One effort in order to increase productivity as well as quality of tomato is through application of different plant breeding methods. The improvement of its characters with high economic values often faces challenge when selecting parents with high combining ability. Therefore, the effective study for parent selection is highly needed. The observation to the performance of hybrid offspring can be conducted using diallel crossing method. This progeny test can be related to the Combining Ability which are very useful in determining the parent combination for the best progeny with potentially high productivity and other selected novel characters (Syukur et al., 2012).  Some information can be obtained from diallel analysis e.g. general combining ability (GCA) and specific combining ability (SCA) from crossing parental lines. GCA is the performance of line as combination of solely crossing with other lines, whereas SCA is the performance of a hybrid line resulted from the cross with other line (Singh and Chaudary, 1979). Combining ability is a measurement of plant genotype ability in crossing to produce superior plants. Combining ability which  is   obtained   from   a   cross  between two parental lines can provide information regarding cross combinations for better heredity (Sudesh and Anita, 2016). The analysis of diallel crossing is needed to predict the additive and dominant effects from a certain population that can be used further to predict the genetic variability and heritability. This analysis is often used for many kind of plant, such as tomato (Farzane, et al. 2013; Saleem et al., 2013; Saputra et al., 2014). The objective of this research was to evaluate the effect of GCA and SCA value on tested tomato.

The experiment was conducted at Shambu campus research site (crossing was done in green house) and hybrid studies at two environments, Hareto and Shambu, western Ethiopia. The area is characterized by mono-modal rainy season (March to September) with a mean annual rainfall of 1700-2000 mm, and an altitude of 1700 - 3000 m.a.s.l. The experiment consisted of 36 materials of tomato, that is, eight parents (tomato varieties obtained through selection and released by Melkasa Agricultural Research Center) (Table 1), and 28 F1 crosses between the parents produced in half diallel cross fashion. The varieties  have   been   selected   based   on   their national performance.

Parental materials were planted in staggered (seven days interval) to synchronize days to flowering. Crossing was done by hand in half-diallel fashion following Fasahat (2016) model I method 2. Emasculation was effected by carefully removing anther by hand without damaging the pistil before crossing and emasculated heads was enclosed in paper bag to protect undesired crossing. Two to three days later, paper bag enclosing on the emasculated heads was opened and the male flowers was gently shacked over the stigma to effect pollination.

Paper bags enclosed again after pollination and kept until fruit setting. The presence of some heritable morphological markers like fruit shape and colors in F1 and respective parents were used to indicate that crossing was done successfully between parents.

The experiment was laid out in simple lattice design at both locations. The spacing between two plots in each replication and between adjacent blocks will be 50 and 100 cm, respectively. With two plant rows of ten individual plant per row, 40 and 100 cm was intra and inter row spacing used at F1 evaluations on 1.4 m × 3.6 m (5.04 m²) area of bed.

Data was recorded on eleven quantitative characters viz., days to 50% flowering, plant height (cm), number of primary branches per plant, number of fruits per plant, average fruit weight (g), number of cluster per plant, and total marketable fruit yield per plant (kg). In addition, some physical parameters  like  fruit  dieter,  fruit  length  and per carp thickness was also measured using Culver caliper and quality traits like fruit density, TSS and pH were also taken.

Statistical analysis

The data was subjected to ANOVA following the standard procedures given by Steel and Torrie (1980). Analysis of variance (ANOVA) of each character was carried out using (SAS, 2008) computer software (version 9.3). Statistically significant difference among the crosses for the character being considered justifies further statistical analysis for that character.

Diallel analysis was carried out according to Fasahat (2016) Method II, Model I (Fixed Model), which involves parents and one-way F1 hybrids (excluding reciprocals).

The analyses of variances for genotypes (parents and F1 crosses) and pooled over environments (Table 2) revealed that mean squares due to genotypes were significant (P<0.05) for all the traits studied in all the environments indicating the presence of inherent variation among the materials. Mean squares due to locations were also found significant (P<0.05) for all the traits studied except fruit diameter indicating the presence of environmental variation among the two studied sites. Mean squares due to genotypes × environment interaction were significant (P<0.05) for all traits except days to 50% flowering, fruit length and fruit diameter indicating genotype by environmental interaction was important. Therefore, farther studies using different additional locations and growing season  is  important  for  selecting important genotypes for both environments and/or specific location.

Performance of genotypes

The mean values of 8 parents and their 28 F1s for 13 yield and yield related traits showed significant differences (Figure 1). In this study, both the crosses and the parents showed high variation in their mean performances for most of the characters. Significant differences among genotypes for all the characters in tomato crosses and parents were also reported by Farzane et al. (2013) and Kumar et al. (2013). Presence of significant differences among genotypes for all the characters, allowed combining ability analysis (Singh and Chaudhary, 1979).

In total, TSS parents range from 3.28 (P3) to 4.38 (P8) and the crosses range from 2.75 crosses (P1×P8) to 4.88 (P5×P6) (Figure 1). This indicates the presence of variation in the materials. The present result is in close agreement with that of Thakur et al. (1996) who present some wild tomato accessions attaining very high (11-15%) concentrations of soluble solids, common processing tomato cultivars exhibit moderate soluble solids contents ranging between 4.5 and 6.25%. Seasonal variation as well as horticultural practices may affect tomato soluble solids content commonly expressed in degrees Brix. Larger degree Brix values are  frequently correlated with greater tomato product yield.

Estimation of general combining ability (GCA) effects  

Based on the results from analysis of variance for combining ability, the GCA effects of parents on yield and component characters were estimated (Table 3). Nature and magnitude of combining ability effects provide guideline in identifying the better parents and their utilization. The GCA effects of the parents (Table 3) revealed that none of the parent found to be good general combiner for the character number of primary branch per plant.

The negative combining ability effect indicated the genotypes or cross combiner contributed to decreasing performance in certain characters while the positive combining effect indicated the genotypes or cross combiner contributed to increasing performance in certain characters. GCA and SCA positive effects are used during genotype selection with high yield. In contrast, GCA and SCA negative effect are used during genotype selection towards pathogen resistance (Yustiana et al., 2013).

General combining ability has direct relationship with narrow sense heritability and represents fixable portion (additive and  additive  ×  additive  interaction)  of  genetic variation, thus helps in selection of parents suitable for hybridization (Geleta and Labuschagne, 2006; Saleem et al., 2009)  to develop cultivars with desired traits of interest. In present studies, P8, P1 and P2 were rated as best general combiner and can be used as donors for quality traits through multiple crossing programmes. To decreasing performance in certain characters while the positive combining effect indicated the genotypes or cross combiner contributed to increasing performance in certain characters.

An overall appraisal of GCA effects revealed that among parents P8 was found to be the best parent as it gave good general combining ability consistently in all the environments for number of marketable fruit per plant followed by parent P1. P2 was also found to be good combiner for other traits studied viz., pericarp thickness, marketable yield per plant, and individual fruit weight. The second most desirable parent was observed to be P7 which revealed significant desirable GCA effects for number of fruit cluster per plant, number of marketable fruit per plant, and fruit length.  P6 was also a promising parent for inclusion in breeding programme as it revealed good general combing ability for marketable yield per plant, producing earlier flowering time and producing intermediate plant height with controllable branches.

Estimates of specific combining ability (SCA) effects 

Specific combining ability is the manifestation of non-additive component of genetic variance and associated with interaction effects, which may be due to dominance and epistatic component of genetic variation that are non-fixable in nature. The estimates of specific combining ability of 28 crosses for 14 characters were presented in Table 4. It is observed that a total of 16 crosses exhibited positive and significant SCA for yield per plant. The cross combinations showing high negative SCA effect for days to flowering (earliness) were out of the 28 crosses studied, nearly 50% of the crosses, P4 × P1 (poor × good), P1 × P5 (good × average), P1× P7 (good × good), P2 × P7 (poor × poor) and P5 × P7 (poor × good), P3 × P8 (average × good), P6 × P8 (poor × good), P2xP4 (good × average) and P4 × P5 (good × average) in pooled environment expressed significant negative SCA effects indicating their good specific combining ability (Table 4).

The promising combinations for number of marketable fruit per plant were P5 × P7 followed by P3 × P6 and P3 × P7. It is observed that majority of the crosses with high SCA for number of marketable fruit per plant were involved with high/low or average/low combining parents. But very few crosses showing low/low general combiners showed high SCA. In contrast, showing high negative SCA for number of marketable fruit per plant were P1×P3, P4×P6, P6×P7, P2×P8 and P5×P8. For plant height, estimates of SCA are desirable and the good specific combiners were P1×P3, P1×P5, P1×P6, P4×P2, P2×P8, P4×P8 and P7×P8 which recorded negative SCA. In distinction to these results, crosses combinations P×P2, P1×P3, P3×P4, P4×P5, P5×P6, P1×P8 and P6×P8 revealed high positive and significant SCA in plant height indicating height increment that may be desirable in that as height increase, thus number of primary branches and number of fruit cluster per plant also increase. This has positive correlation for fruit yield though it needs other practices in requiring additional cost for stacking.

The best specific combiners for fruit cluster per plant were P1×P2, P2×P4, P3×P5, P3×P7, P5×P6, and P3×P6 which recorded positive and significant SCA effect. While, cross combinations like P1×P3, P4×P6, P2×P8, P5×P8, and P7×P8 showed significant and undesirable SCA effect on fruit cluster per plant.

Economic yield (fruit yield) is the most complex character governed by polygenic gene. Among the 28 crosses, 13 crosses showed positive and desirable SCA for fruit yield per plant. The cross combinations viz., P1×P2, P2×P3, P2×P4, P2×P6, P3×P6, P3×P7, P4×P5, P5×P7, P1×P8, P3×P8, P5×P8 and P6×P8 showed significant and positive SCA for weight of marketable fruit per plant. While, crosses combinations viz., P1×P4, P1×P5, P1×P6, P2×P5, P2×P7, P3×P5, P4×P6, P7×P7, P6×P7, P2×P8, P4×P8 and P7×P8 recorded significant and negative SCA effect on fruit yield per plant. This result is in agreement with the finding of Shankar et al. (2013).

Three crosses viz., (P1×P8, P3×P8 and P3×P6) showed high SCA effects for most of the yield components. Among these crosses, one of the parents had significant negative GCA effects, revealing that non-additive gene effect played predominant role in their expression and is worthwhile for exploitation of heterosis.

In this investigation, when majority of the characters are considered at a time for improvement of yield components, non-additive gene effect is more predominant than additive gene effects. Thus, heterotic breeding is the most practicable approach in improvement of yield. These results are in agreement with Farzane et al. (2013), Saleem et al. (2013) and Yustiana et al. (2013).

The top selected crosses combination involved both parents with positive × positive GCA effect indicating involvement of more additive gene effects in their heterotic performance. Thus, they may be further improved upon through conventional selection methods like pedigree or recurrent selection. The other crosses which showed significant positive SCA effects were P3×P7 and 'P6×P8. In these crosses the positive SCA effects were mainly due to positive × positive combiners.

In both, the crosses involving positive × positive combination genetic interaction might be of additive × additive type. The category of positive × positive GCA effects played an important role in the expression of favorable and significant SCA effects. Thus, choice of parents based on combining ability is sound proposition (Sharma et al., 2009). From this investigation we can propound that, information regarding general combining ability (GCA) effects of the parents is of prime importance, as it helps in successful prediction of genetic potentiality of crosses, which yield desirable individuals in segregating populations of self-pollinated crops. In general, specific combining ability is associated with interaction effects, which may be due to dominance and epistatic components of variation that are non-fixable in nature. Hence, it can be utilized in development of hybrid varieties.

Fruit length, fruit diameter, fruit density, percentage of TSS, power of hydrogen (pH) and pericarp thickness are some of the quality attributes for fruit crops including tomato. Pericarp is one of the important qualities of fruit crop as it increases the longevity of the crop especially during transportation. Some crosses viz., (P1×P8, P6×P8, P1×P2, P2×P5, P2×P6, P4×P7, P4×P5 and P3×P6) showed high SCA effects for pericarp thickness. Among these crosses one of the parents had significant negative GCA effects, (P3 and P6) revealing that non-additive gene effect and the other parents (P1, P2 and P8) had significant negative GCA effects, played predominant role in their expression and is worthwhile for exploitation of heterosis. This result is getting support from the finding of Hannan et al. (2007), Saleem et al. (2009) and Singh et al. (2010).

Both additive and dominant gene action types play an important role in controlling yield and yield component in tomato, but additive gene action was  more  prominent  in controlling yield per plant, number of fruit, individual fruit weight, fruit length, fruit diameter and fruit thickness. Tomato genotype P8 proved to be the best general combiner for yield and yield components. The presence of both additive and non-additive variances suggested the utilization of certain genotypes and crosses for future breeding work. Therefore, use of diallel mating with recurrent selection and integration with pedigree selection can be suggested as breeding program to exploit both additive and non-additive gene effect for the genetic improvement of the characters of tomato. Thus, parental genotypes having good GCA like P1, P2, P3 and P6 and specific crosses showing high SCA like P1×P2, P2×P3, P2×P4, P2×P6, P3×P6, P3×P7, P4×P5, P5×P7, P1×P8, P3×P8, P5×P8 and P6×P8 should be included in multiple crosses for tangible improvement of marketable fruit yield in tomato.

The authors have not declared any conflict of interests.