hybrids over three environments.