Genetic Parameters of Some Quantitative Traits in Cotton Using Triple Test Cross Analysis

INTRODUCTION:

Estimating the genetic components of the studied traits of any cotton population is important for planning an appropriate and effective breeding program. The early attempts to partition the genetic variance were done by (Fisher, 1918), classified the genetic variance into three components, additive, dominance and epistasis. Which, developed by Hayman and Mather (1955), where they indicated that epistasis can also classified to three components additive x additive, additive x dominance and dominance x dominance. Triple Test Cross technique (Kearsey and Jinks, 1968) provides un-ambiguous estimates for epistasis and in the lack of epistasis, un-biased estimation of additive and non-additive components also remains unaffected by differences in allele frequencies, degree of inbreeding as well as correlation. Successful breeding program is limited by the portion of genotypic variance due to additive gene effect as well as additive x additive epistatic interaction; because these two types of gene effect can only be retained by subsequent inbreeding. While if non-additive gene portion is larger than additive ones, the improvement of the characters studied required intensive selection through later generations; when there were significantly epistatic effects, the possibility of obtaining desirable segregates through inter-mating in early generations can led to breaking undesirable linkage group or by adoption of recurrent selection for rapid improvement (Esmail, 2007).

Al-Hibbiny et al., (2020) revealed that fixable type was most important epistatic effect than non-fixable type for all studied traits. For all of the studied traits, both additive and dominant components were significant. Lint cotton yield/plant, lint index and seed index were confirmed the presence of over-dominant, although, the degree of dominance was less than unity, confirming the occurrence of partial dominance for all studied traits. Except for lint yield/plant, lint index, and seed index, additive gene action was more essential in influencing inheritance than dominance one.

          Hassan et al., (2022) showed that (i) type of epistasis (additive x additive) showed significantly for some yield components and fiber quality traits, except for micronaire reading. While, (additive x dominance) as well as (dominance x dominance) demonstrated significant for seed cotton yield / plant, lint cotton yield/plant, lint percentage and uniformity index. The (i) type as compared to (j+l) type showed higher values for all the studied traits, with the exception of micronaire reading. Both additive and dominance were important for controlling the traits, except boll weight, micronaire reading and pressley index, which controlled with additive genetic effect. On the other side, additive component was higher than dominance component for all traits. Degree of dominance for all studied traits was less than unity, indicating partial dominance.

           El-Shazly et al., (2023) The results revealed that all genotypes, parents, crosses, and parents vs. crosses mean squares were extremely significant for all tested features, with the exception of micronaire reading in the crosses. The findings demonstrated that additive effects had a comparatively minor role in the emergence of these traits as compared to non-additive effects. The results indicated that the hybridization programme would be effective in improving the majority of the attributes studied.

The present investigation was undertaken to detect the presence of epistasis and to estimate the additive and dominance components of genetic variation of some quantitative traits in cotton. 

MATERIALS AND METHODS:

Study Area

Triple test cross (TTC) experiment was conducted out during three growing seasons (2021 to 2023) at Sakha Agricultural Research Station Kafr El-Sheikh Governorate Egypt.

Genetic materials and experimental procedures         

Ten cotton lines included wide genotypes, Giza 80 (L₁), Giza 86 (L₂), Giza 87 (L₃), Giza 88 (L₄), Giza 90 (L₅), Giza 92 (L₆), 10229 (L₇), Pima S₇ (L₈), Karshenky (L₉) and Pima S₆ (L₁₀) were crossed to 3 testers; Giza 94 (T₁), Giza 96 (T₂) and their F₁ hybrid (T₃). The origin, pedigree and category of these genotypes were presented in (Table 1). Thus the experimental materials comprised of 13 parental genotypes, 20 single cross including T₁ and T₂, and 10 three-way crosses involving T₃.

Table 1. Origin, pedigree and category for the thirteen parental cotton genotypes

During 2023 growing season, the 43-genotype evaluated in a randomized complete block design (RCBD) with three replications. Each replicate contained three rows for each genotype. Row was 4 m long, and 0.70 m width and 40 cm between hills with one plant left / hill. All agricultural practices were adopted through the growing seasons.

Ten guarded plants from each plot were used individually to collect data for the following traits: seed cotton yield (g) / plant (SCY/P), lint cotton yield (g)/ plant (LCY/P), lint percentage (L %), boll weight (g) (BW), seed index (g) (SI), lint index (g) (LI), micronaire reading (MR), pressley index (PI), 2.5% span length (mm) (UHM) and Uniformity index (UI%), these traits were estimated at the Cotton Technology Laboratories, Cotton Research Institute, ARC, Giza, Egypt.

Statistical and genetic analysis:

Analysis of variance was done as outlined by Singh and Chaudhary (1999) Epistasis detection was carried out according to the method outlined by Kearsey and Jinks (1968) and is based on the genetic model;

L_ijk = M + G_ij + R_k + E_ijk

Where,

L_ijk = Phenotypic value of cross between tester i and line j in k replication.

M = Overall mean of all single and three way crosses.

G_ij = Genotypic value of cross between tester i and line j.

R_k = Effect of k^th replication.

E_ijk = Error.

The mean squares for L_1i + L_2i – 2L_3i deviations was used for epistasis detection. The overall epistasis was partitioned into (i) type of epistasis (additive x additive) and (i + j) type due to (additive x dominance) and (dominance x dominance) gene interactions. The estimation of additive (D) as well as dominance (H) genetic components and the correlation coefficient (r) between sums (L_1i + L_2i + L_3i) and differences (L_1i - L_2i) were obtained to reveal the direction of dominance, according to Jinks and Perkins (1970). The degree of dominance was calculated as . Where, (H) and (D) indicated to dominance as well as additive variance components, respectively.

RESULTS AND DISCUSSION

Analysis of variance for the studied traits are presented in Table (2). Results revealed significant differences for each of genotypes, parents, lines and testers for all the studied traits except, for BW at both lines and testers and UI at testers only. Moreover, hybrids showed significant mean square for all studied traits, this indicating that the parent lines and testers utilized in the current study were divergent, and that significant differences were passed down via the progenies.

Also, significant differences for lines vs. testers were observed for all the studied traits, highlighting the importance of both additive and non-additive types of gene action in influencing these traits. Furthermore, hybrids vs. parents revealed significant differences in all the studied characteristics, similar results were those obtained by (Abou El-Yazied, 2014 ; Dawwam et al., 2016 ; El-Mansy et al., 2020 ; Amer, 2020 ; Said et al., 2021).

Data concerning that mean performance of the tested genotypes (13 parents, 20 single crosses as well as 10 three-way crosses) are exhibited in Table (3). The L₁ (Giza 80) gave the highest values for SI and LI, while L₂ (Giza 86) gave the best means for BW, L₃ (Giza 87) recorded the highest values for PI, L₆ (Giza 92) had the best values for SCY/P, MR, 2.5% SL and UI, while, L₁₀ (Pima S6) had the best means for L %. While, for testers, T₁ (Giza 94) had the highest values for BW, SI and LI, T₂ (Giza 96) gave the best values for SCY/P, LCY/P, MR, PI, 2.5% SL and UI % although, T₃ (Giza 94 x Giza 96) had the best mean for L%.

The results additionally showed best mean performances for the three-way cross L₆ x T₃ (Giza 92 x (Giza 94 x Giza 96)) for SCY/P, LCY/P and PI. On the other side, the three-way cross L₄ x T₃ (Giza 88 x (Giza 94 x Giza 96)) gave the highest mean values for BW, PI and 2.5% SL. The crosses L₁ x T₁ and L₃ x T₃ [Giza 80 x Giza 94 and (Giza 87 x (Giza 94 x Giza 96)] had the best means for BW. The three-way cross L₁ x T₃ (Giza 80 x (Giza 94 x Giza 96)) gave the best values for L%, SI and LI. While, the crosses L₆ x T₁ and L₄ x T₂ (Giza 92 x Giza 96) and (Giza 88 x Giza 96) had the highest mean value for MR. The cross L₆ x T₁ (Giza 92 x Giza 94) gave the best values for UI%.

Regarding to epistasis, analysis of variance (Table 4) revealed highly significant overall epistasis for all studied traits. Partition of total epistasis into (i) type of epistatic (additive x additive) and (i + j) types of epistasis (additive x dominance) as well as (dominance x dominance) indicated non-significant involvement of (i) type for all studied traits. On the other hand, (i + j) types of epistasis were highly significant for all studied traits. The epistatic type (i) interaction, was detected to be much larger in magnitudes than the other epistatic type (i + j) for all studied traits except for SI, indicating that fixable components of epistasis were more important than non fixable one in the inheritance of these trait. Thus, the breeder should take epistatic into account in producing genetic models for studying quantitative traits. Similar results were obtained by (Hussain et al., 2008 ; Sohu et al., 2010 ; El-Lawendey et al., 2010 ; Saleh, 2013 ; Jayade et al., 2014 ; Dawwam et al., 2016 ; Al-Hibbiny et al., 2020 ; El-Mansy et al., 2020)

The individual epistatic deviations of lines are shown in Table (5). The data showed that the epistatic deviations were exhibited by L₁ (Giza 80) that had significant negative for SCY/P, LCY/P, L%, SI, LI and MR. In contrast, there were significant positive for 2.5% SL and PI.  L₂ (Giza 86) was significant negative for SCY/P, LCY/P, L%, SI, LI, MR and PI. L₃ (Giza 87) was significant negative for all studied traits except, for L% and PI. Regarding L₄ (Giza 88) was negative significantly for all studied traits except, for L% and BW, while gave significant positive epistatic deviations for MR, as well as L₅ (Giza 90) was significant negative for all studied traits except, for BW and UI. On the other hand, L₆ (Giza 92) exhibited significant negative for SCY/P, LCY/P, L%, MR and PI and significant positive for BW, SI, LI and UI. Concerning, L₇ (10229) had significant negative for SCY/P, LCY/P, BW, MR and 2.5% SL but, significant positive for L%, SI, LI and UI. Regarding L₈ (Pima S7) was significant negative for all studied traits except, for L%, 2.5% SL and UI. While, L₉ (Karshenky) had significant negative for all studied traits except, for MR and 2.5% SL. whereas, L₁₀ (Pima S6) had significant negative for all studied traits except, for SCY/P and significant positive for BW and SI. It is evident that all lines exhibited epistatic deviation for most studied traits. Similar results were obtained by (Saleh, 2013 ; Abou El-Yazied, 2014 ; Jayade et al., 2014 ; Al-Hibbiny et al., 2020).

Analysis of variance for sums as well as differences between hybrids (Table 6) indicated that sums item (L_1i+L_2i­) were significant for all traits except, for BW and MR. The differences in items (L_1i – L_2i­) were significant for all traits with the exception of, BW which exhibited insignificant differences. High values of additive genetic variance were found as compared with dominance genetic variance for all studied traits except, for BW and MR. The degree of dominance (√H/D) on the other side was less than unity, suggesting the role of partial or incomplete dominance controlling for all studied traits except, for BW and MR which, showed overdominance (greater than unity). Consequently, it concluded that selection procedures in early generations based on accumulation of additive effects would be successful in improving these traits. Similar results were obtained by(Saleh, 2013 ; Dawwam et al., 2016 ; El-Mansy et al., 2020). Further, the correlation coefficient between the sums (L1i + L2i) and difference (L1i - L2i) were found to be negative and insignificant for SCY/P, LCY/P and 2.5% SL. However, the other traits were positive and non-significant, these results pointed out that the genes with positive and negative dominant alleles were dispersed between testers and didn’t show any proof of directional dominance for these traits. Similar results were obtained by (El-Lawendey et al., 2010) demonistrated non-significant correlation coefficient of sums and differences was found for all traits, revealing that dominant genes were umbidirectional among parents. On the other hand, significant positively additive correlation among lint cotton yield/plant and each of lint index and seed index were also detected.

Table 2. Mean square estimates for the studied traits in triple test cross (TTC)

*& ** significant at 0.05 and 0.01 levels of probability, respectively.

Table 3. Mean performance of the tested genotypes for the studied traits

Table 4. Disclosing the presence of epistasis mean square for the studied traits

*& ** significant at 0.05 and 0.01 levels of probability, respectively.

Table 5. Individual epistatic deviations of ten cotton lines for the studied traits

*& ** significant at 0.05 and 0.01 levels of probability, respectively.

Table 6. Mean square for sums and differences as well as estimates of additive, dominance, degree and direction of dominance for the studied traits

*& ** significant at 0.05 and 0.01 levels of probability, respectively.