Five inbred lines (in at least S6 generation) developed via single seed decent from the five locally grown faba bean cultivars in Sudan (Selaim, Hudeiba/93, Ed-Damar, Bassabier, and Shabah), representing a wide range of genetic variability in their agronomic traits (plant height, number of pods/plant, 100 seed weight, and yield), were used to fulfill the objectives of the present study. These lines were sown in isolated cages for multiplication (by hand tripping) and for cross purposes by hand in all possible combinations (in diallel cross excluding reciprocals) to produce F₁ seeds.

Three experiments were carried out at the Demonstration Farm of the Faculty of Agriculture, University of Khartoum, Shambat (latitude 15˚40'N; longitude 32˚32'E, and altitude 380 meter above sea level). The site is located in the semi-arid zone; the soil is alkaline (pH 8.0); rainfall is about 150 mm per annum and with maximum temperature of about 42˚C in summer and around 21˚C in winter [10] .

In experiment I, the seeds of the lines (parents) and F₁s were grown in seasons 2014/2015-2015/2016 to test the lines for general (GCA) and specific (SCA) coming ability.

In experiment II, the lines were evaluated in seasons 2014/2015-2015/2016 for the degree of cross-fertilization as described by Link [11] using the black hilum color as morphological marker. On average, from each entry 192 progeny were screened by year.

In experiment III, an equal number of seeds (25 seeds) from each of the five inbred lines [Hudeiba/93 (L₁), Ed-Damar (L₂), Shabah (L₃), Bassabeir (L₄) and Selaim (L₅)] were blended in all possible combination. Accordingly, sixteen blends/mixtures (M1 to M16) were developed. The line blends (one from all lines, five blends from four lines and ten blends from three lines) called Syn-0, were grown in season 2014/2015 under open pollination in spatial isolation to produce Syn-1seed. The seed of the lines blend (Syn-0) and that of Syn-1 were grown in season 2015/2016 and evaluated for yield and yield components.

Complete randomized block design with three replicates was used to execute the experiments. The gross plot size in the two experiments was 6.3 m² consisting of 3 ridges each 3 m in length and 70 cm apart. One seed was planted per hole with the spacing of 10 cm along the ridge. The experimental plots were irrigated every 14 days and weeded three times using hand hoe. The experiment was carried out in the presence of pollinators; here honeybee (Apis mellifera sp.) was introduced in the field of the experimental area.

Data were collected from the parameters of: 1) days to flowering, determined when 50% of the plants of each entry open the first flower, 2) Plant height measured in cm at the end of the flowering period, 3) Number of pods per plant, 4) 100-seed weight (g), and 5) seed yield t/ha. The data were subjected to analysis of variance (ANOVA) according to the method described for the randomized complete block design (RCBD). The computer program SAS-1997version 9.0 was used for the analysis of variance.

The heterotic effects of F₁ crosses were estimated as a percentage over mid parent [12] as follows:

Relative heterosis (RH) (%) = F 1 − Midparent Midparent × 100

Data for general combining ability (GCA) and specific combing ability (SCA) were analyzed as described by [13] for method 2 model I (all possible combination excluding reciprocal) as follows:

Y i j = μ + G C A i + G C A j + S C A i j + e i j k l

where:

Y_ij = Observation of ith parent in the jth block;

µ = population mean;

GCA_i = general combining ability (g.c.a) effect of the ith parent;

GCA_j = general combining ability (g.c.a) effect of the jth parent;

SCA_ij = specific combined effect (s.c.a) of two parents;

e_ijkl = experimental error.

The standard error required for testing the significance of general combining ability effects of the parents (p) and differences of GCA effects were obtained as:

S.E._ĝi = ( ( P − 1 ) p ( p + 2 ) σ e 2 ) 1 / 2

and S.E._{(ĝi - ĝj)} = ( 2 p + 2 σ e 2 ) 1 / 2

Three errors required for testing the significance of specific combining ability effect of the parents (p) and differences of SCA effects were estimated as:

S.E._ŝij = ( P 2 + P + 2 ( p + 1 ) ( p + 2 ) σ e 2 ) 1 / 2

S.E._{(ŝij - ŝik)} = ( 2 ( P + 1 ) ( p + 2 ) σ e 2 ) 1 / 2

and S.E._{(ŝij - ŝkl)} = ( 2 P p + 2 σ e 2 ) 1 / 2

Mean squares from the analysis of variance for the studied traits in the 15 entries (parental lines and their F₁s) revealed highly significant differences, except for number of days to 50% flowering which exhibited non-significant differences (data not shown). Comparing the crosses performance with their corresponding parental lines, none of the lines exceeded hybrid performance in any of the studied traits (Table 1). For the lines, plant height was in the range from 74.40 cm for Hudeiba/93 to 70.00 cm for Seliam with an average of 72.50 cm. The performance of the F₁s, plant height ranged from 88.70 cm (Hudeiba/93 × Shaba) to 76.00 cm (Bassabier × Selaim) with an average of 81.50 cm. Number of pods per plant in the lines was in the range from 18.37 registered by Seliam to 27.00 in Bassabier and averaged to 23.20. In F₁s, the cross of L2 × L4 gave the highest number of pod per plant and cross L3 × L5 the lowest one. Although the line Shabah had the highest 100-seedweight (71.40g) and yield (3.17 t/ha) compared the other lines, the cross of Ed-Damar × Bassabier gave the highest number of pods/plant and yield (5.60 t/ha).

Mean squares from the analysis of variance for GCA and SCA for the studied traits are presented in Table 2. The results indicated that the effect of general combining ability (GCA) for parental lines and specific combining ability (SCA) for crosses among the lines were highly significant (p ≤ 0.01) for all of the studied traits; the GCA was greater than SCA. Among the lines, Bassabier exhibited the highest positive GCA for number of pods per plant (1.29) and yield t/ha (0.08) (Table 3). Line Hudeiba/93 gave the highest positive GCA of 0.94 for plant height; Shabah line exhibited the highest (0.51) positive GCA for 100-seed weight, but the highest (−1.64) negative GCA for number of pods per plant. On the other hand, Selaim exhibited the highest (−0.19) negative GCA for yield t/ha.

For SCA, the magnitude of the crosses varied for the different traits e.g. the cross of Ed-Damar × Bassabier (L₂ × L₄) showed the largest (33.52 cm) positive SCA effect for plant height and number of pods per plant (26.42); the cross of Shabah × Selaim (L₃ × L₅) exhibited the largest (26.02) positive SCA effect for 100-seed weight (Table 4). Moreover, Hudeiba × Ed-Damar (L₁ × L₂), Ed-Damar × Shabah (L₂ × L₃) and Shabah × Bassabier (L₃ × L₄) showed the highest (2.00) positive SCA effect for yield t/ha.

** = significant difference at 0.01.

** = significant difference at 0.01.

** = significant difference at 0.01.

Values of heterosis percentage relative to mid parents for the studied traits are presented in Table 5. The mean values were ranged from 10.50 for 100-seed weight to 67% for yield t/ha. Heterosis in plant height was in the range from 6% to 21.2% with an average of 12%. The highest plant height heterosis of 21.2% was given by the cross L₁ × L₃ followed by the crosses of L₁ × L₂ (17%), L1 × L₄ (16.15%) and L₂ × L₄ (15%). Number of pods per plant exhibited heterosis values ranged from 50% (L₁ × L₄) to 89% (L₂ × L₄) with an average of 64%. Six crosses (L₁ × L₂, L₁ × L₃, L₁ × L₅, L₂ × L₃, L₂ × L₅ and L₃ × L₅) showed values of heterosis ranged from 61% to 69.23% and the crosses of L₄ × L₅ and L₃ × L₄ registered heterosis values of 54% and 55.5%, respectively. For100-seed weight, heterosis was in the range from 4.25% (L₃ × L₅) to 23.2% (L₁ × L₄) with an average of 10.5%. Heterosis for yield t/ha was in the range from 58% (L₄ × L₅) to 82% (L₂ × L₄) with an average of 67%. One cross (L₁ × L₅) gave a heterosis of 72%, the remainder of the crosses showed a heterosis values ranged from 60% to 68% (Table 5).

Table 6 shows the mean, range and mean sum of squares for the degree of cross-fertilization of the five lines. There was significant (p < 0.01) variation among the inbred lines for the degree of cross-fertilization. Amongst their progenies a total of 692 individuals were white and 266 were black. The degree of cross-fertilization was in the range from 28.1% (Shabah) to 46.9% (Ed-Damar) with an average of 36%.

**Significance at 0.01.

Mean squares from the analysis of variance (Table 7) showed highly significant differences (p ≤ 0.01) among the entries (lines blends in Syn-0 and Syn-1) for the studied traits. Generally, the increase in mean performance of lines blends for the studied traits in Syn-1 relative to Syn-0 was 24% for plant height, 13% for number of pods per plants and yield (t/ha); however, no significant increase in 100-seed weight of Syn-1 compared to Syn-0 was observed. For the performance of the lines blend, the tallest plant height was recorded by M8 in Syn-0 (76.6 cm) and M13 (92.9 cm) in Syn-1. In Syn-0, M3, M7 and M15 gave the highest number of pods per plant (25), whereas M2 and M13 registered the highest number of pods per plant (29) in Syn-1. M2 gave the highest yield of 3.4 t/ha for Syn-0 and 3.96 t/ha for Syn-1. On the other hand, the lowest yield of 2.13 t/ha and 2.39 t//ha was given by M11 in Syn-0and Syn-1, respectively (Table 6). The blend M2 was formed from four lines, namely Hudieba/93, Ed-Damar, Shabah, and Bassabier. The increase in performance of the four lines in Syn-1 compared to Syn-0 was 13% for plant height and number of pods per plant and 14% for yield t/ha. Compared to the average yield t/ha of the pure stand of the lines (3.07), the increase in yield of M2 was 11% in sy-0 and 29% in Syn-1. Compared with individual yield t/ha of the pure stand of the lines, 2.99 t/ha for Hudeiba/93 and Bassabier, 3.17 for Ed-Damar, 3.27 t/ha for Shabah and 2.96 t/ha for Selaim, (Table 1), the performance of Syn-0 in M2 surpassed the yield of pure stands of Hudeiba/93 and Bassabier by 14%, Ed-Damar by 7%, Shabah by 4%, and Selaim by 15%. For Syn-1, and as a result of increasing heterozygosity, the yield increase in M2 compared with the corresponding yield of the lines grown in pure stands was 32% of Hudeiba/93 and Bassabier, 25% of Ed-Damar yield, and 21% of Shabah yield and 34% of Selaim yield.

To improve yield potentials of a crop, in breeding program, it is important to create new combinations of genes to produce genotype with trait performance that is superior to current genotypes at the target environment [14] . In the present study, the highly significant variations among the parental lines and their possible F1-hybrids for the studied traits provide evidence for the presence

of a wide genetic variability among the lines and F1-hybrids produced from them; indicating that the genetic improvement is possible in the present genetic material. Moreover, the wide range in the mean performance for the traits: plant height, number of pods per plant, 100-seed weight and yield; among the entries accompanied with high heritability (broad-sense) of 0.71 - 0.95 confirm the presence of sufficient genetic variability and possibility of selection and improvement of yield and yield traits in the present faba bean germplasm. Similar results were reported by [15] [16] [17] .

In any breeding program, it is very important to know the combining abilities of the inbred lines that are used as parents in hybrids [9] [18] . Combining ability helps the breeder to identify the best combiners which may be hybridized either to exploit heterosis or to build up favorable fixable genes [19] [20] .

In the present investigation, the highly significant mean squares for both general combining ability (GCA) and specific combining ability may indicate the importance of both additive and dominance gene effect in the inheritance of the studied traits. Moreover, the higher mean squares from the analysis of variance for GCA of the traits than of SCA showed the great contribution of the additive effects of genes in the expressions of the traits. Therefore, selection will be effective for improvement of yield and yield traits in the present genetic material. Several researchers have reported on the significance of both general and specific combining ability effects on yield and yield traits in faba bean and found similar results, e.g. [20] [21] .

Form the results, the highest GCA given by Hudeiba/93 line for plant height, number of pods per plant and number of seeds per pod and the highest GCA exhibited by Shabah line for 100-seed weight as well as that showed by Bassabier line for pods per plant and yield (t/ha), indicate that these lines are the good combiner for the mentioned traits; therefore the improvement of such traits is rather predictive and more heterosis and high yield performance could be expected upon crossing. These findings are in agreement with those reported by [16] [21] [22] . Moreover, the positive SCA exhibited by the crossing L₂ × L₄ for plant height and pods per plant and that positive significant SCA effect given by the cross of Shabah × Selaim for 100-seed weight as well as the desirable positive SCA exhibited by all crosses for grain yield t/ha indicated that both additive and non-additive effects of the genes played a role in controlling faba bean yield as reported by [15] [17] [23] Therefore, as Hudeiba/93, Bassabier and Shabah lines exhibited the highest GCA and highest SCA upon crossing, they represent the presence of suitable genetic material for a breeder to build a future program to improve faba bean yield through selection and hybridization. The average amount of heterosis present in plant height (12.33%), number of pods per plant (64%), 100-seed weight (10.50%) and in yield t/ha (67%), indicate the presence of sufficient amount of heterosis, which could be used as a base for an effective breeding program for improvement of yield of the local faba bean gene in the semi-arid zone of Sudan. similar results were reported by other workers, e.g., [16] [17] [23] [24] [25] .

In the present study the high degree of cross-fertilization of 36% and high heterosis (67%) exhibited by the lines under the study as well as their highly significant positive GCA and SCA indicate the presence of the prerequisites for production of synthetic cultivars. For the number of parental lines, the high and significant performance for yield and yield traits obtained by mixture M2 (the mixture of Hudeiba/93, Bassabier, Ed-Damar and Shabah lines) in Syn-0 and Syn-1 compared to performance of each of the line, indicate that synthetic cultivars could be developed from these four lines. Moreover, the increase in overall yield performance from 3.40 t/ha in Syn-0 to 3.96 in Syn-1 indicates the effect of heterozyosity as well as the joint effect of heterogeneity and hetrozygosity in Syn-1, which is expected to increase and give high yield performance in Syn-2 and Syn-3 [3] as maximum heterosis is not achieved in Syn-1 but could be realized in later generations [8] . [8] pointed out that the maximum amount of heterosis will take until Syn-4 to realize nearly maximum amount of heterosis as synthetic exploited more additive gene action compared with hybrid which exploited non additive genes [9] [26] . Moreover, and for maximum synthetic yield, it is recommended to increase the degree of cross-fertilization before commercialization as reported by [8] and [27] .