Crop Science 42:1876-1881 (2002)
© 2002 Crop Science Society of America
CROP BREEDING, GENETICS & CYTOLOGY
Selection for Large Seed and High Protein in Two- and Three-Parent Soybean Populations
Brian J. Alt,
Walter R. Fehr* and
Grace A. Welke
Dep. of Agronomy, Iowa State Univ., Ames, IA 50011. Journal Paper No. J-19632 of the Iowa Agric. and Home Econ. Exp. Stn., Ames, IA
* Corresponding author (wfehr{at}iastate.edu)
 |
ABSTRACT
|
|---|
Soybean [Glycine max (L.) Merr.] cultivars with large seed size and high protein content are desirable for the production of tofu and other food products. The objectives of this study were to evaluate the amount of genetic variability and the effectiveness of single-plant selection for seed size and protein in two- and three-parent soybean populations. Two parents with large seed and high protein (LSHP) and one parent with normal seed size and protein (N) were used to produce three population types: LSHP x N, LSHP x LSHP, and LSHP x (LSHP x N). Four sets of the three population types were evaluated with different parents in each set. For each of the populations, 100 random F2 plants and 10 plants of each parent were harvested at Ames, IA, and their seed size and protein were measured. The F3 progeny of the 100 F2 plants and 10 entries of the three parents of each set were evaluated in replicated tests at two Iowa locations. The percentage of F2:3 lines with seed size greater than or equal to the smallest LSHP parent and protein greater than or equal to the lowest LSHP parent in the set averaged 4% for the LSHP x N, 88% for the LSHP x LSHP, and 31% for the three-parent crosses. Single-plant selection was not considered cost effective in the LSHP x N populations because the percentage of acceptable segregates was so small or in the LSHP x LSHP populations because the percentage of acceptable segregates was so high. For the three-parent populations, single-plant selection was most effective when the F2 plants were ranked for the two traits and those with the most favorable rank were selected for evaluation as F2:3 lines. LSHP x LSHP and three-parent populations should be the most useful for developing LSHP cultivars.
Abbreviations: LSHP, large seed high protein content • N, normal seed size and normal protein content
|
INTRODUCTION
|
|---|
LARGE SEED and high protein (LSHP) are desirable traits for soybean seed used in the production of some food products, including soymilk and tofu. The preferred levels of the two traits vary among breeders and food processors. In the breeding program at Iowa State University, a seed size of 
200 mg sd-1 and protein of 38% on a moisture basis of 130 g kg-1 are preferred for LSHP cultivars.
Three of the population types considered for breeding LSHP cultivars are the cross between a LSHP parent and one with normal seed size and protein (N), the cross between two LSHP parents, and a three-parent population in which the F1 from a LSHP x N cross is mated to a second LSHP parent. The N parents are conventional cultivars or lines that yield more than the LSHP parents. For improving the seed yield of LSHP cultivars, the preference would be to select in LSHP x N populations if they contained an adequate number of segregates with acceptable seed size and protein. Unfortunately, studies of segregation in soybean populations developed from parents with major differences in seed size or protein indicate that it was difficult to recover the seed size or protein content of either parent (Ting, 1946; Weber, 1950; Bravo et al., 1980; Cianzio and Fehr, 1982; Carpenter and Fehr, 1986; Cianzio and Fehr, 1987; Johnson et al., 2001). Two-parent populations from LSHP x LSHP crosses would be expected to produce a high percentage of lines with acceptable seed size and protein. The frequency of acceptable LSHP segregates from LSHP x (LSHP x N) crosses has not been reported.
It would be desirable to select among individual plants for seed size, protein, or both before their progeny are evaluated as lines in replicated tests. Johnson et al. (2001) indicated that selection among F2 plants for small seed size was not cost effective in crosses between small-seeded parents because the majority of segregates had acceptable seed size or between small-seeded and normal-sized parents because the frequency of small-seeded segregates was so low. They found that selection among F2 plants for small seed size in three-parent populations was effective for increasing the frequency of F2:3 lines with acceptable size. The effectiveness of single-plant selection for LSHP in segregating soybean populations has not been reported. The objectives of this study were to evaluate the genetic variability for seed size and protein content and the effectiveness of single-plant selection for LSHP in two-parent and three-parent populations.
|
MATERIALS AND METHODS
|
|---|
The eight LSHP parents from Iowa State University used in the study and their maturity groups were A97-774007 (II), A97-774049 (I), A97-774065 (II), A97-874007 (II), A97-874014 (II), A97-874052 (II), A97-874063 (II), and A97-974005 (II). The N parents were ‘IA2038’ and ‘IA2050’ of Maturity Group II from Iowa State University, ‘AP1995’ of Maturity Group I developed by AgriPro Seeds (Ames, IA), and ‘9306’ of Maturity Group III developed by Pioneer Hi-Bred International, Inc. (Des Moines, IA).
Four sets of two-parent and three-parent populations were developed from the parents (Tables 1 and 2)
. The LSHP x N crosses were made in March 1998 at the Iowa State University–University of Puerto Rico soybean breeding nursery at Isabela, Puerto Rico. The F1 seeds and the parents were planted in May 1998 at the Agronomy Research Center near Ames, IA. Pubescence color and molecular-marker analysis was used to confirm that the F1 plants were hybrids. The F1 plants from the LSHP x N crosses were used as the male parent in crosses to a second LSHP parent to form the three-parent populations. The LSHP x LSHP crosses also were made during the summer.
View this table:
[in this window]
[in a new window]
|
Table 1. Frequency for seed size of 100 F2:3 lines compared with their parents for three soybean population types in four sets.
|
|
View this table:
[in this window]
[in a new window]
|
Table 2. Frequency for protein of 100 F2:3 lines compared with their parents for three soybean population types in four sets.
|
|
The F1 seeds from the LSHP x LSHP and the three-parent crosses were planted at Isabela in October 1998 to obtain F2 seeds. The F1 plants were confirmed as hybrids by pubescence color or molecular-marker analysis. The F2 seeds from the F1 plants of each LSHP x LSHP cross were harvested in bulk. The F1 plants of the three-parent crosses were harvested individually.
The F2 seeds of all the populations and the parents were planted in May 1999 at the Agronomy Research Center near Ames. A total of 670 seeds from each two-parent population, 27 F2 seeds from each of 30 F1 plants from each of the three-parent populations, and 60 seeds of each parent were planted at 10 seeds m-1 in rows spaced 0.69 m apart. At maturity, plants of each parent and plants of Maturity Group II, the predominant maturity in the populations, were individually harvested. The seed size and protein content of 100 random F2 plants of the two-parent populations, four plants from each of 25 F1 families of the three-parent populations, and 10 plants of each of the parents were measured. Seed size expressed in milligrams per seed was determined by dividing the weight of all the seeds of a plant by the number of seeds. Protein content expressed on a 130 g kg-1 moisture basis was measured with a whole-grain near-infrared reflectance analyzer.
Four sets of F2:3 lines and parents were grown at the Agronomy Research Center and the Burkey Farm near Ames in 2000. Each set of 330 entries included the 100 lines derived from the individual plants that were evaluated for seed size and protein from each of the three population types and 10 entries of each of the three parents. Each set was a separate experiment grown in a randomized complete-block design with two replications at each location. The soil type at both locations is a Nicollet loam (Fine-loamy, mixed, superactive, mesic Aquic Hapludoll). Twelve seeds were planted in single-row plots 76 cm long with 102 cm between rows and a 107-cm alley between the ends of plots. The plots were harvested with a plot combine. Seed size for each plot was measured by weighing 400 random whole seeds. Protein content was measured on the same sample with the near-infrared reflectance analyzer. A missing plot value was calculated for all entries that did not have 400 seeds.
The data for each set of populations were analyzed separately for seed size and protein by the general linear model (GLM) procedure of the SAS software (SAS Institute, 1992). Locations, replications, and genotypes were considered random effects. The 10 entries of each of the three parents in a set were excluded from the analyses of variance.
For each cross, broad-sense heritability estimates on a single-plant basis were computed in standard units by the phenotypic correlation of the seed size or protein of F2 plants with the mean seed size or protein of their F2:3 lines over locations by the correlation (CORR) procedure of the SAS software (Frey and Horner, 1957; SAS Institute, 1992). Heritability estimates on a single-plant basis also were computed by regressing the mean seed size or protein of the F2:3 lines across locations on the seed size or protein of the F2 plants by means of the regression (REG) procedure of the SAS software (SAS Institute, 1992). Estimates of heritability on a plot and entry-mean basis were computed from variance component estimates for the F2:3 lines combined across the two locations (Hallauer and Miranda, 1995).
Within each set, frequency distributions were used to compare the three population types. For seed size, the F2 plants and F2:3 lines in each population were classified as larger than the largest LSHP parent, equal to the largest LSHP parent, greater than or equal to the smallest of the LSHP parents, or smaller than the smallest of the LSHP parents. For protein content, the F2 plants and F2:3 lines in each population were classified as greater than the LSHP parent with the highest protein, equal to the LSHP parent with the highest protein, greater than or equal to the LSHP parent with the second highest protein, and less than the LSHP parent with the second highest protein. An F2 plant value was considered equal to a parent if it was within ± two standard deviations (2 SD) of the parent mean, and the F2:3 line value was considered equal if it was within ± two standard errors of the mean (2 SEM).
The reliability of single-plant selection for seed size and protein was evaluated by computing the percentage of errors that occurred with selection for each trait. Acceptance and rejection errors were computed for all the populations with and without ranking the plants for a trait. Acceptance error for seed size or protein occurred when an F2 plant was equal to or greater than 
± 2 SD of the smallest or lowest protein LSHP parent in the set, but the line derived from that plant did not have a mean seed size or protein equal to or greater than the ± 2 SEM of the smallest or lowest protein LSHP parent in the replicated test. Rejection error for seed size or protein occurred when an F2 plant was smaller or lower in protein than the ± 2 SD of the smallest or lowest protein LSHP parent in the set, but the line derived from that plant had a mean equal to or greater than the ± 2 SEM of the smallest or lowest protein LSHP parent in the replicated test. For evaluating the reliability of single-plant selection when the rank of F2 plants was considered, the plants of the three-parent populations were ranked for seed size or protein, divided into groups of 10 from the largest size or highest protein to the smallest size or lowest protein, and the acceptance and rejection errors were computed for each decile. Ranking was not considered for the LSHP x N populations because the number of acceptable segregates for the two traits was so low or for the LSHP x LSHP populations because the majority of the segregates were acceptable.
|
RESULTS AND DISCUSSION
|
|---|
There were significant (P 
0.05) differences in seed size among the F2:3 lines within each of the 12 populations. The average percentage of F2:3 lines with seed size equal to or larger than the smallest LSHP parent in a set was 92% for the LSHP x LSHP, 41% for the three-parent, and 14% for the LSHP x N populations (Table 1). Transgressive segregation for seed size larger than the largest LSHP parent in a cross was observed in three of four LSHP x LSHP populations and averaged 8%. Transgressive segregation for larger seed size was observed in only one of the four crosses for each of the other population types.
Significant differences for protein content were observed among the F2:3 lines in each of the populations. The average percentage of lines with protein equal to or greater than the LSHP parent with the lowest protein in a set was 95% for the LSHP x LSHP, 66% for the three-parent, and 33% for the LSHP x N populations (Table 2). Transgressive segregation for greater protein than the highest LSHP parent in a cross was observed in each of the LSHP x LSHP, two of the LSHP x N, and one of the three-parent populations.
The phenotypic correlation coefficients between seed size and protein for the F2:3 lines ranged from 0.16 to 0.57 with a mean of 0.37. The positive correlation would favor the selection of LSHP segregates. The frequency of F2:3 lines that had seed size equal to or larger than the smallest LSHP parent and protein equal to or greater than the LSHP parent with the lowest protein averaged 88% for the LSHP x LSHP, 31% for the three-parent, and 4% for the LSHP x N populations (Table 3)
. Based on the same criteria for the two traits, the range in percentage of acceptable lines was 82 to 95% for the LSHP x LSHP populations, 19 to 44% for the three-parent populations, and 1 to 7% for the LSHP x N populations.
View this table:
[in this window]
[in a new window]
|
Table 3. Frequency of F2:3 soybean lines with acceptable seed size and protein for three population types in four sets.
|
|
The frequency of acceptable lines for seed size and protein in the three population types has a direct bearing on the evaluation of single plants to identify those most likely to meet the selection criteria in subsequent generations of testing. The frequency of acceptable lines was so high in the four LSHP x LSHP populations that the time and cost of evaluating single plants does not seem to be justified. Furthermore, many of the F2 plants rejected based on their seed size or protein were acceptable when evaluated more extensively as lines (Tables 4 and 5)
. For example, of the 24 F2 plants in the four LSHP x LSHP populations that were rejected for seed size, 18 (75%) of them should have been accepted based on their progeny performance. For the 376 F2 plants that were accepted for seed size, only 25 (7%) should have been rejected. Acceptance error was low for protein in three of the four LSHP x LSHP populations, and rejection error was high in the four populations (Table 5). Of the 370 F2 plants accepted for protein in the four populations, only 4% should have been rejected. Of the 30 F2 plants rejected, 25 (83%) of them were acceptable as F2:3 lines.
View this table:
[in this window]
[in a new window]
|
Table 4. Acceptance and rejection errors associated with selection for seed size among 100 F2 soybean plants based on the mean seed size of their F2:3 lines for each of 12 soybean populations.
|
|
View this table:
[in this window]
[in a new window]
|
Table 5. Acceptance and rejection errors associated with selection for protein among 100 F2 soybean plants based on the mean protein of their F2:3 lines for each of 12 soybean populations.
|
|
Single-plant selection in the LSHP x N populations may not be justified because the frequency of acceptable lines for seed size and protein was low and the acceptance error was high. On the average, only 4% of the lines were acceptable for the two traits (Table 3). Of the 167 F2 plants accepted for seed size from the four populations, 135 (81%) should have been rejected (Table 4). Of the total 267 plants accepted for protein, 181 (68%) should have been rejected (Table 5).
The effectiveness of single-plant selection for seed size and protein in the three-parent populations was evaluated by two procedures. The first procedure was to test the progeny of all plants that met the criteria for seed size and protein without considering the rank of plants for the traits (Tables 4 and 5). With that procedure, a total of 336 plants were accepted for seed size, of which 178 (53%) should have been rejected. Of the 64 plants rejected for seed size, 7 (11%) should have been selected. For protein, 356 plants were accepted, of which 115 (32%) should have been rejected. Of the 44 plants rejected, 21 (47%) should have been accepted.
The second procedure for single-plant selection in the three-parent populations was to rank the plants independently for seed size and protein, identify those that met the criteria on the basis of comparison with the parents, and advance for further testing only those of the highest rank, even though some plants of lower rank may have met the selection criteria (Tables 6 and 7)
. For the three-parent population in Set 3, all the lines with acceptable size were from plants that ranked in the top 40%, even though 78% of the F2 plants met the selection criteria. If the largest 40% of the plants were accepted in each of the populations, 70 (44%) of the 160 plants in the four populations should have been rejected, which is an improvement over the 53% acceptance error for the procedure in which ranking was not considered.
View this table:
[in this window]
[in a new window]
|
Table 6. Acceptance error associated with F2 plant selection for large seed size in three-parent soybean populations from four sets.
|
|
View this table:
[in this window]
[in a new window]
|
Table 7. Acceptance error associated with F2 plant selection for high protein in three-parent soybean populations from four sets.
|
|
Determining the percentage of the ranked plants to save for protein was more difficult than for seed size because there was not a distinct trend in the percentage of acceptance error for the 10 deciles (Table 7). Nevertheless, saving plants that ranked the highest for protein was an advantage over the first procedure. For example, by selecting plants that ranked in the top 50% of the populations for protein, 56 (28%) of the 200 plants saved in the four populations should have been rejected, compared with 32% acceptance error for the first procedure.
The heritability estimates observed for seed size and protein were comparable to those reported in other studies (Weber, 1950; Cianzio and Fehr, 1982; McKendry et al., 1985; LeRoy et al., 1991; Cober et al., 1997; Johnson et al., 2001). The range of phenotypic correlation coefficients among the 12 populations was 0.31 to 0.66 for seed size, all of which were significant (P 0.01). The range of correlation coefficients for protein was from 0.10 to 0.51, and only the coefficients of 0.10 and 0.16 were not significant. For the regression of F2:3 lines on the F2 plants, the range was 0.14 ± 0.04 to 0.50 ± 0.06 for seed size and 0.06 ± 0.06 to 0.44 ± 0.06 for protein. Heritability on a plot basis ranged from 0.20 ± 0.02 to 0.59 ± 0.03 for seed size and 0.33 ± 0.03 to 0.67 ± 0.04 for protein. The heritability estimates on an entry-mean basis for two replications at two locations were 0.45 ± 0.06 to 0.85 ± 0.05 for seed size and 0.62 ± 0.05 to 0.88 ± 0.05 for protein.
Two-parent LSHP x LSHP and three-parent (LSHP1 x N) x LSHP2 populations would be useful for the development of LSHP cultivars. The use of N cultivars or lines with above-average seed size or protein would enhance the usefulness of LSHP x N populations. Single-plant selection would be most cost effective in three-parent populations. The best results would be obtained by ranking individuals for the two traits and selecting those with the highest rank, instead of selecting all individuals equal to one of the LSHP parents in the cross.
|
NOTES
|
|---|
Project No. 3732 and supported by the Hatch Act, State of Iowa, Iowa Soybean Promotion Board, and Raymond F. Baker Center for Plant Breeding.
Received for publication November 26, 2001.
|
REFERENCES
|
|---|
- Bravo, J.A., W.R. Fehr, and S.R. Cianzio. 1980. Use of pod width for indirect selection of seed weight in soybeans. Crop Sci. 20:507–510.[Abstract/Free Full Text]
- Carpenter, J.A., and W.R. Fehr. 1986. Genetic variability for desirable agronomic traits in populations containing Glycine soja germplasm. Crop Sci. 26:681–686.[Abstract/Free Full Text]
- Cianzio, S.R., and W.R. Fehr. 1982. Genetic variability for soybean seed composition in crosses between high- and low-protein parents. J. Agric. Univ. P.R. 66:123–129.
- Cianzio, S.R., and W.R. Fehr. 1987. Inheritance of agronomic and seed composition traits in Glycine max x Glycine soja crosses. J. Agric. Univ. P.R. 71:53–63.
- Cober, E.R., H.D. Voldeng, and J.A. Frégeau-Reid. 1997. Heritability of seed shape and seed size in soybean. Crop. Sci. 37:1767–1769.[Abstract/Free Full Text]
- Frey, K.J., and T. Horner. 1957. Heritability in standard units. Agron. J. 2:59–62.
- Hallauer, A.R., and J.B. Miranda. 1995. Quantitative genetics in maize breeding. Iowa State Univ. Press, Ames, IA.
- Johnson, S.L., W.R. Fehr, G.A. Welke, and S.R. Cianzio. 2001. Genetic variability for seed size of two- and three-parent soybean populations. Crop Sci. 41:1029–1033.[Abstract/Free Full Text]
- LeRoy, A.R., S.R. Cianzio, and W.R. Fehr. 1991. Direct and indirect selection for small seed of soybean in temperate and tropical environments. Crop Sci. 31:697–699.[Abstract/Free Full Text]
- McKendry, A.L., P.B.E. McVetty, and H.D Voldeng. 1985. Inheritance of seed protein and seed oil content in early maturing soybean. Can. J. Genet. Cytol. 27:603–607.
- SAS Inst. 1992. SAS guide for personal computers. 6th ed. SAS Inst., Cary, NC.
- Ting, C.L. 1946. Genetic studies on the wild and cultivated soybeans. J. Am. Soc. Agron. 38:381–393.
- Weber, C.R. 1950. Inheritance and interrelation of some agronomic and chemical characters in an interspecific cross in soybeans, Glycine max x G. ussuriensis. Iowa Agric. Exp. Stn. Bull. 374.
TOP
ABSTRACT
INTRODUCTION
