Printer Friendly

Analysis of the genotype x environment interaction of barley grain yield (Hordeum Vulgare L.) under semi arid conditions.


Barley is grown as a rainfed crop on the Algerian high plateaus. In this region, most of the rain falls during the autumn- winter period, leading to the development of water and high temperature stresses during the grain filling phase. Under such growing conditions, grain yield is usually variable. Barley breeders aimed to develop varieties which perform relatively well over the target region. However the presence of genotype x environment interaction (G x E) hinders selection progress. Differential yield responses are attributed to differences in phenology, growth habit, vernalization and photoperiodic requirements [1]. To mitigate this problem, stratification of environments and multi environment trials are necessary to identify adapted and high yielding genotypes [2]. Lin and Binns [3] defined adaptation as yield consistency in space while yield stability represents consistency of genotype performances over time. Statistical methods for determining stability and adaptation of crop cultivars tested across diverse environments are used to assist breeders in selecting higher yielding and adapted genotypes [4]. The objectives of this study were to assess the magnitude of grain yield G x E interaction encountered by barley cultivars grown under semi arid conditions of the high plateaus of Algeria and to analyse their yield stability.

Materials and methods

Plant Material and Experimental Design:

Barley trials were conducted as a part of the breeding program of the Field Crop Institute (ITGC) Experimental Farm, located near Setif town, on the eastern high plateaus of Algeria. The experiment was laid down at two locations, Setif (36[degrees]12'N, 5[degrees]24'E, 1023 m asl) and Sersour (35[degrees]56' N, 5[degrees]32' E, 920 m asl), during five cropping seasons (1996, 1998, 1999, 2000, and 2002). The test sites were in a semi- arid area with cold winter, and hot and dry spring [5]. 13 barley genotypes were evaluated, including Saida and Tichedrett, 6-row type Algerian cultivars; Rihane, a 6-row type, Soufara, Rahma, and Tissa, 2-row type varieties and advances breeding lines from the Icarda breeding program; [Acsad.sup.176], a 6-row type Syrian released cultivar; Barberousse, Jaidor, Express, and Plaisant, 6-row type French varieties; and Tina and Begonia, 6-row type Spanish cultivars. These genotypes were sown in a randomized complete block design with four replicates in plots of 12 [m.sup.2] (6 rows, 10 m long and 0.20 m row spacing). Trials were sown in the second half of November at a rate of 250 viable seeds per [m.sub.2]. Plots, which had been fallow the year before planting, were fertilized with 100 kg [ha.sup.-1] of triple super phosphate (46% [P.sub.2][O.sub.5]). A 100 kg [ha.sup.-1] of urea

(35% N) was broadcasted at the onset of jointing growth stage (GS 31, Zadoks et al., [6]). Weeds were controlled chemically, when appropriate, with GranStar [Methyl Tribuneron], at 12 g [ha.sup.-1] rate.

Variables Recorded and Statistical Analysis:

Plots were combine-harvested at maturity; grain yield was determined and analyzed for each environment separately to test for the genotype effect. In the combined analysis of variance, each combination of location x growing season was treated as an environment (E) and the data were analysed according to the following model :[Y.sub.ijk] = m + [G.sub.i] + [E.sub.j] + [R(E).sub.jk] + [(G x E).sub.ij] + [e.sub.ijk], where is the observation of the ith variety G in the jth environment E in the kth replication R nested within environment E; m is the general mean, [e.sub.ijk] is the variation due to random error, associated with ith genotype, in the kth block of the jth environment and [(G x E).sub.ij] is the genotype by environment interaction [7]. The G x E interaction was partitioned according to the joint regression [8, 9] and the additive main effect and multiplicative interaction (AMMI) models [10]. The coefficient of regression (b), the deviation from regression ([S.sub.di]), the score on the first Interaction Principal Component Analysis (IPCA), and the Shukla stability variance [11] were used as measures of stability. The reliability index (I) was determined to estimate, on the basis of the distribution of yield values observed across the environments tested, the lowest yield expected for a given genotype. It was calculated according to the procedure outlined by Annichiarico [12]. The analyses were performed with cropstat. 7.2 free software [13] using the cross site analysis procedure.

Results and discussion

Grain Yield Variation Across Environments:

Total growing season rainfall, September 1st to the end of June, varied from 185.9 to 477.8 mm. The Setif location received, on average, more rainfall than Sersour location, during the course of the experience (Figure 1). The analysis of the variance of the individual trials (data not shown) indicated a significant genotype effect, suggesting the existence of potential useful genetic variability. The sampled environments showed a large range of grain yield, varying from 0.97 to 5.5 t ha-1, with an overall average of 3.04 t ha-1. This is suggestive of high variation in the growth conditions leading to differential genotypic response to the environmental changes. The combined analysis of variance indicated significant environment and genotype main effects and a significant genotype x environment interaction (Table 1).

The environment (E) was the most important source of variation, accounting for 91.93% of the treatment sum of squares, while the genotype (G) and genotype x environment (GxE) interaction effects explained 1.98% and 6.08%, respectively (Table 1). Samonte et al., [14] reported that E, G and GxE explained 55.4, 17.8 and 26.7% of the total sum of square, respectively. Farshadfar and Sutka [15] reported 86.0, 2.0 and 12.0% for the same source of variation. The magnitude of the GxE interaction sum square was three times larger than the genotype sum square. This indicated that substantial differences in genotypic response across environments existed, suggesting to conduct multiple-environment trials to be able to identify superior cultivars for the target region.

Joint regression and AMMI analyses of grain yield G x E interaction:

The regression analysis explained only 26.80% of the interaction, in terms of linear response to the environment with significant deviations from linearity (Table 1). This indicated that regression analysis was enable to fully explain the G x E interaction pattern. These results corroborated findings of Farshadfar and Sutka, [15]. According to Powell et al., [16], the slope values are indicative of stability or sensitivity to environmental changes, and in association with yield level, they are indicative of adaptability. The test of the slopes indicated that [Acsad.sub.176], Tichedrett and Saida, had b values significantly smaller than unity (Table 2). These entries had below average yield, above average stability, and exhibited specific adaptability to low yielding environments (Table 2, Figure 2). Kang [17] mentioned that regression coefficient below 1 provided a measure of greater resistance to environmental change, and this is an indication of above average stability.

The slope of Rahma was significantly greater than unity, suggesting increasing sensitivity to environmental change, associated with below average stability. This genotype had above average grain yield, exhibiting a specific adaptability to favourable environments. The remaining entries had regression coefficients equal to unity, indicating average stability, among which Soufara and Tina had above average grain yield and thus showing general adaptability over the tested environments (Table 2, Figure 2). Test of the mean square due to deviation from regression against the pooled error indicated that all entries had [S.sup.2.sub.di] significantly different from zero, and thus were unstable according to Eberhart and Russell [9]. The first four terms of the AMMI analysis were significant, explaining 84.72% of the GxE interaction sum of squares (45.97%, 24.12%, 12.87%, and 1.76%, for the 1st, 2nd, 3rd and 4th interaction principal component axis, respectively), with 62% of the GxE interaction degrees of freedom, and leaving a non-significant residual (Table 1). Compared with the 26.8% of the interaction sum of squares accounted for by the regression model, the AMMI model appeared more effective at capturing a large portion of the G x E sum of squares. The [AMMI.sub.1] biplot displayed large differences in the additive main effects of genotypes and environments and in the pattern of the GxE interaction. Low yielding environments were [SR.sub.98] and [SR.sub.99] and high yielding ones were [ST.sub.96] and [ST.sub.99]; and similarly high yield genotypes were Soufara and Rahma and low yielding ones were Plaisant and Jaidor (Figure 3). Large [IPCA.sub.1] scores, negative or positive, are indicative of interaction while low [IPCA.sub.1] scores (values close to zero, the origin) are indicative of little to no interaction. Tina, Express, Jaidor, Tissa, Barberousse and Rihane had [IPCA.sub.1] scores close to zero (origin); they contributed little to the interaction (Figure 3).

Soufara, Rahma, Begonia, Plaisant, Tichedrett, Saida and [Acsad.sub.176], had large [IPCA.sub.1] scores; they contributed significantly to the interaction. Soufara and Rahma differed from Begonia and Plaisant in yield main effect; and differed from Tichedrett, Saida and [Acsad.sub.176] in yield main effect and interaction. [ST.sub.99], [SR.sub.99] and [SR.sub.98] environments were most discriminating as indicated by their large vectors. [ST.sub.02] was the least discriminating and genotypes ranking at this environment will be consistent with the ranking based on genotype main effects. Soufara and Rahma were more adapted to high yielding environments ([ST.sub.99]), while Tichedrett, Saida, [Acsad.sub.176] showed specific adaptation to low yielding environment ([SR.sub.99], [SR.sub.98]) (Figure 3). The GxE interaction pattern displayed by the AMMI1 biplot indicated large variation between growing seasons among locations. However, the intra location variation was relatively smaller at the Sersour location where the growing seasons were grouped compared to those of Setif location. This indicated that specific adaptation to Sersour location is more likely to be effective, while specific adaptation to Setif location will always be subjected to greater year-to-year variation. The AMMI2 biplot contrasted, along the [IPCA.sub.1] axis, the high yielding environment ([ST.sub.99], [SR.sub.96] and [ST.sub.96]) with their positive [IPCA.sub.1] scores with the low yielding environments ([SR.sub.00], [SR.sub.02], [ST.sub.00], [SR.sub.98]) having negative [IPCA.sub.1] scores (Figure 4). Similar dispersion pattern along the [IPCA.sub.1] was observed for the genotypes. However the group of genotypes adapted to high yielding environments was subdivided, along the [IPCA.sub.2] axis, into two subgroups: Express, Plaisant, and Begonia, with positive [IPCA.sub.2] scores specifically adapted to [ST.sub.99] environment, and Rahma and Soufara with negative scores on the [IPCA.sub.2] and specifically adapted to [SR.sub.96] environment (Figure 4). Tichedrett, [Acsad.sub.176] and Saida showed specific adaptation to [SR.sub.00], [SR.sub.02] and [ST.sub.00] environments. Tissa, Rihane, Barerousse, Jaidor and to some extend Tina had low [IPCA.sub.1] and [IPCA.sub.2] scores, exhibiting low interaction and general adaptability to the tested environments (Figure 4). The GxE pattern revealed by AMMI biplots indicated that the tested barley genotypes are narrowly adapted, and no genotype was found to have high performances in all environments.

Stability Variance, Environmental Variance and Reliability Index:

In the presence of variable abiotic stressors, a desirable, ideal genotype has consistent, good performance in the target region. According to the Shukla's stability variance Barberousse, Jaidor, Plaisant, Soufara, Tichedrett and Tina were stable, while [Acsad.sub.176], Begonia, Express, Rahma, Rihane, Saida, and Tissa were rated as highly unstable (Table 2). Among the stable entries, Soufara and Tina associated low stability variance with above average grain yield (Table 2). Stability variance was not significantly correlated with yield main effect at P=0.05. The use of a yield reliability index facilitates genotype recommendation, as the mean yield and the yield stability are combined into a unique measure of genotype merit [18].The reliability index ranked the genotypes according to the lowest average yield expected under harsh growing conditions. The most reliable genotypes were [Acsad.sub.176], Tichedrett and Saida, (Table 2). These genotypes had below average grain yield, and showed adaptability to low yielding environments (Figures, 3 and 4). Soufara and Tina had an intermediate reliability index, above average yield, a low stability variance, exhibiting adaptation to high yielding environments (Table 2, Figures 3 and 4). Begonia, Express and Plaisant were less reliable as their index values were low (Table 2). These findings suggested recommending Soufara for high yielding environments (Setif site) and [Acsad.sub.176] to low yielding ones (Sersour site). Based on the [AMMI.sub.2] model, the expected yields of Soufara and [Acsad.sub.176] were determined for the yield range observed through the environments tested, according to the method outlined by Zobel et al., [19]. The results confirmed the narrowed adaptation exhibited by both cultivars (Figure 5)


The presence of the genotype x environment interaction was indicated by changes in relative rankings over environments. The regression analysis explained 26.8% of the GxE interaction sum of squares, compared to 84.72% explained by the AMMI model. Regression analysis indicated that [Acsad.sub.176], Tichedrett and Saida had below average yield, above average stability; and specific adaptability to low yielding environments. Rahma had increasing sensitivity to environmental changes, associated with below average stability, and a specific adaptability high yielding environments. All entries were unstable according to the variance of the deviation from regression. The AMMI1 biplot displayed large differences in the additive main effects for genotypes and environments and in the pattern of the GxE interaction Tina, Express, Jaidor, Tissa, Barberousse and Rihane contributed little to the interaction. Soufara and Rahma Begonia and Plaisant Tichedrett, Saida and [Acsad.sub.176] were the vertex cultivars and [ST.sub.99], [SR.sub.99] and [SR.sub.98] environments were the most discriminating. Soufara and Rahma, classified as unstable according to Shukla's stability variance, with intermediate reliability index; were more adapted to high yielding environments. Tichedrett, Saida and [Acsad.sub.176], were classified as stable according to Shukla's stability variance, were highly reliable and showed specific adaptation to low yielding environment. The GxE pattern revealed by the AMMI analysis indicated that the tested barley genotypes are narrowly adapted, and no genotype was found to have high performances in all environments.






[1.] Van Oosterom, E.J., D. Kleijn, S. Ceccarelli, M.M. Nachit, 1993. Genotype x environment interactions of barley in the Mediterranean region. Crop Sci., 33: 669-674.

[2.] Basford, K.E., M. Cooper, 1998. Genotype-environment interactions and some considerations of their implications for wheat breeding in Australia. Aust. J. Agri. Res., 49: 153-174.

[3.] Lin, C.S., M.R. Binns, 1988. Genetic properties of four stability parameters. Theor. Appl.Genet., 82: 205-09.

[4.] Lin, C.S., M.R. Binns, L.P. Lefkovitch, 1986. Stability analysis: Where do we stand? Crop Sci., 26: 894 -900.

[5.] Bouzerzour, H., M. Dekhili, 1995. Heritabilities, gains from selection and genetic correlations for grain yield of barley grown in two contrasting environments. Field Crops Research, 41: 173-178.

[6.] Zadocks, J.C., T.T. Chang, C.F. Konzak, 1974. A decimal code for the growth stages of cereals. Weed Res., 14: 415-421.

[7.] Shukla, G.K., 1972. Some statistical aspects of partitioning genotype environmental components of variability. Heredity, 29: 237-245.

[8.] Finlay, K.W., G.N. Wilkinson, 1963. The analysis of adaptation in a plant breeding programme. Austral. J. Agr. Res., 14: 742-754.

[9.] Eberhart, S.A., W.A. Russell, 1966. Stability parameters for comparing varieties. Crop Sci. 6:36-40.

[10.] Gauch, H.G., R.W. Zobel, 1997. Identifying mega-environments and targeting genotypes. Crop Sci., 37: 311-326.

[11.] Steel, R.G.D., J.H. Torrie, 1980. Principles and procedures of statistics. McGraw-Hill Books, New York.

[12.] Annicchiarico, P., 2002. Genotype x environment interactions: Challenges and opportunities for plant breeding and cultivar recommendations. FAO Plant Production and Protection Paper No. 174. Food and Agriculture Organization, Rome.

[13.] Cropstat 7.2. 2008. Free Software package for windows, International Rice Research Institute, IRRI, Manila.

[14.] Samonte, S.O.B., L.T. Wilson, A.M. McClung, J.C. Medley, 2005. Targeting cultivars onto rice growing environments using AMMI and SREG GGE Biplot analyses. Crop Sci., 45: 2414-2424.

[15.] Farshadfar, E., J. Sutka, 2006. Biplot analysis of genotype-environment interaction in durum wheat using the Ammi model. Acta Agronomica Hungarica, 54: 459-467.

[16.] Powell, W., P.D.S. Caligari, M.S. Phillips, J.L. Jinks, 1986. The measurement and interpretation of genotype by environment interaction in spring barley (Hordeum vulgare L.). Heredity, 56: 255262.

[17.] Kang, M.S., 1993. Simultaneous selection for yield and stability in crop performance trials: Consequences for growers. Agron. J., 85: 754-757.

[18.] Eskridge, K.M., 1990. Selection of stable cultivars using a safety-first rule. Crop Sci., 30: 369-374.

[19.] Zobel, R.W., M.J. Wright, H.G Gauch, Jr., 1988. Statistical analysis of a yield trial. Agron. J., 80: 388-393.

(1) Zahia Kadi, (1) Farrah Adjel and (2) Hamenna Bouzerzour

(1) Biology Department, Faculty of Sciences, Oum El Bouaghi University, 04000 (Algeria)

(2) Biology Department, Faculty of Sciences, Setif University, 19000 (Algeria)

Corresponding Author

Zahia Kadi, Biology Department, Faculty of Sciences, Oum El Bouaghi University, 04000 (Algeria) E-mail:
Table 1: Combined analysis of variance, Joint regression and AMMI
analyses of the genotype x environments interaction of grain yield
of 13 barley genotypes grown at two locations during 5 cropping

Source of variation             df           SS           MS

Treatment                       129          1131.68      8.77
Environment (E)                 9            1040.38      115.60
Block/E                         30           1.66         0.06
Genotype (G)                    12           22.45        1.87
G x E                           108          68.85        0.64
Heterogeneity of regressions    12           18.45        1.54
Deviation from regression       96           50.40        0.52
[IPCA.sub.1]                    20           31.65        1.58
[IPCA.sub.2]                    18           16.61        0.92
[IPCA.sub.3]                    16           8.86         0.55
[IPCA.sub.4]                    14           1.21         0.08
Residual                        40           1.72         0.04
Pooled error                    360          24.66        0.07
Total                           519          1158.02      2.23

Source of variation             F-value      F-test       %SS

Treatment                       125.2        **           --
Environment (E)                 1926         **           91.93
Block/E                         0.81         ns           --
Genotype (G)                    2.94         **           1.98
G x E                           9.30         **           6.08
Heterogeneity of regressions    2.96         **           26.80
Deviation from regression       7.43         **           73.20
[IPCA.sub.1]                    7.52         **           45.97
[IPCA.sub.2]                    4.38         **           24.12
[IPCA.sub.3]                    2.62         **           12.87
[IPCA.sub.4]                    2.00         *            1.76
Residual                        0.57         ns           --
Pooled error                    --           --           --
Total                           --           --           --

ns, **, ** = non significant and significant effect at 5 and 1%

Table 2: Mean grain yield (GY, t [ha.sup.-1]), coefficient of
regression (b), mean of squares deviation from regression
([S.sup.2.sub.di]), IPCA1 scores, environmental variance
([S.sup.2.sub.i]), Shukla's stability variance
([[sigma].sup.2.sub.i]) and reliability index (I, t [ha.sup.-1]) of
the 13 barley genotypes evaluated at 2 locations during 5 cropping

Genotype         GY                b                 [S.sup.2.sub.di]

Acsad176         2.94 (b)          0.780 (-)         0.160 **
Barberousse      2.94 (b)          0.974             0.0500 *
Begonia          3.02 (b)          1.161             0.180 **
Express          2.93 (b)          1.044             0.130 **
Jaidor           2.86 (b)          0.955             0.0500 *
Plaisant         2.79 (b)          1.125             0.140 **
Rahma            3.40 (a)          1.182 (+)         0.130 **
Rihane           3.09 (b)          1.037             0.0600 *
Saida            2.93 (b)          0.797 (-)         0.130 **
Soufara          3.51 (a)          1.130             0.210 **
Tichedrett       2.87 (b)          0.796 (-)         0.120 **
Tina             3.24 (a)          1.033             0.120 **
Tissa            2.99 (b)          0.986             0.120 **

Genotype         [IPCA.sub.1]        [S.sup.2.sub.i]

Acsad176         -0.850              1.489
Barberousse      -0.098              2.153
Begonia          0.610               3.156
Express          0.249               2.535
Jaidor           -0.048              2.070
Plaisant         0.496               2.940
Rahma            0.484               3.221
Rihane           -0.126              2.441
Saida            -0.737              1.522
Soufara          0.465               3.027
Tichedrett       -0.572              1.513
Tina             0.228               2.482
Tissa            -0.103              2.264

Genotype         [[sigma].sup.       I

Acsad176         0.175 **            0.93
Barberousse      0.072 ns            0.53
Begonia          0.209 **            0.10
Express          0.234 **            0.31
Jaidor           -0.007 ns           0.49
Plaisant         0.008 ns            -0.03
Rahma            0.161 **            0.45
Rihane           0.198 **            0.52
Saida            0.130 *             0.90
Soufara          -0.003 ns           0.65
Tichedrett       0.078 ns            0.85
Tina             0.066 ns            0.65
Tissa            0.184 **            0.52

GY means followed by the same letter are not significantly different
from each other according to the least significant difference test
(LSD5%) at the 0.05 probability; (-), (+): slope significantly
smaller than 1.0, and significantly greater than 1.0, respectively,
the remaining slopes were not significantly different from unity;
ns, *, ** = [S.sup.2.sub.di] and [[sigma].sup.2.sub.i] significant
at 0.05 and 0.01 probability level, respectively.

Fig. 1 Growing season accumulated rainfall
during the course of the experience.

Environments    Rain (mm)

ST96              475.8
ST98              477.8
ST99              353.1
ST00              384.6
ST02              215.9
SR96              310
SR98              345
SR99              280.5
SR00              305
SR02              185.9

Note: Table made from bar graph.
COPYRIGHT 2010 American-Eurasian Network for Scientific Information
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2010 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Title Annotation:Original Article
Author:Kadi, Zahia; Adjel, Farrah; Bouzerzour, Hamenna
Publication:Advances in Environmental Biology
Article Type:Report
Geographic Code:4EUSP
Date:Jan 1, 2010
Previous Article:Comparison of different delivery system of Trichoderma and Bacillus as biofertilizer.
Next Article:Sowing dates and density evaluation of Amaranth (cv. Koniz) as a new crop.

Terms of use | Privacy policy | Copyright © 2022 Farlex, Inc. | Feedback | For webmasters |