p "a V {T ’33 LL n.» .. L. «U. L - - Q» «E. “ISL inn” Hm]: Du“ Emu “Wyn Fun .r:u I... z” E, . , n .u. F... 9. n2. 3 nu,“ HF an H” aw , ,. u um mg. L Tu 2L fin flu» is 11111 it... a. «\w . A. an... m .:_. AL... 8 ":1 hm.” mum”... $1»! Lu”. 00 «an. "H... L)... Ln... = ¢ [Hutu ..... .- .A.l. r FE... ., nun A...“ 3 53.. «NW iv 5!. .3. Dr. . .. m «i, Auk ‘5. n. U .5“ sh ”Hm :Hh I: “L. .m3 of», L! Mun. “nu .u nu ma a...“ a u h: Hm 5.... 2.1. :r. _ I O l L r I u E 7:% r." V 'i AUVXQI'I THE-£815 ABSTRACT INHERITANCE OF RESISTANCE TO POWDERY MILDEW IN CUCUMIS MELO by Richard Roland Harwood A detailed study of the inheritance of resistance to powdery mildew (Erysiphe cichoracearum D.C.) in the cantaloupe variety Seminole showed resistance to be governed by two partially dominant genes. A greenhouse seedling screening was used to determine levels of resistance, with the degree of mildew infection being rated visually on a 5-category scale. The scale was adjusted to near-linearity, but the population distribu- tions were found to be anormal. A partitioning method based on non— parametric statistics was used to separate environmental from genetic components of variance. This method was especially effective in handling low levels of resistance where environmental effects caused an overlap of the two’parents. Genetic comparison of several sources of powdery mildew resistance revealed the existance of genes different from the previously reported Pm1 and sz. The U. S. Department of Agriculture Plant Introduction 124111 which was used as a source of resistance to race 2 of the mildew in Calif- ornia contained a single dominant gene giving excellent resistance in Michigan. This gene was designated Pm3. The two genes in Seminole were found to be different from Pml, sz, or Pm3 and were designated as Pm4 and Pms. Other genes giving good resistance were noted, but their de- signation was delayed pending further analysis of their relationship to the reported genes. Evidence was presented for the existance of genes giving very low levels of resistance which added to that of the major genes. The genes Pm1 and sz were found to act in an epistatic manner, with Pm1 giving good resistance in Michigan and sz giving no resis- 2 tance. Pm1 and Pm acting together gave increased resistance. Pre- liminary screening with mildew race 2 in California indicated that re- sistance was controlled by the epistatic interaction of Pm1 and sz. INHERITANCE OF RESISTANCE TO POWDERY MILDEW IN CUCUMIS MELO By Richard Roland Harwood A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Horticulture 1967 ACKNOWLEDGMENTS The author wishes to express his appreciation to Dr. Deran Markarian for his guidance and encouragement throughout the course of study, and to Dr. M. W. Adams, Dr. C. M. Harrison and to Dr. C. E. Peterson who have contributed in many ways to the value of his graduate program. My sincere thanks are extended to my family for their patience and encouragement. TABLE OF CONTENTS INTRODUCTION. . . . . . . . LITERATURE REVIEW . . . . . . . Controlled Pollinations . . . The Pathogen. . . . . . . Screening. . . . . . . . Genetics of Resistance . . . Genetic Material . . . . . METHODS . . . . . . . . . . Crosses . . . . . . . . Screening. . . . . . . . Evaluation of Mildew Infection. Adjustment of the Rating Scale. Uniformity of Screening . . . GENETIC ANALYSIS OF SEMINOLE RESISTANCE. GENETIC SURVEY OF RESISTANCE SOURCES. Test for Allelism with Pm and Pm SUMMARY . . . . . . . . . . LITERATURE CITED . . . . . . . iii "U D: 00 m \OOLQUJUJNNNH Table Table Table Table Table Table LIST OF TABLES Page Distribution of resistance ratings for check pOPUJ-ations. O O O O O O O O O O O O O O O 15 Distribution of resistance ratings for Seminole crosses . l7 Observed and computed means for the genotypes of the Seminole crosses . . . . . . . . . . . . . . 18 Distribution of resistance ratings for homozygous Seminole x Delicious 51 F3 populations. . . . . . . 23 Distribution of resistance ratings for crosses with California resistance sources. . . . . . . . . . 26 Distribution of resistance ratings for crosses with . non-California resistance sources. . . . . . . . . 27 iv Figure 1. Figure 2. Figure 3. LIST OF FIGURES Powdery mildew resistant muskmelons of the south- eastern U. S. . . . . . .. . . . . . . Muskmelon powdery mildew resistance derived from California sources. . . . . . . . . . . Populational distributions of the crosses of Delicious 51 x Seminole . . . . . . . . . . . . Page 20 INTRODUCTION A study was begun in 1964 at the Michigan Agricultural Experiment Station to determine the inheritance of resistance to powdery mildew (Erysiphe cichoracearum D.C.) of the cantaloupe variety Seminole. The early results of this study were reported by Harwood (8), Harwood and Markarian (7), and by Markarian and Harwood (16). Subsequent work brought about the refinement of technique and the verification of results. The scope of the study was then broadened to include the observation of representative genetic material from the known sources of mildew resistance.' The results reported here are not conclusive concerning all of the, genetic factors which influence powdery mildew resistance in Cucumis 221g. It is hoped they will serve as an outline of the methods and genetic materials for subsequent work designed to reach this end. The work is reported in four parts: The sources of genetic ' material used; the deve10pment of a resistance evaluation technique; the analysis of resistance in Seminole and the genetic evaluation of resis- tance from other known sources. LITERATURE REVIEW Controlled Pollinations The technique for hand pollination of muskmelons is one of the first problems which must be considered before starting a muskmelon genetic study. Muskmelons are rather difficult to work with, both from the standpoint of the induction of female flowers and the setting of fruit. A knowledge of the pattern of fruit set is an essential prerequisite to the manipulation of the plant. Rosa (24) described the basic pattern of fruit set and sug- gested a method of pruning to increase set. Wolf and Hartman (33) im- proved upon this technique. Whitaker and Pryor (29) reported some increase in set using growth regulators. The most recent analysis of muskmelon fruit set patterns was that of McGlasson and Pratt (18). Successful hand pollinations may range from 102 to 70%. The Pathogen A review of the morphology and host range of the pathogen, Erysiphe { .gichoracearum D.C. was done by Harwood (8). The features of fungal mor- \ \ \ 'phology and behavior which especially pertain to the current work are its I dissemination by wind-blown asexual conidia and its temperature sensitivity for multiplication. The races of mildew seem to be rather uniform in distribution. Race 2 or at least a race similar to it in pathogenicity appears across the en- tire southern United States. Response in Costa Rica has been reported as being similar (19). Mildew in Michigan and New York is similar to race 1 in pathogenicity, thus resistant material from any source gives resistance in these areas. Screening The basic screening technique used was patterned after a greenhouse technique developed by Pryor et. a1. (23) and was reported by Markarian and Harwood (16). Genetic§70f Resistance The first important commercial variety of muskmelons having resis— tance to powdery mildew was PMR 45, which was released in 1934 (13). Jagger et. a1. (12) reported its resistance as being due to a single dom- inant gene which they named Pml. In 1938 a new biotic form of powdery mildew was reported (11) which was known as race 2. Since that time several varieties having resistance to this new race have been released. A study of the resistance to race 2 by Bohn and Whitaker (3) has revealed the presence of a single dominant gene sz and several epistatic modifiers. AA study in Egypt (26) has indicated the presence of two genes in native Egyptian material, each of which gave good powdery mildew resistance. \ The relationship of these genes to those in American material is not known. K \ Genetic Material R The current study was originally concerned with the resistance of the ‘Florida variety Seminole (32) and the relationship of the genes responsible to Pm1 and sz. It soon became apparent that the inheritance of resistance in the U.S.D.A.-California material was more complex than supposed. The pedigrees of the material were traced (Figures 1 and 2) in an attempt to more objectively select material for evaluation in determing the genetics of resistance in the major sources. Figure l. A. Powdery mildew resistant muskmelons of the southeastern U.S. Pedigree of Florida varieties and others such as Delta Gold (Louisiana) and Edisto (South Carolina) having the same source of res stance. Resistance was derived from PI 124112 (Pm , Pm ) and probably from Louisiana 8-2 and 7-1 as well as some "minor" genes being derived from Smith's Perfect and others. Pedigree of the Florida variety Floridew and its source of resistance in PI 223637. POWDERY MILDEW RESISTANT MUSKMELONS OF THE SOUTHEASTERN U.S. an. «24:12 HEARTS or GOLD (GA. 29554) smTR's PERFECT 7 smTH's PERFECT / SANFORD 9 HONEY ROCK / GEORG'A 47 “95” LOUISIANA 0’2 ’ 7' I /’ / ’ -’ . DELTA GOLD EDISTO (LA. IDGO) FLORIGOLD RIO SWEET “962) SEMINOLE u “960) I I i FLORISUN P.I. 223637 HONEY DEW FLORIDEW “962) .M‘... -.a - '4 I 3}.“ Figure 2. Muskmelon powdery mildew resistance derived from California Sources. The known sources and their probable gene cantributions were California 525, Pm , PI 79376, Pml, Pm,.and PI 124111, Pm3. Varieties developed from this material at other stations include Howell Spartan at Michigan, Homegarden at Louisiana and Wescan and Perlita at Texas. MUSKMELON POWMRY MILDEW RESISTANCE [IRIVED FROA CALIFORNIA SOURCES RI. 79370 HALES IEST CALIFORNIA 525 PERSIAN PMR BO PIR 40 IROOUGS RESISTMT CANTALM mu ””7”. mum's PERFECT m a mason RI. Tum /PIR 0 all 7 M C HALE STERILE I In) /FIR a m. Tum warm" \I sun um scan 365:: new 30524 36425 n P3 n so P 7 In P O nan as Rlo COLD "st0 sum: /mn «o IKE STE”! 'ECCAR 0.8.01. 57'04 PERLITA cameo new“ AF TE R CONN IDOI _ g - The outline which provided the basic information on the California material was published by Bohn (4). This information is supplemented in the present report by the pedigrees of material derived from the Califor- nia material by the major muskmelon breeding centers. The first powdery mildew resistant variety was PMR 45 (13). This was followed in 1943 by the race 2 resistant PMR #5 (31), also referred to as U.S.D.A. #5. In 1946 the varieties PMR #6 and #7 were released (30). In 1949 a source of male sterility (mslmsl) was released (5) having the. same pedigree as #5, #6, and #7 and the same mildew resistance. In 1951 the variety Georgia 47 was released, having derived its resistance from a Plant Introduction line closely related to the one from which later Califor- nia material derived its resistance (28). There is some confusion on this point, since the source of resistance listed in the release was given as PI 29554, whereas the U.S.D.A. summary (27) lists the source as PI 124112. 'This variety has served as a source of resistance for many of the lines developed for the southeastern U.S. such as Edisto (9) in 1957 and Delta Gold and Seminole (32) in 1960. Varieties subsequently developed from the California material in- cluded Rio Gold (6) in Texas, Homegarden (10) in Mississippi, Wescan (12) and Perlita (20) in Texas. Other California melons included Male Sterile #2 (2) and Campo and Jacumba (1). The variety Howell Spartan has PMR 45 as a parent (17). The Varieties Floridew (15) and Florisun(14) have resis- tance derived from Louisiana lines which probably originated in Georgia 47. Information on the PI lines which were furnished by the U.S.D.A. plant collection center at Experiment, Georgia, is contained in a U.S.D.A. Cucumis melo summary (27). METHODS Crosses The methods used in making the crosses and in screening for resis- tance were essentially the same as reported for the earlier portion of the study (8). All plants used in making crosses were grown in the greenhouse in 8" clay pots. They were pruned to a single main runner which was trained on a stake to a height of three feet. Laterals were allowed to develop be- ginning at the 4th and 5th node. These were pruned, in turn, at the 3rd or 4th node to enhance growth of the female or the perfect flower at the first and second nodes of the laterals. Since the flowers at these nodes are the strongest and the most likely to set fruit after pollination (24), the train- ing and pruning of the plant was designed to maximize their growth and de- velopment. Crosses were made by emasculating the perfect flowers of the female parent the day before opening. The corollas were left intact in this pro- cess. Insect control in the greenhouse was good, but the corollas were held closed both before and after pollination by #1 Tip Tap paper clips as further precaution against pollen contamination. Whenever possible the crosses were planned so as to use either a gynoecious or a monoecious fe- ‘ male parent. The two major advantages to this approach are the elimination 'of emasculation and its source of pollen contamination, and the increased percentage of successful pollinations on the stronger female flowers pro- duced. The overall estimate of successful pollinations under greenhouse conditions in the summertime on this type of plant was in excess of 75%. This is well above results reported for field crosses and was an important factor in making the 500 crosses for this study. _ 9 _ _ 10 _ Screening The screening for resistance was essentially the same as used previously (7,8). The seedlings were grown in 3 ft. x 50 ft. green- house soil benches in the winter when house temperatures could be low- ered as desired. The seeds were planted directly into the soil in rows running across the bench, with 150 rows per bench and 20 seeds per row. The temperature was held at 75 F. until the primary leaves were 1/2 inch in diameter, then lowered to 60 F. for the remainder of the screening. Inoculation was achieved by breath-blowing of conidia from heavily in- fected source plants over the seedlings at the time the temperature was lowered. Ratings of the degree of infection were made three weeks after inoculation using the previously reported 5-category rating scale (Markarian and Harwood (16), Harwood (8)). In screenings conducted dur- ing the months of October and November when there was abundant sunshine, greenhouse air temperature of 60-65 F. seemed optimum for mildew develop- ment and at the same time limited seedling growth. In December and January screenings when there were few days of sunshine, it was nec- essary to maintain daytime air temperature between 70 and 73 F. in order to have satisfactory mildew development. This was attributed to the fact that under the influence of direct sunshine, leaf temperatures may aver- age 5-8 F. higher than air temperature when there is little air movement in the greenhouse. In cloudy weather the leaf temperature may be lower than ambiant air temperature because of radiation. Since the growth of mildew is dependent upon the microclimate of the leaf surface, factors which affect this microclimate become important in the resistance eval- uation process. _ 11 - Evaluation of Mildew Infection A major difficulty with evaluation of plant material for amount of fungal growth occurs with the absence of a satisfactory quantitative method for measuring the amount of mildew present. It would be possible, although tedious, to measure the surface area affected, but this would not account for differences in the type of fungal growth. Most pathologists and genet- icists have therefore resorted to visual evaluation of the degree of infec- tion (23). Some, such as Bohn and Whitaker (3) have used microsc0pic eval- uation to detect minor differences in degree and type. Two major difficul- ties are encountered when microscopic evaluation is attempted. First, the method is slow and requires detailed evaluation of each plant, which limits the number which can be handled. vSecondly, environmental variation may be sufficiently great to limit the effectiveness and need for detailed eval- uation. In adopting an evaluation method consideration should be given both to the magnitude of the genetic and environmental effects and to numbers of plants required. A third and important factor is the adaptation of the method to statistical analysis. A major difficulty with a rating scale whereby plants are assigned to categories by a visual evaluation is that the scale of the categories is not necessarily mathematically linear nor even continuous. This, of itself, precludes the use of means, standard de- viations, etc. in analysis of the results. When such a scale is used,the categories are generally placed into more or less arbitrary groupings and labeled "resistant", "moderately resistant", and "susceptible." Although this method is useful in many cases, it has obvious statistical weaknesses. _ 12 _ Adjustment of the Rating Scale In a standard type of evaluation of a metrical character, for example the height of plants in a homogeneous pepulation, measurements of the height of each plant are made. The scale used is linear and continuous by nature of the measuring device used. If the individual heights are plotted on a frequency diagram it could be expected that the variation would be normally distributed about a mean value. If this same population of plants were rated according to a short, medium, high type of classification in such a manner that the rating procedure was entirely objective, the rating cate- gories could be adjusted to a continuous and linear scale by adjusting their values so that variation was distributed normally. This was done in deter- mining the criteria for the 0-4 rating scale used on the melon seedlings. Several homogeneous populations totaling about 5,000 individuals were rated several times using different criteria for establishing the categories. Where the populations showed abnormal distributions, categories were added, deleted or altered in such a way as to adjust the distributions to normality or as near as possible. Importance was given to the identification of distinctive criteria upon which to base the various categories in order to achieve a maximum of objectivity in the ratings. The problem of locating homogeneous populations which fell in the center of the rating scale arose in the early screenings. As a result, heavy emphasis was placed on the parental populations which could not be ex- pected to show true normal distributions in this case, because they had mean values close to the end categories of 0 and 4. This indicated that there was an accumulation of individuals at the end points of the scale, which “In A‘.-‘- It“ - 13 _ might have fallen further out if the scale had been extended. Environ- mental variation was thus not a random effect, but was determined par- tially by the limitation of the scale. These parental distributions, then, could only be used as indicators of approximate linearity of scale because of the large proportion of their numbers falling in the open- interval categories. A With the screening of large F3 populations of the Seminole x Delicious 51 cross, several F3 families were seen which were probably homozygous and which did fall in the central region of the rating scale (Table 4). Homozygosity was determined by selecting those populations 1, P2 and their F1. A test for symmetry and kurtosis in this population, following the method re- whose variance was less than the average of P ported in Snedecor (25) shows a value of g1 - 1.17 for the moment of the third power of deviation about the mean. This gives a t value of 5.086, including highly significant asymmetry in the distribution. Likewise the g2 value of 3.01 with its t of 6.689 shows highly significant kurtosis. If these deviations were caused by abnormalities in the scale, cor- rections should be possible by adjustment of the rating values. A close analysis of the F1 and F3 populations 28, 31, 33, 40 and 45 shows them all to be skewed to the lower values. Most of them also show positive kurtosis. With the means of these populations falling from values of 0.83 to 3.19, the asymmetry covers the whole range of scale. A correction of scale for one population would increase asymmetry in others. It would thus appear that this asymmetry is not solely a function of abnormality of scale. Since these populations contain different genetic bases for mildew resistance it - 14 _ can be assumed that the environmental effects are different for different genes. The distribution of these F3 populations does indicate, however, that the scale is sufficiently linear to permit the use of means and stan- dard deviations even though the distributions are not normal. The genetic analysis must therefore be based on non-parametric methods which make use of order statistics. Uniformity of Screening The uniformity of screening is shown by the representative data of Table 1. The p0pulations of Delicious 51 were single row populations which were planted every 10 rows in the benches as a check for mildew infection. Each population was comprised of approximately 20 plants. These data were taken from the single benches having the greatest variability. As can be seen, the coefficients of variation for these small groups are 152 and 11% for the 1965 and 1966 screenings respectively. In the segregating pap- ulations of the different crosses of Gynoecious (USDA 5 x Seminole) each population totaled about 200 individuals, giving a much smaller standard deviation and coefficient of variation. These data attest to the uniformity of the screening for both homo- zygous and segregating populations. This uniformity is seen to hold from year to year as well as within a single screening. On the basis of these results the data were totaled over all screenings for the analysis of Seminole resistance. _ 15 _ Table 1 Frequency Distribution of Resistance Ratings (Number of plants per category) Population Mean Variance Average 0 1 2 3 4 of means Del. 51 (1965) 4 14 3.78 0.160 " 7 12 3.63 0.145 " 5 11 3.68 0.214 3.62 i_.56 " 2 16 3.88 0.980 cv - .154 " 3 10 7 3.20 0.460 " 4 13 3.76 0.180 " 9 11 3.55 0.248 " 9 8 3.47 0.265 " Total 3 50 92 3.61 0.280 Del. 51 (Nov.'66 l 13 4 3.16 0.194 " Bench 6) 1 16 3 3.10 0.248 " 1 l 18 1 2.91 0.276 " 1 10 6 3.29 0.325 3.35 i .37 " 12 6 3.33 0.222 cv - .110 " 1 8 8 3.41 0.268 " l 5 14 3.65 0.328 " 23 4.0 0 " Total 1 6 82 65 3.37 .403 3.37 i_.63 Gyn. (USDAxSEM) 123 27 7 37 1.01 2.495 " 117 38 1 7 41 1.10 2.514 " 121 59 2 2 57 1.23 2.618 1.17 1; .109 " 114 43 0 1 54 1.13 2.418 cv - .093 " 111 39 3 6 55 1.32 2.855 " 87 41 l 6 38 1.23 2.572 GENETIC ANALYSIS OF SEMINOLE RESISTANCE In an early analysis of the Seminole x Delicious 51 crosses (8), an empirical method was used to partition the variance of the populations using frequency distributions. This method was similar to that proposed by Powers (22) in dealing with what he called Type I data. No consider- ation was given to normality of the distributions or to the means or stan- dard deviation. The method used here is somewhat similar to the Power's Type III method of partitioning variance where classes are grouped according to the means and variances. The mean values of the Seminole x Delicious 51 and Seminole x Gynoecious crosses (Table 2) were used to determine mean values for each genotype in a two-gene model (Table 3). The Delicious 51 x Seminole F3#10 had been previously selected as being homozygous for one of the genes for resistance in Seminole. It was, in turn, crossed to the susceptible Gynoecious parent, and the F2 and backcross populations produced. The pro- posed genotypes and their observed mean values are listed in Table 3. These observed values were then used to give mean values for the genotypes in the Delicious 51 x Seminole F backcrosses to both parents. In the backcross l to Seminole, for instance, only the value for the genotype aaBb was needed. Since an observed value was available for the mean of the backcross popula- tion, the mean of this genotype could be determined. The same was done for the backcross to Delicious 51. With the means for these genotypes determined, an F2 was constructed as a test of the hypothesis. This was done by multiplying the mean for each genotype by its expected F2 frequency, then totaling the values to arrive at _ 16 _ I .w-m» - _ 17 _ Table 2 Frequency Distribution of Resistance Ratings for Seminole Crosses (Number of plants per category) Population Rating Mean Variance 0 1 2 3 4 Seminole 158 25 15 0 0 .277 .354 Delicious 51 0 l 6 82 65 3.37 .403 De1.xSem. F1 27 41 81 l 0 1.37 .607 Del.xSem. F2 332 166 388 . 184 111 1.65 1.296 (Del.xSem.) Del.B1 12 30 95 58 25 2.13 1.026 (Del.xSem.) Sem. B1 154 12 72 19 0 .86 1.150 De1.x Sem. F3(Total) 960 2130 2979 1557 1105 1.97 1.376 Del. x Sem. F3 10 6 36 69 8 0 1.66 .459 Gyn.x F3 10 F1 1 46 53 39 0 1.93 .630 Gyn.x F3 10 F2 2 22 33 106 10 2.58 .682 0 31 14 103 3 2.52 .700 Gyn. (Gyn.xF3 10)B1 _ 13 _ Table 3 Observed and Computed Means for the Genotypes of the Seminole Crosses Population Genotype Mean Del. x Sem. F3 10 AA 1.68 Gyn. x F3 10 F1 A3 1.93 Gyn. x F3 10 F2 2.58 AA 1.68 Aa 1.93 aa 3.37 Gyn. (Gyn. x F310) 2.53 Aa 1.93 aa 3.37 (De1.x Sem.) Sem. .86 AABB .28 AaBB 1.lO* AABb - .90* AaBb 1.37 (De1.x Sem.) Del. 2.13 AaBb 1.37 Aabb 1.93 aaBb 2.00* aabb 3.37 Del.x Sem. F2 1.65 AABB .28 AaBB 1.10 AABb .90 AAbb 1.68 AaBb 1.37 Aabb 2.31 aaBB 1.99 aaBb 2.00 aabb 3.37 *Computed values .. _ 19 _ an overall mean. The results are as follows: Calculated mean Observed mean (Gyn x F310) F2 2.23 2.58 i_.]09 (Del x Sem) Sem .91 .86 i..109 (Del x Sem) Del 2.17 2.13 i;.109 (Del x Sem) F2 1.59 1.65 i .109 Any deviations from the observed are well within the limits of error. The bimodal aspect of the F2 distribution lends further sup- port to the two—gene hypothesis. The model shows gene A to have 70% dominance and gene B to be somewhat more completely dominant. The environmental effect computed from the average of the variances of the F1 and both parents was 0.454. The observed genetic variance in the F2 was thus 0.942. The calculated genetic variance of the F2 as determined from the means of the component genotypes by the Powers method, was 0.499. This discrepancy will be discussed later. The heritability in the F2 as calculated from the Mather model was 67%. An F3 p0pulation of 8731 individuals was screened. This came from 62 F2 plants which were rated for mildew both in the seedling and the mature plant stage in the greenhouse. The totals are given in Table 2 and plotted in Figure 3. The 62 F2 plants do not represent a completely un- biased sample of the F2 population, however, as some incompatibility was present in the material so that some plants were difficult to self or pro- duced few viable seeds.“ Consequently, more of the plants of these particular phenotypes were chosen so that all of the genetic material would be repre- sented in the F3. The individual F3 families which showed less variance than “u"d. _ 20 _ Figure 3. Populational distributions of the crosses of Delicious 51 x Seminole. - “nus “a”- b' A. Resistant parent Seminole, susceptible parent Delicious 51, 3. and their F1, Delicious 51 x Seminole. B. Delicious 51 x Seminole F and F3. The F2 shows the Z-gene bimodal distribution. C. Backcrosses of the F to the resistant and to the susceptible parent. The backcross to the resistant i , parent shows the bimodal effect of two genes but the backcross to the susceptible merely shows a skewness toward the susceptible. DMM~ -~.-< PERCENT OF THE POPULATION (DEL X SEN, PI 30 20 PERCENT OF THE POPULATION NILDEW RATING (DEL X DENIES (DEL X CEIIFZ CO4 NILDEW RATING (DEL X CENIX DEL (DEL X CENIX SEN PERCENT OF THE POPULATION O -< N U .1 NILDEW RATING _ 22 _ the mean of the parents and F were assumed to be homozygous. These are 1 listed in Table 4, beginning with the most resistant. The mean values COVor the entire rating scale. As has previously been demonstrated (Table 1), the expected standard deviation in mean values of large populations is about i 0.109. Since there are so many different values for these F3 pOp- ulations, it would seem likely that there are additional genetic factors involved. There is, in fact, a good deal of evidence to support this hy- pothesis. It is the opinion of Dr. W. Bohn* of the U.S.D.A. laboratory at La Jolla, California, that there are numerous "minor" genes for mildew re- sistance scattered throughout the Cucumis melo species. Dr. Henry Munger* at Cornell has noted that crosses of Delicious 51 to a resistant source seem to have slightly higher resistance than do crosses of Iroquois x resistant. This same effect was noticed in field screenings in East Lansing in 1965, where the Delicious 51 x PMR 45 F1 had no mildew at all, while the Iroquois x PMR 45 F1 was rated at 1/2, having a few isolated mil- dew colonies late in the season. The Gynoecious x PMR 45 F1 also was rated at 1/2. These differences were not evident in a subsequent greenhouse screening, although the screening technique had not been refined. Iroquois and Delicious 51 by themselves have shown no difference in re- sistance in the field, but in the greenhouse Delicious 51 was rated at 3.5 while Iroquois, which was adjacent to it in the screening, was rated at 3.8. In the Delicious 51 x Seminole F3 several of the populations were rated significantly lower in resistance than was Delicious 51 (Table 4). This *Personal Communication. _ 23 _ Table 4 Frequency Distribution of Resistance Ratings for Homozygous Seminole x Delicious 51 F P0pulations (Number of plants per category) Population 0 1 2 3 4 Mean FZRating Variance F3 # 47 147 34 2 .21 0/0* .184 30 112 91 14 .55 0/0 .425 56 57 65 16 .70 1/0 .441 50 38 144 3 .81 0/0 .187 45 34 136 5 .83 0/0 .195 31 13 166 31 7 1.15 1/0 .312 36 114 79 1.40 2/0 .242 40 4 106 84 1 1.42 3/0 .294 20 51 153 1.75 2/0 .188 59 12 11 167 _ 9 1 1.88 0/0 .345 34 1 5 197” 6 1 2.0 1/0 .090 55 5 39 20 2.23 3/1 .336 33 8 6 178 21 2.99 3/3 .276 28 1 3‘ 16 156 30 3.02 4/3 .325 46 6 166 32 3.12 3/2 .245 21 60 130 3.57 4/4 .222 .\ 38 4 53 110 3.60 4/3 .281 54 4 5 37 167 13.72 4/3 .360 53 1 1 36 154 3.79 4/3 .209 61 1 9 179 3.94 4/2 .066 52 7 39 129 35 4 1.95 1/0 .554 *Seedling rating/Mature plant rating _ 24 _ would seem to indicate segregation of a more susceptible type. The fact that only one population was seen having as high a resistance as Seminole indicates an additional factor or factors in its resistance also. If only the two genes were responsible for its high resistance, four of the 62 F2 plants could be expected to give Seminole type resistance in their F3 pro— geny. These additional minor genes, one each in Seminole and in Delicious 51, would also account for the many different levels of resistance in the homozygous F3 populations. There were, in addition to the homozygous F3 populations, several such as #52 (Table 4) whose variance was slightly higher than the homozy- gous, but which was not as high as for a p0pulation with segregation for a major gene. These p0pu1ations were probably homozygous for gene B but contained a segregating minor gene. One additional indicator of increased genetic variance was seen in the previously mentioned discrepancy between observed and calculated genetic variance in the F2. The calculated variance was based on a two-gene model, while that observed for the F2 was significantly higher because of the added effect of the minor genes in both Seminole and Delicious 51. ARI—5. GENETIC SURVEY OF RESISTANCE SOURCES In an attempt to evaluate the available genetic material having mil- dew resistance, a series of crosses was made between the resistant sources in question and the California sources. The plan was to look for suscep- tible types in the test-crosses in a test for allelism. Obtaining homo- zygous sources of resistance proved to be the greatest difficulty. The sources were checked for homozygosity by making their Fl 3 with Gynoecious and then testing these for uniformity. Of the 35 PI lines having resis- tance, only 5 had plants homozygous for resistance. Resistant Fls from the segregating lines were selfed to provide homozygous individuals for crossing at a later date. The pedigree of the California mildew program as adapted from Bohn (4) is presented in Figure 2. Some of the varieties produced from different breeding have been included in the diagram. Resistance of PMR 45 to race 1 of mildew has been reported (12) as being a single dominant gene, Pml. This has been verified in the current study (Table 5). The PMR 45 backcross to Gynoecious segregatéd 1:1, indicating the presence of a single dominant gene. Bohn (3) re- ported inheritance to race 2 of the mildew in P2 - P9 as being con- trolled by a single dominant gene and two or three epistatic modifiers. This single dominant gene was named Pm2. Data from the current study indicate that this is probably partially correct. - 25 - _ 26 _ Table 5 Frequency Distribution of Resistance Ratings of California Type Resistance (Number of plants per category) . , 2 Population 0 l 2 3 4 Ratio Chi P USDA 5 l8 Gynoecious x USDAS F1. 91 1 Gyn.(Gyn.xUSDA5)B1 95 3 8 72 18 1:1 0 1.0 PMR 45 20 Gyn.(Gyn.xPMR 45)B1 33 1 42 1:1 1.316 .2 Gynoecious 2 8 3 Gyn. x Wescan F 91 Gyn. x Wescan F2 116 2 46 3:1 .813 .3 Gyn.(PMR 45 x Wescan) 152 Gyn.(USDAS x Wescan) 117 Gyn.(Gyn.x Wescan) 54 2 2 55 2 1:1 .039 .8 Gyn. x Perlita F1 120 Gyn. x Perlita F2* 117 l 22 3 Gyn.(PMR 45 x Perlita) 135 Gyn.(USDAS x Perlita) 139 Gyn.(Gyn.x Perlita)Bl* 95 7 55 24 1:1 2.92 .07 Gyn. x PI 124111 F 111 12 Gyn. x PI 124111 F2* 57 .122 28 ll Gyn.(USDAS x PI 124111) 22 8 3 5 3:1 .316 .5 Gyn.(Gyn.x P1124111) 102 48 1 149 1 1:1 0 1.0 *These populations received poor screening because of low incidence of mildew. - 27 _ Table 6 Frequency Distribution of Resistance Ratings of Sources other than California Material (Number of plants per category) Population 0 l 2 3 4 Ratio Chi2 P PMR 45 x Seminole Fl 46 USDAS x Seminole F1 45 Gyn(USDAS x Sem.) 673 247 17 29 282 Gyn(PMR 45 x F #10) 71 5 7 63 1 Gyn(USDAS x F3 10) 92 9 12 48 1 Gyn. x PI 234607 F1 166 Gyn. x PI 234607 F2 127 26 3 Gyn(PMR 45xPI 234607) 142 Gyn(USDA5xPI 234607) 146 Gyn(Gyn.xPI 234607) 244 66 10 117 6 3:1 5.95 .02 Gyn. x PI 236355 F1 96 Gyn. x PI 236355 F 138 16 4 10 15:1 .875 .3 Gyn(PMR 45xPI 236355) 65 8 3 Gyn(USDA5xPI 236355) 153 Gyn(Gyn.xPI 236355) 84 24 26 4 3:1 .785 .3 Gyn. x Bellgarde F1 77 Gyn. x Bellgarde F2 34 39 9 12 3:1 .354 .5 Gyn(PMR 45xBe11.) 82 20 2 9 25 3:1 .196 .6 Gyn(USDA5xBell.) 89 34 5 19 20 3:1 .169 .6 Gyn(Gyn.xBe11.) 7 53 1 53 12 1:1 .286 .6 Gyn. x PI 179901 F 100 Gyn(PMR 45 x PI 173901) 117 Gyn(USDAS x PI 179901) 128 Gyn(Gyn.x PI 179901) 98 44 5 24 3:1 .02 5.86 _ 28 _ In evaluations in Michigan resistant material from any other area has been resistant. The race of mildew present is probably race 1 or at least a race similar to it in its pathogenicity. Test for Allelism With Pm1 and Pm2 In the study of the genetics of resistance of different sources the test for allelism with Pm1 and sz were conducted by making the test- 1 2 . ‘ . cross Susc(Pm or Pm source x re31stant source be1ng evaluated). The variety PMR 45 was used as a source for Pm1 and PMR 5 as a source for sz. The results of these tests conducted with the varieties Wescan and Per- lita and the PI lines 234607 and 179901, all having resistance to race 2, predictably showed no segregation when tested against PMR 5 (Table 5 and 6), indicating the presence of sz. They also showed no segregation with PMR 45. This indicated the presence of Pm1 or of a gene so closely 1 that no crossovers were seen in 1000 testcross individuals linked to Pm with these varieties. Since Wescan and Perlita derived their resistance indirectly from PMR 5 it can be assumed that this variety also contains Pml. On this basis, the cross Gyn(Gyn x USDA 5) was screened here in East Lansing, giving a nearly perfect 1:1 ratio as shown in Table 5. This would seem to indicate the presence of a single dominant gene. The 95 plants which were completely clean of mildew were removed from the screen- ing bench and planted in peat pots. They were then hardened for two weeks for shipment to California. During this period about half of them showed signs of mildew growth. Several of the plants died in shipment, but those remaining, when tested in California to race 2 of the mildew, _ 29 - segregated 24 susceptible - 21 resistant*. The susceptibles were said to have PMR 45 type resistance and the resistant ones USDA 5 type. It can thus be concluded that USDA 5 resistance to race 2 is governed by two genes, Pm1 and another dominant gene hypostatic to Pml. This second gene gave no resistance by itself, at least in Michigan, but in combina- tion with Pml gave increased resistance to race 1 and a moderate level of resistance to race 2. A second, but unlikely hypothesis is that sz acts alone but gives no resistance to the Michigan race of mildew. In comparing all other material with PMR 45 and USDA 5, then, the test was against the gene Pm1 only. The second gene in USDA 5 could not be distinguished in a cross to PMR 45 by screening with race 1. The Texas varieties Wescan and Perlita, having been derived from PMR 6, showed no segregation when tested against PMR 45 and both showed 1:1 segregation in the Gynoecious backcross. They are both resistant to race 2, so it can be concluded that they contain Pm1 and its epistatic modifier. The PI 124111, when tested against USDA 5, segregated 3:1, in- dicating the presence of two separate genes in the F1' These genes are Pm1 and a different independent gene in 124111. This gene by itself gives a level of resistance in Michigan equal to that of any other source, and in addition, gives at least some resistance to race 2 (27). *The California screening was done by Mr. Joseph Principe in the U.S. Department of Agriculture laboratory of Dr. G. W. Bohn at La Jolla. _ 30 - The variety Campo was tested also, buC unfortunately the screening as indicated by the Delicious 51 check row was not severe enough to be re- liable. The P3 line has been reported by Bohn (3) as having the best resistance to race 2 of all of the California lines. It seems likely that this line contains Pml, its epistatic modifier and the gene from P1 124111. Seminole, as shown from the pedigree in Figure 1, derived its re- sistance from Georgia 47. The screening results of the cross Gyn(USDA 5 x Sem) were presented in detail in Table l and in summary in Table 6. On the basis of a two-gene hypothesis for Seminole and Pm1 in USDA 5, the follow- ing genotypes could be expected in equal proportions in the testcross pro- geny: Pml pml 'Aa Bb l l Pml pml aa Bb Pml pml Aa bb Pm pm aa bb pmi pmi Aa Bb pm pm aa Bb 1 l pm1 pm1 Aa bb pm pm aabb Since the first four contain Pml, their expected mean is zero. The second four are exactly the same as expected in the cross Delicious 51 (Del x Sem). The observed mean for this cross was 2.13. This gives an ex- pected mean for the above testcross p0pu1ation of 1.07. The observed value (Table l) was 1.17 1;.109. We can thus conclude that genes A and B of Seminole are not linked to Pml. The Del 51 x Sem F3 #10 when crossed to USDA 5 and the F1 testcrossed to Gynoecious, would give the following genotypes in equal preportions: Pm: pm1 Aa Pm pml aa pmi pmi Aa pm pm aa - 31 _ The mean of the Pml types, as before, is zero (as seen from the PMR 45 x Gyn Fl - Table 5), and the mean of the second two types is the same as that of the cross Gyn(Gyn x F 10) which is 2.52. The expected 3 mean for this p0pulation thus is 1.26. The observed mean is 1.12 and for the similar PMR 45 testcross, 1.44, both of which are well within the limits of variability of the calculated value. This further indicates the non-allelic relationship of gene A and Pml. Seminole is reported to have PI 124112 in its pedigree. This source is probably of a different geno- type from P1 124111, since Seminole has no one major gene for mildew re- sistance. Other material tested showed additional genes for resistance (Table 6). PI 234607 from South Africa showed no segregation with PMR 45 or USDA 5, but showed evidence of having two genes by the 3:1 ratio in its backcross. The resistant plants for all major genes fall in categories 0 and 1 and for the susceptible allele, in 2 and 4. PI 236355 from England showed a similar pattern, with some evidence that one of its genes might be linked to Pml. The second gene in both cases may or may not be similar to that in P1 124111. The Bellgarde crosses showed the presence of a single gene not linked to Pml. The PI 179901 showed linkage or allelism to Pml, but its backcross indicates the presence of at least two dominant genes. S UMMARY The distribution of environmental variance of mildew infection in Cucumis melo does not follow that expected for a normal distribution. The statistics used in the analysis of mildew ratings must therefore be adapted to a non—parametric distribution. The screening of a wide range of genetic types has shown variation in the seedling response to mildew infection. Some types for instance are less prone to cotyledonary infection than others, even though the mature plant or its progeny may be more susceptible. Minor adjustments in the rating scale may be necessary for evaluation of these types. A study of the inheritance of resistance to powdery mildew in the variety Seminole has shown resistance to be governed by two "major" and probably one "minor" gene. Gene "A" was shown to have 70% dominance and gene B somewhat more complete dominance. Their effects are partially 4 additive. The symbols Pm4Pm are prOposed for the dominant alleles of gene A and PmSPmS for those of gene B. The race 2 resistant variety PMR 5 and those derived from it were 2 shown to contain the gene Pml. The gene Pm gave a high level of resis- tance in combination with Pml. sz appears to act in an epistatic manner 1 with Pm to give resistance to race 2. A single dominant gene giving a high level of resistance was found in the PI 124111. This gene was shown to be not linked to either Pm1 or 2. The symbols Pm3Pm3 are proposed to designate its dominant allele. Pm Other genes giving resistance in Michigan were identified but their naming is to be delayed pending a more complete analysis of their relation- - 32 _ _ 33 _ ship to Pm3, Pm4 and Pms. From the evaluation of the available genetic material giving resis- tance to powdery mildew, conclusions may be drawn concerning the suitabil- ity of the material as a source of resistance in a Michigan breeding pro— gram. It has been found (16) that low levels of resistance as determined by a greenhouse seedling evaluation give effective "commercial" resistance 5 would thus be suitable. It is more convenient and desirable, however, to use the more potent genes. Pml, for in the field. The genes Pm4 and Pm instance, gives a much cleaner segregation in backcross progeny and is much easier to select as well as giving somewhat better field resistance. sz is rather difficult to handle. Very close evaluation is required for its detection in the presence of Pml. Since there is some merit to breed- ing for resistance to race 2 as a precautionary measure, the selection of 3 Pm might be wise. This gene should be rechecked by itself for resistance to race 2 before it is used. 10. ll. 12. 13. LITERATURE CITED Bohn, G. W., G. N. Davis, R. E. Foster and T. W. Whitaker. 1965. Campo and Jacumba, new cantaloupe varieties for the southwest. C81. Ago JUly. 8_lOo and J. A. Principe. 1964a. Second male-sterility gene in the muskmelon. Jour. Hered. .22: 211-215. and T. W. Whitaker. 1964b. Genetics of resistance to powdery mildew race 2 in muskmelon. Phytopath. 543 587-591. . 1961. Inheritance and origin of nectarless muskmelon. Jour. Hered. 2: 233-237. and T. W. Whitaker. 1949. A gene for male sterility in the muskmelon. A. S. H. S. 3: 309-314. Godfrey, G. H. 1953. Rio Gold, a new disease resistant cantaloupe. Tex. Ag. Exp. Rep. 1613: 1—3. Harwood, R. R. and D. Markarian. 1966. Genetics of resistance to powdery mildew in the Michigan cantaloupe breeding program. Proc. 17th Int. Hort. Congress. 454-455. . 1966. Inheritance of resistance to powdery mildew in the cantaloupe variety Seminole. M.S.U. Master's Thesis. 1966. Hughes, M. B. 1957. Edisto Cantaloupe, a new variety for the South. S. C. Circ. 111. Ivanoff, S. S. 1957. The homegarden cantaloupe, a variety with combined resistance to downy mildew (Psendoperonospera cubensis). powdery mildew (Egysiphe cichoracearum), and aphids (Aphis gossypii). Phytopath. 423 552—556. Jagger, J. C., T. W. Whitaker and D. R. Porter. 1938a. A new biotic form of powdery mildew on muskmelons in the Imperial valley of California. Plant Dis. Reptr. 22; 275-276. . 1938b. Inheritance in Cucumis melo of resistance to powdery mildew (Ervsiphe cichoracearum). PhytOpath. g§5 671. and G. W. Scott. 1937. Development of powdery mildew resistant cantaloupe no. 45. U. S. D. A. Circ. 441: 1-5. 14. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. Jamison, F. S., J. Montelaro and J. D. Norton. 19628. Two new cantaloupe varieties for Florida growers: Florigold and Florisum. Univ. Fla. Circ. S-139. - 1962b. Floridew. Univ. Fla. Circ. S-138. Markarian, D. and R. Harwood. 1967. The inheritance of powdery mildew resistance in Cucumis melo L. 1. Identification of greenhouse conditions necessary for epiphytosis and the correlation of apparent genetic resistance to field conditions. M. S. U. Quart. Bull. 49-37: 404-411. and J. D. Downes. 1966. Muskmelon F hybrid Howell 1 Spartan. M. S. U. Ag. Exp. Sta. Mimeo. Mc Glasson, W. B. and H. K. Pratt. 1963. Fruitset patterns and fruit growth in cantaloupe (Cucumis melo L., var. reticulatis NaUdo)o A. 8. Ho So .8__3_: 485-5050 Meneses, J. A. and W. J. Wittbank. 1962. Varietal yield trials of melon (C, melo) at the Socorrito experimental farm. Bot. Teh. Minist. Agric. Costa Rica. 31, Patterson, R. E. 1964. Perlita cantaloupe. Tex. A. & M. Ag. EXP. Sta. L-6410 1963. Wescan, a cantaloupe adapted to Texas conditions. Tex. A. & M. Ag. Exp. Sta. L-588. Powers, L. 1963. The partitioning method of genetic analysis and some aspects of its application to plant breeding. In Statistical Genetics and Plant Breeding. Nat. Acad. Sci. Nat. Res. Council. Washington. 280-318. Pryor, D. E., T. W. Whitaker and S. N. Davis. 1946. The develOp- ment of powdery mildew resistant cantaloupe. A. S. H. S. 415 347-356. Rosa, J. T. 1924. Fruiting habit and pollination of cantaloupe. A. S. H. S. .21: 51-57. Snedecor, G. W. 1950. Statistical Methods. Iowa State Press. Ames, Iowa. 174-177. Stino, K. R., W. A. Warid and M. A. Abdelfattah. 1961. Inheritance of resistance to powdery mildew in melon. Cairo Univ. Ag. Bull. 209. 27. 28. 29. 30. 31. 32. 33. U. S. D. A. 1965. A summary of reports on the resistance of plant introductions to diseases, insects and nematodes: Cucumis melo. Mimeo. Van Haltern, F. 1951. Georgia 47, mildew 0. Southern Seedsman. .14 (ll): 31, 45. Whitaker, T. W. and D. E. Pryor. 1946. Effect of plant growth regulators on the set of fruit from hand-pollinated flowers in Cucumis melo L. A. S. H. S. 48; 417—422. and S. N. Davis. 1946. Cantaloupe 6 and 7. Southern Seedsman. 2_(2): 14, 54. , D. E. Pryor and S. N. Davis. 1943. Cantaloupe #5. Western Grower and Shipper. l§_(9): 8, 22-23. Whitner, B. F. 1960. Seminole, a high—yielding good quality, downy and powdery mildew-resistant cantaloupe. Fla. Ag. Exp. Sta. Circ. S-122. Wolf, E. A. and J. D. Hartman. 1942. Plant and Fruit pruning as a means of increasing fruit set in muskmelon breeding. A. S. H. S. .40: 415—420.