Multi-Environment Evaluation of Soybean Genotypes in Nepal: Application of AMMI and GGE Biplot for Improving Crop Stability and Productivity

1. Introduction

Soybean (Glycine max (L.) Merrll is a self-pollinating diploid crop with 2n=2x=40 chromosomes and a genome size of 1100 Mbp[2]) that has been known to play an important role in the world of agricultural and nutrition. Locally called Bhattmas or Bhatta in Nepali, it is also called the golden bean, miracle bean, crop of the planet, and gods sent golden bean because of its large scale of usability, and is thus known as the Golden gift of nature to mankind. Soybean, being the major crop in the global agricultural economy is the most important seed legume. It contributes a quarter of the world stock of edible oil and about two-thirds of the protein concentrate in the world that is fed to livestock. At the moment, it is the leading oilseed in the country and in the world arena. It is a multifaceted nutritional food crop that is a source of low-cost high-quality protein (40-50 percent), high digestibility vegetable oil (20 percent) with zero cholesterol, carbohydrates/ soluble sugars (35 percent), and ash (5 percent). It contains also a lot of vitamins B, C, E and minerals, and has medicinal properties [9,16,17,11,13,14, 15]. In addition, soybean has one of the high rates of biological fixation of nitrogen and can fix up to 180 kg N/ha symbiotically. This legume is widely grown in Nepal, and it is specifically in the Mid Hills where the legume is grown that it covers nearly 80 percent of the soil and the total production of soybean in the country [6]. It is widely cultivated in the Terai and inner Terai areas as a mono crop because of its great yield capacity and intense need of soy meal and cake meal by the poultry and the livestock industry. Soybean is a great supplement food crop to cereals, and particularly in developing countries where malnutrition is a widespread situation. It can be used in the processing of seeds into baby foodstuffs and other products that are available in the market to diabetic patients, which include soy milk, tofu and soy sauce. The green pods are eaten in fresh form like vegetables and the dry seeds are usually roasted or fried into healthy snacks. Its green leaves can be used as green manure and fodder. Soybean proteins have many health advantages such as lowering the cholesterol levels, reducing the symptoms of menopause and lowering the chances of many chronic ailments, including cancer, heart disease and osteoporosis. Soybeans contain isoflavones (daidzein and genistein) which help in reducing cholesterol, decreasing cancer risk, and alleviating the probability and consequences of osteoporosis.

The soybean crop is also very adaptive to a wide range of agro-ecological environments since it can grow at altitudes as low as 500 to 1500 meters above the sea level [6]. It is the third largest acreage and second largest production grain legume in Nepal after lentils and black gram. It is commonly cultivated in rice bunds or intercrops like the maize in the mid-hills. The hectares of soybean in the world amount to 137,927,342, which produces 388,097,784.67 tons, with a production of 2.81 tons/ha [25]. In Nepal, soybean is planted in 24, 921 hectares, yields 35, 138 tons, and the average productivity is 1.410 tons/ha [12]. The number of acres of soybean acreage in Nepal increased by a factor of 10 between 2015-16 and 2020-21, the production level rose by 11 percent, and the productivity by 4 percent. The same period saw Nepal import 1,275, 201 tons of soybean oil worth $608.865 million USD and export 331,764.5 tonnes worth $59.65 million USD. Also, imports also consisted of 79730.9 tons of soybean grains valued at 4.9799 USD, 810.7 tons of soybean sauce worth 0.587 USD and 244337 tons of soybean cake valued at 15.4905 USD. These data show that Nepal is reliant on imports of raw materials which are essential in their feed and food sectors, which means that they are not self-reliant. The environmental variability also has a great impact on the growth of soybean and causes large levels of genotype-and-environment-interactions (G×E) which leads to inconsistent genotypic responses. Genotypes may not be able to perform consistently relative to each other in different environments because of the existence of large Gx E interactions on a trait such as seed yield [8]. Multi-environment trials (MET) have therefore been extensively used by plant breeders to compare the relative performance of genotypes across multiple target environments. These experiments play an important role in the determination of the adaptability and the stability of the genotypes and thus help in the selection of the better and more robust types [10]. Ideal stable genotypes are those genotypes that have the genetic ability to perform well in favorable conditions and are also able to give satisfactory yields in unfavorable conditions. Many statistical techniques such as regression analysis have been created to test interaction patterns of G×E [7]. These include, most successfully, the Additive Main Effects and Multiplicative Interactions (AMMI) and the Genotype-Genotype by Environment interaction (GGE), which are helpful in breaking down the additive (linear) and multiplicative (bilinear) aspects of G×E interaction, providing useful insights into the interpretation of multi-environment data to breeding programs. A number of statistical models have been put forward to use the positive G + E interactions and assist in breeding decisions in terms of cultivar selection and recommendation to particular environments. Thus, the research will evaluate the soybean genotypes with the adaptability and stability of grain yield under various conditions by the AMMI and GGE bi-plot models with a soybean breeding improvement model.

2. Materials and Methods

2.1 Experimental sites

Soybean yield trials were carried out in the field of seven different environments across the country, of which five of them (Surkhet, Rampur, Khajura, Dailekh, Doti) in the 2019/2020 crop seasons and five different environments (Salyan, Doti, Dailekh, Khajura and Nawalpur) in the 2020/21 crop season. Geographical details are depicted in the

2.2 Experimental details

2.2.1 Experimental lay out and research design: This experiment received an array of eighteen soybean genotypes determined in 2019 season and 16 different soybean genotypes determined in the rainy season of 2020, which were designed using a Simple Alpha Lattice Design. In 2019 and 2020 respectively the experiment was repeated with 3 x 6 block replications and 4 x 4 block replications respectively.

2.2.2. Sowing: Sowing was carried out on 1st to the end of July 2019- 2020 in the monsoon season and the crop geometry was 50cm x 10 cm (row to row x plant to plant) with seed being sown manually. The depth of planting was 4-5cm, and the rate of seeding was 60 kg/ha, which was the recommended rate. Each plot was 12 m², or 4-meters x 3 meters, and in six rows. A total number of 36 plots were used in the research. In 2019 and 2020, the size of the blocks had an area of 216 and 192 m², respectively. The extra thinning of the plants was done following the initial weeding after sowing of 25 days.

2.2.3 Fertilizers: The chemical fertilizer dosage recommended was to be applied as a basal dose at 20 kg nitrogen (N), 40 kg phosphorous (P₂ O₅) and 20 kg potassium (K₂ O) per hectare at the final land preparation.

2.2.4. Intercultural operations: Pendimethaline herbicide was undertaken before 3 days following sowing in the field. At 25 days of sowing, first manual weeding was performed and at 45-50 days of sowing, second manual weeding was carried out, and earthing up was then done.

2.2.5 Disease pest: Disease and pest incidences were rated using protocol that had been developed by plant pathology research centre. The disease-pest occurrences were controlled in terms of need to apply the fungicides and insecticides, in the event that it becomes economically disadvantageous. Spodolure laden pheromone traps were also placed to reduce the infestation of the Spodoptera in the field.

2.2.6 Plant genetic materials: The genetic materials used in this experiment were all acquired in International Institute of Tropical Agriculture (IITA), Nigeria and National Agriculture Genetic Resources Centre (NAGRC- Gene Bank) and variety puja were standard check with eighteen and sixteen soybean genotypes (Annex i and ii).

Annex: i Soybean genotypes used for multi-environment yield trial at different locations, 2019

Annex: ii Soybean genotypes used for multi-environment yield trial at different locations, 2020

Data Collection:

The quantitative traits were measured on 10 randomly selected plants, and then IBPGR Descriptors (Anonymous, 1985). The quantitative data was measured in plant basis (plant height cm), pods/plant (P/P) and seeds/ pod (S/P) and quantitative parameters of morphology data were entered as days to 50 percent flowering (DF), days to maturity (DM), 100-seed weight in gram (HSWT), and yield/ gram (GY). The results of these samples were used to estimate the performance of each genotype with regard to the traits of interest.

Statistical analysis

The mean values of the traits were determined and used as the input in the analysis of variance (ANOVA). In order to estimate the genetic parameters, R-stat version and GEA-R software were used to analyze the results with the main emphasis on the main effects of environment (E), genotype (G), and their interplay (GE). Such a major GE interaction variance led to the further division of this component. Further analysis of the GE interaction was performed with the help of GGE biplot model [19]. The GGE biplot has been drawn on the basis of the first two principal components (PC1 and PC2) which were calculated using the data of environment-oriented yields [20].  The GGE model used was:

Where Yij is measured mean of genotype i(=1,2,….,n) in environment j(=1,2…,m), µ is the grand mean, βj is the main effect of environment j, µ + βj being the mean yield across all genotypes in environment j, λ1 and λ2 are the singular values (SV) for the first and second principal component (PC1 and PC2), respectively, ξi1 and ξi2 are eigenvectors of genotype I for PC1 and PC2, respectively, ŋ1j and ŋ2j are eigen vectors of environment j for PC1 and PC2, respectively, εij is the residual associated with genotype i in environment j. PC1 and PC2 eigen vectors cannot be plotted directly to construct a meaningful biplot before the singular values are partitioned in to the genotype and environment eigenvectors. Singular value partitioning was implemented by,

Where f1 is the portion factor for PC1. The f1 can range between 0 and 1. To visualize relationship among genotypes, the GGE biplot based on genotype metric (that is, f=1; S.V.P=1) is appropriate and environment metric (f=0; S.V.P=2) GGE biplot is important to visualize relationship among environments. So, the following formulae from equation (1) were formulated to generate the GGE biplot:

If the data were environment-standardized, the common formulae to generate the GGE biplot were as follows:

Where sj is the standard deviation in environment j, i=1, 2…., k, gi1and e1j are PC1 scores for genotype i and environment j, respectively. In the present study we used environment standardized model, Equation (4).

Results and Discussion:

Combined analysis of variance for yield and yield attributing traits

The overall ANOVA among the varying environments showed the existence of very significant differences in the soy bean grain yield and the allied characteristics such as the days to flowering, days to maturity, plant height and hundred seed weight, the differences of which were mostly as a result of genotypic variation (G). There were also great disparities in agronomic characteristics like grain yield, days to flower, days to maturity, and the quantity of seeds to the pod as a result of genotype-by-environment interaction (GEI). The environments were found to have a significant difference on Agronomic traits days to flowering, days to maturity, plant height, and hundred seed weight (Table 2). The environment led to the highest increase in the proportions of grain yield parameters (27.84%), and genotype led to the low percentage of grain yield parameters (16.83), and interaction between genotype and environment (55.31) in 2019. The environment is the most likely source of the variation in the parameters of grain yield in 2020 with 65.8 percent of the variation attributed to the environment, 5.14 percent of the variation attributed to genotype and 29.04 percent of variation attributed to genotype x environment interaction. The combined AMMI analysis of the Gollobs test of the locations over the years showed that the environment explained the highest percentage (37.39) of variation in the parameters of grain yield whereas genotype only explained 7.027% and the genotype x environment interaction only explained 55.571% of total variation. The results of analyzing the data revealed a range of yield production of soybean genotypes investigated between 1135 kg ha-1 and 1831 kg ha-1. The overall yield of all the test genotypes was 1485 kg ha^-1 of grain. Genotypes TGX1485-ID had the highest grain yield of 1831 kg ha^-1, genotype G1873 had the highest grain yield of 1668 kgha^-1 and SB0095 had the highest grain yield of 1543kgha^-1 respectively (Table 3). A combined analysis of variance was done to explain the main effect and to measure the interactions between and among the sources of variability. Table 2 provides the results of the pooled analysis of variance. Genotypes(G) and GEI showed significant differences (p≤ 0.01, p≤0.05) in the genotype (G) and GEI of the genotype of grain yield, flowering days and maturity days.

GGE bi-plot analysis

The two most significant sources of variation to evaluate cultivar in a multi environment trials are the genotype main effect (G) and the genotype x environment interaction (GE), which are referred to as GGE [22]. A GGE bi-plot is used to visualize the genotypic main effect (G) as well as the genotype by environment interaction (GE) of a genotype- by-environment dataset [20]. This is a bi-plot that works exceptionally well with mega-environment analysis because it is able to visually depict the genetic correlations among environments and the which-won-where patterns. It makes it easy to evaluate the test environments in terms of their discriminating ability and representativeness. Besides, the bi-plot helps to assess the genotype based on their average performance and stability in various mega- environments. The current data set revealed that GGE bi-plot graphically represented interrelationships between the test locations in addition to the genotypes and PC 1 and PC 2 accounted 79.9% with PC 1 = 61.5% and PC 2 = 18.4% of the variation in the yield of grain.

Mega-environment analysis

GGE biplot is used to create the best polygons to see or visualize the pattern of interaction of genotype x environment [21]. The polygon view affirms of the format of Which-won-where, a visualization of which would be useful to approximate the potential presence of various mega-environs in the target environment [23,20,24,]. Figure 1 is a polygon illustration of eight soybean genotypes using seven environments. With this biplot, a polygon was drawn between the vertex genotypes which were furthest away in the various directions which led to the other genotypes being put in the interior of the polygon. The genotypes of the vertex are GC8234GC-13, TGX1485-ID, SB0122 and TGX1805-31F. The polygon will be broken into four sectors, and one of these sectors will not have any test environments. The analysis of the GGE biplot suggests the presence of location specific adaptability: GC8234GC-13 is better performed in Nawalpur and Sarlahi; TGX1485-ID is more adapted to Khajura, Salyan, Surkhet and Rampur; SB0122, G1873, and Kavre is more likely to be productive in Doti and Dailekh; TGX1805-31F does not have a particular environment preference. It is interesting to note that Doti in the two years has continuously performed within the same sector and this may indicate that it has been able to perform at the same location. This trend indicates a recurrent mega-environmental analogous situation of where-won-which, and it substantiates the demarcation of mega-environment. Notably, the conditions of defining mega-environments focus on the pattern of similar environments (which-where) as opposed to a mere similarity in the environment [23,21]. Therefore, Doti can be regarded as a unique mega-location to conduct soybean variety examination and assessment.

Evaluation of Test Environments for Genotype Discrimination and Representation

The main aim of test-environment evaluation is to establish the type of environment that can differentiate superior genotypes in a mega-environment. They ought to have an ideology test environment that has the capability to discriminate between genotypes and also a high level of representativeness to the greater mega-environment [22]. Figure 2 shows the discriminating power verses representativeness of the GGE biplot, through environment centered scaling [23]. In the present analysis the single values are completely assigned to the environment scores (SVP = 2). In the case when the data are not scaled (scaling = 0), the length of the line between an environment marker and the biplot origin deals with the discriminating power of the environment. The longer the vectors in the environment, the better they discriminate among the genotypes whereas the very short vectors give minimal information about the genotype differences [22]. Doti, Surkhet, Khajura, and Salyan became the most discriminating in this research. Nevertheless, the vector of Doti was nearer to the concentric circle meaning that it provides little information regarding the genotypic variation. The second important feature of test environment analysis is its reflectiveness of the mega-environment. This is plotted using the angle between the environment vector and the average environment axis (the abscissa). Assuming that the SVP = 2, the cosine of this angle is close to the correlation coefficient between the performance of the genotype in that environment, and the overall performance. The smaller the angle, the greater the representativeness [24, 22]. Under this criterion, Doti, Surkhet, Khajura and Salyan were found to be more representative environments where soybean regional trials can be carried out. It is important to note that, the vectors of Doti and Surkhet were very close to the average environment axis and therefore they were the most appropriate to use in the selection of better soybean genotypes to use in these growing regions.

Genotype Evaluation

Average Yield and Stability Performance of Soybean Genotypes in Comparison with an Ideal

To determine the stability and mean yield of eight soybean genotypes, the biplot analysis was done to determine the position of the genotypes and their placement behind an ideal genotype as shown in Figure 3. The line which goes through the origin and is parallel to the average tester coordinate (ATC) in this biplot is a representation of the overall average performance of the genotypes [21]. The ATC suggests that there is a perpendicular line between the ATC and the origin, which would mean that the genotypes are stable, and the further the genotypes are on the axis of stability, the more stable they are. The genotype possessing the best mean yield and absolute stability will be described as an ideal genotype and will fall at the middle of the concentric circle of the biplot. This is because this optimum is characterized by the longest vector of all of the genotypes and a minimal projection on the ATC axis indicating optimum performance and stability. Genotypes G1873, Kavre, and TGX1485-ID were identified to be the most desirable based on these criteria as they fall in the middle of the concentric circle which means that they have the highest mean yield but at the same time have the highest stability in the testing environment. On the other hand, other genotypes were prioritized based on the Euclidean distance between them and this ideal point: SB0095, Puja, SB0122, and GC8234GC-13, the genotypes that were most distant were regarded as less desirable because they perform and are less stable.

Discussion

There was performance of 18 soybean genotypes in five locations in the 2019 rainy season and 16 soybean genotypes in different locations in the 2020 rainy season in the present study. This however was not true because the genotypes differ amongst the years and as such, only eight genotypes were shared and thus combined to analyze the data on analysis of variance, AMMI, and GGE biplot. The analysis of variance between the environments and years indicated that the environment significantly affected the yield of soybean grains and accounted about 37.39 percent of the total variation. The model of genotype-by-environment interaction (GE) explained the almost twice the environment contribution and seven times the genotypic effect (7.07) of the total variation. This significant GE effect implies that there are different mega-environments, with each having different genotypes [21]. This has also been observed with Soybean in Nigeria [26], suggesting that performance of a genotype of soybean in one environmental setting is not stationary in another setting because of strong GE interaction. The observations in Table 3 of the soybean suggest that the GE interaction is of crossover type. To be more precise, there were five environments out of seven that produced various best-performing genotypes. This difference makes the process of selection more difficult and the development of genotype advice in breeding programs more complicated [5]. Huge and substantial GE interactions are recorded as well in soybean research carried out in Nepal, Ethiopia, and other African locations [3, 4, 8, 18], which are consistent with the results of Asfaw et al. (2009). As far as particular genotypes are concerned, GC8234GC-13, TGX1485-ID, SB0122, and TGX1805-31F were defined as the vertex genotypes, which means that they are more adaptable to specific conditions. GGE biplot analysis has effectively shown that the location-specific genotype GC8234GC-13 was more adaptive to the conditions in Nawalpur; TGX1485-ID was more adaptive to the conditions in Khajura, Salyan, Surkhet, and Rampur; and genotypes SB0122, G1873 and Kavre were more adaptive to the surroundings in Doti and Dailekh. It is worth noting that Doti in both years were concentrated within one sector and this may indicate uniformity of performance of the genotype in this place. The “Which-Won-Where” analysis divided the sites of testing into two mega-environments where one consisted of three sites with G-1873 being the most successful genotype.

Conclusion

Findings of the study showed that genotype x environment (G×E) interaction had a strong influence on the performance of the soybean yield and then the main effects of the genotype and the environment. Importantly, the magnitude of interaction between G and E was two to four times more than the magnitude of environmental and genotypic effects, respectively. The pattern of GxE interaction was shown to be of crossover type with a difference in the ranking of the genotypes in different environments. GGE biplots turned out to be one of the effective means of analyzing and demonstrating the trends of G x E interaction in soybean multi-environment trials and especially regarding test environments and genotype performance. According to the biplot result, genotypes G1873, Kavre, and TGX1485-ID fell at the center of the concentric circle meaning that the genotypes are characterized by large mean yield and large stability in the environments where the tests were carried out. They can therefore be regarded as the best genotypes to grow soybean. Also, there were existing mega- environments, Doti and Surkhet, which were close to the center of the biplot and on the average environment axis. This location suggests that such regions are appropriate in the selection and planting of better genotypes of soybean to farm in as they are the best locations to produce soybean.

Acknowledgements

The authors are grateful to the hardworking soybean research team of the Grain Legumes Research Program, which is involved in an earnest and sincere way. We owe the success of the multi-location trials to the highly valuable contributions of all the scientists, officers, technicians, chiefs, directors, and coordinators. We also wish to thank the Executive Director of NARC and the whole management team with their support and guidance that remained unchanged.

References

[1]. Agarwal, D. K., Billore, S. D., Sharma, A. N., Dupare, B. U., & Srivastava, S. K. (2013). Soybean: Introduction, improvement, and utilization in Indiaproblems and prospects. Agricultural Research, 2(4), 293-300.  https://doi.org/10.1007/s40003-013-0088-0

[2]. Arumuganathan, K., and E. D. Earle, 1991. Nuclear DNA content of some important plant species. Plant Mol. Biol. Rep. 9: 208–219.

[3]. Asfaw A, Alemayehu F, Gurum F, Atnaf M (2009). AMMI and SREG GGE biplot analysis for matching varieties onto soybean production environments in Ethiopia. Sci. Res. Essay 4(11):13221330.

[4]. Bueno RD, Borges LL, Arruda KMA, Bhering LL, Barros EG, Moreira MA (2013). Genetic parameters and genotype x environment interaction for productivity, oil and protein content in Soybean. Afr. J. Agric. Res. 8(38):4853-4859.

[5]. Comstock RE, Moll RH (1963). Genotype-environment Interactions. In: “Statistical Genetics and Plant Breeding”,(EdS.): Hanson WD and Robinson HF. National Academy of Sciences–National Research Council Publ. 982, NAS-NRC, Washington, DC, pp. 164–196.

[6]. Darai, R., D.B. Gharti and S. Subedi. 2016. Host resistance breeding against the virus diseases of soybean in Nepal. Int. J. Environ. Agric. Res., 2(10): 9-13.

[7].  EBERHART, s. A., AND RUssELL, w. A. 1966. Stability parameters for comparing varieties. Crop Science, 6, 36-40.

[8]. Gurmu F, Mohammed H, Alemaw G (2009). Genotype x Environment interactions and stability of soybean for grain yield and nutrition quality. Afr. Crop Sci. J. 17(2):87-99.

[9]. Hymowitz, T. (1970). On the domestication of the soybean. Econ. Bot. 24, 408–421. doi: 10.1007/BF02860745

[10].  Jha, P.N. Gupta , G. , Jha, P. , Mehrotra , R.  2013. Association of rhizospheric / endophytic bacteria with plants: a potential gateway to sustainable agriculture. Greener J. Agric. Sci., 3 (2013), pp. 73- 84

[11]. Medic, J., Atkinson, C., and Hurburgh, C. R. (2014). Current knowledge in soybean composition. J. Am. Oil Chem. Soc. 91, 363–384. doi: 10.1007/s11746- 013-2407-9

[12]. MoALD, 2021. Statistical information on Nepalese agriculture.

[13]. Rajendran, A., and Lal, S. K. (2020). Assessing the need of pre-germination anaerobic   stress-tolerant varieties in Indian soybean (Glycine max (L.) Merrill). Natl. Acad. Sci. Lett. 43, 593–597. doi: 10.1007/s40009-020-00937-9

[14]. Rajendran, A., Lal, S. K., Raju, D., and Ramlal, A. (2022). Associations of direct and indirect selection for pre-germination anaerobic stress tolerance in soybean  (Glycine max). Plant Breed. doi: 10.1111/pbr.13048

[15]. Ramlal, A., Dey, N., Sharma, U., and Rajendran, A. (2022). “In-silico studies to reveal the potential inhibitory capacity of soy isoflavonoids against angiotensin converting enzyme,” in Proceedings of the coreference: Sustainable future for humanity: The new learning curve, eds H. Chopra, V. Ghuriani, P. M. Arora, S. Babbar, and P. Baweja(Mumbai: Imperial Publications), 155167.

[16]. Rodrigues, L. R., Forte, B. D. C., and Bodanese-Zanettini, M. H. (2006). Isolation and culture of soybean (Glycine max L. Merrill) microspores and pollen grains. Braz. Arch. Biol. Technol. 49, 537–545. doi: 10.1590/S1516-89132006000500002

[17]. Singh, R. J., Kim, H. H., and Hymowitz, T. (2001). Distribution of rDNA loci in the genus Glycine Willd. Theor. Appl. Genet. 103, 212–218.

[18]. Tukamuhabwa P, Asiimwe M, Nabasirye M, Kabayi P, Maphosa M (2012). Genotype by environment interaction of advanced generation soybean lines for grain yield in Uganda. Afr. Crop Sci. J. 20 (2):107-115.

[19]. Yan W (2001). GGE Biplot-A Windows application for graphical analysis of multi environment trial data and other types of two- way data. Agron. J. 93:1111–1118.

[20]. Yan W, Hunt LA, Sheng Q, Szlavnics Z (2000). Cultivar evaluation and mega-environment investigation based on GGE biplot. Crop Sci. 40:597-605.

[21]. Yan W, Kang MS (2003). GGE Biplot Analysis: A graphical tool for breeders, geneticists, and agronomists. CRC Press, Boca Raton, FL.

[22]. Yan W, Kang MS, Ma B, Wood S, Cornelius PL (2007). GGE biplot vs. AMMI analysis of genotype-by-environment data. Crop Sci. 47:643-655.

[23]. Yan W, Rajcan I (2002). Biplot evaluation of test sites and trait relations of soybean in Ontario. Crop Sci. 42:11–20.

[24]. Yan W, Tinker NA (2006). Biplot analysis of multi-environment trial data: Principles and applications. Can. J. Plant Sci. 86: 623–645.

[25]. FAOSTAT (2021) Food and Agriculture Organization Corporate Statistical Database. http://www.fao.org/faostat/en/#data/QC

[26]. Jandong EA, Uguru MI, Oyiga BC (2011). Determination of yield stability of seven soybean (Glycine max) genotypes across diverse soil pH levels using GGE biplot analysis. J. Appl. Biosci. 43:2924-2941.