There were significant differences among the parents as well as their F₁s for resistance to pod borer, H. armigera, both under field and laboratory conditions (Tables 3 and 4), justifying the selection of parents for this study. In general, the mean pod borer damage was lower in F₁s than the parents, but the variability in pod borer damage between the F₁s was greater than the parents. There were substantial differences in gca effects of the parents (as indicated by σ² g) for pod borer damage, which resulted in progenies (F₁s) with differential abilities to resist pod borer damage (Tables 3 and 4). Both σ² g and σ² s were significant for pod borer damage in straight crosses. Only σ² g was significant in reciprocal crosses, suggesting the importance of both additive and non-additive type of gene action. However, greater magnitude of σ² A (17.39 and 1.42) than σ² D (3.93 and 1.21) clearly indicated the preponderance of σ² A in the inheritance of resistance to H. armigera (Tables 5 and 6). Gowda et al. [14,26] and Singh et al. [27] reported that additive and dominance genetic variances, respectively, were predominant for pod borer resistance in earlyand medium-maturity genotypes. However, additive as well as dominance components of genetic variances were equally important in the inheritance of pod borer resistance in the late-maturity genotypes. Salimath et al. [28] reported the involvement of both additive and non-additive type of gene action in the inheritance of pod borer resistance. The parental lines used in the current study mostly belonged to early-and medium-maturity groups, and the results suggested the predominance of additive type of gene action.

In the straight crosses, ICC 506EB, ICC 12477, ICC 12478, ICC 12479 and ICCV 2 were the best general combiners with significant negative gca effects and low pod borer damage (Table 7). Under laboratory conditions, ICC 506EB and ICC 12477 suffered low leaf damage rating with significant and negative gca in detached leaf assay (Table 8). Earlier studies have shown that ICC 12478, ICC 12479, ICC 14876, ICC 506EB and ICC 12477 have negative gca effects, and have a good ability to transmit additive genes to decrease pod borer damage [14,18]. These parents can be involved to generate useful variability for selecting lines with resistance to H. armigera. The hybrids ICC 506EB × ICC 3137, ICC 12476 × ICC 3137, ICC 12477 × ICC 4918, ICC 12479 × ICC 3137 and ICC 3137 × ICCV 2 showed significant and negative sca effects, and were good specific combiners for resistance to pod damage by H. armigera. In detached leaf assay, the hybrids ICC 12477 × ICCC 37, ICC 3137 × ICCV 2 and ICC 4918 × ICCV 2 showed significant and negative sca effects for antibiosis resistance to H. armigera (Table 9). It is interesting to note that these crosses involve one parent with excess of

Table 3. Estimates of mean squares and variances due to general combining ability (GCA) and specific combining ability (SCA) effects for F₁ chickpea hybrids under natural conditions in the field (9 × 9 full diallel, Griffing 1956).

d.f. = Degrees of freedom; ^*,** = Significant at P = 0.05 and 0.01 respectively; σ² g: Variance due to gca, σ² s: Variance due to sca, and σ² r: Variance due to reciprocal effects. σ² A: Additive variance, σ² D: Dominance variance, and PR: Predictability ratio.

Table 4. Estimates of mean squares and variances due to GCA and SCA for resistance to Helicoverpa armigera in chickpea (9 × 9 full diallel, Griffing 1956) (based on detached leaf assay).

d.f. = Degrees of freedom; ¹Damage rating (1 = <10%, and 9 = >80% leaf/pod damage); ^{**, ***} = Significant at P = 0.01 and 0.001, respectively; σ² g: Variance due to gca, σ² s: Variance due to sca, σ² r: Variance due to reciprocal effects, and σ² e: Error variance; σ² A: Additive variance, σ² D: Dominance variance, and PR: Predictability ratio.

Table 5. Nature of gene action for pod damage by Helicoverpa armigera and grain yield in chickpea.

Table 6. Nature of gene action for antibiosis component of resistance to Helicoverpa armigera (based on detached leaf assay).

^*Damage rating (1 = <10% leaf area damaged, and 9 = >80% leaf area damaged).

Table 7. Estimates of general combining ability (GCA) effects of nine chickpea genotypes for pod damage by Helicoverpa armigera and grain yield under natural conditions.

Table 8. Estimates of general combining ability (GCA) effects of nine chickpea parents for resistance to Helicoverpa armigera (based on detached leaf assay).

^*Damage rating (1 = <10% leaf area damaged, and 9 = >80% leaf area damaged). ^*, ^**, ^*** = Significant at P 0.05, 0.01 and 0.001, respectively. PR: Predictability ratio.

recessive genes, and the other with equal frequency of dominant and recessive genes [14], indicating that the parents involved in the crosses are diverse for nature of genes for pod borer resistance. It is desirable to exploit the crosses with significant sca effects, involving parents contrasting for gca effects, and for reaction to pod borer damage (resistant and susceptible). Recurrent selection in the population developed by random mating of the pod borer resistant parents in high-yielding background used in a diallel would facilitate accumulation of favorable gene combinations in the homozygous and heterozygous state [14].

The sca effects for the reciprocal crosses were nonsignificant, indicating the importance of both additive and non-additive gene effects for pod borer resistance. The cross ICCV 2 × ICC 3137 showing significant and negative sca effects was good specific combiner for resistance to pod borer damage. Similar results have been

Table 9. Estimates of specific combining ability (SCA) effects of F_Is for antibiosis component of resistance to Helicoverpa armigera (based on detached leaf assay).

¹Damage rating (1 = <10% leaf area damaged, and 9 = >80% leaf area damaged). ^{*, **, ***} = Significant at P 0.05, 0.01 and 0.001, respectively.

reported by Singh and Paroda [29]. Better progenies have been obtained from the crosses involving diverse parents for gca effects in groundnut [30]. Langham [31] has provided the evidence for the usefulness of divergent parents for maximizing potential transgressive lines in cross derivatives in rice. Effectiveness of pedigree selection for pod borer resistance in chickpea has also been reported by Sharma et al. [32] and Dua et al. [22]. Singh et al. [33] developed pod borer resistant chickpea line, ICCV 7 using pedigree selection of the lines derived from a cross between H 208 and BEG 482. However, caution is necessary while using pedigree selection for pod borer resistance, considering the existence of tight linkage between susceptibility to fusarium wilt (Fusarium oxysporum f. sp. ciceri) and resistance to pod borer in chickpea. Biparental matings in early segregating generations of a multiple cross involving a few pod borer and fusarium wilt resistant genotypes in high yielding background would provide increased opportunity for recombination, which will facilitate disruption of tight linkage between the genes for these traits [34,35]. Germplasm lines (ICC 12478, ICC 12479 and ICC 14876) having stable resistance to H. armigera and moderate yield potential [36] could be used in enhancement of pod borer resistance in elite agronomic background.

There were significant differences among the parents as well as their F₁s for survival of H. armigera larvae (Table 4). In general, the mean larval survival was lower in F₁s than in parents. Greater magnitude of σ² D (0.50) than σ² A (0.23) clearly indicated the preponderance of σ² D in the inheritance of antibiosis component of resistance (% larval survival) to H. armigera larvae (Table 4). In the straight crosses, ICC 12476, ICC 12477 and ICCV 2 were the best general combiners with significant negative gca effects for larval survival (Table 8). The hybrids ICC 12476 × ICC 12478 and ICC 12477 × ICC 12478 showed significant and negative sca effects and were good specific combiners for resistance to survival of H. armigera larvae (Table 9).

There were significant differences among the parents as well as their F₁s for weight of the H. armigera larvae. In general, the mean larval weights were lower in F₁s than in the parents. Greater the magnitude of σ² D (50.21) than σ² A (38.37) clearly indicated the preponderance of σ² D in the inheritance of antibiosis to H. armigera in terms of weight gain by the H. armigera larvae in detached leaf assay (Table 4). In the straight crosses, ICC 506EB, ICC 12476, ICC 12477 and ICC 12478 were the best general combiners with significant negative gca effects and low larval weights (Table 8). The hybrids ICC 12476 × ICC 3137, ICC 12476 × ICC 506EB, ICC 12478 × ICCC 37, ICC 12479 × ICC 506EB, ICC 3137 × ICC 506EB and ICC 4918 × ICCV 2 showed significant and negative sca effects, and were good specific combiners for antibiosis resistance (weight gain) to H. armigera (Table 9).

In detached leaf assay, resistance to leaf feeding was governed by additive gene action. These results are in accordance with field observations by Gowda et al. [14,26] and Singh et al. [27]. Antibiosis component of resistance (based on larval survival and larval weight) was governed by non-additive type of gene action (Table 6).

There were significant differences among the parents and the F₁ crosses (straight and reciprocal crosses) for grain yield (Table 3). The importance of both σ² g and σ² s was evident with predominance of the latter, which is amply reflected from a much higher magnitudes of σ² D than σ² A. Similar results have been reported by Deshmukh and Patil [37] and Gowda et al. [14]. The predominance of σ² D indicated the importance of non-additive gene action. Since gca effects are the manifestation of additive properties of genes, parents selected based on gca effects will be useful for developing breeding lines with high grain yield. Parents with good general combining ability such as ICC 4918 could be used in breeding programs. Gowda et al. [14] reported that ICC 12476 possessed an excess of dominant genes, ICC 12426, ICC 12478 and ICC 4918 possessed an excess of recessive genes, while ICC 506EB possessed equal frequency of dominant and recessive genes for grain yield under un-protected conditions. Parents with significant and positive gca effects are diverse for the nature of gene action for grain yield. The combining ability variances were significant for both gca and sca. The predictability ratio of 0.23 showed that gca alone was not sufficient for predicting the performance of single cross progenies. Of the two genetic parameters, the magnitude of σ² D was more than σ² A, which emphasized that non-additive gene action, was involved in inheritance and expression of grain yield per plant (Table 5). These findings are in conformity with earlier reports [33-44]. However, the reports of Gowda [45], Asawa and Tewari [46], Sandhu et al. [47], and Gowda and Bahl [48] are contradictory to the present findings, which indicated the involvement of additive genetic variance. Singh et al. [49], Singh and Ocampo [50], Annigeri et al. [51], Sarode [52] and Girase [53] reported the importance of additive as well as non-additive genetic variances. The crosses ICC 12476 × ICCC 37, ICC 12477 ×

Table 10. Maternal effects for seeds per pod in chickpea.

^{*, **} = Maternal effects significant at P 0.05 and 0.01, respectively.

ICC 4918 and ICC 12478 × ICC 12479 with highly significant and positive sca effects were good specific combiners for grain yield. Interestingly, one of the parents of these crosses possessed an excess of dominant genes, while the other parent possessed an excess of recessive genes [14], suggesting the need to have genetic diversity for increased heterosis for grain yield. Theoretical investigations substantiating the necessity of parental diversity for better performance of crosses [54,55] lend adequate support to these practical considerations. Although there was a good correspondence between pod borer damage of crosses and their sca effects, the crosses (involving parents with contrasting gca effects) with significant sca effects need to be exploited for deriving superior lines for grain yield.

The significance of both σ² g and σ² s indicated the importance of both σ² A and σ² D in the reciprocal crosses. However, the higher magnitude of σ² D than σ² A is a clear evidence for predominance of σ² D in the inheritance of grain yield. Crosses ICC 12477 × ICC 506EB, ICC 3137 × ICC 506EB, ICCC 37 × ICC 506EB, ICCV 2 × ICC 12476 and ICCC 37 × ICC 3137 with highly significant and positive sca effects were good specific combiners for increased grain yield. The grain yield is predominantly under the control of non-additive gene action, irrespective of the maturity groups in desi type chickpea. Due to predominance of non-fixable genetic variation coupled with low heritability, it has not been possible to achieve breakthrough in increasing the chickpea productivity. Lack of sufficient variability (due to its strictly inbreeding behavior) is one of the reasons for limited progress in increasing chickpea productivity to a desired level [36]. The use of conventional breeding methods such as pedigree, single seed descent, and bulk methods are associated with the weakness of causing rapid homozygosity and low genetic variability, especially in the presence of linkage blocks and inverse relationships among the desirable traits [56]. Breeders need to exploit both additive and non-additive gene effects, besides disrupting undesirable associations and uncovering concealed variability. Biparental mating of segregants in a multiple crossing scheme might be useful in disrupting the undesirable linkages [34,35,57].

There was no maternal inheritance for pod borer damage and grain yield. The cross, ICC 12476 × ICCC 37 showed positive and significant sca effects for seeds per pod, but ICCC 37 × ICC 12476 showed negative and significant sca effects for number of seeds pod⁻¹, suggestive cytoplasmic inheritance for the number of seeds per pod (Table 10).