Keywords: Additive gene action, dominance gene action, epistasis and triple test cross.

Transcription

1 Detection of Epistasis and Estimation of Additive DETECTION OF EPISTASIS AND ESTIMATION OF ADDITIVE AND DOMINANCE COMPONENTS OF GENETIC VARIATION USING TRIPLE TEST CROSS ANALYSIS IN RICE (ORYZA SATIVA L.) ABSTRACT Muhammad Yussouf Saleem * Babar Manzoor Atta * Akbar Ali Cheema * Zahid Mukhtar ** Muhammad Ahsanul Haq * Genetic analyses were performed to uncover the supremacy of additive, dominance and epistasis genetic variances following triple test cross analysis involving three testers (P 1, P 2 and F 1 ) and four lines of rice. Epistasis was found to be an integral part of genetic variation for days to flowering, plant height, number of tillers per plant and yield per plant. The partitioning of total epistasis revealed that i type (additive x additive) were highly significant for days to flowering whilst j and l type (additive x dominance and dominance x dominance) were important for plant height with predominant effect of i type interaction. j and l type epistasis also played significant role in the inheritance of number of tiller per plant and yield per plant respectively. The additive and dominance effects were highly significant for number of grains per panicle and grain weight per panicle with the exception of 1000-grain weight where dominance effects were non-significant coupled with highly significant additive effects. The degree of dominance was less than unity, indicating partial dominance for number of grains per panicle, grain weight per panicle and 1000-grain weight. The direction of dominance was observed towards less grain weight per panicle. Non-allelic interactions registered for days to flowering, plant height, number of tillers and yield per plant can be manipulated through recurrent selection technique for the improvement of these traits. The predominance of additive gene action for number * Nuclear Institute for Agriculture and Biology (NIAB), Faisalabad, Pakistan. ** National Institute for Biotechnology and Genetic Engineering (NIBGE), Jhang Road, Faisalabad, Pakistan. of grains per panicle, grain weight per panicle and 1000 grain weight suggest that the selection may be delayed to later segregating populations for the improvements of yield through yield components in rice. Keywords: Additive gene action, dominance gene action, epistasis and triple test cross. RESUMO Análises genéticas foram realizadas para descobrir a supremacia de variâncias genéticas aditivas, de dominância e epistáticas seguido do triple test cross envolvendo 3 testes (P 1, P 2 e F 1 ) em 4 linhagens de arroz. A epistasia foi considerada como uma parte integrada da variação genética para dias até a floração, altura da planta, número de afilhos por planta e produção por planta. A distribuição da epistasia total revelou que o tipo i (aditivo x aditivo) foi altamente significante para dias até a floração enquanto que os tipos j e l (aditivo x dominante e dominante x dominante) foram importantes para a altura da planta com efeito predominante da interação tipo i. A epistasia tipo j e l também teve um papel significante na herança de número de afilhos por planta e produção por planta, respectivamente. Os efeitos aditivos e de dominância foram altamente significantes para o número de grãos por panícula e peso do grão por panícula exceto o peso de 1000 grãos, onde os efeitos de dominância foram não significativos agrupados com efeitos aditivos altamente significantes. Com relação ao grau de dominância, houve indicação de dominância parcial para número de grãos por panícula, peso do grão por panícula e peso de 1000 grãos. A direção da dominância foi observada como sendo para a de menor peso do grão por panícula. Interações não alélicas registradas para dias até a floração, altura da planta, número de afilhos e produção por planta podem ser manipulados através da técnica de seleção recorrente para o melhoramento nestes tratamentos. A predominância da ação de genes aditivos para o número de grãos por panícula, peso do grão por panícula e peso de 1000 grãos sugere que a seleção pode ser utilizada em programas de melhoria de produção de arroz. Palavras-chave: ação de gene aditivo, ação de gene dominante, epistatia e triple test cross.

2 Detection of Epistasis and Estimation of Additive INTRODUCTION Rice (Oryza sativa L. 2n = 24) is one of the most important cereal crop of Pakistan and is staple diet of the population in major rice growing areas. It is second foreign exchange earning crop after cotton, grown on thousand hectares with annual production of thousand tones, out of which 40-45% is exported with foreign exchange return of 559 million US $. Basmati rice contributes major portion (364 million US $) of foreign exchange. The average yield of rice is low 2013 Kg/hectares (ANONYMOUS, ) as compared to the rice yield of leading countries such as USA, China, Japan, Australia, Korea and Thailand. One of the main constraints for low yield is insufficient understanding of gene action governing yield and most of the yield components (quantitative traits), which decide the choice of selection to be followed for their improvement aiming to obtain higher yield. While estimating genetic components of variance for a metrical trait from second degree statistics, three difficulties inevitably arise. First it is assumed that non-allelic interactions are absent. This assumption may be true for some characters but not for others. Therefore, these analyses rarely provide a valid test of this assumption. Second, estimates of dominance components invariably have much larger standard errors than the corresponding additive components. Third, additive and dominance components are differentially affected by linkage and correlated gene distribution in the parents and are only comparable in the unlikely event of population sample being in linkage equilibrium. A good genetic model is that which enables the breeder to have precise and unbiased estimates of all the components of genetic variance. To over come these difficulties, a design which is a simple extension of design III of COMSTOCK and ROBINSON (1952) was proposed by KEARSEY and JINKS (1968). This design is known as Triple Test Cross (TTC), provides not only an efficient estimates of dominance but also an unambiguous test for epistasis. In the present study, triple test cross analysis has been carried out to investigate the gene action in rice for some important quantitative traits using the TTC model suggested by KETATA et al., (1976) in which the testers L 1 and L 2 were crossed to a number of lines instead of F 2 individuals as suggested and illustrated by KEARSEY and JINKS (1968) and KHATTAK et al., (2002). MATERIALS AND METHODS Two basmati rice genotypes, Basmati-385 and DM-25 designated as L 1 and L 2, respectively were used as tester genotypes. They were the two extreme high and low genotypes for most of the traits and hybridized in a combination of Basmati-385 x DM-25 in summer season (MAY-NOVEMBER,1998). The resulting F 1 was the third tester designated as L 3. Four true breeding genotypes, DM-107-4, DM , Niab rice-1 (NR-1) and Niab Irri-9 were crossed with three testers (L 1, L 2 and L 3 ) in next year summer season (MAY-NOVEMBER, 1999). The distinctive characteristics of all the genotypes are presented in Table 1. The testers were used as females in the entire triple test cross combinations. The experiment thus consisted of 6 inbred lines (2 testers and 4 inbred lines), 9 single crosses and 4 three-way crosses. The material was planted in randomized complete block design with three replications at the research filed of Nuclear Institute for Agriculture and Biology (NIAB), Faisalabad, Pakistan during 2000 summer season. The mean temperature and relative humidity was 29 C and 58% respectively during May-November. A plot size of 0.6 m 2 (single row plot of 2- meter length) was assigned to each entry in each replication. The plant-to-plant spacing between and with in the rows were kept 20 cm respectively. The experimental material was bordered by standard rice variety Basmati 385. Research field soil was sandy loam. The plots received 72.6 Kg/ha of N and 23.4 Kg/ha of P. ½ of the N was applied at the time of transplanting while remaining in two increments; ¼ part after days and the other ¼ after days of transplanting. All of the P was applied at the time of transplanting. Weeds were removed by weedicide Machette 60EC [Butachlor] and Ronstar [Oxadiazon] used at the rate of 2.0 l/ha and 3.5 l/ha respectively, with in 4 days of transplanting. Rice is water loving plant. Water level was kept inch at the time of transplanting and then gradually increased to 3 inch for 25 days after transplanting. Irrigation was discontinued for few days for proper aeration and then again continued; however it was completely stopped before 15 days of harvesting. The trial was protected from insect pest leaf roller and stem borer by application of insecticide Talstar 10EC [Bifenthrin] and Padan 4G 500 ml/ha and 22 Kg/ha respectively. The data were collected from 10 randomly selected plants per replication for the following traits. i) Days to flowering. ii)plant height (cm). iii)number of tillers per plant.

3 Detection of Epistasis and Estimation of Additive iv)number of grains per panicle. v)grain weight per panicle (g). vi)1000-grain weight (g). vii)yield per plant (g). Analysis of variance The analysis of variance was performed following the method described by SINGH and CHAUDHARY (1999) to determine the significance of treatments and to partition the treatment effect so as to determine the significance variations among the hybrids, parents, lines, testers, P 1 + P 2 vs. F 1, P 1 vs P 2, lines vs. testers and hybrids vs. parents for each trait using the TTC technique. Test for epistasis The detection of epistasis was performed according to SINGH and CHAUDHARY (1999). The test of significance of the difference (L 1j + L 2j 2L 3j (j = genotypes), provides information about the presence or absence of epistasis. Therefore the L 1j + L 2j 2L 3j for each line (genotype) and each replication was first computed (a replication thus consisted of 4 values each for a genotype) and then tested. The total epistasis for 4 degree of freedom was calculated as uncorrected genotype (lines) sums of square [S ((L 1j + L 2j 2L 3j ) 2 ]/n on the total number of replications. The total epistasis was partitioned into two components. The correction factor c.f. = [S (L 1j + L 2j 2L 3j )] 2 /n measures mainly the epistasis of additive by additive type (i type) for one degree of freedom and corrected genotypes sums of squares [S ((L 1j + L 2j 2L 3j ) 2 /n- c.f] mainly measures the j + l type (additive x dominance and dominance x dominance) for 3 degrees of freedom. The sum of square associated with the interaction of epistasis with blocks (replications) was calculated as the difference between total sum of squares and type of epistasis (total s.s total epistasis / i type epistasis / j + l type epistasis). Each of three types of epistasis was tested against their respective interaction with blocks. However before testing individual epistasis, the homogeneity of the interaction was first tested. As there were only 2 variances (i x block and j + l x block) homogeneity was tested as under: F (2, 6) = Mean square of i type interaction / Mean square of j and l type interaction. Where the homogeneity of the interaction variances was not significant, the i and j + l type epistasis were also tested against the pooled error, i.e., total epistasis x block interaction. Individual genotypic epistasis The individual contribution of each line to the total epistasis was determined and tested for significance according to KETATA et al., (1976) for those traits in which the total epistasis was significant. The mean value (S L 1j - L 2j 2L 3j ) / r (where r is number of replications) of each genotype for a trait was tested using a t-test with 8 degree of freedom as follows: t = Mean/S.E. Where S.E. = (Error mean square/replication) 1/2 Additive-dominance model The traits where total epistasis effects were not detected by either test, an additive-dominance model was fitted to the data as out lined by the KEARSEY and JINKS (1968) and JINKS et al., (1969). Estimation of additive variance component (D) The sum of L 1j + L 2j for each genotype was calculated replication wise and subjected to analysis of variance as follows: Source of variation df MS Expected Replication r-1 MS r Genotype sum (L 1j - L 2j ) n-1 MS s σ 2 e + 2rσ 2 s Error (n-1)(r-1) Ms e σ 2 e Where r = Replications; n = Genotypes; MS r, MS s, Ms e = Mean squares of replications, genotypes (sums) and error, respectively; s 2 e and s 2 s = Expected mean square of error and genotypes (sums). The observed mean squares were substituted into the equations as follows: s 2 s = (MS s - MS e )/2r s 2 s = (1/4)D D = 4(MS s - MS e )/2r Estimation of dominance component (H) The difference in L ij - L 2j for each genotype was calculated replicationwise and subjected to analysis of variance as follows; Source of variation df MS Expected Replication r-1 MS r Genotype difference (L 1j - L 2j ) n-1 MS d σ 2 e + 2rσ 2 d Error (n-1)(r-1) Ms e σ 2 e

4 Detection of Epistasis and Estimation of Additive Where r = Replication; n = Genotypes; MS r, MS d, Ms e = Mean squares of replication, genotype (differences) and error, respectively; s 2 e and s 2 d = Expected MS of error and genotypes (differences). The observed mean squares were substituted into the equations as follows: s 2 d = (MS d - MS e )/2r s 2 s = (1/4)H H = 4(MS d - MS e )/2r Average degree of dominance Average degree of dominance was calculated as (H/D) 1/2, where H and D are the dominance and additive variance components respectively. Direction of dominance (r s,d ) Direction of dominance was determined by calculating the linear correlation coefficient r s,d between the sum (L 1j + L 2j ) and the genotypic differences (L 1j - L 2j ) for all genotypes. Significant positive and negative correlations would indicate a predominant direction towards decreasing and increasing values of the trait, respectively (JINKS et al., 1969). All the triple test cross calculations were performed using MSTAT-C package (Michigan State University and Agricultural University of Norway). RESULTS AND DISCUSSION Analysis of variance The analysis of variance for different agronomic traits in rice is presented in Table 2. Days to flowering, plant height, number of grains per panicle, grain weight per panicle, 1000-grain weight per panicle and yield per plant showed highly significant differences among the treatments and between the first parent (L 1 ) and second parent (L 2 ). Highly significant differences between the treatments and significant between the first and second parent were registered for number of tillers per plant. The highly significant differences among the treatments indicated considerable genetic variations existed in the lines, testers and hybrids included in this study. The significant differences between the two parents clearly disclosed that L 1 and L 2 testers are the extreme high vs. low selections and would provide an estimate of additive and dominance variation with equal precision, as reported by KEARSEY and JINKS (1968). Detection of epistasis Presence of epistasis (Table 3) was evidenced by the significance of variance (L ij + L 2j 2L 3j ). The total epistasis was found to be highly significant for plant height and yield per plant but significant for days to flowering and number of tillers per plant. Further partitioning of total epistasis into, i type (additive x additive) and J and l type (additive x dominance and dominance x dominance) interactions showed that i type interaction was highly significant for days to flowering while both [i] and [j and l] type interactions were highly significant for plant height but with predominant effect of i type interaction. J and l type interaction was found to be significant for number of tillers and yield per plant. The greater magnitude of i type epistasis than j and l types has a special significance in rice, being a self-fertilized crop where a linear directional and fixable component (i type epistasis) of genetic variation can be most easily exploited for days to flowering and plant height through standard hybridization and selection procedures. The j and l type epistasis are non-directional and unfixable by selection under self fertilization and therefore would not be favorable for developing pure lines for number of tillers and yield per plant however it can be used in the development of hybrids (SUBBRAMAN and RANGASAMY, 1989). Though i, j and l type epistasis were simultaneously significant for plant height, but higher value of i type epistasis suggests that pure lines for this trait may be developed. The i type epistasis has also been found to be more important than j and l type epistasis in wheat (SINGH and SINGH, 1976; PANDEY and SINGH, 2003). Although some earlier workers (NEERAJ et al., 1993; VIJAYAKUMAR et al., 1996) indicated evidence of epistasis for all the traits investigated but in our studies we determined epistasis in three traits only i.e., days to flowering, plant height and number of tillers per plant. The discrepancy relating to our study might have resulted from environmental influences. It is emphasized that only one environment was studied; but genotype-environment interactions may have some influence on the epistatic effects. Such influences have been reported elsewhere in wheat and mungbean (KETATA et al., 1976; KHATTAK et al., 2001; SUNIL and SINGH, 2004). The presence or absence of epistasis may depend upon the environment in which the plant material has been evaluated and thus, it may not always be related to the inherent capacity of a genotype. Similarly JINKS and PERKINS (1970) and SUNIL and SINGH (2003) reported that the components of variance changed to different degrees over the environment. The environment influences have also been reported in wheat (PAWAR et al., 1994).

5 Detection of Epistasis and Estimation of Additive The epistatic deviations of individual lines for days to flowering, plant height, number of tillers and yield per plant are presented in Table 4 to show the direction of dominance and relative magnitudes and to identify the lines, which interacted with L 1 and L 2 to produce significant deviations. Line DM was inert and did not contributed significant portion to the total epistasis. However DM and NR-1 accounted for maximum positive portion to the total epistasis for days to flowering and plant height respectively. NR-1 and Niab-Irri-9 were the major contributors of negative epistatic potion to the total epistasis for yield and number of tillers per plant respectively. The influence of lines on non-allelic interaction of these traits indicated that the manifestation of epistasis is determined to some extent by the lines employed for the study. The limited number of lines used in the study might fail to detect non-allelic gene action, which in fact is a part of the genetic system (BURTON, 1968; KETATA et al., 1976). The optimal experimental size required to detect epistasis through TTC depends largely on the gene dispersion in the tester parents (POONI et al., 1980). Therefore several lines and extreme high vs. low testers (L 1 and L 2 ) should be used in such studies aimed at the detection of epistasis and for estimation of additive and dominance component of variation with equal precision by the TTC technique. Additive and dominance components The estimates of additive and dominance genetic components, degree of dominance and direction of dominance for those traits which were not significantly affected by epistasis are presented in Table 5. Additive and dominance effects were equally important for number of grains and grain weight per panicle but for 1000 grain weight, only additive genetic effect was significant. The magnitude of additive variance was higher for all the traits since the presence of common alleles in testers increases the magnitude of the additive component, usually the magnitude of additive component is higher than that of dominance component for most of the quantitative traits (SINGH et al., 1997).The degree of dominance [(H/D) 1/2 ] was less than unity, indicating partial dominance for number of grains per panicle, grain weight per panicle and 1000 grain weight, however it was highest (0.75) for number of grains per panicle and lowest (0.10) for 1000-grain weight. The result is in partial agreement with the findings of SUREK and KORKUT (1998) who reported incomplete dominance for 1000 grain weight but over dominance for the number of grains per panicle from diallel analysis of some yield components in rice. The significant positive correlation (Table 5) for grain weight per panicle indicated the direction of dominance towards fewer genes responsible for this trait. The non-significant correlation for rest of the traits indicated that these traits did not supply any evidence for directional dominance in rice. Non-allelic interactions, depicted for days to flowering, plant height number of tillers and yield per plant can be manipulated through recurrent selection technique for the improvement of these traits. Recurrent selection has also been suggested for non-allelic inherited traits in rice (SUBBARAMAN and RANAGASAMY, 1989; VIJAYAKUMAR et al., 1996), wheat (SHARMA et al., 1995) and mungbean (KHATTAK et al., 2001), since both dominance and additive gene effects were significant for number of grains per panicle and grain weight per panicle. Simple selection procedure in the early generations may not contribute significantly to the improvement of these traits. The additive components in these traits can be successfully exploited through pedigree method of selection because of major contribution of additive gene effects in late generations of segregating populations. However for 1000-grain additive gene effects can be exploited in early generations. For exploitation of all type of gene effects, bi-parental approach inter se crossing and /or recurrent selection may be practical for developing high yielding rice lines in advanced generations as suggested by KHATTAK et al., (2001). REFERENCES ANONYMOUS. Agricultural Statistics of Pakistan. Area, production, yield and export of some important crops. Ministry of Food, Agriculture and Livestock (Economic Wing), Islamabad, Pakistan. p.17, 209, BURTON, G.W. Heterosis and heterozygous in pearl millet forage production. Crop Science, 8, p , COMSTOCK, R.E.; ROBINSON H.F. Estimation of average dominance of genes. Heterosis. Iowa State College Press, Ames, Iowa. Chapter 30, JINKS, J.L.; PERKINS, J.M. A general method of detecting additive, dominance and epistatic components of variation for metrical traits: III. F and backcross 2 population. Heredity, 25, p , JINKS, J.L.; PERKINS, J.M.; BREEZE, E.L. A general method of detecting additive, dominance and epistatic components of variation for metrical traits: II.

6 Detection of Epistasis and Estimation of Additive Application to inbred lines. Heredity, 24, p , KEARSEY, M.J.; JINKS, J.L. A general method of detecting additive, dominance and epistatic variation for metrical traits. I. Theory. Heredity, 23, p , KETATA, H.; SMITH, E.L.; EDWARDS, L.K.; MCNEW. R.W. Detection of epistatic, additive and dominance variation in winter wheat. Crop Science, 16, p. 1-4, KHATTAK, G.S.S.; HAQ, M.A.; ASHRAF, M.; MCNEILLY, T. Genetic basis of variation of yield and yield components in mungbean (Vigna radiata (L.). Hereditas, 134, p , KHATTAK, G.S.S.; HAQ, M.A.; ASHRAF, M.; TAHIR, G.R. Triple test cross analysis for some morphological traits in mungbean (Vigna radiata (L) Wilezek. Euphytica, 126, p , NEERAJ, K.; MANI, S.C.; CHANDRA, S. Triple test cross analysis for yield and yield components in rice (Oryza satival L.). Indian Journal of Genetics and Plant Breeding, 53, p , PANDEY, D.P.; SINGH, M. Triple test cross analysis in bread wheat [Triticum aestivum (L.) Thell]. Crop Research (Hisar), 26, p , PAWAR, I.S.; YUNUS, M.; SINGH, S.; SINGH, V.P. Detection of additive, dominance and epistatic variation in wheat using triple test cross method. Indian Journal of Genetics and Plant Breeding, 54, p , POONI, H.S.; JINKS, J.L.; POONI, G.S. A general method for the detection and estimation of additive, dominance and epistatic variation for metrical traits. IV. Triple test cross analysis for normal families and their self. Heredity, 44, p , SHARMA, S.K.; MULTANI, D.S.; BAGGA, P.S. Triple test cross analysis of kernel bunt resistance in wheat (Triticum aestivum L.). Indian Journal of Genetics and Plant Breeding, 55, p.13-15, SINGH, R.K.; CHAUDHARY, B.D. Biometrical methods in quantitative genetic analysis. Kalyani Pub. Ludhina, New Delhi, Revised Ed. p , SINGH, U.P.; GANESH, M.; SRIVASTVA, C.P. Detection of epistasis and estimation of genetic variation applying modified tripe test cross analysis using two testers in pea (Pisum sativum L.). Indian Journal of Genetics and Plant Breeding, 57, p , SINGH, S.; SINGH, R.B. Triple test cross analysis in two wheat crosses. Heredity, 37, p , SUBBSARAMAN, N.; RANAGASAMY, S.R.S. Triple test cross analysis in rice. Euphytica, 42, p , SUNIL, K.; SINGH, D. Gene effects and genotype x environment interaction at various growth stages of different biomass characters in Indian mustard (Brassica juncea L. Czern Coss.). National Journal of Plant Improvement, 5, p , SUNIL, K.; SINGH, D. Detection and estimation of component of genetic variation for root characteristics, plant height, and seed yield at various growth stages and environments in Indian mustard (Brassica juncea L. Czern Coss.). Brassica, 6, p.41-45, SUREK, H.; KORKUT, K.Z. Diallel analysis of some quantitative characters in F 1 and F 2 generations in rice. Egyptian Journal of Agricultural Research, 76, p ,1998. VIJAYAKUMAR, S.B.; KULKARNI, R.S.; MURTHY, N. Triple test cross analysis in rice. Indian Journal of Genetics and Plant Breeding, 56, , Table 1 - Distinctive characteristics of genotypes used in the studies. Days to Yield per Genotype Origin Growth type flowering plant (g) Basmati 385 TN1 x Basmati 370 Intermediate DM 25 Mutant of Basmati 370 Tall DM Mutant of Basmati 370 Dwarf DM Mutant of Basmati 370 Tall NR-1 Mutant of Magnolia x Johna- Tall Niab Irri-9 Mutant of IR-6 Dwarf

7 Detection of Epistasis and Estimation of Additive Table 2 - Analysis of variance of agronomic traits in rice genotypes Source of d.f Mean Squares variaton Plant Number Number of Grain Days to 1000-grain Yield per height of tillers grains per weight per flowering weight(g) plant (g) (cm) per plant panicle panicle (g) Replication ** * Treatments ** ** 3.14** ** 1.95** 34.86** ** Hybrids ** ** ** 1.71** 33.18** ** Parents ** ** 6.96** ** 1.38** 40.94** 53.50** Lines ** ** 10.72** ** 2.37** 65.44** 65.04** Tester ** ** 4.53* ** 0.52** 23.45** 47.87** P 1 + P 2 vs F ** 50.00** ** ** 15.89** P 1 vs P ** ** 5.42* ** 1.04** 37.50** 79.86** Lines vs Tester ** ** ** * 30.15** Hyrids vs Parents ** ** ** 8.00** 16.94** ** Error *, ** = Significant at 0.05 and 0.01 levels, respectively. Table 4 - Epistatic deviations of individual rice genotypes exhibiting significant differences. Genotype Days to flowering Plant height No. of tillers per plant Yield per plant (g) DM * DM NR * * Niab-Irri * 0.44 * = Significant at 0.05 level. Table 5 - Estimates of additive (D) and dominance variance (H) components, degree of dominance (H/D) 1/2 and direction of dominance (r s,d ) for agronomic traits in rice showing non-significant epistasis. Trait D H (H/D) 1/2 r s,d Number of grains per panicle ** ** Grain weight per panicle (g) 6.52** 0.43** * 1000-grain weight (g) ** *, ** = Significant at 0.05 and 0.01 levels, respectively. Table 3 - Analysis of variance for the test of epistasis for some agronomic traits in rice Source d.f Days to flowering Number Plant of tillers height(cm) per plant Mean Squares Number of grains per panicle Grain weight per panicle (g) grain weight (g) Yield per plant (g) Total epistasis * ** 11.24* ** i type epistasis ** ** J+l type espistasis ** 14.96* ** Total epistasis x replicates i type epistasis x replicates J+l type espistasis x replicates *, ** = Significant at 0.05 and 0.01 levels, respectively.