SciELO - Scientific Electronic Library Online

 
vol.19 issue3Changes in leaf appearance and developmental phases associated with breeding progress in six-rowed barley in the Pannonian PlainGenetic progress of the peach breeding program of Embrapa over 16 years author indexsubject indexarticles search
Home Pagealphabetic serial listing  

Services on Demand

Journal

Article

Indicators

Related links

Share


Crop Breeding and Applied Biotechnology

Print version ISSN 1518-7853On-line version ISSN 1984-7033

Crop Breed. Appl. Biotechnol. vol.19 no.3 Viçosa July/Sept. 2019  Epub Oct 31, 2019

http://dx.doi.org/10.1590/1984-70332019v19n3a43 

ARTICLE

Genotype x environment interaction analysis of multi-environment wheat trials in India using AMMI and GGE biplot models

Charan Singh1 

Arun Gupta1 

Vikas Gupta1  * 
http://orcid.org/0000-0002-7640-4816

Pradeep Kumar1 

R Sendhil1 

BS Tyagi1 

Gyanendra Singh1 

Ravish Chatrath1 

GP Singh1 

1Indian Institute of Wheat and Barley Research, 132001, Haryana, India.


Abstract

Fifty wheat genotypes were evaluated at nine diverse locations in India to identify high-yielding and stable genotypes. The analysis of variance based on additive main effects and multiplicative interaction (AMMI) indicated significant genotype, environment and genotype - environment (GE) interactions, with a total variation of 5.99, 20.23 and 73.77%, respectively. A biplot-AMMI analysis and yield stability index incorporating the AMMI stability value and yield in a single non-parametric index were used to discriminate the genotypes with highest and stable yield; the genotypes G135, G125, G104, G112 and G144 were found to be promising. Two mega environments (ME) were identified based on GGE (genotype and GE interaction) biplot analysis and the genotypes G119 and G120 and G107, G148 and G146 performed best in the mega-environments ME I and ME II, respectively. Both approaches allowed the identification of stable genotypes (G112 and G135), which can be included in the national testing program, with a view to release a new variety.

Keywords: AMMI; biplot; GGE; stability; wheat

INTRODUCTION

Worldwide as well as in India, bread wheat (Triticum aestivum L.) is the second most important food grain crop after rice. The importance of a sustained increase in wheat production and productivity for food security is well recognized in India, where wheat is a major staple food crop for the ever-increasing human population. The development of high-yielding genotypes coupled with resistance/tolerance to diverse biotic and abiotic stresses will be decisive to meet the demand for food grain. Multilocation trials are a key component of selection for stable and best-performing genotypes in different environments (Ahmadi et al. 2012, Oral et al. 2018, Tekdal and Kendal 2018). The grain yield, the final product of any crop, is determined by the genotypic potential (G), environmental effect (E) and the genotype x environment (GE) interaction (Yan and Kang 2002). In case of an effect of the GE interaction, the selection of genotypes based on the mean yield is inadequate (Sharifi et al. 2017).

A wide variety of methods to detect genotypes with a stable performance across environments were described in the literature. Most of them use regression analysis, sum of squared deviations from regression, Principal Component Analysis (PCA), cluster analysis and Additive Main effects and Multiplicative Interaction models (AMMI). The AMMI model uses ANOVA (analysis of variance) to test the main effects of both genotypes and environments and PCA to analyse the residual interaction component. The GGE biplot is a powerful model, with a graphical representation of identification of the best performing cultivars across the environments (Yan and Kang 2002). These graphical options facilitate the identification of high-yielding, stable genotypes, particularly in multi-environment trials. Moreover, yield stability and wide adaptation are increasingly important as the climate at specific locations is becoming more variable over the years. The AMMI biplot approach has been used for the identification of stable genotypes in multi-environment trials of wheat and barley (Oral et al. 2018, Tekdal and Kendal 2018). Both GGE and AMMI models have also been used to study the interaction component in multi-environment trials to identify stable wheat genotypes (Ahmadi et al. 2012, Kendal and Sener 2015, Vaezi et al. 2017, Oral and Kendal 2018). In this study, genotypes derived from the CIMMYT Elite Spring Wheat Yield Trial (ESWYT) were evaluated for grain yield across different environments to stratify the wheat genotypes according to the environmental conditions, for specific recommendations. The objectives of this study were to: i) analyse the G×E interaction on the grain yield of 50 wheat genotypes using AMMI and GGE biplot models; ii) identify high yielding and stable wheat genotype(s) across environments and to; iii) identify suitable genotype(s) for each environment.

MATERIAL AND METHODS

Experimental material and multi-environment wheat trials

Elite Spring Wheat Yield Trials (ESWYT) consisting of 50 genotypes including one local check were planted at nine test locations (Tables 1 and 2) in India in the winter (Rabi) growing season of 2016/17. The trial was arranged in a randomized complete block design, with two replications per location. The code and pedigree of all genotypes are listed in Table 3. Each genotype was planted in a plot with six 6-m rows, with a row-to-row and plant-to-plant distance of 20 cm and 10 cm, respectively. The recommended management practices were followed for aising strong and healthy crops. The grain yield data were recorded as total grain weight per plot after harvesting and the values extrapolated to kg ha-1.

Table 1 Details of the different locations of evaluation of wheat genotypes 

Code Location Latitude Longitude Mean sea level (m) Annual rainfall (mm) Mean grain yield (kg ha-1)
E1 Karnal 290 4’ N 760 59’ E 253 654.3 7193
E2 Hisar 290 10’ N 750 46’ E 215 287.4 5507
E3 Pantnagar 290 30’ N 790 31’ E 243 991.6 6014
E4 Gurdaspur 320 30’ N 750 21’ E 264 1282.5 6709
E5 Delhi 280 39’ N 770 13’ E 227 815.4 4651
E6 Ludhiana 300 54’ N 750 51’ E 252 576.8 6037
E7 Indore 220 43’ N 750 51’ E 550 776.8 4889
E8 Pune 180 31’ N 730 51’ E 562 1458.1 6010
E9 Dharwad 150 27’ N 750 0’ E 724 580.7 3509

Table 2 Monthly temperature pattern during the growing season at the different locations 

Location Temperature (0C) during growing season
Oct. Nov. Dec. Jan. Feb March April
Max Min Max Min Max Min Max Min Max Min Max Min Max Min
Karnal 34.0 15.0 28.1 7.8 25.0 6.1 21.5 3.9 24.7 5.3 29.6 11.0 37.9 17.2
Hisar 36.7 15.5 31.0 6.4 26.8 4.5 22.9 2.6 27.9 4.6 31.8 10.5 39.2 18.0
Pantnagar 33.5 14.5 28.6 8.5 23.2 7.2 20.5 4.2 27.4 5.9 32.2 10.7 37.1 16.2
Gurdaspur 33.5 14.8 27.2 8.5 24.0 7.7 16.0 4.6 24.0 6.3 31.7 11.0 36.2 16.1
Delhi 35.3 14.8 29.8 7.4 23.9 6.1 21.4 3.0 28.3 4.6 31.8 11.5 38.2 15.4
Ludhiana 34.9 16.2 28.0 7.4 22.7 7.3 22.0 5.3 25.5 5.6 29.9 12.2 39.6 18.0
Indore 35.1 20.3 32.9 12.4 27.9 7.5 29.7 7.8 34.5 9.9 36.9 13.9 41.5 18.2
Pune 31.9 20.3 32.1 12.4 30.2 8.0 31.1 9.6 33.6 10.0 36.4 14.0 37.5 17.5
Dharwad 29.8 19.1 30.1 15.0 29.6 10.7 30.3 13.3 33.3 16.1 35.3 16.4 36.4 20.6

Table 3 Pedigree details of 50 wheat genotypes 

SN Pedigree details
G101 LOCAL CHECK
G102 KACHU #1
G103 KUTZ
G104 CHIPAK
G105 MUCUY
G106 MUTUS/DANPHE #1/4/C80.1/3*BATAVIA//2*WBLL1/3/C80.1/3*QT4522//2*PASTOR
G107 CROC_1/AE.SQUARROSA (205)//BORL95/3/PRL/SARA//TSI/VEE#5/4/FRET2/5/TRCH/SRTU//KACHU
G108 BAV92//IRENA/KAUZ/3/HUITES/4/DOLL/5/SERI.1B//KAUZ/HEVO/3/AMAD*2/4/KIRITATI
G109 KACHU/SAUAL/3/TACUPETO F2001/BRAMBLING//KIRITATI
G110 KACHU/SAUAL/5/SERI.1B//KAUZ/HEVO/3/AMAD*2/4/KIRITATI
G111 KACHU/SAUAL/3/TRCH/SRTU//KACHU
G112 TRCH/SRTU//KACHU/5/SERI.1B//KAUZ/HEVO/3/AMAD*2/4/KIRITATI
G113 TRCH/SRTU//KACHU/3/TRCH/SRTU//KACHU
G114 HUW234+LR34/PRINIA*2//SNLG/3/KINGBIRD #1/4/BAJ #1
G115 WHEAR//2*PRL/2*PASTOR/4/2*WBLL1/KUKUNA//TACUPETO F2001/3/UP2338*2/VIVITSI
G116 BOKOTA/3/UP2338*2/KKTS*2//YANAC
G117 UP2338*2/SHAMA/3/MILAN/KAUZ//CHIL/CHUM18/4/UP2338*2/SHAMA/5/COPIO
G118 SAUAL/WHEAR//SAUAL/3/PBW343*2/KUKUNA*2//FRTL/PIFED
G119 UP2338*2/VIVITSI/3/FRET2/TUKURU//FRET2/4/MISR 1/5/TUKURU//BAV92/RAYON*2/3/PVN
G120 CAL/NH//H567.71/3/SERI/4/CAL/NH//H567.71/5/2*KAUZ/6/WH576/7/WH 542/8/WAXWING/9/ATTILA*2/PBW65//PIHA/3/ATTILA/2*PASTOR/10/UP2338*2/KKTS*2//YANAC
G121 NELOKI/5/FRET2/KUKUNA//FRET2/3/TNMU/4/FRET2*2/SHAMA/6/KINGBIRD #1//INQALAB 91*2/TUKURU
G122 SHA7//PRL/VEE#6/3/FASAN/4/HAAS8446/2*FASAN/5/CBRD/KAUZ/6/MILAN/AMSEL/7/FRET2*2/KUKUNA/8/TRAP#1/BOW/3/VEE/PJN//2*TUI/4/BAV92/RAYON/5/KACHU #1/9/COPIO
G123 ELVIRA/CHIBIA//DIAMONDBIRD/4/2*MARCHOUCH*4/SAADA/3/2*FRET2/KUKUNA//FRET2
G124 ATTILA*2/PBW65/5/CNO79//PF70354/MUS/3/PASTOR/4/BAV92/6/TRCH/SRTU//KACHU/7/UP2338*2/KKTS*2//YANAC
G125 ROLF07/SAUAL/3/TRCH/SRTU//KACHU/4/ROLF07/SAUAL
G126 ROLF07/SAUAL/3/TRCH/SRTU//KACHU/4/ROLF07/SAUAL
G127 TACUPETO F2001/BRAMBLING/5/NAC/TH.AC//3*PVN/3/MIRLO/BUC/4/2*PASTOR*2/6/TRCH/SRTU// KACHU
G128 TACUPETO F2001/BRAMBLING/5/NAC/TH.AC//3*PVN/3/MIRLO/BUC/4/2*PASTOR*2/6/WAXWING/ SRTU//WAXWING/KIRITATI
G129 TACUPETO F2001/BRAMBLING/5/NAC/TH.AC//3*PVN/3/MIRLO/BUC/4/2*PASTOR*2/6/WAXWING/ SRTU//WAXWING/KIRITATI
G130 TACUPETO F2001/BRAMBLING/5/NAC/TH.AC//3*PVN/3/MIRLO/BUC/4/2*PASTOR*2/6/WAXWING/ SRTU//WAXWING/KIRITATI
G131 KAUZ//ALTAR 84/AOS/3/MILAN/KAUZ/4/SAUAL/5/SERI.1B//KAUZ/HEVO/3/AMAD*2/4/ KIRITATI/6/KACHU/SAUAL
G132 KACHU/SAUAL*2/3/TACUPETO F2001/BRAMBLING//KIRITATI
G133 KACHU/SAUAL*2/3/TACUPETO F2001/BRAMBLING//KIRITATI
G134 KACHU/SAUAL*2/5/SERI.1B//KAUZ/HEVO/3/AMAD*2/4/KIRITATI
G135 KACHU/SAUAL*2/4/ATTILA*2/PBW65//PIHA/3/ATTILA/2*PASTOR
G136 KACHU/SAUAL/4/VARIS/MISR 2/3/FRET2/KUKUNA//FRET2/5/KACHU/SAUAL
G137 KACHU/SAUAL/4/VARIS/MISR 2/3/FRET2/KUKUNA//FRET2/5/KACHU/SAUAL
G138 AMUR*2/3/HUW234+LR34/PRINIA//UP2338*2/VIVITSI
G139 C80.1/3*BATAVIA//2*WBLL1/5/REH/HARE//2*BCN/3/CROC_1/AE.SQUARROSA (213)//PGO/4/HUITES*2/6/TRCH/SRTU//KACHU
G140 FRET2*2/SHAMA//PARUS/3/FRET2*2/KUKUNA*2/4/TRCH/SRTU//KACHU
G141 TRCH/SRTU//KACHU/3/WAXWING/PARUS//WAXWING/KIRITATI/4/TRCH/SRTU//KACHU
G142 TRCH/SRTU//KACHU*2/4/WBLL1/KUKUNA//TACUPETO F2001/3/UP2338*2/VIVITSI
G143 CROC_1/AE.SQUARROSA (205)//BORL95/3/PRL/SARA//TSI/VEE#5/4/FRET2/6/MTRWA92.161/ PRINIA/5/SERI*3//RL6010/4*YR/3/PASTOR/4/BAV92
G144 QUAIU #2/BAVIS #1
G145 BECARD #1/4/SOKOLL/3/PASTOR//HXL7573/2*BAU
G146 WBLL1*2/BRAMBLING/3/SOKOLL//SUNCO/2*PASTOR
G147 SOKOLL/92.001E7.32.5//SOKOLL/EXCALIBUR
G148 BAVIS #1*2/4/PASTOR//HXL7573/2*BAU/3/SOKOLL/WBLL1
G149 CROC_1/AE.SQUARROSA (224)//OPATA/3/PASTOR/4/2*SOKOLL/3/PASTOR//HXL7573/2*BAU
G150 PASTOR//HXL7573/2*BAU/3/SOKOLL/WBLL1/4/SUNCO/2*PASTOR//EXCALIBUR/5/W15.92/4/PASTOR//HXL7573/2*BAU/3/WBLL1

Statistical analysis

The AMMI analysis was carried out with the adjusted mean grain yield to assess the relationships among genotypes, locations and G×E interaction, based on the model described by Zobel et al. (1988) and Crossa (1990). The AMMI and GGE biplot package in R software (R Core Team 2013) were used for the analyses. The AMMI stability indices; AMMI distance (Di) and AMMI Stability Value (ASV) were calculated by the procedure proposed by Zhang et al. (1998) and Purchase et al. (2000), respectively. Stability per se might not be the only selection parameter because the most stable genotypes do not necessarily have the best yield performance (Mohammadi and Amri 2007). We decided to incorporate both yield and stability in a single index to classify stable genotypes. The genotype stability index (GSI) considered the ranks of the genotype yields across environments and AMMI stability values. This index incorporates the yield mean and stability index in a single criteria and is calculated as:

GSI = RASV+RY

where RASV is the rank of ASV and RY the rank of mean genotype yield of all environments.

The data were graphically analysed to interpret the GxE interaction to identify stable and adaptive genotypes by the GGE biplot, as described by Yan and Tinker (2006). The biplots were generated from the first two PCAs, without scaling, centering (2) or singular value partitioning (SVP) (2). The lines that connect the test environment to the biplot origin are called environment vectors and the cosine of the angle between the vectors of two environments approximates the correlation between them (Yan et al. 2007).

RESULTS AND DISCUSSION

GE analysis by AMMI model

The average grain yield of the genotypes over locations ranged from 4936 kg ha-1 (115) to 6279 kg ha-1 (107). Genotype G106 yielded highest at two locations (Karnal and Hisar); G119 at three (Karnal, Pantnagar and Gurdaspur) and G107 at two locations (Indore and Pune) (Table 4). The AMMI model is widely used in stability analysis as it provides an initial diagnosis of the model to be fit into multi environmental evaluation, allows a partitioning of the GxE interaction and explains patterns and relationships between genotypes and environments (Zobel et al. 1988, Crossa et al.1990). The AMMI analysis of variance for grain yield showed that 73.77% of the total sum of squares was attributable to environmental, only 5.99% to genotypic and 20.23% to GxE effects (Table 5). 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, indicating that environment has a strong influence on grain yield (Tonk et al. 2011, Munaro et al. 2014, Alam et al. 2015). The magnitude of the GxE interaction sum of squares was 3.37 times higher than that for genotypes, indicating that there were substantial differences in genotypic response across environments, in agreement with previous reports (Tonk et al. 2011, Alam et al. 2015, Vaezi et al. 2017). The multiplicative variance of the treatment sum of squares due to interaction was partitioned into seven significant interaction principal components. The first two PCs explained 46.62% of the total variation, in which the contribution of PC1 was 27.94% and that of PC2 18.68. Therefore, AMMI1 (IPCA1 vs additive main effects) and AMMI2 (IPCA2 vs IPCA1) biplots were generated to illustrate the genotype and environment effects simultaneously (Figure 1). The AMMI 1 biplot indicated that the environments E1, E4, E6, E9 and E3 were high yielding locations with high additive genotypic main effects, while the yields in the other environments were below the environmental mean. The scatter plot of the genotypes in this biplot indicated that genotype G107, followed by G119, G120, G130, G106, G104, G117, G136, G150, G101, G125 were the 10 highest yielding genotypes. The AMMI2 biplot indicated that the environments E2, E4 and E6 were discriminatory and located far away from the biplot origin. The genotypes G111, G135, G137, G134 and G130 were located close to the origin and proved highly stable, although their mean yields were on the lower side and should therefore not be recommended. Similar results regarding the stability of genotypes due to low IPCA1 values were recorded elsewhere (Mohammadi et al. 2013, Oral et al. 2018). The genotypes G128, G146, G148, G105, G141, G103, G124, G108, G127, located far away from the origin, were highly unstable and expressed a higher GE interaction (positive or negative).

Figure 1 AMMI bi-plot model showing relationship among: a) test environments and genotypes based on grain yield; and b) among IPC-1 and grain yield. 

Table 4 Mean grain yield (kg ha-1) of 50 wheat genotypes at nine locations in India 

Genotypes Karnal Hisar Pantnagar Gurdaspur Delhi Ludhiana Indore Pune Dharwad Mean CV
G101 6409 6389 6772 6488 4987 6610 4864 7578 2709 5867 2485
G102 6934 4132 5305 6013 4986 5032 4674 6332 3442 5205 2098
G103 7865 5000 6867 6504 3688 5488 4779 6385 4114 5632 2437
G104 7523 6563 6957 6357 4862 6062 4747 6932 3683 5965 2119
G105 7788 5764 6415 5713 4111 6289 4742 6463 3718 5667 2264
G106 8173 6875 6592 7369 4362 6132 4964 6313 3089 5985 2645
G107 7475 6597 5447 7125 5243 6995 5831 7910 3887 6279 2039
G108 6265 4132 5875 7000 4612 5549 4288 5169 4472 5262 1872
G109 7507 4688 6142 6688 4361 5059 4979 6497 3873 5532 2205
G110 6431 4271 5919 6225 4459 5555 5267 6616 3913 5406 1839
G111 7453 5174 5640 6788 4681 6080 4768 6075 3353 5557 2209
G112 8292 5695 6005 6313 4584 5847 4709 5822 3483 5639 2374
G113 6675 4722 6360 6319 3278 4785 4874 6566 3258 5204 2597
G114 7246 6771 6288 6688 4306 6887 5013 5582 3701 5831 2147
G115 6363 4688 5232 5913 3653 3892 5282 5900 3500 4936 2147
G116 6283 5278 6212 5107 4702 6159 4475 6450 4069 5415 1643
G117 7384 6042 6439 7194 5049 6662 5013 6294 3663 5971 1999
G118 7165 4306 5744 6707 5056 5332 5322 5041 2878 5283 2379
G119 8057 6181 7109 8188 3952 6919 5183 6538 3764 6210 2601
G120 7644 6597 6544 7500 5612 7019 4820 6707 3145 6176 2329
G121 6694 5834 6770 7732 4993 6684 4708 5416 3337 5796 2308
G122 7534 5764 5367 6557 6361 6038 5354 5372 2828 5686 2262
G123 7204 4931 5715 6188 3785 4023 4448 5335 3403 5003 2467
G124 7935 5347 6012 6788 3820 3980 4982 4779 4011 5294 2640
G125 7532 6598 6385 6657 4910 5723 4945 5851 3860 5829 1918
G126 6790 5174 5670 7375 4667 6082 5207 5794 4264 5669 1741
G127 6650 5035 5240 8269 4674 6640 5139 5510 3894 5672 2307
G128 6607 6841 5750 7250 5181 6679 5157 5469 3439 5819 2023
G129 6577 5417 5270 7151 4848 6580 5650 5719 3467 5631 1944
G130 7890 5868 6157 7907 5132 6515 5324 7204 2633 6070 2699
G131 6667 5451 6397 6600 4771 6327 4774 4985 2388 5373 2542
G132 7486 4931 5819 7250 4299 6452 4749 6203 3715 5656 2323
G133 6646 4479 6429 7563 5104 5967 4733 5760 3181 5540 2377
G134 7357 4306 5258 6107 4868 6130 4482 5957 2669 5237 2591
G135 7430 5591 6657 7000 5014 6302 4671 5991 3649 5811 2079
G136 7782 5486 6704 7750 4660 6235 4699 6075 4129 5946 2226
G137 7228 5451 6135 6569 5083 5464 5275 5822 3100 5570 2065
G138 7131 4549 6297 6801 4472 5390 4811 5900 3147 5388 2368
G139 6367 4861 6218 6450 4403 4819 5102 5457 3182 5206 2038
G140 7032 5695 4884 6575 4354 7034 4891 6103 3243 5534 2339
G141 7282 4375 6548 6363 4410 5525 4369 6991 3539 5489 2479
G142 6582 5729 6454 5544 4563 6410 4705 6060 3407 5495 1948
G143 7263 3854 5912 6013 4188 5729 4483 5429 4338 5245 2113
G144 7127 6250 5513 6432 5104 6353 5080 5619 4094 5730 1591
G145 5896 6771 6220 7419 4514 6667 4679 5300 4690 5795 1819
G146 7291 5695 4715 4957 5278 7292 3820 5463 2883 5266 2735
G147 7544 6042 5565 6676 4528 6974 4696 5913 3446 5709 2285
G148 7921 6875 5500 5207 4931 6182 5104 6291 3113 5680 2396
G149 7767 6597 5060 6113 4702 6363 4896 5597 3558 5628 2202
G150 7511 5695 6217 8000 4417 6960 4951 5969 3149 5874 2619
Mean 7193.1 5507.1 6014.0 6709.2 4651.6 6037.4 4889.6 6010.1 3509.4
Min 5896 3854 4715 4957 3278 3892 3820 4779 2388
Max 8292 6875 7109 8269 6361 7292 5831 7910 4690

Table 5 AMMI analysis of grain yield for 50 wheat genotypes grown in nine environments in India 

Source of variation df SS MS %TSS
Environments 8 100810.13 12601.27* 73.77
Genotypes 49 8190.60 167.16* 5.99
G x E 392 27645.86 70.53* 20.23
PC1 56 7724.79 137.94* 27.94
PC2 54 5165.13 95.65* 18.68
PC3 52 3670.32 70.58* 13.28
PC4 50 3261.15 65.22* 11.80
PC5 48 2962.25 61.71* 10.71
PC6 46 2228.09 48.44* 8.06
PC7 44 2000.27 45.46* 7.24
Error 450 9154.99 20.34

* Significant at 0.001% probability

Eight sectors were observed and genotype G143 clustered with E9 and E3, indicating repeatable performance. Genotype 110 clustered with E3, 144 with E5 and 146 with E2, indicating that these genotypes are stable in the respective environments. The genotypes 111 and 137 were relatively closer to the biplot origin and could be good enough for E7 while 117, 130 and 144 were relatively closer to the biplot origin and could be good enough for E5, with average adaptation. Environment (E7) contributed most to the phenotypic stability of these genotypes (Figure1b), whereas E5, E9 and E3 contributed most to the GxE interactions. Tekdal and Kendal (2016) reported 10 sectors with respect to the mega environments and few lines were recommended for each environment.

The AMMI-based stability parameter, the AMMI stability value (ASV), was calculated based on the first two PCAs to produce a balanced measurement between them, and can be useful in situations where the two first IPCs explain a considerable part of the GxE interactions (Table 5). According to ASV, genotypes G111, G131, G112, G129, G137, G132, G105, G126, G135 and G142 were identified as stable for having lower ASV values, whereas genotypes G115, G123, G120, G107 and G119 were identified as being more unstable (Table 6). According to parameter Di, the values of G137, G111, G135, G132 and G112 were the lowest, whereas that of G146 was highest, followed by G124, G115, G119 and G123. The genotypes G125, G135, G104, G112, G136, G144, G105, G132, G126 and G129 were the most stable and high-yielding genotypes based on the genotype stability index, which takes both the overall mean yield and ASV into consideration (Table 6). The AMMI based AMMI stability parameters were used to screen durum wheat for identification of stable lines and were found to be adequate for the identification of stable genotypes (Mohammadi and Amri 2013, Alam et al. 2015). Similarly, AMMI stability parameters were also used to identify stably performing barley lines in Iran and were found to be promising in the identification of stable barley lines (Vaezi et al. 2017).

Table 6 Mean grain yield (kg ha-1) of 50 genotypes in nine environments and estimates of AMMI stability parameters 

Genotypes Gm RGm PC 1 PC 2 SIPCA Di ASV RASV GSI
G101 58.67 10 12.25 -3.39 367.75 19.18 18.56 31 41
G102 52.05 47 -16.84 -5.11 375.93 19.39 25.61 43 90
G103 56.32 26 -7.15 7.09 265.54 16.30 12.79 17 43
G104 59.65 7 9.47 -1.32 268.11 16.37 14.17 19 26
G105 56.67 23 1.11 -5.77 161.31 12.70 6.00 7 30
G106 59.85 5 13.60 4.60 337.79 18.38 20.78 36 41
G107 62.79 1 19.44 -1.72 507.44 22.53 29.02 47 48
G108 52.62 43 -15.28 6.63 396.49 19.91 23.71 40 83
G109 55.32 33 -10.52 5.16 191.37 13.83 16.51 29 62
G110 54.06 37 -12.77 0.82 250.15 15.82 19.05 34 71
G111 55.57 30 -1.45 -0.67 10.77 3.28 2.26 1 31
G112 56.39 25 0.00 -3.47 67.52 8.22 3.47 3 28
G113 52.04 48 -16.20 4.32 393.93 19.85 24.53 41 89
G114 58.31 11 12.00 -1.94 222.81 14.93 17.98 30 41
G115 49.36 50 -24.90 -2.46 653.41 25.56 37.18 50 100
G116 54.15 36 -6.19 -11.25 204.80 14.31 14.54 20 56
G117 59.71 6 10.83 3.31 129.74 11.39 16.47 28 34
G118 52.83 41 -12.64 0.55 238.49 15.44 18.85 33 74
G119 62.10 2 15.91 16.35 573.16 23.94 28.80 46 48
G120 61.76 3 20.17 2.20 430.53 20.75 30.13 48 51
G121 57.96 15 8.52 8.33 206.72 14.38 15.19 23 38
G122 56.86 19 4.48 -9.21 198.69 14.10 11.38 16 35
G123 50.03 49 -21.31 -0.36 561.80 23.70 31.76 49 98
G124 52.95 40 -16.63 6.45 671.59 25.92 25.60 42 82
G125 58.29 12 6.01 -0.71 141.31 11.89 8.98 11 23
G126 56.69 22 -1.49 6.38 92.39 9.61 6.75 8 30
G127 56.72 21 2.52 10.50 374.97 19.36 11.15 14 35
G128 58.19 13 13.25 -1.78 299.89 17.32 19.83 35 48
G129 56.31 27 2.76 0.08 135.20 11.63 4.11 4 31
G130 60.70 4 14.08 6.93 352.04 18.76 22.09 39 43
G131 53.73 39 -0.56 -2.55 70.33 8.39 2.68 2 41
G132 56.56 24 -0.02 5.42 48.18 6.94 5.42 6 30
G133 55.40 31 -4.46 9.15 215.79 14.69 11.31 15 46
G134 52.37 45 -8.88 -8.02 247.68 15.74 15.47 26 71
G135 58.11 14 4.55 3.76 35.56 5.96 7.75 9 23
G136 59.47 8 5.34 12.19 181.87 13.49 14.56 21 29
G137 55.70 29 -2.95 -1.12 10.59 3.25 4.53 5 34
G138 53.88 38 -10.16 4.51 133.61 11.56 15.79 27 65
G139 52.06 46 -14.72 1.74 227.33 15.08 22.00 38 84
G140 55.34 32 5.14 -7.45 129.47 11.38 10.69 13 45
G141 54.89 35 -9.80 2.99 296.90 17.23 14.90 22 57
G142 54.95 34 -0.34 -8.93 90.61 9.52 8.94 10 44
G143 52.45 44 -17.42 0.11 309.95 17.61 25.96 45 89
G144 57.30 17 5.43 -6.17 121.70 11.03 10.17 12 29
G145 57.95 16 9.71 5.04 309.91 17.60 15.32 25 41
G146 52.66 42 2.58 -25.63 735.85 27.13 25.91 44 86
G147 57.09 18 8.57 -4.69 102.82 10.14 13.60 18 36
G148 56.80 20 7.58 -18.65 505.04 22.47 21.80 37 57
G149 56.28 28 6.55 -11.76 259.10 16.10 15.28 24 52
G150 58.74 9 10.84 9.50 227.25 15.07 18.74 32 41

Gm-Genotype mean yield, ASV-AMMI stability value, Di- AMMI Distance, GSI -Genotype stability Index

GGE biplot analysis

The GGE biplot analysis was used to identify the best line of each environment and assess the stability of the lines. The most attractive feature of GGE biplots is the ‘which-won-where’ analysis, in which crossover GE interaction, mega-environment differentiation and specific genotype adaptation are graphically represented (Rakshit et al. 2014, Oral et al. 2018). The visualization of a ‘which-won-where’ pattern in multi-environment trials is essential to study the possible existence of different mega-environments in a region (Yan and Tinker 2006). The vertex genotypes were the most responsive for being located at the greatest distance from the biplot origin. The genotypes with either the best or poorest performance in one or all environments were considered responsive (Yan and Tinker 2006), falling within the sectors. In the biplot, the equality l.ine divides the graph into six sectors and nine environments were retained in two sectors (Figure 2), probably due to latitudinal and longitudinal differences. The test locations could be partitioned into two mega environments, one with E1, E8, E3, E4 and E7 and the second with E5, E2 and E6. In the first mega environment, the genotypes G119 and G120 were the winning genotypes and genotypes G112, G107, G148 and G146 in the second. There were strong correlations between environments located within the same sector, and variation in the genotype performance within environments indicated strong environmental influence and the existence of mega environment (Oral et al. 2018). The GGE biplot is a tool of data visualization that allows an evaluation of environments due to the discriminative ability and representativeness of the GGE view, which is an advantage over the AMMI biplot analysis (Yan et al. 2007, Aktas 2016).

Figure 2 GGE bi-plot showing: a) “which-won-where” pattern for genotypes and environments; b) discriminating ability and representativeness of environments for grain yield; c) the relationship between mean grain yield and wheat stability. 

Identification of ideal genotype based on GGE biplot analysis

The relationship among test environments was studied based on environment-centered (centering, 2) and environment-metric preserving (SVP, 2) without scaling option. Regarding grain yield, E2 and E6 were the most discriminating environments, whereas E5, E1 and E8 were the most representative environments, indicating their adequacy as test environments for multi-environmental trials (Fig. 2b). Environment E8 was closest to the mean environment, followed by E1 and E5. The genotype ranking in the closest to average, i.e., the most representative environments (E1 and E8), showed that genotype G107 yielded highest, followed by G120, G106 and G130. For selection of generally adapted genotypes, E2 was found most suitable based on both descriptiveness and representativeness, while E1 and E8 were found to be most suitable based on representativeness for grain yield.

According to Yan and Tinker (2006), an ideal genotype should have both high mean yield and high stability within a mega-environment. In fact, an ideal genotype should have the highest PC1 score (high yielding ability) and lowest (absolute) PC2 score (high stability) (Rakshit et al. 2014, Yan and Tinker 2006). The genotypes were ranked for ideal grain yield performance, in other words, with high yield performance and stability across the nine locations (Figure 2c). The biplot defined genotypes with longest vectors coupled with zero G×E, represented by dots and arrows, as stable and high yielding. The ideal genotype 129 was stable as its projection on the AEA was close to zero. Other promising genotypes near the ideal genotype were G111, G131, G135 and G112. The low yielding genotypes (G115, G123, G143, G102, G139 and G113) were located far away from the ideal genotype. Kendal and Sener (2015) also identified mega environments for durum wheat in Turkey and identified suitable stable and ideal genotypes for grain yield and quality traits. Both the AMMI and GGE Biplot approaches proved equally effective in the identification of stable and high yielding genotypes (G129, G111, G131, G135 and G112), as also reported by Aktas (2016). On the other hand, the GGE biplot has the advantage of a higher discriminative ability and representativeness of the GGE plot than the AMMI biplot.

CONCLUSION

The GE interaction along with the genotype and environment main effects among 50 genotypes evaluated at nine locations were found to be significant.

Both approaches, AMMI and GGE biplot, allowed the identification of common genotypes (G129, G111, G131, G135 and G112) that are stable and high yielding across all locations.

The genotypes G112 and G135 were identified as high yielding and stable across all nine locations. Therefore, these lines can be included in the national testing program, to be released as a variety.

ACKNOWLEDGEMENT

The authors acknowledge the support of CIMMYT, for providing advanced breeding lines from the CIMMYT Elite Spring Wheat Yield trial (ESWYT). Authors are also thankful to wheat breeders for their support in conducting ESWYT trials at different locations in India.

REFERENCES

Ahmadi J, Mohammadi A and Najafi-Mirak T (2012) Targeting promising bread wheat (Triticum aestivum L.) lines for cold climate growing environments using AMMI and SREG GGE Biplot analyses. Journal of Agricultural Sciences and Technology 14: 645-657. [ Links ]

Aktas H (2016) Tracing highly adapted stable yielding bread wheat (Triticum aestivum L.) genotypes for greatly variable South-Eastern Turkey. Applied Ecology and Environmental Research 14: 159-176. [ Links ]

Alam MA, Sarker ZI, Farhad M, Hakim MA, Barma NCD, Hossain MI, Rahman MM and Islam R (2015) Yield stability of newly released wheat varieties in multi- environments of Bangladesh. International Journal of Plant & Soil Science 6: 150-161. [ Links ]

Crossa J (1990) Statistical analysis of multi-location trials. Advances in Agronomy 44: 55-85. [ Links ]

Kendal E and Sener O (2015) Examination of genotype × environment interactions by GGE biplot analysis in spring durum wheat. Indian J. Genetics 75: 341-348. [ Links ]

Mohammadi R and Amri A (2007) Comparison of parametric and non-parametric methods for selecting stable and adapted durum wheat genotypes in variable environments. Euphytica 159: 419-432. [ Links ]

Mohammadi R and Amri A (2013) Genotype x environment interaction and genetic improvement for yield and yield stability of rainfed durum wheat in Iran. Euphytica 192: 227-249. [ Links ]

Munaro LB, Benin G, Marchioro VS, Franco FA, Silva RR, Lemes da Silva C and Beche E (2014) Brazilian spring wheat homogeneous adaptation regions can be dissected in major mega environments. Crop Science 54: 1-10. [ Links ]

Oral E, Kendal E and Dogan Y (2018) Selection the best barley genotypes to multi and special environments by AMMI and GGE biplot models. Fresenius Environmental Bulletin 27: 5179-5187. [ Links ]

Purchase JL, Hatting H and Vandeventer CS (2000) Genotype × environment interaction of winter wheat (Triticum aestivum L.) in South Africa: Stability analysis of yield performance. South African Journal of Plant and Soil 17: 101-107. [ Links ]

R Core Team (2013) R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. Available at: < Available at: http://www.R-project.org/ >. Accessed on April 14, 2019. [ Links ]

Rakshit S, Ganapathy KN, Gomashe SS, Swapna M, More A, Gadakh SR, Ghorade RB, Kajjidoni ST, Solanki BG, Biradar BD and Prabhakar A (2014) GGE biplot analysis of genotype × environment interaction in rabi grain sorghum [Sorghum bicolor (L.) Moench]. Indian Journal of Genetics and Plant Breeding 74: 558-563. [ Links ]

Sharifi P, Hashem A, Rahman E, Ali M and Abouzar A (2017) Evaluation of Genotype × Environment Interaction in Rice Based on AMMI Model in Iran. Rice Science 24: 173-180. [ Links ]

Tekdal S and Kendal E (2018) AMMI model to assess durum wheat genotypes in multi- environment trials. Journal of Agricultural Sciences and Technology 20: 153-166. [ Links ]

Tonk FA, Ilker E and Tosun M (2011) Evaluation of genotype x environment interactions in maize hybrids using GGE biplot analysis. Crop Breeding and Applied Biotechnology 11: 1-9. [ Links ]

Vaezi B, Pour-Aboughadareh A, Mohammadi R, Armion M, Mehraban A, Hossein-Pour T and Dorii M (2017) GGE Biplot and AMMI Analysis of Barley Yield Performance in Iran. Cereal Research Communications 45: 500-511. [ Links ]

Yan W and Kang MS (2002) GGE-biplot analysis: A graphical tool for breeders, geneticists, and agronomists. CRC Press, Boca Raton, 288p. [ Links ]

Yan W and Tinker NA (2006) Biplot analysis of multi-environment trial data: Principles and applications. Canadian Journal of Plant Science 86: 623-645. [ Links ]

Yan W, Kang MS, Ma B, Woods S and Cornelius P (2007) GGE Biplot vs. AMMI analysis of genotype-by-environment data. Crop Science 47: 643-655. [ Links ]

Zhang Z, Lu C and Xiang ZH (1998) Analysis of variety stability based on AMMI model. Acta Agronomica Sinica 24: 304-309. [ Links ]

Zobel RW, Wright MJ and Gauch HG (1988) Statistical analysis of a yield trial. Agronomy Journal 80: 388-393 [ Links ]

Received: January 21, 2019; Accepted: April 16, 2019

Creative Commons License This is an open-access article distributed under the terms of the Creative Commons Attribution License