Published online 22 January 2007
Published in Crop Sci 47:11-18 (2007)
© 2007 Crop Science Society of America
677 S. Segoe Rd., Madison, WI 53711 USA
CROP BREEDING & GENETICS
Genetic Mapping Forage Yield, Plant Height, and Regrowth at Multiple Harvests in Tetraploid Alfalfa (Medicago sativa L.)
Joseph G. Robinsa,
Gary R. Bauchanb and
E. Charles Brummerc,*
a USDA-ARS Forage and Range Research Lab., Logan, UT 84322
b USDA-ARS Soybean Genomics and Improvement Lab., Beltsville, MD 20705
c Raymond F. Baker Center for Plant Breeding, Iowa State Univ., Ames, IA 50011; E.C. Brummer current address: Center For Applied Genetic Technologies, Univ. of Georgia, Athens, GA 30602
* Corresponding author (brummer{at}uga.edu)
 |
ABSTRACT
|
|---|
Hybrids between Medicago sativa subsp. falcata and M. sativa subsp. sativa can result in high levels of heterosis for alfalfa forage production. However, commercially viable alfalfa cultivars also must have acceptable performance for other agronomic traits, including regrowth following harvest and appropriate autumn dormancy. In this study, we characterized an F1 population derived from the cross of the two subspecies for the genetics of forage production, forage height, and forage regrowth at three harvests in two Iowa locations for 1 yr. Quantitative trait loci (QTLs) controlling these traits were mapped on a genetic linkage map developed from this population. Broad-sense heritabilities and genetic correlations among the traits were high. Marker alleles associated with QTLs for these traits were contributed by each parental genome, suggesting that both subspecies of cultivated alfalfa may contribute to the improvement of these traits. Multiple regression models accounted for between 11 and 44% of the trait variation. The most marker–trait associations were detected during the first harvest. Marker effects were generally similar across harvests. Markers associated with autumn regrowth tended to be different from those associated with regrowth earlier in the season, suggestive of the identification of autumn dormancy effects. The winter hardiness gene MsaciB is a candidate locus for yield and regrowth, being associated with yield across three harvests and with autumn regrowth. A subset of alleles was associated with more than one of the traits. Major alleles for the three traits appear to reside primarily on linkage groups (LGs) 3, 4, 5, and 7.
Abbreviations: LG, linkage group • QTL, quantitative trait loci
|
INTRODUCTION
|
|---|
ALTHOUGH ALFALFA (Medicago sativa L.) is the most important forage crop in North America (Barnes et al., 1988), breeding efforts over the last 20 years have resulted in little to no gain in alfalfa forage production in the upper Midwestern USA (Riday and Brummer, 2002a). Yield gains that have occurred appear to be primarily due to improved disease resistance (Lamb et al., 2006). Semihybrid cultivars, developed by hybridizing two distinct populations in a foundation seed field, are one possible method to increase yield (Brummer, 1999), and high levels of heterosis for alfalfa forage production have been identified between particular crosses of the two alfalfa subspecies M. sativa subsp. sativa and M. sativa subsp. falcata (Riday and Brummer, 2002a, 2005). The former provides the elite, purple-flowered germplasm typically used in applied alfalfa breeding programs; the latter has the typically unimproved, yellow-flowered germplasm that has received less attention by breeders in the USA outside of the northern Great Plains (Berdahl et al., 1989). Although M. sativa subsp. falcata germplasm has many desirable characteristics, including excellent winter hardiness, it also has slow regrowth following harvest, pronounced autumn dormancy, and prostrate growth, all of which negatively affect hybrid progeny (Riday and Brummer, 2002b, 2004). Thus, while the exploitation of heterosis between crosses of these two subspecies has the potential for improving alfalfa forage production, widespread use of M. sativa subsp. falcata germplasm will not occur until intrapopulation improvement of other important agronomic characters has been accomplished.
Desirable alfalfa cultivars should have not only high forage production, but also rapid regrowth after harvest and erect tall growth to ease mechanical harvesting. Although each of these traits would be desirable in alfalfa cultivars, their genetic characterization and the genetic relationships among them are not well documented. Genetic dissection of these traits may enable the development of a more efficient selection program. Like many traits of agronomic importance, forage production, height, and regrowth exhibit continuous variation and thus are quantitative traits likely controlled by many genes. If QTL mapping could identify the regions of the alfalfa genome that control these traits, the possibility of using a marker-assisted selection program to aid introgression of wild germplasm and to improve selection efficiency becomes possible.
The objective of this study was twofold: (1) to characterize forage production, height, and regrowth at multiple harvest dates and their correlations in a tetraploid alfalfa population derived from a M. sativa subsp. falcata x subsp. sativa cross and (2) to identify the genomic regions underlying these traits and the correspondence among the locations for the three traits using the molecular genetic map and marker data created for this population (Robins et al., 2007).
|
MATERIALS AND METHODS
|
|---|
Plant Materials
Two parental genotypes, ABI408 and WISFAL-6, were used. ABI408 (ABI Alfalfa, Inc., Lenexa, KS) represents elite M. sativa subsp. sativa germplasm and exhibits good phenotypic performance for forage yield and resistances to multiple pests. WISFAL-6 represents semi-improved M. sativa subsp. falcata germplasm from the WISFAL germplasm (Bingham, 1993). The latter typically exhibits good winter hardiness but expresses a high level of autumn dormancy and poor forage production. Crossing WISFAL-6 (as the female parent) and ABI-408 (as the male parent), with vacuum emasculation, resulted in a segregating, full sib F1 population consisting of 200 genotypes (Brummer et al., 2000). We then clonally propagated the two parent genotypes, the 200 F1 genotypes, and 10 check genotypes in the greenhouse at Ames, IA, during the winter of 1997–1998.
Experimental Design
The location of field experiments and transplanting dates were the Agronomy and Agricultural Engineering Research Farm west of Ames, IA, on 19 May 1998, and the Northeast Research Farm south of Nashua, IA, on 22 May 1998. The experimental design at both locations was a quadruple 
-lattice consisting of 4 complete blocks. Within each complete block were 15 incomplete blocks each containing 14 plots, for a total of 840 plots in the entire experiment. Each plot consisted of five clones of the assigned genotype. The spacing was 30 cm between plants within a plot, 60 cm between plots in the same row, and 75 cm between rows.
Data Collection
Plots were harvested three times in 1999 (early June [H1], early July [H2], and early September [H3]) by removing all aboveground biomass with rice sickles to 
7.5 cm above the soil surface. The forage from each plot was weighed in the field using a milk scale. Subsamples consisting of several handfuls of forage were collected from each plot, weighed wet, dried in forced-air driers at 60°C for 5 d, and weighed dry to determine moisture percentage. The plot wet weights were adjusted using the average moisture percentage of all plots to determine dry matter yield. The number of plants in each plot was counted 1 wk after each harvest, and dry matter (g plant–1) of each plot was determined.
Preceding each harvest, the tallest and shortest clone in each plot were measured (cm) and averaged for a plot height value. Two to 3 wk following each harvest, regrowth was determined by measuring the tallest and shortest clone in each plot (cm) and averaging the values. First harvest regrowth was not measured at Nashua. Thus, the height of regrowth assessed the rapidity of growth following harvest, whereas the measurement at harvest documented mature plant height.
Phenotypic Data Analysis
The MIXED procedure of the SAS statistical software package (Littell et al., 1996) was used to calculate least-square means for each of the measured traits on a harvest basis. The model designated complete blocks (nested within locations) and incomplete blocks (nested within complete blocks and locations) as random effects. Genotypes, locations, and the two-way genotype-by-location interaction were fixed effects. Although the genotype-by-location effect was generally statistically significant, it was several times smaller in magnitude than the genotype effect (data not shown); therefore, we combined data across both locations for analysis.
Parent and check cultivars were removed and the progeny data were analyzed using an all-random model and the MIXED procedure of SAS (Littell et al., 1996; SAS Institute Inc., 1999) to calculate variance components of genotype, genotype–environment interaction, and residual error along with the corresponding variance–covariance matrices. The variance–covariance matrices were then entered into the IML procedure of SAS to calculate the heritabilities (Holland et al., 2003) and standard errors of the heritabilities. Heritability was defined both on an individual plot basis and on an entry mean basis (i.e., across all replications at both locations). Because the phenotypic data came from clonal ramets of a full sib population, the estimates of heritability are broad-sense estimates. The genetic variation includes all the components corresponding to an autotetraploid species, including higher-order intralocus (digenic, trigenic, and quadragenic) and epistatic interactions (Rumbaugh et al., 1988).
The MIXED and IML procedures were also used to calculate phenotypic and genetic correlations, with corresponding standard errors (Holland, 2006). The complete model was analyzed with genotype, location, and their interaction being fixed, and the variance–covariance structure was calculated. For each harvest pairwise correlations were calculated between each of the three traits. Then, for each trait separately, correlations were calculated between the three harvests (e.g., the correlation of yield betwen H1 and H2).
Mapping and Markers
All mapping procedures were described by Robins et al. (2007). Marker alleles are identified by the marker name (see Robins et al. 2007 for probe and primer sources) followed by a letter indicating the parental genome that contributed the allele ("a" from WISFAL-6, "b" from ABI408, or "c" from both parents) and a distinct number for each individual allele.
Marker–Phenotype Associations
Marker–phenotype associations, using the genetic linkage map described previously (Robins et al., 2007), were identified using single-factor analysis of variance. For each marker allele and trait, the least-square means of individuals containing the allele were contrasted with those of individuals without the allele. Alleles were declared to be significantly associated with a trait at a p value 
0.01. Because this is the first study examining potential QTL for these traits in alfalfa, we were more concerned with identifying genomic regions possibly associated with the trait than with false-positive identification. However, to identify associations with stronger statistical support, we also used a nonparametric permutation test (Churchill and Doerge, 1994) based on the familywise error rate of Westfall and Young (1993).
Multiple regression models developed using the REG procedure of SAS with the stepwise selection option were used to determine markers that best explained the trait variation. Alleles that were identified as being associated with one of the traits based on single-marker analyses were placed in the model, and those alleles that explained the most phenotypic variation (at a p value 0.05) were retained. All QTL analyses were conducted on each harvest–trait combination.
|
RESULTS
|
|---|
The results described in this article represent analyses by harvest averaged over two Iowa locations for a single year; the total biomass production for this year was examined in a previous article (Robins et al., 2007). We limited this analysis to the first postestablishment year because these data had smaller standard errors than the data of subsequent years due to hand harvesting and fewer effects of plant mortality.
Quantitative Genetic Analysis of Forage Yield, Forage Height, and Forage Regrowth
The population exhibited substantial variation for most traits (Table 1). For each trait, transgressive segregants were present in both directions except for regrowth following H1 and H2. In cases where the parents differed, ABI408 generally was more agronomically desirable—higher production, taller, and faster regrowing—than WISFAL-6, except that WISFAL-6 had higher production in H3. The population mean value for height and yield was generally similar to that of the ABI408 parent and was higher than that of WISFAL-6 for H1 and H2 yield and H2 height. WISFAL-6 outyielded the population mean for H3. The improved performance of WISFAL-6 over ABI408 and the population mean during H3 was likely due to the long interval between H2 and H3, which would have given the falcata parent adequate time to compensate for its generally slower regrowth.
View this table:
[in this window]
[in a new window]
|
Table 1. Mean values of yield, plant height, and regrowth, heritabilities, and variance components for each of three harvests during 1999 averaged across two Iowa locations.
|
|
Heritabilities were moderate to high for all traits at all harvests (Table 1). While these are broad-sense heritabilities, the high values indicate the relative importance of genetic factors compared to environmental factors on the traits of interest. This was also true of the plot-basis heritabilites, although these values were lower. The importance of dominant gene action on the heritability could not be determined. Thus, even if the results of this study were considered as characteristic of all M. sativa subsp. falcata x sativa crosses, estimating the potential gain from selection on these traits would be difficult.
Genetic variance components were generally small, although the variance associated with forage production during H1 was large (Table 1). The variance associated with regrowth was always small, reflecting the relatively narrow range of values for this trait in the population. The variance associated with genotype was approximately 2 to 30 times larger than the variance associated with the genotype-by-location interaction, depending on the harvest, supporting our decision to pool data across locations for analysis.
Phenotypic and Genotypic Correlations
Phenotypic correlations between yield and height were moderate at each harvest (Table 2). Phenotypic correlations between yield and regrowth were consistently positive but low across all three harvests. A similar pattern was observed between height and regrowth, although the correlations tended to be stronger than those between yield and regrowth. In each case, the genetic correlations were higher than the corresponding phenotypic correlations (Table 2). Most of the genetic correlations were positive and high (rG > 0.70), suggesting that selection for any one of these traits would likely improve the others as well.
View this table:
[in this window]
[in a new window]
|
Table 2. Phenotypic and genetic correlations among yield, plant height, and regrowth measured at each of three harvests during 1999 averaged across two Iowa locations.
|
|
Correlations between harvests for yield and height showed that, in general, the highest values were observed between H1 and H2, followed by those between H2 and H3 (Table 3). The lowest correlations were between the data from H1 and H3. The phenotypic correlations of regrowth between harvests were generally low (r < 0.32). In all instances, the genetic correlations were higher than the phenotypic correlations.
View this table:
[in this window]
[in a new window]
|
Table 3. Phenotypic and genetic correlations among harvests for yield, plant height, and regrowth during 1999 averaged across two Iowa locations.
|
|
Marker–Phenotype Associations
The single-marker analysis identified 32 alleles associated with yield, 23 with height, and 31 with regrowth during at least one of the three harvests (Tables 4
–6, Fig. 1
). For each trait, both parents contributed alleles with positive and negative phenotypic effects. For height, all alleles with negative effects either were contributed by ABI408 or were present in both parents. In the latter case, whether WISFAL-6 contributed alleles for shorter plants is unknown because we cannot determine if both parents also contained the QTL.
View this table:
[in this window]
[in a new window]
|
Table 4. Marker alleles associated with yield at each of three harvests during 1999 averaged across two Iowa locations with the allele effects (g plant–1) based on single-marker analysis of variance.
|
|
View this table:
[in this window]
[in a new window]
|
Table 5. Marker alleles associated with plant height at each of three harvests during 1999 averaged across two Iowa locations with the allele effects (cm) based on single-marker analysis of variance.
|
|
View this table:
[in this window]
[in a new window]
|
Table 6. Marker alleles associated with regrowth at each of three harvests during 1999 averaged across two Iowa locations (except harvest 1, which is in Ames only) with the allele effects (cm) based on single-marker analysis of variance.
|
|
View larger version (77K):
[in this window]
[in a new window]
|
Fig. 1. Consenus genetic linkage maps of F1 progeny of the cross between WISFAL-6 and ABI408. Loci containing alleles associated with height, regrowth, or yield in at least one harvest are listed in boldface type and noted by the trait(s) with which they were associated (H, height; R, regrowth; Y, yield) to the right.
|
|
All but three alleles associated with H2 yield were also identified in H1. Four of seven alleles identified in H3 were not associated with H1 or H2 yield. Interestingly, the marker uga083 was associated with yield in all three harvests, but by an allele with negative effect from ABI408 in H1 and H2 and by an allele with positive effect from WISFAL-6 in H3 (Table 4). For both plant height and regrowth, a suite of different alleles was identified at each harvest, with limited overlap (Tables 5 and 6). Eighteen alleles were associated with autumn regrowth (H3), but only eight were identified for each of the other regrowth periods. This result was likely observed because regrowth at this time period, unlike the others, was influenced not only by inherent differences in regrowth potential among genotypes but also by differences among genotypes in autumn dormancy.
Although a number of alleles—18 for yield, 7 for height, and 3 for regrowth—were associated with the traits during more than one harvest, only 4 alleles—3 for yield and 1 for height—were associated with a trait for all three harvests. The alleles identified across multiple harvests tended to fall on the LGs with the largest phenotypic effects, particularly LGs 3, 4, and 7 for yield, LGs 5 and 7 for height, and LG 5 for regrowth. To identify those QTLs with the strongest statistical support, a nonparametric permutation test was used to control the familywise error rate. Although many of the alleles previously identified did not remain after the permuation test, at least one allele was identified for each of the traits of interest (Tables 4
–6).
Common Markers for Multiple Traits
A number of alleles on LGs 3, 4, 5, and 7 were associated with QTLs for more than one of the traits analyzed in this study (Table 7). Two alleles on LG 7 from ABI408 (bn2_21e3v14b2 and uga772b1) and one on LG 4 from both parents (aa660573c3) showed associations with both yield and height during more than one harvest. Alleles associated with multiple traits were typically found for yield and height or for height and regrowth, but only rarely for the more agronomically desirable combination of yield and regrowth. In only two instances (aa660573c3 and uga671a1) did alleles have a positive effect on both yield and regrowth during the same harvest. The remaining alleles associated with QTLs for both yield and regrowth typically exhibited the association in different harvests. Alleles negatively associated with yield in H1 or H2 were positively associated with H3 (autumn) regrowth. Because increasing both yield and regrowth are desirable, this relationship is unwelcome but could be due to linkage or pleiotropy.
View this table:
[in this window]
[in a new window]
|
Table 7. Alleles associated with more than one trait across three harvests during 1999, averaged across two Iowa locations.
|
|
Multiple Regression Models
Alleles remaining in multiple regression models explained between 11% (H2 regrowth) and 44% (H1 yield) of the phenotypic variation associated with the traits (Table 8). The allele afct45a1 from LG 7 of WISFAL-6 explained the largest amount of variation for any trait (20% of H1 yield). LGs 1 and 8 were the only LGs that did not contribute an allele contained in the models for at least one of the harvest–trait combinations. Consistent with the single-marker analysis, it appears that LG 7 plays an important role in controlling these traits.
View this table:
[in this window]
[in a new window]
|
Table 8. Multiple regression models and the cumulative phenotypic variance explained by the model for each harvest and trait combination, partial R2 for each marker, and p value of markers remaining in the models.
|
|
|
DISCUSSION
|
|---|
Given the moderate to high broad-sense heritability estimates we obtained, each of these traits is under strong genetic control in this population. Although substantial variation existed for genotypic effects for each of the traits from each harvest, in general the variance components were not large, particularly for regrowth (Table 1). Even though the F1 population mean never significantly outperformed either the mid-parent or high parent for any of the traits, transgressive segregation was present for most of the traits at most harvests.
In this study, both phenotypic and genetic correlations between the three traits were generally moderate to high (Table 2). A previous analysis of relationships among traits, conducted on hybrid progeny of crosses between M. sativa subsp. falcata and M. sativa subsp. sativa, found highly positive phenotypic correlations between height and regrowth, low to moderate positive correlations between yield and height, and no correlation between yield and regrowth (Riday and Brummer, 2005). Increasing each of these traits is a goal of most applied alfalfa breeding programs, so the positive correlations are favorable. Because genetic correlations are due to either linkage or pleiotropy, at least some genomic regions underlying these traits are likely common to all three of the traits. Similarly, the positive genetic correlations for the same trait between harvests suggested similar genetic control for the trait throughout the year (Table 3).
This first experiment examining height and regrowth QTLs and their connection with yield has identified regions on each of the eight alfalfa LGs that were associated with at least one of the traits in at least one harvest. Not surprisingly given the genetic correlations, we identified three markers associated with yield, height, and regrowth, nine with both yield and height, five with both yield and regrowth, and five with both height and regrowth. Of particular interest is LG 7, in which numerous QTLs for yield, plant height, and regrowth were identified across harvests. One particular marker, MsaciB, had a WISFAL-6-derived allele negatively associated with biomass yield in all three harvests. This allele was also positively associated with regrowth in only the third harvest. The MsaciB gene is known to be related to winter hardiness (Monroy et al., 1993) and could serve as a candidate gene for autumn regrowth and for yield.
Both parents contributed alleles with positive and negative effects for yield on LG 7. Given our marker-analysis approach, we cannot determine the number of loci involved with yield on this LG. However, it appears, based on marker locations, that more than one locus is likely present. In either parent, the marker alleles linked to the negative and positive trait alleles were contained on separate cosegregation (homologous) groups (data not shown). Thus, the heterozygosity present in alfalfa means that a locus (or set of linked loci) may contain QTL alleles with differing effects spread across homologs. Linkage groups 3 and 4 also contained markers associated with yield in more than one of the harvests. Although the magnitude of their effect in this population appeared lower than the loci on LG 7, they may be starting points for dissection of yield in other populations given the regularity with which they are associated with the trait.
The identification of the same markers for yield at multiple harvests in this experiment support the contention in our previous article (Robins et al., 2006) that markers for total biomass yield that were identified only in this year (1999) are truly linked to QTLs and not spurious associations. Seventeen loci were associated with total biomass yield in 1999 in Iowa (Robins et al., 2007), and all were also detected in this experiment. Of these 17, 2 (MsaciBb1 and aw693871c3) were detected only in a single harvest; the remaining 15 alleles were detected in two or three harvests. Thus, even though markers may be associated with biomass yield only in a single year, within that year, marker–trait associations at multiple harvests lend credence to the validity of the QTL association.
Because genetic correlations among harvests were high for each of the three traits, we expected several alleles to be associated with a single trait across harvests. This was not always observed, especially with yield between H1 and H3 or between H2 and H3. Several reasons can explain these observations. First, at least some different loci may be controlling yield throughout the year, producing allele-by-harvest interactions. If this is the case, then marker-assisted selection would need to consider different loci to improve total seasonal yield. However, over half of the total yield was produced at first harvest in this population and year, so markers for first harvest yield are probably of most interest. Second, given that heritabilities declined in H2 and H3 compared to H1, we can infer that some of the associations we observed at H1 might not be found in later harvests due to an increase in uncontrolled variation among individuals. Thus, the QTLs found at one harvest might not be detected due to phenotyping limitations later in the year. Third, some of the marker–trait associations may be false positives. Those associations found only at one harvest for one trait, and not linked to other markers also showing associations with the trait, could be falsely identified. While caution is warranted about some of these associations, given that this is the first study examining QTLs for these traits, it is useful to cast a wide net to identify possible genomic regions involved in expression of them.
Linkage group 7 also contains trait–allele associations for height and regrowth, in at least some environments and harvests. Several other LGs—LGs 3, 4, and 5—contain QTLs for height and regrowth in one or more harvests. Because different associations are present at different harvests, the best candidate markers to use in a marker-assisted selection program to improve these traits is unclear. Regrowth, in particular, appears to be controlled by different regions of the genome depending on the time of year being considered.
Marker alleles associated with multiple traits tended to have effects in a similar direction (positive or negative) in all traits, clearly a useful attribute from a selection perspective. A notable exception to this trend is with alleles on LG 7 from WISFAL-6. Several alleles exhibited a negative association with yield in at least one of the harvests, but a positive association with regrowth during H3. For at least these alleles, improvement of one of these traits may not necessarily result in improvement for the other traits and could potentially result in decreased performance.
Some caveats are in order regarding these results. First, we are examining a single population derived from only two genotypes, one of which is from a M. sativa subsp. falcata population not widely used in commercial alfalfa breeding. Thus, other loci and other alleles are likely to be present in different genetic backgrounds. An association mapping approach to quantify the presense of QTLs within breeding populations is an obvious next step. Second, single-marker analysis can identify genomic regions associated with complex traits, but because it confounds recombination with genotypic value, it cannot precisely localize QTLs on the genetic map or clearly determine the number of QTLs in a region (Bernardo, 2002). Third, this experiment was conducted on spaced plants, not sward seedings, which may bias the results in terms of improvement that would be seen in actual commercial plantings. Ultimately, more experimentation is needed both to confirm the QTLs identified here, as well as to more precisely locate them to use them in marker-assisted selection.
|
ACKNOWLEDGMENTS
|
|---|
We thank Heathcliffe Riday, Meenakshi Santra, M. Maroof Shah, Cara Lea Council, and Ryan Kunz for assistance. This research was supported by USDA-NRI (97-35300-4573), USDA-IFAFS (00-52100-9611), and Hatch Regional Research Project NE-1010 (all to E.C.B.) and an Iowa State University Plant Sciences Institute Fellowship (to J.G.R.).
Received for publication July 7, 2006.
|
REFERENCES
|
|---|
- Barnes, D.K., B.P. Goplen, and J.E. Baylor. 1988. Highlights in the USA and Canada. In A.A. Hanson et al. (ed.) Alfalfa and alfalfa improvement. Agron. Monogr. 29. ASA, CSSA, and SSSA, Madison, WI.
- Berdahl, J.D., A.C. Wilton, and A.B. Frank. 1989. Survival and agronomic performance of 25 alfalfa cultivars and strains interseeded into rangeland. J. Range Manage. 42:312–316.
- Bernardo, R. 2002. Breeding for quantitative traits in plants. Stemma Press, Woodbury, MN.
- Bingham, E.T. 1993. Registration of WISFAL alfalfa (M. sativa subsp. falcata) tetraploid germplasm derived from diploids. Crop Sci. 33:217–218.[Free Full Text]
- Brummer, E.C. 1999. Capturing heterosis in forage crop cultivar development. Crop Sci. 39:943–954.[Abstract/Free Full Text]
- Brummer, E.C., M.M. Shah, and D. Luth. 2000. Reexamining the relationship between fall dormancy and winter hardiness. Crop Sci. 40:971–977.[Abstract/Free Full Text]
- Churchill, G.A., and R.W. Doerge. 1994. Empirical threshold values for quantitative trait mapping. Genetics 138:963–971.[Abstract]
- Holland, J.B. 2006. Estimating genotypic correlations and their standard errors using multivariate restricted maximum likelihood estimation in SAS Proc MIXED. Crop Sci. 46:642–654.[Abstract/Free Full Text]
- Holland, J.B., W.E. Nyquist, and C.T. Cervantes-Martinez. 2003. Estimating and interpreting heritability for plant breeding. Plant Breed. Rev. 22:9–111.
- Lamb, J.F.S., C.C. Sheaffer, L.H. Rhodes, R.M. Sulc, D.J. Undersander, and E.C. Brummer. 2006. Five decades of alfalfa cultivar improvement: Impact on forage yield, persistence, and nutritive value. Crop Sci. 46:902–909.[Abstract/Free Full Text]
- Littell, R.C., G.A. Milliken, W.W. Stroup, and R.D. Wolfinger. 1996. SAS system for mixed models. SAS Institute, Cary, NC.
- Monroy, F.A., Y. Castonguay, S. Laberge, F. Sarhan, L.-P. Vézina, and R.S. Dhindsa. 1993. A new cold-induced alfalfa gene is associated with enhanced hardening at subzero temperature. Plant Physiol. 102:873–879.[Abstract]
- Riday, H., and E.C. Brummer. 2002a. Forage yield heterosis in alfalfa. Crop Sci. 42:716–723.[Abstract/Free Full Text]
- Riday, H., and E.C. Brummer. 2002b. Heterosis of agronomic traits in alfalfa. Crop Sci. 42:1081–1087.[Abstract/Free Full Text]
- Riday, H., and E.C. Brummer. 2004. Morphological variation of Medicago sativa subsp. falcata genotypes and their hybrid progeny. Euphytica 138:1–12.[CrossRef]
- Riday, H., and E.C. Brummer. 2005. Relationships among biomass yield components within and between subspecies of alfalfa [Online]. Available at http://www.medicago-reports.org/ (verified 22 Nov. 2006). Medicago Genetic Reports. Vol. 4.
- Robins, J.G., D. Luth, T.A. Campbell, G.R. Bauchan, C. He, D.R. Viands, J.L. Hansen, and E.C. Brummer. 2007. Genetic mapping of biomass production in tetraploid alfalfa (Medicago sativa L.). Crop Sci. 47:1–10.[Abstract/Free Full Text]
- Rumbaugh, M.D., J.L. Caddel, and D.E. Rowe. 1988. Breeding and quantitative genetics. In A.A. Hanson et al. (ed.) Alfalfa and alfalfa improvement. Agron. Monogr. 29. ASA, CSSA, and SSSA, Madison, WI.
- SAS Institute, Inc. 1999. SAS/STAT user's guide, version 8. SAS Institute, Inc., Cary, NC.
- Westfall, P.H., and S.S. Young. 1993. Resampling-based multiple testing: Examples and methods for p-value adjustment. John Wiley & Sons, New York.
This article has been cited by other articles:
|
|
|
|
 
J. J. Ariss and G. J. Vandemark
Assessment of Genetic Diversity among Nondormant and Semidormant Alfalfa Populations Using Sequence-Related Amplified Polymorphisms
Crop Sci.,
November 7, 2007;
47(6):
2274 - 2284.
[Abstract]
[Full Text]
[PDF]
|
|
|
TOP
ABSTRACT
INTRODUCTION
