| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Dep. of Agronomy and Plant Genetics, Univ. of Minnesota, 411 Borlaug Hall, 1991 Upper Buford Circle, St. Paul, MN 55108-6026
* Corresponding author (berna022{at}umn.edu).
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONREFERENCES |
|---|
Abbreviations: BSSS, Iowa Stiff Stalk Synthetic • VTA, testcross additive variance • VTAA, testcross additive-by-additive variance • VTC, testcross variance
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONREFERENCES |
|---|
The average yield in different crops continues to increase by about 1 to 2% each year and these linear increases across time demonstrate that advanced cycle breeding has not reached a yield plateau (Poehlman and Sleper, 1995, p. 4). Advanced cycle breeding, however, systematically reduces genetic variation (Rasmusson and Phillips, 1997; Tanksley and McCouch, 1997). From a population genetics standpoint, crossing two inbreds to form a breeding population at each cycle creates a bottleneck, which is defined as a severe reduction in the number of parents that are mated to form the next generation. Assuming no dominance and no epistasis, the additive genetic variance in a bottlenecked population decreases as a linear function of the inbreeding coefficient (Falconer and Mackay, 1996, p. 264). Bottlenecks would therefore lead to a continuous loss of genetic variation in breeding populations from cycle to cycle.
Concern about the loss of genetic variation in crops has increased (Tanksley and McCouch, 1997; Hoisington et al., 1999). No research to systematically monitor the changes in genetic variance during advanced cycle breeding has been reported. Our objectives were to estimate the loss, if any, in genetic variance during advanced cycle breeding in maize, and to compare these genetic variance estimates with those expected under a testcross additive model.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONREFERENCES |
|---|
Seeds of BSSS were obtained from Iowa State University. Seeds of B73 and B37 were obtained from stocks maintained at Purdue University. Seeds of Lo904 and Lo916 were obtained from the Instituto Sperimentale per la Cerealicoltura, Bergamo, Italy. Six populations (BSSS, B73 x B37, and Lo904 x Lo916, each at two selfing generations, S0 and S2) were synthesized. The BSSS-S0 was formed by random mating 180 S0 plants, and the B73 x B37-S0 and Lo904 x Lo916-S0 populations by random mating 180 F2 plants. Random plants from the BSSS-S0, B73 x B37-S0, and Lo904 x Lo916-S0 populations were selfed for two generations to create S2 families. The resulting populations were denoted by BSSS-S2, B73 x B37-S2, and Lo904 x Lo916-S2. Having populations at two selfing generations allowed us to calculate a pooled variance estimate with the assumption of no epistasis, and to test an epistatic model by considering the two estimates separately.
Three non-BSSS inbreds (LH185, LH295, and a proprietary Syngenta inbred) were used as testers (Table 1). These three testers were chosen so that the resulting testcrosses could be evaluated in the maturity zone of the testing environments. An individual S0 plant was crossed as the male parent to two plants of a tester, and the resulting seeds were bulked as an S0 testcross. A single S2 plant from each S0 family was likewise crossed as the male parent to two plants of a tester to obtain an S2 testcross.
|
The plants were grown in two-row plots, each row 6.7 m long and spaced 0.76 m apart, at a plant population density of 68678 plants ha–1. Data recorded on plot were grain yield (Mg ha–1, adjusted to 155 g H2O kg–1), grain moisture (g kg–1), stalk lodging (percentage of stalk lodged plant per plot), plant height (cm, distance from the soil surface to the tip of the tassel), and ear height (cm, distance from the soil surface to the highest ear-bearing node). Grain yield and grain moisture were recorded for the LH185, Syngenta, and LH295 testcrosses. Stalk lodging was recorded for the LH185 and Syngenta testcrosses, and plant height and ear height were recorded for the LH295 testcrosses.
Expected Genetic Variance
A zero inbreeding coefficient was assumed for the BSSS population. The inbreeding coefficient of B73 x B37 (FB73xB37), corresponding to the coefficient of coancestry between B73 and B37 (fB73, B37), was calculated from the average of three studies (Godshalk et al., 1990; Messmer et al., 1991; Bernardo, 1993) as summarized by Bernardo (1993). The parental contribution of B73 and B37 to Lo904 and Lo916 was based on a study with 195 SSR markers (Bernardo et al., 2000). The inbreeding coefficient of Lo904 x Lo916 (FLo904xLo916), corresponding to the coefficient of coancestry of Lo904 and Lo916 (fLo904, Lo916), was further calculated as described by Bernardo et al. (2000).
Because testcross genotypic values are additive (Hallauer and Miranda Filho, 1988, p. 28; Bernardo, 2002, p. 79), the expected testcross variance can be calculated as
 |
Statistical Analysis and Observed Genetic Variance
The linear model for the LH185 and Syngenta testcrosses at each selfing generation was
 |
 |
ijm (ijkm) is the random error. The environments and testcrosses were considered as random effects. The mean squares from the ANOVA were equated to their expectations to solve for the testcross variance (VTC) (SAS Institute, 1990).
The VTC is a function of VTA and VTAA. We considered a testcross additive model and a testcross epistatic model. For the testcross additive model, VTA was solved from
 |
 |
A 90% confidence interval on the pooled VTA was obtained based on the method of Ting et al. (1990) and Burdick and Graybill (1992)(p. 56). The variance estimates were regarded as significantly different from zero if their confidence intervals did not include zero. A significant difference between two estimates was declared if their confidence intervals did not overlap.
|
RESULTS AND DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONREFERENCES |
|---|
For grain yield, the pooled estimates of VTA across two selfing generations were all positive (Table 2). The sizes of VTA in BSSS, B73 x B37, and Lo904 x Lo916 showed a similar pattern for the LH185 and Syngenta testcrosses: a decrease from BSSS to B73 x B37, followed by an increase from B73 x B37 to Lo904 x Lo916. The pooled estimates across three testers also followed the same pattern (Table 2, Fig. 1) . For the LH295 testcrosses, the sizes of VTA showed a different pattern: an increase from BSSS to B73 x B37, followed by a decrease from B73 x B37 to Lo904 x Lo916. The differences, however, were not significant among the pooled estimates of VTA across two selfing generations for each tester and across the three testers.
|
|
For stalk lodging, plant height, and ear height, the pooled estimates of VTA across the two selfing generations followed a common pattern: a decrease in B73 x B37 and a further decrease in Lo904 x Lo916 (Table 2, Fig. 1). For stalk lodging, the pooled estimates of VTA across two testers decreased significantly from BSSS to B73 x B37 (Table 2). For plant height and ear height, the pooled estimates of VTA across the two selfing generations decreased significantly from BSSS to B73 x B37 (Table 2).
Across all five traits, the behavior of VTA can be summarized as follows. As expected, VTA decreased from BSSS to B73 x B37 for all traits. But VTA in Lo904 x Lo916 was not always less than VTA in B73 x B37. The VTA in B73 x B37 was lower than expected for all traits (Fig. 1). The VTA in Lo904 x Lo916 was lower than expected for grain moisture, plant height, ear height, and stalk lodging, but not for grain yield.
The classical additive model predicts a decrease in genetic variance after a population bottleneck. Several studies, however, showed that additive genetic variance after a population bottleneck was significantly higher than expected from the additive model (Bryant et al., 1986; Wade et al., 1996; Cheverud et al., 1999). Our results showed that traits react differently to population bottlenecks. At a given bottleneck, VTA may decrease for one trait but increase for another. Meanwhile, for a given trait, VTA may decrease after one bottleneck but increase after the next bottleneck. Reviews of experimental and theoretical studies suggested that bottlenecks affect traits in different ways depending on the relative amount of additive and nonadditive genetic variances (Wade et al., 1996; Meffert, 1999). Fertility in flour beetle [Tribolium castaneum (Herbst)], for example, showed increased genetic variance after the bottleneck, but pupal weight showed decreased genetic variance as predicted by an additive model (Wade et al., 1996). Conversion of additive-by-additive epistatic variance into additive variance was proposed as a mechanism for maintaining additive variance in a bottlenecked population (Cockerham and Tachida, 1988; Goodnight, 1988; Whitlock et al., 1993; Cheverud and Routman, 1996). For example, for two loci with duplicate dominant epistasis, random fixation of either locus would lead to an average increase of 300% in additive variance (Bernardo, 2002, p. 115). The changes in VTA for grain yield and grain moisture in our study deviated prominently from a continuously decreasing trend after two population bottlenecks, which was expected from the testcross additive model. This deviation suggested that testcross nonadditive effects, which can be converted into additive genetic variance during a bottleneck, might have been involved.
Testcross Additive versus Testcross Epistatic Models
For all traits examined, the testcross epistatic model did not fit the data for any tester because it failed to give positive estimates of both VTA and VTAA (data not shown). Theoretical studies suggested that even a small epistatic variance could have substantial effect in maintaining additive variance in a bottlenecked population (Goodnight, 1988; Bernardo, 2002, p. 116). The negative estimates of VTA or VTAA in the testcross epistatic model did not allow us to verify the maintenance of additive variance through the conversion of epistatic variance. Although physiological epistasis undoubtedly exists, the general conclusion drawn from empirical studies is that realistic estimates of epistatic variance are difficult to obtain (Hallauer and Miranda Filho, 1988, p. 140).
General reasons that lead to the difficulties in detecting a significant epistatic variance were summarized by Bernardo (2002)(p. 141). First, a complex mating design is needed to obtain multiple covariances between relatives. Second, epistatic variances are by definition expected to be smaller than additive variance and dominance variance. Third, epistatic variances are difficult to separate from additive variance and dominance variance. The large sampling error variance associated with estimates of epistatic variance is also another reason for nonsignificance of epistatic variance (Mackay, 2001); the nonsignificance of epistatic variance does not suggest an absence of epistasis.
Genetic Variance and Advanced Cycle Breeding
Advanced cycle breeding is generally believed to improve mean performance at the cost of reducing genetic variance. Our results confirmed the improvement in mean performance: higher grain yield and fewer plants lodged in Lo904 x Lo916 than in BSSS, and acceptable grain moisture, plant height, and ear height in Lo904 x Lo916 (Table 3). The testcross means (pooled across selfing generations and testers) indicated a 30% increase in grain yield from BSSS to B73 x B37, and a 9% increase in grain yield from B73 x B37 to Lo904 x Lo916. The results in testcross means for grain yield and grain moisture were in agreement with the improvement documented for BSSS, B73, B37 (Troyer, 1999), Lo904, and Lo916 (Bertolini et al., 1991).
|
Molecular marker studies of progenitors and elite lines from BSSS (Messmer et al., 1991; Bernardo et al., 2000) confirmed that the number of alleles per locus has been reduced in B73 x B37 and further reduced in Lo904 x Lo916 compared with the original BSSS. In practice, advanced cycle breeding involves crossing many pairs of elite inbreds, for example, B14 and B84, in addition to B73 and B37. This would result in a slower rate of fixation of alleles than in BSSS to B73 x B37, and B73 x B37 to Lo904 x Lo916. But, if the inbreds all belong to the same genetic background, those at advanced cycles would have an increasingly higher proportion of loci with fixed alleles, which could eventually lead to a lower genetic variance even if additive genetic variance is maintained during the first few cycles. For any crop species, it is perhaps premature to predict an exhaustion of genetic variance, but broadening the genetic base beyond the current level would help to sustain long-term genetic gains (Hallauer, 1990; Hoisington et al., 1999). Brown (1975) suggested that, on a global basis, only about 5% of the available maize germplasm is used commercially. The amounts of available but unused germplasm for other major crops are also huge (Hoisington et al., 1999). A strategy of maintaining favorable gene combinations through advanced cycle breeding, while increasing genetic variation through introgressing new germplasm into elite germplasm pools, has led to promising results in wheat (Carver and Rayburn, 1994), tomato (Lycopersicon esculentum Mill.; Bernacchi et al., 1998), and barley (Peel and Rasmusson, 2000). But because of its proven success, breeders, particularly in private companies, will probably continue to practice advanced cycle breeding until it becomes obvious that genetic variance has been exhausted and that new germplasm sources are needed.
Received for publication April 16, 2003.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  D. A. Tabanao and R. Bernardo Genetic Variation in Maize Breeding Populations with Different Numbers of Parents Crop Sci., September 23, 2005; 45(6): 2301 - 2306. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| 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 | |||
