Ann. appl. Bid. (1976), 83, 139-147 Printed in Great Britain

Breeding for reduced seed coat fuzz in Upland cotton (Gossypium hirsutum)

BY N . L. INNES* Namulonge Research Station, Kampala, Uganda, and

Cotton Research Corporation, London

AND R. H. WIMBLE Rothamsted Experimental Station, Harpenden, Herts, ALg 2JQ

(Accepted 14 November 1975)

SUMMARY

The inheritance of seed coat fuzz was studied in two half diallel sets of crosses of Upland cotton. One with F4 selections from an inter-varietal cross showed a significant level of non-additive variance attributable to dominance and non-allelic interaction. In the other, using inbred varieties of diverse origin, the genetic control of seed fuzz was adequately accounted for by an additive-dominance model with no interaction.

Genotypic correlations between seed coat fuzz, yield and lint quality characters, calculated for both diallel sets and for two other groups of breed- ing material, showed good agreement within each experiment between parents and hybrids or between parents and progenies but no consistent pattern between experiments. The results serve to emphasize the risks in extrapolating correlations from one group of breeding material to another.

A useful level of reduced fuzz has been obtained in selections from the AH breeding programme and the genetical investigations indicate that a further reduction may be possible, thereby leading to easier handling of seed, speedier and cheaper ginning, low levels of seed coat nep and better seed germination.

I N T R O D U C T I O N

In Upland cotton (Gossypium hirsutum) the degree of fuzz on the seed is a varietal characteristic that may vary from a dense woolly cover to virtual absence of fuzz. A totally glabrous form of seed is rarely found in commercial varieties of Upland cotton and the term ‘naked’ or ‘black’ has often been used to describe seed with an apical tuft. The range of seed fuzz types has been illustrated by Low (1968) who had pre- viously outlined the many advantages of cotton seed with a reduced cover of fuzz (Low, 1962). Among these advantages were quicker germination, easier oil extraction, speedier and easier ginning, economy in storage and transport and elimination of seed coat nep (particles of seed ripped off during ginning of seed cotton which pass into the lint and cause problems during spinning, weaving and dyeing). Selection work by Low (1968) at Serere, Uganda, in stocks variable for degree of seed coat fuzz indicated that strains with reduced seed coat fuzz gave lint yields which compared favourably with

* Present address: National Vegetable Research Station, Wellesbourne, Wanvick, CV35 gEF.

N. L. INNES AND R. H. WIMBLE 140

their fully fuzzed counterparts. He emphasized, however, that there was a trend towards reduced lint length and increased fibre coarseness, leading to a marked drop in yarn strength in selections with little fuzz. Similarly Costelloe (1970) concluded that intense selection for a reduction in seed coat fuzz could lead to a reduction in ginning percentage (the ratio of lint to seed cotton) and to an increase in the coarseness of lint. He therefore advocated that the cotton breeder should be wary of selecting too in- tensely for a reduction in seed coat fuzz. Furthermore, Costelloe concluded that fuzzed varieties were nearly always likely to exceed their non-fuzzed counterparts in productivity.

In this paper the inheritance of seed coat fuzz is further examined using the results of two half diallel trials. One of these, known here as AH(67), had twelve parents which were selections from an inter-varietal cross. The other, WRC F,, had five inbred parents of diverse origins. The relationship between seed coat fuzz and other charac- ters of economic importance is also studied using these results together with those of two other trials.

MATERIALS A N D M E T H O D S

The derivation of the twelve AH(67) parents in the 12 x 12 half diallel set of crosses has been described by Innes (1973) and Innes, Wimble & Gridley (1975). Sixty-six hybrids from the half diallel set and the twelve parents were included in a 9 x 9 balanced lattice square with five replicates at Namulonge in 1969. Also included in this trial were the commercial variety BPA, and the two parents, A(61)21 and Barhop from which the twelve F, lines were derived. Plot size was 4'32 m2 with two plants per stand in two rows at a spacing of 0.9 m between and 0-3 m within rows.

The 1968 trial of ten F, populations and five inbred parents from a 5 x 5 half diallel- set (the WRC diallel) was a 4 x 4 balanced lattice with five replicates, in which was also included commercial BPA. Plot size was 8.1 m2 with two plants per stand in two rows at a spacing of 0.9 m x 0.3 m.

Lint yields were estimated from the weight of seed cotton harvested per plot and ginning percentage. The lint used for testing for fibre characters was obtained from a composite sample from all picks (Innes et al. 1975) and yield components were estimated from a random sample of fifteen bolls harvested from each plot at the time of the second pick. Seed fuzz grade was scored on a plot basis, using seed from the fifteen boll sample, on a scale of I (naked) to 9 (fully fuzzed).

Correlation matrices for yield, yield components and seed fuzz grade were calculated for the diallels and for two other sets of data. Both sets were obtained from two 7 x 7 balanced lattice squares, each with four replicates, at Namulonge in 1970. In the first trial were included thirty-four F , progenies of ALOA (70) which were derived from the double cross (Bar 24/4 x Acala 1517) x (A(61)21 x Coker 100W/2), four entries of A(61)21, four of Coker 100W/2, four of BPA and three of an ALOA Bulk F,. In the second trial there were forty-five CA(70) progenies, derived from crosses between BP52 and Albar 51 stocks and four entries of BPA68. Plot size in both trials was 3'24 m2 with two plants per stand in two rows at a spacing of 0.9 m between and 0.3 m within rows.

For genetical analysis of the seed fuzz data, Hayman's (1954) analysis of variance and Jinks' (1954) w, regression method were used. Genotypic correlations for the

141 Breeding for reduced seed coat fuzz in cotton parents and for the hybrids (or progenies) were calculated from the 'between parents' or 'between hybrids' mean squares and products from analyses of variance and covariance. Environmental correlations were similarly calculated using the mean squares and products for the interaction between entries and replicates. In all analyses the rows and columns of the lattice design were ignored.

All statistical analyses were carried out at Rothamsted Experimental Station on an ICL System 4-70 computer using GENSTAT (Nelder, 1975).

RESULTS

Genetical anuijves 12 x 12 half diallel set of A H crosses. The mean values of seed fuzz grade of the

twelve parents of this diallel ranged from 4.3 to 6.7 (Table I). As has been noted the parents derive from A(61)21 and Barhop, the mean values of which in the same experi- ment were 6.5 and 6.4 respectively, while that of BPA was 6.0. Only two of the twelve AH(67) parents exceeded the A(61)21 and Barhop means and then only slightly, while the mean of the twelve, 5-73, was below them.

The analysis shows that while much of the variance could be accounted for by

Table I . Mean seed fuzz grade of AH(67) parents and hybrids iu the 12 x 12 half diallel"

AH(67) I 2

5'1 4.0 5.2 4.0 4.8 4.0 4.0 5'4 5'4 6.0 5'9 5'2 4'9

5'3 5 '2 j.2 6.0 5'3 4'7 6.0 5.8

5'9 5'6

3 4 6 7 1 1

5.1 4'0 4'3 5.0 5.0 6.0 5.8 5.9 6.4 6.6 5.0 5.3 6.0 6.0 5'7 5.0 5 '2 6.2 6.2 6.0 5.2 5'2 6.0 6.1 6.0 5.6 5'1 6.0 6.0 5.6 5.2 5.0 6.0 6.0 6.0 6.0 6.0 6.3 6.8 6.0

S.E. of a mean 0.12.

* Parental values in italics.

13 17 18 19 20

6.1 6.0 6.0 5.8 6.0 5 9 5.9 6.1 6.0 6.1 6.2 6.8 6.0 6.1 6.7

Table 2. Hayman analysis of variance for seed fuzz grade in 12 x 12 half diallel set of AH c~osses

Source D.F. M.S. V.R.

U

b bi bz b3

a x blocks b x blocks

b, x blocks b, x blocks b, x blocks

I1 66

I I1

54 44

264 4

44 216

**, *** P<o.or , <O.OOI respectively.

142 N. L. INNES AND R. H. WIMBLE additive parental differences, there were also significant non-additive effects (Table 2). Transgressive segregation, in the direction of less fuzz,was shown in the cross A(61) 21 x Barhop (Innes, 1973) and there is evidence here of a further reduction in fuzz. Pedigree selection in the A(61)21 x Barhop material over several generations beyond the F4 indicated that the AH parents were largely homozygous for alleles controlling the degree of fuzz.

Tests for an additive-dominance model of the type described by Mather & Jinks (1971) were carried out, using the statistics W, and V,. The slope of the regression of W , on V , was consistently low in all blocks, averaging only 0.6. There were large differences between arrays in (W, + K), suggesting the presence of dominance, but the low slope indicated that a simple additive-dominance model would not be adequate. An analysis of variance of (q-v) showed significant differences between arrays thereby confirming the inadequacy of an additive-dominance model. In view of the transgressive segregation in the original intervarietal cross, epistasis resulting from complementary gene action cannot be ruled out.

Table 3. General Combining ability (GCA) estimates for twelve AH(67) parents and values for mean potence in 12 x 12 half diallel set

Parent I 2 3 4 6 7 1 1 13 17 18 19 20

GCA -0.46 -0.44 -0.51 -0.69 0.20 0.53 0.04 0'24 0'21 0.16 0.11 0.61

S.E. of a GCA estimate 0,035. Parental mean, 5.73; hybrid mean, 5-58; mean potence, -0.15.

Table 4. Mean seed fuzz grade of parents and crosses in the WRC F," A(61)6 BP52 Reba Barhop Coker

A(61)6 5'6 BP52 6.4 7'0

Barhop 5 *8 7'0 6.0 4.4 Coker 4'2 5 '0 5 '0 4'2 3'0

Reba 5'8 6.0 6.0

S.E. of a mean 0.18. * Parental values in italics.

General and specific combining abilities (Griffing, 1956) were calculated. The general combining abilities (Table 3) identify parents AH(67)1,2,3 and 4 as forming a low fuzz group. The difference between the mean of a cross and the mean of its two parent means has a standard error of (312)) times that of a single mean in Table I. Using this standard error, 0.15, all the crosses between parents AH(67) I, 2, 3 and 4 can be seen to be below their respective mid-parent values, some greatly so. By con- trast, other crosses such as 4 x 20 and 17 x 20 are significantly above their mid-parent values.

5 x 5 half diallel set, WRC F,. The mean values of seed coat fuzz of the five parents of the diallel (Table 4) range from 3.0 for Coker to 7.0 for BP 52. Although the mean fuzz grade of the (Coker x BP52)F2 falls exactly halfway between the parental value, in other Coker crosses, such as Coker x Barhop, the means fall below the mid-parental values. The general combining ability estimates (Table 5 ) serve to emphasize the value

Breeding for reduced seed coat fuzz in cotton I43

Table 5. General combining ability (GCA) estimates for jive WRC F2 parents and values for mean potence in 5 x 5 half diallel set

GCA 0’01 0.72 0’21 0.34 - 1.28 A(61)6 BP52 Reba Barhop Coker

S.E. of a GCA estimate 0.08. Parental mean, 5’59; hybrid mean, 5.54; mean potence -0.05.

Table 6 . Hayman analysis of variance for seed fuxx grade in 5 x 5 half diallel set of WRC F, crosses

Source D.F. M.S. V.R.

U

b bi bz b,

bl x blocks b, x blocks b, x blocks

(I x blocks b x blocks

4 I 0

I

4 5

16 40 4

16 20

+*, *++ P < 0.01, < 0.001, respectively.

8

I

6

5

s4

3

1 2 3 4 5 6 V,

Fig. I . Graph of W, against V, for seed fuzz grade in 5 x 5 half diallel WRC Fa (mean data over blocks). I , A(61)6; 2 , BP52; 3, Reba W296; 4, Barhop; 5 , Coker ~ o o W / z .

I44 N. L. INNES AND R. H. WIMBLE of Coker as a source of genes for reduced seed coat fuzz. The analysis of variance shows that additive parental differences were much the largest source of variation in the half diallel table but that, nevertheless, some non-additive effects were also present (Table 6). Since the parents were inbred, tests for an additive-dominance model (Mather & Jinks, 1971) were carried out. The graph of W, against V , after removal of environmental effects is given in Fig. I . The slope of the regression was very close to unity, consistently in all blocks. The analysis of variance of (W, + V,) gave significant differences between parents (P < 0.01) indicating dominance. The close proximity of the line to the limiting parabola indicates that dominance is relatively weak. Partial dominance is also indicated because the regression line intersects the W, axis above the origin. From Fig. I, it is concluded that parent 3, Reba, had the highest pro- portion of dominant genes and parent 4, Barhop, the least, as also shown by the means of the parents and crosses in Table 4. The absence of a significant correlation between parent means and (w + q) indicates that dominance is ambidirectional.

Table 7. Mean valus and ranges of seed fuxz grades by trial and data partition Trial

AH WRC F2 ALOA CA &&&- Mean Range Mean Range Mean Range Mean Range

Hybrid populations 5'58 4-7 5'54 4-7 6.28 5-8 6.00 5-8

S.E. o'org - 0.055 - 0'022 - 0'015 - Parents or controls 5'73 4-7 5'59 3-7 5'21 4-7 5.75 5-6 S.E. 0'035 - 0.078 - 0'034 - 0.050 -

or progenies

Correlations of seed fuxz grade with yield components and lint quality measurements Mean values and ranges of seed fuzz grades from each of the trials are given in

Table 7 and genotypic correlations of seed coat fuzz with yield components and lint quality measurements in Table 8. Environmental correlations were almost all neg- ligible. The genotypic correlations for parents and hybrids in each of the two diallels were, with few exceptions, very close. By contrast, they varied greatly from experiment to experiment. For example, seed fuzz grade and lint per seed were positively cor- related in the AH, ALOA and CA trials but negatively correlated in the WRC. In the CA trial there was evidence of a positive association between seed fuzz density and lint yield but the opposite was found in the WRC trial. The very large negative correla- tion of seed fuzz grade with micronaire value in the WRC F, trial was due mainly to the parent Coker, that had a large micronaire value, which reflects the maturity and coarseness of its lint, and little if any fuzz.

D I S C U S S I O N

The inheritance of seed coat fuzz in the WRC material is adequately accounted for by an additive-dominance model with a preponderance of additive genetic variance. The high negative correlation between lint yield and seed coat fuzz indicates that it should be possible to select heavy yielding lines with little fuzz as found in the Coker

Q &

> &

Table 8. Genotypic correlations of nine components of yield and lint quality with seedjum grade, by trial and data partition 2

ou - 2

\ L

7 A OQ Wt of I seed

rAH WRC ALOA C k

Seed cotton per boll Seeds per boll r 7

AH WRC ALOA CA AH WRC ALOA CA All entries 0'52 -0'42 0.42 -0.48 0.27 0.26 -0.05 -0.67 0.61 -0.55 0.41 0.34 Parents 0.58 -0.51 - - 0'21 0.31 - - 0.68 -0.66 - Hybrid populations or progenies 0.63 -0.30 0.15 -0.48 0.38 0'34 -0.40 -0.68 0.63 -0.40 0.48 0.35 3

Lint per seed Ginning % Lint yield f , r

AH WRC ALOA CA AH WRC ALOA CA 'AH w R c A ALOA c i -0.28 ' 0'53 0'5 I -0.29 0.13 0.42 0.42 -0.18 -0'47 -0.00 0.55 % All entries 0'39

Parents 0.52 -0.30 - Hybrid populations or progenies 0'43 -0.24 0.28 0.56 -0.31 0.18 -0.08 0.46 -0'12 -0'79 -0.37 0.60 R, - 3 - - 0 ' 3 5 -0'44 - - 0'21 0'12 -

All entries Parents Hybrid populations

Tensile strength - AH WRC 0.39 0'12 0.64 0.19 0.32 0.05

Staple length &

AH WRC 0.06 - 0.23

0'10 -0.24 0'00 -0.22

Micronaire value &

AH WRC - 0.09 0.92 - 0'00 0.96 -0.12 0.89

146 N. L. INNS AND R. H. WIMBLE parent. Unfortunately, among the varieties used in the wide range crossing programme there were considerable differences in resistance to local pests and diseases, as well as in lint quality. Pedigree selection had provided by the F, only a few lines that were likely to meet the necessary standard in terms of disease resistance, lint quality and lint yield. However, although there was no directed selection for reduced seed coat fuzz there were, among the more promising lines, low fuzz grade derivatives from crosses involving Coker.

In reviewing the literature on the effects of selection for reduced seed coat fuzz on lint production and quality, Costelloe (1970) emphasized that there was conflicting evidence regarding the feasibility of selecting for reduced fuzz without adversely affecting yield and lint quality. The inconsistent pattern of correlations between seed coat fuzz and other economic characters in the different populations stresses further the dangers of extrapolating results from one breeding population to another.

Within the AH material it has been possible to select heavy yielding lines with reduced fuzz and lint quality equal to, or better than, that of the parents and of the commercial variety, BPA. Moreover, after five successive generations of growing AH(67) lines there has been no reversion of tufted seeded types to a more fully fuzzed condition. These results are in contrast to those obtained by Arnold, Innes & Church (1969) and Arnold, Innes & Gridley (1971) in pedigree lines with tufted seed selected from BPA. Jones & Riggs (1969) also found it difficult, within Albar breeding stocks, to stabilize true breeding lines with little or no seed fuzz, but appeared to have little difficulty in establishing lines that bred true for moderate to heavy degrees of fuzz. These difficulties seem to have been overcome by Jones & Fielding (1971, 1973) who obtained a range of progenies covering the classes 2-8. Using such progenies it should be possible to resolve the controversy about the effect of seed fuzz on yield and lint quality within Albar breeding stocks.

There appears to be sufficient scope for further reducing fuzz in AH material, as shown by results from some of the AH crosses and the considerable contribution of additive genetic variance. The AH (67)1 to 4 group provides a source of genes likely to contribute to a further reduction in fuzz. It is encouraging to find that, although these four lines are unlikely to be accepted by farmers because of their small bolls, they out- yield BPA and have lint of similar quality (Innes, 1973). However, only more breeding work will show if a further reduction can be achieved without a concomitant adverse effect on other characters of importance. The level of reduced fuzz already obtained represents a very useful contribution and could lead to easier handling of seed, speedier and cheaper ginning, lower levels of seed coat nep and better germina- tion of seed provided bigger-bolled recombinants with the attributes of the AH(67) I to 4 group can be obtained by recurrent selection in the synthetic population derived from the AH diallel.

Breeding for reduced seed coat f u z z in cotton I47

REFERENCES

ARNOLD, M. H., INNES, N. L. & CHURCH, J. M. F. (1969). Progress Reports from Experimental

ARNOLD, M . H., I-, N. L. & GRIDLEY, H. E. (1971). Progress Reports from Experimental

COSTELLOE, B. E. (1970). Inheritance and effects upon lint of seed coat fuzz in Upland cotton.

GRIFFING, B. (1956). Concept of general and specific combining ability in relation to diallel

HAYMAN, B. I. (1954). The analysis of variance of diallel tables. Biometrics 10, 235-24. INNES, N. L. (1973). Promising selections from inter-varietal crosses at Namulonge. Cotton

Growing R e v i m 50, 296-306. INNES, N. L., WIMBLE, R. H. & GRIDLEY, H. E. (1975). Estimates of genetic parameters for lint

quality in Upland cotton (Gossypium hirsutum). Theoretical and Applied Genetics 46,249-256. JINKS, J. L. (1954). The analysis of continuous variation in a diallel cross of Nicotiana rustica

varieties. Genetics 39, 767-788. JONES, G. B. & FIELDING, J. (1971). Progress Reports from Experimental Stations, Uganda, 1970-71. Cotton Research Corporation, London.

JONES, G. B. & FIELDING, J. (1973). Cotton Research Reports, Uganda, 197192. Cotton Research Corporation, London.

JONES, G. B. & RIGGS, T. J. (1969). Progress Reports from Experimental Stations, Uganda, 1968-69. Cotton Research Corporation, London.

Low, A. (1962). Progress Reports from Experimental Stations, Uganda, 1961-62. Cotton Research Corporation, London.

Low, A. (1968). Developments towards tufted seed in varieties of Gossypium hirsutum. Cotton Growing Review 45, 101-113.

MATHER, K. & JINKS, J. L. (1971). Biometrical Genetics. London: Chapman and Hall. NELDER, J. A. (1975). Genstat Reference Manual. Inter-University/Research Council Series,

Stations, Uganda, 1968-69. Cotton Research Corporation, London.

Stations, Uganda, 1970-71. Cotton Research Corporation, London.

Cotton Growing Review 47, 8-19.

crossing systems. Australian Journal of Biological Sciences 9, 462-493.

Report No. 3, 3rd Edition.