Simultaneous Selection Index as a Tool for Identication of Stable High Yielding Maydis Leaf Blight Resistant Maize Prebreeding Lines

Maize is a crop possessing high adaptability however, large differential genotypic responses have been reported when evaluated under multiple environments. Using randomized complete block design with two replications a total of 169 teosinte derived maize backcross inbred lines (BILs) were evaluated in three different environments namely, E2, E4 and E6 for maydis leaf blight (MLB) resistance and grain yield. Out of these, 73 BILs were identied displaying resistance to MLB in at least one of the environments and were subjected to additive main effect and multiplicative interaction (AMMI) analysis and genotype and genotype X environment (GGE) biplot analysis for identication of lines showing stable and high MLB resistance and grain yield. Highly signicant effects of genotype, environment and genotype X environment interaction (GEI) were observed for both the traits studied. AMMI ANOVA for percent disease index (PDI) revealed that highest percentage of total sum of squares (SS) was attributed to GEI (40.55%) while 32.86% and 26.59% was contributed by genotype and environment, respectively. For grain yield largest contribution of 68.02% towards SS was done by genotype component followed by GEI (17.50%) and E (14.48%). GGE biplot analysis identied two mega environments for both PDI (E2, E4/E6) and grain yield (E2/E4, E6). Based on AMMI stability value (ASV), genotype MT-90 (32) was observed to be most stable for PDI. While for grain yield highest stability was displayed by genotype MT-83 (28). Simultaneous selection index (SSI) helped in identication of ten stable high yielding MLB resistant genotypes namely, MT-120 (45), MT-14 (2), MT-166 (62), MT-148 (55), MT-190 (72), MT-37 (9), MT-19 (3), MT-114 (42), MT-77 (27) and MT-94 (35) which could be used in future breeding programmes either as donor of MLB resistance and grain yield or after combining ability analysis these genotypes could be used as parents for development of superior yielding MLB resistant hybrids.


Introduction
Maize, exhibiting highest genetic yield potential amongst different cereals thereby earning for itself the title "queen of cereals" is a crop of global repute. Grown in more than seventy countries (Anonymous 2018) along with rice and wheat maize provides 60% of the world's energy intake (Anonymous 2008) and contributes 39% of grain production globally. In Indian context, maize is the third largest produced and consumed crop after wheat and rice (Kumar et al. 2013) and contributes signi cantly towards the poultry industry (Hellin and Erenstein 2009). Like other crops, maize cultivars are also confronting with various factors during different developmental phases to realize its genetic potential while grown commercially at farmers' eld or grown at experimental site. Consequently, the crop growth and development is severely affected leading to sub-optimum outputs. Apart from abiotic factors that constitute all the non-living components and interacts right from the seed germination to maturity stages, maize crop is also affected by biotic factors, many of them are bene cial while others are harmful, interfering at different stages, from germination to maturity and during storage, leading to loss of green biomass, green ears including baby corn and sweet corn and, grain yield. Diseases caused by a fraction of biotic factors are 61 in numbers and have been reported to cause about 13.2% loss in economic product per annum (Payak and Sharma 1985;Kumar et al. 2013). Maydis leaf blight (MLB) is one of the diseases signi cantly affecting maize production in India and abroad.
A disease of historical signi cance due to its epidemic propositions in 1970 in the US, MLB is caused by a necrotrophic ascomycete fungi Bipolaris maydis. The disease can cause as high as 70% yield losses to maize production (Mubeen et al. 2017). Almost all the yield losses in India are attributed to race O of the pathogen as race C is con ned only to China (Wei et al. 1988) and race T affects Texas male sterile cytoplasm (cmsT) maize (Carson et al. 2004) which is not widely cultivated in the country. Though chemical fungicides can be used for controlling the disease, yet the additional input requirement and the associated ill effects necessitate the search of novel genes and alleles and their deployment to incorporate gene-based protection mechanism in the plant. In fact, development of tolerant genotypes by integrating novel alleles/genes seems to be the most feasible, attractive, cost effective and long-term alternative for management of MLB in maize. Resistance to MLB in maize is reported to be both qualitative (Faluyi and Olorode 1984;Zaitlin et al. 1993;Chang and Peterson 1995)and quantitative (Pate and Harvey 1954;Kumar et al. 2016) in nature. The development of MLB resistance using quantitative genetic factors is advantageous in many ways as it not only provides resistance against many pathogenic races but also prevents the evolution of new and more virulent pathogen variants. A number of MLB resistant varieties have been derived using maize germplasm belong to primary gene pool however prospects of wild relatives and biological progenitor of maize in incorporating resistance to MLB has remained largely unexplored. The narrow genetic base of cultivated maize is an important factor limiting the breeding of new maize varieties for high-yield and disease resistance (Wallace et al. 2014). Sourcing resistance from wild relatives is advantageous because they possess a plethora of novel resistance alleles on account of being exposed to di cult environments from times immemorial and still surviving, ourishing and thereby continuously evolving under the existing and emerging climatic conditions. The breeding goals would be easier to address if the enormous genetic variation present in wild progenitors is available to the breeders in a form, they could use in their breeding programs.
Teosinte (Zea mays ssp. parviglumis), the nearest and most probable wild progenitor of maize is interfertile with maize and produces viable hybrids Singh et al. 2017) and in certain parts of the world is still believed to be exchanging genes with maize naturally. Teosinte can be used as donor of pre-domestication alleles for the improvement of maize with respect to different traits (Liu et al. 2016;Kumar et al. 2019;Adhikari et al. 2019Adhikari et al. , 2021Singh et al. 2021;, Sahoo et al. 2021). This could be done by creation of diverse prebreeding lines which would serve as donor of novel genetic variation and could be employed to breed for high value characteristics such as MLB disease resistance. Only a single study has been performed by Lennon et al. (2017) in which maize wild relative Zea mays spp. parviglumis was used as a donor of MLB resistance alleles.
Development of resistant genotype is essential, however, besides being resistant, a genotype must be high yielding and stable in performance in order to be commercially viable and substantially accepted by maize growers over a wide range of agroclimatic conditions. Yield alike MLB resistance is a complex quantitative trait and greatly in uenced by external environment which may result in shift in scale or rank of the genotype performance when grown in diverse environments (Dia et al. 2016a). Differential genotypic response with respect to yield and disease resistance when grown in diverse environments have also been observed by Aina et al. (2007) and Ssemakula and Dixon (2007). Popularly called as the GxE interaction, the differential sensitivity of genotype performance to environments complicates the identi cation of superior genotype across the environments (Dia et al. 2016b). The presence of GxE interaction reduces the correlation between genotype and phenotype therefore slows the progress due to selection (Chalwe et al. 2017). Therefore, the nature and magnitude of GxE interaction must be taken into consideration during the identi cation of superior genotypes. Multi environment trials are an effective tool for identi cation of GxE interaction. A number of different methods (Wricke 1962;Eberhart and Russell 1966;Perkins and Jinks 1968;Shukla 1972;Francis and Kannenberg 1978;Lin and Binns 1988) are available for stability assessment however, the most commonly used methods for studying GxE interaction is the additive main effect and multiplicative interaction (AMMI) model and the genotype and genotype x environment (GGE) biplot (Yan et al. 2001).
The GGE biplot is a statistical tool utilized for examining the performance of genotypes tested in different environments. The which-won-where biplot helps in identi cation of mega environments and the winning genotypes for each mega environment (Yan 2001 . Selection of genotypes with stable MLB resistance and grain yield over a range of different environmental conditions and in the presence of variable disease pressure is an ideal strategy for making optimal progress in resistance breeding (Gyawali et al. 2019). Keeping all these observations in perspective a study was planned to generate diverse pre breeding lines by crossing maize inbred line DI-103 with MLB resistant teosinte (Zea mays spp. parviglumis) and using the MLB resistant inbred lines so produced for AMMI analysis for grain yield and MLB reaction score parameters with objective to identify genotypes that displays stable and high grain yield combined with resistance to MLB. Lines identi ed to be high yielding and MLB resistant can be used as potential germplasm in maize breeding programme.

Experimental materials
The genetic materials for the experiment were developed at Norman E. Borlaug Crop Research Centre, G. B. Pant University of Agriculture and Technology, Pantnagar, Uttarakhand, India. Wild relative of maize, teosinte (Z. mays ssp.parviglumis) was crossed as pollen parent with a superior maize inbred line DI-103 used as a seed parent. The F 1 s thus produced were subjected to a single generation of backcrossing with DI-103 followed by sel ng for four consecutive generations leading to development of BC 1 F 5 backcross inbred lines (BILs) population. A total of 169 BILs were subsequently utilised for the purpose of experimentation.

Experimental Layout
The evaluation of experimental material consisting of 169 BC 1 F 5 lines was done in Kharif 2018-2019. Each line was planted in a single row 2 m long and 75 cm apart. These lines were evaluated in Randomized Complete Block Design with two replications in three different environments namely E2, E4 and E6. The details of study environments along with magnitude of whether parameters during the crop season is presented in table 1.
Observation procedure Data on MLB reaction and grain yield was recorded for all the 169 BILs in three different environments. For the purpose of screening of BILs for MLB response by creating arti cial epiphytotic conditions in E2 and E4 the causal organism Bipolaris maydis, was aseptically isolated from locally collected infected maize leaves. The pathogen was initially cultured on sterilized solidi ed Potato Dextrose Agar (PDA) medium containing petri plates. For the purpose of mass multiplication of the pathogen the autoclave bags containing sorghum grains were seeded with 1.5x1.5cm rectangular pieces of Bipolaris maydis mycelium containing solidi ed PDA medium under aseptic conditions and incubated at a temperature of 25-27 0 C for 10 days with intermittent shaking after every 2-3 days to facilitate uniform mycelium growth on grains. After shade drying 15-20 mycellium covered sorghum grains were placed in the leaf whorls of 35 days old plant. The eld was frequently sprayed with water for effective spore germination and disease development. The entire process of mass multiplication and eld inoculation of pathogen have been presented in Fig.   1. For natural spread of MLB in E6, a susceptible check was sown after every ten BILs and also as a border crop so as to ensure uniform disease inoculum for all the lines. Data on disease severity was recorded 35 days after inoculation for each line in all three environments by following 1 to 9 rating scale of Hooda et al. (2018) ( Table 2). Disease rating was converted into percent disease index (PDI) by using the following formula: In order to assess grain yield per plant, ears from three plants per entry per replication per environment were harvested after physiological maturity and weighed. Average grain yield over three plants was calculated and expressed in gm and further utilized for analysis.

Statistical Analysis
Combined Analysis of Variance Combined Analysis of variance with respect to both PDI and grain yield was done. For both the traits the data was heterogenous at 5% level of signi cance as revealed by Bartlett's Chi-square test therefore, the data was transformed before combined ANOVA was carried out.
AMMI and GGE biplot analysis AMMI analysis was conducted for both PDI and grain yield in order to divide the total variation into variation due to genotype, environments, and interaction as per the model given by Gauch and Zobel (1996) as mentioned below: where, Yger = Performance of genotype g in environment e for replicate r, µ = grand mean, αg = genotype mean deviation (genotype means minus grand mean), βe = environment mean deviation, n = number of principal component analysis (PCA) axes retained in the model, n = singular value for PCA axis n, ygn = genotype eigenvector values for PCA axis n, δen= environment eigenvector values for PCA axis n, ρge = residuals, Eger = error term.
Both AMMI and GGE biplot analysis was done with the help of GEA-R (Genotype x Environment Analysis with R for Windows) software version 2.0.

AMMI stability value (ASV)
Stability of genotype with respect to MLB resistance and grain yield was assessed across environments by means ASV as suggested by Purchase et al. (2000). ASV was calculated as per the formula given below and was used to rank genotypes according to trait stability. Lower ASV signi es more stable genotypes.

Genotype Stability Index (GSI)
The rank of genotype (RT) based on overall mean of the genotype for a speci c trait across seasons and ASV rank of the genotype for that particular trait (RASV) were added up to calculate the GSI as given below and suggested by Chalwe et al. (2017).

GSI= RASV + RT
Genotype with lowest GSI was considered to be superior and stable for the trait under consideration.
Simultaneous Selection Index (SSI) GSI for both MLB resistance and grain yield were summed up to calculate SSI for each genotype.

SSI= GSI MLB resistance + GSI grain yield
A genotype with lower SSI was considered to be superior and stable with respect to both MLB resistance and grain yield.

Results And Discussion
Categorisation of BC 1 F 5 lines into different disease resistance classes when evaluated in multiple environments Differential response for MLB resistance was observed amongst the evaluated lines ( Fig. 2) (2008) also observed variability for northern leaf blight resistance in the teosinte derived maize introgression lines. As the parental maize inbred line DI-103 was susceptible to MLB, the MLB resistance in the derived BILs were thought to be the result of teosinte genomic introgression. In accordance with our study, the near isogenic lines containing teosinte genomic introgression in the background of the maize inbred B73 were also developed by Liu with different GLS resistance alleles sourced from teosinte. The unimproved germplasm though a source of novel resistant alleles is scarcely used in breeding programme due to associated inferior characteristics as low productivity, high plant and ear height and high stalk and root lodging (Teixeira and Guimarães 2021). The MLB resistant prebreeding lines developed in this study can serve as a link between unimproved and improved germplasm and can be utilized effectively in future breeding activities as source of MLB resistance. When all three environments were considered together a total of 73 BILs which showed resistance in at least one of the environments were identi ed. As the objective of our study was mainly to detect superior and stable genotypes for MLB resistance and grain yield and also to reduce clutter in biplots, based on the preliminary MLB resistance evaluation, a subset of only 73 resistant BILs were subsequently utilised for combined ANOVA, AMMI analysis, GGE biplot analysis and computation of GSI MLB resistance , GSI grain yield and SSI.

Combined Analysis of Variance for PDI and grain yield
Combined Analysis of Variance of the shortlisted 73 resistant BILs was done for PDI and grain yield using the data obtained after evaluation under three environments. Combined ANOVA (Table 4) across the environments revealed presence of highly signi cant differences (P < 0.01) for genotypes, environment and GXE interaction (GEI) for both PDI and grain yield. In accordance with our study signi cant G, E and GEI effects were also reported by Sibiya et al. (2012) for grey leaf spot disease resistance and grain yield in elite African maize germplasm. (Gyawali et al. 2019) also reported signi cant G and GEI while evaluating barley genotypes in multiple environments for resistance to spot form of net blotch. For grain yield, signi cant G, E and GEI in maize genotypes were also detected by Mafouasson et al.
(2018) and Aktas and Ure (2020). Highly signi cant environmental effect demonstrated that the experiments were carried out under differing climatic conditions affecting both PDI and grain yield of genotypes. The presence of signi cant GEI indicates that MLB resistance response and the yield of the evaluated BILs varies with the environments which reduces the correlation between genotype and phenotype for a speci c trait thereby making the selection of genotype less effective and hence ultimately reducing the speed of crop improvement. Therefore, reducing the magnitude of GEI is important. Reduction in GEI can be achieved by means of stability analysis and selection of only those genotypes which shows wider adaptability (Yaghotipoor and Farshadfar 2007).
AMMI analysis of variance for PDI and grain yield AMMI ANOVA for PDI revealed that highest percentage of total sum of squares (SS) was attributed to GEI (40.55%) while 32.86% and 26.59% was contributed by genotype and environment, respectively (Table 5). GEI was further divided into interaction principal components axes (IPCA) the rst two of which were statistically signi cant. The rst IPCA (IPCA1) had maximum share of 64.09% whereas IPCA2 had share of 35.91% in the total GEI sum of square. In accordance with our study, signi cant G, E and GEI was also observed by Persaud et al. (2019) for sheath blight resistance in rice with GEI contributing as high as 39.52% to the total SS. They also identi ed two signi cant principal components contributing greater than 70% of the total GEI effect. Highly signi cant G and GEI were also observed for spot form of net blotch disease in the evaluated 340 barley genotypes by Gyawali et al. (2019). As GEI effect was larger than the genotype effect the genotypes were ascribed to show variable performance for MLB resistance in different environments which makes the identi cation of stable genotypes more crucial.
For grain yield, AMMI ANOVA revealed the largest contribution of 68.02% towards SS by the genotype main effect followed by GEI (17.50%) and E (14.48%) as presented in Table 5. The contribution by GEI to SS was 3.88 times smaller when compared to genotype component however it was greater than contribution due to environment. Two statistically signi cant principal components namely, IPCA1 and IPCA2 accounted for 96.25% and 3.75% of the GEI effect, respectively. Signi cant G, E and GEI was also observed earlier by Mafouasson et al. (2018) while evaluating yield stability of single cross maize hybrids. A large SS for genotypes displayed that the diverse nature of genotypes with greater differences among the mean grain yield was responsible for causing most of the variations. GEI was 3.89 times smaller than that for the sum of square for G. Further, sum of square due to G x E was 1.2 times larger than that for E. This clearly showed that grain yield was mainly ascertained by the genotype and GEI and E had a very little role to play thereby suggesting the possible existence of different genotype groups (Mohammadi et al. 2011). It is evident from table 5 that for both PDI and grain yield using biplots in interactions was very advantageous as the rst two PCA axes explained 100% of the total interaction effect. The conclusion drawn on genotype stability based on these two axes was therefore very reliable. Different researches have also indicated that GEI was precisely predicted by two PCAs (Gauch and Zobel 1996;Nayak et al. 2008;Mukherjee et al. 2013). Accordingly, the stability values and index were calculated by retaining two PCA axes in the model for both the traits.
Predictions from AMMI1 biplot AMMI1 biplot consists of main effects for the trait in question displayed on the X axis and IPCA1 scores displayed on the Y axis. On the basis of mean of the trait and the IPCA1 scores the genotypes and environments are plotted on to the biplot. The central vertical line on the biplot denotes grand mean of the genotypes for the trait while the horizontal line indicates an IPCA1 value of zero. IPCA1 scores are an indication of genotype stability. Lesser the IPCA1 score more stable the genotype. An IPCA1 score of zero indicates highly stable genotypes across all the environments. The larger the IPCA scores, either positive or negative, the more speci cally adapted a genotype is to certain environments (Mafouasson et al. 2018). AMMI1 biplot (Fig. 3a) between IPCA1 and PDI revealed that four genotypes namely MT-40 (10), ,  and  falling almost on vertical line had PDI equivalent to the PDI grand mean of all genotypes while seven genotypes namely , , , , ,  and  falling almost on horizontal line had equal GEI effects with IPCA1 scores of almost zero signifying a small interaction with the environments and were considered to be the most stable with respect to MLB resistance. Out of these MT-148 (55) was located on the extreme left and therefore was identi ed as line having high degree of resistance and lower uctuations to MLB. On the other hand, MT-172 (64) was located on the extreme right showing stable and susceptible disease response. The highly resistant genotype  identi ed in the present study could be used as potential source for disease resistance while MT-172 (64) could be used as susceptible host for producing pathogen inoculum for MLB screening notwithstanding the diverse environments. Similar identi cation and utilisation of stable blast resistant and susceptible genotypes was also proposed by Persaud and Saravanakumar (2018).
A signi cantly larger number of genotypes recorded IPCA1 scores of -0.667 to 0.334 leading to clustering of the genotypes on the biplot. Genotypes MT-176 (66) and  showed the highest IPCA1 score of 1 and -0.759, respectively and therefore were speci cally adapted.

Predictions from AMMI2 biplot
In the AMMI2 biplot for PDI (Fig. 4a), more responsive genotypes and environments are displayed away from the origin. E2 was the most differentiating environment followed by E4 and E6. Genotypes MT-9 (1), MT-25 (4) (64) were closer to E6 indicating their speci c adaptation to the respective environments and therefore varieties or hybrids derived from these genotypes can be recommended for cultivation in the respective environments.

GGE biplot analysis for PDI and grain yield
The magnitude of GEI was further studied by genotype and genotype x environment (GGE) biplot analysis. The rst two principal components (PCs) of the GGE biplot accounted for 85.5% (PC1 = 54.5%, PC2 = 31.0%) of the total variation for PDI over three environments, while for grain yield the PCs explained a total of 99.2% (PC1 = 80.5%, PC2 = 18.7%) of the total variation.
The average environment coordination (AEC) view of GGE biplot for PDI (Fig. 5a) revealed that E6 was the most discriminating environment followed by E4 while E2 was least discriminating. The average environment axis (AEA) was used to show the ideal test environment for MLB. The E6 was placed closest to the "average environment" and was considered suitable for MLB resistance screening. However, E4 and E2 being situated away from the AEA were least representative of the ideal environment. For grain yield, E4 was identi ed as the most discriminating environment followed by E2 while E6 was least discriminating environment (Fig. 5b). E4 on account of being located closest to the AEA was also identi ed as the most representative of the ideal environment and was followed by E2.
The E6 was the least discriminative environment for grain yield. In Fig. 5a and Fig. 5b, the genotypes dispersed away from the origin and distanced from each other were different with respect to MLB resistance and grain yield, respectively. These genotypes were least stable and contributed to both G and GEI. While genotypes situated near to the origin were highly stable and contributed insigni cantly to both G and GEI as discussed by Persaud et al. (2019) while evaluating rice genotypes in multiple environments for resistance to sheath blight.
An ideal test environment that can be recommended for testing of genotypes for a speci c trait must have high discriminating ability for the genotypes as well as be true representative of all the environments (Yan and Kang 2002;Tekalign et al. 2017). In our study E6 had high discriminating ability for MLB resistance of genotypes and was also an ideal representative of all the environments as evident from the AEC view of GGE biplot for PDI. The E6 can therefore be recommended for selecting generally adapted genotypes for MLB resistance. For grain yield E4 could be used for selecting broadly adapted genotypes. The E4 was discriminating but was not an ideal representative of test environment this could therefore be used to select speci cally adapted genotypes for MLB resistance as proposed by Yan and Tinker (2005).

Which-won-where biplot for PDI and grain yield
The which-won-where biplot con rmed that only two mega environments existed for PDI (Fig. 6a). The E2 constitutes one mega environment while E4 and E6 present in same sector constitute the second mega environment. Genotypes  and  were the vertex genotypes in the sector that had environments E4 and E6. As these genotypes had high PC1 scores they were most susceptible to MLB in the environment they were located in. While  was the vertex genotype for the sector containing environment E2 and had negative PC1 score indicating it to be the most resistant in E2. The other genotypes in the polygon view situated near the origin displayed low positive or negative PC1 scores signifying moderate resistance to moderate susceptibility and showed reduced responsiveness when compared with vertex genotypes (Tekalign et al. 2017) in a particular mega environment. For grain yield also two mega environments were observed (Fig. 6b). One mega environment consisted of E2 and E4 and MT-183 (67) was the vertex genotype while E6 was a part of second mega environment with MT-61 (18) and MT-72 (24) as vertex genotypes displaying highest grain yield in their respective environments. These genotypes can therefore be recommended as high yielding genotypes for their respective mega environments. Two mega environments for grain yield with different winning genotypes were also detected by Erdemci (2018) while evaluating chickpea genotypes in multi environment trials.
Stability of genotypes based on ASV for PDI and grain yield ASV was calculated in order to estimate stability of genotypes with respect to MLB resistance and yield under three environments ( Table 6). The genotypes having lower ASV were said to be more stable. For PDI, the genotype MT-90 (32) with lowest ASV of 0.193 was observed to be the most stable. Other genotypes with lower ASV scores were , , , ,  and

Identi cation of superior genotypes
Genotypic selection index (GSI) for MLB resistance and grain yield and simultaneous selection index was calculated for 73 BILs (Table 7). Based on the GSI for PDI, ve genotypes namely , , ,  and  which had the lower GSI PDI score of 7, 11.5, 20.5, 21 and 32.5 amongst the evaluated BILs, exhibiting high stability and increased MLB resistance across environments were identi ed (Table 8). The genotypes MT-120 (45), MT-166 (62), MT-14 (2), MT-37 (9) and MT-57 (15) having lower GSI of 13, 17, 18, 25.5 and 31, respectively, for grain yield were identi ed. These genotypes possessed better adaptability and higher grain yield (Table 8). Summing up the GSI for PDI and GSI for grain yield, the genotype exhibiting the lowest score was considered to be the best as it combined stability and best trait mean for grain yield and resistance to MLB. Based on simultaneous selection index (SSI) ( Table 8) (27) and 55,57.5,58,58.5,77.5,81.5,82,88 and 88, respectively, were identi ed to be promising. Such genotypes could be potential germplasm for development of high yielding MLB resistant genotypes or can be used as parent in generating biparental mapping population.

Conclusion
The present investigation identi ed a total of 73 BILs displaying resistance to MLB in at least one of the environments. AMMI1 biplot identi ed MT-148 (55) as genotype having high degree of resistance and lower uctuations to MLB while MT-172 (64) showed stable and susceptible disease response. The resistant genotypes can therefore be used as potential donor for stable disease resistance trait and the susceptible line can be used for producing pathogen inoculum for MLB screening notwithstanding the diverse environments. MT-120 (45) as revealed by AMMI1 biplot could serve as source of high and stable grain yield. The study was also able to classify different environments into various mega environment, as well as to identify the most discriminating ideal environments for assessing the variability with respect to MLB resistance and grain yield. For both PDI and grain yield two mega environments were identi ed. Mega environments are composed of environments falling in the same sector and consisting of same winning genotypes. Genotypes MT-60 (17) and  were winning genotypes in the sector that had environments E4 and E6 for PDI. These genotypes showed highest MLB susceptible reaction in these environments. Therefore, if MLB screening is undertaken in E4 or E6 these genotypes can be used as susceptible checks.  was the winning genotype for the sector containing environment E2 and showed resistant response. This genotype would hence be important MLB resistance donor for breeding resistant genotypes speci cally adapted for E2. For development of high yielding varieties speci cally for rst mega environment consisting of E2 and E4, MT-183 (67) could be used as donor while for E6  and  could be used as donor for high yield. The study ultimately identi ed ten genotypes namely, , , , , , , , ,  and  as the best genotypes containing both high and stable MLB resistance and grain yield. These genotypes can serve as an important source of MLB resistance and grain yield in future breeding programmes and after analysis of combining ability can also serve as parent for development of hybrids or synthetic varieties characterized with resistance to MLB and possessing high yield. As the studied BILs fall into highly susceptible, susceptible, resistant and highly resistant categories it could be easily predicted that MLB resistance in these BILs were governed by both oligogenes and polygenes. However, this study did not aim to dissect the genes or the mechanisms accompanying MLB resistance. Hence further research is needed in this aspect to identify novel genes and QTLs governing MLB resistance so that the information can be helpful in development of reliable MLB speci c molecular markers. The developed molecular marker will then be useful in reliable identi cation of resistant cultivars without the need to evaluate them in eld trials which would ultimately speed up the process of resistance breeding and varietal development.      E4  64  MT-2, MT-3, MT-8, MT-12, MT-15, MT-16, MT-17, MT-18, MT-22,  MT-24, MT-25, MT-32, MT-41, MT-45, MT-46, MT-55, MT-59, MT-60, MT-61, MT-62, MT-63, MT-65, MT-66, MT-67, MT-68, MT-85,  MT-86, MT-87, MT-91, MT-100, MT-102, MT-105, MT-108, MT-112, MT-118, MT-119, MT-121, MT-125, MT-126, MT-127, MT-128, MT-129, MT-131, MT-133, MT-134, MT-136, MT-140,     Mass multiplication and arti cial inoculation of Bipolaris maydis in eld.

Figure 2
Differential response of BILs to Bipolaris maydis infection.
Page 28/29  Average environment coordination (AEC) view of GGE for PDI (a) and grain yield (b) Figure 6 which-won-where view of GGE for PDI (a) and grain yield (b).