Identification of variability for agronomically important traits in rice mutant families

Bragantia, Campinas v. 75, n. 1, p.41-50, 2016 AbstrAct: The increase of yield potential in new rice (Oryza sativa L.) varieties has been a major challenge for genetic improvement. The generation of mutants, followed by their characterization, constitutes a great possibility to isolate and select genes and genotypes that present agronomic traits of interest. This study aimed to evaluate the effect of chemical mutagen ethyl methane sulphonate (EMS) on agronomic traits in 340 M3 families of rice derived from BRS Querência cultivar. Seeds from each family and the original genotype BRS Querência were sown in the experimental field, and the characters main panicle length, main panicle weight, main panicle grain weight, flag leaf width and plant height were evaluated. Data were subjected to analysis of variance (p ≤ 0.05), PlAnt breeding Article


intrOdUctiOn
Rice (Oryza sativa L.) is the second most-produced cereal in the world, representing a staple food source for more than half of the world's population.In 2012, Brazil was the largest rice producer (11.5 million tons) out of the Asian continent (FAOSTAT 2014), being a country with high export potential.
Rice consumption and cultivation in different regions, followed by selection of specific types of rice (variation in amylose content and/or aroma), generated a remarkable phenotypic diversity in the genus Oryza (Harland 1975).The genetic variability constitutes the principle of evolutionary processes and plant breeding, enabling its creation through mutations or expansion through hybridization, followed by selection of genotypes (Jennings et al. 1981;Jalata et al. 2011).
Mutation induction assists breeding programs by increasing variability and therefore obtaining genotypes with agronomically important traits, more adapted to different environmental conditions (Serrat et al. 2014).The spontaneous mutation rate in higher plants is very low, ranging from about 10 −5 to 10 −8 (Jiang and Ramachandran 2010).Accordingly, mutation induction is an important strategy to increase the frequency of mutations.
The use of physical and chemical mutagens may be of utmost importance for the development and incorporation of new genes and/or alleles of agronomic interest and may result in more stable genotypes, better adapted to climatic and soil conditions in Brazil.Numerous studies have shown positive results obtained with the use of mutagens, thus generating a genotype that may have agronomically desirable traits (Maluszynski et al. 1998;Martins et al. 2007).In Brazil, there are already commercial rice varieties obtained by Empresa de Pesquisa Agropecuária e Extensão Rural de Santa Catarina (EPAgRI) through mutation induction: the first, called SCS 114 Andosan, and the second, SCS 118 Marques (Ishiy et al. 2006;Schiocchet et al. 2014).
Mutation can be induced by treatment with either physical mutagens, such as ionizing radiation, or chemical mutagens, including alkylating substances, for example, ethyl methane sulphonate (EMS) (Predieri 2001;Wu et al. 2005;Talebi et al. 2012).Among chemical mutagens, EMS is the most widely used due to its ease of use and potential for producing high frequency of nucleotide substitution, generating a relatively high density of irreversible mutations (Henikoff and Comai 2003;Talebi et al. 2012).
Thus, this study evaluated the effect of the chemical mutagen EMS on agronomic traits in 340 M 3 families of rice derived from BRS Querência.

Obtaining the mutant families
Mutant genotypes were obtained by treating seeds of the rice variety BRS Querência with the chemical mutagen EMS at 1.5% (v/v) (0.15 M).Previous studies indicate that concentrations of 1.5 and 1.6% EMS were efficient in generating mutant populations of rice (Wu et al. 2005;Till et al. 2007).The variety BRS Querência, of early cycle, consists of modern Americantype plants, with smooth leaves and grains.It has high tillering capacity, strong stems and stands out for presenting long panicle with many fertile spikelets (SOSBAI 2012).
The treatment consisted of soaking a sample of 40,000 seeds in distilled water.Seeds remained submerged for 6 h and then were subjected to the mutagen for 2 h.This soak time in distilled water was set in a preliminary test, in which seeds of BRS Querência were soaked in distilled water for 24 h and weighed every hour until constant weight.This procedure aimed to standardize the water content of seeds and activate their metabolism to favor the mutagenic action of the product.Soon after the application of the mutagenic agent, seeds remained for 1 h in running water and 1 h in still water (Coimbra et al. 1999(Coimbra et al. , 2005)).Afterwards, these seeds were sown in experimental field for generation advance.

Generation advance
Seeds of the M 1 generation were sown in 2009 for generation advance in an experimental field.Plants harvested in bulk gave rise to the M 2 population.
In the following year, in experimental field, 20,000 M 2 seeds were sown, aiming at the generation advance and obtaining a population of 1,500 plants.Each plant was harvested individually, giving rise to M 3 families.

Field trial
In an experimental field, ten seeds of BRS Querência (control) were sown as well as a sample of 340 families randomly chosen, following a completely randomized design.Of these, we evaluated four plants of each family and the control, at random, where each plant was considered an experimental unit, totaling four replications.
For the field trial, the minimum cultivation system was employed according to the technical recommendations for irrigated rice (SOSBAI 2012).
Plants were evaluated for the traits: main panicle length (PL), in cm; main panicle weight (PW), in g; main panicle grain weight (PgW), in g; flag leaf width (FLW), in cm; and plant height (PH), in cm.Data were subjected to analysis of variance (p ≤ 0.05), and the comparison between the means was performed by Dunnett's test at a 5% level of significance.Subsequently, multivariate analysis was run for the separation of families using cluster analysis by the Tocher's method (Rao 1952), based on the generalized Mahalanobis distance D 2 , as well as the canonical variable analysis (Cruz and Regazzi 1997).The relative contribution of each variable in the determination of genetic dissimilarity was detected by the method proposed by Singh (1981).Statistical analyses were performed using the genes software (Cruz 2001).

resUlts And discUssiOn
The results presented by analysis of variance (Table 1) evidenced significant differences at 5% significance by F-test between mean values of mutant families for all traits, demonstrating the efficiency of mutation induction with EMS at 1.5%.
Results of Dunnett's test for the five traits studied are listed in Tables 2 to 6.These data indicated genetic variability among the families studied, i.e. for all the traits there were families with mean values significantly different from the control BRS Querência.
Considering the main panicle length (Table 2), seven families presented mean values higher than that of the control (24.72 cm), and only the M 3 213 family showed smaller main panicle length, while the others remained in the same length class of the control.
The panicle length is positively related to the number of grains per panicle and consequently is associated with grain yield (Ashfaq et al. 2012;Aditya and Bhartiya 2013;Ranawake et al. 2013).In this way, the mutagenic agent used promoted a significant increase in the main panicle length of some families (M 3 331, M 3 335, M 3 332, M 3 277, M 3 217, M 3 336 and M 3 218), which may be correlated with an increased grain yield (Table 2).
As for the main panicle weight (Table 3), four families (M 3 36, M 3 34, M 3 28 and M 3 2) had mean values higher than the control (3.19 g) and only one family (M 3 277) showed a lower mean.The other families exhibited the main panicle weight significantly similar to the control.
The main panicle weight directly and positively affects grain yield and is correlated with the number and weight of grains per panicle (Senapati et al. 2009;Abd Allah et al. 2010;Li et al. 2014), representing a desirable trait in rice breeding programs.
Families showing increased main panicle weight also presented a higher value of main panicle grain weigh than that of the control, suggesting that mutation interfered with the accumulation of photoassimilates (Tables 3, 4).With regard to main panicle grains (Table 4), EMS caused an increase in the trait, with six mutant families exhibiting a better performance than the control.Whereas the control had a mean grain weight of 2.89 g, superior mutant families presented a mean value of 4.40 g.This result observed in these mutants can be related to a greater yield, since the grain mass is a major component of yield (Jeng et al. 2006;Song et al. 2007;Hasan et al. 2013;Karim et al. 2014).It is worth emphasizing that this increase in the main panicle grain weight will be relevant provided that there is no reduction in other components of grain yield.

M 3 family with mean values lower than that of the control brs Querência
No family had a mean value lower than the control
Our results suggest that possibly the mutagen promoted changes in alleles that confer short height, causing this increase.The SD1-Semidwarf1 gene, isolated from semidwarf rice, has shown to be related to gibberellin biosynthesis (Monna et al. 2002;Sasaki et al. 2002), which is a key hormone for controlling plant height.It should be noted that the mutation sd1 is considered recessive (Ashikari et al. 2002).
Comparing the results for the five traits evaluated in the study, plant height was the most affected by the mutagen EMS, with a higher number of families (52.05%) differing from the control BRS Querência (Table 6).
According to results obtained by Martins et al. (2005), the induction of ionizing mutation was effective in reducing plant height in rice genotypes.The use of EMS, at a dose of 0.04 M, in different rice mutant generations leads to an increment of the trait plant height (Siddiqui and Singh 2010).
In general, cereal breeding programs tend to select short plants, because tall plants without thick stems usually have low yield potential and are more susceptible to lodging (Ni et al. 2000).However, families with increased height should not be immediately discarded, and other traits should also be considered, such as stem thickness.It should be emphasized that taller plants have the benefit of reducing the amount and quality of weeds (Balbinot Junior et al. 2003).
Many factors influence the induced mutation rate, such as concentration of mutagen, mutagen exposure time, washing time, types of organs treated (seed or pollen) and species or genotype chosen.Seed structure may vary from species to species, which can influence the efficacy of the mutagenic agent.For example, in paddy rice, the chemical product used must penetrate the husk first and then interact with the embryo, inducing mutation (Xin et al. 2008).
The cluster analysis by Tocher's method (Table 7) revealed the formation of seven groups.The first one grouped most families of mutants together with the control BRS Querência; the second group was composed of five families (M 3 331, M 3 332, M 3 335, M 3 217 and M 3 277); the third one grouped three families (M 3 44, M 3 143 and M 3 160); the fourth (M 3 218 and M 3 336) and fifth (M 3 89 and M 3 219) groups were formed by two families each, while the sixth and seventh groups had one family each (M 3 213 and M 3 289, respectively).
Regarding the relative contribution of each trait to the genetic dissimilarity among mutant families (Table 8), the 340 families evaluated as well as the control presented, in descending order of contribution, the traits: main panicle length, main panicle grain weight, flag leaf width, plant height and main panicle weight.Among them, the traits main panicle length, main panicle grain weight and flag leaf width contributed with 69.49% to genetic dissimilarity, though the traits plant height and main panicle weight contributed with only 30.51%.
The trait that contributed most to the genetic dissimilarity was the main panicle length (31.07%), with values ranging from 36.07 cm (M 3 331 family) to 18.62 cm (M 3 213 family).The second most important variable for genetic dissimilarity was the main panicle grain weight (19.31%), with values ranging from 4.58 g (M 3 36 family) to 1.67 g (M 3 81 family).
The variances (eigenvalues), percentage variances and cumulative variances of canonical variables are presented in Table 9.As the estimates of eigenvalues, corresponding to the first three canonical variables, obtained 80.41% of the total variance of the analyzed traits, it was possible to explain satisfactorily the variability among the evaluated mutant families and represent them in a three-dimensional scatter plot (Figure 1).
The graphic dispersion indicated the distinction of mutant families into groups and can be used as a strategy to select different families to be used in artificial crosses in breeding programs.The groups formed by the graphic dispersion of scores (Figure 1) followed the same trend of the groups obtained by the Tocher's optimization method (Table 7), with the exception of group III (Tocher), which was not separated by the canonical variable analysis, appearing clustered with genotypes from group I.
There is genetic dissimilarity among rice families, and the traits contributing most to it are the main panicle length, main panicle grain weight and flag leaf width.

table 1 .
Summary of the analysis of variance for the traits main panicle length, main panicle weight, main panicle grain weight, flag leaf width and plant height of 340 M 3 mutant families and control BRS Querência.Federal University of Pelotas, Pelotas (RS), Brazil, 2015.

and 213 M 3 family with mean values lower than that of the control brs Querência 277table 2 .
Distribution of 340 M3 mutant families in relation to the control BRS Querência for the main panicle length.Federal University of Pelotas, Pelotas (RS), Brazil, 2015.

table 3 .
Distribution of 340 M 3 mutant families in relation to the control BRS Querência for the main panicle weight.Federal University of Pelotas, Pelotas (RS), Brazil, 2015.

table 4 .
Distribution of 340 M 3 mutant families in relation to the control BRS Querência for the main panicle grain weight.Federal University of Pelotas, Pelotas (RS), Brazil, 2015.

table 5 .
Distribution of 340 M 3 mutant families in relation to the control BRS Querência for the flag leaf width.Federal University of Pelotas, Pelotas (RS), Brazil, 2015.

and 213 M 3 family with mean values lower than that of the control brs Querência
No family had a mean value lower than the control

table 6 .
Distribution of 340 M 3 mutant families in relation to the control BRS Querência for the plant height.Federal University of Pelotas, Pelotas (RS), Brazil, 2015.

table 7 .
Grouping of 340 M 3 mutant families and the variety BRS Querência by Tocher's method based on the Mahalanobis distance, considering traits evaluated in M 3 families.FederalUniversity of Pelotas, Pelotas (RS),Brazil, 2015.

table 8 .
Estimates of the relative contribution of each trait for genetic divergence between rice mutant families, based on the partitioning of total D 2 .Federal University of Pelotas, Pelotas (RS), Brazil, 2015.
S.j = Estimates of the relative contribution of each trait.

table 9 .
Variances (eigenvalues), percentage variances and cumulative variances of canonical variables, aiming to estimate the dissimilarity among rice mutant families.Federal University of Pelotas, Pelotas (RS),Brazil, 2015.