Hot pepper (Capsicum annuum L.), of the family, Solanaceae (2n = 24), is an important spice and vegetable crop (Gogula, 2015); it covers 67.98% of all the area under vegetables produced in Ethiopia (CSA, 2011/2012). The country produces paprika and Capsicum oleoresins for export market. Because of its wide use in Ethiopian diet, the hot pepper is an important traditional crop mainly valued for its pungency and color. The crop serves as source of income particularly for smallholder producers and also contributes significantly to house hold food security in many parts of rural Ethiopia (Shiferaw and Alemayehu, 2014). Combining ability of genotype is the ultimate factor determining future usefulness of the lines for hybrid development (Hallauer and Miranda, 1988). At the same time, it also elucidates the nature of gene actions involved in the inheritance of characters. General combining ability (gca) is attributed to additive gene effects and additive x additive epistasis and is theoretically fixable. On the other hand, specific combining  ability  (sca)  attributable to non-additive gene action may be due to dominance or epistasis or both and is non-fixable. The presence of non-additive genetic variance is the primary justification for initiating the hybrid programme (Pradhan et al., 2006; Acquaah, 2007; Skoric et al., 2000).

The mean sum of squares due to gca and sca and their variance ratio are indicator of the nature of gene action (Patil, 2012). In other words, relatively larger gca/sca variance ratio demonstrates importance of additive genetic effects and a lower ratio indicates predominance of dominance and/or epistatic gene effects (Fasahat et al., 2016). High sca effects resulting from crosses where both parents are good general combiners (good gca × good gca) may be ascribed to additive × additive gene action (Dey et al., 2014; Kumar et al., 2015). The high sca effects derived from crosses including good × poor general combiner parents (Kaya and Atakisi, 2004; Dey et al., 2014) may be attributed to favorable additive effects of the good general combiner parent and epistatic effects of poor general combiner, which fulfills the favorable plant attribute. High SCA effects manifested by low × low crosses (May et al., 2010) may be due to dominance × dominance type of non-allelic gene interaction produced over dominance, thus being non-fixable (Skoric et al., 2000; Vasal et al., 1986). The diallel mating design is the most important for gca and sca. In terms of amount of information, it is more informative than North Carollina designs (Acquaah, 2007).

Despite the long history of hot pepper production and its importance in Ethiopia, the research work done so far in variety improvement was almost nil. Although, such genetic studies have been made in various crops, including pepper, in various parts of the world, little or no effort has been made on pepper under Ethiopian condition to exploit the existing potential. In this study, therefore, an attempt was made to generate information on six introduced and four released cultivars of pepper crossed in half-diallel fashion with the following objective:

1. To identify promising  parents  with  better  gca  effects and single cross hybrids of pepper with promising sca effects.

2. To investigate the type of gene actions involved in traits for hybridization.

Description of the study areas

The field experiment of F₁ crosses with the parental materials was conducted at six different environments namely Wolaita Soddo, Alaba and Humbo that represent major pepper growing areas in the Southern Ethiopia for two cropping seasons in 2015 to 2016.

Experimental materials

The experimental materials consist of 10 parents (six introduced genotypes from Asian Vegetable Research and Development Center (AVRDC) located in Taiwan and four Ethiopian released varieties). The local parental materials were collected from Melkasa Agricultural Research Center (Table 1).

General (g_i) and specific combining ability (S_ij) effects

The following effects were obtained:

(i) General combining ability (GCA) effect of the i^th parent

g_i = (X_i· + X_ij – 2 X_··/P) P+2

(ii) Specific combining ability (SCA) effect of ij^th cross.

S_ij = [X_ij – (X_i· + X_ij + X_j· + X_jj) P+2 + 2X_··/(P+1) (P+2)]

ANOVA of general combining ability (gca) and specific combining ability (sca)

Combining ability analysis revealed highly significant (P ≤ 0.01) effects of gca and sca for all traits considered. Gca x e mean squares were significant for all traits but the magnitudes were consistently smaller than the respective gca mean squares. The sca x e effects were significant for all traits except fresh fruit yield and fruit dry weight. Similarly, the magnitudes of sca x e mean squares were smaller than that of sca for all  traits except fruit wall thickness (Table 2). The result indicated that the magnitude of mean squares for gca was higher than the mean squares of sca, suggesting that additive gene effects were more important than non-additive gene effects for these traits. This is in agreement with Nsabiyera et al. (2013) and Geleta and Labuschagne (2006) who indicated that gca effects were higher than sca effects on seven pepper parents.

General combining ability study (gca)

The estimates of gca effects of the parents presented in Table 3 indicated significant desirable and undesirable effects depending on the trait under consideration. The analysis revealed that among the introduced genotypes, except parents 6 (AVPP0206) and 10 (AVPP59328), the rest exhibited significant and positive GCA effects in most of the cases including fresh fruit yield and dry weight. Whereas, genotypes AVPP0206 and AVPP59328 showed undesirable  gca in  almost all growth and yield traits. The Ethiopian genotype ‘Marekofana’ showed significant positive gca effects in the desired direction for fruit diameter, seed number per fruit, fruit thickness and fresh fruit yield, whereas other Ethiopian genotypes had undesirable gca effects in most of the traits considered (Table 3). Similarly, Fekadu et al. (2009) found that 'Marakofana' was the best general combiner for yield and quality traits among 12 Asian and Ethiopian genotypes studied. Geleta and Labuschagne (2006) also noticed more or less similar result.

The gca effects for fruit length were highly significantly positive (9.24) for AVPP0514 followed by 4.12 for AVPP0512. In contrary, the lowest gca for fruit length (-8.96) was recorded by AVPP59328. Similarly, fruit diameter has both positive and negative gca values with the highest highly significant gca (2.17) recorded for parent Marekofana, whereas the lowest negative value (-1.28) was recorded for parent Melkashote. Fruit number per plant had highly significant positive gca  (5.07)  and  negative  gca  (-4.33)  values  for parents Melkashote and AVPP0105, respectively. Fresh fruit (pod) yield had positive gca value (1843.87) recorded by parent AVPP0514 followed by (930.97) parent Mareko fana and that of highly significant negative gca (-1344.28) followed (-1322.61) by parents Melka shote and AVPP59328, respectively. In most cases, the gca for fruit dry weight was non-significant except that recorded by parent AVPP0514 (202. 41) (Table 3). Consistent result was reported by Adarsh and Kumari (2015) in that out of 10 parental chilies evaluated, four were found to be good general combiners for fruit yield and other related traits. Nsabiyera et al. (2013) reported the existence of gca and sca effects on hot pepper traits. In support of the current result, additive and non-additive gene actions were reported for most agro-morphological and quality traits on Capsicums (Adarsh and Kumari, 2015). Parents with significant negative gca effects are considered desirable and should be selected in a breeding program for traits that need reduced expression  in   the   progeny   including   days   to flowering and fruit maturity, pedicel length and disease incidence. In such cases, parents 6 (AVPP0206) and 10 (AVPP59328) could be selected. In contrast, parents with significant positive gca effects were considered desirable for traits that require increased expression in the progeny. Thus, among released Ethiopian pepper varieties, Melka awaze and Melkazala were selected as parents for plant height, canopy width and primary branch number per plant. Marekofana variety is best combiner for fruit diameter, seed number per fruit, pericarp thickness and fresh fruit yield. Thus, the result indicated that there were introduced and local pepper genotypes which were considered to be the best general combiners for growth, yield and quality.

Estimates of specific combining ability effects of crosses

The current result revealed that highly significant sca  values  were  recorded   both   positively  and negatively by hybrids in traits measured. Out of 45 crosses, only 10 for fresh and 14 for dry weight showed significant sca effects, and among these, only 5 for fresh fruit yield and 7 for dry weight showed desirable (positive) sca effects. The cross combinations of Melka awaze x AVPP0206, Marakofana x AVPP0514, Melkazala x AVPP0105, AVPP9813 x AVPP0105 and AVPP0514 x AVPP59328 were the top desirable crosses for fresh pod yield and dry weight as well (Table 4). The result also indicated that the highest desirable sca value (13410.28) for fruit yield was recorded by crosses between introduced genotypes (AVPP0514 x AVPP59328) followed by sca value (6043.56) obtained by cross made by Ethiopian and introduced genotypes (Marakofana x AVPP0514). This indicates that there are potentials for the production of hybrids from Ethiopian local hot pepper genotypes. Besides the above listed crosses, sca value for dry weight was highly significant for cross Melkashote x AVPP9813. The best three top crosses for  dry  weight  include  AVPP0514  x AVPP59328 (2679.56), Marakofana x AVPP0514 (866.80) and Melkashote x AVPP9813 (840.97) (Table 4).

Furthermore, the result revealed that the parent genotype AVPP0514 has positive significant gca, whereas AVPP59328 has highly significant negative gca effects for yield and other related traits. Sca effect is highly significant in desirable direction. This indicates the strong contribution of additive gene effect of AVPP0514 for the cross and hence should be selected as best combiner for hybrid production in improving yield in pepper production. Statistically significant positive or negative sca effects for fruit yield and other traits showed that the crosses performed better or poorer than what would be expected from the gca effects of their respective parents (Table 4). Parents with various gca effects are combined to produce crosses varying in sca effects. This result is consistent with the findings reported by some researchers (Lohithaswa et al., 2000; Marchesan et al., 2009).

In the case of fruit quality traits such as fruit length and width, some of the hybrids with highly significant and positive SCA effects include  Melka awaze x AVPP0206, Marakofana x AVPP0206, Marakofana x AVPP0514, Melkashote x AVPP0105, Melka shote x AVPP59328, Melkazala x AVPP0512, for fruit length; Marako fana x AVPP0206, Marakofana x AVPP0514, AVPP9813 x AVPP0512, AVPP9813 x AVPP0105 and AVPP0514 x AVPP0105 for fruit width. Fruit number per plant had significant positive SCA effects in crosses, such as Melka awaze x AVPP0105, Melkashote x AVPP9813, Melkashote x AVPP59328 and AVPP0514 x AVPP59328. In most cases, the result indicated that, SCA effects of fruit length, width and fruit number had much more related effects with that of fruit yield (Table 4). Navhale et al. (2014) and Lohithaswa et al. (2000) observed and reported almost similar finding.

Moreover, the result showed that the ratios of gca to sca mean squares were greater than that for all traits, except fruit dry weight indicating the influence of additive gene actions in the inheritance of the characters (Table 4). This is again in line with most of the previous findings, suggesting that additive variance (gca) might have played more significant roles in the  expression  of both plant height and fruit length with contrasting result in other traits (Reif et al., 2007; Fekadu et al., 2009). Similarly, Geleta and Labuschagne (2006) reported that the magnitude of mean squares for gca was higher than that of sca in traits studied in introduced and Ethiopian hot pepper genotypes.

The authors have not declared any conflict of interests.

The authors thank Wolaita Sodo University and Jimma University College of Agriculture and Veterinary Medicine for their valuable contribution in both facilitation and financial support. Their recognition also extends to Melkasa Agricultural Research Center, especially vegetable program staff for their cooperation in providing planting materials, and AVRDC for the same. Moreover, they thank Prof. Derbew Belew for his unreserved contribution during material transfer from AVRDC.