Crop Science 41:345-350 (2001)
© 2001 Crop Science Society of America
CROP BREEDING, GENETICS & CYTOLOGY
Comparative Efficiency of Two Breeding Methods for Yield and Quality in Rice
Dimitrios A. Ntanosa and
Demetrios G. Roupakiasb
a National Agric. Res. Foundation, Cereal Institute, Thermi-Thessaloniki, 570 01 Greece
b Dept. of Genetics and Plant Breeding, School of Agriculture, Aristotelian Univ. of Thessaloniki, 540 06, Greece
Corresponding author (cerealin{at}mail.otenet.gr)
 |
ABSTRACT
|
|---|
A rice (Oryza sativa L.) breeder must choose a selection method that facilitates the simultaneous improvement of yield and quality. This study was conducted to compare the effectiveness of honeycomb selection (HCS) and panicle-to-row selection (PRS) in two F2 rice populations. Both populations were advanced to the F6 generation by both methods. Five F6 lines with high yield potential and good grain quality were selected from each population and selection method and were tested with a check cultivar (Strymonas) in a randomized complete block design for 2 yr near Kalochori, Thessaloniki, in Greece. In population GW 1992, significant differences in grain yield occurred between lines derived from the two methods as well as between selected lines and the check. In GW 2002, the differences were significant only between the HCS-derived lines and the check. In both populations, significant genotypic differences were attributed to selection methods, and to methods vs. the check, for all quality traits studied except for total milling yield. The average superiority of the HCS method over the PRS method was 6 (GW 1992) and 5% (GW 2002) for grain yield, 18 and 9% for grain vitreosity, 1% (both populations) for grain length, and 3 and 2% for grain length/width ratio. These results indicate that honeycomb selection for yield and quality applied during generations was more effective than panicle-to-row selection applied in later generations.
Abbreviations: CPS, Conventional pedigree selection • HCS, Honeycomb selection • PRS, Panicle-to-row selection
|
INTRODUCTION
|
|---|
COMBINED IMPROVEMENT for high yield and quality is the aim of all rice breeders. A high yielding rice cultivar will not be accepted by the rice miller or the consumer unless its quality is acceptable. To achieve this goal, a rice breeder should choose a breeding method which will facilitate the simultaneous improvement of these traits.
The different approaches to selecting in rice segregating populations are focused on the generation during which selection for yield is first applied and the potentially confounding role of plant competition. Nagai (1962) pointed out that rigorous pedigree selection in rice in early generations might result in the loss of desirable genotypes which could be selected in later generations as homozygotes. Sakai (1951) mentioned that competition between individuals in a segregating population is considerable and results in the reduction of selection reliability. Thus, he considered that bulk selection might be preferred to pedigree selection. De Pauw and Shebeski (1973) pointed out that single-plant selection in early generations was effective for qualitative traits in wheat (Triticum aestivum L.) and ineffective for quantitative traits such as yield. The same was confirmed in barley (Hordeum vulgare L.) (Hanson et al., 1979). Simmonds (1979) reported that classical pedigree selection was effective in early generations (under competition) only for traits with high heritability such as grain size. In contrast, McKenzie and Lambert (1961) and Sneep (1977) suggested that selection could be applied for high-yielding genotypes in the F2 and successive segregating generations. Indeed, Mitchell et al. (1982), Lungu et al. (1987), Roupakias et al. (1997), and Batzios (1997) reported successful selection for yield in F2 populations of durum wheat (Triticum turgidum L. var. durum), spring wheat, faba bean (Vicia faba L.), and cotton (Gossypium hirsutum L.), respectively, when the plants were grown in a honeycomb design. Fasoulas (1988)(and 1993) predicted that honeycomb pedigree selection at low plant density in an F2 population should be effective for both yield and quality traits.
The objective of this study was to evaluate the effectiveness of two selection methods, honeycomb selection and panicle-to-row selection, for yield and four quality traits (total milling yield, grain vitreosity, grain length and grain length/width ratio) in two F2 rice populations.
|
MATERIALS AND METHODS
|
|---|
The materials used were the F2 generation of two rice hybrids, GW 1992 and GW 2002, maintained by Golden West Seed Co. Inc., Oregon. Field experiments were conducted at the Experimental Station of the Cereal Institute in Kalochori, Thessaloniki (40°33' N lat, 23°00' E long, 0 m alt). The soil was silty loam (Aquic Xerofluvents) with a pH of 7.5 and 1.6% organic matter.
Honeycomb Selection
During the spring of 1989, 1607 F2 plants from GW 1992 and 963 F2 plants from GW 2002 were grown in two (Fig. 1)
unreplicated blocks with one check in a honeycomb design (Fig. 3.26 in Fasoulas and Fasoula, 1995). Seeds were soaked in water for 24 h before sowing. In each hill, five seeds were sown by hand. After seedling establishment, the plants were thinned to one plant per hill. The plant-to-plant spacing was 100 cm. One day before sowing the soil was flooded with no more than 2 cm (in depth) of water. The water level was increased progressively following seedling establishment and maintained at about 10 cm (in depth) until 15 d before harvest. The field was fertilized with 75 kg N ha-1 (in three increments), 17 kg P ha-1, and 31 kg K ha-1, which was applied by hand broadcasting. The first 25 kg ha-1 of N and the whole amount of phosphorus and potassium were applied before sowing. The second increment of N (30 kg ha-1) was applied at the early beginning of tillering stage and the third one (20 kg ha-1) prior to panicle initiation. The field was kept free of weeds by hand hoeing. Fungicides and insecticides were not needed.
View larger version (44K):
[in this window]
[in a new window]
|
Fig. 1. Flow diagram illustrating honeycomb selection (HCS) and panicle-to-row selection (PRS) in the F2 through F6 generations in two rice populations, GW 1992 and GW 2002
|
|
In each honeycomb design, seed yield was determined individually for each plant. For selection, the center of a moving grid comprising 19 plants in the GW 1992 population and seven plants in the GW 2002 population was moved from plant to plant. A particular plant was selected if its yield exceeded the yield of the remaining plants within the grid (5.3% selection pressure in the GW 1992 population and 14.3% in the GW 2002 population). Border plants were evaluated with a lower selection pressure since the moving circle was incomplete. Thus, 79 F2 plants were selected from GW 1992 and 113 F2 plants from GW 2002. These plants were further evaluated for four quality traits: total milling yield, grain vitreosity, grain length, and grain length/width ratio.
Twenty eight F2 plants from each cross were advanced to the F3 generation on the basis of high yield and acceptable quality. The F3 progenies of the 28 F2 plants were planted separately for each cross as F2:3 families together with a check, Strymonas, (as three entries) in two replicated (R-31) honeycomb designs (one per population) with 31 replications (Fig. 2)
. Strymonas is a locally adapted, long-grain Japonica-type rice cultivar combining satisfactory grain yield and good quality traits. Selection in the F3 conducted according to procedures used in the F2, except that selection pressure applied in both F3 populations was 14.3%. Selection pressure in the F4 and F5 generations was 5.3% in both populations (Fig. 1). Finally, the best five F5:6 lines, determined on the basis of yield and quality traits, were selected from each population for further evaluation. The HONEY microcomputer program (Batzios and Roupakias, 1997) was used for individual plant selection in each cycle.
Panicle-To-Row Selection
From each honeycomb design of the F2 populations, one panicle per selected plant was collected, and seeds were planted in a single 1.0- by 0.4-m row containing 20 to 25 plants. Thus, 1567 F2:3 families were seeded from population GW 1992 and 810 from GW 2002. Cultural practices were similar to those described for the honeycomb design. In all generations, negative selection was applied for late maturity, disease susceptibility, and long awns. From each of the remaining rows, seed from one main-stem panicle from the middle of the row were collected and seeded in a row the following year. As a result, 1333 F3:4 families were grown from GW 1992 and 673 from GW 2002. In the F5 generation, the number of families was reduced to 892 in GW 1992 and 568 in GW 2002 and for every 10 rows, a row of the check Strymonas was seeded. Yield of each family was expressed as a percentage of a moving average (Knott, 1972). Finally, 15% of the rows with the highest yield (compared with the moving average) from each population were selected and advanced to the next generation. Thus, in the F6 generation, 135 F5:6 families from GW 1992 and 85 from GW 2002 were seeded and evaluated as in the F5 generation described above. Families which exceeded the yield of the moving average by at least 10% were selected. Thus, 40 F6 families were selected from GW 1992 and 23 F6 families from GW 2002. These families were further evaluated for quality traits. From each population, the best five F5:6 lines with high yield potential and good quality traits were selected for further evaluation.
Measurement of Quality Traits
All measurements were carried out on grain with a moisture content of 140 g kg-1. The total milling yield (head and broken rice) was estimated from one sample for the individual plants during the selection procedure and from two samples (100 g) of cleaned rough rice from the selected lines under evaluation. After cleaning, the grain moisture content was determined by a Steinlite moisture meter (Fred Stein Laboratories, Inc., Atchison, KS) and the samples were then dried until their moisture reached 140 g kg-1. The samples were hulled by a mill of Olmia type and their milling yield was determined according to the standard procedure for rice grading. The grain vitreosity was estimated on two samples of 50 milled grains. For this, the grains were placed on a glassy table lit with a 60-W electric lamp. Grains with short spots of pearl were considered as chalky. The traits were expressed in percent. The grain length and width were evaluated from two random samples of 50 milled grains each with a micrometer, and the length/width ratio was calculated.
Evaluation of Selected Lines
Two field experiments (one per population) were carried out for 2 yr (1994 and 1995) in a randomized complete block design with four replications. Each experiment included the best five lines selected from each selection method and the rice cultivar Strymonas as a check. Rice was seeded directly in the soil by hand at a rate of 5 g/m in plots of six rows, 4 m long and 25 cm apart. The two external rows of each plot were not harvested. Yield, total milling yield, grain vitreosity, grain length, and grain length/width ratio were measured.
Orthogonal comparisons were used to compare the groups of the best five lines selected by each selection method as well as the members of each group among themselves and with the check variety (Steel and Torrie, 1980).
|
RESULTS AND DISCUSSION
|
|---|
Significant differences were observed among the selected lines for grain yield in population GW 1992 and for quality traits in both populations (Table 1). Significant differences in all traits were also observed from year to year in population GW 1992. In GW 2002, however, the differences from year to year were significant only for grain yield and total milling yield (data not shown). In addition, year x selected lines interaction was significant for all the traits in both populations.
View this table:
[in this window]
[in a new window]
|
Table 1. Selected mean squares from the analysis of variance of 10 lines derived by two selection methods from two rice populations, and for one check cultivar, tested in one location for 2 yr in Greece
|
|
Grain Yield
Across years the grain yield of the HCS-selected lines was significantly higher than that of the PRS-selected lines and the check cultivar in population GW 1992. The differences were significant only between the HCS-selected lines and the check in population GW 2002 (Tables 1 and 2). Furthermore, four of the five HCS lines and two of the five PRS lines in population GW 1992, and one line from each method in the population GW 2002, yielded significantly more than the check cultivar (Tables 3 and 4). In GW 1992, the best two HCS lines, namely No. 2 and 5, yielded significantly more than two PRS lines. Nevertheless, the best PRS lines (No. 3 and 5) yielded on average 96.0% of the grain yield yielded by the best two HCS lines (No. 2 and 5). In GW 2002, although results of the analysis of variance did not show overall differences among lines in grain yield (Table 1), mean comparisons revealed that one PRS line, namely No. 5, was inferior compared with four HCS lines (No. 1, 2, 3, and 4) and two PRS lines (No. 1 and 2). Nevertheless, the best PRS line (No. 1) yielded 101.4% of the yield produced by the best HCS line (No. 3).
View this table:
[in this window]
[in a new window]
|
Table 2. Means of selected lines derived from honeycomb (HCS) and panicle-to-row (PRS) selection methods for yield and four quality traits in two rice populations
|
|
View this table:
[in this window]
[in a new window]
|
Table 3. Means for grain yield and four quality traits in 2 yr of lines selected from the population GW 1992 by HCS and PRS
|
|
View this table:
[in this window]
[in a new window]
|
Table 4. Means for grain yield and four quality traits in 2 yr of lines selected from the population GW 2002 by HCS and PRS
|
|
Grain Quality Traits
Across years, the lines selected by the HCS had significantly higher values in three of the four quality traits compared with the PRS lines and the check in both populations (Tables 1 and 2). For the same traits, the PRS lines had significantly higher values than those of the check cultivar in both populations. No significant differences were observed in total milling yield between the group of HCS lines and PRS lines, or between these groups and the check cultivar, in either population (Tables 1 and 2). This could be due to less total genetic variation from which to select for total milling yield as compared with the other quality traits.
In GW 1992, the two HCS lines with the best yielding ability, namely No. 2 and 5, had also significantly higher values in three quality traits (grain vitreosity, grain length and grain length/width ratio) than the best PRS line (No. 3) (Table 3). In addition, the PRS line No. 3 had undesirably low grain vitreosity value (34.3). In GW 2002, the highest-yielding HCS line (No. 3) had significantly higher values in total milling yield, grain length, and grain length/width ratio than the highest-yielding PRS line (No. 1) (Table 4). Thus, among the HCS and PRS lines selected from GW 2002, the HCS line No. 3 was the most appropriate one as a potential cultivar.
On the basis of grain yield, the number of lines exceeding the yield of the check, grain quality traits, and the number of lines with good yielding ability and good grain quality, the HCS method showed superiority over the PRS method for both populations. Similar results have been reported by Gill et al. (1995) for selection in mungbean [Vigna radiata (L.) Wilczek]. Furthermore, four of the five F6 HCS lines selected from GW 1992 originated from one F2 plant and the five F6 HCS lines selected from GW 2002 originated from three F2 plants. In contrast, in both populations, all five F6 PRS lines originated from different F2 plants. None of the selected F6 PRS lines originated from the same F2 plants from which the selected HCS lines were obtained. This, together with the better performance of the HCS lines, indicates that HCS for yield and quality traits was effective in selecting superior plants in the F2 and subsequent segregating generations. Effective early generation HCS has been reported by other workers (Lungu et al., 1987; Roupakias et al., 1997; Batzios, 1997).
Early generation selection for yield has generally been found to be ineffective by conventional pedigree selection (McGinnis and Shebeski, 1968; Knott, 1972; DePauw and Shebeski, 1973; Hanson et al., 1979). Theoretically, panicle-to-row selection should be more effective than conventional pedigree selection. Useful variability is maintained from generation to generation since only negative selection is carried out in the segregating generations and selection for yield is applied when the materials are in advanced homozygosity. In this study the lines selected by PRS were inferior to the ones selected by HCS. This might indicate that the best genetic combinations are not guaranteed in later generations when the materials are advanced by one panicle and negative selection. If this is true, then one could repeat the statement made by Valentine (1979) "if yield is to be maximized, no opportunity for selection in early generations should be lost." Thus, HCS is favored because it is the only method reported so far to be effective in early generation selection for yield (Lungu et al., 1987; Roupakias et al., 1997; Batzios, 1997).
The advantage of HCS over PRS in selecting better lines might be attributable to methodological differences between them. First, all F2 plants in PRS are advanced to the F4 generation via a single panicle and negative selection for yield, while selection for yield begins in the F5 generation. In contrast, HCS selection begins in the F2 and is based on individual plant performance for yield in an environment of low plant density (Fasoula and Fasoula, 1997a). The second difference is related to the allocation of genotypes. With PRS, families were planted in rows, while in HCS, single plants were planted in an equilateral triangular lattice pattern (Fig. 2, and Fig. 9.13 in Fasoula and Fasoula, 1997b) and therefore plants of a family were distributed evenly within the experimental area. Finally, the selection pressure that is applied with HCS, beginning with the first generation (F2), is relatively higher. Early-generation selection, combined with individual-plant performance for yield and application of the moving circle (Fasoulas and Fasoula, 1995) during selection, apparently led to more effective selection with HCS method. It could be argued, however, that the advantages of honeycomb selection are limited because of greater demands for time, labor, and land, particularly in the earlier generations. According to our preliminary estimation, this additional demand does not exceed 20% and therefore, it is not considered a serious limitation to the use of HCS in rice breeding programs.
Received for publication November 11, 1999.
|
REFERENCES
|
|---|
- Batzios, D.P. 1997. Effectiveness of two selection methods in the breeding of cotton (Gossypium hirsutum L.). Ph.D. diss. Aristotelian Univ. of Thessaloniki, Greece.
- Batzios, D.P., and D.G. Roupakias. 1997. HONEY: A microcomputer program for plant selection and analyses of the honeycomb designs. Crop Sci. 37:744–747.[Abstract/Free Full Text]
- De Pauw, R.M., and L.H. Shebeski. 1973. An evaluation of an early generation yield testing procedure in Triticum aestivum L. Can. J. Plant Sci. 53:465–470.
- Fasoula, D.A., and V.A. Fasoula. 1997a. Competitive ability and plant breeding. Plant Breed. Rev. 14:89–138.
- Fasoula, D.A., and V.A. Fasoula. 1997b. Gene action and plant breeding. Plant Breed. Rev. 15:315–374.
- Fasoulas, A.C. 1988. The honeycomb methodology of plant breeding. Dep. of Genetics and Plant Breeding, Aristotelian Univ. of Thessaloniki, Greece.
- Fasoulas, A.C. 1993. Principles of crop breeding. Dep. of Genetics and Plant Breeding, Aristotelian Univ. of Thessaloniki, Greece.
- Fasoulas, A.C., and V.A. Fasoula. 1995. Honeycomb selection designs. Plant Breed. Rev. 13:87–139.
- Gill, J.S., M.M. Verma, R.K. Gumber, and J.S. Brar. 1995. Comparative efficiency of four selection methods for deriving high-yielding lines in mungbean [Vigna radiata (L.) Wilczek]. Theor. Appl. Genet. 90:554–560.
- Hanson, P.R., G. Jenkins, and B. Westcott. 1979. Early generation selection in a cross of spring barley. Z. Pflanzenzuchtg 83:64–80.
- Knott, D.R. 1972. Effects of selection for F2 plant yield on subsequent generations in wheat. Can. J. Plant Sci. 52:721–726.
- Lungu, D.M., P.J. Kaltsikes, and E.N. Larter. 1987. Honeycomb selection for yield in early generations of spring wheat. Euphytica 36:831–839.
- McGinnis, R.C., and L.H. Shebeski. 1968. The reliability of single plant selection for yield in F2. p. 410–415. In K.W. Finlay and K.W. Shephero (ed.) Proc. 3rd Int. Wheat Genet. Symp., Canberra. 1968. Australia.
- Mckenzie, R.I.H., and J.W. Lambert. 1961. A comparison of F3 lines and their related F6 lines in two barley crosses. Crop Sci. 1:246–249.[Free Full Text]
- Mitchell, J.W., R.J. Baker, and D.R. Knott. 1982. Evaluation of honeycomb selection for single plant yield in durum wheat. Crop Sci. 22:840–843.[Abstract/Free Full Text]
- Nagai, I. 1962. Japonica rice its breeding and culture. Yokendo LTD, Tokyo.
- Roupakias, D., A. Zesopoulou, S. Kazolea, G. Dalkalitses, A. Mavromatis, and T. Lazaridou. 1997. Effectiveness of early generation selection under two plant densities in faba bean (Vicia faba L.). Euphytica 93:63–70.
- Sakai, K. 1951. Studies on individual selection and selective efficiency in plant breeding. Jap. J. Breed. 1:1–9.
- Simmonds, N.W. 1979. Principles of crop improvement. Longman, London.
- Sneep, J. 1977. Selecting for yield in early generation of self-fertilizing crops. Euphytica 26:27–30.
- Steel, R.G.D., and J.H. Torrie. 1980. Principles and procedures of statistics: A biometrical approach. 2nd ed. McGraw-Hill, New York.
- Valentine, J. 1979. The effect of competition and method of sowing on the efficiency of single plant selection for grain yield, yield components and other traits in spring barley. Z. Pflanzenzuchtg 83:193–204.
TOP
ABSTRACT
INTRODUCTION
