Genetic Relationships of Crown Rust Resistance, Grain Yield, Test Weight, and Seed Weight in Oat

mained effective in North America longer than 5 yr after the resistance genes were released in pure-line Integrating selection for agronomic performance and quantitative cultivars (Holland, 1997). Monogenic resistances can be resistance to crown rust, caused by Puccinia coronata Corda var. avenae W.P. Fraser & Ledingham, in oat (Avena sativa L.) requires overcome rapidly by new races of the pathogen that an understanding of their genetic relationships. This study was conemerge because of the selection pressure exerted by ducted to investigate the genetic relationships of crown rust resistance, large areas of uniformly resistant hosts (Harder and grain yield, test weight, and seed weight under both inoculated and Haber, 1992; Kolmer, 1997). Evolution of new races in fungicide-treated conditions. A Design II mating was performed becrown rust populations can occur via accumulation of tween 10 oat lines with putative partial resistance to crown rust and mutations in asexual populations or by sexual recombinine lines with superior grain yield and grain quality potential. Progenation on the alternate host, buckthorn (Rhamnus canies from this mating were evaluated in both crown rust-inoculated thartica L.), which occurs naturally in North America and fungicide-treated plots in four Iowa environments to estimate (Chong and Kolmer, 1993; Dinoor et al., 1988). genetic effects and phenotypic correlations between crown rust resisMethods proposed to improve the durability of crown tance and grain yield, seed weight, and test weight under either infection or fungicide-treated conditions. Lines from a random-mated poprust resistance in oat include gene pyramiding, gene ulation derived from the same parents were evaluated in three Iowa deployment, multiline breeding, and selection for partial environments to estimate heritabilities of, and genetic correlations resistance. Gene deployment, wherein breeders in difbetween, these traits. Resistance to crown rust, as measured by area ferent regions agree to release cultivars with different under the disease progress curve (AUDPC), was highly heritable sets of resistance genes, should exert disruptive, rather (H 0.89 on an entry-mean basis), and was favorably correlated with than directional, selection pressure on the pathogen grain yield, seed weight, and test weight measured in crown rustpopulation (Frey et al., 1973). Multilines are expected inoculated plots. AUDPC was unfavorably correlated or uncorrelated to exert stabilizing, rather than directional, selection on with grain yield, test weight, and seed weight measured in fungicidethe pathogen (Frey, 1982). Both gene deployment and treated plots. To improve simultaneously crown rust resistance, grain multiline breeding strategies depend upon the availabilyield, and seed weight under both lower and higher levels of crown rust infection, an optimum selection index can be developed with the ity of large numbers of effective resistance genes, which genetic parameters estimated in this study. are not currently available for crown rust resistance in oat. Compared with resistance from a single major gene, gene pyramiding may enhance the durability of resistance because it should be more difficult for virulence C rust, one of the most widespread and damagto two or more major resistance genes to develop in a ing diseases of oat (Harder and Haber, 1992), can single fungal genotype. The combination of resistance reduce grain yields (Endo and Boewe, 1958; Frey et al., genes Pc38 and Pc39 was released in the Canadian culti1973) and grain quality traits such as seed weight and vars Dumont, Riel, and Robert, and in the North Dagroat percentage (Simons and Browning, 1961; Simons kota cultivars Steele and Valley (McMullen and Patet al., 1979). Host plant resistance is the most economiterson, 1992). This combination of genes was no more cal control measure of oat crown rust (Harder and Hadurable than typical single gene resistances. Simultaneber, 1992). ous virulence to both Pc38 and Pc39 became frequent The most common form of resistance exploited by in Canadian rust populations after the release of cultioat breeders to date completely prevents reproduction vars with this gene combination (Chong and Kolmer, of the fungus on the host and segregates as a single 1993). gene. Monogenic resistance historically has not rePartial resistance should be more durable than racespecific, complete resistance because selection pressure J.B. Holland, USDA-ARS, Plant Science Research Unit, Dep. of on the rust population is reduced (Simons, 1972). EvoluCrop Science, North Carolina State Univ., Box 7620, Raleigh, NC tion of virulence to partial resistance is expected to be 27695-7620; G.P. Munkvold, Dep. of Plant Pathology, Iowa State slower than to complete resistance, although Puccinia Univ., Ames, IA 50011. Journal Paper No. J-18658 of the Iowa Agric. and Home Economics Exp. Stn., Ames, IA, Project No. 3368 and recondita Roberge ex Desmaz populations responded to 3260. Received 1 Sept. 2000. *Corresponding author (james_holland@ ncsu.edu). Abbreviations: AUDPC, area under the disease progress curve; SCA, specific combining ability. Published in Crop Sci. 41:1041–1050 (2001). 1042 CROP SCIENCE, VOL. 41, JULY–AUGUST 2001 artificial selection for a shorter latent period on partially yield, test weight, and seed weight under inoculated conditions and grain yield, test weight, and seed weight resistant wheat hosts, suggesting that fungal populations can evolve to overcome quantitative partial resistance under rust-free conditions as separate traits that can be included along with disease severity in a selection index. (Shaner et al., 1997). Durability of resistance is impossible to prove except in retrospect, but Stuthman (1995) The objectives of this study were to (i) develop an oat population segregating for quantitative genes affecting noted that the oat cultivar Portage has maintained its high level of partial resistance to crown rust from the grain yield, test weight, and seed weight, and resistance to crown rust; (ii) test for both additive and nonadditive time of its release in 1960 to the present. Two major impediments to selection for partial resisgenetic effects on resistance to crown rust, measured as AUDPC; and (iii) estimate heritabilities and genotypic tance to rust diseases are the difficulty in measuring partial resistance accurately and the generally quantitaand phenotypic correlations of crown rust severity, grain yield, seed weight, and test weight measured in crown tive inheritance of partial resistance. Area under the disease progress curve (AUDPC), which is based on rust-infected plots; grain yield, seed weight, and test weight measured in plots without substantial crown rust measurements of the percentage of leaf area infected made periodically during the growing season, is a useful infection; and grain yield, seed weight, and test weight tolerance ratios. The results of this experiment will measure of partial disease resistance in the field, but is very labor-intensive to measure. Partial resistance can guide future efforts to develop durably resistant oat cultivars with good agronomic performance. also be characterized by its components, including latent period, infection efficiency, infectious period, and spore production (Brake and Irwin, 1992; Parlevliet, 1979). MATERIALS AND METHODS Partial resistance is usually polygenically inherited and Inoculum often has low heritability, although there are exceptions to this (Brake and Irwin, 1992; Parlevliet, 1979; SiRace nonspecific partial resistance can be indistinguishable mons, 1972). from race-specific complete resistances controlled by separate genes if a heterogeneous pathogen population is used as the Simons (1972) suggested that measurement of tolerinoculum (Parlevliet, 1992). In such a situation, a race-specific ance traits is a more reliable method to identify genogene with major effects on resistance would prevent infection types with partial resistance than is direct measurement by a part of the inoculum population, resulting in a disease of partial resistance. Simons (1966) estimated tolerances reaction that may appear to be partial resistance. Therefore, for grain yield and seed weight of many oat genotypes we used a single isolate of P. coronata, isolate 345, from the grown in hill plots as ratios of the traits measured in Iowa State University P. coronata collection as inoculum for plots inoculated with P. coronata to the traits measured all experiments. This isolate is compatible with many A. sativa in rust-free plots in the same experiment. Heritabilities hosts (Wise and Gobelman-Werner, 1993). of yield and kernel weight tolerance ratios were large Urediniospores stored in evacuated glass tubes in liquid N enough to allow progress from selection for tolerance were heat-shocked at 40 C for 10 min and increased in the greenhouse by inoculating plants of the susceptible oat cultiin either trait (Simons, 1969); however, tolerance was, var, Markton. Urediniospores were collected by aspiration in some instances, associated negatively with yield pofrom greenhouse plants and used immediately for field experitential (Simons, 1985). Rosielle and Hamblin (1981) sugments or desiccated and stored at 80 C in microfuge tubes. gested that tolerance to stress will generally be correFor field inoculations, urediniospore suspensions (approxilated negatively with yield potential, and that selection mately 10 mL 1 ) were prepared in sterile distilled water with for tolerance is often not an appropriate breeding strat0.20% (v/v) Tween 20. egy. Carson and Wicks (1989) suggested that selection for yield under disease stress is expected to result in Population Development increased disease resistance and grain yield potential in Ten cultivars and lines were selected on the basis of previous the absence of disease stress. Selection for grain yield field evaluations to serve as donors of putative partial resisunder northern leaf blight [caused by Exserohilum turcitance genes (“rust resistance donor parents,” Table 1, Fig. 1). cum (Pass.) Leonard and Suggs] and diplodia stalk rot MN841810 and MN841823 are experimental lines developed [caused by Stenocarpella maydis [Berk.] Sutton] disease at the University of Minnesota that have exhibited durable stress in a maize population resulted in significant inpartial resistance to crown rust. UQP4-1 and UQAsc-1 are creases in grain yield in the absence of disease stress experimental lines developed at the University of Queensland, and in disease resistance, but did not significantly imAustralia. UQP4-1 is a selection from the cross of ‘Panfive’, prove grain yield under disease stress (Carson and which has good partial resistance to crown rust (Brake and Irwin, 1992), and ‘Panfour’. UQAsc-1 is a selection from the Wicks, 1993). cross of Panfive and ‘Ascencao’. H632-518 was reported preOat cultivars that produce high grain yields with good viously to have good seed weight tolerance to crown rust grain quality under both crown rust-free environments (Simons et al., 1987). Ten cultivars and lines with excellent and crown rust-conducive environments are ideal for grain yield and agronomic performance but lacking complete North Central U.S. production environments, where the resistance to predominant races of crown rust in Iowa were disease is endemic, but varies in intensity from year to also selected to serve as donors of favorable alleles for grain year. We suggest that an appropriate strategy to develop yield and other agronomic traits (“yield donor parents,” Table such cultivars with potentially more durable resistance 1, Fig. 1). to crown rust is to evaluate genotypes under both inThe 10 resistance donor parent lines were mated to the 10 yield donor parent lines using a Design II mating scheme oculated and rust-free conditions and to consider grain HOLLAND & MUNKVOLD: OAT AGRONOMIC PERFORMANCE AND QUANTITATIVE CROWN RUST RESISTANCE 1043 Table 1. Means of yield donor and rust resistance donor parent oat lines and cultivars and Markton check for crown rust AUDPC and agronomic traits measured under crown rust inoculation and in plots treated with a systemic fungicide to limit crown rust infection, estimated from four environments in 1997 and 1998. Traits measured in crown Traits measured in Means over rust rust-inoculated plots fungicide-treated plots treatments 100-seed Test Grain 100-seed Test Heading Line AUDPC Grain yield weight weight yield weight weight date Height g m 2 g kg m 3 g m 2 g kg m 3 dap† m Yield donor parents Armor 205 159 1.97 319 281 2.65 405 65 0.98 Brawn 33 258 2.93 371 315 3.17 398 68 0.94 Don 58 227 2.45 406 241 2.55 416 63 0.90 Hazel 28 216 2.61 392 280 2.73 413 66 0.97 IAR30-20 57 199 2.49 363 244 2.63 379 63 0.98 Ogle 137 215 2.48 336 290 2.94 405 65 0.99 Prairie 53 275 2.52 372 333 2.78 394 66 0.98 Premier 199 151 2.18 346 255 2.57 453 66 0.96 Sheldon 193 162 2.28 340 271 2.70 414 62 1.03 Starter 181 151 2.52 342 234 2.83 407 63 0.96 Group mean 115 201 2.44 359 274 2.75 409 65 0.97 Rust resistance donor parents Calibre 51 215 2.44 370 271 3.14 415 71 1.06 H632-518 102 122 2.57 401 158 2.67 402 62 1.05 Jim 30 264 2.67 421 334 2.91 424 63 1.02 Milton 30 229 2.44 376 314 2.76 429 68 0.95 MN841810 47 200 2.72 385 254 2.96 418 67 0.99 MN841823 97 148 2.64 304 228 2.93 333 67 1.08 Moore 102 188 2.28 375 278 2.81 419 67 1.07 UQP4-1 122 117 2.22 321 164 2.33 372 61 0.93 UQAsc-1‡ 140 113 2.06 – 133 2.20 – 63 0.96 Portage 29 140 2.58 369 206 2.66 408 68 1.18 Group mean 75 174 2.46 369 234 2.74 402 66 1.03 Resistance vs. yield Parents *** *** NS * *** NS NS *** *** Markton‡ 265 63 1.63 – 245 2.32 – 71 1.10 LSD 0.05 44 67 0.25 31 67 0.25 29 2 0.06 * P 0.05. *** Mean of rust resistance donor parents significantly different from mean of yield donor parents at the P 0.01. NS, no significant difference between mean of rust resistance donor parents and yield donor parents. † Days after planting. ‡ UQAsc-1 and Markton did not produce sufficient grain to measure test weight. (Hallauer and Miranda, 1988). F2 and F3 bulk progeny from and inoculated them as seedlings with P. coronata isolate 345 to test for the presence of major resistance genes in the pareneach of these 100 matings were grown as entries in field evaluations of the Design II mating in 1997 and 1998. In addition, tal lines. All parental lines, except the cultivar Jim, exhibited a fully susceptible seedling reaction. Jim was completely resisunrelated F1’s from the Design II mating were intermated to produce 83 full-sib families (4-way crosses) (Fig. 1). S0 plants tant to the isolate as a juvenile plant. The greenhouse screening was performed after the population development crosses were from each 4-way cross were grown as spaced plants in Aberdeen, ID, and S1 seed was harvested separately from each S0 made, therefore, Jim and all of its progeny were included in field evaluations. These entries were included in some of the plant. A single, randomly chosen S1 plant from each of two randomly chosen S0 plants per cross was grown in the greenstatistical analyses to provide information on complete and incomplete block effects, but they were eliminated from analyhouse in autumn, 1997. S2 seed was harvested separately from each S1 plant. Ten randomly chosen S2 progeny per S1 plant ses designed to estimate correlations or heritabilities, or to test for additive and nonadditive genetic effects (Fig. 1). were grown in the greenhouse in spring, 1998. S3 seed descended from a common S1 parent was harvested in bulk to form S1:3 families. Four of the 4-way crosses produced only Design II Mating Experiment one S0 plant, so a total of 162 S1:3 families were developed in this way. We also developed 36 F3:5 families from the biparental F2 progenies from each of the 100 matings in the Design II crosses using the greenhouse in the same way. These additional crossing experiment, along with each of the parent lines plus families were included to make the allelic contribution of each the susceptible check cultivar, Markton, were included as enoriginal parent to the population approximately equal. The tries in a field experiment grown in 1997 at the Iowa State F3:5 and S1:3 families together constituted 198 lines representing University Agronomy and Agricultural Engineering Research the random-mated population used to estimate heritabilities Farm, Boone Co., IA. The experimental design was a splitplot with inoculation treatment [either inoculation with P. and genotypic and phenotypic variances and covariances. The expected amounts of inbreeding and heterogeneity within and coronata or treatment with the systemic fungicide triadimefon, 1-(4-chlorophenoxy)-3,3-dimethyl-1-(H-1,2,3-triazol-1-yl)-2between families are identical for F3:5 and S1:3 families, so we refer to all lines in the random-mated population as S1:3 families. butanone, to prevent crown rust disease] as the whole-plot factor and genotype as the sub-plot factor. The experiment Selection for polygenic partial rust resistance is predicted to be effective only in populations lacking major genes conferwas replicated twice, each replication of a treatment was designed as an 11 by 11 square lattice. Plots were hills of 30 ring complete resistance to the inoculated isolate (Cox, 1995). Therefore, we grew the 20 parental lines in the greenhouse seeds each planted on a grid and spaced 0.3 m in perpendicular 1044 CROP SCIENCE, VOL. 41, JULY–AUGUST 2001 Fig. 1. Development of oat populations using a Design II mating of 10 “rust resistance donor” parents and 10 “yield donor” parents, followed by natural self-fertilization to produce F2 and F3 bulk progenies and F3:5 families from biparental crosses and intermating followed by selfing to produce S1:3 families from 4-way crosses. directions. Each plot occupied a 0.09-m area. Experiments was measured on the inoculated plots only. A second fungicide application was made 30 d after the first. The fungicide-treated were surrounded by two rows of border hills of the crown rust susceptible cultivar, Markton. Soil type at this location was plots were monitored to time reapplication of the fungicide when crown rust began to appear. a Nicollet loam (fine-loamy, mixed, mesic Aquic Hapludoll). Three tillers in each plot in the inoculated treatment whole Heading date (date after planting on which the first nodes on half of the plants in the plot had emerged completely above plots were inoculated with isolate 345 of P. coronata at the 3 to 4 leaf stage of development (Zadoks growth stage 13–14, the flag leaf) and plant height at maturity (excluding awns) were measured on each plot. All plants in a plot were bundled Zadoks et al., 1974) by injecting approximately 0.2 mL of a urediniospore suspension into each stem. Plants in the border together at harvest and dried at ambient temperature for at least 1 wk, after which the plants were threshed and grain hills were also inoculated. The fungicide treatment plots were not inoculated, but were sprayed with the systemic fungicide yield was measured on each plot. One hundred-seed weight was measured on each plot by averaging the weights of two triadimefon to prevent crown rust disease. One fungicide application (500 g a.i. in 815 L of H2O ha 1 ) was made at the 4 samples of 100 seeds. After weighing seeds, the grain from each of the three plots of an entry-whole-plot treatment comto 5 leaf stage (Zadoks growth stage 14–15, Zadoks et al., 1974) with a motorized backpack sprayer. bination was bulked together to provide sufficient seed for measuring test weight. Percent of leaf area infected was scored visually by a modified Cobb’s scale for cereal rust (Peterson et al., 1948) on the F3 seeds harvested from the F2 bulk entries in the 1997 experiment were used to replicate the experiment in three flag leaf and second leaf of four tillers in every plot in the inoculated treatment. Disease severity ratings were made on locations in 1998: the Agronomy and Agricultural Research Farm; the Hinds Research Farm, north of Ames, Story Co., four dates, at 3to 4-d intervals, after symptoms appeared on flag leaves. Mean percent leaf area infected, averaged over IA; and the Iowa State University Northern Research Farm, near Kanawha, Hancock Co., IA. The experimental design both flag and second leaves, was computed for each plot on each rating date. AUDPC was then computed for each plot, and execution were the same as in 1997, but three replicates were used at each location. Soil types were Coland clay (fineby means of the formula of Bjarko and Line (1988). AUDPC HOLLAND & MUNKVOLD: OAT AGRONOMIC PERFORMANCE AND QUANTITATIVE CROWN RUST RESISTANCE 1045 loamy, mixed, mesic Cumulic Endoaquoll) at Hinds Farm were computed by the delta method (Lynch and Walsh, 1997). Genotypic and phenotypic correlations among traits were estiand Canisteo (fine-loamy, mixed, mesic Typic Endoaquoll) at Kanawha. mated by the multivariate analysis of variance option in SAS Proc GLM (SAS Institute Inc., 1990). Standard errors of the correlations were estimated following Mode and Robinson Statistical Analysis of Design II Experiment (1959). Analysis was performed by SAS Proc Mixed (SAS Institute Inc., 1997), considering whole plot treatments and entries to be fixed effects, and environments, complete blocks, and inRESULTS complete blocks to be random effects. This overall analysis Design II Experiment was used to estimate the main effects of whole plot treatments and entries, and the entry-treatment combination means. Main Entries differed significantly for all traits measured, effects and interactions between entries and treatments were and entry treatment interaction was significant (P tested for significance in this analysis. The Design II mating 0.0001) for 100-seed weight and test weight, but not for analysis was conducted on the 100 F2 and F3 progeny entries grain yield (P 0.23). The Design II analysis indicated only. To take advantage of the complete and incomplete block that rust resistance donor parents varied significantly for information provided by the parents and check entries, withingeneral combining ability for all traits tested—AUDPC environment split-plot analyses were conducted, and means adjusted for block effects were obtained for each entry-treat(P 0.01), grain yield (P 0.01), 100-seed weight (P ment combination. The adjusted means of the progeny entries 0.02), and test weight (P 0.006) in inoculated plots; from each environment (excluding progenies of Jim) were and grain yield (P 0.001), 100-seed weight (P 0.001), then analyzed as a Design II by means of Proc Mixed, considerand test weight (P 0.0001) in fungicide-treated plots. ing resistance donor parents, yield donor parents, and their Yield donor-parent general combining ability was also interactions to be fixed effects, and considering environments, a significant (P 0.002) source of variation for all traits and interactions of environments with other factors to be except for grain yield in fungicide-treated plots (P random effects. Phenotypes measured in different treatments 0.06). The significant general combining ability variation were considered to be different traits. For example, grain yield in the progeny indicates that additive genetic effects are measured in inoculated plots was considered to be a separate significant for all traits in this population. Yield parent trait from grain yield measured in fungicide-treated plots. Tolerance ratios were computed for grain yield, 100-seed weight, rust resistance parent interaction was a highly significant and test weight as the ratio of the trait mean estimated in in(P 0.01) source of variation for all traits, indicating oculated plots to the trait mean estimated in fungicide-treated that nonadditive specific combining ability effects were plots. Correlations among traits were estimated on the basis important for these traits in this population. of progeny entry means over environments. The yield donor-parent lines per se had higher mean grain yield than the rust resistance parents under both Random-Mated Population Evaluation inoculated and rust-free conditions (Table 1). The rust In 1998, 198 S1:3 families representing a random-mated popresistance parents had better mean resistance to crown ulation were evaluated along with the parents, the susceptible rust disease (lower AUDPC) and higher mean test check cultivar, Markton, and the crown rust resistant check weight under crown rust inoculation than the yield docultivar, Gem. The 220 entries were arranged as an alpha nor parents (Table 1). Several of the yield donor parents, lattice with 11 entries within each of 20 incomplete blocks such as Brawn and Hazel, however, had crown rust within each of three complete replications within each treatresistance equivalent to the most resistant resistance ment at each location. The experimental designs, execution, donor parents. and locations were otherwise identical to the Design II experiEntry mean values for grain yield, 100-seed weight, ment in 1998. and test weight under crown rust inoculation were correlated either positively or not significantly with the same Statistical Analysis of Random-Mated traits measured in fungicide-treated plots (Table 2). Population Evaluation AUDPC was correlated negatively (r 0.42 to r The experiment consisted of both “fixed” effect entries 0.63) with these traits measured under crown rust inoc(parents and checks), and “random” entries (S1:3 families repulation, indicating that higher levels of resistance (mearesenting the random-mated population). Therefore, a first sured as lower AUDPC scores) tended to be associated analysis was performed using SAS Proc Mixed in which entries with higher grain yield and grain quality traits under and treatments were considered fixed effects, and locations, inoculation. Of the traits measured in fungicide-treated complete blocks, and incomplete blocks were considered ranplots, however, only mean 100-seed weight was negadom effects. This analysis was used to estimate the main effects of whole plot treatments and entries, the entry-treatment comtively correlated (r 0.35) with mean AUDPC. Test bination means, and the standard errors for mean compariweight in the fungicide-treated plots was positively corsons. The means of each S1:3 family for each trait-treatment related (r 0.30) with AUDPC, indicating that entries combination adjusted for block effects were obtained for each with higher levels of resistance (lower AUDPC) tended location. These means (excluding those of the 36 lines deto have lower test weight in fungicide-treated plots. scended from Jim) were then analyzed using Proc Mixed, Grain yield and 100-seed weight in the fungicide-treated considering families and environments to be random effects. plots were positively correlated with mean heading date Heritabilities were estimated on an entry-mean basis from (r 0.34 and r 0.37), but mean heading date did not these data, and on a plot basis from the original plot data, have a significant relationship with AUDPC or grain but excluding the parents, checks, and lines descended from Jim. Approximate standard errors of the heritability estimates yield and quality traits under inoculation (Table 2). 1046 CROP SCIENCE, VOL. 41, JULY–AUGUST 2001 Table 2. Correlations among 100 F2 and F3 oat progeny means from crosses between yield donor and rust resistance donor parent lines for traits measured under crown rust inoculation and in plots treated with a systemic fungicide to limit crown rust infection and for tolerance ratios, estimated from design II mating experiment in four Iowa environments in 1997 and 1998. Traits measured in crown Traits measured in Means over rust rust-inoculated plots fungicide-treated plots Tolerance ratios treatments 100 Seed Test Grain 100 Seed Test Grain 100 Seed Test Heading weight weight AUDPC yield weight weight yield weight weight date Height Traits measured in crown rust-inoculated plots Grain yield 0.61 0.34 0.58 0.61 0.56 NS 0.52 0.20 0.28 NS NS 100-seed weight 0.39 0.63 0.26 0.63 NS 0.42 0.62 0.53 NS NS Test weight 0.42 NS NS 0.40 0.38 0.29 0.71 NS NS AUDPC NS 0.35 0.30 0.47 0.45 0.65 NS NS Traits measured in fungicide-treated plots Grain yield 0.54 0.25 0.34 0.23 NS 0.37 NS 100-seed weight NS NS 0.23 NS 0.34 NS Test weight NS 0.30 0.37 NS NS Tolerance ratios Grain yield 0.47 0.51 0.27 0.21 100-seed weight 0.52 0.43 NS Test weight NS NS Means over rust treatments Heading date 0.19 † NS, not significant at P 0.05. All other correlations were significant at P 0.05. Random-Mated Population Experiment These results are generally congruent with those from the Design II experiment. Grain yield measured under The main effect of crown rust inoculation on the S1:3 crown rust inoculation was positively genetically correfamilies of the random-mated population and the parenlated with grain quality traits under the same conditions, tal and check line entries was to reduce grain yield 40% and with grain yield and quality traits in fungicide(P 0.07), 100-seed weight 17% (P 0.05), and test treated plots (Table 5). Seed weight measured under weight 21% (P 0.05). Entry main effects and entry inoculation was positively genetically correlated with treatment interaction effects were highly significant seed weight but not grain yield or test weight measured (P 0.0001) for all three traits, indicating that entries in fungicide-treated plots (Table 5). Therefore, grain differed both for mean grain yield, seed weight, and test yield measured under crown rust inoculation exhibited weight and also for responses of these grain phenotypes favorable genotypic correlations with all of the other to crown rust infection. The significant entry treattraits measured. ment interaction observed in the random-mated population contrasts with the result of the Design II experiment, but the interactions observed in the random-mated DISCUSSION population are applicable to predictions of selection Evaluating Partial Resistance to Crown Rust response because this is the Hardy-Weinberg equilibrium population to which selection will be applied. The Race-specific resistance genes may have affected our entry treatment interactions suggest that the grain estimates of crown rust resistance in this experiment; phenotypes measured under different inoculation treattherefore, we cannot be certain that our ratings reflected ments can be considered separate variables that are only partial resistance. Evidence for this is that the cultiaffected in part by unique sets of genes. Entries also var Jim was rated as only partially resistant in the field differed highly significantly (P 0.0001) for AUDPC when inoculated with isolate 345, whereas Jim was comfor crown rust. Several families were identified that pospletely resistant to isolate 345 as a seedling in greensessed favorable combinations of crown rust resistance house inoculations. Natural inoculum is prevalent in and grain yield and grain quality (Table 3; Fig. 2). Iowa and is impossible to exclude from field trials; thereHeritabilities on an entry-mean basis were high for fore, it is likely that races with virulence genes different all traits, ranging from 0.62 for grain yield under crown than isolate 345 were a part of the inoculum population. rust infection to 0.89 for AUDPC (Table 4). HeritabilitNevertheless, we suggest that isolate 345 dominated the ies on a plot basis tended to be much lower (Table 4). inoculum population because it was artificially inocuAUDPC was genotypically and phenotypically negalated before natural infection was observed in this extively correlated with grain yield, 100-seed weight, and periment or in surrounding oat plots. When it is not certain that the effects of race-specific major-effect retest weight under crown rust inoculation (Table 5). Lower values of AUDPC indicate higher levels of resissistances have been excluded from partial resistance measurements, Parlievliet (1992) suggested that selectance to crown rust, therefore higher levels of crown rust resistance likely contributed to increased grain yield tion be practiced against both the most resistant and least-resistant genotypes. and grain weight when plots were inoculated with crown rust. On the other hand, AUDPC was not significantly Timing of the ratings is critical to accurately measuring AUDPC. Crown rust severity increased quickly on correlated with grain yield and 100-seed weight and was correlated unfavorably (both genotypically and phenothe most susceptible genotypes; ratings on the same genotype taken only 3 d apart differed greatly in some typically) with test weight in fungicide-treated plots. HOLLAND & MUNKVOLD: OAT AGRONOMIC PERFORMANCE AND QUANTITATIVE CROWN RUST RESISTANCE 1047 Table 3. Means of selected S1:3 oat lines and parental and check lines for crown rust AUDPC and agronomic traits measured under crown rust inoculation and in plots treated with a systemic fungicide to limit crown rust infection, estimated from three Iowa environments in 1998. Traits measured in crown Traits measured in Means over rust rust-inoculated plots fungicide-treated plots treatments Grain 100 Seed Test Grain 100 Seed Test Heading Line AUDPC yield weight weight yield weight weight date Height g m 2 g kg m 3 m 2 g kg m 3 dap† m Experimental lines IA97009-2 5 171 2.73 322 192 2.90 311 55 1.07 IA97027-1 79 144 2.78 317 241 3.34 372 53 0.90 IA97038-2 25 150 2.86 312 173 3.50 353 52 0.91 IA97045-1 14 128 3.08 309 184 3.13 356 52 0.81 IA97062-2 40 180 2.82 366 236 3.31 387 54 0.95 IA97066-2 67 163 2.62 318 272 3.21 348 54 1.03 IA96498-2 67 177 2.59 305 240 3.08 403 56 0.94 IA96508-1 75 150 2.72 287 287 3.07 371 57 0.96 Parental lines Brawn 27 175 2.64 286 291 3.26 463 60 0.97 Don 28 146 2.34 351 223 2.86 400 54 0.91 Hazel 15 143 2.32 298 271 2.69 353 62 1.03 Ogle 159 155 2.46 292 282 3.12 375 55 0.96 Prairie 47 215 2.33 299 254 3.09 379 57 0.93 Premier 277 104 2.03 258 209 2.78 396 54 0.87 MN841810 31 137 2.49 301 158 2.83 330 57 0.98 Portage 42 143 2.47 325 258 2.88 383 56 1.06 Check lines Markton (susceptible)‡ 288 16 1.53 – 164 2.54 324 58 1.01 Gem (resistant) 12 172 3.05 324 214 3.37 360 56 0.98 LSD 0.05 71 65 0.33 57 65 0.29 57 1 0.10 † Days after planting. ‡ Markton did not produce sufficient grain under crown rust infection to estimate test weight. cases (Fig. 2). We observed that highly infected leaves senesced leaves on the later rating dates, and therefore we tended to rate the less infected plants of the most tended to senesce more rapidly than uninfected leaves. This caused some bias in our estimation of AUDPC, susceptible genotypes at later dates. This effect is illustrated by the disease progress curve exhibited by the because we were unable to rate the highly infected, Fig. 2. Mean crown rust severity ratings of selected experimental oat lines and cultivars at each of four rating dates in Boone County, IA in 1998. Within each date, mean disease severity was estimated by measuring disease severity on four flag leaves and four second leaves within each of three replicate plots. † Days after planting. 1048 CROP SCIENCE, VOL. 41, JULY–AUGUST 2001 Table 4. Heritability estimates (and their standard errors) of oat grain yield, 100-seed weight, test weight, and AUDPC measured in crown rust-inoculated plots; and grain yield, 100-seed weight, and test weight measured in plots treated with a systemic fungicide to limit crown rust infection. Estimates were based on 162 random S1:3 oat families evaluated in three Iowa environments in 1998. Traits measured in crown Traits measured in rust-inoculated plots fungicide-treated plots Grain 100-seed Test Grain 100-seed Test yield weight weight AUDPC yield weight weight Heritability on a plot-basis 0.22 (0.04) 0.58 (0.03) 0.56 (0.06)† 0.57 (0.04) 0.26 (0.04) 0.40 (0.04) 0.46 (0.06)† Heritability on an entry mean-basis 0.62 (0.06) 0.89 (0.02) 0.80 (0.04) 0.89 (0.02) 0.72 (0.04) 0.81 (0.03) 0.72 (0.05) † Test weight was measured on samples of grain bulked over replicate plots at a location, therefore this is heritability on a sample-basis. susceptible check Markton, which had a mean severity values not significantly lower than the well-adapted but crown rust-susceptible cultivar, Ogle (Table 1). H632of 61% on the third rating date in the random-mated population experiment at the Boone County location 518, which was released as a crown rust tolerant germplasm line (Simons et al., 1987), exhibited relatively in 1998, but only 34% on the fourth rating date in the same environment (Fig. 2). This phenomenon tended good tolerance to crown rust infection in terms of test weight, but was quite low-yielding, and did not have to bias our estimates of AUDPC of the more susceptible cultivars downward. outstanding crown rust resistance (Table 1). The poor adaptation of these lines to Iowa combined with only In opposition to the effect of senescence, interplot interference probably resulted in higher AUDPC values mediocre resistance to crown rust suggests that they did not contribute useful germplasm to the population. for the more resistant genotypes than would occur in larger field plots (Patanothai et al., 1975). The inoculated border rows were intended to provide a uniformly Gene Action and Heritability high level of inoculum for secondary infection. We deThe progeny from the Design II mating were tested sired to make selections under conditions of high disin the F2 generation in one environment and in the ease severity. F3 generation in three environments, making precise Despite these difficulties, AUDPC was a good meainterpretation of the rust parent yield parent interacsure of the rate of crown rust development on different tion (specific combining ability, SCA) difficult. The genotypes, as shown by our ability to detect statistically progenies tested were not highly inbred, therefore domisignificant differences among genotypes for AUDPC. nance gene effects may have contributed to SCA. It is The F-statistics for genotype effects were higher for likely that epistatic effects contributed substantially to AUDPC than for disease severity taken at any single the SCA variation for AUDPC and grain yield and 100date within locations in the random-mated population seed weight under both disease treatments, because for evaluation, and AUDPC integrates information from these traits, the yield parent rust resistance parent multiple dates into a single variable. Therefore, AUDPC interaction was highly significant even when the F2 genprovided the most appropriate measurement of resiseration data were excluded. Stuthman and Stucker tance to crown rust in this population. (1975) reported significant SCA variation for grain yield in highly inbred oat progeny lines, which must have been Parent Selection the result of epistatic effects, rather than dominance effects. The rust resistance parents, on average, had better resistance to crown rust disease than the yield parents, The high entry-mean heritabilities for all traits measured (Table 4) suggest that all traits should respond but there were several rust resistance donor parents that exhibited relatively poor crown rust resistance. The well to selection on the basis of line means. Heritabilities on a plot basis for grain yield, particularly, were much Australian parent lines, for example, had mean AUDPC Table 5. Genotypic and phenotypic correlation estimates (and their standard errors) among grain yield, 100-seed weight, test weight, and AUDPC measured in crown rust-inoculated plots; and grain yield, 100-seed weight, and test weight measured in plots treated with a systemic fungicide to limit crown rust infection. Estimates were based on 162 random S1:3 oat families evaluated in three Iowa environments in 1998. Genotypic correlations are given in the upper right half of the table, phenotypic correlations are given in the lower left. Traits measured in crown Traits measured in rust-inoculated plots fungicide-treated plots Grain 100-seed Test Grain 100-seed Test yield weight weight AUDPC yield weight weight Traits measured in crown rust-inoculated plots Grain yield – 0.51 (0.09) 0.58 (0.10) 0.63 (0.09) 0.49 (0.11) 0.24 (0.12) 0.27 (0.13) 100-seed weight 0.45 (0.07) – 0.38 (0.09) 0.40 (0.08) 0.10 (0.10) 0.76 (0.05) 0.10 (0.10) Test weight 0.47 (0.08) 0.34 (0.09) – 0.44 (0.09) 0.04 (0.12) 0.07 (0.11) 0.41 (0.12) AUDPC 0.47 (0.07) 0.37 (0.07) 0.33 (0.10) – 0.08 (0.10) 0.07 (0.10) 0.31 (0.10) Traits measured in fungicide-treated plots Grain yield 0.35 (0.08) 0.08 (0.08) 0.06 (0.11) 0.07 (0.08) – 0.27 (0.10) 0.54 (0.09) 100-seed weight 0.21 (0.08) 0.66 (0.05) 0.09 (0.11) 0.05 (0.08) 0.25 (0.08) – 0.17 (0.11) Test weight 0.20 (0.09) 0.08 (0.09) 0.29 (0.10) 0.24 (0.08) 0.46 (0.07) 0.20 (0.08) – HOLLAND & MUNKVOLD: OAT AGRONOMIC PERFORMANCE AND QUANTITATIVE CROWN RUST RESISTANCE 1049 lower (0.22 and 0.26, Table 4), indicating the importance ease resistance. We would make the same prediction for this population, because the observed pattern of of replication and multiple-environment testing for evaluating grain yield. The relatively high heritability for genetic correlations suggests that selection for grain yield under disease stress alone will result in correlated AUDPC, both on a plot basis (0.57) and on an entrymean basis (0.89) suggests that evaluation of percentage improvements in the other traits of interest. Grain yield under disease stress and grain yield in the of leaf area infected on two leaves on each of four plants per plot within each of several rating dates was absence of substantial disease stress can be considered distinct traits that may be under the control of different an effective method to distinguish the partial crown rust resistance of oat lines. This result contrasts with the sets of genes. Population improvement for multiple traits can be achieved by different methods, including conclusion of Simons (1972) that heritability of partial resistance tends to be low. The genotypic variance that independent culling, tandem selection, or index selection (Young, 1961) If relative economic values for imconstitutes the numerator of these heritability estimates is an estimate of (3/2) A (1/32) D (11/4)D1 (41/ provement are known, then optimal selection indices will be superior to all other methods of multiple trait im64)D*2 (1/64)H* (9/4) AA, where terms are defined in Nyquist (1991). Given the evidence for epistatic gene provement (Falconer and Mackay, 1996; Young, 1961). To estimate relative weights for an optimal selection effects in this population from the Design II experiment, the estimates of heritability are likely biased upward by index, genotypic and phenotypic variances and covariances among the traits included in the index must be additive additive epistatic variance, and perhaps by other nonadditive components of genetic variance. In known or estimated (Baker, 1986). Evaluation of random S1:3 lines from the random-mated population peraddition, genotype year interaction variances are confounded with the genotypic variance component estimitted estimation of the relevant genotypic and phenotypic variances and covariances. Optimal index selection mates since the lines were evaluated only in one year. has not been widely used in plant breeding because Therefore, predictions of response to selection based relative economic values associated with different traits on the heritability estimates presented in Table 4 are are generally unknown, and genetic and phenotypic coprobably overestimates of the true response to selection. variances between traits are not routinely estimated. Carson and Wicks (1989) suggested that selection for Trait Correlations yield under disease stress would be a good alternative The absence of strong unfavorable correlations to index selection for yield under stress and nonstress among most traits indicated that simultaneous improveenvironments and for disease resistance because it ment of grain yield, test weight, and seed weight should would avoid these problems, and it would have the addinot be unduly difficult in this population. The correlational advantage of eliminating the time and expense tions between mean tolerance ratios and grain yield required to make multiple disease severity ratings. and grain quality measures illustrate the typical pattern Given that we have already measured AUDPC for described by Rosielle and Hamblin (1981). Under stress the lines in this population and estimated the genotypic (in this case, crown rust infection), grain yield, seed and phenotypic covariances, however, development and weight, and test weight were correlated positively with use of a selection index is most appropriate. In lieu of tolerances for those traits (Table 2). With negligible assigning relative economic values to each of the traits, crown rust stress in the fungicide-treated plots, grain one could create an aggregate breeding value that yield, seed weight, and test weight were negatively or weights standardized improvements in grain yield, seed not significantly associated with tolerance (Table 2). weight, and AUDPC measured under crown rust inocuThus, selection for increased tolerance to crown rust lation, and grain yield and seed weight measured in resistance would likely result in selection for genotypes fungicide-treated plots equally. A difficulty with selecthat perform better under infection, but worse in the tion for crown rust resistance will be to avoid selecting absence of disease. Crown rust disease is endemic in genotypes with race-specific major-effect resistance the U.S. North Central region, but varies in intensity genes. While we attempted to minimize the effects of among years and among sites within states. Therefore, these genes, they may still have affected our results. oat cultivars that perform well under varying levels of Since the index incorporates four other traits in addition crown rust disease are desired. Selection for crown rust to AUDPC, selection pressure for crown rust resistance disease tolerance would likely not contribute to this would be reduced and therefore the likelihood of selecting only major-effect resistance genes would be lesgoal. sened. Advanced generation lines developed from the Selection for increased levels of crown rust resistance population could be tested later to determine if they (lower AUDPC) rather than tolerance in this populacontain race-specificic resistance genes. Another possition would be expected to result in improved grain yield, bility is to eliminate those lines with the highest levels seed weight, and test weight under infection, but would of crown rust resistance before developing the selecnot improve grain yield or 100-seed weight in diseasetion index. free conditions, and would result in lower test weight in disease-free conditions. Carson and Wicks (1989) sugACKNOWLEDGMENTS gested that selection for yield under disease stress will generally result in improvements in yield in both disThis research was supported in part by The Quaker Oats Co. We thank Dr. Vanessa Brake of the University of Queensease-stressed and disease-free environments and in dis1050 CROP SCIENCE, VOL. 41, JULY–AUGUST 2001 Sorrells (ed.) Oat science and technology. Agron. Monogr. 33. land and Dr. Kurt Leonard, USDA-ARS Cereal Rust LaboraASA, Madison, WI. tory for donating experimental germplasm, and Dr. Darrell Mode, C.J., and H.F. Robinson. 1959. Pleiotropism and the genetic Wesenberg, USDA-ARS Small Grains Research Facility, for variance and covariance. Biometrics 15:518–537. growing out the S0 plants. We thank John Shriver, Jessica Nyquist, W.E. 1991. Estimation of heritability and prediction of selecBetts, Nick Chambers, and Sara Helland for assistance with tion response in plant populations. Crit. Rev. Plant Sci. 10:235–322. disease ratings in the field. Parlevliet, J.E. 1979. Components of resistance that reduce the rate of epidemic development. Euphytica 17:203–222. Parlevliet, J.E. 1992. What is durable resistance, a general outline. p.REFERENCES23–39. In T. Jacobs and J.A. Parlevliet (ed.) Durability of disease resistance. Kluwer, Dordrecht, the Netherlands.Baker, R.J. 1986. Selection indices in plant breeding. CRC Press, BocaPatanothai, S., L.J. Michel, and M.D. Simons. 1975. Comparison ofRaton, FL.different hill-plot designs for evaluating quantitative response toBjarko, M.E., and R.F. Line. 1988. Quantitative determination of theoat crown rust. Crop Sci. 25:803–806.gene action of leaf rust resistance in four cultivars of wheat, Triti-Peterson, R.F., A.B. Campbell, and A.E. Hannah. 1948. A diagram-cum aestivum. Phytopathology 78:451–456.matic scale for estimating rust intensity on leaves and stems ofBrake, V.M., and J.A.G. Irwin. 1992. Partial resistance of oats to P.cereals. Can. J. Res. 26:496–500.coronata f. sp. avenae. Aust. J. Agric. Res. 43:1217–1227.Rosielle, A.A., and J. Hamblin. 1981. Theoretical aspects of selectionCarson, M.L., and Z.W. Wicks, III. 1989. Theory of selection for yieldfor yield in stress and non-stress environments. Crop Sci. 21:943–of plants in disease-stress environments. Phytopathology 79:70–76.946.Carson, M.L., and Z.W. Wicks, III. 1993. Response of a maize syntheticSAS Institute Inc. 1990. SAS/STAT user’s guide, vers. 6, 4th ed. SASto S1 recurrent selection for grain yield in a disease-stress environ-Institute, Cary, NC.ment. Maydica 38:193–199.SAS Institute Inc. 1997. SAS/STAT software: Changes and enhance-Chong, J., and J.A. Kolmer. 1993. Virulence dynamics and phenotypicments through release 6.12. SAS Institute, Inc., Cary, N.C.diversity of Puccinia coronata f.sp. avenae in Canada from 1974 toShaner, G., G. Buechley, and W.E. Nyquist. 1997. Inheritance of1990. Can. J. Bot. 71:248–255.latent-period of Puccinia recondita in wheat. Crop Sci. 37:748–756.Cox, T.S. 1995. Simultaneous selection for major and minor resistanceSimons, M.D. 1966. Relative tolerance of oat varieties to the crowngenes. Crop Sci. 35:1337–1346.rust fungus. Phytopathology 56:36–40.Dinoor, A., N. Eshed, and E. Nof 1988. Puccinia coronata, crown rustSimons, M.D. 1969. Heritability of crown rust tolerance in oats. Phyto-of oat and grasses. Adv. Plant Path. 6:333–344.pathology 59:1329–1333.Endo, R.M., and G.H. Boewe. 1958. Losses caused by crown rust ofSimons, M.D. 1972. Polygenic resistance to plant disease and its useoats in 1956 and 1957. Plant Dis. Reptr. 42:1126–1128.in breeding resistant cultivars. J. Environ. Qual. 1:232–240.Falconer, D.S., and T.F.C. Mackay. 1996. Introduction to quantitative Simons, M.D. 1985. Crown rust. p. 131–172. In A.P. Roelfs and W.R.genetics, 4th ed. Longman Technical, Essex, UK.Bushnell (ed.) The cereal rusts: Vol. II, diseases, distribution, epide-Frey, K.J. 1982. Multiline breeding. p. 43–71. In I.K. Vasil et al.miology, and control. Academic Press, Orlando, FL.(ed.) Plant improvement and somatic cell genetics. Academic Press, Simons, M.D., and J.A. Browning. 1961. Seed weight as a measure ofNew York.response of oats to crown rust infection. Iowa Acad. Sci. 68:114–118.Frey, K.J., J.A. Browning, and M.D. Simons. 1973. Management of Simons, M.D., L.J. Michel, and K.J. Frey. 1987. Registration of threehost resistance genes to control diseases. Zeits. Pflanzenkrank-oat germplasm lines resistant to the crown rust fungus. Crop Sci.heiten. Pflanzunschutz. 80:160–180.27:369.Hallauer, A.R., and J.B. Miranda, Fo. 1988. Quantitative genetics in Simons, M.D., V.L. Youngs, G.D. Booth, and R.A. Forsberg. 1979.maize breeding. 2nd ed. Iowa State Univ. Press, Ames, IA.Effect of crown rust on protein and groat percentages of oat grain.Harder, D.E., and S. Haber. 1992. Oat diseases and pathological tech-Crop Sci. 19:703–706.niques. p. 307–425. In H.G. Marshall and M.E. Sorrells (ed.) Oat Stuthman, D.D. 1995. Oat breeding and genetics. p. 150–176. In R.W.science and technology. Agron. Monogr. 33. ASA, Madison, WI.Welch (ed.) The oat crop: Production and utilization. ChapmanHolland, J.B. 1997. Oat improvement. p. 57–98. In M.S. Kang (ed.)and Hall, London, England.Crop improvement for the 21st century. Research Signpost, TrivanStuthman, D.D., and R.E. Stucker. 1975. Combining ability analysisdrum, India.of near-homozygous lines derived from a 12-parent diallel cross inKolmer, J.A. 1997. Virulence dynamics and genetics of cereal rustoats. Crop Sci. 25:800–803.populations in North America. p. 139–156. In I.R. Crute et al. (ed.) Wise, R.P., and K.S. Gobelman-Werner. 1993. Resistance to oat crownThe gene-for-gene relationship in plant-parasite interactions. CABrust in diploid and hexaploid Avena. Plant Dis. 77:355–358.International, New York.Young, S.S.Y. 1961. A further examination of the relative efficiencyLynch, M., and B. Walsh. 1997. Genetics and analysis of quantitativeof three methods of selection for genetic gains under less restrictedtraits. Sinauer Associates, Inc., Sunderland, MA.conditions. Gen. Res. 2:106–121.McMullen, M.S., and F.L. Patterson. 1992. Oat cultivar development Zadoks, J.C., T.T. Chang, and C.F. Konzak. 1974. A decimal code for the growth stages of cereals. Weed Res. 14:415–421.in the U.S.A. and Canada. p. 573–612. In H.G. Marshall and M.E.