| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
a INIA La Estanzuela, CC 39173, 70000 Colonia, Uruguay
b Dep. of Agronomy and Plant Genetics, Univ. of Minnesota, St. Paul, MN 55108
c Universidad de la República, Garzón 780, 12900 Montevideo, Uruguay
* Corresponding author (jdiaz{at}inia.org.uy)
 |
ABSTRACT |
|---|
 TOPNOTES ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONREFERENCES |
|---|
|
INTRODUCTION |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONREFERENCES |
|---|
Partial resistance is a form of incomplete resistance, characterized by a reduced rate of epidemic development despite a susceptible or high infection type (Parlevliet, 1978). This resistance is assumed to be largely race nonspecific and inherited polygenically (Parlevliet, 1985). Since race nonspecific resistance is expressed by the absence of significant genetic interactions between host and pathogen genotypes (Parlevliet, 1985), the use of this type of resistance in crop cultivars should prevent major shifts in the composition of the pathogen population increasing the durability of the disease resistance (Geiger and Heun, 1989).
Recurrent selection is a population improvement procedure that increases the frequency of desirable alleles through repeated cycles of selection and systematic recombination. This breeding method has been widely used for improving quantitative traits in cross-pollinated crops. The expression of partial resistance to fungal pathogens is usually controlled by several genes with additive gene action (Veillet et al., 1996; Geiger and Heun, 1989; Parlevliet, 1978). Although these attributes suggest that recurrent selection could be an appropriate and effective breeding method for improving disease resistance in plants, it has received only limited attention in self-pollinated crops. Recurrent selection has been used successfully to increase the resistance in soybean [Glycine max (L.) Merr.] to Phytophthora rot (caused by Phytophthora megasperma Drechs. f. sp. glycinea Kuan and Erwin) (Walker and Schmitthenner, 1984), in barley (Hordeum vulgare L.) to leaf rust (caused by Puccinia hordei Otth.) (Parlevliet and Van Ommeren, 1988; Reinhold et al., 1993), and powdery mildew (caused by Erysiphe graminis DC. f. sp. hordei Em. Marchal) (Parlevliet and Van Ommeren, 1988), in wheat (Triticum aestivum L.) to powdery mildew (caused by Erysiphe graminis f. sp. tritici) (Abdalla et al., 1989), and scab (caused by Gibberella zeae Petch.) (Jiang et al., 1994), and in oat to barley yellow dwarf virus (Baltenberger et al., 1988).
A recurrent selection program to improve oat grain yield was initiated at the University of Minnesota in 1968 (Stuthman and Stucker, 1975). Payne et al. (1986) and Bregitzer et al. (1987) studied this population after three cycles of selection and reported an average grain yield gain of 3.8 and 4.5% per cycle, respectively. Reysack et al. (1993) estimated an average response per cycle of 7.1% after four cycles and Pomeranke and Stuthman (1992) reported average gains of 7.9% after five cycles of recurrent selection for grain yield. De Koeyer and Stuthman (1998) evaluated the response to seven cycles of selection and estimated a realized gain per cycle of 3.1%. After the completion of the seventh cycle of recurrent selection for grain yield, this adapted and high-yielding oat population was subjected to four rapid recurrent selection cycles (Walker and Schmitthenner, 1984; Frey et al., 1988) for partial resistance to crown rust.
The objectives of this research were to: (i) determine the effectiveness of rapid cycle recurrent selection as a method for improving partial resistance to oat crown rust in a high-yielding breeding population; (ii) assess the indirect effect of selecting for disease resistance on three other plant traits: grain yield, maturity, and plant height; and (iii) estimate phenotypic and genetic correlations among traits and broad-sense heritabilities.
|
MATERIALS AND METHODS |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONREFERENCES |
|---|
A rapid cycle improvement scheme, similar to the one proposed by Walker and Schmitthenner (1984) and Frey et al. (1988), was followed in Phase II. The duration of a complete recurrent selection cycle was 1 yr. The S0:2 progenies of the 21 selected plants were crossed in the fall and the S0 plants were grown during winter to produce S0:1 seed for the following field selection trials. The 63 S0:1 full-sib families were evaluated in replicated hill plot field experiments grown at one or two locations. In years with two selection experiments, one location was artificially inoculated with a mixture of crown rust races collected the previous year at the Minnesota buckthorn (Rhamnus cathartica L.) nursery. A detailed description of this nursery can be found in Al-Kherb et al. (1987). After flag leaves were fully expanded, the upper portion of the experiment's canopy was inoculated on three consecutive evenings with a gas powered backpack mist blower. Fresh crown rust urediniospores were applied at a rate of 15 g ha-1 suspended on lightweight mineral oil (Soltrol 170, 4 L ha-1). All selection experiments had two complete replicates. The level of partial resistance to crown rust was assessed by visually rating the proportion of diseased flag leaf area using the modified Cobb's scale (Peterson et al., 1948). Ten leaves were assessed per plot to estimate the average disease severity. Crown rust was assessed two, three, or four times after the appearance of the first symptoms on the flag leaves of the susceptible check. The multiple crown rust readings were then used to calculate the AUDPC as described by Shaner and Finney (1977). The 21 parents for the next recurrent selection cycle were selected by choosing the plants with lowest disease severity from each of the 21 families with the lowest AUDPC values.
The presence of race-specific resistance genes effective against part of the pathogen population can be confounded with partial resistance, especially when field experiments are artificially inoculated with a mixture of races or preformed under natural infection (Parlevliet, 1985). To test for the presence of effective race-specific resistance genes, 7-d-old plants of the 12 genotypes that compose the base population and a random subset of 40 recurrent selection parents were inoculated with 10 single pustule isolates of variable virulence. Inoculated plants were placed in a dew chamber overnight at 18°C, and then transferred to a greenhouse where temperature varied from 18 to 28°C. The infection types were read 12 to 15 d after inoculation. Susceptible or high infection types were observed for 85% of all oat-genotype x crown-rust-isolate combinations.
Evaluation Experiments
Six evaluation experiments were conducted to determine the direct and indirect response to 11 cycles of recurrent selection. Table 1 presents information on year, location, number of replications, planting and harvest dates, pesticide applications, and artificial inoculation dates, for each of the six trials. Experiments were conducted at two locations: Rosemount and St. Paul, MN. Soil type at both locations was a Waukegan silt loam (fine-silty over sandy or sandy-skeletal, mixed, mesic Typic Hapludoll). The experimental design used was a randomized complete block design with a maximum of three or five replications (Table 1). The experimental unit was a hill plot planted with 30 seeds and spaced in a 30-cm grid. Each replication was surrounded by two rows of hills to control border effects. All experiments were hand weeded and one was protected from fungal foliar diseases with one application of 220 g ha-1 of the systemic fungicide tebuconazole (Table 1). The systemic insecticide Imidacloprid (1-[(6-Chloro-3-pyridinyl)methyl]-N-nitro-2-imidazolidinimine) was used in 2000 to protect plants from aphids. Half of the evaluation experiments were artificially inoculated with a mixture of crown rust races collected the previous year at the Minnesota buckthorn nursery and following the inoculation procedure previously described.
|
Statistical Analyses
The homogeneity of error variances across experiments was evaluated using Hartley's F
-max test (Hartley, 1950). AUDPC and grain yield obtained F-max values above the critical F-max ( = 0.05), while no error variance heterogeneity was detected in flowering date and plant height. Scatterplots of error variances against trial means were visually inspected to discover possible associations between mean and variance. This type of association was found only for the AUDPC data. The Shapiro-Wilks' W, the kurtosis, and the skewness were estimated, by means of SAS Proc Univariate (SAS Institute, 2000), to detect deviations from the normal distribution. The Shapiro-Wilks' W test indicated that all distributions departed slightly from normality. The kurtosis and the skewness showed that distribution of AUDPC data was clearly not normal. The Box-Cox power transformation series (Box and Cox, 1964) was tested on the AUDPC data as a possible solution to achieve error variance homogeneity, to break the association between mean and variance, and to attain normality. The square root transformation had the lowest mean square error, reduced the F-max value by half (4.5 vs. 2.3), and eliminated the association between mean and variance. This transformation also reduced the kurtosis from 2.3 to 0.25 and the skewness from 1.2 to 0.2. No power transformation improved error variance homogeneity of the grain yield data (F-max = 4.0).
Combined analyses of variance of the square root transformed AUDPC, and nontransformed grain yield, flowering date, and plant height data for the six evaluation experiments were conducted by SAS Proc GLM (SAS Institute, 2000). Environments and replicates within environments were considered random effects and recurrent selection cycles and checks were considered fixed effects. The significance of main effects and interactions was evaluated by appropriate F tests. All main effects and interactions were found significant at the 0.01 level. The Azzalini-Cox test, described by Baker (1988), was used to evaluate the presence of crossover interactions. The critical value for this test was calculated following the modifications suggested by Cornelius et al. (1992). The number of significant crossover interactions detected, using a 0.05 interaction-wise , were 8 for squared root transformed AUDPC, 33 for grain yield, 10 for flowering date, and 8 for plant height, out of 450, 675, 270, and 270 total interaction contrasts, respectively. These numbers are below the expected values under the null hypothesis, indicating absence of significant genotype x environment crossover interaction in the four dependent variables studied.
Generalized least-squares means across environments and corresponding standard errors for the seven recurrent selection cycles and four check cultivars were estimated by the restricted maximum likelihood procedure of SAS Proc Mixed (SAS Institute, 2000). Means were compared by two-tailed t-tests of pairwise differences with no adjustment for multiple comparisons and a 0.05 comparisonwise . Removal of check cultivars from the data set before analysis reduced the transformed AUDPC treatment x environment interaction variance by 87% and the average standard error of a difference between two transformed AUDPC means by 60%. Consequently, to increase the power of detecting significant differences among recurrent selection cycle AUDPC means, only the seven recurrent selection parental cycle means were compared. The direct and indirect responses to recurrent selection for the two population improvement phases were estimated by SAS Proc Mixed (SAS Institute, 2000). Checks were excluded from the data set and regression coefficient estimates and standard errors for the four response variables were computed by considering the recurrent selection cycle number as the only fixed effect in the mixed model. The statistical significance of the linear regression coefficients was evaluated with a t-test and the average change per cycle was calculated as the regression coefficient divided by the corresponding cycle zero mean. When the quadratic regression coefficient was significant, the average response per cycle was estimated as the difference between the mean of the last cycle minus the cycle zero mean divided by the number of recurrent selection cycles.
Phenotypic and genetic correlation coefficients among response variables were estimated for each recurrent selection cycle. Phenotypic correlations were determined using genotype means across environments. Genetic correlations were calculated by means of genetic variance and covariance estimates obtained from the analysis of the multienvironment trial by Proc Mixed (SAS Institute, 2000) considering all effects random. Genetic covariances between traits were estimated as CovGab = 
, where S2G
is the genetic variance of the compound observation a + b (Kempthorne, 1957). The statistical significance of the correlation coefficients was tested as t
= 
(Kearsey and Pooni, 1996). Broad sense heritability estimates were computed on an entry mean basis for each recurrent selection cycle as 
. The numbers of replicates (r) and environments (e) used to estimate H are presented in Table 1. Exact 90% confidence intervals for H were calculated by the method of Knapp et al. (1985). Approximate values for the numerator and denominator of the F-statistic (M1/M2) were obtained from the estimated components of variance as M1 = re 
and M2 = M1 - re S2G.
|
RESULTS AND DISCUSSION |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONREFERENCES |
|---|
|
|
A continuous positive response in grain yield was observed during Phase I. The grain yield of Cycle I-0 parents is comparable with the yield of Starter, Portage, and MN 841801, and Cycle I-4 has a grain yield similar to Belle and significantly lower (P < 0.05) than Cycle I-7 and Phase II Cycle parents (Table 2). The Phase I average gain in grain yield per cycle was 8.1%. This direct response estimate is similar to the average gains reported by Reysack et al. (1993), Pomeranke and Stuthman (1992), and De Koeyer et al. (1993) for the same oat population after four, five, and five cycles of recurrent selection for grain yield, respectively. Payne et al. (1986) and Bregitzer et al. (1987) studied this population after three cycles of selection and reported smaller realized gains (3.8 and 4.5%), comparable to the 3.1% average response per cycle estimated after seven cycles of selection by De Koeyer and Stuthman (1998). All seven selection response estimates were obtained from combined hill plot experiments. Pomeranke and Stuthman (1992) and De Koeyer et al. (1993) also studied the response to recurrent selection for grain yield using two-row plots. Pomeranke and Stuthman (1992) found similar gains for hill and row plots (7.9 and 7.5%, respectively), while De Koeyer et al. (1993) estimated a higher selection response for hill plots (11.1 vs. 5.4%). The phenotypic correlations between hill and row plot means estimated by Pomeranke and Stuthman (1992) and De Koeyer et al. (1993) were 0.85 and 0.68, respectively.
In Phase II of this study, recurrent selection parents were selected for partial resistance to oat crown rust. The indirect response in grain yield to four cycles of recurrent selection for partial resistance was not significant (Fig. 1b). When the six evaluation experiments were separated into two groups on the basis of the presence or absence of crown rust disease, the estimated grain yield responses for Phase I and Phase II did not differ from the general averages presented previously (Table 3) . The lack of response in grain yield during Phase II could be attributed to moderate crown rust epidemics in all but one of the six evaluation experiments. 40% was the average final disease severity of the susceptible check (Starter) in the experiments with moderate epidemics. In Exp. 5 (Table 1), the average final crown rust severity was 89%, and the Phase II progress in resistance to crown rust was reflected in an average grain yield gain of 4.7% cycle-1. The positive association between partial resistance to crown rust and grain yield under high disease pressure has been documented previously. Holland and Munkvold (2001) studied this association in artificially inoculated field experiments, and report a genotypic correlation between AUDPC and grain yield of -0.63.
|
|
Broad-Sense Heritability Estimates
Hanson (1963) defined heritability as the fraction of the selection differential expected to be gained when selection is practiced on the declared reference unit. This definition emphasizes the importance of describing the basis (plant, plot, or replicated plot) used in the estimation. Table 5
presents broad-sense heritability estimates computed on an entry-mean basis for each recurrent selection cycle, and the corresponding reference units are presented in Table 1.
|
Luke et al. (1975) and Holland and Munkvold (2001) reported the only other heritability estimates for partial resistance to oat crown rust of which we are aware. Luke et al. (1975) evaluated crown rust resistance on the progeny of a resistant by susceptible cultivar cross and found a single-plant broad-sense heritability value of 87% using indirect estimates of environmental variation. Holland and Munkvold (2001) studied partial resistance in a random-mated population derived from crossing 10 partial resistance donor lines and 10 grain yield donor parents, and report an entry-mean broad-sense heritability estimate of 89% based on three environments. High heritability values for partial resistance to plant diseases are frequently observed in resistant by susceptible crosses. Qi et al. (1998) studied partial resistance to leaf rust in 103 recombinant inbred lines from a partially resistant by susceptible barley cross and reported a broad-sense heritability of 82% based on three replications at one environment. Bjarko and Line (1988) estimated broad-sense heritabilities for resistance to wheat leaf rust (caused by Puccinia triticina Eriks.) using data from one experiment with three replications. The highest heritability (92%) was obtained for the cross of the highly resistant by the susceptible cultivar, while the cross between the two slow-rusting varieties had the lowest heritability estimate (44%).
Grain yield heritabilities varied greatly between recurrent selection cycles and this fluctuation did not follow any apparent tendency (Table 5). Our evaluation experiments failed to produce precise estimates of the genetic variances for grain yield, and the poor quality of these estimates was reflected in the variability of the broad-sense heritabilities and genetic correlations between grain yield and other traits. Frey and Holland (1999) also estimated broad-sense heritabilities for oat grain yield after nine cycles of recurrent selection for increased groat-oil content, and their combined analysis of three replicated hill plot field trials produced similar and equally variable grain yield heritability estimates.
The flowering date and plant height broad-sense heritabilities presented in Table 5 are based on four environments and 10 replicates. Heritability estimates for flowering date ranged from 92 to 96% and from 69 to 91% for plant height. Our data set suggests that two replicates at two environments could provide a satisfactory assessment of the flowering date, while plant height would require four replicates at two environments. With these reference units the broad-sense heritability estimates would range from 83 to 92% for flowering date and from 66 to 87% for plant height.
This study demonstrated the usefulness of rapid cycle recurrent selection as a population improvement procedure capable of effectively increasing the level of partial resistance to crown rust in an adapted and high-yielding oat population. Although higher heritability estimates and greater gains per cycle could be obtained if the evaluation and selection were conducted on inbred lines, our results indicate that selection for partial crown rust resistance in early generations can produce adequate gains per recurrent selection cycle with a minimum cycle length. The use of hill plots as the experimental unit in the selection field trials did not hinder the ability of accurately assessing crown rust resistance. The benefit of additional replication more than compensated for the potential interplot interference effect caused by the plot type. In addition, our results confirm that the previous seven cycles of recurrent selection for grain yield produced a continuous positive response for grain yield. Single-trait recurrent selection induced correlated changes in other agronomically important variables during the two breeding phases. Therefore, proper correction measures should be implemented to prevent undesirable indirect responses in flowering date and plant height while selecting for grain yield or partial resistance to oat crown rust. Finally, broad-sense heritability estimates indicate that four replicates at two or three environments are required to assess properly partial resistance to crown rust, while satisfactory assessment of flowering date and plant height would demand two and four replicates at two environments, respectively.
|
ACKNOWLEDGMENTS |
|---|
|
NOTES |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONREFERENCES |
|---|
Received for publication August 10, 2001.
|
REFERENCES |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  T. S. Rani and R. L. Ravikumar Genetic Enhancement of Resistance to Alternaria Leaf Blight in Sunflower through Cyclic Gametophytic and Sporophytic Selections Crop Sci., March 1, 2007; 47(2): 529 - 534. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. Long, J. B. Holland, G. P. Munkvold, and J.-L. Jannink Responses to Selection for Partial Resistance to Crown Rust in Oat Crop Sci., April 25, 2006; 46(3): 1260 - 1265. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| 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 |
||||
