Generation Mean Analysis for Yield and Yield Component Traits in Inter-specific Cross of Rice (Oryza sativa L.)

Rice (Oryza sativa L., 2n = 2x = 24) is the primary food source for billions of people in Asia and Africa, accounting for half of the world’s population (Chukwu et al., 2019) and also the principal source of income for unprivileged rural people. India occupies the world’s largest area under rice with 42.5 million ha. and is the second highest producer with 106.65 million tonnes followed by China, contributing to 21 per cent of global rice production. It has a vital role in the food and livelihood security of the country. However, productivity of rice is only 2.54 tonnes ha^-1 as against the global average productivity of 3.28 tonnes ha^-1 due to cultivation of rice under wide range of ecological situations coupled with high pressure of biotic and abiotic stresses particularly bacterial leaf blight, blast and low soil Phosphorous situations (Kumar et al., 2017). Considering the current growth rate in population, the supply projection falls short of expected demand of 121.6 m tons by the year 2030 and 137.3 m tons by the year 2050. This is far below the current growth rate of production (0.36%) in comparison to population growth rate of 1.63 per cent. In order to achieve this target, the productivity of rice has to be brought to the level of 3.3 tons ha^-1, from the present scenario (IIRR, 2015). To reach the desired production level, rice varieties must be created that have a yield advantage of about 20% over commonly cultivated varieties (Kumar et al., 2017). To obtain the desired genetic improvement towards the development of better lines, it is crucial to collect information about genetic architecture of quantitative traits having wider adaptability contributing to grain yield and quality (Barhate et al., 2016).
              
Rice yield is the outcome of multiplicative interaction of several component characters. Therefore, while breeding of high yielding varieties, breeders usually face the problem of selection of desirable parents. In general, parents are selected on the basis of their per se performance, but many times high yielding genotype may/may not transmit its superiority to progeny. Hence, critical choice of parents is most important, particularly for improvement of complex quantitative characters such as yield and its components (Kacharabhai, 2015). Partitioning of genetic variance into its all the probable components i.e, additive, dominance and all types of epistasis with regard to individual cross therefore assumes immense value in formulating an effective and sound breeding programme (Alvarez-Castro and Le Rouzic, 2015). Among the common approaches followed to understand the nature of gene effect, generation mean analysis using first degree statistics is an accurate one and gives detailed account of various gene effects and also the quality of the genes carried by the parents (Pujar et al., 2022). It is also an important statistical tool for identification of epistasis using various basic generations from a cross between two parents (Bano et al., 2017). An earlier study showed that understanding the inheritance pattern of the genes involved in desirable traits in cereal crops, including rice, is essential for efficient breeding approaches to genetically boost yield potential (Askander, 2020; Ganapati et al., 2020 and Sharma et al., 2022). The present study was carried out in this context to estimate different gene effects in the inheritance of yield and its related traits through generation mean analysis.

Plant material and experimental design
 
The present study was carried out from Kharif 2019 to Rabi 2021-22 at Agricultural College Farm, Agricultural College, Bapatla, Acharya N G Ranga Agricultural University (ANGRAU), Guntur. The experimental material was generated from the cross between the rice genotypes, namely, YH3 and AKDRMS 21-54. The female parent, YH3 is improved MTU 1121 (Sri Druthi), a high yielding and popular Rabi rice variety of ANGRAU, by the incorporation of tolerance to low soil phosphorus, through introgression of pup1 gene, while the male parent, AKDRMS 21-54 is improved Akshayadhan through the introgression of Xa21 and Pi54 genes for bacterial leaf blight and blast tolerance. Five generations (P₁, P₂, F₁, F₂ and F₃) were developed from the cross, YH3 × AKDRMS 21-54 as detailed in Fig 1 and were evaluated during Rabi 2019-20. The crossing was initiated during Kharif 2019, between the female parent YH3 (possessing Pup1) and male parent AKDRMS 21-54 (possessing Xa21 and Pi54) for development of F₁’s. True F1’s were identified with the help of gene-specific markers and advanced to F₂ during Rabi 2019-20 by selfing. F₂ plants, homozygous and positive for the target traits, were advanced to F₃ generations through selfing.
   

    
The experiment was laid in a randomized complete block design with six replications during Rabi 2021-22. The parental lines and F₁’s, F₂’s and F₃’s were planted in 2 rows and 4 rows each of 4m length at a spacing of 25 cm × 10 cm. Data were recorded on 10 plants in case of parents and F₁’s 50 plants of F₂’s and F₃’s per replication.The phenotypic traits were assessed on randomly selected plants from each individual entry in the segregating generations for 12 quantitative traits namely, days to 50 per cent flowering (days), shoot length (cm), plant height (cm), productive tillers per plant, flag leaf length (cm), flag leaf width (cm), panicle length (cm), filled grains per panicle, total grains per panicle, spikelet fertility (%), test weight (g) and grain yield per plant (g).
 
Statistical analysis
 
The morphological data recorded was subjected to statistical analysis using R software (R Core team, 2013). Different statistical tests like co-efficient of variation (CV), standard error (SE), critical difference (CD) and analysis of variance (ANOVA).
       
Adequacy of scale was tested to fulfil the conditions, namely, additivity of gene effects and independence of heritable components from non-heritable ones. The test of first condition gives information about absence or presence of gene interactions. The test of adequacy of scales is essential because in most of the cases the estimation of additive and dominance components of variances is made assuming the absence of gene interaction. The generation mean analysis was performed according to Hayman (1958) and Jinks and Jones (1958) for the estimation of genetic components of variation, epistasis model and gene effects in two steps (i) testing for epistasis to determine the presence or absence of inter-allelic interaction and (ii) estimation of gene effects, variances and the type of epistasis involved.
       
Scaling test for A, B, C and D scales as suggested by Hayman and Mather (1955) and Mather and Jinks (1971) was applied to test the adequacy of simple additive-dominance model but since, back cross is absent, only C and D scales were computed as follows:
Scale C = 4F₂ - 2F₁ - P₁ - P₂ = 0
Scale D  = 4F₃ - 2F₂ - P₁ - P₂ = 0
     
When the scale is adequate, the values of C and D should be zero within the limit of their respective Standard Errors.
Variances of above scales:

VC = 16V(F̅₂) + 4V(F̅₁) + V(P̅₁) + V(P̅₂) = 0
         
VD = 16V(F̅₃) + 4V(F̅₂) + V(P̅₁) + V(P̅₂) = 0

Standard errors of the above scale:
S.E for C scale =
 
S.E for D scale =
The ‘t’ values were calculated as follows:
t cal for C-test = Scale C/S.E of C scale
t cal for D-test = Scale D/S.E of D scale
              
The calculated value of ‘t’ was compared with tabulated value of ‘t’ at 5% level of significance (Singh and Chaudhary. 1985). In each test, the degree of freedom is sum of the degrees of freedom of various generations (total number of observations-total number of replications) involved.

The analysis of variance revealed very significant genotype differences for each of the twelve quantitative traits, indicating that the presence of sustainable amount of variability for these yield attributing traits (Table 1).
 

 
Mean performance of traits in different generations
 
The character wise mean performances of the five generation materials P₁, P₂, F₁, F₂ and F₃ for 12 traits were presented in Table 2. F₁ along with two segregating populations (F₂ and F₃) flowered and matured earlier compared to P₁ parent, which is desirable for further selections. Shoot length and plant height of F₂ and F₃ were observed to be intermediate to the parents. These findings are in agreement with earlier reports of Singh et al., (2015a) for plant height. The number of productive tillers per plant, flag leaf length and width in segregating populations (F₂ and F₃) was intermediate to the parents. The less difference for panicle length was observed between parents. Among all generation materials panicle length was comparable to both parents. The F₂ and F₃ segregants possessed higher number of filled and total grains per panicle than parents. The number of filled grains, total grains per panicle and spikelet fertility in F₁ was higher than both parents. The test weight of segregating generations was intermediate to the parents and desirable for consumer preference. Most of the above results of present investigation are conformity with the findings of Ghritlahre et al., (2018). The YH3 parent (P₁) had recorded higher grain yield per plant compared with AKDRMS 21-54, while the F₁ yielded more grain yield compared with the parents, but the two segregating populations (F₂ and F₃), grain yield per plant were lesser than F₁ generation.
 

 
Estimates from scaling tests
 
Scaling tests were performed to understand the adequacy of simple additive dominance model (Table 3). The scaling test showed significance for both C and D scales for grain yield and all yield component traits studied, indicating the presence of epistasis, The significance of C scale suggests [dd] type of epistasis. The significance of D scale reveal [aa] type of epistasis, significance of C and D scales indicate presence of both [aa] and [dd] type of epistasis (Kearsey and Pooni, 1996). The results are in broad agreement with the reports of Ghritlahre et al., (2018).
 

 
Estimation of gene effects based on five generation means
 
Digenic non-allelic interaction model with five parameters namely, m, d, h, i and l revealed that the epistatic interaction model was found adequate to explain the gene action in the traits studied in the present investigation. The estimates of gene effects clearly illustrate high variation in the observed traits (Table 3 and Fig 2-3). The dominance (h) and dominance × dominance (l) gene effects displayed opposite signs for the traits viz., shoot length, plant height, flag leaf length, flag leaf width, panicle length, filled grains per panicle, total grains per panicle and spikelet fertility indicating duplicate epistasis. The results are in agreement with the findings of Sudeepthi (2020) for shoot length; Ghritlahre et al., (2018) for plant height; Muthuvijayaragavan and Murugan (2019) and Jondhale et al., (2018) for flag leaf length and flag leaf width; Palaniraja (2017) for panicle length; Divya et al.,  (2014) for filled grains per panicle; Subbulakshmi et al., (2016) for total grains per panicle and Sand and Lal (2014) for spikelet fertility. In contrast, days to 50 per cent flowering, productive tillers per plant, test weight and grain yield per plant reported same sign for (h) and (l), indicating complementary epistasis. The results are in agreement with the findings of Savitha and Kumari (2015) for days to 50 per cent flowering; Subbulakshmi et al., (2016) for productive tillers per plant; Rani et al., (2015) for test weight and Krishna et al., (2018) grain yield per plant. On contrary, Singh et al., (2015b) had reported opposite sign for days to 50 per cent flowering, test weight and yield per plant, indicating duplicate epistasis.
 

 

       
The classification of gene interactions depends on the magnitudes and signs of the estimates of dominance and dominance × dominance effects, when there are many pairs of interacting genes (Mather and Jinks, 1982). The sign associated with the estimates of (d) and (h) indicates the parent that concentrates the highest number of genes for increasing the trait (Falconer, 1964). The positive sign for (d) was observed in the traits, days to 50 per cent flowering, flag leaf length, flag leaf width, panicle length, filled grains per panicle, total grains per panicle, spikelet fertility, test weight and grain yield per plant, indicating that YH3 contributed positively to these traits, as compared to AKDRMS 21-54. Further, the negative sign for (h) was observed in the traits, days to 50 per cent flowering, shoot length, plant height and panicle length indicating that the dominance was towards the resistant parent AKDRMS 21-54 as observed earlier by Ghritlahre et al., (2018).
              
In general, the choice of an appropriate breeding method for improvement of quantitative characters largely depends on the nature of gene action. Knowledge of the way genes act and interact will determine which breeding system can optimize gene action more efficiently and will help to elucidate the role of breeding systems (Hallauer and Miranda, 1988). Hayman’s generation mean analysis is considered as the best method to obtain information on the nature and magnitude of gene action (Gunasekar et al., 2018).

The generation mean for most of the characters showed the importance of both additive and dominance type of gene effects. Among the epistatic gene effects, the additive x additive gene interaction was predominant for shoot length, flag leaf length and flag leaf width. The gene interaction is associated with homozygosity and hence is fixable in nature and therefore, selection for these traits will be very effective in early generations and breeding procedures, such as pedigree method of breeding may be adopted for isolation of desirable lines, while for all other traits, including grain yield per plant, dominance x dominance gene interaction was found to be pre-dominant. The gene interaction is not fixable and hence, population improvement approaches, such as bi-parental mating and recurrent selection, followed by isolation of purelines in later generations would be effective.

The financial support provided in the form of INSPIRE JRF and SRF Fellowships by the Department of Science and Technology (DST), Ministry of Science and Technology, New Delhi, Government of India, through order No: DST/INSPIRE Fellowship/[IF180552]Dated: 24 July, 2019 to the first and corresponding author, for pursuing full time Doctoral research (Ph.D.) program of Acharya N G Ranga Agricultural University, Lam, Guntur, at Agricultural College, Bapatla andhra Pradesh and ICAR-IIRR, Hyderabad, Telangana, India is acknowledged.

None.