Crop Science Grow Your Career with CSSA
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

Published online 24 January 2006
Published in Crop Sci 46:303-311 (2006)
© 2006 Crop Science Society of America
677 S. Segoe Rd., Madison, WI 53711 USA
This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF) Free
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Stendal, C.
Right arrow Articles by Jung, G.
Right arrow Search for Related Content
PubMed
Right arrow Articles by Stendal, C.
Right arrow Articles by Jung, G.
Agricola
Right arrow Articles by Stendal, C.
Right arrow Articles by Jung, G.
Related Collections
Right arrow Other Forage Crops
Right arrow Crop Genetics

CROP BREEDING, GENETICS & CYTOLOGY

Marker-Assisted Selection for Neutral Detergent Fiber in Smooth Bromegrass

C. Stendala, M. D. Caslerb,* and G. Jungc

a Dep. of Agronomy, Univ. of Wisconsin, Madison, WI 53706-1597
b USDA-ARS, U.S. Dairy Forage Research Center, 1925 Linden Dr. West, Madison, WI 53706-1108
c Dep. of Plant Pathology, Univ. of Wisconsin, Madison, WI 53706

* Corresponding author (mdcasler{at}wisc.edu)


   ABSTRACT
TOP
 NOTES
 ABSTRACT
INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
 REFERENCES
 
Neutral detergent fiber (NDF) is the most effective and measurable predictor of animal intake. About 70% of the variation in animal production is attributed to differences in intake potential of livestock feed. Sixteen molecular markers associated with NDF in four divergently selected smooth bromegrass (Bromus inermis L.) populations were selected from a previous study for further examination for possible use in a marker-assisted selection (MAS) breeding program. The objectives of this experiment were to confirm the association of previously identified markers to NDF concentration in smooth bromegrass populations and to develop a marker-based selection index. Marker and NDF data were analyzed and marker indices were constructed using data from the current and previous studies. Index scores were used to rank genotypes and create selection differentials on the basis of phenotypic data. Marker frequencies calculated on a subpopulation basis between the previous and current studies were highly unrepeatable. Nevertheless, of the 14 groups of marker indices, seven of which accounted for pedigree structure, one marker index appears to provide the greatest potential for use in MAS to reduce NDF across all four populations of smooth bromegrass. Where pedigree structure is known, selections made using this index would lead to the largest expected response per year and eliminate the need to collect phenotypic data for as long as linkage relationships remain intact. Where pedigree structure is unknown, a general marker index may be used, or phenotypic data can be utilized along with a marker index. However, inclusion of phenotypic data would necessitate a cost–benefit analysis.


   INTRODUCTION
TOP
 NOTES
 ABSTRACT
 INTRODUCTION
MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
 REFERENCES
 
SMOOTH BROMEGRASS is a cool-season forage grass that was introduced to North America in 1884. It gained recognition during the droughts of the 1930s because of its ability to survive these conditions and its subsequent use in revegetation of the Great Plains and midwestern USA (Casler et al., 2000). Smooth bromegrass is an auto-allo octoploid (2n = 8x = 56) (Vogel et al., 1996) that displays self-incompatible reproduction (McKone, 1985). Fifty years of breeding has increased its forage yield by 5 to 10% (Vogel et al., 1996). The slow progress in the breeding process is most likely due to a combination of complex inheritance, restricting genetic gains, as well as a lack of concentrated breeding efforts (Casler et al., 2000).

Breeding for forage quality traits should improve the quality of livestock products as well as the efficiency of livestock production (Burton, 1973). Forage quality is a product of intake and digestibility, the former being of more importance (Fahey and Hussein, 1999, Mott and Moore, 1969). About 70% of the variation in animal production is attributed to differences in intake potential of livestock feed (Crampton et al., 1960). Research reviewed by Minson and Wilson (1994) revealed a negative correlation between voluntary forage intake and forage fiber concentration.

Coarse insoluble fiber, corresponding to neutral detergent fiber (NDF), is necessary for promoting rumen function (Van Soest et al., 1991). The NDF fraction, consisting of cellulose, hemicellulose, and lignin (Van Soest et al., 1991), is the most effective and measurable predictor of animal intake (Van Soest, 1994). A higher level of animal preference, or palatability, is also associated with lower neutral detergent fiber concentration (Falkner and Casler, 1998). The concentration of NDF is a heritable trait and genetic progress has been achieved in smooth bromegrass (Casler, 1999); reed canarygrass, Phalaris arundinacea L. (Surprenant et al., 1988); and timothy, Phleum pratense L. (Claessens et al., 2004). Casler (1999) reported the realized heritability for NDF concentration in smooth bromegrass as 0.31. Divergent selection for NDF concentration in smooth bromegrass resulted in a direct response of 0.7 to 1.3% of population means (Casler, 2002). Reducing NDF concentration may lead to increased voluntary intake by livestock (Minson and Wilson, 1994) and increased particle breakdown in the rumen, allowing digesta to pass more quickly through the rumen, thereby decreasing the time needed to stimulate appetite (Van Soest, 1994).

Significant forage quality improvements have been made in smooth bromegrass by phenotypic recurrent selection methods (Casler and Vogel, 1999; Casler, 1999; Diaby and Casler, 2005). Recurrent selection is useful for a forage trait such as NDF with low heritability because it will utilize much of the additive genetic variance in a population (Hallauer and Miranda, 1991, p. 169). Recurrent selection is accomplished by evaluating many plants for the trait or traits of interest, selecting the superior performing plants, then intermating these plants, thereby increasing the frequency of favorable alleles in subsequent generations (Comstock, 1996, p. 55). This method has been widely applied and shown measurable results for many traits in maize (Buendgen et al., 1990; Stojsin and Kannenberg, 1994). Nevertheless, phenotyping hundreds or thousands of plants for any forage quality trait, including NDF, is expensive and very time consuming, sometimes adding an extra year to each recurrent selection cycle.

Lande and Thompson (1990) stated that molecular markers could be used to increase the rate of improvement of economically important quantitative traits over that of phenotypic selection alone. Furthermore, their research showed that the efficiency of MAS is greatest for characteristics with low heritability. Xie and Xu (1998) concluded that marker-aided recurrent selection could speed the breeding process by the selection of immature plants and increase the efficiency of selection over recurrent phenotypic selection when phenotypic traits are difficult or costly to measure. Dudley (1993) suggested first identifying associations between marker alleles and quantitative trait loci (QTL), then using those associations to develop a selection scheme.

Previous research used random amplified polymorphic DNA (RAPD) to detect polymorphic markers to associate with QTL for NDF in divergently selected smooth bromegrass populations (Diaby and Casler, 2005). These authors identified numerous examples of significant correlated selection responses. One marker (AG14.0825) accounted for much of the variation in NDF across selection cycles in all four germplasm pools, while other markers explained much of the variation across selection cycles in only one or two populations. The markers identified by these authors as having the greatest potential for MAS had significant population-specific or generalized selection responses and were relatively unaffected by drift. Sixteen of these RAPD bands, varying for direction of selection response and effects of drift and selection on marker frequencies, were selected for further examination because of significant associations with NDF concentration (Diaby and Casler, 2005). The objectives of this experiment were to confirm the association of these molecular markers to NDF concentration in smooth bromegrass populations and to develop a marker-based selection index. Because these markers were identified at the population level, their utility for MAS must be confirmed before application in a selection program.


   MATERIALS AND METHODS
TOP
 NOTES
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
RESULTS AND DISCUSSION
 CONCLUSIONS
 REFERENCES
 
Germplasm and Pedigree
Divergent selection for NDF was performed on four populations of smooth bromegrass. The four populations–Alpha, Lincoln, WB19e, and WB88S–were unrelated to each other except that the Wisconsin populations Alpha and WB19e shared approximately 25% of their alleles (Casler et al., 2000), and a population selected for high digestibility out of Lincoln was also one out of four parents of WB19e (Diaby and Casler, 2005). Lincoln is a land race cultivar created from a wild collection from Hungary, and WB88S is a Syn1 strain-cross created from five accessions collected in the Altai Mtns. of southern Russia (USDA-ARS, 1990).

The first cycle of divergent selection for NDF was completed as described by Casler (2002) creating low NDF (C-1) and high NDF (C+1) subpopulations. Diaby and Casler (2005) conducted the second cycle of divergent selection for NDF creating C-2 and C+2 subpopulations for each of the four populations. The 10 most extreme individuals from a population of 300 (Cycle 1) or 350 (Cycle 2) plants were selected on the basis of their phenotype. Selection units were unreplicated plants on 0.9-m centers. Selection was based on a single harvest of leaf tissue in August (Cycle 1) or May (Cycle 2). Plants from the original germplasm were designated as C0, resulting in five subpopulations for each population. A total of 370 plants, representing a random sample of 12 to 26 plants from each of these 20 subpopulations, were grown and maintained in a greenhouse. Plants that were sufficiently large were clonally propagated between December 2001 and May 2002.

Field Evaluation
All plants were transplanted into the field in a spaced-plant nursery on 0.9-m centers in June 2002 at Arlington, WI. The experimental design was a randomized complete block design with two clonal replicates. Clones were blocked according to a split-plot randomization restriction within each replicate, with populations as whole plots and clones as subplots. Plants were clipped twice during the establishment year, but no samples were collected. Weeds were controlled by a combination of tillage, hand weeding, and application of 1.12 kg ha–1 alachlor [2-chloro-N-2,6-diethylphenyl)-N-(methoxymethyl)-acetamide] with 0.07 kg ha–1 imazethapyr {(±)-2-[4,5-dihydro-4-methyl-4-(1 methylethyl)-5-oxo-1H-imidazol-2-yl]-5-ethyl-3-pyridinecarboxylic acid}. Plants were fertilized with 56 kg N ha–1 in early spring 2003, early spring 2004, and following every clipping.

Starting in 2003, leaf blades and sheaths were harvested from each plant. Harvests were taken in May, June, August, and October in 2003 and 2004 when the canopy height ranged from 20 to 30 cm. The frequent harvest schedule and the use of leaf blades and sheaths were meant to simulate use of a management-intensive rotational grazing system. Plants were hand-clipped leaving a stubble height of 9 cm. The tissue samples were collected in paper bags and dried at 60°C. Dried samples were ground through a 1-mm screen with a Wiley-type mill. All samples were scanned by near-infrared reflectance spectroscopy (NIRS). Subsets of 80 samples from each year were selected on the basis of cluster analysis of wavelength reflectance data (Shenk and Westerhaus, 1991). These samples were analyzed for NDF concentration by using the procedure outlined in Van Soest et al. (1991), omitting the {alpha}-amylase step. The NDF values were predicted for all samples using two calibration equations: SEcal = 12.8 g kg–1, R2cal = 0.93, SEval = 17.7 g kg–1, and R2val = 0.87 for 2003 data and SEcal = 10.8 g kg–1, R2cal = 0.95, SEval = 14.9 g kg–1, and R2val = 0.90 for 2004 data.

Molecular Markers
Tissue samples were collected from each genotype for DNA isolation as described in Skroch and Nienhuis (1995b). Approximately 0.5 g of tissue was ground in microcentrifuge tubes with the addition of 500 µL of potassium ethyl xanthogenate and ceramic beads in a Bio101 Savant FP120 Fast Prep. The samples were incubated in a water bath at 65°C for 30 min. The samples were centrifuged to separate the tissue from the buffer, which was then collected in a new microcentrifuge tube. The nucleic acids were precipitated for 30 min at room temperature after adding a 6:1 solution of 95% (v/v) ethanol and 7.5 M ammonium acetate. The supernatant was discarded after centrifugation followed by the addition of 300 µL of a 100 µg mL–1 RNase solution then incubated for 1 h at 37°C. Any remaining plant material was eliminated by centrifugation and collection of the supernatant. The DNA was reprecipitated for 30 min at room temperature following the addition of a 10:1 solution of 95% ethanol and 3 M sodium acetate. The DNA was pelleted through centrifugation and washed for two cycles in 70% ethanol. After drying, the pellet was hydrated in TE buffer. Concentrations of DNA were quantified with a Hoefer Dyna Quant 200 fluorometer, then diluted to 4 ng/µL with a solution of tartrazine in TE buffer.

Sixteen RAPD bands were selected for further examination from an earlier study (Diaby and Casler, 2005) because of their significant association with NDF concentration (Table 1). These markers were chosen on the basis of their performance for individual populations and, with the exception of AG14.0825, were not among those reported by Diaby and Casler (2005). Within each of the four populations, two bands with positive effects on NDF and two bands with negative effects on NDF were selected. Within each pair of bands one was selected for minimal drift effect and one was selected for maximum selection effect by the methods described by Wilson (1980). Markers were selected for maximum selection effect to achieve the maximum expected selection response. Other markers were selected for minimal drift effect to ensure that observed marker associations were not due to sampling error resulting from restricted population size. One of the 16 bands (AG14.0825) was selected because it showed an association with NDF in all four smooth bromegrass populations. The R2 values from the previous study (Diaby and Casler, 2005) measured the proportion of additive genetic variance explained by each marker.


View this table:
[in this window]
[in a new window]
  		Table 1. Statistics of markers selected from the experiment of Diaby and Casler (2005) for association with neutral detergent fiber (NDF) in four smooth bromegrass populations and utilized in the current experiment.

 
Polymerase chair reactions (PCR) were run as described in Johns et al. (1997) in a Genemate Genius thermal cycler. Each RAPD primer was run on all samples. The cycling conditions were as follows: 91°C for denaturation, 42°C for annealing, and 72°C for elongation. Temperature holds were 60 s for denaturation, 15 s for annealing, and 70 s for elongation for the first two cycles. The remaining 38 cycles were set the same with the exception of the denaturation step being set for 15 s. The 10-µL polymerase chain reaction amplification contained the following concentrations: 50 mM Tris, pH 8.5, 20 mM KCl, 2 mM MgCl2, 500 µg mL–1 bovine serum albumin, 2.5% ficoll 400, 0.02% xylene cyanol, 100 µM dNTPs, 0.4 µM of 10-mer oligonucleotide, 2 ng µL–1 of DNA template, and 0.6 units of Taq DNA polymerase.

The amplification products were separated by electrophoresis in a 20 cm x 25 cm gel box using 1.5% (w/v) agarose gel in 1x Tris-acetate-EDTA (TAE) buffer. Gels were run at 300 V for 2 h then stained in ethidium bromide and destained in distilled water. Gels were illuminated on a ultra-violet transilluminator and documented using the Syngene digi-genius photo-documentation package. Relevant bands were scored by hand (0 for absence or 1 for presence of the marker). Frequencies for RAPD marker bands were obtained for each marker in each subpopulation. Marker frequencies were correlated with marker frequencies determined by Diaby and Casler (2005).

Statistical Analyses
Because of missing plants, occasional plant mortality, and a high frequency of samples of insufficient size, the population of genotypes was reduced to a subset of 227 genotypes that were present in at least one replicate for all eight harvests (Table 2). A generalized least squares analysis of variance was conducted on the NDF data to generate least squares means for phenotypic selection. All factors were fixed except for blocks, which were assumed to be a random effect. Phenotypic selection, using least squares means over two replicates and eight harvests, was used as a control selection criterion. Selection differentials were computed from the five genotypes ranked highest in NDF and the five genotypes ranked lowest in NDF within each population. Selection differentials were tested for significance using a contrast within an analysis of variance for each population.


View this table:
[in this window]
[in a new window]
  		Table 2. Number of smooth bromegrass plants analyzed for each of 20 subpopulations.

 
For each population, 16 generalized least squares analyses of variance were conducted on the NDF data, one for each of the 16 markers. The sum of squares for genotype and genotype x harvest were partitioned into 1 df for marker class and g–2 or (g–2)(h–1) df for genotypes within marker classes (g = number of genotypes, h = number of harvests). The selection differential for each marker was computed as the difference in least-squares-mean NDF between marker classes. The proportion of variation attributed to the marker was computed as the sum of squares for the marker divided by the sum of squares for genotypes. Individual marker confirmation was conducted by comparing selection differential, marker P value, and R2 from the current study to the selection response, selection P value, and R2 from Diaby and Casler (2005).

Fourteen types of marker selection indices were defined according to Lande and Thompson (1990). The potential contribution of phenotypic data was ignored in developing indices because a goal of this research was to investigate the possibility of substituting marker data for phenotypic data. The molecular score (Ii) for the ith genotype was calculated as

Formula
where cj is a coefficient of 1 or –1 depending on the sign of the observed selection response for the jth marker from the Diaby and Casler (2005) study, Rj2 is the proportion of additive genetic variance explained by the jth marker in the Diaby and Casler (2005) study, mij is the score (0 for absence or 1 for presence) of the jth marker on the ith genotype, and n is the number of markers included in a given marker index.

Marker indices were generated in pairs, where each member of a pair of indices were based on identical markers and marker index scores. For odd-numbered indices, plants that tied in marker index scores were separated by choosing plants at random. For even-numbered indices, plants that tied in marker index scores were separated by pedigree, preferentially choosing genotypes from lower-NDF or higher-NDF selection cycles (the order of preference was –2, –1, 0, +1, and +2 for low NDF, the opposite for high NDF). This is analogous to Lande and Thompson's (1990) idea of using information from relatives to estimate the breeding value of an individual for a low heritability trait.

Marker indices 1 and 2 included all 16 markers. Marker indices 3 and 4 included all markers that had a significant effect within a population, as determined by the individual marker tests previously described. Marker indices 5 and 6 included five population-specific markers selected for use in this study (Table 1). Marker indices 7 and 8 included all markers that had a significant effect within a population, as determined by the individual marker tests, and had a selection differential with the proper sign as determined by Diaby and Casler (2005). Marker indices 9 and 10 included only those population-specific markers from Table 1 that had a significant selection differential with the proper sign. Marker indices 11 and 12 included all markers that were significant in more than one population for the current study. Marker indices 13 and 14 included all markers that were significant in more than one population in both the current study and the Diaby and Casler (2005) study.

For all marker indices, the selection differential was computed as the difference between NDF least squares means of the top five vs. the bottom five genotypes ranked by net molecular score. P values and R2 values for each marker index were computed from a contrast of the top five vs. the bottom five genotypes, exactly as described for phenotypic selection.


   RESULTS AND DISCUSSION
TOP
 NOTES
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
CONCLUSIONS
 REFERENCES
 
RAPD Marker Reproducibility
Frequencies for only two of 16 markers, calculated on a subpopulation basis, were significantly correlated between this study and that of Diaby and Casler, 2005 (AF06.1750 with r = 0.45, P < 0.05; A09.1500 with r = 0.70, P < 0.01). Thus, marker frequencies between Diaby and Casler (2005) and the current study were not repeatable. The two studies used the same DNA isolation methods, RAPD reaction conditions, and gel electrophoresis equipment. Therefore, the lack of repeatability could only be attributed to differences in the number of plants in each population and different people scoring the bands. Although having fewer plants is likely to alter the marker frequencies slightly, it would not change them to this degree on its own. More likely, because different people scored the results, the perspective of the person scoring could lead to large variation in marker frequencies between studies. It is possible that the polyploid nature of the smooth bromegrass genome may contribute varying band intensities because of copies of a given band or comigration of similar-size bands amplified from independent loci. A common problem in using RAPD markers is the underscoring of faint bands (Xu et al., 1995; Skroch and Nienhuis 1995b). A more reliable system for scoring genetic differences, where scoring is independent of the perspective of the person conducting the scoring, would be preferable and most likely yield more consistent results. Dot-blot hybridization could be used to improve the accuracy and sensitivity of scoring RAPD bands as well as reduce the time and labor involved (Penner et al., 1996).

Before initiating this experiment, we recognized the potential for low reproducibility of RAPD markers (Skroch and Nienhuis, 1995a). We attempted to avoid this problem by developing sequenced characterized amplified region (SCAR) markers from the RAPD markers identified by Diaby and Casler (2005), using techniques described by Scheef et al. (2003). Despite repeated attempts to excise and isolate a single unique nucleotide sequences from a single RAPD band, all excised DNA samples of any given molecular weight contained multiple unique nucleotide sequences. Therefore, we were unable to obtain any single unique nucleotide sequences to correspond to any of the RAPD bands scored by Diaby and Casler (2005). The size and complexity of the smooth bromegrass genome may be an important factor contributing to this phenomenon.

Analysis of NDF Phenotype
The mean square for harvests was the largest in the analysis of variance, indicating large differences among harvests for mean NDF (Table 3). Despite the vegetative growth stage of all plant samples, the inferences from this study span a relatively wide range in mean NDF values for the eight harvests (403–501 g kg–1). Means of the four germplasm sources (Alpha, WB19e, Lincoln, and WB88S) were not significantly different from each other, because of a large range of NDF values within each germplasm source. This may be partially due to the parallel success of divergent selection within each of the four populations (Diaby and Casler, 2005).


View this table:
[in this window]
[in a new window]
  		Table 3. Generalized least squares analysis of variance conducted on the neutral detergent fiber data for 227 smooth bromegrass genotypes present in at least one replicate for all eight harvests.

 
Genotype and genotype x harvest interaction effects were both significant (P < 0.0001), but the mean square for genotype was much larger than that for genotype x harvest interaction. Therefore, the genotype x harvest interaction had relatively little biological significance in comparison to the variation between genotypes. This result was largely expected because numerous previous experiments have similarly shown genotype x environment interactions to be relatively unimportant for NDF of smooth bromegrass (Reich and Casler, 1985; Casler and Vogel, 1999; Casler, 2002). Least squares genotype means over eight harvests were utilized in all analyses.

Confirmation of Individual Markers
The generalized least squares analyses of variance for 64 marker-population combinations revealed that 51 of 64 possible marker-population combinations had significant (P < 0.05) marker effects (results not shown). Twenty-one of these marker-population combinations were significant (P < 0.05) in a direction consistent with results of Diaby and Casler (2005), across 11 of the 16 markers used in this study (Table 4). Marker P values showed greater statistical significance in our study because the test was based on a first-order statistic (marker-group means), while the hypothesis tests of Diaby and Casler (2005) were based on a second-order statistic (linear regression slopes), which are measured with inherently less precision. The low R2 values for our study were a result of the large phenotypic variability for NDF within marker classes and can be considered, in a general sense, as the estimated heritability of the individual marker as an index of NDF. These results indicate that individual markers accounted for very small amounts of phenotypic variability for NDF on an individual-plant basis and have little or no utility by themselves.


View this table:
[in this window]
[in a new window]
  		Table 4. Marker statistics for markers found to be significant (P < 0.05) and in directional agreement with results of Diaby and Casler (2005).

 
Of the 21 marker–population combinations showing significance and having the proper sign, six of these were among the population-specific markers identified in Diaby and Casler (2005) as having the greatest potential utility for MAS. However, 30 of the 64 marker–population combinations were significant (P < 0.05) in the opposite direction to the results of Diaby and Casler (2005) and seven of these were among the population-specific markers identified in that study. Although some of the marker associations were found to be valid, more were found to be significant in the wrong direction, indicating more-or-less random correspondence between the two studies based on the entire set of 16 markers.

Nevertheless, there were some trends evident in Table 4 to indicate that some markers showed a greater potential for use in MAS than the remaining markers. Seven of the 16 markers studied showed a significant association with NDF in the direction consistent with the previous study (Diaby and Casler, 2005) in more than one population. Three of these markers, AF06.1750, A09.1250, and A18.0600, were also significant in multiple populations of the Diaby and Casler (2005) study. This agreement provides a limited measure of validation for these markers, indicating their potential value for MAS within multiple populations. Previous studies concluded that marker-based selection could be effectively applied in populations other than the mapping population for {alpha}-amylase activity in barley (Hordeum vulgare L.) and water-soluble carbohydrate content in perennial ryegrass (Lolium perenne L.) (Ayoub et al., 2003; Humphreys, 1992). It is clear from these studies that some linkage relationships between markers and functional loci are consistent across diverse germplasms, indicating evolutionary conservation for some marker-trait linkage relationships.

Marker Indices
The truest measure of validation of the marker indices consisted of indices derived from only those population-specific markers recommended by Diaby and Casler (2005). One population, Lincoln, had a significant positive selection differential (Table 5) and two populations had a significant negative selection differential for marker index 5 (MI-5). Therefore, using the strictest sense of validation of the previous study, only one of four populations could be considered to have a validated marker index. Selection on the basis of MI-5 would lead to changes in the wrong direction for the other three populations.


View this table:
[in this window]
[in a new window]
  		Table 5. Statistics for phenotypic selection or marker-index (MI) selection criteria based on markers previously identified by Diaby and Casler, 2005 (D&C).

 
Using pedigree data to break ties in the marker index score created significant improvement for MI-5 (Table 5). Both Lincoln and WB88S had significant selection differentials in the desired direction and only one population had a significant selection differential in the wrong direction. These results were expected, because MI-6 placed emphasis on plants that derived from selection cycles with lower mean NDF. Thus, pedigree data combined with previously identified population-specific markers resulted in a degree of validation for two of four populations. Pedigree information has been successfully incorporated into a marker-breeding program in forest trees using clonal propagules to associate markers with the phenotypic trait of interest (Bradshaw and Foster, 1992).

Marker indices 1 through 4 were based on larger numbers of markers, based on the idea that more marker data might improve the validation (Table 5). Marker indices 1 and 2, incorporating all markers, each showed a significant selection differential for the WB88S population but also showed a significant selection differential for Alpha in the wrong direction. Marker indices 3 and 4, using between 11 and 14 significant markers in each population, showed significant (P < 0.05) selection differentials in the wrong direction for two populations each. The effect of using pedigree information is not evident in the selection differentials when contrasting MI-1 with MI-2 and MI-3 with MI-4 because the numbers of markers was sufficiently high that there were few ties between plants within populations. Adding additional markers, regardless of the significance level of their association with NDF in either study, did not improve their performance in this study. Gimelfarb and Lande (1994) similarly concluded that increasing the number of markers included in the index does not necessarily result in higher selection efficiency.

Marker indices 7 and 8 used markers with a significant (P < 0.05) marker effect and a sign corresponding to those in Diaby and Casler (2005). Both marker indices had significant selection differentials, in the proper direction, for all four populations. The selection differential nearly doubled in the WB88S population when considering pedigree information (MI-8 vs. MI-7). Selection differentials for these indices ranged from 11 to 58% for MI-7 and 11 to 56% for MI-8 relative to phenotypic selection. While these marker indices do not represent true validation of the Diaby and Casler (2005) study, they do represent a certain agreement in marker–NDF associations between the two studies. Thus, limiting the number of population-specific markers recommended by Diaby and Casler (2005) by including only those with significant marker effects in the proper direction, we achieved more functional marker indices and a limited measure of validation. Marker indices including only a few markers may result in efficient selection progress for the phenotypic trait provided the markers explain a large proportion of the phenotypic variability (Schön et al., 2004; Knapp, 1998).

The selection differentials using MI-9 were significant for all four populations, but vast improvements were observed as a result of using pedigree information to break ties in MI-10. This was to be expected because these indices were based on only one or two markers per population and there were many ties. Marker indices included markers AG14.1100 and AF06.1750 in Alpha, AF14.1200 in WB19e, AG14.0825 in Lincoln, and A09.1250 and A18.0600 in WB88S. The selection differential for Alpha increased over six times by using pedigree information. Selection differentials for MI-9 explained 9 to 38% of control phenotypic selection differentials, while selection differentials for MI-10 explained 36 to 58% of phenotypic selection differentials. The amount of marker index selection differential relative to phenotypic selection differential for each population is especially remarkable considering it is being measured against phenotypic selection based on means over two replicates and eight harvests. Measuring NDF phenotype with this degree of precision required a tremendous amount of labor, considerably more than would be required to isolate DNA and score a relatively small number of markers. These results are similar to those of a previous study in which a large proportion of the variation for water-soluble carbohydrate concentration in perennial ryegrass was attributed to a single marker locus (Humphreys, 1992).

Selection differentials using MI-11 were significant in two populations, but three of the four MI-11 indices were improved when pedigree information was used to break ties in MI-12. Selection differentials for MI-13 and MI-14 showed similar, but more striking results, with significant values in all four populations and a slight improvement from the use of pedigree in all four populations. Although MI-13 and MI-14 selection differentials were smaller than for MI-9 and MI-10, they were more widely applicable across smooth bromegrass populations. These markers provided a measure of validation of the Diaby and Casler (2005) study because marker P values used to eliminate meaningless markers and R2 values used to build the marker indices were all taken from the previous study. These results agree with those of Ayoub et al. (2003) and Humphreys (1992) demonstrating that marker selection can be applied successfully in a population other than the population used to detect the marker-trait associations.

Marker index 10, across all four populations, appears to provide the greatest potential for use in MAS to reduce NDF of smooth bromegrass. These indices should be validated by intercrossing plants chosen by the index, then testing progeny populations selected for high or low values of the index relative to the parent population as a control. Phenotypic selection for NDF should also be employed as an additional control. Following validation, it is likely that the linkage blocks could remain intact for a few generations of breeding before the marker associations need to be re-evaluated because linkage blocks are difficult to disrupt during recombination in smooth bromegrass because of its complex inheritance. Also, many linkage blocks in a perennial forage such as smooth bromegrass remain intact because of its slow evolution in nature. Future studies should utilize a marker system such as dot-blot hybridization (Penner et al., 1996) that gives consistent results independent from the perspective of the person doing the scoring to avoid discrepancies like those observed between this study and prior research (Diaby and Casler, 2005). Selecting plants on the basis of these marker indices could also reduce the amount of time required per cycle of selection, leading to greater response per year than for current methods of phenotypic recurrent selection.

In most cases, populations will not have the pedigree structure that will allow for the breaking of ties. In these cases, marker index 13, using only markers significant in multiple populations, identified in both the current and Diaby and Casler (2005) studies, can be used to perform selection. Another way of using these indices would be to perform marker analysis on all individual plants and to phenotype pooled related samples, which could be accomplished within any family selection scheme. Selecting plants from families with lower NDF concentration could break marker index score ties and enable the use of marker indices conceptually similar to marker index 14 with the benefit of a higher predicted selection response. Another way of utilizing phenotypic data would be to optimize the selection index for maximum response per cycle by weighting the coefficients of marker data using the amount of additive genetic variance explained by the markers and weighting phenotypic NDF data with the heritability of NDF (Lande and Thompson, 1990). However, gathering and analyzing phenotypic data would add cost and time per cycle that might not be worth the added response. A cost–benefit analysis of gathering phenotypic data should be performed before beginning such a MAS selection program.


   CONCLUSIONS
TOP
 NOTES
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
REFERENCES
 
The association of several RAPD markers with NDF, first observed on a population basis, was confirmed in this study of phenotypic variability for NDF of individual plants. These associations hold promise for using marker selection as a substitute or supplement for phenotypic selection in breeding for reducing NDF concentration of smooth bromegrass. Use of marker indices would allow identification of putative low-NDF plants at the seedling stage, potentially speeding along the selection process. Alternatively, seedlings could be prescored using mild selection pressure for marker indices, reducing the number of plants to be phenotyped. Validation of these marker indices will require completion of the selection process initiated in this study–intermating selections identified by the marker indices and by phenotypic analysis, conducting a field study of parental and selected populations, and comparing populations selected by marker indices with those selected by NDF phenotype.

Lack of repeatability between two people scoring RAPD bands will prevent long-term use of the RAPD technique for marker selection. The lack of reproducibility of RAPD marker scores did not prevent confirmation of some markers but most likely reduced the number of effective markers and the effectiveness of the remaining markers. The use of dot-blot hybridization would allow continued use of these potentially valuable PCR-based primers, but improve their reproducibility, reducing their sensitivity to personal interpretation. Alternative marker systems, such as PCR-based amplified fragment length polymorphic (AFLP) markers may prove to be more reliable for applications in marker selection.


   NOTES
TOP
 NOTES
ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
 REFERENCES
 
Research partially supported by the Univ. of Wisconsin, College of Agric. and Life Sciences and Hatch formula funds. Mention of a trademark or brand name does not imply endorsement over any other product by the USDA-ARS or the University of Wisconsin.

Received for publication February 17, 2005.


   REFERENCES
TOP
 NOTES
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
 REFERENCES
 





This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF) Free
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Stendal, C.
Right arrow Articles by Jung, G.
Right arrow Search for Related Content
PubMed
Right arrow Articles by Stendal, C.
Right arrow Articles by Jung, G.
Agricola
Right arrow Articles by Stendal, C.
Right arrow Articles by Jung, G.
Related Collections
Right arrow Other Forage Crops
Right arrow Crop Genetics

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