MSU LIBRARIES n . RETURNING MATERIALS: Place in book drop to remove this checkout from your record. .Elfl§§ will be charged if book is returned after the date stamped below. THE INFLUENCE OF UNNECESSARY VIRULENCE GENES ON THE REPRODUCTIVE FITNESS OF ERYSIPHE GRAMINIS F. SP. TRITICI By Charlotte Ruth Bronson A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Botany and Plant Pathology 1981 ABSTRACT THE INFLUENCE OF UNNECESSSARY VIRULENCE GENES ON THE REPRODUCTIVE FITNESS OF ERYSIPHE GRAMINIS F. SP. TRITICI By Charlotte Ruth Bronson Vanderplank postulated that virulence genes in excess of those required to attack a particular host line will reduce the fitness of a pathogen to thrive and produce successful offspring. Vanderplank called this phenomenon "stabilizing selection“. The objective of this research was to determine whether or not unnecessary virulence genes reduce the reproductive fitness of Erysiphe graminis f. sp. tritici, causal organism of powdery mildew of wheat. Two isolates of g. graminis f. sp. tritici, MS-l and Mo-lO, were tested for virulence on 20 differential wheat lines. The results suggested that the isolates differed by at least four virulence genes (g genes). Analysis of progeny from a cross of MS-l and Mo—lO confirmed this observation. No linkages were detected between the P loci; however, ‘locus P? was found to be linked to a locus controlling mating type (mag). MS-l was assigned the genotype 32 33a 34 515889 fly. Mo-lO was assigned the genotype 32 33a E4 315889 _m_a_t_-. The relative fitnesses of isolates MS-l and Mo-10 were estimated by growing them as a mixture on the susceptible wheat variety Chancellor in growth chambers and measuring changes in their relative frequency over six to twelve conidial generations. Isolate Mo—lO decreased in frequency in the population at a rate that indicated that it was 19 to 32 percent less fit than MS-l. Charlotte Ruth Bronson The relative fitnesses of progeny from a cross between MS-l and Mo-IO were estimated in similar mixture studies to determine whether the difference in parental fitness was attributable to virulence genes. Analysis of twenty-four progeny showed that the large difference in fitness between the parental isolates segregated from the known virulence genes and that the virulence genes had no significant effect on fitness. Thus, no support was found in this system for Vanderplank's hypothesis that unnecessary virulence genes reduce the reproductive fitness of plant pathogens. TABLE OF CONTENTS LIST OF TABLES ..........OOOOOOOOOOOOOOO..........OOOOOOOOOOOOOOOOOO LIST OF FIGURES ............OOOOOOOOOOOOOOO...0..OOOOOOOOOOOOOOIOOOO INTRODUCTION AND LITERATURE REVIEW ......OOOOOOOOCOOOCCOOO0.....0... Stabilizing Selection ......................................... Evidence for and Against Stabilizing Selection ................ The Segregation Test .......................................... Biology Of Powdery Mildew of Wheat ............................ MATERIALS AND METHODS .........OOOOOOOOOOOOO......OOOOOOOOOOOOOOOOOO Seed Stocks ................................................... Powdery Mildew Isolates: Source and Maintenance .............. Powdery Mildew Isolates: Classification by Infection Types ... Powdery Mildew Isolates: Classification by Mating Type ....... CTOSS 0f MS'I and MO‘IO 00000000000000.000.00.000...0.000000... Determination of the Relative Fitnesses of Mildew Isolates .... RESULTS 00.0.00.........OOOOOOOOOOOOOOOO0.0.0.0.........OOOOODOOOOOO Parental Isolates: Genotype Assignments ...................... Parental Isolates: Determination of Relative Fitness ......... Progeny: Genotype Assignments and Genetic Analysis Of Viru1ence ......OOOOOOOO....O.......COOOOOOOOOOOOOO0...... Progeny: Determination of Relative Fitness and Its RelationShip to VirUIence Genes OOOOOOOOOOOOOOO......OOOOOOOO DISCUSSION .........OOOOOOOO............OOOOOOOOOOOOOOO00.00.0000... APPENDIX A: Discrete Selection Model for Strains of Asexually Reproducing Organisms Competing in a complex MiXture .....OOOOOOCOOOOO...OIOOOOOOOOOOOO APPENDIX B: Evidence for Environmental Effects on Ralative Fitness 0............OOOOOOOOO0.0.000000...... APPENDIX C: Data Use for Calculation of Selection C0€ff1CientS 0..........OOOOOOOOOOOOOOOI000.00.000.00.0 LITERATURE CITED oooeoeo00000000000000.0000...00000000000000.0000... ii Page iii V 64 68 75 81 LIST OF TABLES Table Page 1 Wheat lines used to classify the virulence phenotypes of isolates of E. graminis f. Sp. tritici .................. 15 2 Infection typesa of E. raminis f. Sp. tritici isolates on differentiai wheat lines 7 days after inoculation onto wheat seedlings in the greenhouse ......... 28 3 Relative fitnesses and selection coefficients for E. graminis f. Sp. tritici isolates MS-l and Mo-lD ........... 31 4 Progeny of E. graminis f. Sp. tritici isolates MS-l and Mo-10: infection types as observed under laboratory conditions and genotype assignments ............. 32 5 Infection types of isolates MS-l and Mo-lD of g. graminis f. Sp. tritici and twenty-four progeny from a cross of MS-l and Mo-ID on differential wheat lines on which segregation was observed (after 7 days under greenhouse conditions) ..................................... 34 6 Mating types of E. graminis f. Sp. tritici isolates as determined by the presence (yes) or absence (no) of cleistothecia 4 weeks after inoculation .................... 35 7 Proposed genotypes of E. graminis f. Sp. tritici isolates MS-l, Mo-10 and their progeny ..................... 36 8 Segregation of alleles for virulence and mating type in the progeny of E. graminis f. Sp. tritici isolates "5-1 and "0'10 ......OOOOOOOOOOOOOO......OOOOO...0.0.00.0... 37 9 Linkage analysis of the progeny of E. graminis f. Sp. tritici isolates MS-l and Mo-10 ............................ 39 10 Selection coefficients for the progeny of E. graminis f. Sp. tritici cultures MS-l and Mo-10. Isolates used as the standard for calculation in each set were defined as having a selection coefficient of zero .......... 50 11 Statistical analysis of selection coefficients of the progeny of E. graminis f. sp. tritici isolates MS‘I and ”0.10 oooo0.00000000000000000000000000000...ooooooo 56 iii Table 12 13 14 Page Average selection coefficients associated with alleles for virulence and mating type in the progeny of isolates MS-l and Mo-10 of E. graminis f. sp. trit1c1. ......OOOOOOO0......00.000000000000000...0.00 57 Effect of continuous light on the growth of isolates of E. graminis f. sp. tritici ..................... 73 Data used to calculate isolate frequencies and selection coefficients for cultures of E. graminis f. Sp. tritici. The data are recorded as t e number of individuals of each isolate per sample ........... 76 iv LIST OF FIGURES Figure Page 1 Sequence of events in tests for the relative fitnesses of cultures of E. graminis f. Sp. tritici. The number on the pot indicates the number of days since planting ............................................ 24 2 Frequencies of E. raminis f. sp. tritici isolates MS-1 (open squares) and Mo-10 (clos Ed circles) during Six to twelve conidial generations on the susceptible wheat line Chancellor ......................... 30 3 Frequencies of isolates of E. raminis f. sp. tritici during Sixteen conidiai generations on the susceptible wheat line Chancellor (experiment 4) .......... 41 4 Frequencies of isolates of E, graminis f. Sp. tritici during Sixteen conidia generations on the susceptible wheat line Chancellor (experiment 5) .......... 42 5 Frequencies of isolates of E. gr raminis f. sp. tritici during Sixteen conidia generations on the susceptible wheat line Chancellor (experiment 6) .......... 43 6 Frequencies of isolates of g. graminis f. Sp. tritici during Sixteen conidia generations on the susceptible wheat line Chancellor (experiment 7) .......... 44 7 Frequencies of isolates of E. graminis Sf. Sp. tritici during Sixteen conidial generations on the susceptible wheat line Chancellor (experiment 8) .......... 45 8 Frequencies of isolates of_§. graminis f. sp. tritici during Sixteen conidial generations on the susceptible wheat line Chancellor (experiment 9) .......... 46 9 Frequencies of isolates of E. graminis f. Sp. tritici during Sixteen conidia generations on the susceptible wheat line Chancellor (experiment 10) ......... 47 10 Frequencies of isolates of E. graminis f. Sp. tritici during Sixteen conidia generations on the susceptible wheat line Chancellor (experiment 11) ......... 48 Figure 11 12 13 14 Page Frequencies of isolates of E. graminis f. Sp. tritici during sixteen conidia generations on the susceptible wheat line Chancellor (experiment 12) ......... 49 Linearization of curves from a theoretical selection experiment. a) Change in frequency of isolate A (s=.00) and isolate B (s=.22). b) Plot of transformed data in which p is the frequency of isolate A and 9.15 the frequency of isolate B ............. 66 Frequency of isolate MS- 1 of E. minis f. Sp. tritici in a mixture of MS-l and Mo-IU during 17 weeks on the susceptible wheat variety Chancellor. Vertical barS represent 95% confidence intervals for the frequency of MS-l ................................. 70 Correlation between growth inhibition during exposure to continuous light and reduced competitive ability as measured by selection coefficients in twenty-six cultures of E. graminis Sf. Sp. tritici (rL .67) ........... 74 vi INTRODUCTION AND LITERATURE REVIEW Since the discovery of genetically controlled disease resistance at the turn of the centry, plant breeders have been incorporating genes for resistance into crop plants (Day, 1974). Unfortunately, inherited resistance to many diseases is effective for only a few years as a result of the appearance of new races of the pathogen. The capacity of plant pathogens to adapt to genetic changes in the host has concerned both plant breeders and plant pathologists. In 1953, Flor noted that races of flax rust, Melampsora lini, with wide host ranges were relatively uncommon. Watson (1958) noted a Similar phenomenon with Puccinia graminis f. Sp. tritici, causal organism of stem rust on wheat. Based on these observations and others, Vanderplank (1963, 1968) hypothesized the existence of a phenomenon that he hoped would offer a solution to the problem of adaptable pathogens. He called the phenomenon "stabilizing selection". This dissertation will discuss this hypothesis, its importance, and methods of verification. Finally, I will describe my efforts to test Vanderplank's hypothesis with Erysiphe graminis f. Sp. tritici. Stabilizing Selection The term “stabilizing selection" is used in population genetics to describe selection against extreme phenotypes (Strickberger, 1976). Its effect is to reduce the variability of the population without altering the average phenotype. As pointed out by Crill (1977), Vanderplank used the term to describe a specific type of directional selection. According to Vanderplank (1968): "When a gene for vertical resistance is present in a host plant, races of the pathogen must have enough virulence to match this gene, if they are to survive. But when the gene is absent, virulence to match it is unnecessary; and stabilizing selection operates in favor of races of the pathogen without unnecessary virulence." "Unnecessary virulence reduces the fitness of a race to survive.“ Vanderplank postulated that the genes for virulence that permit a pathogen to attack resistant host lines have a deleterious effect on the ability of the pathogen to grow and reproduce. In the absence of resistant host plants, these virulence genes are unnecessary and pathogen races with them decrease in frequency in favor of races without unneces- sary virulence genes. This presumably prevents the continuous accumulation of virulence genes in the pathogen and “stabilizes" its genotype. For the purposes of this dissertation the term “stabilizing selection" will refer to directional selection against unnecessary virulence genes in plant pathogens as hypothesized by Vanderplank. Vanderplank applied his concept to both obligate and nonobligate plant pathogens and suggested mechanisms to explain the effect of unnecessary virulence genes. According to Vanderplank (1968), when obligately parasitic plant pathogens mutate from avirulence to virulence, they modify their normal metabolic pathways. The substitute pathways are presumably less efficient and the pathogen is therefore less fit. Pryor (1977) argued that because virulence is often recessive, mutations to virulence constitute a loss of normal gene function. Thus, the “cost of virulence" is that the pathogen becomes less fit. For pathogens that spend part of their life cycle in a saprophytic phase and for which virulence can be a positive function, such as the ability to make a toxin (Scheffer, 1976), Vanderplank noted that the characteristics that make an organism a good saprophyte may be different than those that make it a good parasite. Therefore genetic changes that increase parasitic ability may decrease saprophytic ability and stabilizing selection will occur in the saprophytic phase. According to Vanderplank, not all virulence genes have a debilita- ting effect on a pathogen. If a virulence gene reduces the fitness of a pathogen dramatically, the corresponding host resistance gene is, in Vanderplank's words "strong". If the virulence gene has little or no effect on fitness, the host resistance gene is “weak“. Knott (1971) disagrees with the concept of using the behavior of genes for virulence in the pathogen to describe genes for resistance in the host. The “weak gene" concept has also been attacked by Nelson (1972) who points out that it can be used to dismiss data that do not support the hypothesis. If virulence genes really do make a pathogen significantly less fit, our ability to control plant disease through the use of various strate- gies of gene deployment may be enhanced (Browning and Frey, 1969; Frey g; 31., 1977; Marshall, 1977; Parlevliet, 1981; Vanderplank, 1968). When the resistance of cultivars "breaks down", that is, when pathogens with the virulence genes necessary to attack those cultivars become common, the cultivars could be removed from commercial use. The frequency of virulence genes should then decline and the resistance genes in those cultivars could eventually be reused, thus recycling the available resistance genes (Person, 1967). Stabilizing selection, if it is strong enough, will also permit long-term disease control through the use of mixtures of plants with different genes for resistance (multilines and cultivar mixtures) and through the use of more than one resistance gene in the same host line (multigene varieties). Presumably, a pathogen with the many virulence genes required to attack a large number of the components in a mixture or attack a host with many resistance genes would not be sufficiently vigorous to cause an epidemic. Numerous articles have been written on the subject of stabilizing selection, including reviews that discuss the available evidence, the theoretical reasonableness of the hypothesis and the potential usefulness of stabilizing selection (Brown, 1975; Browning and Frey, 1969; Crill, 1977; Leonard, 1980; Nelson, 1972; Person, Groth and Mylyk, 1976; Watson, 1970). Terms representing stabilizing selection have even been incorpo- rated into models describing disease develOpment and the p0pulation dynamics of plant pathogens (Barrett, 1978; Fleming and Person, 1978; Groth, 1976; Groth and Person, 1977; Kiyosawa, 1977; Leonard, 1969b; Leonard, 1977a; Leonard, 1980; Marshall and Pryor, 1978; Trenbath, 1977). In spite of the great deal of thought generated by the concept and the amount of effort expended to prove the hypothesis, there is no convincing evidence for the existence of stabilizing selection as Vanderplank defined it. Evidence for and against stabilizingselection The objective of studies on stabilizing selection is to determine whether or not unnecessary virulence genes reduce the reproductive fitness of plant pathogens. Stabilizing selection can be inferred by: 1) observing a low frequency of unnecessary virulence genes in a population when a high frequency might otherwise be expected (by means of race surveys), 2) observing a decline in the frequency of individuals with unnecessary virulence genes compared to individuals without those genes (by means of competition studies), or 3) comparing the fitness attributes (infection efficiency, latent period, etc.) of individuals possessing unnecessary virulence genes to that of individuals lacking them. Evidence for or against stabilizing selection has come almost exclusively from the first two types of studies. Since both of these involve inferences based on gene frequencies, we must understand the various phenomena that can affect the frequency of virulence genes. Factors which can influence gene frequencies are listed below. 1. Mutation pressure: Mutation may occur from avirulence to virulence or from virulence to avirulence. In the absence of other forces, the frequency of virulence is determined by relative mutation rates in both directions. The author knows of no plant pathogen for which Spontaneous mutation rates in both directions are known. 2. Selection pressure: Selection may occur for either virulence or avirulence. Selection for virulence on resistant hosts iS well documented (Johnson, 1961). Selection for avirulence (stabilizing selection) remains unproven. 3. Emmlgration pressure: The movement of genes into one population from another is called immigration or gene flow. AS many plant pathogens are easily disseminated from one place to the next, the frequency of virulence in a particular locality will be determined not only by the selection pressures in that area, but also by the frequency of immigration, and the gene frequencies in the area from which the immigrants come. 4. Founder effect: When a new population of parasities is initiated by a small fraction of some larger population, as often happens when a new disease is introduced into an area, that new p0pulation, due Simply to sampling error, often has different gene frequencies than the parent population. 5. Random genetic drift: Only a small random fraction of the propagules from one generation survive to the next. Since the p0pulation is finite, sampling error causes the frequencies of genes to fluctuate. They whll, in the absence of mutation, eventually become fixed at either 0% or 100%. The smaller the population, the more of an effect sampling error will have. 6. Meiotic drive: Meiotic drive refers to the production of abnormal gametic ratios due to meiotic abnormalities. Its importance for plant pathogens is unknown. 7. Linkage disequilibrium: Linkage disequilibrium is a condition in which genes at different loci are not associated randomly with one another. Imagine a population of organisms begun by two individuals (AB and ab). If the distance between locus a and locus b is greater than or equal to 50 map units and if there is no interaction between the loci (no epistasis), after one sexual cycle, the population will consist of approximately equal frequencies of AB, Ab, aB and ab individuals. The population will be in linkage equilibrium. However, if the loci are linked (less than 50 map units apart), or if epistasis favors the AB and ab arrangements, after sexual recombination the AB and ab types will be more frequent than Ab and a8, and the population will be in a State of linkage disequilibrium. The rate of approach to equilibrium is determined by the recombination frequency (map distance), the number of sexual or parasexual cycles per unit time, the amount of crossbreeding and the amount of epistasis. Epistasis can prevent linkage equilibrium entirely. Thorough discussions and mathematical models of these phenomena can be found in population genetics texts (Crow and Kimura, 1970; Spiess, 1977; Wilson and Bossert, 1971; Wright, 1969). Because of the complex nature of the evolutionary forces that affect virulence genes, as described above, it has been difficult to adequately test Vanderplank's hypothesis. Perhaps the strongest argument for stabilizing selection is based on the fact that selection of resistant varieties is possible. Vanderplank noted that mutations to virulence are common and yet the frequency of virulence can be initially low, as indicated by the ability of breeders to select for resistance. He believed that the factor keeping the frequency of virulence low is selection against unnecessary virulence genes. Even if his argument is correct, the reduction in fitness attributable to virulence genes need only be great enough to counter the net mutation rate to virulence. Thus, Stabilizing selection may be a very weak force of little practical value in the presence of selection for virulence, Such as, in the presence of hosts with genes for resistance 03 genes). Surveys of parasite populations have provided no definitive proof either for or against stabilizing selection (Crill, Jones and Burgis, 1973; Luig and Watson, 1970; Mac Key, 1973; Mac Key, 1976; Martens, et al., 1970; Roelfs and Groth, 1980; Watson, 1970; Wolfe, 1973; Wolfe and Schwarzbach, 1978). Earlier race surveys were reviewed by Vanderplank (1968). Conclusions from the surveys have been contradictory. This may be because, in many cases, virulence genes are in a condition of linkage disequilibrium with other genes in the genome. This problem may be exacerbated by rapid shifts in host genotypes and an infrequency or lack of sexual recombination in the pathogen. Roelfs and Groth (1980) demonstrated that the population of Puccinia graminis f. Sp. tritici in eastern and central U.S.A. was composed of genetically distinct clusters of races. The lack of sexual recombination in this population has apparently prevented linkage equilibrium. Even in populations in which linkage disequilibrium is presumably not so severe as in the asexual population of E. graminis, race surveys have been contradictory. Mac Key (1973) reported that Swedish races of Erysiphe graminis f. Sp. tritici with no or single genes for virulence were highly over-represented in the pathogen population, suggesting that unnecessary virulence genes were disadvantageous. 0n the other hand, Wolfe (1973) reported that the p0pulation of_E. graminis f. sp. tritici in England possesses a high frequency of unnecessary virulence genes. Evidence has accumulated (Jenkyn and Bainbridge, 1978) that E. graminis can survive from year to year in cool-temperate climates without undergoing a sexual cycle. To my knowledge, the actual extent of sexual recombination in these populations is unknown. There have been numerous controlled selection studies on susceptible hosts. Most of the forces acting on virulence genes can be effectively controlled by this method. The only factors that can cause Significant change in the frequency of virulence genes are selection against virulence and linkage of virulence genes to other genes that have an effect on fitness. Unfortunately, in most studies there was no control over linked genes. Two races of a pathogen were simply allowed to compete on a susceptible host variety for a number of asexual generations. Competition studies prior to 1968 were reviewed by Vanderplank (1968). Numerous experiments designed to test Vanderplank's hypothesis have been performed since that time (Brown and Sharp, 1970; Martens, 1973; Mortensen, 1974; Nelson and Scheifele, 1970; Ogle and Brown, 1970; Osoro and Green, 1976; Scheifele £3 21-: 1968; Scheifele and Nelson, 1970; Watson and Luig, 1968). Results of these studies are contradictory and inconclusive due to the influence of the entire pathogen genome on fitness. Mortensen (1974) compared two races of E. graminis f. Sp. tritici, one with the virulence genes to attack wheat lines with resistance genes £91, £92, Em3a, Em3b, Em4,‘Mle and mlr and the other with the corresponding avirulence alleles. When grown on susceptible wheat no evidence of reduced competitive ability was found in the virulent isolate. However, the real number of virulence genes in different races cannot be determined by testing them on a limited set of host lines (Wolfe and Schwarzbach, 1978), thus leaving conclusions about the influence of virulence genes on the rate of reproduction open to question. Some workers in plant pathology have assumed that, in a mixture of randomly collected isolates or in mixtures of progeny from a cross, any virulence gene is associated as frequently with any other genes influencing fitness as with their alternate alleles. Thus, on the average, those "other" genes have no effect on the observed behavior of virulence genes. Blanco and Nelson (1972) tried to eliminate the effects of linked genes by following gene frequency changes in experimental populations of Helminthosporium maydis consisting of mixtures of race 0 and race T isolates collected from the field. A tacit assumption was made that the population from which these isolates were collected was in linkage equilibrium and that the collections were representative of the population. They found that race T decreased in frequency on normal cytoplasm corn, suggesting selection against unnecessary virulence. In a 10 similar attempt to eliminate the effect of linked genes, Leonard (1969a) obtained a heterogenous population of Puccinia graminis f. Sp. avenge by making collections of the pathogen after sexual recombination on barberry. When a mixture of these isolates was grown on oats over several generations, the frequency of avirulence in the population increased, again suggesting selection against unnecessary virulence. Unfortunately, linked genes may be more important than previously assumed in determining the outcome of competitions among progeny. Hiura (1978) reported the results of competitions among progeny from a cross of Erysiphe graminis f. Sp. tritici (powdery mildew of wheat) and Erysiphe graminis f. Sp. agropzri (powdery mildew of wheatgrass). When mixtures of the progeny were grown on wheat, the frequency of isolates virulent on both wheatgrass and wheat declined, whereas the frequency of isolates virulent on only wheat increased. Following the logic used in other studies, this is evidence for stabilizing selection, since the virulence genes needed to grow on wheatgrass are unnecessary on wheat. However, at the same time, the frequency of isolates virulent on 5 to 6 wheat varieties increased from 10% of the population to over 90% of the population in 15 generations. The increase was at the expense of isolates virulent on fewer varieties. Since genes for adaptation to wheat came from the same parental culture as genes for virulence on Specific wheat varieties, some of these genes must be linked. Thus, a likely explanation for these results is that when selection occurred for adaptation on wheat, unnecessary genes for virulence on Specific wheat varieties were carried along, and unnecessary genes for virulence on wheatgrass were lost. The best way to test for stabilizing selection is to use isolates 11 that are identical except for the gene(s) being tested. Watson (1970) reported variable and inconclusive results with induced mutants of Puccinia graminis f. Sp. tritici. Watson and Singh (1952) studied the competitive ability of field-collected races of Puccinia graminis f. sp. tritici that they thought may have arisen from one another as mutations. They found that the races with wider host ranges were less fit. However, when Osoro and Green (1976) performed an experiment with the same organism, again using field-collected races that had presumably arisen from one another as mutations, they found no correlation between fitness and virulence. In practice, tests with presumed mutants should be confirmed by progeny tests, since mutations may have occurred at many loci. In the absence of isolates identical except for their virulence genes (true isolines), near-isolines (congenic lines) may be used. Two isolates that differ in their virulence genes are repeatedly backcrossed until lines are obtained that have a large fraction of their genome in common. This reduces but doesn't eliminate the probability that the isolates differ in genes (other than virulence genes) that affect fitness. Leonard (1977b) tried this approach with Bipolaris maydis (Helminthosporium maydis) to see if the ability to produce toxin reduced the fitness of the pathogen when growing on normal cytoplasm corn. The results were inconclusive because of a lack of replication and a lack of statistically significant evidence for differences in fitness in the 1 selection experiment. The labor involved in backcrossing pathogen cultures has limited the application of this method. 12 The Segregation Test The discussion of various approaches tried in the past to study stabilizing selection indicates that no method tried so far has provided an unequivocal answer to the question of whether unnecessary virulence genes reduce reproductive fitness. In this work, I have used the segre- gation test, a method pr0posed by Ellingboe (1976) to separate correlated phenomena. To my knowledge, this is the first time this method has been used to test Vanderplank's hypothesis. Two cultures were chosen that displayed differences in virulence and in fitness that were suggestive of stabilizing selection. A cross was made and the progeny were tested. If reduced fitness is controlled by a virulence gene, reduced fitness and virulence will segregate together. If the difference in fitness is not controlled by the virulence/avirulence locus, segregation will occur in some of the progeny. The disadvantage of the segregation test is that it requires increased effort in actually testing fitness. The relative time and effort involved in repeated testing of fitness versus repeated backcrossing of the pathogen must be taken into consideration when choosing the most appropriate approach for a particular organism. Biology of Powdery Mildew of Wheat The organism used in this study was Erysiphe graminis D. C. f. sp. tritici E. Marchal, which causes a disease commonly known as powdery mildew of wheat. E, graminis is an obligately parasitic ascomycete that occurs throughout the world, but is especially serious on wheat in cool-temperate regions. Because of its importance, it has received considerable study (Jenkyn and Bainbridge, 1978). An excellent collection of review articles on powdery mildews, including Erysiphe sp., is available (Spencer, 1978). 13 Mildew appears on the above ground parts of wheat plants as white colonies of mycelia. The mycelia are strictly superficial except for haustoria, which are absorbtive organs within the epidermal cells. Conidia are produced in profusion on upright conidiophores. The masses of conidia are easily air-borne and thus give the colonies a powdery characteristic. Unlike many fungal plant pathogens, neither high humidity nor free moisture are required for either sporulation or infection. Conidia released from sporulating colonies are passively carried by wind currents. Those deposited on susceptible tissue produce new sporulating colonies in as few as 4 to 5 days under favorable environmental conditions. Conidia of E. graminis are short-lived, lasting only a few hours to a few days at laboratory temperatures. Both the conidia and the colonies which produce them are haploid. E, graminis is heterothallic (Powers and Moseman, 1956, 1957). When opposite mating types are present on the same plant, crossing occurs and cleistothecia develop among the mycelia. Cleistothecia appear on the leaf surface as small black specks on a white mycelial mat. Moistening of the cleistothecia causes maturation of asci and subsequent release of ascospores, which, as the products of meiosis, have new combinations of genes and are believed to be the main source of genetic variability. Ascospores infect and form colonies in the same manner as conidia. Methods of storing, propagating, and crossing powdery mildew are available (Moseman, 1959). In addition, congenic wheat lines are available to determine the virulence Spectra of the different mildew cultures (Briggle, 1969). MATERIALS AND METHODS Seed Stocks Seeds of the wheat lines used in this study were provided by R. A. Kilpatrick, Beltsville Agricultural Research Center, Beltsville, Maryland. The near-isogenic powdery mildew differential host lines (Table 1) were developed by L. W. Briggle by backcrossing powdery mildew resistance genes (Em_genes) from various wheat varieties into the susceptible cultivar Chancellor (Briggle, 1969; Moseman, 1973). Seed lots of Chancellor, and lines CI 14118, CI 14120 and CI 14123 with powdery mildew resistance genes £92, Empa and Em4, respectively, were tested for purity at the initiation of the project. Because selection studies were to be done on Chancellor, it was important to have no seed with Em genes in the Chancellor lot. Therefore, the seed was ,tested for purity repeatedly during the course of the work. Tests for purity (absence of off-type seed) were done by inoculating seven-day-old seedlings with the isolates MS-l and Mo-10 of E, graminis f. Sp. tritici and observing infection type 7 days later. Out of 976 Chancellor seedlings inoculated with MS-1 and 245 inoculated with Mo-10, no resistant plants were detected. Infrequent off-type seed were found in the other wheat lines. Powdery Mildew Isolates: Source and Maintenance Two isolates of E. graminis f. Sp. tritici (MS-1 and Mo-10) and their progeny were used in this study. MS-1 and Mo-10 were isolated from 14 15 Table 1. Wheat lines used to classify the virulence phenotypes of isolates of E. graminis f. sp. tritici. Near-isoline Resistance or Cultivara Gene Designation Parentageb CI 12333 - Chancellor CI 14114 Pml Axminster/8*Cc CI 14115 P_ml CI 13836/8*Cc CI 14116 Pml AsII/8*Cc CI 14117 PHD. Norka/8*Cc CI 14118 P52 Ulka/8*Cc CI 14119 2E2 CI 12632/8*Cc CI 14120 Pm3a Asosan/8*Cc CI 14121 Efi3b Chul/8*Cc CI 14123 Pm4 Khapli/8*Cc CI 14124 EEA Yuma/8*Cc CI 14189 - Transec/8*Cc CI 15520 - Khapli/8*Cc CI 15886 - Triticale/8*Cc CI 15887 - Shearo/8*Cc CI 15888 - Michigan Amber/8*Cc CI 15889 - CI 4546/8*Cc CI 17739 - CI 7742/8*Cc CI 17760 - CI 3008/8*Cc PI 367698 - Kavhaz aCereal Investigations accession no. (CI) or Plant Introduction no. (PI). bThe symbol “/8*Cc“ indicates that the wheat line named was backcrossed to Chancellor eight times. 16 Michigan wheat fields and kept in separate controlled environment chambers. MS-l was maintained on the susceptible wheat cultivar Little Club. Mo-10 was maintained on line CI 14120. The environmental conditions for culture maintenance have been published elsewhere (Nair and Ellingboe, 1965). In addition to maintenance in growth chambers, cultures were stored in the laboratory in separate isolation chambers made of glass lamp chimneys. ApprOpriate genotypes of wheat were planted in soil in 237 ml capacity wax paper cups (hereafter indicated simply as "cups“). Glass lamp chimneys (Corning No. 845310) were placed on the soil to cover the seedlings before they emerged. The taps of the chimneys were covered with a filter consisting of either a thin layer of cotton batting between single layers of cheesecloth or a double layer of tissue (Kimwipes or Kleenex). The filters allowed evaporation from the isolation chambers yet prevented movement of mildew conidia into or out of the chambers. The cups were placed on shelves in the laboratory under Gro-lux lights (Sylvania F40-GRO-WS) (about 1 x 104 ergs sec'1 cm‘2 at soil level). The temperature within the lamp chimneys was 2411°C during the light cycle (14 hr) and 21:1°C during the dark cycle (10 hr). These light and temperature conditions are indicated when the text refers to "growth under lights in the laboratory". Seven days after planting, the chimneys were removed, the seedlings inoculated with conidia and the plants again covered with the chimneys. Mildew colonies appeared 4 to 5 days later. Cultures not in frequent use were stored for periods of 4 to 6 weeks in a refrigerator. About 6 cm3 of vermiculite were placed in the bottom of glass tubes that were 30.5 cm tall and 2.5 cm in diameter. Three seeds of an appropriate wheat line were placed on top of the 17 vermiculite. The tubes were plugged with a foam rubber stopper and watered from below through a 0.5 cm hole in the bottom of the tube. After growth for seven days under lights in the laboratory, the plants were inoculated with isolates of E. graminis f. sp. tritici. After an additional 1 to 5 days, the tubes were transferred to a refrigerator at 612°C. Continuous lighting was provided by a 15 watt cool white fluorescent bulb. Mildew cultures were transferred as necessary before death of the host plant and thus death of the culture. Transfers (inoculations) were made by dusting conidia from one plant to another or by rubbing the leaf surface of a new plant with a leaf bearing sporulating colonies. Because mildew conidia are produced in profusion and are air-borne, transfers were carried out in a positive pressure transfer hood. At times when the laboratory was free of mildew and it was desirable to keep mildew from escaping into the laboratory area, transfers were made in a negative pressure fume hood. Uninoculated control plants were used to monitor for contamination. All mildew cultures were purified immediately prior to use in exper- iments by isolation of Single colonies. Conidia were lightly dusted onto 7-day-old Chancellor wheat seedlings. After 5 days, when tiny colonies began to appear, several pieces of leaf (0.5 to 1 cm), each bearing a single colony, were excised. Each piece was placed in a small petri dish and floated on distilled water for 3 days. Preliminary experimentation Showed that the probability of transferring possible contaminants (mildew conidia from other sources that might land on the leaf surface during the cutting process) was minimal 3 days after cutting. This observation is in keeping with the short life span of conidia under laboratory 18 conditions and the observation that the minimum time required for infection and initiation of sporulation at 21°C is about 4 days. After 3 days in isolation, each colony was classified by infection type, as described below. Preliminary testing demonstrated that virtually all cultures obtained by this method were genotypically pure. Those colonies with the phenotype of the original culture were selected and used for further work. Off-type colonies were rarely found. Powdery Mildew Isolates: Classification by Infection Types Genotypes were assigned to isolates based on their infection type (phenotype) on wheat lines with different Em genes. The infection types were determined in both laboratory and greenhouse tests. Three seeds of each wheat line were planted in soil in cups in the laboratory. Up to 6 lines were planted in a circle with Chancellor in the center.‘ The cup was covered with a chimney. After 6 days growth under lights in the laboratory, the Chancellor seedling was inoculated with the culture to be tested. After 7 more days, when the mildew was sporulating heavily, the cup and chimney were Shaken to distribute conidia to the surrounding wheat lines. 0n the fourteenth day after the first inoculation the infection type on each of the wheat lines was compared to that on Chancellor. Some of the wheat lines were known to have low levels of resistance that may not be expressed in the laboratory where conditions are favor- able to mildew and unfavorable to wheat. However, according to J. G. Moseman (personal communication), resistance can be detected on 7-day-old seedlings grown and inoculated in the greenhouse. Therefore, infection tests were repeated in the greenhouse. Three seeds of each wheat line 19 were planted in soil in cups in the greenhouse (21t4°C). Three lines were planted per cup. After 7 days, the wheat seedlings were placed in inoculation chambers and dusted with conidia. After 12 hours the cups were removed. After an additional 7 days, infection types were recorded -and compared to the infection type on Chancellor. Noninoculated control plants remained mildew free. Because of the subtle nature of some of the differences in infection type, the wheat lines were assigned random numbers and the experiment was repeated. Infection types were recorded without knowledge of the isolate or wheat line. Identical results were obtained. Leijerstam (1972) suggested the gene symbols V and A for virulence and avirulence, respectively. In this study I have chosen to use the symbolism suggested by Ellingboe (1976). Isolates able to infect a particular host line were considered virulent on that line and were assigned the gene symbol ERA. Isolates with a reduced ability to infect a host line were considered avirulent on that line and were given the gene symbol "3}. The choice of a capital "Bf to symbolize avirulence iS arbitrary and does not imply dominance. The dominance characteristics of genes in Erysiphe graminis are not known. Powdery Mildew Isolates: Classification bnyating Type Mating types were assigned to isolates on the basis of their ability to cross with isolates MS-l or Mo-10. Three seeds of Chancellor wheat were planted in soil in cups and a chimney was placed on top of each cup. After 7 days growth under lights in the laboratory, the seedlings were inoculated with the two cultures to be tested. Sporulating vegetative colonies appeared in 5 days. If the two isolates were of opposite mating 20 type, dense mats of mycelia began to appear 10 days after inoculation and cleistothecia developed about 4 days later. Four weeks after inoculation the presence or absence of cleistothecia was recorded. Mo-10 was arbitrarily designated mating type “-" and MS-l mating "+". The designations are arbitrary with respect to any mating types established by other workers, as Standards for comparison were not available. Cross of MS-l and Mo-10 Genotypically pure cultures of MS-l and Mo-10 derived from Single colonies were crossed in quantity to permit collection and analysis of their progeny. Three seeds of Chancellor were planted in soil in each of eight 25 cm diameter clay pots. The seeded pots were held in a- controlled environment chamber at 2011°C during the day (14 hr) and night (10 hr). Light was provided by a combination of fluorescent (2.5 x 104 ergs sec‘1 cm’z) and incandescent lights (1.5 x 104 ergs sec'1 cm'z). The chamber was isolated from possible sources of mildew to prevent con- tamination. After Six weeks, the plants were dusted with condia of MS-l and Mo-10. After the first appearance of cleistothecia at two weeks, the day temperature was raised by 0.5°C per day to a final day temperature of 26°C. The fungus does not grow well at high temperatures; therefore, this step was included to prevent premature death of the plants due to excessive fungal growth. However, the procedure was not necessary for the production of cleistothecia capable of releasing viable ascospores (Moseman, 1959; Bronson, unpublished). Six weeks after inoculation, the leaves bearing cleistothecia were harvested, air-dried, and stored in a paper bag at 21:1°C. The freshly harvested cleistothecia had immature, undifferentiated 21 asci. Ascospore formation and release were induced by a modification of methods suggested by J. G. Moseman (personal correspondence) and Moseman and Powers (1957). Isolated cleistothecia, or dried mycelial mats bearing cleistothecia, were removed fron the leaf surfaces and placed on filter paper moistened with distilled water. The filter paper was placed on and adhered to the inside of the top of a 6 cm diameter Petri dish. Pieces of leaf from 7-day-old Chancellor plants were floated in a small amount of benzimidazole solution (20 ppm) in the bottom of the dish. Benzimidazole acts as a cytokinin to retard the senescence of detached leaves. Preliminary experimentation Showed that benzimidazole reduced colony development Slightly at 20 ppm and this effect was more pronounced at higher concentrations. The dishes were held in a plastic bag either in the dark or in diffuse room light at 2111°C. Care was taken to avoid bright lights which resulted in heating and condensation on the leaf surfaces, both of which inhibit fungus development. Every two days the top lid of the Petri dish (bearing the cleistothecia) was placed on a new bottom dish supplied with fresh leaf pieces. The exposed leaves were covered with a lid and placed under lights in the laboratory to allow the ascospores that had fallen on the leaves to infect and form colonies. Ascospores were released from the cleistothecia 3 to 9 days after moistening, with peak release between day 5 and 7. This coincided with the time of ascospore maturation as observed microscopically. No mildew developed on leaves placed under the cleistothecia during the first two days, indicating no contamination from conidia clinging to the cleistothecia or mycelial mats. Tiny mildew colonies appeared on the leaves 5 days later. Leaf pieces (0.5 to 1 cm), bearing what appeared to be single colonies, were 22 excised. Preliminary testing showed that these colonies were frequently mixtures of two or more genotypes, perhaps indicating that the ascospores fell in clumps onto the leaf surface. For this reason purification was required. After a first cycle of purification by single colony isolation no mixtures were found. However, a second single colony isolation was performed as added insurance against mixtures. The purified cultures were assigned an isolate number, tested for infection type on the differential lines and Stored for later use. Determination of the Relative Fitnesses of Mildew Isolates The relative reproductive fitnesses of the various isolates were estimated by their increase or decrease in frequency in a mixed population over a number of generations. Fitness was expressed quantitatively by means of a relative fitness term (W) and a selection coefficient (s=1-W). Relative fitness was defined as the number of successful offspring produced per parent colony per generation for a particular isolate divided by the number of successful offspring produced per parent colony per generation for an isolate used as a standard. A successful offspring was defined as one which survives to reproduce. The selection coefficient for a particular isolate is a measure of the relative inability of the isolate to produce successful offspring. A method for defining and estimating these terms for two cultures was provided by Leonard (1969a). The method was extended (Appendix A) to include any number of isolates. All tests were performed in triplicate in three controlled environ- ment chambers (Sherer-Gillett models CEL 37-14 and CEL 25-7HL). Air flow into and out of the chambers was reduced as much as possible to reduce the likelihood of contamination. External air vents were plugged and the 23 opening of the chamber covered with a clear plastic Shield. A tube inserted in the shield was used to water the plants. A passageway in the Shield could be opened when needed. Light was supplied by a combination of fluorescent bulbs (2.5 x 104 ergs sec"1 cm‘z) and incandescent bulbs (1.5 x 104 ergs sec‘1 cm'z). Temperature within the chambers was 2011°C during both the day (14 hrs) and night (10 hrs). Temperature and rela- tive humidity were monitored with hygrothermographs that were calibrated weekly with a sling psychrometer. Relative humidity generally fluctuated between 55% and 90% in the chamber used for experiments 1, 4, 7 and 10; between 40% and 95% for experiments 2, 5, 8 and 11; and between 40% and 90% for experiments 3, 6, 9 and 12. Cultures to be tested for fitness were purified and increased. Chancellor seeds were planted in soil (25 to 30 per cup) and 10 cups were placed in each chamber. When the plants were 6 days old, approximately equal frequencies of all cultures to be tested at one time were dusted . onto the plants. In experiments 1, 4, 5 and 6, freshly harvested conidia were weighed out in equal amounts. In experiments 2, 3 and 7 through 12, leaves bearing approximately equal numbers of sporulating colonies were used as sources of inoculum. Neither method was particularly effective in insuring that all cultures appeared in equal frequencies in the first generation. Old plants were removed from the chambers and new plants added in a sequence that resulted in the transfer of mildew from one set of plants to the next by the wind currents in the chamber and thus from one asexual generation to the next (Figure 1). This system resulted in discrete generations of mildew. The population size (number of colonies in each generation) was generally over 2,500 in the initial generation and over 10,000 in succeeding generations. 24 .mcwu=e_a mu=_m mace ea Lucas: mew mmuuu_v=_ won use co cones: ugh ._u_u_gu .nm .u m_:psoem .m mo mme=u_=u a we mammw:p_» m>_uupme use cow mama» :_ mucm>m mo muemaomm .H me=m_u 25 The frequency of each isolate in each chamber was estimated at the initiation of the experiments and at intervals of 14 or 28 days thereafter. Preliminary trials with various methods of estimating isolate frequency failed to detect a method that was both simple and reliable for this system. Therefore, the following method was employed. Mildewed plants were temporarily removed from the chambers and used to lightly dust pots of 7-day-old Chancellor seedlings in a settling chamber. These plants were covered with a chimney and placed under lights in the laboratory. After 5 days, leaf sections with tiny, individual colonies were excised, then floated on distilled water for 3 more days, and tested for infection type in the laboratory on host lines CI 14118 (EmZ), CI 14120 (Em3a), CI 14123 (BEA) as well as Chancellor (CI 12333) as described earlier. The frequency of each genotype in the sample was calculated. Approximately 100 colonies were identified per sampling time per chamber. In Spite of precautions to prevent contamination, air-borne conidia could have entered the chambers from unknown sources. If these conidia represented races that were relatively unfit, they would probably not be detected in samples or influence results. However, if a very fit isolate had established itself in a chamber early in the experiment, it might increase to the extent to which it would appear in samples and be mistaken for one of the cultures being tested. For this reason, samples of all isolates still detectable in the chambers (3 samples per genotype, if available) were tested in the laboratory for infection type on all differential wheat lines. No off-type cultures were detected. Data were plotted as ln (g) versus generation where q is the frequency of a less fit isolate and p is the frequency of a more fit 26 isolate. Assuming the relative fitness of an isolate does not vary over succeeding generations, data plotted in this manner will yield a straight line except for deviations due to sampling error. Least squares linear regression was used to estimate the slope (ln W) and thus the relative fitness (W) and selection coefficient (s=1-W). An example is provided in Appendix A. RESULTS The objective of this research was to determine whether or not certain unnecessary virulence genes (9 genes) have a debilitating effect on Erysiphe graminis f. Sp. tritici. This was accomplished by determining whether or not a culture with 2 genes was less fit than a culture with with corresponding E ggngg (avirulence alleles) and by testing to see whether or not the reduced fitness and 2 genes segregated together. Parental Isolates: Genotype Assignments The two isolates used in this study, MS-l and Mo-10, were chosen because they were known to differ in host range and mating type (A. H. Ellingboe and J. L. Clayton, personal communication). Opposite mating types were required to permit the crossing of the cultures. Genotypes were assigned to the cultures based on their infection types on wheat lines with different Em genes. Infection types are shown in Table 2. MS-1 and Mo-10 gave Similar infection types on all lines except those with genes 2mg, Em3a, and Em4, and on CI 15889. MS-I displayed reduced virulence on CI 15889 that was detectable in the greenhouse, but not in laboratory tests. Based on this information, it was concluded that were were at least 5 gene differences between MS-1 and Mo-10 and the following presumptive genotypes were assigned. MS-I 32 33a _P_4 £15889 flit.“ Mo-10 £2 p_3a 34 315889 mai- 27 Table 2. Infection typesa of E. differential wheat TTnes 28 seedlings in the greenhouse. graminis f. sp. tritici isolates on days after inoculation onto wheat Infection Types C.I. or P.I. Resistance number Gene MS-l Mo-10 Progeny 12333 - 4 4 4 14114 Pml 0,1 0,1 0,1 14115 Em1 0,1 0,1 0,1 14116 Emu 0,1 0,1 0,1 14117 EmI 0,1 0,1 0,1 14118 BE? 2 4 segregating 14119 Pm2 2 4 segregating 14120 Pm3a 0,1 4 segregating 14121 'EEBb 0,1 0,1 , 14123 Pm4 0,1b 3-c segregating 14124 Efih 0,1b 3"c segregating 14189 - 4 4 4 15520 - 3cmd 3chld 3cmd 15886 - 4 4 4 15887 - 0,1 0,1 0,1 15888 - 4 4 4 15889 - 3d 4 segregating 17739 - 333d 3-,3d 3-,3 17760 - 3-,3d 3-,3d 3',3d 367698 - 0,1,3=d 0,1,3=d 0,1,3“d aSymbolism used to describe infection types: 0 no symptoms 1 small, light-green flecks with no sporulation 2 distinct necrotic and chlorotic spots with no or scant sporulation 3= few colonies with scant sporulation 3' scant sporulation 3 reduced sporulation 3chl reduced sporulation with a definite chlorosis 4' slightly reduced sporulation 4 heavy sporulation, no chlorosis. bOccasionally appeared as 3’ when tested in the laboratory. cAlways appeared as a 4' when tested in the laboratory. dConmonly appeared as 4 when tested in the laboratory. 29 Parental Isolates: Determination of Relative Fitness The relative fitnesses of MS-I and Mo-10 were determined by mixing the cultures and observing changes in their relative proportions over several generations on the susceptible cultivar Chancellor. By this method, differences in absolute fitness (the number of progeny colonies produced per parent colony per generation) occurring as a result of reduced growth or survival at any stage were detected and expressed as the relative fitness of the less fit isolate compared to the more fit isolate. A preliminary experiment (described in detail in Appendix B) revealed no difference in fitness between isolates MS-l and Mo-10. However, when the experiment was repeated using slightly different environmental conditions, as described in the Materials and Methods section, a clear difference in fitness was detected. Results of three tests for fitness using the selective conditions are shown in Figure 2 and Table 3. Calculation of the relative fitnesses and the selection coefficients indicate that Mo-10 is 19 to 32 percent less fit than MS-I under these conditions. That is, it produces 19 to 32 percent fewer successful offspring per generation than does MS-l. This result is in keeping with Vanderplank's hypothesis that pathogens with unnecessary virulence genes are less fit than pathogens without those genes. Progeny: Genotype Assignments and Genetic Analysis of Virulence Progeny of MS-1 and Mo-10 were obtained to determine if reduced fitness segregated with virulence, and to verify genotype assignments. Progeny were tested in the laboratory for infection type on CI 14118 (Egg), CI 14120 (Em3a), CI 14123 (394) and CI 12333 (Chancellor) and presumptive genotypes were assigned (Table 4). 30 3823.35 me: name: «33333 2» .5 2.35883 .mmecu @325 3 5.9.1.223 76.9.3 ammo—0V 3.3.. new $9.33. :83 Tm: 3:39. 3.5.2» ...... .2. 25.5.5 .u .8 335:3: :23;on or Np Q Q o 9‘. o . o\ // 1 O .. ‘ l o a o I o a a a a r u N ~I s ~ ~ I m not \ l l V. o ‘9 o “2 o m acoetmaxm oé 9‘. a V. o ‘9 c Q . O OJ cotoeocow or up a v c .J . . u u . q 0110/ or .. II I. I a I: x J I! m .. m. , ... x. .A ~ at. in. . 0.110. l N acme—emaxm Aauenbug .N p.53... gotcha—00 N— a v o u - u u - - ‘l‘ll‘z ss”; I ’1 xx x (4 / [Nd ,,. . .. . 4. .... to m 1 b A n .... m. .. .. md b .. a ss h a $6 \\\ III us DIIID\\Q IG~ 0.9 a u=o3_cmaxm 31 Table 3. Relative fitnesses and selection coefficients for E, graminis f. Sp. tritici isolates MS-l and Mo-10. Relative fitnessa Selection coefficienta Experiment (W) (S) 1. MS-l 1.00 0.00 Mo-lO .80 0.20 20 MS-l 1000 0000 MO-lO .81 0.19 30 MS']. 1000 0000 MO-IO .68 0.32 aData and method for calculation of relative fitness terms and selection coefficients are provided in Appendices A and C. 32 Table 4. Progeny of E. graminis f. Sp. tritici isolates MS-l and Mo-IO: infection types as 0 served under laboratory conditions and genotype assignments. Infection Type on Host Plant With Number Class .EEZT EmSa Emfl_ Genotype Observeda Percent 1 4 4 4‘ p2 p3a.p4 21 13.0 2 2 4 4- 32 p_3a p_4 20 12.3 3 4 o 4- p_2 _P_3a £4 25 15.4 4 4 4 o 22 23a 34 28 17.3 5 2 o 4- 32 _P_3a p_4 18 11.1 6 4 o o p_2 33a 34 18 11.1 7 2 4 o 32 33a 34 22 13.6 8 2 o 0 £2 E3a g4 10 6.2 TOTAL 162 aRatio of number of offspring observed in each class is not significantly different from 1:1:1:1:1:1:1:1 (Chi-square = 9.95). 33 Twenty-four of these progeny were selected for use in tests for relative fitness. They consisted of three each of the eight classes found in the laboratory tests. They were tested further for infection type in the greenhouse on all wheat lines and also tested for mating type. Infection types are Shown in Tables 2 and 5. Segregation was observed in the progeny only on wheat lines on which the parental cultures differed. Note also that the progeny had the phenotype of either one or the other parents. No intermediate types were discernible. Results of mating tests are shown in Table 6. The progeny were either of two mating types. No progeny were observed to mate with both or neither of the parents. Based on these observations, genotype assignments were made (Table 7). i The assignment of genotypes was based on the assumption that each of the differences in host range between MS-l and Mo-10 was conditioned by a single, distinct gene, that is, that mildew has a gene-for gene relaionship with wheat (Flor, 1955, 1956, 1971). To verify this assumption, it is necessary to Show that only one gene is responsible for each difference in phenotype by observing segregation ratios. It is also necessary to Show that the genes are distinct by Showing that segregation occurs between them. Statistical analyses of the frequencies of the various classes of offspring were made. Ratios of avirulent to virulent offspring on wheat lines with resistance genes Em2, EmBa or Em4 did not differ significantly from 1:1. This suggests Single locus control over each of these phenotypes (Table 8). Hypotheses that two loci control virulence, as indicated by either a 1:3 ratio if virulence is dominant or a 3:1 ratio if virulence is recessive, were not supported by the data. Since virulence on CI 15889 and mating type were determined on non-random 34 Table 5. Infection types of isolates MS-l and Mo-10 of E. graminis f. Sp. tritici and twenty-four progeny fron a cross 0 - and Mo-lD on differential wheat lines on which segregation was observed (after 7 days under greenhouse conditions). CI 14118 CI 14119 CI 14120 CI 14123a CI 14124a CI 15889b Pm2 Pm2 Pm3a Pm4 Pm4 Parents MS-l 2 2 0,1 0,1 0,1 3 MS-10 4 4 4 3' 3' 4 Progeny #52 4 4 4 3' 3' 4 #59 2 2 4 3' 3‘ 4 #70 4 4 0,1 3’ 3 4 #65 4 4 4 0,1 0,1 4 #92 2 2 0,1 3' 3' 3 #51 4 4 0,1 0,1 0,1 4 #78 2 2 4 0,1 0,1 3 #89 2 2 0,1 0,1 0,1 4 #116 4 4 4 3‘ 3' 3 #105 2 2 4 3‘ 3‘ 3 #117 4 4 0,1 3‘ 3‘ 3 #121 4 4 4 0,1 0,1 3 #102 2 2 0,1 3' 3' 4 #124 4 4 0,1 0,1 0,1 4 #125 2 2 4 0,1 0,1 3 #138 2 2 0,1 0,1 0,1 3 #146 4 4 4 3' 3' 4 #150 2 2 4 3' 3' 3 #106 4 4 0,1 3' 3‘ 3 #141 4 4 0,1 0,1 3 #143 2 2 0,1 3' 3' 3 #127 4 4 0,1 0,1 0,1 4 #140 2 2 4 0,1 0,1 4 #149 2 2 0,1 0,1 0,1 3 33' reaction appeared as 4‘ under laboratory conditions. b3 reaction appeared as 4 under laboratory conditions. 35 Table 6. Mating types of E. raminis f. Sp. tritici isolates as deter- mined by the presence (yes) or absence (no) of cleistothecia 4 weeks after inoculation. Isolate MS-l Mo-10 Mating Typea Parents MS-l no yes + MS-10 yes no - Progeny #52 yes no - #59 no yes + #70 yes no - #65 yes no - #92 yes no - #51 yes no - #78 no yes + #89 no yes + #116 yes no - #105 no yes + #117 yes no - #121 yes no - #102 no yes + #124 yes no - #125 no yes + #138 no yes + #146 yes no - #150 yes no - #106 no yes + #141 yes no - #143 no yes + #127 yes no - #140 yes no - #149 no yes + aMS-l was arbitrarily designated mating type +. All other isolates were assigned mating types with respect to MS-l. 36 Table 7. Proposed genotypes of E. graminis f. sp. tritici isolates MS-l, Mo-10 and their progeny. Isol ate Genotype Parents MS-l E2 E3a E4 E15889 3191+ MS-10 32 33a 34 215889 mat: Progeny #52 32 33a 24 315889 ma_t_- #59 E2 3a 34 315889 MT #70 22 _3a 4 £15889 E- #65 2 3a _4 15889 mai- #92 _2 3a 4 _15889 M- #51 2 E3a 4 15889 E}? #78 _2 3a E4 _15889 ma_t+ #89 E2 _3a E4 215889 flail” #116 32 23a 24 E1 5889 ma_t- #105 E2 3a 24 P15889 £85.” #11 7 p_2 _3a 4 E15889 M- #121 $2 3a _4 E1 5889 9.3.1? #102 _2 3a 4 315889 El?“ #124 2 E3a 4 15889 mg;- #125 2 3a E4 15889 IE1?" #138 E2 _3a E4 E1 5889 3132+ #146 32 23a 24 315889 93E- #150 E2 3a 24 P 15889 @- #106 p_2 _3a 24 E15889 FEET #141 2 3a E4 E15889 m_aE- #143 _2 3a 4 E15889 113.15.“ #127 2 E3a 4 215889 mg;- #140 _2 3a E4 £15889 mat- #149 P2 _3a P4 E15889 E} 37 Table 8. Segregation of alleles for virulence and mating type in the progeny of E. graminis f. Sp. tritici isolates MS-l and Mo-lO. Hypothesized Phenotype Ratioa Observed Chi-square E,p_ avirulent:virulent 0n host lines 1:1 70:92 2.72 n.s. with E92 1:3 27.69 ** 0n host lines 1:1 71:91 2.23 n.s. with Em3a 1:3 29.63 ** 0n host lines 1:1 78:84 0.15 n.s. with Em4 1:3 45.07 ** 0n c1 15889b 1:1 13:11 0.04 n.s. 3:1 4.50 * .mggf:mgg- mat+:mat- mating typeb 1:1 10 14 0.38 n.s. 1:3 2.72 n.s. 3? 1 ratio suggests Single locus control of the phenotype. A 3:1 or 1: :3 ratio suggests that two independent loci control the phenotype. bConclusions about the number of loci controlling virulence on CI 15889 and controlling mating type may not be strictly valid since progeny analyzed for these traits were not selected at random with respect to other virulence genes. *significant at P50.05 **significant at P50.01 38 progeny, the number of loci controlling these characters cannot be determined with certainty. However, the simplist interpretation of the data is single gene control over each character. No evidence was found to reject the hypothesis that the loci controlling reaction to Em2, Egpa and En4 are unlinked (Table 9). Evidence was f0und for linkage between the mating type locus and the locus for reaction t°.EEZ- These two loci may be linked to the locus controlling reaction to CI 15889, but the sample Size was too small to provide proof. The evidence presented in Tables 8 and 9 suggests that the assigned genotypes were valid in that the differences in phenotype between the parents are controlled by single, separate loci. Progeny: Determination of Relative Fitness and Its Relationship to Virulence Genes The twenty-four selected progeny were used to determine whether the differences in fitness between MS-l and Mo-10 segregated with any one or with a combination of.p genes. Eight progeny, representing one each of the eight possible combinations of genes for virulence on £929 Empa and Emfl, were analyzed at one time in each of three growth chambers. The progeny were selected at random with respect to virulence on CI 15889 and to mating type because, at the time of selection, these characteristics were not known. The procedure used to measure fitness was the same as that used for the parental cultures, except that eight isolates, rather than two, were present in each chamber at one time. The experiments were allowed to run for 16 generations. Samples were taken every two to four generations. The fitness of each isolate was calculated relative to the most fit isolate in each set of eight. 39 Table 9. Linkage analysis of the progeny of E. graminis f. sp. tritici isolates MS-l and Mo-10. """‘ Map Units Percent (95% Confidence Hypothesis Recombinants Interval)a Chi-square Loci E2 and E3a unlinked 52.5 44.7 - 61.1 5.85 ns Loci E3a and E4 unlinked 57.4 49.7 - 66.0 6.25 ns Loci E2 and E4 unlinked 51.9 43.8 - 60.2 3.44 ns Loci E2 and E15889 unlinked 37.5 19.2 - 57.1 1.67 ns Loci E3a and E15889 unlinked 54.2 32.2 - 74.6 0.33 ns Loci E4 and E15889 unlinked 54.2 32.2 - 74.6 0.33 ns Loci E2 and mg; unlinked 16.7 4.8 - 36.4 11.33* Loci Efia and ng unlinked 41.7 22.7 - 60.7 1.33 ns Loci E4 and ng unlinked 50.0 29.0 - 71.6 .67 ns Loci-E15889 and ESE unlinked 37.5 19.2 - 57.1 2.33 ns aBinomial confidence interval. *significant at P15 0.05. 40 The results of relative fitness tests on the progeny are shown in Figures 3 through 11. Selection coefficients are presented in Table 10. Inspection of the graphs reveals that the isolates were not equally frequent in generation 0. The initial frequency of an isolate does not affect its calculated fitness or selection coefficient, since these are determined by relative, not absolute, changes in frequency. However, selection coefficients for the less frequent isolates were probably not estimated as well aS for the conmon isolates because of limitations in the sampling method. Note that the initial frequency of the isolates did not determine their behavior. Both rare and comnon types increased in frequency, though fit types tended to appear more comonl y in the initial generation, perhaps due to selection for them prior to initiation of the experiment. Casual inspection might indicate that selection occurred more slowly among the progeny than between the parents. This impression is an artifact of the lower frequencies of each isolate in the chamber. In fact, comparison of selection coefficients shows that fitness differences in some cases were about the same for the progeny as for the parental cultures. Within each set of eight progeny, there was fair reproducibility in the results fron chamber to chamber. This indicates that there were true differences in fitness between the isolates and that random drift was not responsible for the major increases or decreases in isolate frequency. Because of the large population size (generally in excess of 10,000 colonies per chamber per generation), genetic drift was not expected to have a major effect. } In a mixture of two isolates with constant fitness, theoretical 41 co..o.o:oo o. «p o v / IIJ a 6%. NM 2: .323230 a. up 0 Q l l l l l l o d N d ‘1 0* Aauonboag w Q6 0‘ 1. O) o 6 4311041603; .6 .Am ucwe_cmaxmv :o mco_uccmcmm pawu_:oo :mmux_m mc_e=v _uwu_ep .um £0230600 Op «p o C .e mp:PEacm an/uL/HLLI- a and Nu m: 30:32.09 up up a v cm 6% Nu o: 5:33.30 o o. a. o v o «6 \./.\ «.0 \/.\. to m. o\o .3 m. b o\ b . odm can ... em 6%. Na 8 9. yaw Q. 80:05.800 0 Op N. O C O .6 N6 .0 u. ‘6 u. m m o u Odfm 05f” ad ad O' p Di .3 o mocmcw.o=_p poms: m_n_uamum:m ecu .u mo mmuepom_ mo mm_ucmacmcm .323 2.00 0— «p 0 v 0 qua-.- O/O‘.\.|I.\ No tom. .5 a new am 8% NM ... Nm* oa 30:20:00 0. «p o v 0 Aauonbug .m we=m_m .Rm acme_emaxmv co muemgw.o=_p wade: w_a_uamum:m we» co m:o_paem=aa pewu_cou cumux_m ucpcsu _u_p_eu .am .4 mp:_5ee .m we mmumpomp co mm_u=m=cmeu 30:22.00 :02! 2.00 30:22.00 .30...” 2.06 o. «w 0 Q o o. u C o «d o o «d /O\ /O/ \ l . .l .8 0110 I '.0 u to u u 0 b b o 1 b n n u 06 u m 1 0.0 w b A .A Q am am Nu ... a and NM ... em 8m Nu ... am 6mm .....m ... a: a: E. . N2 .... 0.. a. o.- co..o.oc.0 , 80:30:00 30:32.00 £3.3on0 3 «p o v o o. «u o v o 3 a. a v a up a v 53000.03; Aauonbug Annabel; Aauonbug .¢ me=m_m 43 .nm nemepequmv co meowuacmcmm pm_uwcou camax_m m:_c:u _u_p_eu .am 80.3.0.36 0 '. 6 9 430001103; am an“ E a: cozozcoo 6. «p 6 Q 6 '. o huonbug o d 62:32.06 up 0 v 6 «.6 10m. u Gum a and NM ... mum 6.. 30:20:06 up 6 v 6 u 99» ‘ am can Nu .... oh* 6.- .» chPEmL 6p co__mu:a;m.m=_p Home: m—n_uqmum=m we» so..o.o..06 co..o::o6 a 2. a. o v o ull-JIl-ll-Ijl-J 6.116 of. 1 N6 6’.\ /\ lfifl.° 19% .Qow m . m 0.0 “ ... ’.O b a emu Na 8 am 6% NM ... 5 3 N: 6.. 80:22.06 80:30:06 5. 6 C 6 6’ Na 6 Q 6 .m co moum_omw mo mo_u:o:cmem ///, 3 8 3 3 a 6mm Na N: O '3 Aauonbug .m «suave 44 .wn acmewemaxmv cop—muemzw.m:_— acme: m~n_uqmum=m we» so m=o_paem:mm —a_u_:ou :mmax_m m:_c=u wo_u_eu .am .e mw=_5mem .m we mmaapom_ we mm_u:m=cmeu .o me=m_u 60:32.06 .32.: ...-6 5:33:06 5:33:06 3 up a v o 3 up a v 6 up up a v 6 1 ul . u u u - N6 «.6 lo/o / \ «.6 to m. '6 m. to m. .6 .6 b u a 0.0m 0.C ” 0.0“ a .e. m .. a .8 NM .. a .m N... mmaa q. m~H* ca e. mafia a; 60:05.860 80:95.860 86:05.860. 86:05.660 3 up a v 6 M— 4 «1’, a Vm/n/uc u 0 0w In. a v 6 o. «p o v 6 6 I” 610/ 1 6/6161. /(\n«6 «d .«d «d O .. L .16“« some .cduu emu l W m l W m . .odw odm new new A A A u J a and Na . 3 am an E 8 em and NM ... cm and Na ... -H¢. NH~* mcfi* .6; as ea as 45 .mm peeswemexmv ce__eeeegm.eeww Heme: e_ewueeemem esp ee meewueeoemm Feweweee eeeuxwm mewcee wewaweu .em .w mweweee .u we meeepemw we meweeeeeeew .w ecemwe 36:32.00 :33. ...-6 .322230 :23. 2.00 a. «w a a 3 a. o v o o. «w o v o 3 a. o v a lad 1 V. O V. 0 V. J 9 O O I . Aguonbug \\ o/. j Annabel; a. 6mm M... ... em and NM ... em 69. Na .8 «m an um ... mmuw mmaw vwfiw . Nofiw Q. e; e. e; gaze-:06 .322236 30:30:06 cot-.3606 e o. a. e c e «d «e m tom tom 0 o m new eem tm mam Mm ed .6 uafiw ea ea 46 .wm pewsweeexev ee meewueemeea _eweweee eeeexwm neweee wewuwee .em .w 30:32.06 6. up a Q 6 qlqnqvqnquqlqnnu 1 L L l q 6 Aauonbug a 3m NM emew co..o.o..06 6.6 6.. 6. N. 6 v 6 30:32.06 up a v 6 mm 6% Nu ... mwew 30:32.06 to u... b 6.6 h em 6%. Nu ... mafin Le eeeegp.mew— Hews: o_e_ueeumem we» mwe_Eeeu .u we mopepemw we mmweeeeemem cot-.....omv .323236 a. e . e. a. e e e «.6 m. c... u m ... em 68. NM ... Neew 6.— 3922236 cot-22.06 «p 6 v 6 6 «.6 «.6 eduu to m gm ed em and NM 8 em and we .6 meew . eeew 6p 6.— Aauonbug Aauonbug .m «gnaw; 47 .RDH peeawceexmv ee__eeeezw.o:._ Heme: m_eweeeumem use so meewueeeeem _eweweeu eemuxwm mewcee _ewuweu .em .w mwewaec .m we meee_emw we mmwuemeamem .m meemwm £0..o3£o6 £0..o3£o6 £0..o3£o6 £0..o3£o6 e. .... o v e e. a. e v e e. a. e e e "I" - '4!!!” /6 6: /6|1|11 e L «.6 1 «.6 U ... m m .6 w . ...m u u m .. m 4 e... m m new. . eem 1 4 aim: aim: 388: 65%: 3.: cc: RN: . m3; .... e.. e. .... 5:33:06 £0.33£o6 . £0..o3£o6 £0..o3£o6 e. a. e v. e e. a. e v e e. a. e o e 2 a. e v e JI—IJIJI-II-l . \.. . o\ «.6 \0 ... «.6 o/o\ .. \o 1 . ...e u. o\. . co m. u. u. .. 6.6 m 4 6.6 m h m a and N... . ... em 6mm N... .6 S: . 8.. .. 6.. 6.. 48 .r. ueeewemexev Le meeeeu we.— pews: m.e.uemem:m ecu ee meewueememm .e.e.eee emeux.m me.eee .u.u.eu .em .w m.c.Eec .u we mmeepem. we mm.ueeeemcm .oH ee:m.u Bonus-66:60 86:666-669 p 86:666-560. . O O. «’86—:“6860' O 6' «p 6 6 J 6- «p 6 v 6 6 «p 6 £+ulru . u 1 - 6 ~inlil...\|..l~. 1 /6 .. «.6 «.6 .. to m. m. m. tam. . a m m . n .. 6.6 m M h 66 b a ...m m. .6 6.... and NM. .6 6... 6% ...... .6 6... 6% NM. .6 mefiw oefiw wmaw meaw q. 66 e. e. £233£06 £233£o6 £233.36 £233.36 6. «p 6 v 6 6. «p 6 v 6 6. «p 6 v 6 6p «— 6 v 6 0.0 h .A M 6m 6mm ...m .6 66.. 49 .N ueEs.emexmv Le meeeeu ee._ Heme: m_e.ueeemem we» ee wee—ueeeeem .e.e.eee eeeex.m me.e=e .e.u.ee .em .w m.e.Eec .m we mean—em. we we.eeeeeegu ... meemwm 89-.050306 GouoobOQ-Co ...—....CCG cop-UbCBOG e. .... e v e e. a. e c e e. a. o v e [/6 6 6 H 4 .«d n ...e u u. m. m. . m . . 6.... 2m NM .6 6m 8m NM .6 6m ...... NM .6 meaw oefifi meaw e. e.. e.. £233£o6 £233£o6 £233£o6 e. N. e v e e. a. o e e e. a. e v e o. a. o v e «e «d .6\6 6 ... J. a. J. .\ ...m ...... . L n n n o o o 6.. .mm Nu... .m .2 Na 3 “ea“ :66 66 50 We 1°' 3:162}2623:2311???5.5835516“"133 .22. Eh?” f‘ sp' Md for calculation in each set were defined as having a selection coefficient of zero. PROGENY SET 1 Isolate Genotype EXP. 4 EXP. 5 EXP. 6 AVE.a 5.0. #52 £2 £3a £4 £15889 - .20 .16 .26 .21 b .05 #59 E2 £3a £4 £15889 + -.08 -.02 .06 -.01 a .07 #70 £2 E3a £4 £15889 - .06 .04 .09 .06 a .03 #65 £2 £3a E4 £15889 - .22 .15 .20 .19 b .04 #92 E2 E3a £4 E15889 - .00 .01 .03 .01 a .02 #51 £2 E3a E4 £15889 - .00 .00 .00 .00 a .00 #78 E2 £3a E4 E15889 + .19 .15 .27 .20 b .06 #89 E2 E3a E4 £15889 + .18 .16 .30 .21 b .08 PROGENY SET 2 Isolate Genotype EXP. 7 EXP. 8 EXP. 9 AVE.a 5.0. #116 £2 £3a £4 E15889 - .09 .16 .25 .17 bc .08 #105 E2 £3a £4 E15889 + .15 .20 --- .18 be .04 #117 £2 E3a £4 E15889 - .22 .33 .36 .30 d .07 #121 £2 £3a E4 E15889 - .04 .07 .19 .10 be .08 #102 E2 E3a £4 £15889 + .04 .06 .18 .09 b .08 #124 £2 E3a E4 £15889 - .00 .00 .00 .00 a .00 #125 32 £3a £4 E15889 + .19 .19 .21 .20 c .01 #138 E2 E3a E4 E15889 + .13 .16 .18 .16 bc .03 51 Table 10. (cont.) PROGENY SET 3 Isolate Genotype EXP. 10 EXP. 11 EXP. 12 AVE.a 5.0. #146 22 £36 £4 £15889 - .11 .13 .06 .10 b .04 #150 32 £3a £4 E15889 - .27 .31 .29 .29 a .02 #106 22 £36 £4 E15889 + .00 .00 .00 .00 a .00 #141 p_2 £36 _P_4 E15889 - .05 .10 .03 .06 b .04 #143 32 £36 £4 E15889 + .21 .22 .14 .19 c .04 #127 p_2 _P_3a _P_4 £15889 - .15 .24 .13 .17 c .06 #140 32 £3a 34 £15889 - .26 .31 .27 .28 a .03 #149 £2 E3a _134 E15889 + .08 .08 .02 ' .06 b .03 aWithin each progeny set, isolates with selection coefficients followed by the same letter are not significantly different than one another as indicated by Student-Neuman-Keuls multiple range test. 52 selection curves are Simple. However, when more than two isolates are present, and even though the fitnesses of the isolates are constant, the curves can be complex because of the simultaneous action of all the isolates. The change in frequency of an isolate from one generation to the next is determined by its fitness relative to the average p0pulation fitness (see Appendix A). Since selection favors the more fit isolates, the average p0pulation fitness increases each generation. Thus, isolates with fitnesses above average (but not the most fit) may at first increase in frequency, while the less fit isolates are being removed from the population, and then decrease in frequency when they, in comparison to the remaining isolates, become relatively unfit. Similarly, the rate at which an isolate increases or decreases may change in succeeding genera- tions. The results with the progeny are suggestive of some of these phenomena, although the true selection curves are obscured by sampling error. In addition, variations in environment over time or interactions between the isolates may have caused fluctuations in the fitnesses of the isolates and made the selection curves even more complex. Sampling errors (errors in estimation of the true frequency of an isolate within a chamber) may account for some of the observed variability in isolate frequency. Deviations from the true frequency may have resulted from random error or a systematic error inherent in the sampling method. Random error can be reduced by increasing the sample size. Unfortunately, an increase in sample Size was technically unfeasible in these experiments. Systematic error is the tendency of the sampling method to detect one isolate more readily than another. If systematic errors are consistent fron one sampling time to the next, they should not affect the overall conclusions of the experiments. However, a 53 fluctuation in the ability to detect the different isolates could account for some of the noticeable deviations from smooth selection curves. As will be discussed later, the isolates of E. graminis f. sp. tritici used in this study appear to be sensitive to environmental stress. This sensitivity may have affected the ability of the sampling method to detect the various isolates. In progeny set #1, two of the most fit isolates appeared to be #59 and #51. Isolate #51 was chosen as the standard for calculation of selection coefficients in this set because of its high fitness and relatively low variability compared to isolate #59. Isolate #51 by definition has a relative fitness (W) of 1.00 and a selection coefficient (5) of 0.00. Isolate #59 was Slightly more fit than #51 in experiments 4 and 5 as indicated by negative selection coefficients, but, overall, there was no significant differences in fitness between the two isolates. Among the progeny in set 1, there was no clear correlation between any gene or numbers of virulence genes and fitness. Note that isolates #59 and #51, which were similar in their fitness, differed by three £ genes. The same is true for isolates #52 and #89. These results do not support Vanderplank's hypothesis. There were problems in maintaining constant light and temperature conditions in the chamber used for experiment 5 between generations 9 and 13, which may account for some of the variability observed in this set. A Student-Neuman-Keuls multiple range test showed the isolates in set 1 could be divided into two fitness classes. This may be indicative of the presence of a single locus controlling the majority of the fitness differences observed. In the second set of eight progeny, the most fit isolate was #124. Note that isolates #121 and #102 are of Opposite genotypes but similar 54 fitness. The failure to obtain a fitness estimate for #105 in the third replication was due to its rapid dr0p in frequency to undetectable levels. Selection pressure appeared to be stronger in the third replicate (experiment 9). The reason for this is not known, but may be attributable to some unknown difference in environmental conditions. A multiple range test showed evidence of roughly 4 classes of fitness, indicating the possibility of two loci controlling fitness. In the third set of eight progeny, the most fit isolate was #106. Again in this set, there was no clear correlation of any gene or number of.£ genes and fitness. Note that isolate #146 and #149 are of exactly Opposite genotype with respect to the genes studied, but are of similar fitness. The reduced variability from generation to generation in this set may be attributable to fewer technical problems with the growth chambers and thus more constant environmental conditions. A multiple range test indicates the presence of 4 classes of isolates, again suggesting the presence of two loci controlling fitness. Comparisons of the selection coefficients from set to set are more difficult to interpret because the standards used for calculating the selection coefficients were different in each set. However, it is obvious that the ranking of the isolates Shifts from set to set. Note that isolates #51, #124, and #127 are identical with respect to the genes studied but ranked differently in the three sets. In addition, there was no correlation between the number of virulence genes in an isolate and its fitness (r2=.001). These results indicate that the large difference in fitness between MS-l and Mo-10 was not due to the £2, £3a, £4, or £15889 genes. Instead, the majority of the fitness differences observed are probably 55 attributable to 1 or 2 unknown genes segregating more or less randomly with respect to the known £ genes. In the presence of unknown genes having a large effect on fitness, it is difficult to detect smaller differences in fitness that may be attributable to the known virulence genes. Nevertheless, the data were analyzed statistically to determine whether or not any of the known virulence genes were associated with reduced fitness. There were 24 experimental units (isolates) assigned to 3 blocks (progeny sets) with 8 combinations of 3 "treatments“ (£2, £3a and £4) in a factorial design. For example, 12 isolates were given a "treatment" of a.£2 gene and 12 received E2. All received unknown “treatments" of unidentified genes. The fitness of each isolate was determined three times (3 chambers). The hypotheses to be tested were whether or not isolates with E genes were on the average different than isolates with £ genes and whether or not there was interaction between these genes. Results are shown in Tables 11 and 12. Perhaps because of the large effects of unknown genes, no significant effects were detected, though there were hints of a decrease in fitness in isolates with £3a, and an increase in fitness in isolates with £2. The lack of Significant interaction terms indicates that there was no detect- able synergism between the loci. Genes for mating type and virulence on CI 15889 were not detected until after the experiments were initiated and were not a part of the factorial design. Strictly speaking, therefore, they cannot be tested for their effect on fitness, because tests for their effects are not orthogonal with the tests for the effects of £2, £36 and £4. By doing these tests, the risk of erroneously showing significant differences is increased. Nevertheless, t-tests for the effect of £15889 and Egg were performed and no significant differences were detected. 56 Table 11. Statistical analysis of selection coefficients of the progeny of E. graminis f. Sp. tritici isolates MS-I and Mo-10. Analysis of Variance Degrees of Source freedom SS MS F-statistic Progeny Set 2 .0079 32 1 .0104 .0104 0.941 n.s. .333 1 .0216 .0216 1.950 n.s. E4 1 .0001 .0001 0.006 n.s. E2 x E3a 1 .0006 .0006 0.054 n.s. E2 x‘E4 1 .0193 .0193 1.739 n.s. E3a x‘E4 1 .0008 .0008 0.074 n.s. E2 x E3a x‘E4 1 .0000 .0000 0.002 n.s. Error 14 .1551 .0111 The following comparisons are not orthogonal to the tests listed above. Degrees of Comparison freedom t-Statistic .E15889 vs. £15889 1 0.719 n.s. E‘.’ VS. &‘ 1 00259 nos. 57 Table 12. Average selection coefficients associated with alleles f0r virulence and mating type in the progeny of isolates MS-l and Average Average Gene Selection Coefficient Gene Selection Coefficient E2 .16 E15889 .15 ‘£2 .11 £15889 .12 E3a .10 [£35 + .13 B36 016 [Ila—t " 014 P4 .14 £4 .13 DISCUSSION The suggestion that virulence genes might reduce the ability of pathogens to survive and reproduce caught the imagination of scientists concerned with breeding for disease resistance. It offered a way to use natural evolutionary forces, the key to a pathogen's ability to adapt, as a weapon against the pathogen itself. The main arguments for Vanderplank's hypothesis were made in his 1968 book, "Disease Resistance in Plants". To date, no test of Vanderplank's hypothesis has been made that has been generally accepted to be unambiguous. This failure is in part attributable to the fact that evidence obtained from observations of agricultural and wild pathogen p0pulations, though suggestive, is circumstantial. It is also attributable to the difficulty of devising controlled selection experiments in which the results are not confused by the linkage of virulence genes to other genes influencing fitness. The difficulty in testing the hypothesis is made worse by Vanderplank's "weak gene-strong gene" argument. Vanderplank argued that some resistance genes are “weak" and that the virulence genes necessary to overcome them have little or no effect on fitness. Since there is no £_£51951 way of knowing whether a resistance gene is weak or strong, in order to reject Vanderplank's hypothesis all virulence genes must be tested for their effects on fitness. The task of testing the effect of even a single gene is difficult. This work demonstrated that two parental isolates of Erysi£he 58‘ 59 graminis f. Sp. tritici which differed in their genes for virulence also differed in their ability to compete on a susceptible host in a way that was suggestive of stabilizing selection. However, a segregation test showed that of the difference in fitness in the parental cultures could not be attributed to the known £ gene differences. Analysis of the host-range of the parental cultures suggested at least four.£ gene differences. The report by Slesinski and Ellingboe (1969) that MS-1 was avirulent on lines W1th.flflle.EEZ:.EE331.EE3P and ‘E£4 and virulent on host lines derived from Michigan Amber was confirmed. Mo-10 was found to be virulent on lines with E£2,.E£3a and .396: Further tests showed that MS-l was avirulent on line CI 15889 whereas Mo-10 was virulent. No other host range differences were detected in this study. The culture Mo-10 did not grow as well on host lines with E£4 as it did on Chancellor. In laboratory tests Mo-10 repeatedly gave a 4' reaction. In greenhouse tests the effect was A more pronounced and Mo-10 gave a 3' reaction. Mo-10 gave a type 4 reaction on Chancellor in both laboratory and greenhouse tests. Similar observations of incomplete virulence on E94 were reported f0r other mildew cultures by Martin and Ellingboe (1976) and Nass (1980). A genetic analysis of the inheritance of virulence confirmed the presence of four independent loci controlling the difference in host range between the parental cultures. The segregation data are consistent with the gene-for-gene hypothesis as stated by Flor (1955, 1956, 1971) in that single genes in the pathogen determine virulence or avirulence on host lines in which resistance is determined by Single genes (Briggle, 1969). Evidence for complementary Single gene control of virulence in E. graminis f. sp. tritici has been reported for two unlinked loci, one 6O controlling virulence on host lines with gene E£1 and the other on host lines with gene E£2 (Powers and Sando, 1957; Powers and Sando, 1960; Powers, Moseman and Sando, 1961). Although only two chromosomes are reported to be present in E. graminis (Kimber and Wolfe, 1966), the loci detected in this study for virulence on wheat lines with genelem , Em3a, and E£4 are apparently unlinked. Similar results were reported by Leijerstam (1972) for loci controlling virulence °".EEZ and E94. Leijerstam reported one locus controlling virulence on Em3a in two crosses and two loci controlling virulence on E£3a in a third cross. He also reported linkage between one of the loci controlling virulence on ‘EQBa and the locus controlling virulence on E£4. The difference in results may be due to the isolates Studied. Further study of the genetics of virulence on E£3a is required. Additional loci in E. graminis Sp. have been reported by Hiura (1978). When the relative fitnesses of MS-l and Mo-10 were compared under growth chamber conditions, Mo-10 was found to be 19 to 32 percent less fit than MS-l. Similarly, the most unfit progeny from a cross between these two isolates were 21 to 30 percent less fit than the most fit progeny in each set. Leonard (1977a) estimated reductions in fitness associated with unnecessary virulence genes from 12 to 39 percent based on his own work with Bi£olaris.££yg1§ (1977b) and Puccinia graminis f. sp. ££££££ (1969b) and 22 to 42 percent based on the work of Watson and Singh with races of Puccinia graminis f. Sp. tritici (1952). The data presented here indicate that while the parental cultures had large differences in fitness, similar to the amounts calculated by Leonard, only a small fraction, if any, of the differences were associated with virulence genes. Instead, the difference in fitness between the parental 61 cultures appeared to be controlled by one or more loci of unknown function. Though a large number of characteristics probably determine overall fitness, the results suggest that only one or two loci were of major importance under these particular experimental conditions. It is possible that the fitness loci also happen to be undetected loci for virulence, but there is no evidence of this at present. As mentioned earlier, Mortensen (1974) reported no differences in the competitive abilities of races of E. graminis f. Sp. tritici with and without £ genes. The difference in results may be attributable to the isolates used or the experimental and environmental conditions employed. My own results (Appendix B) and the results of others (Katsuya and Green, 1967) indicate that fitness may be environmentally dependent. The segregation test was designed to determine whether or not a known difference in fitness between two isolates was conditioned by known £ genes. The data clearly demonstrated that, in this case, the majority of the fitness difference was not conditioned by known £ genes. Unfortunate- ly, in the presence of unknown genes with large effects on fitness, the experimental design used in this study was not effective for detecting possible residual effects of the £ genes. Gene £3a was associated with a slight decrease in fitness, in a manner suggestive of stabilizing selection. Gene £2 was associated with a slight increase in fitness. If these results are indicative of the true effects of.£ genes, stabilizing selection may not be powerful enough to be of significant value in controlling powdery mildew on wheat. Even making the unjustifiably optimistic assumption that all £ genes reduce fitness by the amount associated with £3a, based on the model of Marshall and Pryor (1978), at least 11 wheat lines with different resistance genes would have to be 62 incorporated into a multiline in order to stabilize the racial composition of the pathogen. If most £ genes behave as £2, £4, and £15889 have in this study, stabilization of the pathogen by use of multilines may not be feasible. Further crosses and progeny tests could assist in the detection of residual effects of.£ genes. Certain pairs of progeny, such as #59 and #51, and #146 and #149, are of different genotype but similar fitness, suggesting that these pairs may have the same genes for fitness and that the known £ genes have little or no effect on fitness. A cross between these isolates and an analysis of the progeny could determine whether or not these progeny have the same major fitness genes, whether or not there are any residual differences in fitness, and whether or not the differ- ences are associated with virulence genes. AS gene £3a shows a slight association with reduced fitness, it would be interesting to demonstrate by a further cross and progeny test whether or not the observed effect was real and due to a Slight effect of £3a or due to loose linkage to unknown gene(s) of large effect. Considering that there appear to be at least 1 or 2 genes with large effects, the latter seems a likely possiblity. Demonstration that the association of a particular gene and fitness is not due to a tight linkage with a gene of small effect would require more extensive progeny testing. Although this Study gave no evidence for stabilizing selection, it has not disproven Vanderplank's hypothesis. Controlled selection experiments, out of practical necessity, are limited to analyzing only certain genes under defined environmental and experimental conditions. The selection coefficients obtained are applicable only to the conditions in these experiments and do not necessarily apply to other conditions. 63 Modifications such as continuous (overlapping) generations, growth on adult plants, overwintering, addition of a sexual stage or different light, temperature or moisture conditions may change the results of fitness tests. Evidence for environmental effects on the relative fitnesses of the mildew cultures used in this study are presented in Appendix B. The influence of environment and experimental technique on the competitive ability of other pathogens has been documented (Katsuya and Green, 1967; Osoro and Green, 1976; Thurston, 1961). It is possible that the £ genes studied may reduce fitness but this effect was not detected under these particular experimental conditions. Stabilizing selection, as defined by Vanderplank, may also exist at levels too low to be detected by this method, at other loci, or in other organisms. APPENDIX A DISCRETE SELECTION MODEL FOR STRAINS OF ASEXUALLY REPRODUCING ORGANISMS COMPETING IN A COMPLEX MIXTURE APPENDIX A DISCRETE SELECTION MODEL FOR STRAINS OF ASEXUALLY REPRODUCING ORGANISMS COMPETING IN A COMPLEX MIXTURE A hypothesis about the relative fitness of two strains of asexually reproducing organisms can be tested without a quantitative measure of fitness by simply noting an increase or decrease in the relative frequency of the organisms. However, a quantitative measure of fitness permits more objective comparisons of the results of different experiments and provides numerical values for use in predictive models. In addition, it can provide meaningful statements about relative fitness when more than two strains are present in a complex mixture. The most appropriate measures of fitness are ones with reasonable biological foundations and few assumptions. Leonard (1969a) presented an asexual selection model for use in comparing the fitnesses of two isolates of plant pathogens. His model expressed the frequency of one isolate as a fraction of another and employed a logarithmic transforma- tion to obtain a straight line selection curve. This permitted the use of linear regression to obtain the best estimates of relative fitness, and thus reduced problems associated with sampling error. Leonard's model assumes that generations are discrete, that selection coefficients are constant, and that only two isolates or genotypes are present. The model presented here was derived from Leonard's two isolate case with a trivial extension that allows the model to apply to selection in complex mixtures with any number of isolates. Except for the number of strains, it has the same assumptions as Leonard's model. 64 65 Definitions: frequency a fraction of the colonies in the population that are of one genotype absolute fitness = the number of conidia produced per colony that survive to infect and fOrm colonies N = p0pulation Size . the number of colonies in the p0pulation in any generation t = number of generations S a selection coefficient Model: Isolate (genotype) A B Frequency p q Absolute Fitness WA W. Relative Fitness 1 a:.. Wa= 1—S Generation 1 p a £nWeN a 93W.N (A) 1 powm + q,w,N q‘ powm + q,w,N divide q, by p, and cancel £1. , gnW. , Sn. W Pi Po A Po ‘3 Generation 2 g; . 3; "a, gm “a P2 P1 P0 therefore, by induction: Generation t g; , gm ”7° 9! Po lnfllslnfll+tln W. P1 Po (8) Note that this is the equation for a straight line with a slope of ln W (see Figure 12). 66 1'0 ' on. 4", (I ”.. /" isolate A Frequency O 4 8 12 16 O 4 8 12 16 Generation Generation Figure 12. Linearization of curves from a theoretical selection experiment. a) Change in frequency of isolate A (s-.00) and isolate B (s-.22). b) Plot of transformed data (p is the frequency of isolate A and q is the frequency of isolate B). 67 Leonard stated that p + q = 1, therefore = + t ln W lfl _qT Inf-a: Q Extension to n isolates: To obtain the equation on line B, the assumption that only two isolates are present is unnecessary. If n isolates are present, the denominators in line A become powm + q,w,N + 17,ch + . . . The denominators cancel to give . 3.1.39.1“; ’ hath’ °'° pf p0 p1: p0 Thus for a complex mixture of n isolates, n-1 fitness terms may be calculated, the fitness of the isolate used as a standard having been defined aS 1.0. Adequacy of the Model: The fit of this selection model to the observed data can be seen by inspection of coefficients of determination (see Appendix C). The amount of variability in isolate frequency explained by the model varies between isolates and experiments. Random variability may be attributable to sampling problems, as discussed earlier. However, systematic deviations from the model may indicate non-constant selection coefficients. Changes in relative fitness may be attributable to fluctuations in environmental conditions (as discussed in Appendix B) or frequency dependent selection. Although some of the results with MS-l and Mo-10 are suggestive of frequency dependent selection, further experimentation is necessary in order to determine the source of the deviations. APPENDIX B EVIDENCE FOR ENVIRONMENTAL EFFECTS ON RELATIVE FITNESS APPENDIX B EVIDENCE FOR ENVIRONMENTAL EFFECTS ON RELATIVE FITNESS Variability in the results of relative fitness tests suggested that the fitness of isolates of E. graminis f. sp. tritici may be influenced by environment. The purpose of this appendix is to relate further evidence for environmental effects. Competition Between MS-l and Mo-10 In addition to the competition studies with isolates MS-l and Mo-10 described in the main part of the text (experiments 1 through 3), a Similar experiment was performed under different environmental and experimental conditions with different results. The purpose of the experiment was to develop methods for maintaining E, graminis in a self-sustaining epidemic and to develop methods for measuring the frequency of each of the isolates. For this reason, the methods were not constant throughout the study. The experiment was per- formed in a controlled environment chamber (Sherer-Gillett CEL 512-37). The temperature was 20 t 1°C during the day (14 hr) and 17 1°1 C at night (10 hr), except for a 6 hr period on day 15, when the chamber overheated to 32°C. Light was supplied by a combination of fluorescent and incandescent bulbs for a total of approximately 2.0 x 104 ergs sec‘lcm‘2 at pot level. The relative humidity fluctuated between 55% and 100%. The isolates MS-1 and Mo-IO were inoculated onto 10 cups of 7-day- old Chancellor seedlings as described earlier and the cups were placed in the chamber. During the first 11 weeks, 10 additional cups of 7-day-old 68 69 Chancellor seedlings were added to the chamber 3 times a week. Each cup remained in the chamber 14 days and then was removed. During the last 6 weeks, the schedule for addition and removal of seedlings was the same as for experiments 1 through 12. Thus, generations were continuous during the first 11 weeks and discrete thereafter. Samples of the mildew were taken at intervals throughout the experi- ment. At first the mildew was sampled by using tiny Spatulas to remove clumps of mycelia and conidia from supposedly single colonies on plants in the chamber, then testing them on different wheat lines as described earlier. However, the mildew was so dense on the seedlings that a high frequency of the samples were mixtures of the two strains. The rate of successful transfer of these clumps was also low. Therefore this method of Sampling was dropped and from then on mildew was sampled by lightly dusting conidia from plants in the chamber onto the susceptible wheat Chancellor and selecting well isolated colonies from these plants for testing, as described previously. The number of colonies successfully tested per sampling time varied considerably throughout the experiment (from 12 to 80). Therefore, 95% confidence intervals of isolate frequen- cy were determined to indicate the reliability of each measurement. Results are shown in Figure 13. At no time in the experiment did the frequency of MS-l or Mo-10 differ significantly from 50%. Calcula- tion of selection coefficients using MS-1 as the standard Showed that Mo-10 was about 1% more fit than MS-1 (s = -0.01) but the difference was not statistically significant. For the last 6 weeks, during which time the experimental procedures, but not environmental conditions, were identical to those used in experiments 1 though 3, no selection was observed (5 = 0.00). The experiment was not repeated, so it is not 70 ...m: we heeeecaew any eew mpe>ceuew aueeewweeu umo ueameeeae meme _euwuce> .ce__oeee:o wua.e~> peace apewueoumem use ee mxaa: NH meweee cane: eee Him: we meeust e e. wewuwcu .em .w mwewEeem .u we “in: eeewemw we xeeaecaew .m. eeemwe «Jae; 6w '— Np 6p 0 6 N _1 q u q a u u q u q 1 q . «.6 .6a w v 1111111111111 flotnoouo 11L flit! I \s .M tttttttttttttttt IIIoIIA .1 III \ u 47111:: J 6.6 e 1.. A T111111111 1 6.6 71 possible to conclude with certainty that differences in environmental conditions caused the differences in results. It is also not possible to determine from this information which environmental parameters were responsible. The conditions during the last 6 weeks of this experiment differed from those in experiments 1 through 3 by light intensity, night temperature and humidity. Effect of Environmental Stress on MS-l and Mo-10 Further evidence for the influence of environment on fitness comes from the differential sensitivity of the isolates to environmental stress. This phenomenon was first noted when isolates of E. graminis were accidentally exposed to continuous light. The cultures grew poorly and some appeared to be more inhibited than others. The differential inhibition of the isolates was confirmed in the following experiment. Single colonies of each of the isolates were inoculated onto one of four 5-day-old Chancellor seedlings growing in soil in wax paper cups. The cups were covered with lamp chimneys and placed under Gro-lux lights in the laboratory, as described previously. Six pots of each isolate (controls) received light 14 hr per day. The temperature inside the chimneys was 24 1 1°C during the light phase and 21 i 1°C during the night. Fifteen pots (treatments) were placed under identical continuous lights. The temperature inside the chimneys was 24 i 1°C continuously. After 7 days the pots were shaken to distribute conidia to the noninoculated seedlings. After an additional 6 days, the growth of the cultures was compared by counting the number of visible colonies on the middle 1 cm of the first leaf of 3 seedlings in each pot (excluding the originally inoculated seedling). 72 Results are Shown in Table 13. The cultures exposed to continuous light grew poorly compared to the control cultures. Sensitivity to the treatment varied between the isolates and was significantly correlated with the selection coefficients (Figure 14) with approximately 67% (r2 . .67) of the variability in the selection coefficients associated with sensitivity to the treatment. This suggests that a Similar stress may have been acting in the selection experiments to give differential growth of the isolates, even though the cultures in the chambers appeared vigorous at all times. AS both light duration and temperature were modified by exposing the cultures to continuous light, it is not clear which parameters were responsible for the differential effect. Conclusions The results of these two studies suggest that the isolates of E. graminis f. Sp. tritici used were differentially sensitive to light and/or temperature and suggest that it was this difference in sensitivi- ty, at least in part, that was detected in the relative fitness tests. 73 Table 13. Effect of continuous light on the growth of isolates of E. graminis f. sp. tritici. Average Colonies Per Cm Isolate 14Tfiours 24 hours % Inhibition light per day light per day MS-l 6.4 2.9 55 ”5.10 5.2 Go]. 98 #52 8.4 1.3 85 #59 5.7 4.3 25 #70 7.6 5.3 29 #65 9.4 0.3 97 #92 5.8 5.4 7 #51 6.7 5.6 3 #78 6.8 0.4 94 #89 7.7 1.4 82 #116 6.4 0.2 97 #105 4.4 0.1 98 #117 5.1 0.1 98 #121 7.8 0.1 99 #102 5.9 5.0 15 #124 6.1 5.0 18 #125 7.7 4.2 46 #138 7.3 3.5 52 #146 6.8 5.4 21 #150 7.4 0.1 99 #106 4.5 3.5 22 #141 5.9 2.6 56 #143 6.9 1.1 84 #127 6.3 0.9 86 #140 6.9 0.1 99 #149 8.8 5.1 44 74 am we mac: u—ee eaeeeac x.m1aee eee use. use e. m — m: ace e . eeewueeu e» wewwwmmw meweum—em zewwe.u V we.u.cu 6......” ...... .. eem ewe: wuwuaes awe eewue ee paeeeu . a. mean. .6 66:326. a AXUP _ .UQ 2 .99 . .ee 6. - .UN 6 - C C .61.! ea. \ . . . \\\\\\ \\ \\\\\\ \\\ \\ .. . \\\\\\\\ 1 OP. . \\\\ . S e11111e. m- 11111 e n . . . I39. 0 l u 6 .uN. «J 6 O C 6 I61 nu D . u 1 en. .. ILAYQ. APPENDIX C DATA USED FOR CALCULATION OF SELECTION COEFFICIENTS APPENDIX C DATA USED FOR CALCULATION OF SELECTION COEFFICIENTS Data used to calculate selection coefficients (Tables 3 and 10) and to plot selection curves (Figures 2 through 11) are presented in this appendix. Table 14 lists the number of times each isolate was observed per sample. Isolate frequencies were obtained by dividing this number by the total number of colonies tested in that sample. Relative fitness terms and selection coefficients were calculated by methods described previously. Sl0pes of the regression lines (b) and coefficients of determination (r2) are given. If an isolate was used as the standard for comparison, its fitness was defined as 1.00. Towards the end of the experiments, many cultures were at such low frequencies in the population that they were absent from the samples. Since the logarithm of zero is not defined, these sampling times were not considered. However, because of sampling error, isolates occasionally did not appear in a particular sample, though they were Still in the population at low levels as indicated by their appearance in later samples. In these cases, it was assumed that the number of individuals of that isolate at that sampling time was one. This step was not strictly valid and may have resulted in underestimation of the true selection coefficients in some cases. However, the step was included because it resulted in less underestimation of the true selection coefficients than would have been obtained by omitting all sampling times with zero values, it was a reasonable approximation, and it provided additional sampling times for use in the regression analysis. 75 76 Table 14. Data used to calculate isolate frequencies and selection coefficients for cultures of E. graminis f. Sp. tritici. The data are recorded as the number 0 1n 1viduals of eacfi isolate per sample. EXPERIMENT 1 Generation Isolate o 2 4* 6 8 10 12 b w s r2 MS-l 25 55 91 70 62 63 48 -- 1.00 .00 - Mo—10 73 23 5 17 6 5 5 -.23 .80 .20 .52 EXPERIMENT 2 Generation Isolate 0 2 4 6 8 10 12 b W . s r2 MS-1 44 62 82 75 76 - - -- 1.00 .00 - "0'10 50 25 19 14 15 " ‘ '02]. e81 619 680 EXPERIMENT 3 Generation Isolate o 2 4 6 8 10 12 b w s r2 MS-l 90 74 92 85 - - - a- 1.00 .00 - Mo-10 120 11 8 10 - - - -.39 .68 .32 .66 EXPERIMENT 4 Generation Isolate 0 2 4 6 8 10 12 14 16 b w s r2 #52 24 27 24 22 4 15 3 10 1 -.23 .80 .20 .73 #59 2 2 4 9 14 15 6 10 23 .07 1.08 -.08 .50 #70 12 19 11 13 9 16 13 14 13 -.06 .94 .06 .52 #92 3 5 1 3 6 7 5 9 5 .00 1.00 .00 .00 #51 14 18 13 25 33 24 23 36 39 -- 1.00 .00 -- 77 Table 14. (cont.) #78 19 7 5 6 9 3 3 1 O -.21 .81 .19 .87 #89 8 6 5 O 1 2 2 1 O -.20 .82 .18 .62 EXPERIMENT 5 Generation Isolate 0 2 4 6 8* 10 12 14 16 b w s r2 #52 29 16 9 8 15 13 1 0 0 -.18 .84 .16 .43 #59 17 21 23 12 12 30 17 73 26 .02 1.02 -.02 .02 #70 9 5 8 12 15 15 5 2 14 -.04 .96 .04 .12 #65 5 7 6 5 6 8 1 0 1 -.16 .85 .15 .61 #92 4 8 7 6 10 5 15 3 11 -.01 .99 .01 .01 #51 15 21 39 37 28 17 19 26 55 -- 1.00 .00 - #78 5 10 5 6 8 4 2 0 1 -.16 .85 .15 .69 #89 5 3 2 4 0 1 o 0 0 -.17 .84 .16 .63 EXPERIMENT 6 Generation Isolate 0 2 4 -6___8__10 12 14 16 b W s r2 #52 18 27 8 2 4 - 7 1 0 -.30 .74 .25 .57 #59 12 27 11 20 10 - 15 18 38 -.05 .94 .05 .09 #70 3 13 23 12 11 - 15 5 5 -.10 .91 .09 .25 #55 13 15 9 11 13 - 5 0 3 -.23 .80 .20 .57 #92 10 5 7 18 14 - 10 25 23 -.03 .97 .03 .05 #51 15 3 28 30 40 - 32 51 30 -- 1.00 .00 - #78 8 11 10 5 5 - 1 1 0 -.31 .73 .27 .81 #89 4 5 2 2 O - O O O -.36 .70 .30 .37 78 Table 14. (cont.) EXPERIMENT 7 Generation Isolate 0 2 4 6 8 10 12 14 16 b w s r2 #116 8 13 5 - 10 19 11 6 9 -.10 .91 .09 .50 #105 2 3 1 - 0 1 1 1 0 -.17 .85 .15 .82 #117 7 5 4 - 0 2 1 1 0 -.25 .78 .22 .93 #121 7 19 10 - 61 27 28 26 19 -.04 .96 .04 .11 #102 3 8 19 - 12 21 16 13 11 -.04 .96 .04 .25 #124 8 20 27 - 18 28 47 53 61 -- 1.00 .00 - #125 5 12 1 - 7 . 2 1 2 1 -.21 .81 .19 .62 #138 1 7 3 - 0 2 1 3 1 -.14 .87 .13 .71 EXPERIMENT 8 Generation Isolate 0 2 4 6 8 10* 12 14 16 b w s r2 #116 8 2 10 - 5 6 10 1 7 -.17 .84 .16 .31 #105 2 0 3 - 1 1 1 1 0 -.22 .80 .20 .42 #117 7 6 0 - 1 1 0 0 0 -.41 .67 .33 .82 #121 7 13 1 - 6 10 21 15 14 -.07 .93 .07 .35 #102 3 2 38 - 21 23 14 15 13 -.06 .94 .06 .05 #124 8 30 2 - 68 47 52 67 66 -- 1.00 .00 - #125 5 23 1 - 6 14 8 4 1 -.21 .81 .19 .79 #138 1 2 0 - 1 2 1 1 0 -.18 .84 .16 .52 Table 14. (cont.) 79 EXPERIMENT 9 Generation Isolate 0 2 4 6 8 10 12 14 16 b w s r2 #116 13 14 14 14 16 4 5 1 0 -.29 .75 .25 .89 #105 2 0 0 0 0 0 0 0 0 -- -- -- - #117 12 9 13 0 2 1 0 0 0 -.45 .64 .36 .85 #121 10 18 12 29 16 9 3 4 4 -.21 .81 .19 .88 #102 6 10 13 6 3 9 6 4 1 -.20 .82 .18 .83 #124 11 37 37 54 66 79 80 95 84 -- 1.00 .00 - #125 8 8 1 4 4 2 1 2 0 -.24 .79 .21 .72 #138 2 4 1 1 1 0 2 0 0 -.19 .82 .18 .67 EXPERIMENT 10 Generation Isolate 0 4 8 12 16 b w s r2 #146 28 24 23 12 16 -.11 .89 .11 .88 #150 44 6 3 2 0 -.32 .73 .27 .96 #106 18 17 27 39 46 -- 1.00 .00 - #141 40 13 21 34 32 -.06 .95 .05 .52 #143 20 10 6 4 1 -.24 .79 .21 .98 #127 14 7 5 _3 3 -.17 .85 .15 .97 #140 35 10 4 2 0 -.31 .74 .26 .99 #149 5 14 15 6 6 -.08 .92 .08 .40 Table 14. (cont.) 80 EXPERIMENT 11 Generation Isolate 0 4 8 12 16 b w s r2 #146 28 23 22 18 15 -.14 .87 .13 .65 #150 44 19 3 0 1 -.37 .69 .31 .83 #106 18 5 34 49 45 -- 1.00 .00 - #141 40 17 24 21 33 -.11 .90 .10 .61 #143 20 12 4 3 2 -.25 .78 .22 .79 #127 14 6 6 1 1 -.28 .76 .24 .85 #140 35 13 3 2 0 -.37 .69 .31 .83 #149 5 6 8 6 8 -.08 .92 .08 .33 EXPERIMENT 12 Generation Isolate 0 4 8* 12 16 b w s r2 #146 15 26 31 39 22 -.06 .94 .06 .47 #150 15 10 2 1 0 -.34 .71 .29 .91 #106 6 6 17 15 22 -- 1.00 .00 - #141 17 31 32 38 50 -.03 .97 .03 .23 #143 14 6 7 6 4 -.15 .86 .14 .95 #127 2 9 2 0 2 -.14 .87 .13 .51 #140 19 7 7 1 0 -.32 .73 .27 .98 #149 1 8 9 9 4 -.02 .98 .02 .01 LITERATURE CITED LITERATURE CITED Barrett, J. A. 1978. A model of epidemic devel0pment in variety mixtures. Pages 129-137 in: P. R. Scott and A. Bainbridge, eds. Plant Disease Epidemiology. Blackwells Scientific Publications. 329 pp. Blanco, M. H., and R. R. Nelson. 1972. Relative survival of populations of race 0 and race T of Helminthosporium ma dis on a corn hybrid in normal cytoplasm. Plant Dis. Rep. 56:889- 1. Briggle, L. W. 1969. Near-isogenic lines of wheat with genes for resistance to Erysi£he graminis f. Sp. tritici. Crop Sci. 9:70-72. Brown, J. F., and E. L. Sharp. 1970. The relative survival ability of pathogenic types of Puccinia striiformis in mixtures. Phytopathology 60:529-533. Brown, J. F. 1975. Factors affecting the relative ability of strains of fungal pathogens to survive in populations. J. Aust. Inst. Agri. Sci. 41:3-11. Browning, J. A., and K. J. Frey. 1969. Multiline cultivars as a means of disease control. Annu. Rev. Phytopathol. 7:355-382. Crill, P., J. P. Jones, and D. S. Burgis. 1974. Evaluation of some concepts of variety devel0pment and disease control with host resistance. Plant Dis. Rep. 58:579-593. Crill, P. 1977. An assessment of stabilizing selection in crop variety devel0pment. Annu. Rev. Phytopathol. 15:185-202. Crow, J. F., and M. Kimura. 1970. An Introduction to Population Genetics Theory. Harper and Row, New York. 591 pp. Day, P. R. 1974. Genetics of Host-parasite Interaction. W. H. Freeman, San Francisco. 238 pp. Ellingboe, A. H. 1976. Genetics of host-parasite interactions. Pages 761-778 in: R. Heitefuss and P. H. Williams, eds. Physiological Plant Pathology (Encycl0pedia of Plant Physiology, New Series, Vol. 4). Springer-Verlag, Berlin, Heidelberg, New York. Fleming, R. A., and C. 0. Person. 1978. Disease control through use of multilines: a theoretical contribution. Phyt0pathology 68:1230-1233. 81 82 Frey, K. J., J. A. Browning, and M. D. Simons. 1977. Management systems for host genes to control disease loss. Ann. N.Y. Aca . Sci. 287:255-274. Flor, H. H. 1953. Epidemiology of flax rust in the North Central States. Phytopathology 43:624-628. Flor, H. H. 1955. Host-parasite interaction in flax rust: its genetics and other implications. Phytopathology 45:680-685. Flor, H. H. 1956. The comlementary genic systems in flax and flax rust. AdVe 6811812 6 8329-546 Flor, H. H. 1971. Current status of the gene-for-gene concept. Annu. Rev. Phytopath. 9:275-296. Groth, J. V. 1976. Multilines and “super races“: a simple model. Phytopathology 66:937-939. Groth, J. V., and C. 0. Person. 1977. Genetic interdependence of host and parasite in epidemics. Ann. N.Y. Acad. Sci. 287:97-106. Hiura, U. 1978. Genetic basis of formae speciales in EryiS£he gramini . D.C. Pages 101-127 in: D. M. Spencer, ed. The Powdery Mildews. Academic Press, New York. 565 pp. Jenkyn, J. F., and A. Bainbridge. 1978. Biology and pathology of cereal powdery mildews. Pages 283-321 in: D. M. Spencer, ed. The Powdery Mildews. Academic press, New York. 565 pp. Johnson, T. 1961. Man-guided evolution in plant rusts. Science 133:357-362. Katsuya, K., and G. J. Green. 1967. Reproductive potentials of races 158 and 56 of wheat stem rust. Can. J. Bot. 45:1077-1091. Kimber, G., and M. S. Wolfe. 1966. Chromosome number of Erysi£he graminis. Nature 212:318-319. Kiyosawa, S. 1977. Development of methods for the comparison of utility value of varieties carrying various types of resistance. Ann. N.Y. Acad. Sci. 287:107-123. Knott, D. R. 1971. Can losses from wheat stem rust be eliminated in North America? Crop Sci. 11:97-99. Leijerstam, B. 1972. Studies in powdery mildew on wheat in Sweden. III. Variability of virulence in Erysi£he raminis f. Sp. tritici due to gene recombination and mutation. Statens axtskyddsanstalt. Meddelanden 15:231-248. Leonard, K. J. 1969a. Selection in heterogeneous populations of Puccinia graminis f. Sp. avenae. Phytopathology 59:1851-1857. 83 Leonard, K. J. 1969b. Genetic equilibria in host-pathogen systems. Phytopathology 59:1858-1863. Leonard, K. J. 1977a. Selection pressures and plant pathogens. Ann. NeYe Acade SCie 2873207-2226 Leonard, K. J. 1977b. Virulence, temperature optima, and competitive ability of isolines of race T and 0 of ijolaris maydis. Phytopathology 67:1273-1279. Leonard, K. J., and R. J. Czochor. 1980. Theory of genetic interactions among populations of plants and their pathogens. Annu. Rev. Phytopathol. 18:237-258. Luig, N. H., and I. A. Watson. 1970. The effect of complex genetic resistance in wheat on the variability of Puccinia graminis f. Sp. tritici. Proc. Linn. Soc. N.S.W. 95:22-45. Mac Key, J. 1973. Fitness of virulence genes in ruSts and mildew in absence of host selection pressure. (Abstr.) Genetics 74 (June Suppl. No. 2, Part 2), 169. Mac Key, J. 1976. Genes of virulence and their adaptability in oat stem rust. Page 35 in: Proc. Fourth Int. Cereal Rust Conference. Marshall, 0. R. 1977. The advantages and hazards of genetic homogeneity. Ann. N.Y. Acad. Sci. 287:1-20. Marshall, 0. R., and A. J. Pryor. 1978. Multiline varieties and disease control. I. The dirty crop approach with each component carrying a unique Single resistance gene. Theor. Appl. Genet. 51:117-184. Martens, J. W., R. I. H. McKenzie, and G. J. Green. 1970. Gene-for-gene relationships in the Avena: Puccinia graminis host-parasite system Martens, J. W. 1973. Competitive ability of oat stem rust races in Martin, T. J., and A. H. Ellingboe. 1976. Differences between compatible parasite/host genotypes involving the E94 locus of wheat and the corresponding genes in Erysi£he graminis . Sp. tritici. Phyt0pathology 66:1435-1438. Mortensen, K. 1974. Konkurrenceforsdg med isolater af meldugsvampen, Erysi£he graminis, med forskellige patogenitetsgener. Nordisk Jordbrugsforskning 56:387-388. Moseman, J. G. 1959. Host-pathogen interaction of the genes for resistance in Hordeum vul are and for pathogenicity in Erysi£he graminis f. sp. Horaei. gfiytopathology 49:469-472. Moseman, J. G. 1973. Genetics of resistance to wheat powdery mildew. Pages 435-438 in: E. R. Sears and L. M. S. Sears, eds. Proc. 4th Int. Wheat Genetics Symposium, 6-11 August 1973. Univ. of Missouri. 84 Moseman, J. G., and H. R. Powers, Jr. 1957. Function and longevity of cleistothecia of Erysi£he graminis f. Sp. hordei. Phytopathology 47:53-56. Nair, K. R. S., and A. H. Ellingboe. 1965. Germination of conidia of Erysi£he graminis f. Sp. tritici. Phytopathology 55:365-368. Nass, H. A. 1980. The residual effects of certain defeated powdery mildew resistance genes. M.S. Thesis, Pennsylvania State University, College Park. 22 pp. Nelson, R. R., and G. L. Scheifele. 1970. Factors affecting the over- wintering of Trichometasphaeria turcica on maize. Phytopathology 60:369-370. Nelson, R. R. 1972. Stabilizing racial populations of plant pathogens by use of resistance genes. J. Environ. Qual. 1:220-227. Ogle, H. J., and J. F. Brown. 1970. Relative ability of two strains of Puccinia graminis tritici to survive when mixed. Ann. Appl. Biol. Osoro, M. 0., and G. J. Green. 1976. Stabilizing selection in Puccinia graminis tritici in Canada. Can. J. Bot. 54:2204-2214. Parlevliet, J. E. 1981. Deploying rust resistance. Pages 149-152 in: T. Kommedahl, ed. Proceedings of Symposia, Ninth International Congress of Plant Protection. Vol. 1. Burgess, Minneapolis. 411 pp. Person, C. 1967. Genetic aspects of parasitism. Can. J. Bot. 45:1193-1204. Person, C., J. V. Groth, and 0. M. Mylyk. 1976. Genetic change in host-parasite populations. Annu. Rev. Phyt0pathol. 14:177-188. Powers, H. R., Jr., and J. G. Moseman. 1956. Heterothallism in Erysi£he graminis tritici. (Abstr.) Phyt0pathology 46:23. Powers, H. R., Jr., and J. B. Moseman. 1957. Pathogenic variability within cleistotheia of Erysiphe graminis. Phytopathology 47:136-138. Powers, H. R., Jr., and W. J. Sando. 1957. Genetics of host-parasite zglztionship in powdery mildew of wheat. (Abstr.) Phyt0pathology : 53. Powers, H. R., Jr., and W. J. Sando. 1960. Genetic control of host-parasite relationship in wheat powdery mildew. Phytopathology 50:454-457. Powers, H. R., Jr., J. G. Moseman, and W. J. Sando. 1961. Genetic basis of the host-parasite relationship of wheat and powdery mildew. Recent Advanc. Botany 512-516. 85 Pryor, A. J. 1977. Plant disease resistance and the acquisition of pathogen virulence. Third SABRAD Congress Proc. 1:15-28. Roelfs, A. P., and J. V. Groth. 1980. A comparison of virulence phenotypes in wheat stem rust populations reproducing sexually and asexually. Phytopathology 70:855-862. Scheffer, R. P. 1976. Host specific toxins in relation to pathogenesis and disease resistance. Pages 247-269 in: R. Heitefuss and P. H. Williams, eds. Physiological Plant Pathology. Springer-Verlag, Berlin. Scheifele, G. L., R. R. Nelson, and C. C. Wernham. 1968. Studies on stabilizing selection in Erichometasphaeria turcica (Helmintho- S£orium turcicum). Plant Dis. Rep. 52:427-430. Scheifele, G. L., and R. R. Nelson. 1970. Factors affecting the survival of Trichometasphaeria turcica (Helminthosporium turcicum) on Egg may . Can. J. Bot. 48:160331608. Slesinski, R. S., and A. H. Ellingboe. 1969. The genetic control of primary infection of wheat by Erysi£he graminis f. Sp. tritici. Phytopathology 59:1833-1837. Spencer, 0. M., ed. 1978. The Powdery Mildews. Academic Press, New York, London. 565 pp. Spiess, E. B. 1977. Genes in Populations. John Wiley & Sons, Inc., New York. 780 pp. Strickberger, M. W. 1976. Genetics. 2nd ed., MacMillan, New York. 914 pp. Thurston, H. D. 1961. The relative survival ability of races of Phytophthora infestans in mixtures. Phytopathology 51:748-755. Trenbath, B. R. 1977. Interactions among diverse hosts and diverse parasites. Ann. N.Y. Acad. Sci. 287:124-150. Vanderplank, J. E. 1963. Plant Diseases: Epidemics and Control. Academic Press, Inc. New York, London. 349 pp. Vanderplank, J. E. 1968. Disease resistance in plants. Academic Press, New York. 206 pp. Watson, I. A., and 0. Singh. 1952. The future for rust resistant wheat in Australia. Jour. Australian Inst. Agric. Sci. 18:190-197. Watson, I. A. 1958. The present status of breeding disease resistant wheats in Australia. Agric. Gaz. N.S.W. 69:630-660. Watson, I. A., and N. H. Luig. 1968. The ecology and genetics of host-pathogen relationships in wheat rust in Australia. Pages 227-238 in: K. W. Findlay and K. W. Shepherd, eds. Third 86 International Wheat Genetics Symposium Proc., Canberra, Butterworths. Watson, I. A. 1970. Changes in virulence and population shifts in plant pathogens. Annu. Rev. Phyt0pathol. 8:209-230. Wolfe, M. S. 1973. Changes and diversity in populations of fungal pathogens. Ann. Appl. Biol. 75:132-136. Wolfe, M. S., and E. Schwarzbach. 1978. Patterns of race changes in powdery mildews. Annu. Rev. Phyt0pathol. 16:159-180. Wilson, E. 0., and W. H. Bossert. 1971. Primer of Population Biology. Sinauer Associates, Sunderland, Mass.. 192 pp. Wright, S. 1969. Evolution and the Genetics of P0pulations. Vol. 2. The Theory of Gene Frequencies. Univ. of Chicago Press, Chicago. 511 pp.