Genetic Evaluation and Screening of Diverse Wheat Genotypes for Spot Blotch Resistance


 Production of wheat (Triticum aestivum L.) the main food source of South Asian countries including India faces several constraints including spot blotch caused by Bipolaris sorokiniana resulting in yield loss of 25–43 % depending upon the stage of infection. Fifty genotypes were evaluated for nine quantitative characters and area under disease progress curve (AUDPC) to identify superior genotype with spot blotch resistance. High heritability coupled with moderate to high genetic advance as percent of mean was registered for grains per spike, tillers per square meter, days to 50% heading and days to 50% flowering indicating the characters to be governed by additive genes. Correlation and path coefficient analysis favored days to 50% heading, days to 50% flowering and grains per spike since they had significant positive correlation with yield and simultaneous negative correlation with AUDPC and also conferring highest positive direct effect towards yield. Multiple linear regression (MLR) analysis indicated days to 50% heading to be most sensitive with negative influence on AUDPC. D2 analysis grouped the 50 genotypes into 10 clusters suggesting presence of diversity among the genotypes. Frequency distribution of AUDPC among the genotypes showed more or less normal distribution of the character. Low AUDPC score with acceptable level of yield performance were recorded for the genotypes 29882, 29610, 29473, 29940, 29477, 29748 and 30081. Identification of high yielding and less susceptible genotypes for spot blotch disease in the present investigation offered an opportunity for wheat improvement through selective breeding.


Introduction
Bread wheat (Triticum aestivum L.) one of the oldest cereal crop is regarded as the 'King of Cereals' since it shares a large area under production, high productivity and holds a prominent position in the international food grain trade (Hazra et al. 2019a). It is the main food source of South Asian countries ) and in India it is the principal cereal crop next to rice (Kumari et al. 2020). The overall production of wheat in India has gone up tremendously from 12 Production of bread wheat in South East Asian countries including India still faces various constrains like raising temperature, unexpected hailstorms, erratic and unusual precipitation during February-March (Duveiller 2004), exposing the crop to several diseases and pests including spot blotch or foliar blight of wheat. West Bengal is categorized as a hotspot for the disease because of its mild and short winter, humid climate and late sowing due to delay in harvesting of kharif rice and sometimes excessive soil moisture after rice harvest. Warm and humid climate of this region aggravates the disease which seriously hampers the production of intensive cropping system ). The yield loss may range from 25-43% depending upon the stage of infection and the national yield loss is recorded to be around 18-22% (Acharya et al. 2011). Spot blotch is caused by a hemi biotrophic fungal pathogen Bipolaris sorokiniana (Sacc.) Shoem syn. Drechslera sorokiniana (Sacc.) Subrm and Jain (syn. Helminthosporium sativum) and its teleomorph is Cochliobolus sativus . Typical symptoms of the spot blotch disease appear on the leaves, sheath, nodes and glumes with brown lesions of oval to oblong or elliptical in shape measuring 5 to 10 mm long and 3 to 5 mm wide (Gupta et al. 2018). Other symptoms include darkening of the sub crown region, dark brown lesions on culm, coleoptile, crowns and roots.
This pathogen rst attacks the older leaves at the base of the plant and then progresses upward (Joshi et al. 2002). Lesions on leaves may start from few mm and later it turns into dark brown spots and can extend up to 1-2 cm (Chand et al. 2002). As the disease progresses, the lesions get scattered throughout the leaves and subsequently their size increase to coalesce with each other to form large necrotic spots which result in loss chlorophyll that causes reduction in photosynthetic area of the leaf (Gupta et al. 2018). Sometimes yellowing can be seen due to toxin production from the lesion (Chowdhury et al. 2013). In its severe form the fungal pathogen attacks the spikes forming dark brown to black discolouration around the germinating point of the seed known as "Black Point" (Gupta et al. 2018).
The genetic base of cultivated wheat genotypes has become narrow due to continuous inbreeding (Rehman et al. 2018) and the present agricultural scenario has led to rapid decline in both inter and intra varietal variability as a result of continuous breeding of the elite genotypes. Complete resistance in bread wheat genotypes against spot blotch or foliar blight is still lacking, although low to high levels of resistance have been reported (Rosyara et al. 2007; Singh et al. 2020). In high yielding varieties of wheat resistance to spot blotch is poor and requires rigorous investigation to improve the resistance along with good yield (Joshi et al. 2007;Meena et al. 2014). The existing genetic variability can be exploited by intercrossing among diverse genotypes to isolate superior transgressive seggregants (Baranwal et al. 2012;Shah et al. 2020). Identi cation of superior diverse genotypes with desirable traits and their consequent use in breeding program and establishment of successful selection criteria can be helpful for successful varietal improvement (Hazra et al. 2019b).
The present investigation has been aimed to assess the interrelation between the disease score and other important agro-morphological traits of fty diverse wheat genotypes grown naturally under a hotspot region for the disease in India for formulating an effective selection criterion for spot blotch resistance wheat breeding.

Page 3/18
Fifty bread wheat genotypes (Table 1) collected from International Center for Agricultural Research in Dry Areas (ICARDA), Aleppo, Syria and AICRP-Indian Institute of Wheat and Barley Research (ICAR-IIWBR), Karnal, India was used in the present study to screen for the resistance to spot blotch disease and other agronomic traits. Out of the fty genotypes, HD2967, HD3086 and DBW107 were used as yield check as suggested by Gupta et al. (2017) while Sonalika, vulnerable to spot blotch, was used as a susceptible check variety for spot blotch disease screening according to Turan et al. (2017). The genotypes designated as (C) has been used as check.

Field experimentation
The present investigation was conducted at 'AB' block farm BCKV (Bidhan Chandra Krishi Viswavidyalaya), Kalyani (22°59' N, 88°48' E, and 9.75 m above mean sea level) under the new alluvial zone of West Bengal, India during the Rabi season of two consecutive years 2018-2019 and 2019-2020. This area is primarily considered as hot spot area of spot blotch disease due to prevalent sub-tropical humid climatic condition with annual mean temperature range of 12.5° C − 36.3° C and rainfall of 1120-1500 mm with relative humidity of 50-80%, which is ideal for the development of disease. Fifty test genotypes were sown following randomized block design with three replications. In all the replications each genotype was planted in three rows of 3 m length keeping 18 cm distance between rows. The susceptible check Sonalika was included after every 20 test entries and along the borders to provide the chance of creating equal disease pressure to all the test genotypes. The sowing was done in the rst week of December so that the post anthesis stage is exposed to warm and humid environment which is conducive for disease development (Chaurasia et al. 2000). During both the years, the crop was raised with recommended package of practices. A fertilizer dose of 120-60-40 kg/ha N-P 2 O 5 -K 2 O was applied to the experimental plot. Half nitrogen and full amount of phosphorous and potassium was provided as basal during eld preparation, a quarter of nitrogen was top dressed at 21 days after sowing (DAS) and another quarter at 40 DAS. Five irrigations were applied as recommended in the critical growth stages of the crop (crown root initiation stage, tillering stage, late jointing stage, owering stage and dough stage) although the soil contained su cient organic matter content (0.78%) to retain moisture.

Assessment of agro-morphological traits
The agro-morphological traits assessed in this experiment were plant height (cm), days to 50% heading, days to 50% owering, tillers per square meter, days to maturity, spike length (cm), grains per spike, test weight (g), yield/plant (g). Plant height was measured from the base at ground level to the tip of spike of main tiller excluding awns at maturity. Days to 50% heading and days to 50% owering were counted as the number of days from sowing until 50% of the ear emerges fully from the boot of ag leaf and anthesis occurred in 50% of the ear in each plot respectively. Days to maturity were counted as the number of days from sowing till the grains became hard enough and contained moisture levels near 12%. Test weight and yield/plant was recorded for each test genotype separately by taking weight in electric balance. Grains per spike were counted manually and panicle length was measured by a 30 cm long scale bar with 0.1 cm interval markings after harvest.

Arti cial inoculation of pathogen
To impose an optimum disease pressure for thorough screening of test genotypes, arti cial epiphytotic condition was created beside natural disease occurrence. To have an idea about the progress of disease with time, area under disease progress curve (AUDPC) based on disease severity score was calculated using the following expression (Das et al. 1992): Yi is the disease severity measured on the i th date and (t (i+1) -t i ) is the number of days in between two consecutive dates of disease scoring and n is the number of dates on which spot blotch was recorded.

Statistical analysis
The data obtained from the quantitative parameters of the 50 genotypes grown for two years were subjected to pooled analysis to consider the consistency of response of the test genotypes over the years. Analysis of variance (ANOVA) was performed for the characters under study. Estimates of genetic parameters like genotypic coe cient of variance (GCV) and phenotypic coe cient of variance (PCV) as per Burton (1952) and Burton and De Vane (1953), broad sense heritability (h 2 ) as per Hanson et al. (1956) and genetic advance as a percent of mean as per Johnson et al. (1955) were evaluated using the R-packages, version 3.6.1. Character association expressed in terms of correlation coe cient as per Al-Jiboari et al. (1958) and path coe cient analysis as suggested by Wright (1921) and discussed by Dewey and Lu (1959) were determined by Statistical Package for Agricultural Research (SPAR-I). Character with maximum in uence on AUDPC was determined through multiple linear regression (MLR) using SPSS version 23.0. Genetic divergence among the genotypes was determined by the Mahalanobis' generalized distance (Mahalanobis 1936) as per Rao (1952) using Genres software version 7.01. Skewedness for AUDPC among the genotypes was calculated using SPSS version 23.0.

Results
Analysis of variance (Table 2) revealed signi cant differences among the fty genotypes for all the ten quantitative characters. Heat map analysis ( Fig. 1) arranged the genotypes and the characters into hierarchical clustering simultaneously based on similarity and distances between them and the pattern of colour mosaic indicated the association between the genotypes and the characters. High values (> 20%) of genotypic coe cient of variation (GCV) and phenotypic coe cient of variation (PCV) could not be recorded for any characters. However, among the characters the GCV and PCV values (Fig. 2) were highest for yield per plant followed by tillers per meter square and grains per spike. High heritability in broad sense (> 60%) was observed for all the characters excepting test weight, which was moderate (51.50%). The characters like plant height and days to maturity showed high heritability but their genetic advance (Fig. 2)   For character association study using correlation analysis ( Fig. 3a and 3b   The genotypes could be grouped into 10 clusters (Table 4) based on D 2 square analysis. Cluster VI accommodated maximum genotypes (9), followed by cluster I accommodating 7 genotypes. Cluster II and cluster IX comprised of 6 genotypes each, while cluster X accommodated 5 genotypes and cluster VIII consisted of 3 genotypes while two genotypes grouped in cluster III and cluster IV, each. Maximum intra cluster distance was observed for cluster IV (9.23), followed by cluster VI (9.22), cluster X (8.93) and cluster I (8.26). The intra cluster distance of cluster II and cluster VI were also on a higher side being 7.29 and 7.91 respectively, while minimum intra cluster distance was recorded for cluster III (2.61) followed by cluster IV (2.73). Inter cluster distance was recorded to be maximum between cluster VII and cluster VIII (13.26), followed by cluster III and cluster V (12.03), cluster V and cluster VII (11.98) and cluster VIII and cluster X (11.68). Minimum inter cluster distance was registered between cluster II and cluster IV (6.53) followed by cluster II and cluster VII (7.86). Out of the 10 quantitative characters studied (Fig. 6), yield per plant contributed maximum (26.20%) towards total divergence, followed by days to 50% heading (23.92%), grains per spike (15.18%) and days to 50% owering (14.94%). The contribution of plant height, tillers per square meter and spike length were not evident while the contribution of days to maturity and spike length were also less. The cluster mean analysis (Table 5) revealed the maximum cluster mean for days to 50% heading (73.00) and days to 50% owering (78.78) in cluster VIII while cluster IV exhibited highest mean for grains per spike (50.17) and yield per plant (6.18g). Minimum cluster mean for days to 50% heading (61.56) and days to 50% owering (69.50) were registered in cluster VII and minimum cluster mean for grains per spike (41.78) and yield (4.67) per plant were recorded in cluster VIII and cluster V respectively. The genotypes designated as (C) has been used as check Two genotypes 29872 (1378.33) and 30001 (1375) showed higher AUDPC value than that of Sonalika. Low AUDPC score with acceptable level of yield performance (Fig. 8) with respect to the yield checks HD 2967, HD3086 and DBW 107 were recorded for the genotypes 29882, 29610, 29473, 29940, 29477, 29748 and 30081.

Discussion
Signi cant variation among the genotypes for all the quantitative characters was the testimony of varied parentage of the genotypes taken under study as suggested earlier by Arya et al. (2017a) and different agro-climatic conditions from where the genotypes were obtained (Yadav et al. 2014).
Signi cant variation was further re ected by heat mapping, which not only hierarchically clustered the genotypes according to their distance and similarity but also simultaneously clustered the characters under study and revealed the interaction between the genotypes and characters.
Hierarchical clustering is a powerful tool to agglomerate a group of individuals into a cluster based on their similarity and distance. Heritability depicts the percentage of variability that is transmitted from parents to offspring. However, heritability alone cannot provide a reliable picture for genetic gain (Arya et al. 2017a; Sejake et al. 2020) as evident for characters like plant height and days to maturity which registered high heritability (broad sense) but low genetic advance as percent of mean. Combination of these genetic variability parameters indicated that these characters might not be governed by additive gene action hence, direct selection will not be effective. High heritability combined with high genetic advance re ects governing of the characters by additive genes thus, direct selection for those characters could be rewarding (Poudel et al. 2021). High heritability coupled with high genetic advance as percent of mean recorded for tillers per square meter, grains per spike and yield per plant suggested direct and early generation selection for these characters (Bhanu et al. 2018; Al-Nager et al. 2020). Days to 50% heading and days to 50% owering registered very high heritability with moderate genetic advance as percent of mean indicating selection for these traits might be bene cial through their phenotypic performance (Hailu 2020; Poudel et al. 2021), while moderate heritability and low genetic advance as percent of mean for test weight suggested direct selection for this trait would be non-rewarding.
Selection based on yield alone, a complex character is generally not very effective. Correlation studies provide important information to identify and verify whether the selection for a certain character in uences another one, to quantify indirect gains due to selection of correlated traits and to evaluate the complexity of the traits (Tiwari and Upadhyay 2011). Correlation coe cients for most of the characters at genotypic level were higher than the corresponding coe cients at phenotypic level indicating the presence of inherent genetic relationship among the characters as suggested earlier (Tripathi et al. 2015). Yield per plant registering signi cant negative correlation with AUDPC at both genotypic and phenotypic levels suggested that spot blotch disease was a major biotic constraint for realizing high yield in wheat as reported earlier (Meena et al. 2014;Ayana et al. 2018). Days to 50 % heading, days to 50 % owering, tillers per square meter, spike length and grains per spike revealed signi cant positive correlation with yield per plant at both genotypic and phenotypic level. Days to maturity showed signi cant positive correlation at genotypic level but it was not signi cant at phenotypic level. This might be possible since the genetic advance as percent of mean for days to maturity was low indicating the character to be governed by non-additive genes and hence in uenced by environment (Addisu and Shumet 2015; Hossain et al. 2021). The characters, days to 50 % heading, days to 50 % owering, tillers per square meter and grains per spike revealed signi cant negative correlation with AUDPC at both genotypic and phenotypic level. The negative association between AUDPC and 50 % days to heading and 50 % days to owering suggested a negative relation between the disease severity and duration of the crop (Mahto 2001;Sharma et al. 2006). Earlier reports also suggested negative correlation between disease severity and yield attributing parameters like, number of grains per spike and tillers per square meter (Gilchrist et al. 1991;Sharma et al. 1997;Duveiller 2003, Singh et al. 2008). Non-signi cant correlation between AUDPC and plant height, spike length and test weight indicated a scope of simultaneous improvement of these traits and disease resistance. Selection of the characters having strong positive correlation with yield and simultaneous negative correlation with AUDPC would be rewarding in future wheat breeding programs (Sharma and Duveiller 2003). Correlation alone does not provide a clear picture of character association, since two characters might show correlation as a result of their correlation with a common third one (Poudel et al. 2021). Path coe cient analysis provides the actual information by splitting the correlation coe cients into measures of direct and indirect effects of the set of quantitative characters on yield per plant (Mecha et al. 2017). Grains per spike registered maximum direct positive effect on yield per plant at both genotypic and phenotypic level and its correlation with yield per plant was also very high. Days to 50 % heading and days to 50 % owering also revealed positive direct effects although the values were very low. Other traits which had signi cant positive correlation with yield had negative direct effects. The low value of residual effect at both genotypic and phenotypic level suggested the inclusion of most of the responsible factors for grain yield per plant. AUDPC being complex and dependent on several factors, is di cult and often misleading to estimate empirically and data mining techniques in the form of MLR has been reported to be more accurate and is becoming a new trend in understanding the sensitivity of several characters towards complex traits like AUDPC. (Nourani and Fard 2012). MLR extends linear modelling ideas to a wider class of response types, such as count data or binary responses (Sengupta et al. 2021) and is able to gure out the interrelationship among the input (Quantitative charaters) and the output data (AUDPC) and predict each output with its corresponding output. MLR analysis revealed days to 50% heading to be most sensitive towards AUDPC in a negative direction and the MLR equation suggested that the character contributed 66.1% towards AUDPC. Sensitivity analysis using MLR was in accordance with the ndings of correlation study which indicated days to 50% heading to be negatively correlated with AUDPC. Correlation study along with MLR validated the character days to 50% heading to be most important for spot blotch resistance screening which might be due to the fact that in South East Asian countries the prevalence of spot blotch coincides with the heading and post-heading stage of wheat (Chowdhury et al. 2013). In this context it may be pointed out that breeding for short duration crops will be rewarding since reduction in days to 50% heading can alleviate the impact of spot blotch by escaping the critical stage ( owering).
The genotypes could be grouped into 10 clusters indicating presence of divergence among them. The clustering pattern suggested no parallelism between genetic diversity and geographical origin as recorded earlier in soybean, Glycine max L. (Malik et al. 2011) and tomato, Solanum lycopersicon L. (Debnath et al. 2020). Grouping of the genotypes of same geographical origin into different clusters might be due to change in certain characters as a result of natural or arti cial selection (Narayan et al. 2018). The intra and inter cluster distance among the genotypes indicates the distance among the genotypes in a single cluster and between the genotypes of different cluster respectively. The intra and inter cluster D 2 distance depicted the diversity present within the genotypes in a particular cluster and among the clusters, respectively. Maximum intra cluster distance was recorded for cluster IV, while minimum was recorded for cluster III. High intra cluster distance suggested the genotypes within the cluster had high degree of divergence and would produce desirable breeding materials for attaining maximum genetic advance (Dobariya et al. 2006;Chandramohan et al. 2016) whereas low intra cluster distance suggests presence of homogeneity among the genotypes within the cluster and selection of genotypes within the cluster would be ineffective. High inter cluster distance was recorded between cluster VII and cluster VIII, cluster III and cluster V, cluster V and cluster VII and cluster VIII and cluster X. Higher inter cluster distance suggested that the genotypes grouped in these clusters revealed broad spectrum of genetic diversity and can be utilised in future wheat breeding program to isolate desirable transgressive segregates for developing potential high yielding wheat varieties (Singh et al. 2010;Arya et al. 2017b). Low inter cluster distance between cluster II and cluster IV and cluster II and cluster VII depicted close relationship between the genotypes present in the clusters. Yield per plant, days to 50 % owering, grains per spike and days to 50 % heading contributed majorly towards total divergence. The characters contributing maximum towards divergence might offer good scope of improvement through selection (Anuradha et al. 2020; Kumar et al. 2020). From this context, the cluster mean analysis of the major contributing characters revealed maximum cluster mean for the days to 50 % heading and days to 50 % owering were in in cluster VIII, while the minimum cluster mean for these characters were recorded in cluster IV. Maximum cluster mean for grains per spike and yield per plant were registered in cluster IV and minimum cluster mean for these characters were in cluster VIII and cluster V respectively. The contrasting value of cluster mean for the characters were evident from the high inter cluster distances between the clusters and inter crossing among the genotypes from these clusters might be rewarding.
The disease reaction of wheat genotypes was characterized by their response measured as area under disease progress curve (AUDPC) as suggested earlier by Duveiller et al. (1998) which signi cantly differed among the genotypes. Frequency distribution for AUDPC revealed more or less normal distribution suggesting that the variable was clustered more near the mean. This implied the presence of more number of moderately resistant/ moderately susceptible genotypes than that of highly susceptible or highly resistant ones. No genotypes showed complete resistance for the disease similar to the earlier reports on lack of resistance in the south Asian wheat cultivars by Siddique et al. (2006)  The present investigation screened the genotypes having low AUDPC score for spot blotch disease along with high yield under hot spot region which presented a reliable picture on the disease reaction of the genotypes. Moreover, the present study aimed at understanding the interaction of the genotypes with the quantitative characters through heat map analysis and also developing a selection criterion for screening better genotypes through identi cation of important yield attributing characters with negative correlation with AUDPC with further validation using multiple linear regression model. This kind of selection approach for screening high yielding resistant lines ensures great deal of novelty.

Conclusion
Breeding of high yielding and spot blotch resistance is of utmost importance in wheat especially in the South Asian countries like India. In the present investigation, the characters 50% days to owering, 50% days to heading and grains per spike came out as the major yield attributing characters since they showed high heritability coupled with high genetic advance as percent of mean, signi cant positive correlation with yield per plant and simultaneous signi cant negative correlation with AUDPC and also conferring positive direct effect to yield per plant in path analysis. Multiple linear regression (MLR) identi ed days to 50% heading to be most sensitive towards AUDPC in a negative direction indicating breeding for short duration to be rewarding. D 2 analysis revealed presence of wide divergence among the genotypes by grouping them into ten clusters. Presence of superior genotypes in different clusters of varied diversity broadens the scope of using them in combinational breeding programme to exploit transgressive segregates in positive direction. Among the 50 genotypes, 29882, 29610, 29473, 29940, 29477, 29748 and 30081 revealed acceptable yield performances with low AUDPC value. Identi cation of high yielding and less susceptible genotypes for spot blotch disease in the present investigation offered an opportunity for wheat improvement through selective breeding. Figure 1 Heat map clustering analysis of 50 genotypes across 9 characters. The rows of a microarray heat map represent characters, and the columns represent the genotypes. Each cell is colorized based on the level of expression of that character in that sample Here, G represents genotype followed by numerical as described in Table 1 and Ch represents characters as plant height (Ch1), days to 50% heading (Ch2), days to 50% owering (Ch3), Tillers/sq.m (Ch4), Days To maturity (Ch5), Spike length (Ch6), Grains per spike (Ch7), Test weight (Ch8), and yield/plant (Ch9) Figure 2 GCV, PCV, Heritability (broad sense), Genetic advance as percent of mean of the nine quantitative characters (pooled data of two years) Figure 3 a. Genotypic correlation of yield, and yield contributing traits including AUDPC (pooled data of two years). Signi cance levels are provided by * (5%) and ** (1%) respectively. b. Phenotypic correlation of yield, and yield contributing traits including AUDPC (pooled data of two years). Signi cance levels are provided by * (5%) and ** (1%) respectively.

Figure 4
a. Genotypic path coe cient analysis indicating direct and indirect effects of the independent variables over yield (solid line represents direct effect, while dotted line refers to indirect effects, and the arrow signi es the relatable variables; as pooled data of two years). b. Phenotypic path coe cient analysis indicating direct and indirect effects of the independent variables over yield (solid line represents direct effect, while dotted line refers to indirect effects, and the arrow signi es the relatable variables; as pooled data of two years).

Figure 5
Inter cluster and intra cluster distance of the genotypes (not to scale); (pooled data of two years) Figure 6 Percentage contribution of the characters towards divergence(pooled data of two years)

Figure 7
Frequency distribution of AUDPC among the genotypes(pooled data of two years) Figure 8 Yield per plot and AUDPC score of the selected genotypes as compared to the check varieties (pooled data of two years).