Generation Mean Analysis of Some Economic Traits in Okra (Abelmoschus esculentus L. Moench)

MATERIALS AND METHODS

Experimental design: Six basic sets of generations namely P₁, P₂, F₁, F₂, BC₁ and BC₂ were derived from three crosses involving six contrasting genotypes of okra. Two local cultivars i.e., Cairo Red (HK) and Romy (R) were used in this study as well as four genetically divergent parents of okra that previously in 2009 created and developed by Soher El-Gendy (El-Gendy, 2012). These four genetically divergent parent lines are: P₁, P₆, P₇ and P₉. The crosses were referred as: Cross₁ (HK X R), Cross₂ (P₁ X P₇) and Cross₃ (P₆ X P₉). These crosses among the accessions were sown during summer seasons through 2010 at El-Baramon experimental farm, horticultural institute (Mansoura Research Station, ARC). In summer of 2011, F₁ seeds were planted in the field to produce F₂ seeds; subsequent flower buds were back crossed to produce the first backcross (BC₁) and second backcross (BC₂) generations. Parents, F₁, F₂, BC₁ and BC₂ from each of the three crosses were sown in a complete randomized block design (RCBD) with three replications during summer season 2012 at El-Baramon experimental farm of the horticultural institute (Mansoura Research Station, ARC).

Data from these six generations were recorded on three plants chosen at random from each plot for the following traits: Number of seeds/pod (N. S/p); Weight of 100 seeds (W100 sec g) Number of days to 50% flowering (N. DF); Plant height (PH cm); Number of Branches/plant (NB/p); Number of leaves/plant (N. L/p); Fruit Diameter (FD cm); fruit length (FL cm) and total yield per plant (TY/p kg).

Statistical and genetic analysis: Using SAS software (SAS 9.1), analyses of variances were done for six populations (The two parents, F₁, F₂, BC₁ and BC₂) within each cross with respect to all the studied traits. In addition, analysis of variance according to (RCBD) analysis of variance for the studied traits was made to detect the significance of the observed differences among and within crosses (Singh and Narayanan, 2000).

Estimation of heterosis (H_--): Estimates of Heterosis (%) were computed as the percent deviation of F₁ mean performance over that either mid or better parent as follows according to Abd El-Haleem et al. (2010):

Heterosis from the mid-parents:

where, Heterosis deviation = and Variance of heterosis deviation:

Heterosis from the better-parent:

where, Heterosis deviation = and Variance of heterosis deviation = V_F1+V_Bp.

The t values of T-test were computed as follows:

Estimation of Inbreeding depression (I.D.): Its value was computed as the percent deviation of F₂ mean than their the corresponding F₁ mean from the following equation according to Abd El-Haleem et al. (2010):

The t values of T-test of I.D. were computed as follows:

The scaling tests (A, B and C) and their variances were computed for each trait for testing deviations of segregation from the additive and dominance model of gene effects were computed according to Mather and Jinks (1982) as follows:

where, V_P1, V_P2, V_F1, V_F2, V_BC1 and V_BC2 were estimated according to Scheffe (1959).

The t vales of T-test are calculated as follows:

In each test, the degrees of freedom is sum of the degrees of freedom of various generations involved. Also, the significance of any one of these scales is taken to indicate the presence of non-allelic interaction (Singh and Chaudhary, 1977).

Therefore, the six parameter model is used for estimating various genetic components i.e., m, a, d, aa, ad and dd, which stand for mean effects, additive, dominance, additivexadditive, additivexdominance and dominancexdominance gene effect, respectively. These genetic components and their variances were computed according to Jinks and Jones (1958) and Khodambashi et al. (2012) as follows:

Standard error of these parameters and calculated “t” values were estimated according to Gamble (1962) and Singh and Chaudhary (1977) in a similar manner as done in case of scaling test.

In the absence of non-allelic interaction, the additive-dominance model is adequate. Thus, m, a and d their variances were estimated according to Jinks and Jones (1958) as follows:

Significance of the genetic effects is tested in a similar manner as done in case of scaling test.

Estimation of genetic parameters
Phenotypic coefficient of variability (pcv) and genotypic coefficient of variability (gcv): Its values were computed according to Singh and Chaudhary (1977) as follows:

Heritability in broad sense (H²_bs) was computed according to Mather and Jinks (1982) using the following equation:

where, V_E is environmental variance.

Heritability in narrow sense (H²_ns) was computed for the characters using the following equation (Adeniji and Kehinde, 2012):

Expected Genetic advance under selection (GA%): The estimate of expected genetic advance (response to selection) expressed as percentage of mean at 5% selection intensity (i) (selection differential, K = 2.06) was computed by the following formula according to Eshghi and Akhundova (2010) and Deb and Khaleque (2009):

RESULTS AND DISCUSSION

In this study, yield and other traits were investigated. Therefore, several analyses of variances were made in order to test the significance of differences among crosses as well as populations within crosses. The results of the analysis of variance and the mean squares of yield and other traits of crosses and their populations are presented in Table 1.

The obtained results revealed that most generations within all crosses had significant differences of all studied traits indicating the existence of genetic variation and possibility of selection for yield and yield components. This finding indicates that the further partition of genetic variance to its components and the comparisons between means are valid with respect to the traits under study. In fact the development of any plant breeding program is dependent upon the existence of genetic variability. Furthermore, the efficiency of selection and expression of heterosis also depends largely upon the magnitude of genetic variability present in the plant population (Singh and Narayanan, 1993; Singh and Chaudhary, 1999).

Therefore, the means and standard deviations of the six generations as well as heterosis and inbreeding depression with respect to the three crosses for all studied traits are presented in Table 2.

The variance estimates from data of standard deviation for all studied traits in segregating generations were greater than that of F₁ and the parents. This showing that genetic variability does exist among the generation means in all studied crosses. Similar results were obtained by Adeniji and Kehinde (2012).

The results indicated that line P₆ was the best parent compared to the crosses and its generations in most studied traits, especially in total yield per plant with the mean value of 1.51 kg. Also, indicated that the local cultivar R was the earlier genotype with the mean value of 43.7.

Meanwhile, all of F₁’s were lower than the better-parent or mid-parents value for number of seeds/pod, number of branches/plant, fruit diameter and total yield per plant. These results led to the negative heterosis values for these traits in all studied crosses which indicate that dominance direction was toward the low respective parent in these traits. However, heterosis above the better parent for weight of 100 seeds, plant height, number of leaves/plant and fruit length in cross 1, indicated that the dominance direction was toward the best parent in this cross with respect to these traits. These results were in agreement with those obtained by Abd El-Haleem et al. (2010).

Also, these results showed desirable negative values of heterosis over mid-parents for number of days to 50% flowering trait of all studied crosses. However the cross 2 only showed desirable negative and highly significant of heterosis over mid-parents and better-parent with values of -14.58 and -13.03%, respectively. This result indicates that the dominance direction was toward the best parent in this cross with number of days to 50% flowering trait.

Positive and significant inbreeding depression was observed for plant height trait in cross 1 and cross 3 with values of 30.7 and 30.1%, respectively. This indicates the role of dominant gene action in the inheritance of this trait. This result disagreed those obtained by Khanorkar and Kathiria (2010), while it was in agreement with the result obtained by Senthil Kumar et al. (2005). Also, the positive and significant inbreeding depression was observed for number of days to 50% flowering in cross 3 with value of 18.29%. This indicates the possibility to obtain desirable segregants for earliness in subsequent segregating generations of this cross. This result was in agreement with the results obtained by Khanorkar and Kathiria (2010).

On the other hand, insignificant positive or negative inbreeding depression values were registered in the other traits for all crosses. These indicated that the presence of vigor in F₂ can be attributed to additive and epistatic gene action. Such crosses in these traits are expected to give segregants superior to the better parent in these traits, which may be handled through pedigree breeding method. Similar results were obtained by Surendira Kumar et al. (2004). Low inbreeding depression suggests that increased retention of vigor in F₂'s is expected to be mainly due to accumulation of favorable additive genes (Shukla and Gautam, 1990).

Table 1:	Analysis of variance and mean squares for all studied traits of crosses and their populations
*,**Significant at 0.05 and 0.01 levels probability, respectively, D.F: Days to 50% flowering, NS/p: Number of seeds/pod, W100s g: Weight of 100 seeds, N. DF: Number days to 50% flowering, PH (cm): Plant height, N. B/p: Number of branches plant, N. L/p: Number of leaves/plant, FD (cm): Fruit diameter, FL (cm): Fruit length, TY/p kg: Total yield per plant

Table 2:	Mean performance and standard deviations of parents, F1, F2 and backcross generations in three okra crosses as well as heterosis and inbreeding depression for all studied traits
*,**Significant at 0.05 and 0.01 levels probability, respectively, D.F: Days to 50% flowering, NS/p: Number of seeds/pod, W100s g: Weight of 100 seeds, N. DF: Number days to 50% flowering, PH (cm): Plant height, N. B/p: Number of branches plant, N. L/p: Number of leaves/plant, FD (cm): Fruit diameter, FL (cm): Fruit length, TY/p kg: Total yield per plant

The results of the A, B and C scaling tests for assessing the validity of additive-dominance models are given in Table 3. The values of the A, B and C scaling tests were not significant for all crosses in weight of 100 seeds, fruit diameter, fruit length and total yield per plant traits. The same results recorded for cross 1 with number branches/plant and number of leaves/plant traits and in cross 3 for number of seeds/pod trait. These findings indicate the absence of epistasis (non-allelic interaction) and the additive-dominance model was adequate to demonstrate the genetic variation. Thus, it is important in the inheritance of the above mentioned studied traits in such crosses. The result of weight of 100 seeds trait is in agreement with those obtained by Adeniji et al. (2007). On the other hand, the results of fruit length and total yield per plant traits were opposite with that obtained by Akhtar et al. (2010).

The additive gene effects (a) were significant in cross 2 for weight of 100 seeds traits as well as in cross 1 and 3 for fruit diameter and fruit length traits, while the dominance gene effect (d) was significant in cross 3 for number of seeds/pod trait. This finding, explain the role of both additive and dominance gene effects in inheritance of these traits in such crosses.

The non-allelic interactions were found to be operating in the control of genetic variation among the six generations for all crosses in number of days to 50% flowering and plant height traits. The same result was recorded for cross 2 and 3 with number of branches/plant and number of leaves/plant traits and for cross 1 and 2 with number of seeds/pod trait. These findings indicate the presence of epistasis (non-allelic interaction) and the results were in agreement with those obtained by Akhtar et al. (2010).

The mean effect [m] was significant for all crosses in number of seeds/pod, number of days to 50% flowering, plant height, number of branches/plant and number of leaves/plant traits, except cross 1 for number of branches/plant and number of leaves/plant traits.

Table 3:	The scaling test and estimates of the genetic components: additive, dominance and interaction parameters as well as standard errors in three okra crosses for all studied traits
The scaling tests (A, B and C), M = Mean, [a] = Additive effects, [d] = Dominance effects, [aa] = Additivexadditive effects, [ad] = Additivexdominance effects, [dd] = Dominancexdominance effects *, **: Significant at 5 and 1% level of probability, respectively,D.F: Days to 50% flowering, NS/p: Number of seeds/pod, W100s g: Weight of 100 seeds, N. DF: Number days to 50% flowering, PH (cm): Plant height, N. B/p: Number of branches plant, N. L/p: Number of leaves/plant, FD (cm): Fruit diameter, FL (cm): Fruit length, TY/p kg: Total yield per plant

These results indicated that the above mentioned studied traits in such crosses were quantitavely inherited exactly as what was found in accordance with (Abd El-Haleem et al., 2010).

The magnitude of additive gene effects (a) was small relative to the corresponding dominance effects (d) in most cases, suggesting that pedigree selection method is a useful breeding program for improving these populations. However, the negative value of (d) observed in most cases except in fruit length trait indicated that the alleles responsible for less value of the trait were dominant over the alleles controlling high value. These results are in harmony with those obtained by Alake et al. (2012), Adeniji et al. (2007) and Khattab et al. (2010).

The results indicated that the dominancexdominance effects were greater in magnitudes than additivexadditive and additivexdominance in all cases which recorded non-allelic interaction except in case of cross 1 for number of days to 50% flowering trait. When non-additive portion is larger than additive, the improvement of these traits needs intensive selection through later generations (Khattab et al., 2010).

The dominance gene effect (d) was significant in all crosses for number of seeds/pod trait. These results indicated that the dominance effect (d) was important in the inheritance of this trait. Also, in this trait the additivexadditive (aa) and dominancexdominance (dd) gene effect were significant in cross 1 and 2, which may lead to hinder the progress of selection leading to losses of favorable genotypes during the early generation of selection. According to (Mather and Jinks, 1982), opposite signs of additivexadditive and dominancexdominance indicated prevalence of duplicate epistasis and complementary epistasis.

Table 4:	Magnitude of phenotypic coefficient of variation, genotypic coefficient of variation, heritability and genetic advance in three okra crosses for all studied traits
D.F: Days to 50% flowering, NS/p: Number of seeds/pod, W100s g: Weight of 100 seeds, N. DF: Number days to 50% flowering, PH (cm): Plant height, N. B/p: Number of branches plant, N. L/p: Number of leaves/plant, FD (cm): Fruit diameter, FL (cm): Fruit length, TY/p kg: Total yield per plant

The complementary effect will produce new recombinants capable of improving yield. Therefore, the improving of this character could be achieved through hybrid breeding method.

The phenotypic (PCV) and genotypic (GCV) coefficients of variation, heritability in broad sense (H²_bs), heritability in narrow sense (H²_ns) and expected genetic advance (GA%) were calculated as percent of F₂ mean for all traits are presented in Table 4.

The magnitude of PCV was higher than that of GCV for all the traits indicating that these traits are more sensitive to the environmental conditions (Abd El-Haleem et al., 2010). The GCV and PCV were high for number of branches/plant, number of lecules/pod, fruit diameter, fruit length, total yield per plant and number of seeds/pod for cross 3, ranged from 23.14 (in case of cross 3 for number of seeds/pod trait) to 40.08 (in case of cross 2 for fruit length trait) for PCV and ranged from 19.82 (in case of cross 1for number of leaves/plant trait) to 39.28 (in case of cross 2 for fruit length trait) for GCV. These results indicated that the maximum variability among the genotypes selected for evaluation and thus these traits provide a better selection chance of desirable genotypes (Jindal et al., 2010). Meanwhile, the GCV and PCV were moderate for number of seeds/pod except in cross 3, weight of 100 seeds and plant height ranged from 12.74 to 21.10 for PCV and from 11.05 to 19.10 for GCV. Also, low GCV and PCV were recorded for number of days to 50% flowering.

Heritability was estimated in each of broad and narrow sense. Estimation of broad-sense heritability indicated higher importance of genetic effects in control of traits. Comparison between broad and narrow-sense heritabilities revealed equal importance of additive and non-additive effects in genetic control of traits (Reza, 2012). The higher values of narrow sense heritability for a particular character indicated that it is controlled largely by genes acting in an additive fashion (Khanorkar and Kathiria, 2010). Thus, expected genetic advance is yet another important genetic parameter that aid breeder in a selection program (Shukla et al., 2004). The knowledge of heritability along with expected genetic advance aids in drawing valuable conclusions for selection of breeding methods to be employed for further improvement of the traits.

From Table 4, the results of heritabilities and expected genetic advance indicated that the heritability in broad and narrow sense were generally found to be high in magnitudes in all crosses for number of seeds/pod, weight of 100 seeds, plant height, number of branches/plant, fruit diameter and fruit length, ranged from 72.43% (in case of cross 1 for plant height trait) to 98.03% (in case of cross 3 for fruit diameter trait) for heritability in broad sense and ranged from 72.27 to 97.72% (in the above mentioned cases) for heritability in narrow sense. The analysis of expected genetic advance in percentage of mean for the above mentioned studied traits in all crosses showed that highest expected genetic advance (more than 19%) indicated good response to selection. High heritability estimates accompanied with high estimates of expected genetic advance expected in the next generation for these traits, indicated the preponderance of additive gene action for the expression of these traits which is fixable in subsequent generations. This also provides the evidence that larger proportion of phenotypic variance has been attributed to genotypic variance and reliable selection could be made for these traits on the basis of phenotypic expression (Bello et al., 2012).

Moderate heritability (ranged from 51.50 to 59.35% in broad sense and from 50.90 to 52.88% in narrow sense) accompanied with high expected genetic advance (ranged from 39.15 to 41.15%) were recorded for total yield per plant trait in all crosses, providing little chance for its further improvement. However, care must be taken while breeding for this complex trait as it is considerably influenced by environmental factors. It seems a limited scope of improvement could be achieved for this trait within these genotypes (Bello et al., 2012).

Also, moderately high to low heritability (ranged from 77.92 to 14.81% in broad sense and from 73.75 to 11.11% in narrow sense) accompanied with low expected genetic advance (ranged from 0.79 to 5.55%) were observed for number of days to 50% flowering, which might be due to non-additive gene effects. Hence, little improvement by selection is likely in this trait. This result is in line with those reported by Abdelmageed (2010). Straight forward selection from the segregation population of this trait does not seem to be possible, the genetic variation existed in this trait could be improved successfully following reciprocal recurrent selection. Furthermore, this trait in which over dominance was involved may advantageously be utilized by the breeders to develop hybrid, as suggested by Ojaghi and Akhundova (2010).

From the above mentioned results, high and moderately high GCV, PCV, heritability and GA% of mean in most crosses for all studied traits except number of days to 50% flowering suggested that these traits could be transmitted to the hybrid progeny and phenotypic selection based on these would be effective. Also, reported low GCV, PCV, heritability and GA% for number of days to 50% flowering. This finding was due to the influence of environment on this trait.

CONCLUSION

Variations observed in this research study for all studied traits were indicative of the differences in the genetic constitution of used okra genotypes. These differences led to better chance and large scope of selection and breeding programs. Applying generation mean analysis to three special okra crosses made it possible to conclude that there was very high direct genotypic effects especially additive and non-additive variances that were found to be important in the genetic control of all studied traits. Additive-dominance model was adequate to demonstrate the genetic variation and it is important in the inheritance for yield and yield component studied traits. While, non-allelic interactions were found in the other traits for most crosses. Also, High and moderately high GCV, PCV, heritability and GA% of mean in most crosses for most studied traits except number of days to 50% flowering suggested that the phenotypic selection could be effective for improving these traits.