| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
a Pioneer Hi-Bred International, Inc., P.O. Box 1004, Johnston, IA, 50131-1004
b USDA-ARS and Dep. of Crop Science, North Carolina State Univ., Raleigh, NC 27695-7631
c Dep. of Plant and Soil Sciences., Univ. of Tennessee, Knoxville, TN 37901-1071
* Corresponding author (joe_burton{at}ncsu.edu).
 |
ABSTRACT |
|---|
 TOPNOTES ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONAPPENDIXREFERENCES |
|---|
Abbreviations: HOT, high-oil tester • LOT, low-oil tester
|
INTRODUCTION |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONAPPENDIXREFERENCES |
|---|
In two populations derived from a base population segregating for ms1 male sterility, Priadi (1993) compared half-sib family and single selfed-plant recurrent selection for oil content. Six cycles of recurrent selection resulted in a per cycle linear increase of 1.1 ± 0.3 g kg–1 for single-plant and 0.5 ± 0.2 g kg–1 for half-sib family selection. No correlation was seen between oil and other agronomic traits, although there was a nonsignificant linear decrease in protein content.
Recurrent selection, as practiced above, is rapid in terms of cycles per year. That the mean oil concentration increased indicates that the frequency of genes affecting that trait increased in the population. The problem with selection for seed composition alone is that agronomic quality of the population may decline as the seed component increases (Brim and Burton, 1979). A possible way to prevent this would be to use half-sib recurrent selection and employ a good agronomic cultivar as the tester. At the very least, the improved trait would be expressed in a more agronomic genetic background, i.e., the population tester cross. Possible advantage could be made of this with additional breeding.
Testcross evaluation of unadapted germplasm in soybean was proposed by Kenworthy (1980) as a way to determine the combining ability of plant introductions. Several studies of testcross evaluation of germplasm have been conducted using the F1 generation (Lewers et al., 1998) or advanced generations (Reese et al., 1988; St. Martin and Aslam, 1986). Results on the importance of choice of tester, however, were mixed. Lewers et al. (1998) conducted F1 generation testcross evaluation of soybean germplasm using male sterility (ms6ms6) and seedling marker W1. Significant F1 and F2 midparent heterosis and inbreeding depression for oil content were observed. Hence, there may have been some dominance effects for oil content. The choice of tester was important in detecting the value of parental germplasm.
St. Martin and Aslam (1986) used two adapted lines and two plant introductions (PIs) with good agronomic characteristics as male testers crossed with six adapted lines and six PIs. Lack of female x male interactions for yield and oil content in the study indicated that the choice of tester was not critical. Since there were no differences in yield among their testcross progenies, they concluded that testcross evaluation was not even necessary. Similarly, Reese et al. (1988) used four testers crossed with nine visually desirable lines and 14 PIs selected for yielding ability. The F2 bulk populations were evaluated at three locations. They concluded that the choice of tester was not important since no tester group appeared to have an advantage over another.
A recurrent half-sib selection procedure was devised using a population that was segregating for ms1 male sterility and two testers, one with high-oil concentration and one with low-oil concentration. Seven and four cycles of half-sib selection were completed with these two testers, respectively. The objectives of this study were (i) to investigate the effectiveness of recurrent half-sib selection for increased oil content and evaluate changes in agronomic traits that might be correlated with selection, (ii) to evaluate the effect of tester oil content on selection response, and (iii) to investigate testcross heterosis and inbreeding depression for seed oil.
|
MATERIALS AND METHODS |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONAPPENDIXREFERENCES |
|---|
The germplasm for the first recurrent selection program, which was also the genetic basis for the subsequent recurrent selection program for increased oil content, was derived from a random mating among six lines, ‘Arksoy’, ‘Ogden’, ‘Lee’, ‘Roanoke’, D60-8107, and ‘Jackson’. Roanoke, N45-745, Ogden and Biloxi were grandparents of D60-8107. ‘CNS’ and Biloxi were parents of N45-745. Random mating was facilitated by male sterility induced by Tobacco ringspot virus (Brim, 1973). Four cycles of S1 progeny recurrent selection were undertaken and 10 lines with high seed oil content were selected and used as the base population for the second program.
For the second program, the 10 selected lines were manually mated with genetic male-sterile (ms1ms1) plants (Burton and Brim, 1981). Equal portions of F1 seeds (Ms1ms1) from each of the crosses were composited to form the C0 population. Each cycle of selection consisted of a step of mass selection followed by a step of within half-sib family selection for high seed oil. At the end of the third cycle, 22 plants with high seed oil content and gray pubescence color (tt) were selected from a total of 106 male-sterile plants. Seeds from the 22 selected plants were composited to form the base population, designated as Ca, for the current recurrent selection program. The genotypes in the Ca population were a mixture of Ms1Ms1tt, Ms1ms1tt, and ms1ms1tt.
Recurrent Half-Sib Selection Using Testers
Two testers were chosen for comparing the effects of tester oil content on selection response. One was a high-oil low-protein cultivar, Ransom, and the other was a low-oil high-protein line, N80-2177-2. N80-2177-2 was a line derived from the population, NC-1 (Burton and Brim, 1983). Both testers have tawny pubescence, which is controlled by a dominant gene (TT). Corresponding gray pubescence is controlled by recessive genes (tt). This trait was used as a marker to facilitate tester hybrid evaluation and sib hybrid recognition.
In summer 1981, the base population was planted in single plant hills and bordered with the Ransom tester (Ms1Ms1TT) in a random mating block at Clayton, NC. To minimize contamination from other pollen sources, random mating blocks were separated from adjoining soybeans at least 2 m of open space and no plants were harvested in the outside 2 m on all sides of the block. At maturity, the block was divided into 20 sub-blocks. About 10 male-sterile plants with seeds were randomly harvested from each subblock. Genotypes of the hybrid seeds were a mixed population of Ms1ms1Tt, Ms1ms1tt, and ms1ms1tt. This population was then grown in a winter nursery in Puerto Rico. Seeds harvested from male-sterile plants (half-sib families) in North Carolina were grown in progeny rows. At maturity, oil content of 10 g of seed from the tawny fertile Ransom hybrid plants (Ms1ms1Tt) was measured by nuclear magnetic resonance spectroscopy (NMR) at North Carolina State Univ. to identify high-oil content families (Brim et al., 1967). Seeds of the corresponding fertile gray sib-hybrid plants (Ms1ms1tt) within the rows were bulked to start the next cycle of selection. This bulked population, designated as C0, was used as the parent population for C1. Seven cycles of this recurrent selection for high-oil content were completed. Each cycle resulted in a gray population. These populations were designated as HOT populations because they were developed using a High-Oil Tester.
Another recurrent selection procedure for increased oil content was initiated, using C4 of the HOT population as the parent population, C0, and N80-2177-2 as a tester. Except for the change of tester, the same testing, selecting, and intermating procedures were used. This recurrent selection procedure was conducted for three cycles and the resultant populations were designated as LOT populations because they were developed using the Low-Oil Tester, N80-2177-2.
The average selection differential of the two procedures was 8.2 and 10.5 g kg–1, respectively. The average number of progenies tested in each cycle was 162. In most cycles, between 10 and 14% were selected from each population (Table 1).
|
Therefore, the materials used for evaluation were the second generation bulk-selfed selected progenies of Ca, C0, C1 through C7 gray (tt) fertile plants, and C4 through C7 tester- hybrid plants for the HOT populations; and the second generation bulk-selfed selected progenies of C0, C1 through C3 gray (tt) fertile plants, and C2, C3 tester-hybrid plants for the LOT populations. Equal quantities of seeds from the random selected progenies of each population were composited for each cycle to evaluate selection response. These composite populations were entries in the field experiment. A total of 21 entries were used in the evaluation experiment including the two testers, Ransom and N80-2177-2.
The experiment was conducted in 1997 at three North Carolina locations [Clayton, (plinthic kandidult), Plymouth, (typic umbraquult), and Windblow (arenic paleudult)], in a randomized complete block design with three replications at each location. Each entry was grown in a three-row plot, 5.8 m long by 0.97 m wide. Seeding rate was 26 seeds per meter in rows. At maturity, the center row of each plot was end-trimmed to 4.9 m and harvested for seed yield. Plants in the hybrid plots were separated into gray (sib hybrids) and tawny (tester hybrids) and threshed separately. Data were recorded on performance of the middle row of each three-row plot. Recorded agronomic traits included flowering (all plants in flower) and maturity date (95% of plant mature), plant height (5 plants/plot) and lodging (all plants), seed weight and plot yield. Seed weight, protein and oil content were measured separately for the sib-hybrid and tester-hybrid plants. Protein and oil content were analyzed by near infrared reflectance spectroscopy (NIR) at the USDA North Regional Research Center, Peoria, IL.
Statistical Analysis
The SAS GLM procedure (SAS Institute, 1990) was used for the statistical analysis. Locations were considered to be random effects, and genotypes (cycles) were considered fixed. To test for genotype effects, the Type III mean square of location x genotype was used as the error term. To test location effects, the Type III Rep within location mean square was used as the error term. Residual mean square was used for testing the effect of location x genotype interaction.
For testing progress in selection for oil content in the HOT population, variation among cycles was partitioned into linear and quadratic effects and these effects were tested with the genotype x location interaction mean squares. Since the cycle quadratic x location was significant for oil and protein content, a separate analysis of variance for each location was conducted and the residual mean square was used as an error term for testing differences among genotype effects. To evaluate change of rank in oil content among locations, Spearman's rank correlation analyses were performed (Steel et al., 1997). Since these analyses indicated that there were no significant changes (p > 0.20) in the rank of oil content among locations, a combined analysis for oil content was also performed over locations.
The SAS PROC REG procedure was used for regression of oil content on cycles of selection over locations and by locations for the HOT populations. Realized heritability estimates were calculated by regression of cycle means on accumulated selection differential. The oil increase was fit to a linear and quadratic model. The quadratic term was dropped from the regression model because it was not significant.
The SAS PROC CORR procedure was used for calculating simple correlation coefficients between oil content and other measured traits. The number of sterile plants per plot was used as a covariate in the initial analysis. Because the covariate was not significant, it was dropped from the analysis.
|
RESULTS AND DISCUSSION |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONAPPENDIXREFERENCES |
|---|
|
|
One possible reason for the plateau at C6 and C7 in our experiment may have been small effective population size and low initial genetic variability. Dudley and Lambert (1992) estimated that the number of genes affecting oil content for maize (Zea mays L.) was 69. Rawlings (1980) found that for short-term experiments of 5 to 10 cycles, expected progress began to plateau when the number selected was 16. This is larger than our initial samples. In our selection process, 12 plants were selected from 88 plants tested in the base population and 14 plants were selected from 52 plants tested in the C0 population. Although the number of genes controlling oil content in soybeans is not known, those small testing and sampling sizes in the early cycles of selection may have limited the genetic background for our base population and caused the plateau at C6 and C7 and the low overall progress. Compared with recurrent selection for oil content in maize (Smith, 1908; Dudley and Lambert, 1992), where oil content ranged from 38.4 to 60.2 g kg–1 for the 163 ears selected in the base selection population, the oil content in our base population ranged from 206 to 243 g kg–1. Our population had seven parents, but one of those, the male-sterile source, contributed half of the genetic make-up. So the genetic background might not have been as broad as Burr's White, the base population for the selection of maize oil content. This could be another reason for the lower rate of progress in our program.
Unexpected responses to recurrent selection in soybean as well as in maize have also been reported previously for other traits. Kenworthy and Brim (1979) found a significant yield improvement when selecting for seed yield per se but no significant linear response when selecting for efficiency expressed as the ratio of seed weight to straw weight or for an equally weighted rank index of yield and efficiency. Brim and Burton (1979) found a significant yield decrease in one population and a significant yield increase in another when selecting for percentage of seed protein. Comparing 90 generations of recurrent half-sib family selection for oil and protein in maize (Smith, 1908; Winter, 1929; Dudley et al., 1974; Dudley and Lambert, 1992), changes in oil or protein content were not strictly linear. Both plateaus and decreases were observed while selecting for high oil and plateaus and increases were observed while selecting for low oil for one or several consecutive generations. Therefore, other factors must influence selection results. One of those factors may be environment. For example, Smith (1908) reported severe weather conditions, particularly rainfall, affected oil as well as protein content during the different selection years.
In contrast to the HOT program, oil content did not change significantly (P > 0.05) with selection in LOT populations (Table 2). The average oil content ranged from 202 g kg–1 in C3 to 207 g kg–1 in C2 but the differences were nonsignificant (Table 3). The significant increase in the HOT program and lack of increase in the LOT program indicates that choosing an appropriate tester is important for selection progress. Results suggest that the tester should represent the desired phenotype of selection, which is assumed to have a high frequency of favorable allelic genes. Similar results were obtained by Lewers et al. (1998). In their research, St. Martin and Aslam (1986) and Reese et al. (1988) found the choice of tester not to be critical for parental evaluation. However, the testers used in those studies were all high yielding lines. Therefore, one reason for this discrepancy may be that high yielding testers were not compared with use of low yielding testers in those experiments.
In our study, there were significant location effects for all other traits recorded except flowering date (data not shown). Genotype x location interaction was not significant for protein content, plant height, and seed yield, but was significant for flowering and maturity date, plant lodging, and seed weight (data not shown).
A highly significant negative simple correlation (r = –0.88, p < 0.001) was observed between oil and protein content. When combined over locations, mean protein content in HOT populations decreased from 400 to 391 g kg–1 from C0 to C7 as oil content increased from 202 to 211 g kg–1 (Table 3). The decrease was linear (p < 0.05) at a rate of –1.2 ± 0.6 g kg–1 per cycle. Protein content increased slightly, as oil content remained steady in C6 and C7. From C0 to C5, oil content increased linearly at a rate of 1.7 ± 0.2 g kg–1 per cycle and protein content decreased linearly at a rate of –2.7 ± 0.3 per cycle. In this case, the ratio of change in oil to change in protein was 1.7/2.7 = 0.63. Combined over all seven cycles and three locations, the ratio of change in oil to change in protein was 1.1/1.2 = 0.92. The large negative correlation indicates that a dynamic equilibrium between the two may exist and further illustrates the well-established negative relationship between concentrations of seed protein and seed oil. The 0.63 ratio of change in oil to change in protein from C0 to C5 is similar to a previous estimate by Burton and Brim (1981). The negative correlation between oil and protein content was also indicated for hybrids in this experiment. If there was a positive midparent heterosis for oil content, there was usually negative midparent heterosis for protein content (Table 4).
|
There was no significant (p < 0.05) change in oil content among cycles for tester hybrids or sib hybrids in either the HOT or the LOT populations (Table 2). For HOT populations, Ransom hybrid mean oil content ranged from 213 g kg–1 in cycle four to 216 in cycle six but decreased to 212 in cycle seven; sib hybrid-oil content increased from 204 to 210. For LOT populations, oil content in N80-2177-2 hybrid and in sib hybrids remained the same for the two cycles of selection at the levels of 175 and 203 g kg–1, respectively.
Compared with selfed and sib hybrids in the HOT populations combined over cycles and locations, Ransom hybrids had a significantly (p < 0.01) higher paired difference in oil content of 0.4 and 0.5, respectively (Table 4). For LOT populations, the results were just the opposite. Tester N80-2177-2 hybrid had significantly (p < 0.01) lower paired difference in oil content of 3.0 and 2.9 compared to its corresponding selfed- and sib-hybrid populations (Table 4).
There was significant (p < 0.01) midparent heterosis combined over two cycles of selection for LOT populations. Average midparent heterosis [tester-hybrid mean – (mean of selfed population + tester)/2], was 0.9 (p < 0.001) for oil content and –1.2 (p < 0.01) for protein content. For HOT populations, Ransom hybrids showed a nonsignificant average midparent heterosis of 0.1 in oil content, ranging from –0.1 to 0.4 (Table 4). Evidence of heterosis for soybean oil indicates that there may be some dominance effects for oil content. This finding is different from some previous research (Brim, 1973) but is consistent with results of Lewers et al. (1998). In the LOT populations, there were significant (p < 0.05) differences in sib-hybrid seed weight. There were no significant changes (p < 0.05) for any other sib-hybrid traits measured in either the HOT and LOT populations.
Recurrent half-sib selection has been widely used in maize as a way to improve the specific combining ability of a population (Hallauer and Miranda, 1981), but it has not been used in soybean. Because the soybean breeding product is a pure-line cultivar rather than an F1 hybrid, selection for specific combining ability has never been a breeding goal. Yet, some soybean cultivars have been used successfully as a parent (e.g., Williams, Young, and Hutcheson) in the derivation of many productive cultivars. Thus it would seem that those cultivars have superior general combining ability. Using such a cultivar as tester in a recurrent half-sib selection program might be a way to improve the trait of interest, and improve the combining ability of the populations with the tester. Both favorable additive and additive x additive epistatic gene combinations would be accumulated. Once a desired change in the trait has been achieved, pure-lines with agronomic quality of the tester could be derived from the population x tester cross using modified pedigree selection.
Finally, this experiment further demonstrated that the use of nuclear male-sterility and a marker trait can be useful in facilitating a recurrent half-sib selection procedure that combines self-pollination and testcross production in the same season, as well as selfing and testcross evaluation in the same season. Each cycle of selection requires two seasons that could be completed in a single year if a winter nursery were used. Although environmental conditions may exert a significant effect on oil and protein contents, genotype x environment interactions were not significant for oil content in this study. Therefore, it is suggested that breeding progress for increased oil content can be made with progeny selection in a single environment.
|
APPENDIX |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONAPPENDIXREFERENCES |
|---|
 |
2A is the additive genetic variance, D1 is the covariance between additive and homozygous dominance effects, and 2Y is the phenotypic variance among the selection units. Both 2A and D1 are defined in reference to the noninbred population in linkage equilibrium (Cockerham, 1983). |
ACKNOWLEDGMENTS |
|---|
|
NOTES |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONAPPENDIXREFERENCES |
|---|
Received for publication January 21, 2003.
|
REFERENCES |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONAPPENDIXREFERENCES |
|---|
Related articles in Crop Science:
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| The SCI Journals | Agronomy Journal | Vadose Zone Journal | |||
| Journal of Natural Resources and Life Sciences Education |
Soil Science Society of America Journal | ||||
| Journal of Plant Registrations | Journal of Environmental Quality |
The Plant Genome | |||
in the HOT and LOT population.