| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
a USDA-ARS, Cereal Crops Research Unit, 501 Walnut St., Madison, WI 53726 and Department of Agronomy, University of Wisconsin-Madison
b USDA-ARS, Small Grains and Potato Germplasm Research Unit, 1691 S 2700 W, Aberdeen, ID 83210
* Corresponding author (dmpeter4{at}wisc.edu)
 |
ABSTRACT |
|---|
 TOPNOTES ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONREFERENCES |
|---|
-tocopherol, and the avenanthramides. The genotype x trial interaction was significant only for avenanthramides concentration, whereas the genotype x year interaction was significant for most traits. Principal component analysis biplots illustrate that for protein and oil there was a clear difference between 2000 and the other years. Some genotypes were stable across environments; others responded differently in different environments. Correlation analysis showed several close associations among traits: avenanthramides were correlated with ß-glucan, and oil was negatively correlated with groat physical characteristics and with avenanthramide 2f. The results show that knowledge of the relationships among traits and environments can assist breeders in optimizing both agronomic traits and grain composition simultaneously.
Abbreviations: AT, -tocopherol • AT3, -tocotrienol • AVA, avenanthramide • BG, ß-glucan • GPC, groat percentage • GWT, groat weight • HD, heading date • KWT, kernel weight • PC, principal component • PCA, principal component analysis • PRO, protein • TOC, tocols • TWT, test weight • YLD, yield • 2C, avenanthramide 2c • 2F, avenanthramide 2f • 2P, avenanthramide 2p
|
INTRODUCTION |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONREFERENCES |
|---|
Improvement of agronomic traits has been the primary objective of oat breeders for many years. Breeders have also measured and selected for certain grain physical traits such as test weight, kernel weight, and groat percentage (Forsberg and Reeves, 1992). All these traits are affected by the growing environment as well as by genetic factors, and numerous studies have described the genotype–environment interactions (Humphreys et al., 1994a; Doehlert and McMullen, 2000; Doehlert et al., 2001). Other studies have examined the role of genotype and environment on grain composition, especially ß-glucan, protein, and oil (Brunner and Freed, 1994; Humphreys et al., 1994b; Peterson et al., 1995; Doehlert et al., 2001).
In addition to the traditional nutrients, oat is a source of certain phytochemicals, which may be beneficial for human and animal health because of their antioxidant or other activities. Many of these phytochemicals are phenolics, and their chemistry and occurrence in oat were reviewed by Collins (1986). Peterson (2001) reviewed their antioxidant properties. Two classes of oat phytochemicals, the tocols and the avenanthramides, have received particular attention because of their abundance and properties. The tocols, consisting of tocopherols and tocotrienols, are found in many grains and oilseeds, and they exhibit biological activity known as vitamin E. Tocotrienols also inhibit cholesterol biosynthesis (Pearce et al., 1992), and both tocotrienols and tocopherols are lipid-soluble antioxidants in plasma and lipoproteins (Suarna et al., 1993). -Tocotrienol, which is the predominant tocol in oat and barley (Hordeum vulgare L.), was reported to be superior to -tocopherol in scavenging reactive oxygen species in membranes (Serbinova et al., 1991). Tocol concentration in oat is affected by genotype and by environment (Peterson and Qureshi, 1993). In contrast to the tocols, avenanthramides are unique to oat (Collins, 1989). These compounds, which consist of anthranilic acid or hydroxyanthranilic acid linked to one of the hydroxycinnamic acids through an amide bond, are potent antioxidants in vitro (Peterson et al., 2002; Bratt et al., 2003). Recently, antioxidative and antiatherogenic effects in vivo have been demonstrated (Ji et al., 2003; Chen et al., 2004; Liu et al., 2004). The concentrations of avenanthramides in oat are significantly affected by genotype and growing environment (Emmons and Peterson, 2001).
The first objective of this study was to determine effects of different growing environments on phenotypic traits. The second objective was to determine whether there were significant genotype by environment interactions for the various traits. Third, the relationships among several agronomic and grain quality traits were investigated.
|
MATERIALS AND METHODS |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONREFERENCES |
|---|
|
After harvest, the yield of whole oats was measured by weighing, and test weight was determined. Kernels were dehulled with an impact type dehuller. Kernel and groat weights were determined on 100-grain samples. Groat samples were cleaned, eliminating hulled and broken kernels, and sent to the ARS Cereal Crops Research Unit, Madison, WI, for analysis of their composition.
Grain Composition Analyses
The intact groats were analyzed for protein and oil by near infrared transmittance (NIT) using an Infratec 1255 Food and Feed Analyzer (Foss, Eden Prairie, MN). The NIT instrument was calibrated for protein with a subset of the samples, which were analyzed for N using a LECO FP-428 N analyzer (LECO, St. Joseph, MI). Nitrogen was converted to protein by multiplying by 6.25. A similar subset of calibration samples was extracted with petroleum ether followed by gravimetric analysis for oil. The calibration samples were ground in a Retsch ZM-1 Centrifugal Mill (Brinkmann, Westbury, NY) to pass a 0.5-mm screen. ß-Glucan was analyzed by the Calcofluor method (Peterson, 1991). Protein, oil, and ß-glucan data are single determinations for each sample and are reported on a dry basis. Ground samples were extracted with ethanol for analysis of tocols by HPLC as previously described (Peterson and Qureshi, 1993). Only the two major peaks (-tocopherol and -tocotrienol) were quantified using -tocopherol as an external standard. For avenanthramides analysis, samples were extracted with 800 mL L–1 ethanol, the solvent evaporated, and the residue redissolved in methanol as described (Emmons and Peterson, 2001). The HPLC procedure was modified from earlier studies as follows. The column was a Discovery HS C18, 5 cm by 4.6 mm, 5-µm column (Supelco, Bellefonte, PA). The gradient was 150 mL L–1 acetonitrile for 5 min, then increasing in concentration to 180 mL L–1 at 6 min, 300 mL L–1 at 17 min, and 600 mL L–1 at 18 min. The other phase was 0.01 M NaH2PO4, pH 2.8. The flow rate was 1 mL min–1 and 10 µL of sample was injected. Peaks were detected at 340 nm with a Model 170 diode array detector (Gilson, Middleton, WI). Only the three major avenanthramides, 2c, 2p, and 2f, (corresponding to C, A, and B using previous terminology) were quantified, although other minor putative avenanthramide peaks were present. Tocol and avenanthramide data are means of duplicate determinations expressed on an as-is basis. Avenanthramide data were obtained for the years 2000 and 2001 only.
Statistical Analysis
Both univariate and multivariate statistical approaches were used to evaluate the influence of genotype and environment on these traits and the relationships that may exist among them. While traditional analysis of variance methods are useful for showing the significant differences among main effects and their interactions, the multivariate principal component analysis (PCA) methods graphically illustrate the relationships among environments and genotypes.
Each of the nine environments was analyzed using the GLM procedure of SAS (SAS Institute Inc., 1989). Homogeniety of variances among environments was tested using Levene's test, and this was followed by Welch's ANOVA. A combined analysis of variance for all environments was performed using the GLM procedure of SAS. Trials, years, genotypes, and their interactions were all considered to be random. Variance components attributed to genotypes and their interactions were computed using the VARCOMP procedure of SAS.
Phenotypic and genotypic correlation coefficients among traits were computed using the sums of squares and cross products matrices for genotypes and genotypes x environment generated by the SAS MANOVA procedure (Falconer, 1989).
Principal component analysis, performed with The Unscrambler v. 8.0 software (Camo Technologies, Woodbridge, NJ), was used to examine the genotype x environment relationships for each trait. Principal component analysis models the data in terms of their principal components, which can be visualized graphically (for up to three PC). The computations and background for PCA are described by Esbensen (2002). The data set was averaged across repetitions. The data were mean centered (the mean is subtracted from each variable) and scaled to unit variance (each variable is divided by the standard deviation). The models were validated by the full cross validation method (Esbensen, 2002). The results of the PCA are shown as biplots of PC1 vs. PC2 for each specific trait, where genotypes and environments are represented by markers on the biplot (Yan and Kang, 2003).
|
RESULTS AND DISCUSSION |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONREFERENCES |
|---|
Using a simple one-way model, Levene's test showed that variances among environments were homogeneous for two-thirds of the traits (not shown). The subsequent Welch's ANOVA showed that there were significant differences among the nine environments for all traits.
Table 2 shows the mean values and standard deviations across genotypes for all traits in each of the nine environments. Considering the agronomic traits, yield showed the greatest differences among environments, and heading dates were also quite different. For yield, Aberdeen yielded higher than Tetonia irrigated, which in turn was higher than Tetonia dryland. Plants headed earlier at Aberdeen than Tetonia, a result of the earlier planting dates. The kernel characteristics, test weight, kernel weight, groat weight and groat percentage, although significantly different among environments, did not vary as much. For the grain constituents, the avenanthramides showed the greatest influence of environment; they were about twofold higher at Aberdeen than at the Tetonia trials. ß-Glucan concentration was affected more by year than by trial, showing 20 to 21% lower values in 2001 than in 2000, with intermediate concentrations in 1999. Oil concentration was higher in 1999. For protein and the tocols, the differences among environments were not specifically attributed to either year or trial.
|
Avenanthramide concentrations were markedly lower at all trials than was previously reported from Wisconsin sites (Emmons and Peterson, 2001). Even at these low levels, there was a consistent advantage of the Aberdeen trial over those at Tetonia. Reports from Sweden have shown both relatively high (Dimberg et al., 1996) and low (Bryngelsson et al., 2002) avenanthramide concentrations. While there are significant differences in avenanthramide concentrations among oat cultivars, the growing environment can have an even greater effect (Emmons and Peterson, 2001). For example, oat cultivars grown at Sturgeon Bay produced significantly higher concentrations than at six other diverse sites in Wisconsin. The reason for this difference is unknown. In an experiment in Sweden, oat plants that experienced mild or severe drought stress and were infected with saprophytic fungi and rust had avenanthramide concentrations that were >10-fold higher than the same cultivars in a second parallel experiment that experienced normal rainfall and were not diseased (Mannerstedt-Fogelfors, 2001). However, there were no significant differences between mildly and severely stressed plants in the first experiment, suggesting either that the presence of the fungal infections may have caused the higher avenanthramide concentrations, or that even mild stress is sufficient to elicit a strong response. Avenanthramides in seedling oat leaves are elicited by infection with crown rust spores (Puccinia coronata Corda var. avenae W.P. Fraser & Ledingham) (Mayama et al., 1982; Miyagawa et al., 1995) or other elicitors (Ishihara et al., 1998). It is unknown if grain avenanthramide concentrations are similarly influenced by elicitors. However, one could speculate that avenanthramide concentrations are lower in the relatively disease-free Idaho environment than in Wisconsin and other more humid environments, where disease pressure is common and the grains are colonized by saprophytic fungi. The difference between Aberdeen and Tetonia could be that the warmer temperatures at Aberdeen caused some stress, and this, by some unknown mechanism, led to increased avenanthramide production.
Most of the variance of genotype means was associated with the genotype main effect for all traits except yield, -tocopherol and the avenanthramides (Table 3). Yield had a large variance associated with error. The variance associated with the genotype x year interaction was significant for nine traits and was greater for ß-glucan and protein than for the other traits. For -tocopherol, variance was distributed among genotype main effect, its interaction with year, and the three-way interaction. The genotype x trial interaction variance was uniformly negligible, except for the avenanthramides, where it was significant. As discussed previously, we know that environment affects avenanthramide concentrations in ways that are not yet understood. It is clear from these results that the genotypes did not respond similarly to environmental factors in their avenanthramide concentrations.
|
For the traits oil, protein, kernel weight, and groat percentage, the three 2000 trials (Aberdeen, Tetonia irrigated, and Tetonia dryland) were closely associated and were opposite the 1999 and 2001 trials (PC2). There was no indication for any of the traits that the results of any trial were closely associated across all three years. The results for protein and oil are shown (Fig. 1). The cosine of the angle between the vectors from the biplot origin to the environments approximates the correlation coefficient between them (Yan and Kang, 2003).
|
From these biplots, one can see which genotypes were relatively stable across environments. These generally have low absolute values for PC2, although PC1 may be highly positive or negative, depending on whether their phenotype is high or low for the trait. Other genotypes, whose position on the biplot gives a highly positive intersect for some environmental vectors but a highly negative one for others (high absolute values for PC2), have a significant environmental interaction in their phenotype.
For example, Genotypes 8 and 12 were below average for oil concentration in all environments, whereas 16 and 31 were above average in all environments (data not shown). These four genotypes have high absolute values for PC1 but low values for PC2 (Fig. 1), indicating stability across environments. Genotypes 25 and 28 were much higher than average for oil concentration in the 2000 trials but lower than average in 1999 and 2001. Conversely, Genotype 22 was above average for oil concentration in 1999 and 2001 but below average in 2000. These genotypes all had high absolute values for PC2, and the positions of their markers on the biplot are closely associated with the markers for the environments where they expressed relatively high oil concentrations. It is important to note, however, that the variance for oil concentration associated with PC1 was more than seven times greater than that associated with PC2, so a large absolute value of PC1 is much more significant than a similar value for PC2. A similar analysis can be done for the genotypes and locations on the protein biplot. In this case, the variance associated with PC1 was only 2.6-fold greater than that associated with PC2, so the environmental effects were greater for protein than for oil.
It is essential to keep in mind that the first two principal components of biplots do not account for all the variation. For some of the trait-specific biplots, two principal components make quite good models, that is, kernel weight, groat percentage, protein, and oil. For these, the association of genotype positions in relationship to environments is a good indicator of their performance in that environment. For other traits, additional principal components would have to be considered to more accurately reflect the relative positions.
The biplots can be used to select genotypes that may have favorable combinations of traits for use in a breeding project. For example, if an objective were to increase yield, protein, and oil, genotypes that fall into the two right quadrants of biplots for these traits might be more promising than those that fall into the left quadrants.
The fact that the environments generally were in a similar position with regard to PC1, which has the highest percentage of the explained variance, indicates that for the most part, the genotypes responded to the different environments similarly. This is confirmed by the variance components (Table 3), which were mostly accounted for by genotype. Also, most genotypes fell close to the axis where PC2 = 0. For most traits, genotypes that had high values for PC2 appear to be exceptional, and these are the ones that have a stronger environmental interaction. For example, high-protein Genotypes 2 and 10 had identical mean protein concentrations (194 g kg–1) and nearly identical values for PC1 (Fig. 1), but Genotype 2 (PC2 
0) was more stable across environments than Genptype 10 (PC2 0.8), and therefore might be the better choice as a parent for the high-protein trait.
The associations among traits are shown by the phenotypic correlation coefficients (Table 4). Genotypic correlation coefficients were similar. Yield was negatively associated with protein and -tocopherol. Oil was negatively associated with kernel and groat size, as was noted previously (Holland et al., 2001), and groat percentage. -Tocotrienol, but not -tocopherol, was correlated with oil percentage. Previously, Peterson and Wood (1997) found a highly significant correlation between -tocotrienol and oil concentrations in a group of high-oil oat genotypes. An unexpected result was the positive correlation between each of the avenanthramides and ß-glucan concentration. A recent publication showed incorporation of avenanthramide 2f into the cell walls of oat leaf cells (Okazaki et al., 2004). There is no information on the cellular location of avenanthramides in oat grain, but if, as in leaves, avenanthramides are incorporated into cell walls, the association with ß-glucan seems to be logical.
|
Test weight was correlated with kernel and groat weight and groat percentage. High test weight and groat percentage are generally associated with well-filled kernels but are not necessarily a function of kernel weight (Forsberg and Reeves, 1992).
A random effects model was assumed, since these environments and genotypes are typical for the northwestern USA and western Canada oat growing region. However, these traits would not necessarily have the same associations for oat grown in midwestern or eastern environments. Almost certainly, the genotype performance would be significantly different in unadapted environments.
In conclusion, we have shown that the growth environment affects all traits to some extent, some more than others. Interactions of genotype with the environment are important for most of the traits that were measured, although the main effect of genotype predominated for all traits but the avenanthramides. Bi-plots from PCA can visually demonstrate which genotypes have a strong environmental interaction, and which are stable across environments. Aside from the well known trait associations of yield vs. protein and oil vs. groat (kernel) size, there was a close relationship between avenanthramide and ß-glucan concentrations. Selecting for increased concentrations of either trait might enhance the other; both are considered to be beneficial for oat as human food.
|
ACKNOWLEDGMENTS |
|---|
|
NOTES |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONREFERENCES |
|---|
Received for publication February 4, 2004.
|
REFERENCES |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONREFERENCES |
|---|
Related articles in Crop Science:
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| The SCI Journals | Agronomy Journal | Vadose Zone Journal | |||
| Journal of Plant Registrations | Soil Science Society of America Journal | ||||
| Journal of Natural Resources and Life Sciences Education |
Journal of Environmental Quality |
||||
