Combining Ability in the Rice Lines Selected for Direct-seeding in Flooded Paddy Field
Jong Gun Won and Tomohiko Yoshida
(Faculty of Agriculture, Kyushu University, Fukuoka 812-8581, Japan)
Abstract : To estimate the general and specific combining ability for direct seeding in flooded paddy field, we completed a 4 x 3 factorial crossing (design II) in 1998, and evaluated the 12 F1 hybrids and their 7 parents, including selected lines for yield, yield
components and culm traits in 1999. Generally, parents with good combining ability produce good lines in later generations. Therefore, the basic information about the general combining ability (GCA) and specific combining ability (SCA) is valuable for
breeding cultivars. The effect of GCA was significant for culm length, culm thickness, panicle length, 1000-grain weight and grain yield, which indicated the importance of the additive effects of the genes for these traits. The effects of dominant genes were
significant for culm length, culm thickness and grain yield, indicating that these traits were also controlled by nonadditive gene actions. Mid-parent heterosis ranged from −16.3% to 37% with an average value of 17.3% for grain yield, and from −1.7% to
30.0% with an average value of 13% for culm length, and from −6.1% to 23.9% with an average value of 4.2% for culm thickness. Some parents with positive GCA for yield, culm thickness and panicle length were identified. They would be useful for developing
cultivars adapted to direct-seeding in flooded paddy fields.
Key words : Combining ability, Design II, Direct-seeding, Heterosis, Rice.
Corresponding author: J.G. Won ( Kyongbuk Provincial ATA, 200 Dongho-dong, Buk-gu, Taegu 702-320, Korea).
Abbreviations : GCA, general combining ability ; SCA, specific combining ability.
Combining ability implies the capacity of a parent to produce superior progenies when crossed with another parents. In breeding programs, the information on heterosis and the combining ability of parents and crosses is very important. By analyzing the
combining ability and estimating the degree of heterosis, clues to the nature of gene action, desirable parents and important yield traits may be found (Can et al., 1997). Knowledge of the underlying gene action within the breeding population is a key to
selecting procedures that will maximize the gain in yield. If the additive action of genes is predominant in a self-pollinated species such as rice, the breeder can effectively select the lines at various levels of inbreeding, because the additive effects are readily
transmissible from one generation to another (Gravois and McNew, 1993). Vasal et al. (1993a, b) reported that information on the combining ability and heterosis pattern of CIMMYT`s maize germplasm was of great value to maize breeders worldwide for hybrid
development. In the case of rice, Gravois and McNew (1993) reported that the general combining ability (GCA) was more important than specific combining ability (SCA) in the U.S. southern long-grain rice, accounting for 70% of the variation in yield, 89% in
plant height, 84% in panicle number, 60% in panicle length, and 69% in panicle weight. Xu and Shen (1991) also reported that additive effects were more important than dominance in determining rice tiller number. Murai and Kinoshita (1986) estimated the
general and specific combining abilities for rice yield in Japan, and Kang et al. (1997) estimated those for germinability of rice at a low temperature in Korea.
Currently, the direct-seeded rice has received considerable attention in Japan, Korea and Southeast Asia. In the cultivation by direct seeding in flooded paddy field, because the seeds are sown directly onto the surface of paddy soil, the rice plant lodges
more easily than in the cultivation by other methods. In one study, the lodged semidwarf rice had a 35% reduced yield relative to the plants held erect (IRRI, 1986). Ando (1995) considered that the lodging tolerance and high seedling establishment are the most
important characters for direct seeding. Therefore, development of cultivars tolerant to lodging is very important for yield stability in direct-seeded rice. Miyasaka (1970) concluded in his report that root strength at ripening stage was closely correlated not only
with root lodging but also with culm-breaking lodging. Won et al. (1998) studied the relationship among the culm thickness, root thickness and pushing resistances and reported that the culm thickness had a strong positive correlation with root thickness and
pushing resistance in direct-seeded rice. In some reports, the available genetic resources for direct-seeding were selected by several characters such as low-temperature germinability, root-lodging tolerance, short and strong culm, and panicle-weight type
(Ando, 1995 ; Park et al., 1995). Kim and Vergara (1990) also reported that the ideotype of direct-seeded rice is the panicle-weight type with large panicles. It is thus very important to get some information on the gene action for these characters in order to
develop cultivars for direct-seeding. However, estimation of the combining ability and heterosis in direct-seeded rice is lacking. The objective of this study was, therefore, to analyze the GCA and SCA for yield, yield components and culm traits in several rice
cultivars direct-seeded in flooded paddy field, and to identify promising parents with good general combining ability for use in breeding programs for direct-seeded rice.
Materials and Methods
In the summer of 1998, one 4 x 3 factorial crossing (Design II mating scheme; Comstock and Robinson, 1948) was completed using seven parents. For the complete diallel design, the same parents were used as males and females, but for factorial
crossing (design II) different sets of parents were used as males and females (Table 1). Four lines selected for direct-seeding (L 38, L 42, L 53 and L 76) mainly by lodging tolerance, yield and grain quality (Won et al., 1998) were used as the female parent, and
three cultivars, Ansanbyeo, Hinohikari and Lemont, as the male parent. The resulting 12 F1 hybrids of each cross and seven parents were used in an experiment conducted in 1999 at the experimental field of Kyushu University (lat. N 33.5, long. E 130.4). The
seeds of F1 hybrids and parents were sterilized with 70% ethanol for 30 seconds and 2% sodium hypochloride for 15 minutes successfully, and then soaked in water at 25℃ for 2 days. The pre-sprouted seeds were directly drill-seeded by hand on the surface of the soil withstanding water after the plots were thoroughly puddled and leveled on May 20. After seeding, the plots were irrigated with 2-3 cm water depth continuously and the water was once drained for the root anchorage at 10 days after seeding. There were two replications. Each plot consisted of one 1m row with a 20cm space between rows and 10 seeds were sown in each row.
The total fertilizer applied was, 8, 8 and 8 g m-2 of N, P2O5 and K2O. A half of it was applied at sowing as the basal fertilizer, 30% was top-dressed at the tillering stage and 20% at the panicle initiation stage. Prior to harvest, 10 culms from five plants in the
middle of each row were measured and averaged for culm length, culm thickness, panicle length and panicle number. The culm thickness was measured at the part of 5cm above the soil surface including the leaf sheath. At harvest, five plants at the center of
each row were hand-harvested to determine the 1000-grain weight and single-plant grain yield. Grain yield and 1000-grain weight were measured with head brown rice.
The data were analyzed using the analysis of variance. Variations due to 12 crossings were divided into those due to female parents, male parents and female x male interaction. The expected variation due to female and male parents corresponds to the
general combining ability, and that due to the female x male interaction corresponds to the specific combining ability (Hallauer and Miranda, 1981). Because there are two sets of parents in design II, there are two independent estimates of GCA. Appropriate F-tests
were made to test for the differences among males and among females and for the interaction between males and females. The percentage of the sum of squares for the crosses attributable to GCA and SCA for each trait (Table 4) was calculated as follows;
Sum of squares: percentage for male (%) = [SSm /(SSm+SSf+SSmf)] x 100
Sum of squares: percentage for female (%) = [SSf /(SSm+SSf+SSmf)] x 100
Sum of squares: percentage for male x female (%) = [SSmf /(SSm+SSf+SSmf)] x 100;
where SSm is the sum of squares for male, SSf is the sum of squares for female, and SSmf is the sum of squares for male x female in Table 3.
Both mid-parent heterosis and high-parent heterosis were calculated as follows;
Mid-parent heterosis (%) = [ (F1−MP)/MP ] x 100
High-parent heterosis (%) = [ (F1−HP)/HP ] x 100
where F1 is the performance of the hybrid, MP is the average performance of parents and HP is the performance of the higher parent.
The culm length of parents ranged from 66.2 (Ansanbyeo) to 86.4 cm (L 42) and that of hybrids from 75.1 (L 53 x Ansanbyeo) to 102.2 cm (L 42 x Lemont) (Table 2). The mean culm length was 76.3 cm for parents and 85.6 cm for hybrids. Culm thickness
of parents ranged from 5.7 (Ansanbyeo) to 7.4 mm (L 53) and that of hybrids from 5.8 (L38 x Ansanbyeo) to 8.3 mm (L 42 x Lemont). The mean culm thickness was 6.5 mm for parents and 6.8 mm for hybrids. Grain yield of parents ranged from 15 (L 53) to 26.2 g (L 42) and that of hybrids from 17 (L 38 x Hinohikari) to 31.1 g (L 42 x Lemont). The mean grain yield was 19.8 g for parents and 23.4 g for hybrids. Generally, the mean value for the hybrid was higher than that for the parents except for the number of panicles. The grain yield seemed to be somewhat high in this study because it was the yield of F1 seeds which showed the hybrid vigor and the experimental plot size was small. However, the tendency was not so different among the entries.
Table 3 shows the results of the analysis of variance for crosses, female parents, male parents and their interaction for all of the traits in this study. Mean squares of crosses were significant for all the traits except for the number of panicles. The mean squares for males and females were both significant for culm length, culm thickness, panicle length, 1000-grain weight and grain yield, indicating the importance of additive genetic effects. In males, the mean squares for all traits except for the number of panicles were significant and larger in magnitude than those in females, suggesting that the role of cytoplasmic inheritance for these traits was not so important in rice crosses (Gravois and McNew, 1993). The effects of dominant genes, deducible from the mean squares of male x female interaction, were highly significant for culm length, culm thickness, 1000-grain length and grain yield. This indicates that these traits were also controlled by nonadditive gene actions.
Table 4 shows the details of the percentages of the sum of squares for the crosses attributable to the general and specific combining abilities for the seven traits. About 82% of the variation in yield depending on the cross combination was attributable
to GCA and 18% to SCA. This was in agreement with the report by Gravois and McNew (1993) who used the U.S. southern long-grain rice. More than 90% of the sum of squares for the crosses was attributable to GCA for culm length, culm thickness, panicle length and spikelet number. Brim and Cockerham (1961) reported that the dominance was less important in self-pollinating species, and Gravois and McNew (1993) confirmed it in their study. However, SCA for some traits were also significant, indicating that nonadditive gene actions affected yield and other traits as well. Although additive gene actions accounted for a majority of the genetic variations among hybrids for most traits, a nonadditive gene action, as indicated by SCA, was sizable for panicle number (43%) and
stronger for 1000-grain weight (72%).
The GCA effects for grain yield were positive in Lemont, L 42 and L 76, indicating that the use of these three parents in direct-seeded rice breeding programs could be promising for higher grain yield (Table 5). Lemont, L 38 and L 42 exhibited
positive GCA effects for culm length, which might decrease the lodging tolerance in the crosses. GCA effects for culm thickness were positive in Lemont, L 42, L 53 and L 76, suggesting the importance of these parents for increasing the lodging tolerance in direct-seeded rice breeding programs. The GCA effects for panicle length were positive in Lemont, L 38 and L 42. Lemont showed positive GCA effects for spikelet number and L 42 showed positive GCA effects for panicle number and 1000-grain weight.
High and positive SCA effects for grain yield were observed in L 38 x Lemont, L 42 x Lemont, L 53 x Hinohikari and L 76 x Ansanbyeo (Table 5). These hybrids also showed positive SCA effects for panicle length and panicle number. Among them, hybrids of L 38 x Lemont and L 76 x Ansanbyeo showed negative SCA effects for culm thickness.
Mid-parent heterosis for yield ranged from −16.3% (L 42 x Ansanbyeo) to 37% (L 53 x Hinohikari), with an average value of 17.3% for all hybrids, and high-parent heterosis for yield was somewhat lower (Table 6). Crosses of L 42 x Lemont, L 53 x
Hinohikari and L 76 x Hinohikari showed large and positive high-parent heterosis value. Certain parents, L 76 and Lemont, produced large and positive yield heterosis. The heterosis for the culm length was positive in all hybrids except mid-parent heterosis of L
42 x Hinohikari, and a certain parent such as Lemont produced high heterosis value indicating the negative effect for the lodging. The mid-parent heterosis for culm length ranged from −1.7% (L 42 x Hinohikari) to 30% (L 42 x Lemont) with an average value
of 13%. The mid-parent heterosis for culm thickness ranged from −6.1% (L 53 x Ansanbyeo) to 23.9% (L 42 x Lemont) with an average value of 4.2%. Especially crossing combinations with parents of L 42 or Lemont produced good heterosis for culm
In spite of the long history of studies and advantages such as labor- and cost-saving, the area of direct-seeding cultivation has not become widespread because of the instability of the yield in Japan and Korea. One of the major reasons of the yield instability is lodging in direct-seeding, and no suitable cultivars for direct-seeding are yet available. Because of the segregation of the traits in early generations, breeders usually wait for later generations to obtain uniform lines, which takes a long time to see whether a cross combination will or will not produce superior lines. Good combining ability, which would be detectable in the F1 or F2 generations could be a reliable indicator for breeders. In the present study, the percentage of the sum of squares for the crosses attributable to GCA was higher than that attributable to SCA for yield, culm thickness, panicle length and spikelet number (Table 4). This shows the predominance of additive gene actions, and because additive effects are readily transmissible from one generation to another, the breeder can effectively select these traits at various levels of inbreeding. Therefore, the cultivars with high yield and lodging tolerance may be developed by using the cultivars with a high GCA identified in this study as cross parents. Development of cultivars adapted to direct-seeding will also contribute to the stability of the yield and increase the area of the direct-seeded rice.
Some investigators reported several characters such as germinability at a low temperature germinability, root lodging tolerance, short and strong culm, and panicle weight type as the available genetic resources for direct-seeding (Ando, 1995 ; Park et al., 1995). Kim and Vergara (1990) also reported that the ideotype of direct-seeded rice is the panicle weight type with large panicles, and Won et al. (1998) reported that the culm thickness had a strong positive correlation with lodging tolerance in direct-seeded rice. Therefore, in the present study, we tried to select the cultivars tolerant to lodging having a thick culm, and those with a large panicle and high yield, in priority. However, the parents selected for a particular trait were not always acceptable for other traits. Thus, the
simultaneous selection of rice lines for all traits is difficult. Lemont and L 42 were considered to be good combiners with a positive GCA for overall traits of grain yield, panicle length, culm thickness and other traits, although the positive GCA effects for
culm length were not desirable for lodging tolerance.
In conclusion, GCA was more important than SCA for grain yield, culm length, culm thickness, panicle length and spikelet number, which confirmed the additive gene action for a majority of the genetic variation. Lemont and L 42 were considered to be
good combiners for overall traits of grain yield, culm thickness and other traits, which could be useful for the breeding program of direct-seeding.
Ando, I. 1995. Rice breeding adaptable to direct seeding at NARC in Japan. Korea and
Japan Seminar on Direct Seeded Rice. 139−142.
Brim, C.A. and Cockerham, C.C. 1961. Inheritance of quantitative characters in
soybeans. Crop Sci. 18:533−536.
Can, N.D., Nakamura, S. and Yoshida, T. 1997. Combining ability and genotype x
environment interaction in early maturing grain sorghum for summer seeding. Jpn.
J. Crop Sci. 66:698−705.
Comstock, R.E. and Robinson, H.F. 1948. The components of genetic variance in
populations of biparental progenies and their use in estimating the average degree
of dominance. Biometrics 4:254−266
Gravois, K.A. and McNew, R.W. 1993. Combining ability and heterosis in U.S.
southern long-grain rice. Crop Sci. 33:83−86.
Hallauer, D.S. and Miranda, J.B. 1981. Quantitative Genetics in Maize Breeding. Iowa
State Univ. Press, Ames. 45−113.
IRRI 1986. Program Report for 1985. International Rice Research Institute, Los Banos,
Kang, J.R., Ko, M.S., Kim, H.Y., Lim, S.J. and Kim, S.C. 1997. Genetic analysis of low
temperature germinability of rice. Korean J. Crop Sci. 42:766−777.
Kim J. K. and Vergara B. S. 1990. Tillering behavior of low and high tillering rices.
Korean J. Crop Sci. 35 : 512 - 517.
Miyasaka, A. 1970. Studies on the strength of rice root. II. On the relationship between
root strength and lodging. Proc. Crop Sci. Soc. Japan. 39:7−14.
Murai, M. and Kinoshita, T. 1986. Diallel analysis of traits concerning yield in rice.
Jpn. J. Breed. 36:7−15.
Park, H.W., Moon, H.P. and Choi, I.S. 1995. Breeding of rice varieties for direct seeding
in Korea. Korea and Japan Seminar in Direct Seeded Rice. 143−156.
Vasal, S.K., Srinivasan, G., Pandey, S., Gonzalez, C.F., Crossa, J. and Beck, D.L.
1993a. Heterosis and combining ability of CIMMYT`s quality protein maize
germplasm: I. Lowland tropical. Crop Sci. 33:46−51.
Vasal, S.K., Srinivasan, G., Gonzalez, C.F., Beck, D.L. and Crossa, J. 1993b. Heterosis
and combining ability of CIMMYT`s quality protein maize germplasm: I.
Subtropical. Crop Sci. 33:46−51.
Won, J.G., Hirahara, Y., Yoshida, T. and Imabayashi, S. 1998. Selection of rice lines
using SPGP seedling method for direct seeding. Plant Prod. Sci. 1:280−285.
Xu, Y.B. and Shen, Z.T. 1991. Diallel analysis of tiller number at different growth
stages in rice (Oryza sativa L.). Theor. Appl. Genet. 83:243−249.
Table 1. Origins of the seven parents used in the 4 x 3 factorial crossing.
Table 2. Mean values for several traits in parents and F1 hybrid of rice.
1) Ansanbyeo, 2) Hinohikari, 3) Lemont.
* : Means followed by the same letter are not significantly different at 0.05 level
according to Duncan`s multiple range test.
** : fghij.
Table 3. Mean square values in analysis of variance of culm traits, yield and yield
components of direct-seeded rice.
*, ** : Significant at 5% and 1% level, respectively.
Table 4. Percentages of the sum of squares for the crosses attributable to general (GCA)
and specific combining ability (SCA) for the traits.
1) [SSm /(SSm+SSf+SSmf)] x 100,
2) [SSf /(SSm+SSf+SSmf)] x 100,
3) [SSmf /(SSm+SSf+SSmf)] x 100;
where SSm is sum of squares of Male, SSf is sum of squares of Female, and SSmf is sum
of squares of M x F in Table 3.
Table 5. Estimates of GCA* and SCA** effects for several traits of direct-seeded rice.
1) Ansanbyeo, 2) Hinohikari, 3) Lemont.
* : General combining ability.
** : Specific combining ability.
Table 6. Estimates of mid-parent and high-parent heterosis (%) for several traits of
1) Ansanbyeo, 2) Hinohikari, 3) Lemont.
* : Mid-parent heterosis calculated from 100 x [(F1− MP)/MP], where F1 indicates
performance of hybrid. MP indicates average performance of parents.
** : High-parent heterosis calculated from 100 x [(F1− HP)/HP], where F1 indicates
performance of hybrid. HP indicates average performance of higher parents.