| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
akiro
luCenter for Applied Genetic Technologies, Crop and Soil Sciences Dep., Univ. of Georgia, 111 Riverbend Rd., Athens, GA 30602
* Corresponding author (brummer{at}uga.edu).
 |
ABSTRACT |
|---|
 TOPNOTES ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONREFERENCES |
|---|
Abbreviations: GCA, general combining ability • HS, half-sib • MG, mid-dormancy group • MP, midparent • N, nondormant-derived • S, semidormant • SCA, specific combining ability
akiroluCenter for Applied Genetic Technologies, Crop and Soil Sciences Dep., Univ. of Georgia, 111 Riverbend Rd., Athens, GA 30602
* Corresponding author (brummer{at}uga.edu).
Increasing the yield of alfalfa (Medicago sativa L.) is a desirable breeding goal. The objective of this experiment was to test the hypothesis that hybrids between nondormant-derived alfalfa germplasm selected for winter hardiness and semidormant germplasm will express heterosis for yield. Four semidormant cultivars and four nondormant-derived populations were handcrossed in a half-diallel mating design to form population hybrids. Both seeded and transplanted trials were grown in Iowa between 2003 and 2005. Total biomass yield was measured at four harvests each year. The nondormant-derived germplasm produced lower yields than the semidormant cultivars. The hybrids between groups were intermediate to the parents with little evidence of heterosis in any environment or for any harvest during the year. Although particular crosses produced high yields in certain environments, no general heterotic pattern was observed between the two proposed groups. A Gardner and Eberhart Analysis III indicated the presence of variation for both general combining ability and specific combining ability effects of approximately equal magnitude among these eight populations. The nondormant germplasm sources used here do not form a new heterotic group for midwestern U.S. alfalfa breeding programs.
Abbreviations: GCA, general combining ability • HS, half-sib • MG, mid-dormancy group • MP, midparent • N, nondormant-derived • S, semidormant • SCA, specific combining ability
|
INTRODUCTION |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONREFERENCES |
|---|
The yield increase in alfalfa has been estimated to be 0.15 to 0.30% per year (Hill et al., 1988; Holland and Bingham 1994). However, Riday and Brummer (2002a) suggested that as pest or pathogen pressures force a decline in yield of older varieties, yield improvements are observed relative to old cultivars but that new cultivars are actually only maintaining a constant yield level. Support for this hypothesis has been shown in a multilocation evaluation of yield trends (Lamb et al., 2006). One possible reason for the yield stagnation is that the focus of breeders on traits other than yield, such as pest resistance, has diverted effort away from explicit selection for yield and consequently limited the gain for yield (Hill et al., 1988).
The semihybrid breeding model could capture heterosis as a way to overcome current yield stagnation (Brummer, 1999). The key to success for such program is to identify possible heterotic groups that enable the realization of heterosis in a dependable manner when crossed. We have previously proposed three heterotic groups in the United States: M. sativa subsp. falcata, semidormant M. sativa subsp. sativa, and nondormant M. sativa subsp. sativa (Brummer, 1999). Heterosis between falcata and sativa has been demonstrated (Riday and Brummer, 2002a, 2005; Segovia-Lerma et al., 2004), and the potential of at least one source of nondormant alfalfa, the Peruvian germplasm, as a source of heterosis with semidormant alfalfa has also been suggested (Segovia-Lerma et al., 2004; Maureira et al., 2004). Despite the yield boost falcata provides to hybrids, it includes some undesirable agronomic characteristics, such as slow regrowth, early dormancy, and a decumbent grow habit (Riday and Brummer, 2002b). Therefore, focusing on nondormant germplasm, which has a more desirable palette of traits, may be a more reasonable direction for yield enhancement.
Because nondormant germplasm does not survive winter conditions in the midwestern United States, we developed winter hardy germplasm derived from nondormant cultivars to serve as a new gene pool for yield improvement (Weishaar et al., 2005). Following three cycles of phenotypic recurrent selection for decreased winter injury in central Iowa, populations derived from four nondormant alfalfa cultivars were developed that had winter injury similar to adapted cultivars and hence are able survive the Iowa winter. Because nondormant germplasm derives from different parental material than dormant or semidormant germplasm, we postulated that it may contain alleles for yield that are not present in the typical cultivars grown in the midwestern United States.
We hypothesized that the nondormant-derived germplasm may represent a new heterotic group for the midwestern United States. Therefore, the objective of this study was to test the hypothesis by making population crosses between adapted elite semidormant midwestern cultivars and the new nondormant-derived germplasm and evaluating the hybrids for biomass yield in field trials in Iowa.
|
MATERIALS AND METHODS |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONREFERENCES |
|---|
|
Experimental Design
Both direct-seeded and transplanted field experiments were evaluated. A seeded trial was planted on 23 Aug. 2003 at the Northeast Research Farm, south of Nashua, IA, which has a Readlyn loam soil (fine-loamy, mixed, superactive, mesic Aquic Hapludolls); a companion trial at Ames, IA, failed to establish due to drought. Seeds of each entry were directly planted using a hand planter into plots consisting of three 2.1-m rows spaced approximately 15 cm apart. The seeding rate was 0.75 g per plot (0.25 g or approximately 100 seeds per row). Plots were separated end-to-end by a 1-m fallow border and side-to-side by approximately 35 cm. Transplanted trials were established on 21 Apr. 2004 at the Agronomy and Agricultural Engineering Research Farm west of Ames, IA, on a Nicollet loam soil (fine-loamy, mixed, superactive, mesic Aquic Hapludolls) and at Nashua on 12 May 2004 in an area adjacent to the seeded trial. Each plot consisted of two 3-m-long rows separated by 15 cm; 10 plants spaced approximately 30 cm apart were planted in each row for a total of 20 plants per entry per plot. Plots were separated end-to-end by 1.5 m and side-to-side by 60 cm. The plot designs of all three experiments were 6 by 6 triple 
-lattices with each complete block consisting of six incomplete blocks each including six entries. Randomization for the block design was made using the Alphagen software program. Transplanted plots were mowed approximately monthly without data collection during 2004.
Biomass was harvested with a sickle bar harvester equipped with an electronic weigh system (Swift Machine and Welding Ltd., Swift Current, SK, Canada). Harvests were conducted on 8 June, 8 July, 10 August, and 8 September in 2004; and 31 May, 1 July, 1 August, and 29 August in 2005 for the seeded trial at Nashua. The transplanted trials were harvested on 20 May, 21 June, 25 July, and 22 August 2005 at Ames and on 31 May, 1 July, 1 August, and 29 August 2005 at Nashua. Several subsamples were randomly taken from clipped forage, weighed wet, dried for 5 d at 60 C, and then weighed dry. Whole plots were weighed wet during each harvest, and dry matter yield on a per plot basis was calculated based on the dry matter percentage averaged across subsamples.
Data Analysis
Replications and blocks were considered to be random effects, and entries were fixed. Data for the two experiments (transplanted trial and seeded trial) were analyzed separately. Least square means of each entry at each harvest and location were calculated using the MIXED procedure of the SAS statistical software package (SAS Institute, 1999; Littell et al., 1996).
Intergroup Comparisons
Our first analysis compared the different cross types with the parental populations. The 36 entries were divided into five groups: (i) S parents, (ii) N parents, (iii) S·S hybrids, (iv) N·N hybrids, and (v) S·N hybrids. Mean separations among the five categories were done using a least significant difference. Next, we computed two linear contrasts among means. The first contrast was made between the S·N hybrid mean and the average of the S and N parental means. The deviation percentage between the hybrid and parental means represents average midparent heterosis. The second contrast was made between the S·N hybrid mean and the average of the S·S and N·N hybrid means. The deviation percentage between these two groups represents average midgroup heterosis. The significance of deviation was estimated using contrast statements in SAS (Littell et al., 1996).
Our second analysis computed performance of hybrids both within and between dormancy groups and calculated high parent and midparent heterosis based on mean intra- vs. intergroup performance. This analysis is analogous to that described by Riday and Brummer (2002a); the only difference is that the former experiment compared hybrids of individual genotypes while we used population crosses in this experiment. First, we computed linear contrasts between the mean of each parental population's crosses within its dormancy group to the mean of that population's crosses to the other dormancy group. Second, to determine high parent heterosis on a family basis for each population, we computed linear contrasts between each parental population's inter-dormancy group crosses with the larger of either (i) the family mean of the parental population's intradormancy group crosses or (ii) the mean of the intradormancy group crosses (S·S or N·N) of the dormancy group in which the parental population was not found. The same process was also used for determining low parent negative heterosis on a family basis.
Third, the mean family heterosis of each of parental population can be formulated analogously to Riday and Brummer (2002a), shown briefly as follows:
 |
Overall Combining Ability Analysis
A further analysis of the data, which did not differentiate among parental populations on the basis of their dormancy classification, was conducted according to analyses II and III of Gardner and Eberhart (1966) as clarified by Murray et al. (2003) using a general linear model approach. Entries were divided into varieties (i.e., the parental populations), varieties vs. crosses (which is equivalent to average heterosis), and crosses. Cross effects were further subdivided into general combining ability (GCA) and specific combining ability (SCA). Variance components were computed using ANOVA.
Estimates of diallel effects and their formulas are described in terms of sample and population means. Entry means and their standard errors were computed using analysis of variance with the MIXED procedure (SAS Institute, 1999). The effects of varieties, heterosis, average heterosis, variety heterosis, GCA, and SCA were estimated based on contrasts between parent and cross means using a SAS macro program (Murray et al., 2003).
Throughout the results and discussion section, statistical significance is assessed at the 5% probability level unless indicated otherwise.
|
RESULTS AND DISCUSSION |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONREFERENCES |
|---|
|
Heterosis based on Cross Means and on Half-Sib Family Performance
We computed heterosis observed in S·N crosses in three ways. First, midparent heterosis (MP heterosis) was calculated as the percentage deviation of S·N crosses from the average S and N parental population means. We saw positive MP heterosis only in the seeded trial at Harvest 3 and in the transplanted trial for Harvest 1 and Harvest 4, but otherwise, no evidence of heterosis was present (Table 2). Second, we computed mid-dormancy group heterosis (MG heterosis) as the deviation of S·N hybrids from the average of S·S and N·N hybrids. No evidence for MG heterosis was observed in either trial (Table 2). Collectively, these two results suggest that these particular nondormant germplasm are not suited to repeatedly generating a heterotic response with elite semidormant cultivars.
Third, we analyzed heterosis based on half-sib family performance (HS heterosis) for each parental population, similar to the analysis done by Riday and Brummer (2002a). The HS heterosis values ranged from –18 to 18%. The highest HS heterosis value, 18%, was observed for 5454 in the seeded trial in 2004. Both Innovator+Z and 6420 exhibited positive HS heterosis in one trial in one environment (Table 3 ). The fourth semidormant cultivar, Vernal, produced highly negative HS heterosis (–18%) in the seeded trial of 2004 and (–8%) at Nashua in transplanted experiment. Among the nondormant-derived germplasm, only Magna 8 (C3) produced positive HS heterosis in any environment, 8% in the seeded trial of 2005. GT13R+(C3) and 5939(C3) produced negative HS heterosis of –9% and –6% (P = 0.10), respectively, in the Nashua transplanted experiment. No high or low parent HS heterosis was detected in any environment of either experiment.
|
Estimates of Genetic Effects of Yield
The entire diallel, without regard to the dormancy grouping, was analyzed using the Gardner and Eberhart analyses II and III (Gardner and Eberhart 1966; Murray et al., 2003). Average heterosis (
) was not present in any environment except marginally (P = 0.10) in the Nashua transplanted experiment (Table 4
), as expected based on the previous results. Variance among populations for GCA and SCA was greater than zero in three of four environments, although the Nashua transplanted experiment had marginal statistical significance for both effects (P = 0.10) (Table 4). In general, the variance for GCA effects was slightly larger than variance of SCA effects, except in the seeded experiment in 2005, when GCA effects were not present. In these plant materials, the results suggest that nonadditive genetic effects were nearly as important as additive genetic effects.
|
|
The lack of heterotic response is somewhat puzzling, given that the progenitors of the N populations were adapted to a different geographic region than the S cultivars and derived from different original germplasm introductions, which may suggest different alleles for yield in the two groups. However, previous hybridization results from New Mexico suggested that Peruvian and possibly Chilean germplasm may serve as the best combining germplasm with semidormant germplasm (Segovia-Lerma et al., 2004). Two molecular marker studies suggested that Peruvian is the most genetically distinct of the four nondormant germplasm sources (Peruvian, Chilean, African, and Indian) (Maureira et al., 2004; Kidwell et al., 1994). Only Magna 8 is known to contain Peruvian germplasm, and it also contains more Chilean germplasm than the others; it had a positive heterotic response for total yield in one of the four environments we examined (Table 3). Thus, perhaps the nondormant cultivars used here were not the best choices if developing population crosses that express yield heterosis is a main breeding goal. However, Segovia-Lerma et al. (2003) identified African as the nondormant population most genetically distinct from the major semidormant germplasm sources. Both GT13R+ and 5939 have significant proportions of African germplasm; they showed negative HS heterosis in the Nashua transplanted trial. Thus, relying on genetic distance analyses seems unlikely to identify desirable cross combinations.
In general, the adapted germplasm derived from nondormant sources produced more poorly than the semidormant cultivars, but hybrids were intermediate. We had previously noted that hybrids between semidormant genotypes and genotypes from M. sativa subsp. falcata tended to show midparent, and occasionally, high parent, heterosis (Riday and Brummer, 2002a). Thus, the better parent elevated the hybrid above the midparent value; that was not seen in this case. In the end, it seems odd that nondormant germplasm simply does not have the necessary complementary alleles that could lead a heterotic pattern when combined with midwestern germplasm predominantly derived from M. varia, Turkistan, and Flemish sources (Barnes et al., 1977; USDA, Agricultural Research Service, 2000).
Two other points should be made. First, one could argue that the lack of heterosis is expected because the populations are all very broadly based, making the requisite causes of heterosis, viz., directional dominance and divergent allele frequencies, unlikely to occur reliably. While this may be true, we expected that the N populations would have unique alleles not found in the S populations (or at least not in high frequency) and that the complementation of alleles from the different populations would result in heterosis. The hypothesis of different alleles for yield among dormancy groups could be tested using genetic mapping and genomics. Second, perhaps the process of selecting the N populations through three cycles in Iowa altered the allele frequencies so that they are now the same as those in the S populations. This is certainly a possible explanation of the results. We did not do any explicit selection for yield in forming the N populations; doing so may have resulted in different allele frequencies, but whether heterosis would have resulted is not known.
Both trials—one seeded in short rows and the other a moderately dense transplanted nursery—showed the same big picture of heterotic response: that is, that S·S crosses outyielded N·N crosses with S·N hybrids intermediate and not consistently expressing heterosis. On an individual cross basis, however, there is little similarity between the two nursery types. Thus, extrapolation of results from one nursery type to the other should be done cautiously.
That particular combinations of germplasm performed well in certain environments suggests that a weak heterotic response may be present among some of these germplasm but that this response does not necessarily accord to geographic source of origin. Therefore, a dedicated breeding program designed to create heterotic groups, such as using reciprocal half-sib recurrent selection, may be a better approach toward hybrid breeding in alfalfa than attempting to search for patterns in existing geographically distinct germplasm. Working with populations with a more narrow genetic base may also make yield gains, both through direct selection and through hybrid breeding, more attainable.
|
ACKNOWLEDGMENTS |
|---|
|
NOTES |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONREFERENCES |
|---|
Received for publication December 28, 2006.
|
REFERENCES |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONREFERENCES |
|---|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| 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 | |||
