Abstract

In order to suggest breeding strategies to improve Citrullus mucosospermus (Fursa), 4 × 4 complete diallel cross design involving Bebu, Wlêwlê small seeds 1 (Wss1), Wlêwlê small seeds 2 (Wss2) and Wlêwlê small seeds 3 (Wss3) genotypes was used to assess combining ability and gene actions involved in the inheritance of six seed traits. The F₁ direct and reciprocal crosses plus the parental inbred lines coming from these cultivars were grown in a randomized complete block design with three replications. The results indicated the existence of genetic variation between parental lines for all investigated seed traits. Combining ability analysis exhibited the involvement of both additive and non-additive types of gene actions in the expression of all studied traits, suggesting, doing the selection in C. mucosospermus heterogeneous populations for improving these seed traits. Non-additive gene actions were predominant in the inheritance of investigated traits indicating the possibility of the heterosis exploitation or the postponement of selection to later generations for improving genetically these traits. Bebu appeared the best general combiner for Mass of fresh seed, Mass of dry seed, Mass of 100 seeds, seed length and seed width while, Wss1 and Wss2 are the best combiners for percentage of seed integuments. Therefore, parental lines Bebu, Wss1, Wss2 and crosses with high significant specific combining ability effects are proposed for their incorporation in C. mucosospermus breeding programs. The presence of both GCA and SCA effects suggests the use of recurrent reciprocal selection to improve C. mucosospermus seed traits.

1. Introduction

Citrullus mucosospermus (Fursa), the so-called Egusi melon belongs to Cucurbitaceae family ¹ and is native to West Africa where it was domesticated ². In Côte d’Ivoire, it is the most oleaginous cultivated cucurbit species. C. mucosospermus is prized for its oleaginous seeds. Like other oleaginous cucurbits, its oilseeds are consumed as soup thickener ^{3, 4}. Oil extracted from C. mucosospermus seeds is comestible and is used in cosmetology and pharmaceutic industries ^{5, 6, 7}. Seeds are rich in nutrients such as carbohydrates (10.45 - 26.30%), proteins (21.78 - 30.42%) and lipids (41.78 - 56.08%) ⁸. In addition, seed oil were found to be rich in various unsaturated fatty acids, which possess potential health benefits ⁶.

In spite of its economic, cultural, therapeutic, nutritive importance, C. mucosospermus is a neglected and an underutilized crop because of a deficiency of research and development. Consequently, there is lack of information about genetics and breeding of this important crop in many African countries, particularly in Côte d’Ivoire ^{9, 10}. For instance, yields of different cultivars despite their good adaptation to extremely divergent agro-ecosystems are low ³ and vary from 0.10 to 0.31 t ha^{-1 11}. C. mucosospermus productivity must be improved genetically in consideration of its numerous qualities. The improvement of the production and the uses of this important crop can result in food security and income generation for peasants ¹².

The choice of the most efficient breeding program for genetic improvement of one trait depends on the knowledge of the type of gene actions involved in its inheritance and the genetic control of related trait. One of the biometrical techniques available to plant breeders to obtain information’s on the type of gene actions involved in expression of trait in the crop species is diallel analysis ¹³, which reportedly provided early information on the genetic behavior of a group of parents and their hybrid combinations in the first generation ¹⁴. Diallel analysis represents the best strategy for determining the general (GCA) and specific (SCA) combining abilities which help to identify the best parents and hybrid combinations and, provide sufficient genetic data about character inheritances ^{15, 16, 17}. Indeed, the significance of the GCA reflects additive gene actions while, the significance of SCA reflects the non-additive genetic effects which indicates relevant non-allelic interactions ^{16, 18, 19}. The non-significant of both GCA and SCA values indicates epistasis gene effects play a remarkable role in determining these characters ^{20, 21}. If several studies on GCA and SCA in other Cucurbitaceae species have been realized ^{22, 23, 24, 25, 26, 27}, it is not the case for C. mucosospermus.

This study aims to assess the genetic control of some important seed traits and to study combining abilities (GCA and SCA effects) of four parental lines and their F₁ hybrids in order to suggest breeding strategies for C. mucosospermus.

2. Materials and Methods

Combining ability and gene action analysis were done using four Citrullus mucosospermus (Fursa) accessions collected from four regions of Côte d’Ivoire (Table 1). The genotypes Bebu (B), Wlêwlê small seeds 1 (Wss1), Wlêwlê small seeds 2 (Wss2) and Wlêwlê small seeds 3 (Wss3) were chosen basing on a wide range of morphological diversity in fruit and seed characteristics that prominently distinguish one line from another.

Each of four accessions (B and Wss1, Wss2, Wss3) was purified after four generations of self-pollination to obtain inbred lines, which were planted during the period from January to April 2016 for crossing. At flowering, these four parental inbred lines were crossed in complete diallel cross design to produce 12 F₁ hybrid descents.

Parents and their 12 F₁’s hybrid descents were evaluated at field from May to August 2016 in a randomized complete block design with three replications at Kononfla city located at altitude 243 m above sea level between latitude 6° 37′ 18″ North - longitudes 5° 54′ 37″ West. The mean annual temperature fluctuates from 25 to 30 °C and average of annual rainfall varying between 1500 to 2000 mm. The experimental field area was 2067 m² (53 m × 39 m) and divided into three blocks with 16 plots per block. The area of one plot was 32 m² (8 m × 4 m) and the distance between two plots is 1 m. Only one F₁ family or parental line was sowed by hand on 3 rows within a plot with plant spacing of 2 m × 2 m. Two weeks after sowing, the most vigorous seedlings were selected per hole by separating. The experimental field was regularly weeded during the vegetative growth stage.

Ten plants were selected randomly in each plot giving 30 plants per family for data collection. Six traits (Table 2) related to seeds were collected from three fruits of each randomly selected plant.

Mean values of the parents and their F₁ hybrids were calculated for all studied traits. Analysis of variance (ANOVA) and the post ANOVA test (Turkey test) were performed at 0.05 level of probability to test differences among treatment means. All statistic tests were performed using R Software version 3.3.1 (R Development Core Team 2011).

The analysis of variance of GCA, SCA and reciprocal effects and their respective effect estimates were carried out according to ²⁸ model 1 method 1. The statistical model for the mean value of a cross (i × j) in Griffing’s analysis is:

(1)

Where Y_ij is the mean of (i×j)^th cross over replications k (k = 1, 2, . . ., b); m is the general population mean; g_i and g_j is the GCA effects of i^th and j^th parent respectively; s_ij is the SCA effect of the cross between the i^th and the j^th parents; e_ijkl is the environmental effect that is associated with i_jkl observations; b is the number of replication; and c is the number of samples per replication.

The significance of the estimates of variance due to GCA and SCA was tested using F-values at threshold probabilities of 1% and 5%, while significance of estimates of GCA and SCA was tested using their respective standard error according to ²⁸.

The genetic components were estimated according to ²⁹:

(2)

(3)

Where σ²_GCA is the component due to GCA; σ²_SCA is the component due to SCA; MS_GCA is the variance due to GCA, MS_SCA is the variance due to SCA, MS_error is the mean error, and n is the number of replications.

The ratio σ²_GCA/σ²_SCA was estimated to assess gene actions such as additive and dominance effects. More σ²_GCA/σ²_SCA is raised, more the additive effects of the genes are more important.

3. Results and Discussion

Table 3 presents the mean values of seed characters of C. mucosospermus parental lines and their F₁ hybrids from diallel cross.

The results showed that Bebu (B) seed characters differed significantly from these of Wlêwlê small seeds (Wss) cultivars for all studied traits. Indeed, B stands out from accessions of Wss cultivars by large seeds (long and wide seeds), by higher mass of fresh seeds (MfS), mass of dry seeds (MdS), mass of 100 seeds (MS₁₀₀) and percentage of seed integuments (PSi) (Figure 1). ¹⁰ reported similar results about mean values of seed traits in B and Wss accessions. The results also revealed that Wss accessions (Wss1, Wss2 and Wss3) were distinct one from other, indicating existence of variability among them. Indeed, Wss3 stands out from two other accessions of Wss according to higher MfS, PSi and wider seeds. Accessions of Wss1 and Wss2 are different from each according to MS₁₀₀, Seed length (SL) and Seed width (SW).

The comparison of agronomic performance of F₁ hybrids revealed that hybrids B × Wss3 and Wss3 × B recorded the highest values for trait MfS, MdS, MS₁₀₀ and PSi expressions. Progenies B × Wss1, Wss1 × B, B × Wss2 and Wss2 × B exhibited high values for SL and SW expressions. ²⁶ suggested that progenies with high agronomic performance value for some traits could be selected as candidate genotypes in breeding programs aiming these traits enhancement. Elsewhere, some F₁ hybrids presented mean values equal or superior to those of their best parents for some traits suggested complete or super dominance. This is the case of hybrids from direct and reciprocal crosses B × Wss1, B × Wss2, B × Wss3 and Wss2 × Wss3 for traits MfS and MdS, of hybrids from direct and reciprocal crosses Wss1 × Wss3 and Wss2 × Wss3 for MS₁₀₀, and of hybrids from crosses that involved Wss3 as one of the parents for PSi. The complete or super dominance observed in these crosses for these characters indicates the heterosis effect. Even if the expression of heterosis in one crop implies the potential of producing superior cultivars through selection in segregating populations or heterosis breeding ^{30, 31}, it is not the case for PSi. Indeed, high mean values for PSi, which express the impediment to decorticate seeds, were not desirable. On the other hand, the lower values express easiness to decorticate seeds were desirable. Thus, crosses with low mean values for PSi could be selected as candidate genotypes for the reduction of seed integuments in breeding programs of C. mucosospermus.

The mean squares from analyses of variances for general (GCA) and specific (SCA) combining abilities as well as the ratio σ²_GCA/ σ²_SCA are presented in Table 4.

The data showed that mean squares due to general and specific combining abilities were highly significant (P < 0.001) for all evaluated traits suggesting the presence of both additive and non-additive genetic variance in the expression of these traits. The presence of both additive and non-additive gene actions in the inheritance of these traits indicates that selection may be used in C. mucosospermus heterogeneous populations for improving these traits ³². In concrete terms, the use of reciprocal recurrent selection and/or bi-parental mating suggested for the improvement efficiently of the traits ^{33, 34}. For all studied traits, the magnitude of SCA variance was higher than GCA variance. The ratio σ²_GCA/σ²_SCA ranged from 0.01 to 0.05, which means non-additive effects played a larger role than additive effects in the expression of these traits. The preponderance of non-additive gene actions in the inheritance of these traits suggests the possibility of the heterosis exploitation ^{19, 35}, or the postponement of selection to later generations for improving genetically these traits ^{36, 37}.

Results about GCA effects reveal that B has recorded the highest significant positive values of GCA for all studied traits while Wss1 and Wss2 obtained significant negative values of GCA for PSi. In addition, Wss3 has significant positive values of GCA for PSi (Table 5). Positive values of GCA for PSi expressing the difficulty to decorticate seeds were not desirable and the negative values expressing the easiness to decorticate were desirable. Thus, B was the best general combiner for all studied traits except for PSi that the best general combiners were Wss1 and Wss2. These three inbred lines (B, Wss1 and Wss2) could be used in C. mucosospermus breeding programs. In fact, genotypes with significant GCA effects for a trait indicate that these genotypes contain more genes with additive effects ^{23, 38}, suggesting this trait is heritable and these genotypes have high potential for generating superior offspring ²⁷. Consequently, these genotypes could be used as a good parent for this trait in a breeding program designed to improve that trait ^{24, 29, 38}. High significant values of GCA for one trait reveals that selection and hybridization methods would result in interesting genetic improvement for this trait thanks to desirable genes accumulation of two parents in the targeted genotype ^{17, 26, 30}. Thus, parental lines B (for MfS, MdS, MS₁₀₀, SL and SW) and Wss1 and Wss2 (for PSi) can be used in recombination breeding programs to accumulate their favorable genes responsible for increasing seed yield in promising pure lines.

The estimates of SCA revealed that B × Wss2, B × Wss3 and Wss2 × Wss3 crosses had the highest positives significant values of SCA for MfS and MdS. For MS₁₀₀, higher significant positive estimated values of SCA were recorded in Wss1 × Wss2, Wss1 × Wss3 and Wss2 × Wss3 progenies. The best combinations for PSi were B × Wss1, B × Wss2 and Wss1 × Wss2 crosses since they presented negative significant values of SCA. Crosses B × Wss2 and Wss1 × Wss3 showed the highest positive significant estimates of SCA for SL while Wss2 × Wss3, B × Wss1 and Wss1 × Wss3 crosses exhibited the positive significant estimates of SCA for SW (Table 5). Thus, crosses with significant SCA effects involved parents with various types of general combining ability effects (high x low and low x low). Combinations exhibiting significant positive SCA effects for a trait have the potentials of producing good segregants for selections for these attributes in the higher generations ²⁷. ¹⁹ reported that the most favorable combination is the cross having the greatest SCA, in which at least one parent has greater GCA and is divergent in relation to the other parent of the crossing. ³⁹ emphasized that the diversity in parental GCA-effects plays an important role for the production of F₁ hybrids with significant positive SCA effects. In fact, great positive significant SCA effects for one trait, obtained in crosses which were involving one progenitor with high GCA effect (high x low), suggested the involvement of additive x dominance gene interactions in this trait expression ³⁸. When one parent with high GCA crosses another with low GCA gives high SCA effect, this cross could be advanced further for the isolation of transgressive segregants in order to develop good inbred lines ²⁵. Indeed, these promising crosses may be improved through conventional breeding methods such as bi-parental mating and diallel selective mating, thereafter followed by pedigree method of selection, so as the tight linkage, if any, may be broken and transgressive segregants may be isolated ¹⁵. High significant SCA values resulting from crosses, which involved both parents with low GCA effects, were also observed. This fact is the result of the presence of epistasis (non-allelic interaction) at heterozygous loci, which is not fixed, so it is suggested utilizing these crosses through single plant selection in the later generations ³⁸.

4. Conclusion

In the present study, parental lines shown genetic variation for all examined traits. Both additive and dominance components of variation were highly significant for all studied traits indicating that selection may be used in C. mucosospermus heterogeneous populations for improving these traits. Non-additive gene actions play an important role in the inheritance of all studied traits, which suggests the possibility of the heterosis exploitation or the postponement of selection to later generations for improving genetically these traits. Among parental lines, B is the best general combiner for MfS, MdS, MS₁₀₀, SL and SW while, Wss1 and Wss2 are the best combiners for PSi. For each studied trait, at least two crosses with high significant SCA effects were observed. It is suggested that C. mucosospermus breeders can improve the potential of seed crop by selecting very promising lines in respect of seed traits. Therefore, we suggested parental lines B, Wss1, Wss2 and crosses with desirable significant SCA effects for their including in C. mucosospermus breeding programs. Cause of the presence of both GCA and SCA effects, recurrent reciprocal selection can be used to improve C. mucosospermus.

Acknowledgements

The authors express their gratitude and their thanks to IFS (International Foundation for Science) for its great contribution in the realization of this research works through the project funding.

References