Introduction
Cotton, being the king of fibers in preparing human apparel has played a key role in civilization of mankind. Cotton is providing livelihood directly and indirectly to over 60 million people and accounting for about 16 per cent of India’s export earnings. India has a pride place in the global cotton scenario due to several distinct features such as the largest cotton growing area, cultivation of all the four cultivated species, large area under tetraploid cotton, possibly the only country to grow hybrid cotton involving different species of cotton, native home of old world cultivated cotton and wide diversity in agro-climatic conditions under which cotton is grown. There is maximum diversity in the quality of cotton grown in India ranging from 5s counts to 120s counts.

There is a constant need to develop more potential hybrids and adopt noval approaches for improving hybrid performance. In cross pollinated crops like maize heterotic populations are developed and exploited through population improvement schemes meant for improving combining ability. Such programmes are integral part of hybrid breeding programme and these populations are shared among breeders and used further to obtain more potential hybrids. Studies have shown that even in cotton it is possible to adopt these concepts with suitable modifications in the procedure to suit the mating system of self pollinated crops (Patil and Patil, 2003 and Patil et al., 2007).

In several studies conducted on improving combining ability in often cross pollinated crops like red gram (Patil, 1997), sorghum (Patil and Pandit, 1991 and Madhusudhana, 1993) and cotton (Patil and Patil, 2003, Mallikarjun, 2005, Somashekar, 2006 and Ramakrishna 2008), attempts were made to exploit the potentiality of the simple traditional method of practicing selection in segregating generations (obtained though hybridization) and utilizing the recombinational variability for improving combining ability as a trait. These studies have clearly indicated that combining ability of lines could be improved by following selection for combining ability (as a trait) in segregating generations.

Mallikarjun (2005), Patil and Patil (2003) and Somashekhar (2006) worked on developing intra hirsutum heterotic population for creating recombinational variability for combining ability. The studies have confirmed that it is possible to develop potential intra hirsutum hybrids through exploitation of recombinational variability for combining ability. There are no studies to evidence the possibility of exploiting recombinational variability for combining ability in developing potential inter specific hybrids. Realizing the need for developing potential inter specific (H×B) hybrids, a detailed study was initiated at University of Agricultural Sciences, Dharwad during 2007/08 to identify hirsutum and barbadense genotypes capable of giving potential inter specific hybrids.

Based on this study two barbadense and four hirsutum lines giving best hybrid (H×B) combinations between them were selected. To create recombinational variability, the two barbadense genotypes were crossed to get F₁ and it was advanced to F₄ generation. In the present phase of this continuing study, the F₄ lines of this population ( barbadense x barbadense) is utilized for assessing recombinational variability for combining ability against selected hirsutum testers. Nature and magnitude of variability for combining ability was assessed against each hirsutum tester included in the heterotic box. In this study new population of F₄ lines was developed by crossing DB 533 and DB 534. The improvement seen in the barbadense lines was assessed in terms of productivity and fiber quality traits. The variability for combining ability of these F₄ lines was assessed in the study by crossing them with four hirsutum testers. These testers were decided upon based on evaluation of the hybrid involving parental barbadense lines with these hirsutum testers. The main objectives of this study:

Exploitation of heterotic groups by creation of recombinational variability in G.barbadense F₄ population for ability to combine with selected diverse G. hirsutum testers.

To determine combining ability effects (gca and sca), variances (GCA and SCA), combining ability patterns of these barbadense and hirsutum lines.

1 Conclusion
There are several advantages of evaluating combining ability of lines in early segregating generations. Identification of superior combining lines in early segregating generations would at least avoid undue multiplication and advancing of material of little genetic worth i.e., with low combining ability. Jenkin’s (1935) study revealed that combining ability of a line is heritable and thereby he indicated that potential parents of hybrids can be selected in an early generation of inbreeding. It is well known that plants in early segregating generation have a higher level of heterozygosity, while, the segregating lines in later generations viz., F₅, F₆ and F₇ would be nearly homozygous. Hence, evaluating for combining ability in later generation perhaps would be automatically more reliable. Despite this, it is felt that selection needs to be initiated as early as possible in a generation where the constitution would have reached a satisfactory level of uniformity (homozygosity). In fact, F₄ is a generation in which for the first time even in conventional pedigree method emphasis for the selection (for performance) is laid exclusively on the line mean. It means that F₄ onwards one need not distinguish individual plants performance in a line (Allard, 1960).

2 Material and Methods
2.1 Choice of the material
To create recombinational variability for combining ability (Figure 1), the elite barbadense lines DB 533 and DB 534 were crossed during 2007-2008. During two seasons 2008-2009 and 2009-2010 these barbadense crosses were advanced to F₂ and F₃ generations, respectively. The F₃ lines were evaluated for productivity and fiber quality parameters realizing the emphasis laid on developing ELS (Extra Long Stable) cotton hybrids out of 171 F₃ lines, only those F₃ lines with acceptable fiber strength were utilized in the study on recombinational variability of combining ability.

Figure 1 Schematic presentation of the procedure following in the study to improve combining ability

 
During 2010 -2011 those twenty eight F₄ lines of barbadense cross DB 533 × DB 534 depending on the higher value of fiber tenacity, were crossed with the selected four hirsutum testers viz., DH 98-27 (T₁), ZCH 8 (T₂), 178-24 (T₃) and DH 18-31 (T₄₎ selected based on earlier study. Each barbadense F₄ line was involved in a set of crosses (112 crosses refer to as derived F₁ crosses) were subjected to Line x Tester analysis.

2.2 Crossing programme
The crossing programme was taken up during 2010. The F₄ lines and four common testers were sown on staggered dates. To obtain derived F₁ crosses seeds, the flower buds of the proper size from testers (used as female) were hand emasculated in the evening between 3.00 to 6.00 pm. The emasculated flowers were covered by butter paper packets for avoiding out crossing as well as ensuring their easy identification at the time of crossing. The emasculated flowers were pollinated during the next day morning between 9.30 am and 11.30 am by brushing the pollen from one of the F₄ lines (used as male) on the stigmatic surface. The pedicel of each pollinated flower was tied with price label bearing date and lines number for identification of crossed bolls. In this manner derived F₁ crosses seeds were obtained. Simultaneously, the barbadense population of F₄ lines was selfed and material was advanced to F5 generation during the same year.

2.3 Evaluation of derived F₁ crosses and F₅ barbadense lines
There was a need for improving performance of inter specific hybrids. This was possible through genetic improvement of barbadense varietal lines. So that both productivity and fiber quality of barbadense were improved. An improved barbadense varietal base is essential for improving performance of inter specific hybrids.

The entire experimental material was planted on a medium black soil at College of Agriculture, Dharwad under irrigated condition. All the 53 F₅ (included Suvin variety as check) lines, four hirsutum testers and derived F₁ crosses along with the straight crosses (Bench Mark Crosses (BMC)) and ruling commercial checks (MRC 6918 Bt check and DCH 32 non Bt check) were sown during kharif 2011 in a Randomized Block Design with two replications and a spacing of 90 cm between rows and 60 cm between the plants within a row (Plate 1). Recommended fertilizer doses were applied and other cultural practices were carried out at regular interval. Plant protection measures were taken at appropriate time to control pests and diseases.

Plate 1 General view of the derived F1 crosses at University of Agricultural Sciences, Dharwad

 
To facilitate Line Tester analysis, the crosses obtained were randomized and were sown in one block along with checks, bench mark crosses and parents were sown in adjoining block.

3 Experimental Results and Discussion
This study was aimed at evaluating recombinational variability for combining ability in F₄ generation. To assess variability for combining ability, twenty eight F₄ (Gossypium barbadense L.) lines were crossed with four common diverse testers (Gossypium hirsutum L.) viz., DH 98-27 (T₁), ZCH8 (T₂), 178-24 (T₃) and DH 18-31 (T₄) for use in assessing the variability for combining ability.

In this study an improvement in combining ability is defined (described) as improvement in performance of derived F_1s over respective other crosses involving the tester concerned. When we refer to “performance of derived F₁” we primarily look at on the performance of seed cotton yield as a measure of combining ability. Seed cotton yield is the most important character directly reflecting the economic worth of the crop. Hence, it is more meaningful to assess the combining ability of a line in terms of the seed cotton yield of the F_1s derived from it. The combining ability of each line with four testers was assessed for comparing the derived F_1s with the other crosses and commercial checks.

Combining ability analysis
Analysis of variance for combining ability
The analysis of variance for 14 characters studied for this set are presented in Table 1. Among the lines (males), the mean sum of squares (MSS) were not significant for all the characters except mean boll weight, reproductive points on sympodia and seed cotton yield which showed highly significant differences, while number of bolls per plant and ginning outturn exhibited significant differences. Testers (females) exhibited not significant difference for all the characters except number of bolls per plant and seed cotton yield which recorded highly significant differences, while mean boll weight and transpiration rate showed significant differences. Line x Tester interaction were highly significant differences for number of monopodia per plant, number of bolls per plant, mean boll weight, reproductive points on sympodia, seed index, lint index, photosynthetic rate, stomatal conductance and transpiration rate, while seed cotton yield had significant differences.

Table 1 Analysis of variance for combining ability in derived F1 crosses for different quantitative characters

 
The estimates of variance due to general combining ability (GCA), variance due to specific combining ability (SCA), the magnitude of SCA variances were greater than GCA variance for all 14 characters and the variance ratio was less than half in these traits, indicating that dominance variance was more than additive variance for these characters (Table 2). Patel et al. (2005), Paulo Antnio de Aguiar et al. (2007), Kumboh et al. (2008), Wankhade et al. (2008), Naqib Ullah et al. (2009), Cetin Karademir et al. (2009), Basal et al. (2009) , Deosarkar et al. (2009b) and Mohammad Reza et al. (2010) recorded the same results.

Table 2 Variance due to general and specific combining ability in derived F1 crosses for different quantitative characters

 
Combining ability effects
The estimates of general combining ability effects of females and males presented in Table 3, their specific combining ability effects are presented in Table 4 for all the characters.

Table 3 Estimates of general combining ability effects of parents involved in recombinational variability study for different quantitative characters

Table 4 Estimates of specific combining ability effects of derived F1 crosses for different quantitative characters

 
Seed cotton yield (kg/ha)
Eight lines recorded significant gca effects (Figure 2), of which five lines exhibited positive significant gca effects. The highest gca effect was found by the line DB 533 x DB 534 F₄ IPS 8 (363.15). Among the testers, DH 98-27 had positive significant gca effect (94.65) and the tester DH 18-31 showed positive gca effect (35.44), while 178-24 (-77.43) recorded negative significant gca effect (Figure 2). Two crosses manifested positive significant sca effects, of which the cross DH 98-27 X (DB 534 x DB 533 F₄ IPS 22) (680.34) recorded the highest positive sca effect.

Figure 2 Estimates of general combining ability effects of parents involved in recombinational variability study for seed cotton yield

 
Plant height (cm)
The estimates of general combining ability effect indicated significant differences among the line parents, three lines had significant positive gca and two had significant negative gca. The highest value of gca effect was exhibited by DB 533 x DB 534 F₄ IPS 36 (14.29). All four testers exhibited non significant gca effects. Seven crosses differed significantly for sca effects, of these four had positive sca effects. The cross DH 98-27 X (DB 533 x DB 534 F4 IPS 12) (26.27) had expressed the highest value of sca effect.

Number of monopodia per plant
Nine line parents expressed significant gca effects and the line DB 533 x DB 534 F₄ IPS 48 (0.40) was the best general combiner among the line parents. Among the testers, DH 18-31 showed significant positive gca effect (0.06) for number of monopodia per plant. Twenty seven crosses expressed significant sca effects. Among these, fourteen crosses had positive sca effects and the maximum sca effect was recorded by the cross DH 98-27 X (DB 533 x DB 534 F₄ IPS 23) (0.53).

Number of sympodia per plant
Out of 28 male (Line) parents used in the population based crosses, one line showed significant negative gca effect was shown by DB 533 x DB 534 F₄ IPS 15 (-2.36). Among the females (testers), there are no significant gca effects. Two crosses expressed significant sca effects were shown by DH 98-27 X (DB 533 x DB 534 F₄ IPS 71) and 178-24 X (DB 533 x DB 534 F₄ IPS 48) -4.76 and -4.55, respectively.

Number of bolls per plant
Among the lines nine lines had significant gca effects, out of which five lines recorded positive significant gca effects and DB 533 X DB 534 F₄ IPS 24 (5.52) exhibited highest value of gca effect. Among the testers, DH 98-27 and DH 18-31 recorded positive significant gca effects 1.37 and 0.95, respectively. The tester 178-24 (-1.87) had negative significant gca effect. Two crosses differed significantly for sca effects, these two crosses had positive sca effects and shown by ZCH 8 X (DB 533 x DB 534 F₄ IPS 30) and 178-24 X (DB 533 x DB 534 F₄ IPS 17) 17.97 and 13.87, respectively.

Mean boll weight (g)
The estimates of gca effects of line parents in the population based crosses were found to be significant in two lines, out of which one was positive significant and other was negative significant differences were shown by DB 533 x DB 534 F₄ IPS 52 (0.38) and DB 533 x DB 534 F₄ IPS 30 (-0.47). Among the testers, ZCH 8 recorded positive significant gca effect (0.03), while 178-24 had negative significant gca effect (-0.05). One cross expressed positive significant sca effect shown by ZCH 8 X (DB 533 x DB 534 F₄ IPS 49) (0.88).

Reproductive points on sympodia
Out of 28 male (Line) parents used in the population based crosses, thirteen lines recorded significant gca effects. Among these, five lines showed significant positive gca effects and the cross DB 533 x DB 534 F₄ IPS 15 (0.93) had the highest value of gca effect. Among the testers, ZCH 8 (-0.17) exhibited negative significant gca effect. Twenty four crosses expressed significant sca effects. Among these, thirteen had positive sca effects and the maximum sca effect was recorded by the cross 178-24 X (DB 533 x DB 534 F₄ IPS 12) (1.56).

Sympodial length at 50 % plant height (cm)
The estimates of gca effects of lines revealed positive significant differences for two line parents. These lines are DB 533 x DB 534 F₄ IPS 30 (6.02) and DB 533 x DB 534 F₄ IPS 36 (5.00). Among the testers, there are no significant gca effects. The estimates of sca effects were significant for eight crosses. Of these, six crosses had positive sca effects and the highest sca effect was expressed by the cross DH 98-27 X (DB 533 x DB 534 F₄ IPS 38) (14.62).

Seed index (g)
Eight line parents exhibited significant gca effects, four lines had significant gca in positive direction and highest was recorded by the line DB 533 x DB 534 F₄ IPS 30 (1.30). Among the testers, 178-24 and DH 18-31 showed positive and negative significant gca effects 0.34 and -0.43, respectively. Twenty nine crosses revealed significant sca effects, of these fourteen crosses showed positive sca effects and the highest was displayed by the cross DH 98-27 X (DB 533 x DB 534 F₄ IPS 32) (5.80).

Ginning outturn (%)
Six line parents displayed significant gca effects, of which two lines exhibited significant positive gca effects. Highest gca effect was recorded by DB 533 x DB 534 F₄ IPS 16 (2.93). None of the testers depicted significant gca effect for this character. Six crosses expressed significant sca effects of these, two crosses showed positive sca effects. The cross ZCH 8 X (DB 533 x DB 534 F4 IPS 26) (5.52) recorded highest positive sca effect.

Lint index (g)
Out of twenty eight lines, eight showed significant gca effects, of which three lines had positive gca effects. The highest gca effect was found by the line DB 533 x DB 534 F₄ IPS 17 (1.34). The tester ZCH 8 (-0.27) showed negative significant gca effect and the tester 178-24 (0.21) showed positive significant gca effect. Of the 112 crosses studied, seventeen crosses recorded significant sca effects for this character, of which nine showed significant gca effects in positive direction. The cross DH 98-27 X (DB 533 x DB 534 F₄ IPS 32) recorded highest positive sca effect (1.98).

Photosynthetic rate (µmol CO²m^-2s^-1)
Out of 28 line parents, twenty one lines are having significant gca effects. The male parent DB 533 x DB 534 F₄ IPS 26 (7.36) showed highest positive significant gca effect. Among the testers, 178-24 and DH 18-31 had significant negative and positive gca effects -1.30 and 1.47, respectively. Fourty eight crosses showed significant sca effects, of these twenty six crosses had positive sca effects. The cross ZCH 8 X (DB 533 x DB 534 F₄ IPS 25) expressed the highest sca effect (13.09).

Stomatal conductance (µmol m^-2 s^-1)
Out of twenty eight male lines, sixteen showed significant gca effects, of which eight lines exhibited positive gca effects and the highest gca effect was found by the line DB 533 x DB 534 F₄ IPS 26 (0.31). The tester 178-24 registered significant negative gca effect (-0.07), while ZCH 8 and DH 18-31 revealed significant positive gca effects 0.03 and 0.05, respectively. Of the 112 crosses, fifty two crosses recorded significant sca effects, of which twenty seven showed positive sca effects and the cross ZCH 8 X (DB 533 x DB 534 F₄ IPS 52) (0.58) recorded highest positive sca effect.

Transpiration rate (mmol H₂O m^-2 s^-1)
The estimates of gca effects of lines and testers for transpiration rate trait, six line parents expressed significant gca effects. Out of which three lines showed positive significant gca effects and the highest gca effect was found by the line DB 533 x DB 534 F₄ IPS 55 (1.98). Among the testers, two are with significant gca effects, of which DH 18-31 indicated significant positive gca effect (0.55) and 178-24 indicated significant negative gca effect (-0.56). As regard to sca effects, twenty two crosses recorded significant sca effects, of which ten were significant positive gca and the hybrid DH 98-27 X (DB 533 x DB 534 F₄ IPS 1) registered significant and highest positive sca effect (3.58).

Pooled score for gca effects
Simple pooled gca score method
In this approach, significant gca effect in desirable direction is given positive weightage (+1) and negative weighatage (-1) is given for gca effect in undesirable direction (Arunachalam and Bandopadhyay, 1979). These values are added over different yield influencing characters to arrive at pooled score of gca effects. The inherent disadvantage with this system is that all the parents with significant gca effects in desirable direction get the same score (positive). Hence, it is not possible to quantify the magnitude of difference existing among the genotypes of this group which get a positive score. Therefore, it is necessary develop a system of working out pooled scores of gca by utilizing the actual gca values and ensuring quantification of every possible difference existing in gca effects between only two parents.

Per cent gca method
When the actual gca values are added across characters to arrive at pooled score, problem arises because of difference in unit of measurement of each character. Absolute values of gca effects may be big (plant height) or small (boll weight) depending on the character and if used, the importance of the character may not be projected correctly. If the raw values of gca effects are added across the characters, the character with higher per se effect influence the pooled scores most as against the character with low per se gca values. To overcome this disadvantage, the raw gca values have to be converted into per cent gca values.

Thus, by working out per cent gca values, the minute differences in gca values are also focussed and the possible problem arising out of the differences in unit of measurement, high and lower per se gca values associated with the type of character concerned are overcome.

Weighted per cent gca method
In this method further improvement is brought about in arriving at the pooled gca scores of the parents. In per cent gca method, the per cent gca values are straight away added across the characters which means each character including yield and yield components are all given equal weightage. The experience of the breeders would suggest sometimes that, in arriving the pooled score, it is desirable to attach differential weightages to each of the characters studied depending upon its economic importance, contribution to yield etc. These weightages can be multiplied with per cent gca values of corresponding characters and then added to arrive at the pooled gca score for each parent. In the present study weightages for different yield related characters were worked out by consulting the senior breeder related to cotton viz., Dr. S. S. Patil, Senior Scientist, Agricultural Research Station, Dharwad and Dr.B.C.Patil, principal scientist (Plant physiology), ARS, Dharwad Farm.

In F₄ lines, based on simple pooled gca score method (Table 5), the F₄ lines DB 533 x DB 534 F4 IPS 26, DB 533 x DB 534 F4 IPS 17, DB 533 x DB 534 F4 IPS 48 and DB 533 x DB 534 F₄ IPS 8 (Decreasing order) are recognized as the most potential parents. Among the testers DH 18-31 based on simple pooled gca score method showed the most potential parent. Based on per cent gca method, the line parents DB 533 x DB 534 F₄ IPS 26, DB 533 x DB 534 F₄ IPS 8, DB 533 x DB 534 F₄ IPS 17 and DB 533 x DB 534 F₄ IPS 55 (Decreasing order) were the most potential combiners (Table 6). Among the testers, DH 18-31 based on per cent gca score method showed the most potential parent. Similarly, based on weighted gca method the most potential combiners were found to be the lines DB 533 x DB 534 F₄ IPS 26, DB 533 x DB 534 F₄ IPS 17, DB 533 x DB 534 F₄ IPS 8 and DB 533 x DB 534 F₄ IPS 32 . Among the testers, DH 18-31 based on weighted gca method is the most potential parent (Table 7). The overall combining ability status of F₄ barbadense lines was determined by working out pooled gca score. Three methods namely simple gca score, per cent gca and weighted gca method were used in arriving at the best general combiner lines. The overall ranking from these approaches differed and the weighted gca approach helped in preciese identification of potential combiners.

Table 5 Pooled scores of F4 barbadense lines and hirsutum testers based on Simple pooled gca score

Table 6 Pooled scores of F4 barbadense lines and hirsutum testers based on Per cent pooled gca score

Table 7 Pooled scores of F4 barbadense lines and hirsutum testers based on weighted percent gca method

 
References
Allard, R. W., 1960, Principles of Plant Breeding, John Wiley and Sons, New York., pp. 75-98

Arunachalam, V. and Bandopadhyay, 1979, Are “Multiple cross - multiple pollen” an answer for productive population in Brassica campestiris var. brown Sarson. Theor.Appl. Genet., 54: 203-207
http://dx.doi.org/10.1007/BF00267708
 
Basal, H., Unay, A., Canavar, O. and Yavas, I., 2009, Combining ability for fiber quality parameters and within boll yield components in intra specific and inter specific cotton populations. Spanish J. Agri. Res., 7(2): 364-374 http://dx.doi.org/10.5424/sjar/2009072-428

Cetin Karademir, Oktay Gencer, Emine Karademir and Remzi Ekinci., 2009, Combining ability estimates and heterosis for yield and fibre quality of cotton in line X tester design. Notulac Botanicae Horti Agrobotanici Cluj-Napoca., 37(2): 228-233

Deosarkar, D. B., Jadhav, D. S. and Patil, S. G., 2009b, Combining ability studies for yield and quality traits in cotton (Gossypium hirsutum L.). J. Cotton Res. Develop., 23(2): 183-187

Jenkins, M. T., 1935, The effect of inbreeding and of selection within inbred lines of maize upon the hybrids made after successive generations of selfing. Iowa State J. Sc.i, 3: 429-450

Kumboh, N., Baloch, M. J., Kumbhar, M. B., Khanzada, S. and Jatoi, W. A., 2008, Diallel analysis for estimating combining ability in upland cotton (Gossypium hirsutum L.). Pakistan J. Agril. Engin. and Veter. Sci., 24(1): 27-33
Madhusudhana, R., 1993, Combining ability in segregating generation of sorghum (Sorghum bicolor L. Moench). M. Sc. (Agri.) Thesis, Univ. Agric. Sci., Dharwad, India

Mallikarjun, H. B., 2005, Evaluation of recombinational variability for combining ability in cotton. Ph. D.(Agri). Thesis, Univ. Agric. Sci., Dharwad

Mohammad Reza Zangi and Nadali Babaein Jelodar, 2010, Cytoplasmic and combining ability effects on agro-morphological characters in intra and inter crosses of pima and upland cottons (G. barbadense and G. hirsutum). Internat. J. Biol., 2(1): 94-102

Naqib, U. K., Khan, B. M., Gul, H., Farhatullah, M. B., Aisha, P., Umm-E-Aiman, Muhammad, Z. K. and Zahoor A. S., 2009, Diallel analysis of some quantitative traits in Gossypium hirsutum L.Pakistan J. Bot., 41(6): 3009 -3022

Patel, K.G., Patel, U.G., Patel, J.C., Patel, D.H. and Patel, R.D., 2005, Line x Tester analysis for combining ability of introgresed Asiatic cotton. J. Indian Soc. Cotton Improv., pp.105-109

Patil, S. S., 1997, “Exploiting variability for important economic traits resulting from combinations of early and medium duration pigeon pea varieties” Annual Report 1997 of ICAR ADHOC Scheme

Patil, S. S. and Pandit, V. V., 1991, Use of B R cross for improving combining ability in sorghum. Golden Jubilee Symposium on Research and Education : Current Trends and the Next Fifty Years, held at New Delhi, p. 397

Patil, S. S. and Patil, S. A., 2003, Role of improving combining ability in increasing performance of cotton hybrids. Third World Cotton Research Conference, 9-13 March 2003, held at Cape Town South Africa, pp. 234-238

Patil, S. S., Patil, S. A., Ramakrishna, V., Deepakbabu, H. and Mallikarjun, H. B., 2007, Approaches of improving combining ability for enhancing performance of cotton hybrids. Fourth World Cotton Research Conference, 10-11th September 2007, held at Lubbock, Texas, U. S. A

Paulo Antonio De Aguiar, Julio Cesar Viglioni Penna, Eleusio Curvelo Freire and Leonardo Cunha Melo, 2007, Diallel analysis of upland cotton cultivars. Crop Breed. Appl. Biotechnol., 7(1): 353-359

Ramakrishna, V., 2008, Reciprocal selection for improving combining ability in cotton. Ph.D. Thesis, Univ. Agric. Sci, Dharwad, Karnataka (India)

Somashekhar, D., 2006, Genetic studies on reciprocal selection for combining ability to improve hybrid performance in cotton. Ph.D (Agri.) Thesis, Univ. Agric. Sci, Dharwad, Karnataka (India)

Wankhade, S. N., Patil, S. P., Kayande, N. V. and Burghate, S. K., 2008, Diallel analysis for estimating combining ability of quantitatively inherited traits in upland cotton. J. Cotton Res. Dev., 22(2): 147-149