SELECTION EFFECTS IN RICE AS ASSESSED BY GENETIC ANALYSIS IN SEGREGATING POPULATIONS

Estimation of selection effects on changes of a trait is of vital importance for the success of any plant breeding program, and helps to select the desirable breeding method. Heritability and genetic advance are important selection parameters, and selection success is a reflectance of selection response. To estimate selection effects on rice genetic parameters, a research was conducted using 4 different generations (two parents: Neda and Sadri, and two segregating populations: BC1F1 and BC1S1). After development of a backcross population, a single plant (BC1#4) was selected based on its desirable performance, particularly in heading date and seven other morphological traits. BC1F1 population compared to mid-parent performance showed advance for heading date, plant height, tiller number, hundred seed weight, weight of filled seeds per panicle and grain yield per plant, while mean performance of BC1F1 population compared to BC1S1 population showed advance only for heading date, plant height, tiller number and grain yield. Prevalence of additive genetic effects in controlling panicle weight, hundred seed weight, weight of filled seeds per panicle, plant height and heading date was observed, and in contrast prevalence of non-additive effects in controlling grain yield was observed. High general heritability was observed for most traits, while only heading date and plant height showed a considerable specific heritability (60.7% and 67.5%, respectively), and grain yield showed a relatively low specific heritability (37.0%). High expected genetic advance (∆Ge) was obtained for tiller number (49.4%), followed by grain yield (43.5%) and plant height (35.5%), while the highest real genetic advance (∆Ge) was obtained for heading date (-8.5%) and tiller number (5.4%). High selection success was obtained only for heading date (51.8%). Altogether, the obtained results gave promise for selecting progenies with early maturity and semi-dwarfism in early segregating generations, while they suggested preference of heterosis for improvement of grain yield.


INTRODUCTION
Rice is one of the most important agricultural products in the world earning substantial foreign exchange and is a staple food crop in densely populated Asia DOI: 10.1515/plass-2018-0006 (Alavi et al., 2009). Genetic improvement of this crop plant will serve the mankind living on our planet. Evaluation of important traits with direct or indirect effect on grain yield and sustainability of rice growers is indispensable for successful breeding of rice. Estimation of genetic parameters helps our understanding about gene action, identification of components of genetic variances and, finally facilitates the selection of a desirable breeding method (Ahmadikhah et al., 2007, Immanuel et al., 2013Ahmadikhah et al., 2015). The knowledge of genetic variability present in a given crop species for the character under improvement is of paramount importance for the success of any plant breeding program. Heritability and genetic advance are important selection parameters. Heritability estimates along with genetic advance are normally more helpful in predicting the gain under selection than heritability estimates alone (Bisne et al., 2009, Immanuel et al., 2013. Heritability values have been variable depending upon the genetic nature of genotypes for different studied characters. Vivek et al. (2000) observed high heritability coupled with high genetic advance for harvest index, biological yield per plant and grain yield per plant in evaluation of 39 tropical Japonica rice genotypes. Mishra and Verma (2002) evaluated 16 rice genotypes along with 72 F 1 hybrids and noted high heritability with high genetic advance for flag leaf area and plant height, indicating dominant role of additive gene action. The association of high heritability with high genetic advance was observed for plant height and grain yield per plant by Mahto et al. (2003). Swati and Ramesh (2004) reported high heritability for grain yield while moderate heritability for flag leaf area and plant height. Hosseini et al. (2005) observed 61 percent broad sense heritability for grain yield in rice. Saleem et al. (2008) noted high broad sense heritability and expected genetic advance in response to selection in next generation for all the studied traits. Genetic advance for plant height and yield per plant, calculated equal to19.4% and 14.6%, respectively.
High heritability coupled with high genetic advance was exhibited by harvest index, total number of chaffy spikelets per panicle, grain yield per plant, total number of filled spikelets per panicle and spikelet fertility percentage and selection may be effective for these characters (Bisne et al., 2009). Bisne et al. (2009) obtained 98.7% general heritability for plant height, 89.4% for panicle length, 63.9% for tiller number, 98.0% for 100-seed weight, 98.7% for panicle length and 93.4% for yield per plant. Our objectives were to estimate the genetic parameters of some important traits in response to selection, to estimate the genetic advance and real selection success in rice.

Plant material and trait evaluation
Two parental rice lines, Neda (P 1 ) and Sadri (P 2 ), were crossed in 2007 to produce F 1 hybrid. First generation of backcross (BC 1 ) was produced in 2008 by crossing of F 1 with Neda. BC 1 population (consisted of 25 plants) was sown in 2009. One BC plant was selected based on some desirable morphological characters, particularly early maturation, shorter height and longer panicle. The self-pollinated seeds of this plant (BC 1 S 1 ) were sown next year (spring 2010). Each generation was sown in three replicats with crop spacing of 35 cm x 35 cm. For each parent 30 plants were analyzed, however for two BC 1 and BC 1 S 1 generations, 25 and 78 plants were analyzed, respectively. Eight important quantitative traits including heading date (HD; days from germination to panicle emergence), plant height (PLH; cm), panicle length (PL; cm), tiller number (TN), hundred seed weight (HSW; g), panicle weight (PW; g), filled seed weight of panicle (WFS; g) and grain yield per plant (GY; g plant -1 ) were evaluated on two parents, F 1 , BC 1 and BC 1 S 1 plants.

Studied genetic parameters
Some important parameters were evaluated on plants of each generation (P 1 , P 2 and BC 1 and BC 1 S 1 ) including mean, coefficient of variation (C.V), phenotypic variance (VP), environmental variance (VE), genetic variance (VG), broad-sense heritability (h 2 b), narrow-sense heritability (h 2 n) and genetic advance due selection (∆G). Data were analyzed using GLM procedure and subsequent univariate tests in SPSS software (Kinnear and Collin 2000). Trait means were compared using Duncan multiple test, and graphs were drawn in excel spreadsheet. Mean square of experimental error (EMS) in ANOVA was considered as environmental variance (V).
Degree of dominance (d) was calculated as: where, VA and VD are additive and dominance genetic variances, respectively. Broad-sense heritability (h 2 b ) and narrow-sense heritability (h 2 n) were calculated as: where, VG is genetic variance, VP is phenotypic variance, R is selection response and D is differential of selection. Expected genetic advance (∆G e ) and real genetic advance due selection (∆G r ) were calculated as: where, k is a constant coefficient (here, k was considered equal to 2.06 for selection severity of 5%) and GM is grand mean in the experiment.

Analysis of variance (ANOVA)
Analysis of variance for 4 generations is shown in Table 1. This analysis showed that all eight traits significantly differed in studied generations, indicating that selection had significant effect on mean performance of studied traits. Grain yield h showed highest coefficient of variation (C.V) (13.1%), followed by tiller number (8.8%), while plant height and heading date had the least C.V (~1.9%). Table 1 Analysis of variance (ANOVA) for different traits in the study ** differences are significant at 1% level of probability.
Selection effect on mean performance in BC 1 F 1 generation Mean performance of different traits for two parents, BC 1 F 1 , BC 1 S 1 and selected plant in BC 1 F 1 (BC 1 #4) is shown in Table 2. As seen, mean of BC 1 F 1 population compared to mid-parent mean was advanced in heading date (-1.6 days), plant height (-14 cm), tiller number (9.3), hundred seed weight (0.23 g), weight of filled seeds per panicle (0.56 g) and grain yield per plant (7.7 g). However, its mean compared to mid-parent mean did not differ for the remained traits. Based on better performance, particularly in heading date and plant height one plant (BC 1 #4) in this population was selected to develop next generation  (Table 2). This plant had superiority over mean performance of BC 1 F 1 generation in most other studied traits, too (Table 2; Fig 1) and hence, BC 1 S 1 generation was developed from its self-pollination. As seen in Table 2, mean performance of BC 1 S 1 population compared to BC 1 F 1 population was advanced in heading date (-6.8 days), plant height (2 cm), tiller number (1.1) and grain yield (2.9 g per plant). However, for the remained traits was not observed further progress compared to BC 1 F 1 generation. Mean performance of different traits for best single plants in BC 1 S 1 population is shown in Table 2. As seen, these plants have advantage in all studied traits over all preceded generations including two parents, encouraging possibility for continuing selection to obtain relatively supper rice lines (each harboring one or more desirable traits). Such lines had an improved performance even compared to better BC 1 F 1 plant (BC 1 #4); their superiority in heading date (-2 days), plant height (-11 cm), panicle length (4.5 cm), tiller number (22), grain yield (~101 g) and in the remained traits was considerable over selected BC 1 #4 plant (

Estimation of genetic parameters
Some important genetic parameters of the studied traits are shown in Table 3. As seen, degree of dominance (d) for panicle weight (0.523), hundred seed weight (0.573), weight of filled seeds per panicle (0.664), plant height (0.671) and heading date (0.71) is significantly lower than 1, indicating predominance of additive genetic effects in controlling these traits. However, value of this parameter for panicle length (0.856) and tiller number (0.879) is skewed toward 1, indicating that these traits are controlled by partial dominance. It seems that only grain yield is controlled by non-additive gene effects of dominance nature (d=~1.065). Shahid et al. (2012) also reported that grain yield of autotetraploid rice mainly was regulated by dominance variance. Above results show that improvement of most studied traits is possible via selection in segregating populations, while for grain yield breeders must rely on heterosis vigor (Shahid et al., 2012).
It seems that half traits have a high degree of general heritability (h 2 b ). Four traits such as plant height, heading date, panicle length and tiller number showed 87-91% general heritability (Table 3), a considerable part of which could be due to non-additive effects, because both additive and non-additive variances form the genetic variance. For the remained traits h 2 b exhibited only a moderate value between 69.7-78.9%. In contrast to general heritability, the specific heritability (h 2 n ) is more applicable for selection issues and genetic progress (Zhao et al., 2006, Ahmadikhah et al., 2007. Actual heritability can be equalized to specific heritability when many generations (parents, F 1 , F 2 , BC 1 F 1 etc) are not available to breeders. Therefore, with availability of two nonsegregating generations (two parents) and two segregating sequential generations (such as BC 1 F 1 and BC 1 S 1 in this research), it is possible to obtain selection response and differential of selection and hence, narrow-sense heritability can easily be calculated using their values. Values of selection response, differential of selection and h 2 n have been shown in Table 3. As seen, only heading date and plant height have a considerable h 2 n (60.7% and 67.5%, respectively). This gives promise for selection of progenies with early maturity and semidwarfism in early segregating generations. Grain yield showed a relatively low h 2 n (37.0%), again showing preference of heterosis for improvement of this trait.
The highest real genetic advance (∆Gr) was obtained for heading date (-8.5%) and tiller number (5.4%) followed by grain yield (4.3%), while for the remained traits value of ∆Gr was very low (0.4-2.3%). On the basis of obtained results, selection success was high only for heading date (51.8%; Table 3), giving the promise for further advance in few next generations, as noted by Saleem et al. (2010) and Immanuel et al. (2013).  Table 3 Important genetic parameters of eight studied traits of rice a. The sign of ∆G r for these traits was negative.