| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
a State Plant Breeding Institute, University of Hohenheim, 70593 Stuttgart, Germany
b Institute of Plant Breeding, Seed Science, and Population Genetics, University of Hohenheim, 70593 Stuttgart, Germany
* Corresponding author (melchinger{at}uni-hohenheim.de)
 |
ABSTRACT |
|---|
 TOPNOTES ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
was more important than SCA variance 
for all traits except grain yield and protein concentration. For most traits, GCA x location and SCA x location interaction variances were small relative to 
2GCA and 2SCA, respectively. Genetic correlations between midparent and hybrid performance and between GCA effects and line per se performance showed similar trends, being moderate for grain yield and protein concentration and higher for the other traits. We concluded that grain yield heterosis in winter triticale crosses from parents in the current European germplasm pool is adequate to justify continuing research on hybrid breeding. By selecting parents for combining ability and establishing genetically diverse heterotic groups, a midparent grain yield heterosis of 20% could presumably be surpassed. Further information is needed on F1 seed production and the cytoplasmic male sterility system.
Abbreviations: BPH%, relative better-parent heterosis • CHA, chemical hybridizing agent • GCA, general combining ability • HYB, hybrid performance • LP, line per se performance • MP, midparent value • MPH, absolute midparent heterosis • MPH%, relative midparent heterosis • SCA, specific combining ability
|
INTRODUCTION |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Owing to its genome constitution with one third of the chromosomes from the allogamous rye and its floral biology of large extruding anthers and some degree of outcrossing (Yeung and Larter, 1972; Sowa and Krysiak, 1996), triticale is expected to have more potential for heterosis and hybrid breeding than wheat. Modern rye hybrids displayed substantial midparent heterosis for grain yield (92%) and are widely cultivated in several European countries (Geiger and Miedaner, 1999).
An important prerequisite for establishing a hybrid breeding program is a sufficiently high level of heterosis. Previous reports based on single plants or small plot experiments tended to overestimate heterosis. Trethowan and Darvey (1994) estimated an average of 17% midparent grain yield heterosis in hill-plots. In small plots of 2.5 to 3 m2, Oettler et al. (2001) measured 10.5% midparent heterosis. A recent study used larger plots seeded at normal rates and reported an average of 10.1% midparent grain yield heterosis, but this estimate was based only on 24 hybrids from six female and four male parent lines (Oettler et al., 2003). Hitherto, a large-scale and comprehensive study with genetically diverse material was lacking.
A fundamental issue in hybrid breeding is the choice of parents and identification of superior hybrid combinations. Grzesik and W
grzyn (1998) found GCA effects to be more important than SCA effects in winter triticale. In contrast, Oettler et al. (2003) reported predominance of SCA effects for grain yield and concluded that prediction of GCA from parental performance was moderate.
The objectives of this study were to (i) determine the level of heterosis for eight agronomic traits in 209 winter triticale hybrids, (ii) assess the relative importance of GCA vs. SCA effects, (iii) calculate correlations between GCA and line per se performance, (iv) estimate trait correlations in parents and hybrids, and (v) discuss the prospects for hybrid breeding in winter triticale.
|
MATERIALS AND METHODS |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Statistical Analyses
Ordinary lattice analyses of variance were performed with the data from each experiment in each location. Since the means of the three experiments were not significantly different and the ranges of parent lines were similar, we pooled data across experiments to present the results most parsimoniously. Thus, we assumed three random samples of triticale lines and presented combined analyses with the following model for the adjusted hybrid means from the lattices (Cochran and Cox, 1957):
 |
ijkm = averaged plot residual. A fully random model was assumed. The combined analysis was separated for hybrids and parents because common variances for the two groups could not be assumed. Heritabilities were estimated on an entry-mean basis and their 95% confidence intervals were calculated after Knapp and Bridges (1987). Variance components of female GCA 
and SCA 
were estimated for all traits by standard methods (Bernardo, 2002). Genotypic correlations (rg) were calculated between midparent and hybrid performance and between female GCA effects and their line per se performance (LP). Empirical 95% confidence intervals for these correlations were determined by 2000 bootstrap MANOVA samples after Liu et al. (1997). For each cross combination (P1 x P2), hybrid performance (HYB), midparent value (MP), absolute midparent heterosis (MPH), relative midparent heterosis (MPH%), and relative better-parent heterosis (BPH%) were calculated as follows: MP = (P1 + P2)/2; MPH = HYB–MP; MPH% = (MPH/MP) x 100; BPH% = (HYB–Pmax)/Pmax x 100, where Pmax refers to the higher performing or taller parent. The significance of MPH was tested by a t test using the pooled interaction variances of hybrids x locations and parents x locations as the error term. The minimum and maximum values of MPH were tested by Scheffé's test (Steel and Torrie, 1980).
|
RESULTS |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
|
were significant (P < 0.01) for parents and hybrids and exceeded twice their respective standard errors for all traits (Table 2). Both groups displayed similar 2G for grain yield. For the remaining traits, 2G for the parents was about twice as large as for the hybrids. For parents, estimates of genotype x location interaction variances 
were smaller than 2G for all traits and most of the differences were significant (P < 0.01). Estimates of 
2GCA were significant (P < 0.01) for all traits except protein concentration and were larger than 2SCA for most traits. For grain yield and protein concentration, however, 2SCA surpassed 2GCA. GCA x location and SCA x location interaction variances 
were significant (P < 0.01) in most cases but small relative to 2GCA and 2SCA, except for number of kernels per spike and protein concentration. For most traits, 2GCAxL was larger than 2SCAxL, but both interaction variances were of the same order for grain yield. Heritability estimates ranged from 58.9 to 95.5% among parents and from 50.1 to 91.8% among hybrids.
|
Genotypic trait correlations were moderate to low and similar for parents and hybrids (Table 3). The highest association (rg = –0.6, P < 0.01) was observed between grain yield and protein concentration in both parents and hybrids. In parents, protein concentration was negatively associated with all other traits.
|
|
DISCUSSION |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
The mean relative better-parent grain yield heterosis of 5.0% in the present experiment, which amounts to half the mean midparent heterosis, was in agreement with the results of the earlier study (Oettler et al., 2003) and with the 5.2% measured in spring triticale (Pfeiffer et al., 1998). Information from wheat varies greatly and appears to be influenced by the material tested. Oury et al. (2000) reported 6.5% better-parent heterosis, whereas Dreisigacker (pers. commun.) observed a negative better-parent heterosis of –9.3% in wheat.
In view of the allopolyploid nature of triticale (AABBRR), one might expect a considerable influence on heterosis from the R-genome chromosomes of the cross-pollinating rye. In recent rye hybrids, a relative midparent grain yield heterosis of 92% was observed (Geiger and Miedaner, 1999). The average heterosis of hybrids from parent lines of the current triticale germplasm pool is, however, closer to wheat than to rye. One reason might be that in the allopolyploid triticale there is already "fixed" heterosis in lines caused by epistatic interactions between genes from different genomes, which results only in a moderate level of additional heterosis (Mac Key, 1970). Furthermore, because no systematic hybrid breeding has been conducted hitherto and no distinct heterotic groups exist in triticale, a considerably higher level of heterosis than estimated in our study can be expected with long-term systematic hybrid breeding. The maximum midparent grain yield heterosis of 22.4%, although probably overestimated due to genotype x environment interactions, is an indication of this potential. According to Jordaan (1996), the biggest limitation in wheat for breeding hybrids has been to neglect the basic requirement of developing heterotic groups. A recent study in wheat underlines the benefits of heterotic groups for hybrid performance in an autogamous crop. The maximum midparent grain yield heterosis of 8% in intragroup hybrids was less than half the value (19%) measured in intergroup hybrids (Liu et al., 1999).
The largest heterosis in yield components was observed for 1000-kernel weight. A low or even negative heterosis for number of spikes per square meter is well established in hybrids of small-grain cereals and has been documented earlier for winter triticale (Oettler et al., 2003), spring triticale (Pfeiffer et al., 1998), wheat (Borghi et al., 1988), and for the early rye hybrids (Geiger and Miedaner, 1999). However, the wide range with a maximum of 12% midparent heterosis for this trait indicated potential for considerable improvement, as found in modern rye hybrids with a positive heterosis of 7% (Geiger and Miedaner, 1999).
Preharvest sprouting, indirectly measured by falling number, still remains one of the most serious problems of triticale. Current cultivars show poor sprouting resistance (Oettler, 2002). In hybrids, this defect was exacerbated as revealed by the significant negative heterosis for falling number. Deterioration of preharvest sprouting will hamper farmers' acceptance of hybrid over pure line cultivars in triticale. The significant genetic variation in parents and hybrids, as well as the relative maximum midparent heterosis of 28.7%, showed potential for improvement by systematic breeding for this trait.
A significant midparent heterosis for plant height of 6.7 cm (5.7%) was in conflict with the general breeding objective of reducing the straw length in cereals. However, most hybrids fell within the range of their parents and were only 3.0 cm taller than the check cultivars. Therefore, plant height should have little impact on the potential to commercialize triticale hybrids.
Combining Ability and Prediction of Hybrid Performance
Optimum allocation of resources in hybrid breeding depends on efficient methods for choosing parents and identifying superior hybrid combinations. The analysis of combining ability provided information about the relative importance of GCA vs. SCA effects and gave an indication of the gene action involved in the inheritance of the traits. Estimates of 2GCA were greater than 2SCA for all traits except grain yield and protein concentration, indicating that additive effects were more important than non-additive effects. For grain yield, however, non-additive effects predominated. This was in accordance with our earlier study (Oettler et al., 2003) and with the study in wheat by Oury et al. (2000). Genetic variance of the hybrids, which was only half the size of 2G of the parents for most traits, supports the assumption of only additive gene action. The greater 2G of parents than of hybrids for grain yield, which was also reported for wheat by Borghi et al. (1988), might be the result of dominance effects, a relationship of a tester with some hybrids, or meiotic instability. Even in elite triticale breeding lines and cultivars, cytological disturbances such as univalents are still present and may be more serious in some F1 hybrids (Lelley, 1996).
The close correlation between GCA effects and female line per se performance for all traits except grain yield indicated predominance of additive over dominance effects. Selection of potential parents based on per se performance was effective when traits other than grain yield were considered. In contrast, 2SCA was greater than 2GCA, and rg (GCA, LP) was only moderate for grain yield, indicating that selection for per se performance of parents or GCA was of little predictive value for hybrid performance. Furthermore, rg (MP, HYB) was low for grain yield and only 23% of the variation in hybrid performance was explained by the midparent value. In addition, future experimental testcrosses should be evaluated in multi-environment trials, because 2GCAxL and 2SCAxL were important for grain yield and most other traits. It is conceivable, however, that with the establishment of heterotic groups in triticale, the ratio of 2SCA to 2GCA might be reduced for grain yield and the other traits.
Implications for Hybrid Breeding
Pure lines are currently the predominant type of cultivar in commercial triticale production, and released cultivars are nearly homozygous and homogeneous lines. Breeders have frequently used lines from other programs as parents when developing breeding populations that resulted in a leveling out of the genetic diversity in the European winter triticale germplasm pool (Tams et al., 2004). If hybrid breeding in triticale becomes a long-term goal, diversity between parent lines must again be promoted and heterotic pools should be established. A recent study on hybrid maize from the U.S. Corn Belt demonstrates how two genetically distant heterotic groups evolved from a rather mixed germplasm pool that lacked distinct subgroups (Duvick et al., 2004). This process is, however, a long-term task.
For private plant breeding companies, hybrid seed business will be highly attractive because of the built-in plant variety protection of hybrids. Farmers have to buy new seed every growing season. Schachschneider (2000) estimated that from 1996 to 2000 in Germany 30 to 35% of the triticale area was planted with farm-saved seed.
The decision to embark on a hybrid breeding program and commercialization of hybrids should depend on a number of factors, including fertility, heterosis, trait correlations, and F1 seed production costs. Fertility, which had been a problem in the earlier phases of cultivar development in triticale, has been improved by intensive breeding (Bundessortenamt, 1988, 2004). None of the parents and hybrids in our experiment showed poor fertility.
The amount of midparent grain yield heterosis necessary to make hybrids commercially viable was estimated to range between 6 to 17% for wheat in the UK (Pickett and Galwey, 1997). Schachschneider (1997) regarded a 0.6 to 1.0 Mg ha–1 yield advantage of hybrid wheat over pure lines sufficient for commercial production in Germany. For hybrid triticale, Weißmann and Weißmann (2002) considered a 0.9 to 1.0 Mg ha–1 higher yield enough to justify the production of triticale hybrids for European market conditions. Therefore, the average midparent heterosis of 0.84 Mg ha–1 (10.3%) observed in the present study is encouraging, although relative better-parent heterosis was only half as much. By grouping germplasm into genetically diverse heterotic groups, as is currently under way in our laboratory with the aid of molecular markers (Tams et al., 2004), and by the production of inter-pool hybrids, it appears feasible to reach or even surpass the present maximum midparent heterosis of 1.81 Mg ha–1 (22.4%). Furthermore, because triticale is mainly grown under marginal and stress conditions such as sandy soils, water stress or mineral toxicity (Varughese et al., 1996; Banaszak and Marciniak, 2002), the relative yield advantage of hybrids over pure line cultivars is expected to be even higher. Such a relative yield advantage in stress conditions has been demonstrated in wheat (Jordaan, 1996) and rye (Geiger and Miedaner, 1999). Therefore, with regard to the amount of heterosis, our results are more promising for developing hybrids in triticale than in wheat.
The commercially exploitable yield advantage of hybrids will also depend on the rate of progress in line breeding and the time-scale for the development of hybrid vs. line cultivars. At present, the level of hybrid grain yield is on a par with that of the check cultivars. Forty-six of our hybrids outyielded the mean of the check cultivars by 10% or more. Considering the breeding efforts in line cultivar development for triticale in recent years in Europe (Arseniuk and Oleksiak, 2002; FAOSTAT), substantial progress can also be expected in the future. On the assumption that a CHA is used for inducing male sterility in the seed parent, Pickett and Galwey (1997) estimated that a wheat hybrid could be released 1 yr after its line parents, if the latter had proved suitable as line cultivars. However, if cytoplasmic male sterility had to be introgressed into the female parent lines by recurrent backcrossing, the developmental time lag for hybrids would be considerably longer.
Trait correlations in parents and hybrids were generally similar. All three yield components contributed in the same way to grain yield in both groups. The well-established negative relationship between grain yield and protein concentration was also found in parents and hybrids. Likewise, the breeder must be aware that an increase in grain yield might be in conflict with the breeding objective of reduced plant height. Furthermore, an increase in kernel size might result in a lower falling number that makes genotypes more susceptible to preharvest sprouting.
One of the critical issues for a hybrid program is the cost-effective production of high quality F1 seed. Several factors, including the need to devote large areas to the male parent (which does not produce marketable F1 seed) in seed production fields, difficulties in synchronizing the flowering times of female and male parents, and the production of limited quantities of hybrid seed on the male-sterile female parent, contributed to the failure of hybrid wheat production (Pickett and Galwey, 1997). These authors consider a 6% heterosis to be required for wheat to counterbalance a two-fold cost of hybrid seed relative to pure-line seed at a performance level of 6 Mg ha–1. At four-fold hybrid seed costs, a 17% heterosis would be required to meet the extra costs. The floral biology and a tendency to outcrossing in triticale could help to keep seed costs down. Whereas a male to female ratio from 1:1 to 1:3 is required for adequate F1 seed production of wheat (Pickett and Galwey, 1997; Schachschneider, 1997), a wider ratio can probably be used in triticale. The reduced levels of inputs generally used for triticale, lower seeding rates, and utilization of female parents selected for improved seed set, could also help to reduce seed costs.
Finally, the future of hybrid breeding in triticale depends crucially on a reliable hybridizing system. The CHAs of high sterilizing power, as used in this study, are not licensed for commercial production in the European Union because they are regarded as hazardous to the environment. The cytoplasmic male sterility system from the tetraploid wheat Triticum timopheevi (Zhuk.) Zhuk. with suitable restorers, which is used successfully in some countries for hybrid wheat production (Jordaan, 1996; Duvick, 1999), is currently considered the most promising biological hybridizing system for triticale. Warzecha and Salak-Warzecha (2002) presented the first report on male sterile lines in winter triticale. Commercial release in Australia of a spring triticale hybrid based on the T. timopheevi cytoplasm is expected for 2005 (Darvey and Roake, 2002). However, a final decision on the usefulness of the T. timopheevi cytoplasm for triticale cannot yet be reached. Research in the area of hybridizing systems, including other cytoplasmic male sterility sources, should be given high priority.
In conclusion, grain yield heterosis in winter triticale crosses from parents of the current European germplasm pool appears sufficient to justify continuing work on hybrid breeding. By selecting parents for high combining ability and establishing heterotic groups, this should result in the production of intergroup hybrids with higher levels of heterosis than presently obtained. Further research and more substantial information are needed on F1 seed production, the sterilizing system and also on the growing regimes suitable for hybrids before the prospects for commercial hybrid triticale can be reliably evaluated. Hybrids are not expected to replace line cultivars on a large scale, but they may be good options for more marginal environments.
|
ACKNOWLEDGMENTS |
|---|
|
NOTES |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Received for publication July 29, 2004.
|
REFERENCES |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
grzyn. 1998. Heterosis and combining ability in some varieties of triticale. p. 129–133, Vol. II. In Proc. 4th Int. Triticale Symp., Red Deer, AB, Canada.
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| 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 | |||
