T.‘ .1:J. «3.1.1.1.... run. .31, 3?}. 26:41....:_..\..v£v.4 v , . ‘ . . . . A . . A. . . , . ‘ . . 4 . . . , , . H . . .. .3 , A; . .. , y . V . . v. ‘I. Mr ‘ . . ‘ ‘2 . . . ‘ v . . . ‘ . . , . . . . , Ar ‘ . _. ‘ . 1 . I‘ A ‘ . . .‘ J ‘ r . . ‘ ‘ 1 . . . ‘ .. . : ‘ .1 -. . r... . § le..s.n..x.. .v‘ w . \.y..,. .. . .3: 3... 9. ‘|...,.-..!, .$ vt .- . n: ,1 , , SIS " : THE LESS" my This is to certify that the dissertation entitled THE INHERITANCE OF SEVERAL CHARACTERS IN PAK-CHOI AND PEPPER presented by Kenneth R. McCammon has been accepted towards fulfillment of the requirements for PhD degree in Hamming? WM Mow-M / Major professor Date @VE 4??? MSU is an Affirmative Action/Equal Opportunity Institution 0-12771 ‘IVISSIUJ RETURNING MATERIALS: Place in book drop to LIBRARIES remove this checkout from .—._—- your record. FINES will be charged if book is returned after the date stamped below. " a J “J “g 3"; I "D/ ' THE INHERITANCE OF SEVERAL CHARACTERS IN PAK-CHOI AND PEPPER BY Kenneth R. McCammon A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Horticulture 1983 3| ’41-55?. ABSTRACT THE INHERITANCE OF SEVERAL CHARACTERS IN PAK-CHOI AND PEPPER by Kenneth R. McCammon Studies were conducted on Pak-choi (Brassica rapa L., Chinensis group), to develop a sampling technique for the determination of petiole width, thickness and angle of adaxial curvature, and to investigate the inheritance of these characters. A technique was developed, using the means of the fifth, seventh and ninth petioles, measured at a distance 9 cm from the base of each petiole. Inheritance studies were conducted in the field over two seasons. Petiole width was determined to be simply inherited, with narrow petioles being dominant over wide. Petiole thickness was found to be controlled by two major genes, with thick petioles partially dominant over thin petioles. Dominance gene effects appeared most important, although additive gene effects were also significant. A 9:7 recessive suppressor model best explained the observed ratios. Kenneth R. McCammon Adaxial curvature was also determined to be controlled by two major loci, with curved petioles partially dominant over flat petioles. Both dominance and additive gene effects were involved in the expression of this trait. In addition, significant additive X additive, additive X dominance and dominance X dominance epistatic gene effects were noted. A 9:7 recessive suppressor model best explained the observed ratios. Studies on pepper were conducted to determine the inheritance of the "umbrella" branching habit, a unique habit with possible use in mechanical harvesting. F2 data from crosses between dwarf, indeterminate and umbrella lines showed a recessive inheritance of the dwarf and umbrella pheno- types. Four genes were involved in the expression of the phenotypes studied. The dwarf habit was found to be controlled by two linked recessive genes, Qt, a gene conditioning determinate growth, and £3, a gene controlling the clustered bearing habit. The umbrella habit was controlled by three recessive genes, g3, f3 and £3. The gene gt controls the number of axillary branches produced by the plant. 22 and SE condition indeterminate growth. In the homozygous or heterozygous con- ditions, SE is epistatic to the expression of gt, while 23 is epistatic to the expression of 93. A fourth gene, S3, Kenneth R. McCanmon is a dominant suppressor which acts to suppress the epistatic action of the SE gene. This dissertation is dedicated to my wife, Marlise, and daughter, Mollie... ii ACKNOWLEDGEMENTS The author wishes to express his appreciation to Dr. Shigemi Honma for his encouragement in this research and helpful assistance in the preparation of this manuscript, and for providing me the opportunity to pursue my graduate training under his direction. Sincere gratitude is also expressed to Dr. Ardeshir Ghaderi, Dr. Russell Freed, Dr. Irvin Widders and Dr. Hugh Price for their comments and criticisms in reviewing this manuscript. Final appreciation is extended towards my wife, Marlise, and daughter, Mollie, without whose sacrifices and support, this research would not have been completed. iii TABLE OF LIST OF TABLES . . . . . . . . LIST OF FIGURES . . . . . . . . CHAPTER I: THE INHERITANCE OF IN PAK-CHOI INTRODUCTION . . . . . . . REVIEW OF LITERATURE . . . MATERIALS AND METHODS . . Sampling Technique . Greenhouse Study Field Study . . Inheritance Studies Inheritance Studies RESULTS AND DISCUSSION . . Sampling Technique . Greenhouse Study Field Study . . Inheritance Studies - Petiole Width - 1981. Petiole Width - 1982. Petiole Thickness - l9 Petiole Thickness - 19 iv CONTENTS PETIOLE CHARACTERS 1981 . . . . . . . . . . 8 1982 . . . . . . . . . . 12 . . . . . . . . . . l6 . . . . . . . . . . 18 Pooled Analysis . . . . . 22 . . . . . . . . . . . . . 23 . . . . . . . . . . . . . 26 81. . . . . . . . . . . . 31 82. . . . . . . . . . . . 37 Page Petiole Curvature - 1981 . . . . . . . . . . . 43 Petiole Curvature - 1982 . . . . . . . . . . . 48 SUMMARY AND CONCLUSTION . . . . . . . . . . . . . . 54 CHAPTER II: THE INHERITANCE OF THE "UMBRELLA" BRANCHING HABIT IN PEPPER, (CAPSICUM ANNUUM L.) INTRODUCTION . . . . . . . . . . . . . . . . . . . . 58 REVIEW OF LITERATURE . . . . . . . . . . . . . . . . 61 MATERIALS AND METHODS . . . . . . . . . . . . . . . 68 Hybridization . . . . . . . . . . . . . . . . . 71 MSU 78-101 X MSU 79-221 . . . . . . . . . . . . 71 MSU 79-221 X MSU 74-230 . . . . . . . . . . . . 72 MSU 78-101 X MSU 74-230 . . . . . . . . . . . . 73 Branching and Fruit Number . . . . . . . . . . 73 RESULTS AND DISCUSSION . . . . . . . . . . . . . . . 75 Plant Type and Bearing Habit . . . . . . . . . 75 MSU 78-101 X MSU 79-221 . . . . . . . . . 75 MSU 79-221 X MSU 74-230 . . . . . . . . . 77 MSU 78-101 X MSU 74-230 . . . . . . . . . 81 Branching and Fruit Number . . . . . . . . . . 84 MSU 78-101 X MSU 79-221 . . . . . . . . . 84 MSU 79-221 X MSU 74-230 . . . . . . . . . 88 MSU 78-101 X MSU 74-230 . . . . . . . . . 90 SUMMARY AND CONCLUSIONS . . . . . . . . . . . . . . . . . 92 BIBLIOGRAPHY O O O O O O O O O O O O O O O O O O O O O O 94 APPENDIX 0 0‘ O O O O O O O O O O O O O O O O O O O O O O 98 Table 10. LIST OF TABLES Page Table of Calculated Curvature Index Values with Approximate Corresponding Angles . . . . . . . 11 Means and Standard Deviations for Petiole Width at Individual vs All Distances MSU 78-7A and MSU 78-9A O I O O O O O O O O O O O I O O O O 0 17 Means and Standard Deviations for Petiole Thickness at Individual vs All Distances MSU 78-7A and MSU 78-9A . . . . . . . . . . . . . . 17 Width of Individual Petioles and Various Petiole Combinations for MSU 78-7A and MSU 78-9A . . . . . l9 Thickness of Individual Petioles and Various Petiole Combinations for MSU 78-7A and MSU 78-9A . 20 Angle of Adaxial Curvature of Individual Petioles and Various Petiole Combinations for MSU 78-7A and MSU 78-9A . . . . . . . . . . . . . . 21 Analyses of Variance for Petiole Width, Thickness and Angle of Adaxial Curvature for Years 1981 and 1982 . . . . . . . . . . . . . . 22 Significance of the A,B,C Scaling Test for Petiole Width in the Cross MSU 78-7A X MSU 78-9A, 1981 . . . . . . . . . . . . . . . . . . 26 Significance of the A,B,C Scaling Test for Petiole Width in the Cross MSU 78-7A X MSU 78-9A, 1982 . . . . . . . . . . . . . . . . . . 29 Chi-Square Test for Goodness of Fit for Pooled Populations From the Cross MSU 78-7A X MSU 78-9A Based on a One Gene Hypothesis for Petiole Width 1981 . . . . . . . . . . . . . . . . . . . . 30 vi Table 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. Page Chi-Square Test for Goodness of Fit for Pooled Populations From the Cross MSU 78-7A X MSU 78-9A, Based on a One Gene Hypothesis for Petiole Width, 1982 . . . . . . . . . . . . . . . . . . . . 31 Significance of the A,B,C Scaling Test for Petiole Thickness in the Cross MSU 78-7A X MSU 78-9A, 1981 . . . . . . . . . . . . . . . . . . 34 Gene Effects Estimated Using A Six Parameter Model on the Generation Means for the Cross MSU 78-7A X 78-9A, 1981 . . . . . . . . . . . . . . 35 Theoretical F2 Means for One, Two and Three Gene Pairs Assuming Complete Dominance, 1981. . . . 36 Chi-Square Test for Goodness of Fit for Pooled Populations From the Cross MSU 78-7A X MSU 78-9A, Based on a Two Gene, 9:7 Segregation Hypothesis for Petiole Thickness, 1981. . . . . . . 37 Significance of the A,B,C Scaling Test for Petiole Thickness in the Cross MSU 78-7A X MSU 78-9A, 1982 . . . . . . . . . . . . . . . . . . 40 Gene Effects Estimated Using a Six Parameter Model on the Generation Means for the Cross MSU 78-7A X MSU 78-9A, 1982 . . . . . . . . . . . . 41 Theoretical F2 Means for One, Two and Three Gene Pairs Assuming Complete Dominance, 1983. . . . 42 Chi-Square Test for Goodness of Fit for Pooled Populations From the Cross MSU 78-7A X MSU 78-9A, Based on a Two Gene Pair, 9:7 Segregation Hypothesis for Petiole Thickness, 1982. . . . . . . 42 Significance of the A,B,C Scaling Test for Adaxial Curvature in the Cross MSU 78-7A X MSU 78-9A, 1981 . . . . . . . . . . . . . . . . . . 46 Gene Effects Estimated Using a Six Parameter Model on the Generation Means for the Cross MSU 78-7A X MSU 78-9A, 1981 . . . . . . . . . . . . 47 vii Table Page 22. Theoretical F2 Means for One, Two and Three Gene Pairs Assuming Complete Dominance, 1981. . . . 47 23. Chi-Square Test for Goodness of Fit for Pooled Populations From the Cross MSU 78-7A X MSU 78-9A BASED ON A TWO GENE, 9:7 Segregation Hypothesis for Adaxial Curvature, 1981. . . . . . . 48 24. Significance of the A,B,C Scaling Test for Adaxial Curvature in the Cross MSU 78-7A X MSU 78-9A, 1982 . . . . . . . . . . . . . . . . . . 51 25. Gene Effects Estimated Using A Six Parameter Model on the Generation Means for the Cross MSU 78-7A x MSU 78-9A' 1982 o o o o o o o o o o o o 5.2 26. Theoretical F2 Means for Petiole Curvature for One, Two and Three Genes Assuming Complete Dominance . . . . . . . . . . . . . . . . . . . . . 52 27. Chi-Square Test for Goodness of Fit for Pooled Populations From the Cross MSU 78-7A X MSU 78-9A Based on a Two Gene, 9:7 Segregation Hypothesis for Adaxial Curvature, 1982. . . . . . . 52 28. Frequency Distribution for Petiole Widths (mm) For the Cross MSU 78-7A X MSU 78-9A, 1981 . . . . . 98 29. Frequency Distribution for Petiole Widths (mm) For the Cross Msu 78-7A X MSU 78-9A, 1982 . . . . . 99 30. Frequency Distribution for Petiole Thickness (mm) For the Cross MSU 78-7A X MSU 78-9A, 1981 . . . . . 100 31. Frequency Distribution for Petiole Thickness (mm) For the Cross MSU 78-7A X MSU 78-9A, 1982 . . . . . 101 32. Frequency Distribution for Adaxial Curvature Indexes for the Cross MSU 78-7A X MSU 78-9A, 1981 . 102 33. Frequency Distribution for Adaxial Curvature Indexes for the Cross MSU 78-7A X MSU 78-9A, 1982 . 103 34. Chi-Square Test For Goodness of Fit For Pooled Generations From the Cross MSU 78-101 X MSU 79-221. 75 viii Table Page 35. Chi-Square Test for Goodness of Fit for the Cross MSU 79-221 X MSU 74-230, Based on a 4-Gene, Epistatic Model . . . . . . . . . . . . . . . . . . 78 36. Separation of the X2 Test for Goodness of Fit Into its Components . . . . . . . . . . . . . . . . 79 37. Chi-Square Test for Goodness of Fit to a 9:3:3:1 Expected Ratio for the Cross MSU 78-101 X MSU 74-230 0 O O O O O O O O O O I O O O O O O O O 82 38. Separation of the X2 Test for Goodness of Fit Into Its Components . . . . . . . . . . . . . . . . 83 39. Mean Number of Marketable Fruit and Branches Per Plant for the Various Plant Types, From the Cross MSU 78-101 X MSU 79-221 . . . . . . . . . . . 85 40. Mean Number of Marketable Fruit and Branches Per Plant for the Various Plant Types, From the Cross MSU 79-221 X MSU 74-230 . . . . . . . . . . . . . . 89 41. Mean Number of Marketable Fruit and Branches Per Plant for the Various Plant Types, from the Cross MSU 78-101 X MSU 74-230 . . . . . . . . . . . . . . 91 LIST OF FIGURES Page Schematic Representation of a Pak-Choi Petiole Cross-Section . . . . . . . . . . . . . . . 10 Frequency Distribution for Petiole Widths (mm) From the Cross MSU 78-7A X MSU 78-9A, 1981. . . . . 25 Frequency Distribution for Petiole Widths (mm) From the Cross MSU 78-7A X MSU 78-9A, 1982. . . . . 28 Frequency Distributions for Petiole Thickness (mm) From the Cross MSU 78-7A X MSU 78-9A, 1981. . . . . 32 Frequency Distributions for Petiole Thickness (mm) From the Cross MSU 78-7A X MSU 78-9A, 1982. . . . . 39 Frequency Distributions for Petiole Curvature From the Cross MSU 78-7A X MSU 78-9A, 1981. . . . . 45 Frequency Distributions for Petiole Curvature From the Cross MSU 78-7A X MSU 78-9A, 1982. . . . . 50 Plant Habit of the Three Parental Lines . . . . . . 70 Schematic Representation of the Various Plant Types and Levels of Branching . . . . . . . . . . . 87 CHAPTER I THE INHERITANCE OF PETIOLE CHARACTERS IN PAK-CHOI INTRODUCTION Pak-Choi (Brassica rapa L., Chinensis group) is a member of the mustard family, closely related to Chinese cabbage, though it differs greatly in appearance. It is actually two vegetables in one and is used in a variety of oriental dishes, the leaves prepared like spinach, the petioles like asparagus. The influx of large numbers of Asians into the United States over the past 20 to 25 years, coupled with a greater popularity of ethnic foods by the general public, has led to an increase in the use of and demand for Pak-choi. Commercial production of Pak-Choi has been limited mainly to California, Hawaii and New Jersey, with shipment to other regions of the country for distribution and sale. Reports on total production of Pak-choi are limited. In California in 1971, about 30 acres were produced in the Carmel and Salinas Valleys south of San Francisco. Hawaii produced approximately 196,000 pounds of Pak-Choi in the same year (Yamaguchi, 1973). There has been a steady increase over the years in the number of acres in Hawaii devoted to the production of Pak-Choi, increasing from 84 1 acres in 1975, to 120 acres in 1979, for a total value in 1979 of approximately $360,000 (Statistics of Hawaiian Agriculture, 1979). In View of the large oriental population in the Chicago area, with its close proximity to Michigan, and the high value of the crop, commercial production of Pak-Choi by Michigan farmers could become a very profitable enterprise. However, early attempts at Pak-Choi culture in Michigan often resulted in loss of the crop due to bolting. While Pak-Choi is often sold with small flowers in China and Japan, this trait is undesirable in terms of con- sumer preference in the United States. The major seed sources for Pak-Choi are seed houses located in Japan and Taiwan. Unfortunately, cultivars produced by these com- panies are not well adapted to Michigan or to the needs of U.S. consumers. Acceptance of flowering Pak-Choi in other countries has severely limited interest in selecting for bolt resistant lines. The concern instead lies in the pro- duction of cultivars with wide, thick and succulent petioles. While the major emphasis of the MSU Pak-Choi breeding program has been the development of bolt resistant cultivars, selection has also been directed towards cultivars with wide, thick petioles. As selection pressure for bolt resistance increased, there was a decrease in the number of plants within the population exhibiting these characters. Elucidation of the factors controlling petiole characters such as shape, width and thickness would facili- tate efficient breeding for the development of cultivars exhibiting the desired petiole characteristics. The objectives of the present study were: 1. To develop a technique for measuring petiole shape, width and thickness. 2. To develop a sampling technique appropriate for genetic study. 3. To investigate the inheritance of petiole shape, width and thickness. REVIEW OF LITERATURE There are two main groups of Chinese cabbage, the heading group called Pe-Tsai (Brassica rapa L., Pekinensis group), and the non-heading group called Pak-Choi (Brassica EEEE.L°' Chinensis group). Both Pe-Tsai and Pak-Choi mean "white vegetable", Pe-Tsai is the Peking dialect while Pak= Choi is the Cantonese dialect (Lee, 1982). Pak-Choi (also referred to as Bok Choy and Bok Tsoi in the literature) is a vegetable of ancient origin, believed to have originated in Japan and Eastern China (Vavilov, 1949/1950) or China (Lee, 1982). Early literature from the 5th century A.D. indicates that Pak-Choi was a relatively unimportant crop in China at that time. However, it is now the most common and extensively used leafy vegetable in China (Li, 1969). There is little mention in the literature of the use of Pak-Choi in the U.S. Bailey, in 1894, described Pak-Choi as, "leaf stalks are light colored, sometimes almost ivory white and celery-like," adding that the plant had little use in the U.S. (Bailey, 1922). Discussion of its use as a vegetable crop increased in the 1940's (Smith, 1945; 4 Whallon, 1948), but no work has been reported on the breeding of the crop since then. To the author's knowledge, no work has been reported concerning the petiole morphology of Pak-Choi. A comprehensive search of the literature on other petiole crops such as celery and rhubarb also failed to expose previous research of this type. Frazier, et al, (1960) conducted correlation studies between immature and mature rhubarb, in which they studied seedling petiole color, vigor and habit. In a subsequent study on the forcing of rhubarb, Martin (1962) found high correlations between parental and progeny petiole types, but failed to report any criteria upon which petiole type was defined. MATERIALS AND METHODS Sampling Technique Greenhouse Study This investigation was undertaken to develop a rapid and reliable sampling method for measuring the petiole morphology of the Pak-Choi plant. Two Pak-Choi lines inbred for seven generations, MSU 78-7A, a narrow, thick petiole line, and MSU 78-9A, a wide, thin petiole line were utilized. The lines were derived from the cross, Taisai X Jiang-Mein. In Pak-Choi, the mature plant consists of a single axis, with a whorled leaf arrangement. Seventy-five plants of each of the parents were grown in the greenhouse bed in a completely randomized design with five replicates of fifteen plants each, from February to April, 1980. The petioles of the third, fifth, seventh, ninth and eleventh fully-expanded leaves above the cotyledon were measured when the plants were mature. (Plants were considered mature when they reached marketable size.) Measurements of petiole width and thickness were made with a caliper, at 3, 6, 9 and 12 cm from the base of the petiole, in order to determine the best distance from the base for measuring these characters. Means, variances and standard deviations were calculated from individual plants. Popula- tion means were statistically compared by use of the two-tailed t-test (Steel and Torrie, 1960). Field Study Due to the laborious and time consuming problem of measuring each petiole with a caliper, a more rapid data collecting technique was investigated. A second study was made by stamping a cross-section of each petiole on paper. Seventy-five plants of each line were field grown in a completely randomized design with five replicates of fifteen plants each, from June to July, 1980, in an effort to further refine the procedure by reducing the number of petioles sampled. The petioles of the third, fifth, seventh, ninth and eleventh fully-expanded leaves were measured. Based on the results of the previous study, cross sections were made 9 cm from the base of the petiole, using a single-edge razor blade. The stamp pad was then used to produce an ink impression of the petiole on a data sheet, providing a rapid and permanent recording of the petiole shape for subsequent measurement. Data was obtained for petiole width, thickness and angle of adaxial curvature for each of the five petioles sampled from each plant. A millimeter ruler was used for measuring the thickness and width, while a modified protrac- tor was used to measure the angle. Statistical treatment of the data was similar to the greenhouse grown plants. Data from the two studies were utilized to determine the relation- ship between measurements taken in the greenhouse and field under different environmental conditions. Inheritance Studies - 1981 Two inbred parents, MSU 78-7A (P1) and MSU 78-9A (P2), were used to produce the various populations for the study. Both parents were derived from the cross, Taisai X Jiang-Mein. Reciprocal populations of F1' F2, BC1 and BC2 were produced by hand pollination in the greenhouse. Seeds of parents and progenies were planted in a greenhouse in No. 96 PVC growing trays containing VSP artificial soil mix on May 3, 1981, and transplanted to the field on June 2, 1981, using a randomized complete block design with three replications. Each replicate contained 12 plants of each parental line, 24 plants of each F1 family, 36 plants of each backcross population, and 108 of each F2 population. Plants were grown in rows 61 cm apart and spaced at intervals of 30 cm in the row. Cultural practices were those recommended for commercial production in Michigan. Imprints of the fifth, seventh and ninth petioles, 9 cm from the base, were made in July. Each imprint was measured for thickness (a). FIGURE 1 SCHEMATIC REPRESENTATION OF A PAK-CHOI PETIOLE CROSS-SECTION 10 (C) ; (d) (b) 11 height (b), and outer (c) and inner (d) width (Figure 1). Since the measuring of the adaxial curvature angle was diffi- cult, a curvature index (CI) was determined, using the formula: height (b) - thickness (a) CI inner width (d) The numerator of the above formula expresses the height of the curvature which, when divided by the denominator (inner width), produces a ratio or relative degree of curva- ture; 0 representing no curvature, with a greater degree of curvature as the value increases (Table 1). TABLE 1 TABLE OF CALCULATED CURVATURE INDEX VALUES WITH APPROXIMATE CORRESPONDING ANGLES Curvature Index (CI) Angle (degrees) 0.0 180 0.1 160 0.2 145 0.3 125 0.4 110 0.5 90 0.6 75 0.7 65 0.8 55 0.9 50 1.0 40 Data from the three petioles were pooled to obtain a plant mean prior to analysis. Statistical treatment of the 12 data was similar to the previous experiment. Data was also obtained for bolting stage using the following rating system: (1) vegetative, (2) flower primordia visible, (3) bolted, seedstalk elongation of 2" or greater. Determination of bolting stage was made by visual observation of the apical meristem. Inheritance Studies - 1982 Identical genetic material as used in 1981 and two F3 families were grown in 1982. Due to shortage of seed, some population sizes were limited. Seeds of the parents and progenies were planted in a greenhouse in vermiculite on May 6, 1982, transplanted to PVC growing trays one week later, and transplanted to the field on June 8, 1982. A randomized complete block design with three replications was utilized. Each replicate contained 12 plants of each parental line and F1 family, 18 plants of each F3 family, 36 plants of each backcross family, and 72 plants of each F2 family. Cultural practices were similar to the 1981 experiment. Petiole imprints were made in July. Measurements and data analysis were similar to the 1981 experiment. Analysis of variance was conducted on the data for each character (Little and Hills, 1975). The means of reci- procal populations were tested for homogeneity prior to analysis of the data. The distributions of the segregating 13 generations were continuous, with no apparent division into phenotypic classes. Thus, the data were analyzed biometri- cally. For each character, Mathers' ABC scaling test (Mather and Jinks, 1971), with A = 2561 - El - 51, B = 2562 - Fl — P2, and C = 4F2 - ZFl _ Pl - P2, was applied to the generation means to determine the adequacy of the additive- dominance model. When significant results were obtained from Mathers' scaling test, a generation means analysis was conducted using the method outlined by Mather and Jinks (1971), and notations used by Gamble (1962) to fit a six parameter model. The parameters tested are the mean effect (m), the pooled additive effect (a), the pooled dominance effect (d), the pooled additive X additive effect (aa), the pooled additive X dominance effect (ad), and the pooled dominance x dominance effect (dd). The equations used for estimating these parameters in terms of generation means are: m = F2 a = gal—s62 d = 0.51?1 - 0.5172 + Fl - 452 + 23c, + 2552 aa = ZBCi + ZBCZ - 4F2 ad = -0.551 + 0.552 + 351 - B52 dd = 51 + 52 + 251 + 452 - 4361 - 4352 14 Standard errors of the corresponding generation means were used to test the significance of the various gene effects. Narrow sense heritability, hzns, was estimated by Warners' procedure (1952) as: h2ns = [2VF2 - (VBCl + VBCZX] / VFZ, where VFZ, VBCl, and VBC2 are the variances of the F2, BCl, and BC2 generations, respectively. Standard errors for each hzns estimate were derived by the square root of the formula: V(h2ns) = 2£[(VBc1 + VBC2) 2/df] + (VBC1 2/dec1) + (VBC2 2/dec2)} /VF22. Expected gain from selection (G. S.) was calculated using Allards' formula (1960) as G.S. = k phzns, where k is the selection differential, p is the phenotypic standard deviation of the F2, and h2ns is heritability in the narrow sense. Theoretical F2 means were calculated, based on the number of gene pairs assumed to be differentiating the parents, after the formulas proposed by Powers, et al (1950). Theoretical means were tested for significance from the observed means by the use of a one-tailed t-test (Steel and Torrie, 1960). The minimum number of gene pairs differen- 15 tiating the parents were computed using the methods of Castle (1921) and Burton (1951). To determine if the data from the two years could be pooled prior to genetic analysis, an analysis of variance was conducted using data from both years to test year X generation effects. Pearson's correla- tion coefficients were obtained for the three characters in relation to bolting stage. RESULTS AND DISCUSSION Sampling Technique Greenhouse Study Mean petiole thickness and width data for the parents MSU 78-7A and MSU 78-9A were obtained by measuring the third, fifth, seventh, ninth and eleventh petioles at distances 3, 6, 9 and 12 cm from the base of each petiole. MSU 78-7A, the narrow, thick petiole line, had means of 16.2 mm wide and 7.7 mm thick. MSU 78-9A exhibited wide, thin petioles with means of 19.0 mm wide and 5.5 mm thick. Since sampling of four distances along the petiole was time consuming, it was necessary to determine the proper distance from the base to obtain a valid sample measurement. For petiole width of MSU 78-7A and MSU 78-9A, the means of 3 cm and 12 cm were both significantly different (1% level) from the mean of all four distances (Table 2). The 6 cm and 9 cm means were not significantly different from the mean of all four distances. The data suggest that sampling at 6 cm or 9 cm from the base of the petiole in the MSU 78-7A and MSU 78-9A parents would provide a valid estimate of petiole width. The 't' and probability values 16 17 were obtained by comparing the mean of each individual dis- tance measurement with the mean of all distances combined (3, 6, 9 and 12 cm). TABLE 2 MEANS AND STANDARD DEVIATIONS FOR PETIOLE WIDTH AT INDIVIDUAL VS. ALL DISTANCES MSU 78-7A AND MSU 78-9A Distance Mean . . (cm) (mm) S.D. T Value Probability MSU 78-7A 3 21.8 5.11 7.68 .000 6 16.6 3.37 0.78 .458 9 15.8 2.99 -O.85 .400 12 11.0 2.58 -13.29 .000 3,6,9,12 16.2 3.64 - - MSU 78-9A 3 26.8 3.76 14.41 .000 6 19.7 3.90 1.33 .192 9 18.4 3.23 -1.35 .186 12 12.6 2.21 -l7.6l .000 3,6,9,12 19.0 3.40 - - TABLE 3 MEANS AND STANDARD DEVIATIONS FOR PETIOLE THICKNESS AT INDIVIDUAL VS. ALL DISTANCES MSU 78-7A AND MSU 78-9A Distance Mean S.D. T Value Probability (cm) (mm) MSU 78-7A 3 7.9 1.51 0.78 .458 6 7.8 1.41 0.44 .662 9 7.8 1.37 0.15 .881 12 7.4 1.49 -l.34 .190 3,6,9,12 7.7 1.44 - - MSU 78-9A 3 6.3 1.19 4.89 .000 6 5.7 1.19 1.39 .177 9 5.4 1.18 -O.35 .728 12 4.6 0.99 -5.53 .000 3,6,9,12 5.5 1.15 - - 18 For petiole thickness of MSU 78-7A, none of the four distances from the base of the petiole differed significantly from the mean for all four distances. For MSU 78-9A, the 3 cm and 12 cm distances were significantly different (1% level).from the mean of all four distances. The 6 cm and 9 cm distances were not significantly different and had approximately equal standard deviations. However, the 9 cm distance from the base of the petiole exhibited the mean closest to the total mean, and was thus selected as the proper distance for obtaining a valid estimate of petiole thickness. Based on the results of this study, the 9 cm distance from the base of the petiole was used for sampling petiole width, thickness and the angle of adaxial curvature. Field Study - 1980 For petiole width of MSU 78-7A, means of each indivi- dual petiole per plant were significantly different (1% level) from the mean of all five petioles (Table 4). The mean of petioles 3, 7 and 11 and the mean of petioles 5, 7 and 9 were not significantly different from the five petiole mean. The 5, 7, 9 combination exhibited the lowest standard deviation of the two petiole combinations. For MSU 78-9A, the mean of the third petiole was significantly different (1% level) from the mean of all five petioles.. The mean of the seventh petiole was significantly different (5% level) from the five petiole 19 mean. The mean of petiole numbers 3, 7, and 11 was not sig- nificantly different from the five petiole mean. The mean of petioles numbers 5, 7, and 9 was not significantly different from the five petiole mean and had a similar standard deviation. TABLE 4 WIDTH OF INDIVIDUAL PETIOLES AND VARIOUS PETIOLE COMBINATIONS FOR MSU 78-7A AND MSU 78-9A Petioles Mean S.D. T Value Probability (mm) MSU 78-7A . 3 15.0 2.80 2.78 .009 5 15.5 2.44 5.21 .000 7 15.1 2.70 3.34 .002 9 12.3 2.42 -6.20 .000 11 10.5 1.56 -l7.14 .000 3,7,11 13.8 3.21 -1.17 .245 5,7,9 14.3 3.06 1.22 .225 3,5,7,9,ll 14.1 2.47 - - MSU 78-9A 3 19.8 4.18 -2.79 .008 5 20.4 4.67 -l.56 .128 7 22.9 5.47 2.52 .015 9 21.7 5.71 0.55 .585 11 21.5 6.15 0.32 .750 3,7,11 21.5 5.68 0.48 .632 5,7,9 21.6 5.39 0.97 .334 3,5,7,9,11 21.3 5.33 - - For petiole thickness of MSU 78-7A, individua1;means of petioles 5, 7, 9, and 11 were significantly different (1% level) from the five petiole mean (Table 5). The mean of petiole 3 was significantly different (5% level) from the five petiole mean. The mean of petioles 3, 7, and 11 and the 20 mean of petioles 5, 7, and 9 were not significantly different from the five petiole mean. The 5, 7, 9 combination exhibited the lowest standard deviation of the two petiole combinations. For MSU 78-9A, the mean of the third petiole was significantly different (1% level) from the five petiole mean. None of the other individual petiole means were significantly different from the five petiole mean, and had similar standard devia- tions. Based on data obtained from this study, sampling petioles 3, 7, and 11 or 5, 7, and 9 would provide a reliable estimate of petiole thickness. Table 5 THICKNESS OF INDIVIDUAL PETIOLES AND VARIOUS PETIOLE COMBINATIONS FOR MSU 78-7A AND MSU 78-9A Petioles Mean S.D. T Value Probability (mm) MSU 78-7A 3 9.3 1.56 2.22 .038 5 9.5 1.18 4.30 .000 7 9.4 1.29 3.08 .005 9 8.2 1.47 -3.90 .001 11 7.2 1.20 -10.56 .000 3,7,11 8.7 1.68 1.29 .205 5,7,9 9.0 1.49 1.24 .217 3,5,7,9,1l 8.9 1.36 - - MSU 78-9A 3 5.9 1.48 -3.63 .000 5 6.3 1.75 -l.63 .109 7 7.0 1.75 1.83 .050 9 6.9 1.72 1.35 .185 11 6.9 1.54 1.28 .206 3,7,11 6.5 1.79 -0.64 .523 5,7,9 6.7 1.75 0.85 .399 3,5,7,9,11 6.6 1.66 - - For angle of adaxial curvature, none of the individual 21 petiole means differed significantly from the mean of all petioles for either parent (Table 6). The 5, 7, 9 petiole combination exhibited a lower standard deviation than the 3, 7, 11 combination. Table 6 ANGLE OF ADAXIAL CURVATURE OF INDIVIDUAL PETIOLES AND VARIOUS PETIOLE COMBINATIONS FOR MSU 78-7A AND MSU 78-9A Petioles Mean S.D. T Value Probability (degrees) MSU 78-7A 3 119.9 16.87 0.05 .980 5 118.6 16.45 0.71 .495 7 117.2 15.45 -1.56 .125 9 120.5 17.69 0.28 .777 11 125.0 21.98 1.69 .095 3,7,11 120.2 18.46 0.22 .825 5,7,9 119.3 16.73 -0.59 .551 3,5,7,9,11 120.0 17.57 - - MSU 78-9A 3 145.8 17.33 1.91 .062 5 142.6 19.96 0.51 .612 7 142.5 22.51 0.40 .693 9 140.4 22.88 -0.38 .701 11 138.0 20.94 -1.54 .133 3,7,11 141.8 20.94 0.08 .993 5,7,9 141.9 20.75 0.11 .991 3,5,7,9 11 141.7 20.96 - - Correlation coefficients between data taken from the 9 cm distance in greenhouse and field experiments were .73 for width, .82 for thickness, and .58 for angle of adaxial curvature. All were significant at the .001 level. Based on the results of the two studies on sampling methods, the 22 combination of petioles 5, 7, and 9, measured at a distance 9 cm from the base of each petiole, was used to sample the populations for the various characters under investigation. Inheritance Studies - Pooled Analysis The analyses of variance for all three characters for both seasons showed significant generation, year, and genera- tion X year effects (Table 7). Therefore, the data were not TABLE 7 ANALYSES OF VARIANCE FOR PETIOLE WIDTH, THICKNESS AND ANGLE OF ADAXIAL CURVATURE FOR YEARS 1981 AND 1982 WIDTH Source df MS F Total 1756 Generation 5 3704.82 126.72** Year 1 1429.16 48.88** Generation X Year 5 84.56 2.89* Error 1745 29.24 THICKNESS Source df MS F Total 1756 Generation 5 63.23 17.65** Year 1 432.39 120.68** Generation X Year 5 12.04 3.36** Error 1745 3.58 ADAXIAL CURVATURE Source df MS F Total 1756 Generation 5 4.409 65.07** Year 1 7.178 105.94** Generation X Year 5 .308 4.55** Error 1745 .068 **significant at .01 level *significant at .05 level 23 pooled, and the results of the two experiments are presented separately. Within each year, however, there were no signi- ficant differences between reciprocal population for all traits, and the data were pooled prior to analysis. Petiole Width - 1981 The F1 and F2 frequency distributions were skewed towards the narrow parent, P1, (Figure 2). Seventy-four percent of the F2 population had widths lower than the mid= parent mean, 23.5 mm, (Appendix Table 28) suggesting a partial dominance of the narrow parent over the wide parent. The F1 mean, 20.4 mm, was lower than the midparent mean, 23.5 mm, but was not significant. The F2 mean showed a simi- lar trend. The mean of the BC to P1 fell between the P1 and F1 means, and was significantly different (5% level) from both means. The BC to P2 mean fell closer to the P2 mean, and was significantly different (5% level) from the P2 and F1 means. The parental difference was 14.5 mm. The results of Mather's A, B, C scaling test were non-significant (Table 8). Significance for any of these estimates suggests the presence of epistasis. Two estimates of k, the minimum number of loci con- trolling petiole width were computed. Estimates of 1.06 and 1.14 genes were obtained using the methods of Castle and Burton, respectively. Using the formula proposed by Powers, 24 FIGURE 2 FREQUENCY DISTRIBUTION FOR PETIOLE WIDTHS (mm) FROM THE CROSS MSU 78-7A X MSU 78-9A, 1981 FREQUENCY 75 25 V U! N U! V U! 25 180 60 25 I‘2 ) 1 1 J IO 15 2O 25 3O 35 40 45 50 PETIOLE WIDTH (mm) 26 TABLE 8. SIGNIFICANCE OF THE A, B, C SCALING TEST FOR PETIOLE WIDTH IN THE CROSS MSU 78-7A X MSU 78-9A, 1981 Test A B C ns ns ns ns, non-significant at 5% level et a1 (1950), 3/4(Pl) + 1/4(P2) = F2, a theoretical F2 mean of 19.8 mm was obtained, which was not significantly different from the observed mean (20.5 mm) and suggested that petiole width is determined by a single major gene. Petiole Width - 1982 As in the previous year, the F1 and F2 frequency dis- tributions were skewed towards the narrow parent (P1) (Figure 3). Seventy-one percent of the F2 population had values lower than the midparent mean (25.4 mm), suggesting a partial dominance of the narrow parent over the wide parent. The means of the segregating populations also show similar dominance patterns (Appendix Table .29). The F1 mean (23.0 mm) was not significantly lower than the midparent mean. The F2 mean was slightly higher than the F1 mean, but lower than the midparent mean. The mean of the BC to P1 fell between the F l and P means, and was significantly different (5% level) from 1 27 FIGURE 3 FREQUENCY DISTRIBUTION FOR PETIOLE WIDTHS (mm) FROM THE CROSS MSU 78-7A X MSU 78-9A, 1982 28 25 75 25 FREQUENCY 75 100 50 25 F2 I l l I I I 25 3O 35 4O 45 50 PETIOL E WIDTH (mm) 29 both means. The mean of the BC to P2 fell closer to the P2 mean, and was significantly different (5% level) from the P2 and F1 means. The means of the two F3 populations were 20.8 mm and 19.5 mm. The parental difference in width was 17.1 mm. TABLE 9 SIGNIFICANCE OF THE A, B, C SCALING TEST FOR PETIOLE WIDTH IN THE CROSS MSU 78-7A X MSU 78-9A, 1982 Test A B C ns ns ns ns, non-significant at 5% level As in the previous year, Mather's A, B, C scaling test yielded non-significant estimates, suggesting an absence of epistasis in the expression of petiole width (Table 9). A theoretical F2 mean of 21.1 mm was obtained, using Powers' formula for one gene pair, which was not significantly different from the observed mean of 23.7 mm, suggesting that petiole width is determined by a single major gene. This is supported by the estimates of k, the number of effective factors, obtained using the methods of Castle and Burton. Estimates were for 1.13 and 1.23 genes, respectively 30 The F2 and backcross generations from 1981 were par- titioned into narrow and wide types based on the arithmetic mean of the two parents (23.5 mm), using the chi-square test to test for goodness of fit to a one gene model (Table 10). TABLE 10 CHI-SQUARE TEST FOR GOODNESS OF FIT FOR POOLED POPULATIONS FROM THE CROSS MSU 78-7A X MSU 78-9A, BASED ON A ONE GENE HYPOTHESIS FOR PETIOLE WIDTH, 1981 Generation Observed Theoretical Chi-sq. P ratio ratio Pooled F1 12 :11 1:0 - - Pooled F2 372:131 3:1 .247 .90-.50 Pooled BC to P1 179:16 1:0 - - Pooled BC to P2 61:102 1:1 9.44 .01-.004 Based on the ratios expected with a one major gene model, a good fit was obtained for the F2 in 1981 (P = .90 - .50). The BC2 ratio, however, was skewed towards the wide parent, suggesting the presence of modifiers in P2 affecting the expression of this trait. The BCl ratio of 179:16 suggests agreement with the expectation of the model. In 1982, the F2 population showed an acceptable fit to the 3:1 expected ratio (P = .10 - .09), while the BC2 fit was poor (P = (.001, Table 11). This may be explained by the 31 differential expression of modifier genes in relation to different environmental conditions in the two experiments. TABLE 11 CHI-SQUARE TEST FOR GOODNESS OF FIT FOR POOLED POPULATIONS FROM THE CROSS MUS 78-7A X MSU 78-9A, BASED ON A ONE GENE HYPOTHESIS FOR PETIOLE WIDTH, 1982 Generation Observed Theoretical Chi-Sq. P ratio ratio Pooled Fl 50:4 1:0 - -‘ Pooled F2 214:88 3:1 2.76 .10-.09 Pooled BC to P1 151:1 1:0 - - Pooled BC to P2 26:115 1:1 56.176 (.001 The data indicate that there is dominance for narrow petioles, while wide, thick petiole phenotypes are desired. Therefore, selection of recessive genotypes (exhibiting wide petioles) would be necessary for improving petiole width in Pak-Choi. Petiole Thickness - 1981 The F1 and F2 frequency distributions were skewed towards the thick parent (Pl) (Figure 4). Approximately fifty-nine percent of the F2 population had mean thickness higher than the midparent mean (6.6 mm), suggesting a partial dominance of thick over thin petioles. The means of the various populations support the dominance pattern 32 FIGURE 4 FREQUENCY DISTRIBUTIONS FOR PETIOLE THICKNESS (mm) FROM THE CROSS MSU 78-A X MSU 78-9A, 1981 FREQUENCY 33 75* 25*- 180 — 60 l l 2 4 b 8 IO 12 I4 PETIOLE THICKNESS (mm) 34 (Appendix Table 30). The F1 mean (7.1 mm) was not signifi- cantly higher than the midparent mean. The F2 mean (7.4 mm) showed a similar pattern, but was significantly different (5% level) from the midparent mean. The mean of the BC to P1 fell between the F1 and P1 means, while the BC to P2 mean fell between the F1 and P2 means. The BC to P2 mean was significantly different (5% level) from the P2 mean, but not from the F1 mean or midparent mean. TABLE 12 SIGNIFICANCE OF THE A, B, C SCALING TEST FOR PETIOLE THICKNESS IN THE CROSS MSU 78-7A X MSU 78-9A, 1981. Test A B C ns ** ** **, significant at 1% level ns, non-significant at 5% level Significant values (1% level) were obtained for factors B and C in Mather's A, B, C Scaling test suggesting the possibility of epistasis in the expression of petiole thickness (Table 12). This made it necessary to abandon the three parameter model and use a six parameter model outlined by Mather and Jinks (1971) in order to determine the nature of epistatic effects. The six parameter estimates (Table 13) show that dominance gene effects made the major contribution 35 to variation in petiole thickness. Also, additive X dominance epistatic effects were significant (1% level). TABLE 13 GENE EFFECTS ESTIMATED USING A SIX PARAMETER MODEL ON THE GENERATION MEANS FOR THE CROSS MSU 78-7A X 78-9A, 1981 Effect Estimate m 7.35;.008** a 0.721.033** d 7.65:.321** aa -0.64i.260 ad -0.51i.080** dd -0.931.731 **, significantly different from zero Mather and Jinks (1971) stated that the types of epi- static interaction can be inferred from the relative signs of d and dd. Like signs indicate complementary epistasis whereas opposite signs indicate epistasis of the duplicate, dominant or recessive suppressor types. In this cross, the estimate of d is significantly positive while the dd estimate is negative, suggesting either duplicate, dominant or recessive suppressor epistasis for the expression of petiole thickness. 36 Two estimates of k were computed, yielding values of 0.58 and 0.63, using Castle's and Burton's methods, respec- tively. Theoretical F2 means were calculated, based on the number of factors assumed to be differentiating the parents, after the formulae proposed by Powers, et a1, (1950). In these formulae, P1 is the mean of the dominant parent, P2 is the mean of the recessive parent, and F2 is the theoretical F2 mean. TABLE 14 THEORETICAL F2 MEANS FOR ONE, TWO AND THREE GENE PAIRS ASSUMING COMPLETE DOMINANCE, 1981 No. of genes Theoretical Observed 1 (3/4)f>'l + (1/4)§2 = 7.211.53 7.411.99 2 (15/16)Pl + (1/16)P2 = 7.611.51 7.411.99 3 (63/64)Pl + (1/64)P2 = 7.7.11.60 7.411.99 Theoretical F2 means based on one, two and three gene pair hypotheses were not significantly different from the observed F2 mean (Table 14). The presence of epistatic effects, however, suggest the two gene pair hypothesis. The F2 and backcross generations were partitioned into thick and thin types based on the midparent mean (6.6 mm), resulting in a 9:7 segregation pattern. A chi-square was used to test for goodness of fit to a 9:7, two gene epistatic model. Based on the segregation ratios expected from the proposed 37 model, there was a good fit of both the F2 (P = .50 - .30) and BC2 populations (P = .50 - .10, Table 15). TABLE 15 CHI-SQUARE TEST FOR GOODNESS OF FIT FOR POOLED POPULATIONS FROM THE CROSS MSU 78-7A X MSU 78-9A, BASED ON A TWO GENE, 9:7 SEGREGATION HYPOTHESIS FOR PETIOLE THICKNESS, 1981 Generation Observed Theoretical Chi-Sq. P ratio ratio Pooled Fl 103:29 1:0 - - Pooled F2 296:207 9:7 1.01 .50-.30 Pooled BC to P1 159:36 1:0 - - Pooled BC to P 45:118 1:3 0.60 .50-.10 2 Petiole Thickness - 1982 The F1 and F2 frequency distributions for petiole thickness were skewed towards the thick parent (P1) (Figure 5). Approximately fifty-three percent of the F2 had values higher than the mean of the midparent (7.6 mm), suggesting a partial dominance of thick over thin petioles. The means of the segregating populations support the dominance pattern (Appendix Table 31). The F1 mean was not significantly higher than the mean of the two parents (Fl - midparent = .4 mm). The F2 mean was significantly different (5% level) from the mean of the midparent and also suggested partial dominance. The mean of the BC to P1 fell between the F1 and P1 means, 38 FIGURE 5 FREQUENCY DISTRIBUTIONS FOR PETIOLE THICKNESS (mm) FROM THE CROSS MSU 78-7A X MSU 78-9A, 1982. FREQUENCY IS 25 IO 75 25 75 25 I00 50 25 IO 39 3C2 ’3 I 6 8 IO I2 I4 PETIOLE THICKNESS (mm) 40 while the BC to P2 mean fell between the F1 and P2 means, and was significantly different from the P2 mean, but not from the F1 mean or the mean of the midparent. The two F3 popula- tions exhibited means of 8.2 mm and 8.7 mm. Significant values were obtained for factor C in Mather's A, B, C scaling test (1% level), suggesting the possibility of epistasis in the expression of petiole thick- ness (Table 16). TABLE 16 SIGNIFICANCE OF THE A, B, C SCALING TEST FOR PETIOLE THICKNESS IN THE CROSS MSU 78-7A X MSU 78-9A, 1982 Test A B C ns ns ** **, significant at 1% level ns, non-significant at 5% level Six parameter estimates (Table 17) suggest that although there were significant additive effects, dominance gene effects made the major contribution to the variation in petiole thickness, as noted the previous year. However, the estimates for epistatic effects (aa, ad, and dd) were not significant. This could possibly be due to a cancelling out of positive and negative gene effects, or to a level of epistasis too low to be detected by this test. 41 TABLE 17 GENE EFFECTS ESTIMATED USING A SIX PARAMETER MODEL ON THE GENERATION MEANS FOR THE CROSS MSU 78-7A X MSU 78-9A, 1982 Effect Estimate m 8.171.054** a 0.831.046** d 12.331.360** aa -0.86i.440 ad -0.051.822 dd -0.13:.088 **, significantly different from zero Theoretical F2 means were calculated, based on one, two and three gene models and are presented in Table 18. None of the theoretical means were significantly different from the observed mean. Two estimates of k were computed. Estimates of 0.61 and 0.66 were obtained using Castle's and Burton's methods, respectively. These estimates may be biased by epistatic effects, but suggest that a low number of loci are involved in the expression of petiole thickness. The F2 and backcross generations were partitioned in- to thick and thin types based on the mean of the two parents (7.6 mm), resulting in a 9:7 segregation pattern. The F2 and BC2 populations showed good fits to the ratios expected from this model (P = .so - .30, P = .10 - .05, respectively), 42 TABLE 18 THEORETICAL F2 MEANS FOR ONE, TWO AND THREE GENE PAIRS ASSUMING COMPLETE DOMINANCE, 1982 No. of genes Theoretical Observed 1 (3/4)?1 + (1M)?2 = 8.011.54 8.212.05 2 (15/16)P1 + (V15)?2 = 8.311.54 8.212.05 3 (63/64)Pl + (1/64)P2 = 8.411.554 8.212.05 supporting the hypothesis that two major genes control petiole thickness in Pak-Choi (Table 19). The presence of individuals with thin petioles in the F1 and BC to P1 population could be due to modifiers affecting this trait. TABLE 19 CHI-SQUARE TEST FOR GOODNESS OF FIT FOR POOLED POPU- LATIONS FROM THE CROSS MSU 78-7A X MSU 78-9A, BASED ON A TWO GENE PAIR, 9:7 SEGREGATION HYPOTHESIS FOR PETIOLE THICKNESS, 1982 Generation Observed Theoretical Chi-sq. P ratio ratio Pooled F1 40:14 1:0 - - Pooled F2 160:142 9:7 1.31 .50-.30 Pooled BC to P1 141:11 1:0 - - Pooled BC to P 45:96 1:3 3.60 .10-.05 2 Narrow sense heritability was estimated at 5315 per- cent in 1981 and 60114 percent in 1982, suggesting that improvement can be expected for petiole thickness. Expected 43 gain from selection of the top ten percent of the population was estimated at 10.5 percent in 1981, 13.1 percent in 1982. The following genetic model is proposed: Two major loci, designated A and B determine petiole thickness. The proposed genotype of MSU 78-7A (P1) is AABB (thick), while that of MSU 78-9A (P2) is aabb (thin). The F1, with a geno- type of A322, has thick petioles. The observed F2 and back- cross ratios suggest a two gene, 9:7 epistatic model. A:§: conditions thick petioles. Each gene is epistatic to the other such that recessive homozygosity at either or both loci, conditions thin petioles. Dominance of genes A and B is in- complete, allowing for intermediate petiole thickness. The F2 and backcross genotypes AABB, AABb, AaBB, and AaBb produce thick petioles, while genotypes AAbb, Aabb, aaBB, aaBb and aabb produce thin petioles. Petiole Curvature - 1981 The F1 and F2 frequency distributions were skewed towards the curved parent (P1) (Figure 6). Fifty-five percent of the F2 population had curvature indexes higher than the midparent mean (.32), suggesting partial dominance of curved over flat petioles (Appendix Table 32). The F1 mean was significantly higher than the midparent mean (F1 - midparent = .11). The F2 mean, although not significantly different from the midparent mean, suggested dominance. The mean of BC to P1 44 FIGURE 6 FREQUENCY DISTRIBUTIONS FOR PETIOLE CURVATURE FROM THE CROSS MSU 78-7A X MSU 78-9A, 1981 FREQUENCY 45 I5 45 I5 75 25 75 3‘2 25 I50 50 CURVATURE INDEX 46 was higher than the Pl mean, probably due to transgressive segregation. The BC to P2 mean fell between the F1 and P2 means and was significantly different (5% level) from both populations. TABLE 20 SIGNIFICANCE OF THE A, B, C SCALING TEST FOR ADAXIAL CURVATURE IN THE CROSS MSU 78-7A X MSU 78-9A, 1981. Test A B C n5 ** I ** **, significant factors B and C ns, non-significant A significant value was obtained for factors Bsum.c in Mather's A,B, C scaling test suggesting the possibility of epistasis in determining petiole curvature. The poor fit to the additive-dominance model made it necessary to use a six parameter model to determine epistatic effects. The six para- meter estimates (Table 21) suggest that, while all effects were significant at the 1% level, dominance gene effects made the major contribution to variation in petiole curvature. Estimates of the number of effective factors control- ling adaxial curvature were calculated as 2.38 and 2.09 using Castle's and Burton's methods, respectively. Theoretical F2 means were calculated based on the number of genes assumed to be differentiating the parents, using the formulae of 47 TABLE 21 GENE EFFECTS ESTIMATED USING A SIX PARAMETER MODEL ON THE GENERATION MEANS FOR THE CROSS MSU 78-7A X MSU 78-9A, 1981 Effect Estimate m 0.3491.0001** a 0.2581.0010** d 0.7361.0219** aa 0.1401.0056** ad 0.0921.0016** dd -0.187i.0252** **, Significant at 1% level Powers, et a1. None of the resulting theoretical means were significantly different from the observed F2 (Table 22), how- ever, based on the indication of the presence of epistasis, it appears that more than one gene is responsible for determining adaxial curvature. TABLE 22 THEORETICAL F2 MEANS FOR ONE, TWO AND THREE GENE PAIRS ASSUMING COMPLETE DOMINANCE, 1981 No. of genes Theoretical Observed 1 (3/4)E1 + (1/4)?2 = .4o:.19 .35 1.16 2 (15/16)Pl + (1/16)P2 = .461.21 .35 1.16 3 (63/64)P1 + (1/64)?2 .481.22 .35 1.16 48 The F2 and backcross generations were partitioned into curved and flat types based on the midparent mean (.32), re- sulting in a 9:7 segregation pattern (Table 23). A good fit to a 9:7 epistatic model was observed for both the F2 (P = .90 - .50) and BC2 (P = .50 - .10) populations. TABLE 23 CHI-SQUARE TEST FOR GOODNESS OF FIT FOR POOLED POPU- LATIONS FROM THE CROSS MSU 78-7A X MSU 78-9A BASED ON A TWO GENE, 9:7 SEGREGATION HYPOTHESIS FOR ADAXIAL CURVATURE, 1981. Generation Observed Theoretical Chi-sq. P ratio ratio Pooled F1 120:12 1:0 - - Pooled F2 276:227 9:7 .389 .90-.50 Pooled BC to P1 156:39 1:0 - - Pooled BC to P2 33:130 1:3 1.974 .50-.10 Petiole Curvature - 1982 As in 1981, the F1 and F2 means were skewed towards the curved parent (Pl) (Figure 7). Sixty-three percent of the F2 had values higher than the midparent mean (.45), suggesting a partial dominance of curved over flat petioles. Similar dominance patterns are discernible from the means of the segregating populations (Appendix Table 33). The F1 and F2 means were higher than the mean of the two parents. The mean of the BC to P1 was higher than the Pl 49 FIGURE 7 FREQUENCY DISTRIBUTIONS FOR PETIOLE CURVATURE FROM THE CROSS MSU 78-7A X MSU 78-9A, 1982 FREQUENCY 20 10 I5 75 25 I00 50 I00 50 20 I0 .2 .4 .6 ’ 8 '00 '02 '0‘ CURVATURE INDE X 1.6 1.8 >2.0 51 mean, probably due to tranSgressive segregants as noted in the previous year. The BC to P2 mean fell between the F1 and P2 means and was significantly different (5% level) from the mean of both populations. TABLE 24 SIGNIFICANCE OF THE A, B, C SCALING TEST FOR ADAXIAL CURVATURE IN THE CROSS MSU 78-7A X MSU 78-9A, 1982 Test A B C ns ns ** **, significant at 1% level ns, non-significant at 5% level A significant value was obtained for factor C in Mather's A, B, C scaling test suggesting the presence of epistasis in the expression of petiole curvature (Table 24). Six parameter estimates (Table 25) showed results similar to the previous year. While all effects were significant, dominance gene effects appeared to be the most important in contributing to variation in petiole curvature. Two estimates of k were computed. Estimates of 1.22 and 2.02 genes were obtained using Castle's and Burton's methods, respectively. Theoretical F2 means were calculated using the formulae of Powers, et a1., and are presented in Table 26. As in the previous year, none of the theoretical F2 means were significantly different from the observed mean. 52 TABLE 25 GENE EFFECTS ESTIMATED USING A SIX PARAMETER MODEL ON THE GENERATION MEANS FOR THE CROSS MSU 78-7A X MSU 78-9A, 1982 Effect Estimate m 0.561:.0002** a 0.3201.0007** d 2.220:.0158** aa -0.9481.0006** ad 0.0881.0014** dd 0.2911.0534** ** TABLE 26 THEORETICAL F2 MEANS FOR PETIOLE CURVATURE FOR ONE, TWO AND THREE GENES ASSUMING COMPLETE DOMINANCE No. of genes Observed 1 (3/4)?1 + (1/4)§2 = .57;.2o .56 1.24 2 (15/16)f>1 + (1/16)P2 = .651.20 .56 1.24 3 (63/64)Pl + (1/64)P2 = .681.20 .56 1.24 Th F2 and backcross generations were partitioned in- to curved and flat types based on the midparent mean (.45), resulting in a 9:7 segregation pattern similar to that observed in 1981. Table 27 presents the results of a chi-square test for goodness of fit to a 9:7, two gene epistatic model. 52 TABLE 27 CHI-SQUARE TEST FOR GOODNESS OF FIT FOR POOLED POPU- LATIONS FROM THE CROSS MSU 78-7A X MSU 78-9A BASED ON A TWO GENE, 9:7 SEGREGATION HYPOTHESIS FOR ADAXIAL CURVATURE, 1982 Generation Observed Theoretical Chi-sq. P ratio ratio Pooled Fl 45:9 1:0 " ' Pooled F2 192:110 9:7 6.58 .05-.01 Pooled BC to P1 132:20 1:0 - - Pooled BC to P2 43:98 1:3 2.27 .50-.10 The poor fit of the F2 population to the expected ratio as compared to 1981 was probably due to the differential expression of modifier genes controlling adaxial curvature in 1982. Narrow sense heritability was estimated at 2416 percent in 1981 and 2017 percent in 1982. Expected gain from selec- tion of the top 10 percent of the population was estimated at 39 percent in 1981, 48 percent in 1982. The following genetic model is proposed: Two major loci, designated C1 and C2, determine adaxial curvature. The proposed genotype of MSU 78-7A (P1) is C1C1C2C2 (curved), while that of MSU 78-9A (P2) is clclczcz (flat). The observed segregation ratios suggest a two gene 9:7 recessive repressor model, in which the homozygous recessive condition for either or both loci, conditions formation of a flat petiole. CI_C2_ 53 conditions curved petioles. The F1, with the genotype ClClCZCZ' is curved. Dominance of genes Cl and C2 is incom- plete, allowing intermediate levels of curvature. The F2 and backcross genotypes C1C1C2C2, C1C1C2c2, ClchZCZ and ClClCZCZ produce curved petioles, whereas the genotypes ClClCZCZ' Clclczcz, Clclczcz, clclczcz and clclczcz produce flat petioles. Correlation coefficients between the three characters and bolting stage were calculated from F2 generation data. All correlations were low indicating that there is little or no relation between these characters and bolting. The only signi- ficant (.001 level) coefficient obtained was .20 for petiole width and bolting stage. As petiole width increased, there was a tendency towards premature flower formation. This would suggest that difficulties may be encountered in attempting to develop bolt resistant cultivars with wide, thick petioles. SUMMARY AND CONCLUS IONS A technique for determining width, thickness and curvature of Pak-Choi petioles was developed. Preliminary studies indicated that valid estimates of these characters could be obtained by sampling the fifth, seventh and ninth petioles, at a distance 9 cm from the base of each petiole. The progenies of crosses between MSU 78-7A and MSU 78-9A were evaluated in two field studies (1981 and 1982) to determine the inheritance of petiole width, thickness and curvature. Petiole width was determined to be simply inherited, with narrow petioles being dominant over wide. Petiole thick- ness was found to be controlled by two major genes, with thick petioles partially dominant over thin petioles. Domi- nance gene effects appeared most important in the expression of petiole thickness, although additive gene effects were also significant (1% level). Additive X dominance effects were significant in 1981, but not in 1982. Narrow sense heritability was estimated at 53:5 percent and 60114 percent in 1981 and 1982, respectively. Expected gain from selection was estimated at 10.5 percent in 1981 and 13.1 percent in 1982. 54 55 Ratios obtained by partitioning the segregating generations into thick and thin types based upon the arith- metic mean of the two parents, suggested a two major gene system controlling petiole thickness. A 9:7 recessive suppressor model best explained the observed ratios. MSU 78-7A (P1) was designated AABB (thick), while MSU 78-9A (P2) was designated aabb (thin). The data suggested the presence of modifier genes, which may have affected segre- gation ratios and estimates of various genetic parameters. Adaxial curvature of the petiole was also determined to be controlled by two major loci, with curved petioles being partially dominant over flat petioles. Both dominance and additive gene effects were important in the expression of this trait, however, estimates of dominance gene effects were greater in both years. In addition, significant addi- tive X additive, additive X dominance, and dominance X dominance epistatic gene effects were noted in both years. Narrow sense heritability was estimated at 24:6 percent in 1981 and 2017 percent in 1982. Expected gain from selection was estimated at 39 percent and 48 percent in 1981 and 1982, respectively. Ratios obtained by partitioning the segregating generations into curved and flat types based upon the arith- metic mean of the parental lines, suggested a two major gene system controlling adaxial curvature. A 9:7 recessive 56 supressor model best explained the observed ratios. MSU 78- 7A (P1), was designated C1C1C2C2 (curved), MSU 78-9A (P2) was designated clclczcz (flat). The results of this study show that petiole width, thickness and adaxial curvature are heritable characters. The information obtained on the genetics of these traits and their relation to one another is essential for the development of cultivars adapted to Michigan growing conditions. Of the three characters investigated in this study, the most important is petiole width. Petiole thickness and curvature, though important, are not as apparent as petiole width. For this reason, plants exhibiting narrow petioles are less desirable since the consumer prefers plants with wide petioles. Since wide petioles are recessive, following hybri- dization, one should select for wide petioles for several generations or until such time as the modifiers become homozygous. Selection should include the other characters only after petiole width has been considered. The dominance of thick over thin and curved over flat petioles should aid in the selection of desirable recombinants. Though herit- ability for petiole curvature was low (20-24 percent), it is the least important of the three characters, since consumers prefer wide, thick petioles. The addition of new germplasm to the Pak-choi breeding program will aid in obtaining the 57 desired recombination of bolting resistance and petiole characteristics. CHAPTER II THE INHERITANCE OF THE "UMBRELLA" BRANCHING HABIT IN PEPPER, CAPSICUM ANNUUM L. INTRODUCTION Increase in harvest costs or lack of available labor may necessitate the development of mechanical harvesters for many vegetable crops. The development of a mechanical har- vester for the pepper has been designed based primarily on the architecture of existing pepper cultivars. Problems associated with mechanical harvesting of peppers include difficulty in removing fruit from the plant or from low bearing plants, non-uniform maturity, breakage or branches and uprooting of plants by the action of the harvester. Modifications of existing harvesters to overcome these problems would be a possible solution. Instead, the plant breeder has been charged with the responsibility of attempting to breed a plant adaptable to the harvester. This would require a plant architype exhibiting concentrated fruit set, whereby the fruit would mature uniformly, and with fruit set high off the ground to facilitate harvest operations. It should also possess the necessary horticultural characteris- tics, resistance to diseases and insects, etc. The determinate growth habit reported by Hungarian workers (Kormos and Kormos, 1956; Ferenc, 1970 and Hristov, 58 59 et al, 1975), would appear to fit the description for a plant architype best suited for mechanical harvest. Where- as most pepper cultivars exhibit a dichotomous growth pattern, with fruits set at branching points, and over a long duration, the Hungarian types exhibit a compact, bunchy habit, with a determinate apex and axillary branches producing clusters of fruit over a short duration. The ideal solution would be to breed a pepper plant of this type, adaptable to harvesting with existing or slightly modified machines. The possibility of achieving this goal was realized with the appearance of a particular plant habit among the MSU breeding population. These plants exhibited determinate growth, with a cluster of fruit at the apex, and a subsequent branching pattern which resulted in the production of addi- tional clusters of flowers at the apex of each branch, all maturing at approximately the same time. This plant habit has the advantage of producing the fruit at the periphery of the plant, away from the branch joints. This plant habit could be extremely useful in mechanical harvesting systems, allowing an early hand harvest of the central cluster, followed by a once-over harvest of the remaining fruit upon maturity. Information on the genetic control of this branching pattern would be helpful in designing an efficient breeding 60 scheme for the development of desirable phenotypes. The objectives of the present study were to determine the inheritance of the "umbrella" branching habit in peppers, and to study the relationship between branching and yield. REVI EW OF LI TERATURE Yield of crop plants is a complex trait affected by both genetic and environmental factors. Efforts to improve yield through selection of superior genotypes is based upon phenotypic appearance, which can be greatly affected by environment (Singh and Singh, 1974). Consequently, empha- sis has been placed upon determining the nature and importance of genetic factors contributing to yield in Capsicum species. The genus Capsicum includes many cultivated types of peppers, such as bell peppers, pimentoes, small fruited pungent types for fresh market, and large fruited chili and paprika for dehydration. They are grown all over the world under both natural and irrigated conditions (Arya and Saini, 1977). Due to the fact that peppers are an important part of the diet and spice trade in India, a large portion of the research on various phases of growth and development of Capsicum species has been conducted by researchers in that country. The literature review will be confined to branching and its effects on fruit yield. All studies were conducted 61 62 using cultivars of Capsicum annuum, unless otherwise stated. Singh and Singh (1970) conducted a study with chilies on the interrelationships between fruit yield and characters such as plant height, number of primary and secondary forks, and length, width and number of fruit. They found no signi- ficant correlations between plant height or numbers of primary and secondary forks with fruit yield, and concluded that these characters may have no direct effect on yield. In a later study, Singh and Singh (1974) conducted a path coefficient analysis in which data on plant height, number of branches per plant, days to flower, days to maturity, fruit length and thickness, fruit number per plant, and yield per plant were used for analysis. Contrary to the previous study, the results indicated that the number of branches per plant is the most important component of fruit yield in chili. Soh, et al (1976), in a study of the relationship between yield and its components in chili, produced a seven parent diallel cross population in which they observed days to flower, total number and weight of fruits, weight per fruit, fruit length, width and wall thickness, and plant height. They reported high correlations between total fruit number and yield, and found total number of fruit to be con- trolled by additive gene action. 63 Singh and Singh reported conflicting results in two studies regarding gene action for yield and its components in chili. In the first study (Singh and Singh, 1976a), their results indicated that both additive and dominance gene action were equally important in determining the number of branches per plant, while additive gene action was the major contributing factor in determining fruit number. The subsequent study (Singh and Singh, 1976b) indicated that dominance gene effects contributed substantially to number of branches per plant, while both additive and dominance gene effects were important in determining number of fruits per plant. Epistatic effects were also noted, with dominance X dominance interactions acting in determining the number of branches per plant, and additive X dominance interactions present for fruit number. Awashti, et a1, (1976) also studied components of genetic variability in chili, as well as heritability and genetic advance for various characters. Data was recorded for plant height, number of branches per plant, number of fruits per plant, fruit length, diameter and weight, and fruit yield per plant. The results suggested high herit- ability of all traits except number of fruits per plant, but with low genetic advance for number of branches per plant, indicating non-additive gene action for this trait. 64 Arya and Saini (1977) on the other hand, obtained high heritability estimates and high genetic advance values for fruit number per plant, and high heritability coupled with moderate genetic advance estimates for number of branches per plant. More recently, Bavaji and Murty (1982) obtained high heritability and genetic advance values for both number of branches and number of fruits per plant. Yield was significantly correlated with both traits. Mishra, et al, (1976) observed eight parental lines and their Fl's for several yield contributing traits, among them, number of primary and secondary branches per plant, and number of fruits per plant. The increase in vigor of the Fl's over the better parent was significant for all three characters. Dominance was observed for both primary and secondary branches per plant. Positive association was observed between yield per plant, number of fruits per plant and number of primary branches per plant. Path coefficient analysis conducted by Dutta, et a1, (1979) in chili, indicated that the number of fruits per plant exercises high positive direct effects on fruit yield, while number of primary branches exhibits negative direct effects on yield. In contrast, Rao and Chhonkar (1981) found both number of branches and fruit number per plant to be positively and significantly associated with ripe fruit yield in chili, 6.5 indicating that yield can be improved either by selection of plants with more branches or plants with more fruits. Several investigations of the inheritance of other traits in peppers have been reported. Dale (1931) studied the inheritance of a dwarf plant habit arising from a cross between the cultivars Coral Gem and Anaheim, and determined the trait to be a simply inherited recessive character. Deshpande (1944) conducted an analysis of the inheritance of bunchy habit in chili. This mutant was described as bushy and compact with shortened internodes, and flowers and fruit produced in clusters. The bunchy character was shown to be recessive, and was later termed fasciculate (fa) by Lippert, et al, (1965, 1966). Kormos and Kormos (1956) crossed a fasciculate type of pepper with various non-fasciculate types and recovered F2 progeny exhibiting completely determinate growth. In these plants, the main axis stopped growing early in the season and produced a cluster of fruit, after which no lateral shoots developed. This character was found to be recessive. This plant habit was considered ideal for mechani- cal harvesting, and a breeding program was begun in Hungary to incorporate the determinate trait into desired genotypes. According to Ferenc (1970), the resulting plant types attain a height of 20 to 25 cm, terminate growth, and produce a 56 large number of flowers, from which clusters of fruit develop, and ripen over a short period of time (two to three weeks). Continued selection resulted in the release of an early, simultaneously ripening, determinate cultivar 'Buketen,’ suitable for once-over mechanical harvest (Hristov, 1975). The inheritance of bearing habit in C. frutescens, L. was determined from a cross between two tabasco type culti- vars, LP-l and Almeda, in which the characters cluster vs. non-cluster fruit set were studied. Results indicated that the cluster habit was controlled by a single recessive gene (Barrios and Mosokar, 1972). Two genetic studies have been conducted on branching in peppers. Bergh and Lippert (1975) studied three species, C. baccatum L., C. chinense Jacq. and C. frutescens L., for the inheritance of axillary shooting prior to the first bifurcation or point of branching. They observed mutant plants of C. chinense with extensive prebifurcation axillary shoots and results of intra- and inter-specific crosses with these plants indicated a monogenic recessive basis for the trait, which they termed compact (gt). Considerable environ- mental effects on the expression of axillary shooting were observed. Shifriss and Hakim (1977) studied the inheritance of prebifurcation shooting in C. annuum L. They crossed Santaka, 67 a Japanese cultivar with many axillary shoots developing acropetally along the main stem, and exhibiting the fasci- culate bearing habit, with Csokros Fellalo, an Hungarian bush fasciculate cultivar, devoid of axillary shoots, and Yolo-Wonder Y, with few axillary shoots prior to the first bifurcation. In the F1 and F2 from Santaka X Csokros Fellalo, the progenies showed partial dominance of many over few axillary shoots, while in the F1 and F2 from Santaka X Yolo-Wonder Y, the opposite occurred, i.e., partial dominance of few over many shoots. They concluded that prebifurcation shooting, though a quantitative character, is controlled by relatively few genes with different action, and modified by environmental conditions. MATERIALS AND METHODS Three inbred Capsicum annuum L. lines, MSU 78-101, MSU 79-221 and MSU 74-230 were used in this study (Figure 8). MSU 78-101 is an upright, dwarf banana pepper carrying the "fasciculate" or clustered gene (£3). MSU 79-221 was a selection from the cross MSU 74-190, a yellow long banana pepper, X Csokros Fellalo, a Hungarian upright, yellow bell, also carrying the fasciculate gene. It was selected on the basis of its unique branching habit, termed "umbrella." Upon attaining a height of approximately 15 cm, the apex terminates in a cluster of 2-6 fruits followed by initiation of several lateral branches, each of which produces a termi- nal cluster of fruit, that mature uniformly. The umbrella phenotype has been designated (Em). MSU 74-230, a yellow, long banana, was an F5 selec- tion from the cross between Avelar, and a yellow long banana MSU selection. It is an indeterminate, non-clustered fruiting, yellow banana type, with a prolific branching habit, and produces fruit continuously and, therefore, matures over an extended period. 68 69 FIGURE 8 PLANT HABIT OF THE THREE PARENTAL LINES Top, MSU 78-101, Dwarf. (ctctdtdtfafaSuSu) Center, MSU 79-221, Umbrella. (ctctdtdtfafasusu) Bottom, MSU 74-230, Indeterminate (ctctDtDtFaFasusu) 70 71 Hybridization Crosses were made in the greenhouse to produce reci- procal populations for the genetic studies. One plant of each parental line was selected for hybridization with the other two parents. All generations were produced using these single plants and their Fl's. In the case of MSU 78-101, flowering occurred over a 10-14 day period, followed by very little vegetative development and occasional flowering. Therefore, it was impossible to obtain backcross progenies using MSU 78-101. Seeds of parents and progenies were sown in vermicu- lite on April 13, 1982, transplanted into peat pots on May 1 and 2, and into the field in June, 1982. A completely ran- domized design with three replications was used. MSU 78-101 X MSU 79-221 Due to a short period of flowering of the MSU 78-101 parent, it was impossible to obtain reciprocal generations for this study. Using the dwarf as the female parent, F1’ F2 and BC2 populations were produced. Each replication consisted of 12 plants of each parent, 12 of each F1 family, 72 of the backcross to MSU 79-221, and 96 of the F2 popula- tion. Rows were spaced 1.07 m apart, with plants spaced at intervals of 0.60 m in the row. Cultural practices utilized were those recommended for pepper production in Michigan 72 Plants were classified by growth habit (indetermi- nate vs. determinate) and fruit bearing habit (non-cluster vs. cluster). The means of reciprocal populations were tested for homogeneity prior to analysis of the data. MSU 79-221 X MSU 74-230 Each replication consisted of 12 plants of each parent, 24 of each Fl family, 36 of each backcross population and 96 of each F2 family. Cultural practices were identical to those described for the previous cross as were procedures of data collection and analysis. The means of reciprocal populations were tested for homogeneity prior to analysis of the data. The possibility of linkage necessitated the use of a x2 contingency test in which the test is separated into three components using the following formulae (Bonnier, 1942): Aa segregation component x2 = a+b - 3(c+d) 2 3N (1° of freedom) Bb segregation component x2 = a+c - 3(b-d) 2 3N (1° of freedom) Linkage component x2 = (a - 3b - 3c + 9d)2 9N (1° of freedom) Where classes a, b, c and d = phenotypes AB, Ab, aB, and ab, respectively. Linkage intensities were calculated using the pro- duct method (Fisher and Balmukand, 1928). 73 MSU 78-101 X MSU 74-230 Reciprocal populations of F1 and F2 were produced for use in this study. However, the F1 was backcrossed only to MSU 74-230 due to an absence of flowers caused by the short duration of flowering in the dwarf parent (MSU 78-101). Each replication consisted of 12 plants of each parent, 12 of each F1 family, 72 of the backcross to MSU 74-230, and 96 of each F2 population. All procedures of data collection and analysis were identical to those described previously. Branching and Fruit Number Data were recorded prior to frost for number of fruit at the apical cluster, number of primary, secondary and tertiary branches, and number of fruit per respective branch. Primary branches were identified as those arising directly from the main axis. While secondary branches were those arising from primary branches and tertiary branches those arising from secondary branches. Total fruit per plant (branch fruit plus apical cluster fruit) was calculated for each plant. The F2 population was separated by plant habit and level of branching in order to determine the relationship between branching and the number of marketable fruit produced. Determination of fruit marketability was based upon the size of the mature fruits on the individual plant. 74 The resulting classes were tested for significant differences in number of marketable fruit by means of a two- tailed t-test (Steel and Torrie, 1960). RESULTS AND DISCUSSION Plant Type and Bearing Habit MSU 78-101 X MSU 79-221 The segregating populations were separated on the basis of plant habit prior to genetic analysis. The F2 generation segregated into 103 dwarf : 4O indeterminate 47 umbrella, approximating a 9:3:4 ratio (x2 = .682, P = .80-.70, Table 34). The expected ratio for the backcross to the umbrella parent would be 2 umbrella : 1 dwarf : 1 indeterminate. Out of 188 plants, a ratio of 99 : 49.5 49.5 would be expected. The data show a good fit to the expected ratio (P = .30-.20, Table 34). The observed Fl ratio could not be tested due to the zero expectation for the umbrella class. TABLE 34 CHI-SQUARE TEST FOR GOODNESS OF FIT FOR POOLED GENERA- TIONS FROM THE CROSS MSU 78-101 X MSU 79-221 Generation Observed Theoretical Chi-sq. P Pooled Fl 34:9 1:0 - - Pooled F2 103:40:47 9:3:4 .682 .80-.70 Pooled BC to P2 89:59:50 2:1:1 2.838 .30-.20 75 76 There appear to be four genes involved in the expres- sion of the umbrella habit. Two genes, gt and gt affect plant habit. The Q; gene is a recessive gene conditioning determinate growth, while gt is a recessive gene controlling the number of axillary shoots (Bergh and Lippert, 1975). gt and gt are dominant genes conditioning indeterminate growth, and appear to be epistatic to one another, such that, in the homozygous dominant or heterozygous conditions, gt is epista- tic to the expression of gt, while Qt is epistatic to the expression of gt. A third gene, t3, controls the fasciculate or clustered fruit bearing habit. {3 causes non-clustered fruit set. Assuming that the genotype of MSU 69-221 is ctctdtdtfafa, and that of MSU 78-101 is CtCtdtdtfafa, the expected F2 ratio would be 3 dwarf : 1 umbrella. However, the appearance of indeterminate plants in the F2 suggests the presence of an additional factor(s) in determining the umbrella habit. Modifying the genotype of the dwarf parent to include an additional gene, §Ev a dominant suppressor, would result in a 9:3:4 segregation pattern. In this case, the genotype of MSU 79-221 would be ctctdtdtfafasusu, that of MSU 78-101, CtCtdtdtfafaSuSu. The St gene would act to suppress the epistatic action of gt. The genotypes Ct--dtdtfafaSu—- would produce dwarf plants, genotypes 77 ctctdtdtfafaSu-- and ctctdtdtfafasusu would produce umbrella types, and genotypes Ctdtdtfafasusu and thtdtdtfafasusu would produce indeterminate types. The presence of 9 umbrella plants out of 43 F1 plants is probably due to the effect of modifiers on the expression of the gt gene. The genotype of the F1 would be thtdtdtfafaSusu. Assuming the supressor gene inhibits the action of the Qt gene, modifiers affecting the expression of the gt gene could alter the amount of branching in a given plant, based on the interaction of these modifiers and the gt gene, resulting in a misclassification of a plant as an umbrella type. The proposed genotypes and phenotypes expected from this cross are presented below: CtCtdtdtfafaSuSu 9 Dwarf thtdtdtfafaSuSu CtCtdtdtfafaSuSu thtdtdtfafaSuSu CtCtdtdtfafasusu 3 Indeterminate thtdtdtfafasusu ' ctctdtdtfafaSuSu 4 Umbrella ctctdtdtfafaSusu ctctdtdtfafasusu MSU 79-221 X MSU 74-230 Plant type and fruit bearing habit were used to separate the F2 and backcross progenies into distinct classes for genetic analysis. The F1 plants were indeterminate and non-clustered, suggesting that the umbrella habit was a 78 recessive character. In the F2, the plants segregated into 4 classes: indeterminate, non-clustered; indeterminate, clustered; determinate, non-clustered and determinate, clus- tered. Plant habit appeared to be under a 2 gene dominant epistatic control, resulting in a segregation of 15 indeterminate : l determinate. Also, the non-clustered vs. clustered characters did not fit the expected 3 : 1 ratio. A preponderance of parental types suggested linkage between the determinate and fasciculate genes. This was supported by the poor fit of the data to the proposed (15:1) (3:1) model (x2 = 106.45, P = (.001, Table 35). However, this model gave the best fit of all models tested. The backcross of the F1 to the umbrella parent also showed a poor fit to the expected ratio (P = (.001, Table 35). The preponderance of parental types over recombinants in the backcross generation further suggested linkage between the genes determining plant growth habit and bearing habit. The backcross of the F1 to the dominant parent shows a good fit to the expected ratio. TABLE 35 CHI-SQ. TEST FOR GOODNESS OF FIT FOR THE CROSS MSU 79-221 X MSU 74-230, BASED ON A 4-GENE, EPISTATIC MODEL Generation Observed Theoretical Chi-sq P Pooled F1 136:0 1:0 - _ Pooled F2 440:31:19:20 45:15:3:1 106.45 (.001 Pooled BC to p1 245:43:3o:94 3:3:1:1 203.26 (.001 Pooled BC to P2 394:0 1:0 - - 79 TABLE 36 SEPARATION OF THE X2 TEST FOR GOODNESS OF FIT INTO ITS COMPONENTS Aa component x2 (indeterminate vs determinate=2.l7,P=.50-.10 Bb component x2 (non-clustered vs clustered)=59.89,P=<.OOl Linkage component x2 =33.91,P=<.001 Separation of the x2 test into its components showed a good fit to the expected 15:1 segregation for plant habit, but a poor fit for bearing habit (Table 36). Also, a sig- nificant value was obtained for the linkage component. Linkage studies using the product method indicated a cross- over value of 10.45:t3.82 percent. Therefore, it appears that the poor fit of the data to the proposed model is most likely due to the tight linkage between the genes conditioning indeterminate plant habit and non-clustered fruit bearing habit. Within the determinate class of the F2 and BCl popu- lations, both umbrella and dwarf types were recovered. The following genetic model is proposed: 3 major genes are involved in the inheritance of the umbrella habit. The proposed genotype of the umbrella parent is ctctdtdtfafasusu, where gt is a recessive gene controlling the number of axillary shoots (Bergh and Lippert, 1975). The gt and tg genes are recessive genes for determinate growth and clustered bearing habit, respectively. The proposed genotype 80 of MSU 74-230 is CtCtDtDtFaFasusu. Qt and gt are dominant genes causing indeterminate growth. It appears that, in the homozygous or heterozygous conditions, St is epistatic to the expression of gt, while Qt is epistatic to the expression of gt. However, it is also possible that both dominant genes are inherited independently, resulting in a 9:3:3:l segregation ratio for plant habit, but due to an inability to distinguish the intermediate classes from the dominant class, a 15:1 ratio is observed. The dominant suppressor, St, is in the homozygous recessive condition (gggg) in both parents and thus has no bearing on the expression of these characters in this cross. The proposed genotypes and phenotypes of the F2 generation are tabulated below: CtCtDtDtFaFasusu 45 Indeterminate, non-clustered CtCtDtDtFa susu CtCtDtthaFasusu CtCtDtthafasusu CtCtdtthaFasusu CtCtdtthafasusu thtDtDtFaFasusu thtDtDtFafasusu thtDtthaFasusu thtDtthafasusu thtdtthaFasusu thtdtthafasusu ctctDtDtFaFasusu ctctDtDtFafasusu ctctDtthaFasusu ctctDtthafasusu CtCtDtthafasusu 15 Indeterminate, clustered CtCtDtdtfafasusu CtCtdtdtfafasusu 81 thtDtthafasusu 15 Indeterminate, clustered thtDtdtfafasusu (cont'd) (cont'd) thtdtdtfafasusu ctctDtthafasusu ctctDtdtfafasusu ctctdtthaFasusu 3 Determinate, non-clustered ctctdtthafasusu (Umbrella and dwarf) ctctdtdtfafasusu l Determinate, clustered (Umbrella and dwarf) Dwarf plants arising from this cross would have the genotype ctctdtdtfafasusu or ctctdttha_-susu but would have reduced axillary shoots due to the action of modifier genes.. Recombination of modifiers associated with gt would allow for varying numbers of axillary shoots in different genetic backgrounds, resulting in dwarf plants with few or no axillary branches. MSU 78-101 X MSU 74-230 The plants in the segregating generations were classified by plant type and fruit bearing habit. The F1 plants were indeterminate and non-clustered, suggesting these characters to be dominant to the dwarf, clustered habit, as previously reported. The F2 segregated into 363 indeterminate, non-clusted : 53 determinate, clustered plants. Analysis of each of these two characters showed a segregation ratio of 13:3 for plant type (x2 = 4.33, P = .05-.01) and 3:1 for bearing habit (x2 = 2.95, p = .10-.05), 82 suggesting that bearing habit is controlled by a single gene, as reported by Deshpande (1944), while plant type appears to be controlled by two genes. As discussed in the previous crosses, the Qt and §E genes are present in both parents and are presumably involved in determining plant habit in this cross. The proposed genotype of MSU 78-101 is CtCtdtdtfafaSuSu, that of MSU 74-230 is CtCtDtDtFaFasusu. A cross involving parents with these genotypes would result in a segregation ratio of 39 indeterminate, non-clustered; 13 indeterminate, clustered; 9 determinate, non-clustered; 1 determinate, clustered. A chi-square test for this model gave a poor fit (x2 = 43.27, P = <.001, Table 37). TABLE 37 CHI-SQUARE TEST FOR GOODNESS OF FIT TO A 9:3:3:l EXPECTED RATIO FOR THE CROSS MSU 78-101 X MSU 74-230 Generation Observed Theoretical Chi-sq. P Pooled F1 72:0 1:0 - - Pooled F2 363:68:70:53 39:13:9:3 48.27 (.001 Pooled BC to P2 428:0 1:0 - - The predonderance of parental types in the segregating generations suggested the presence of linkage between the genes for determinate plant habit and fasciculate fruit set. In 2 test into its components, a significant separating the x value was obtained for the linkage component, supporting this view (Table 38). 83 TABLE 38 SEPARATION OF THE X2 TEST FOR GOODNESS OF FIT INTO ITS COMPONENTS Aa component x2(indeterminate vs determinate)= 4.33,P=.05—.01 Bb component x2(non-clustered vs clustered) = 2.95,P=.lO-.05 Linkage component x =40.99,P= <.001 Linkage studies using the product method indicated a crossover value of 31.701:2.58 percent. In the F2 generation, the proposed genotypes and pheno- types expected are tabulated below: CtCtDtDtFaFaSuSu CtCtDtDtFaFaSusu CtCtDtDtFaFasusu CtCtDtDtFafaSuSu CtCtDtDtFafaSusu CtCtDtDtFafasusu CtCtDtthaFaSuSu CtCtDtthaFaSusu CtCtDtthaFasusu CtCtDtthafaSuSu CtCtDtthafaSusu CtCtDtthafasusu CtCtdtthaFasusu CtCtdtthafasusu CtCtDtthafaSuSu CtCtDtthafaSusu CtCtDtthafasusu CtCtDtdtfafaSuSu CtCtDtdtfafaSusu CtCtDtdtfafasusu CtCtdtdtfafasusu CtCtdtthaFaSuSu CtCtdtthaFaSusu CtCtdtthafaSuSu CtCtdtthafaSusu 39 Indeterminate, non-clustered 13 Indeterminate, clustered 9 Determinate, non-clustered 84 CtCtdtdtfafaSuSu 3 Determinate, clustered CtCtdtdtfafaSusu Branching and Fruit Number MSU 78-101 X MSU 79-221 The F2 population was separated into the following classes: dwarf (branchless); dwarf with l-2 primary branches; indeterminate without secondary branches; indeterminate with- out tertiary branches; umbrella without secondary branches and umbrella without tertiary branches. The umbrella pheno- types exhibited primary branches which arose from axillary nodes rather than from bifurcation of the apical meristem. Significant differences in number of marketable fruit per plant were observed between the branchless dwarf classes and all other classes. There was no significant difference in the mean number of marketable fruit per plant between the dwarf class with primary branches and the indeterminated class without secondary branches. No significant differences were observed between the other classes for mean number of marketable fruit per plant (Table 39). A11 differences were significant at the .05 level. .m mnsmflm Eoum AmImv monocmun >Hmwuumu I m9 monocmun mnmocoomm I mm monocmun mumfiflum I m 85 I ~.m «.4 oa~.o b.5H em Ame ozv mflamunss Amy I I >.v om>.o m.vH I ma Amm ozv maamunfio Amy I m.v o.v 0mm.o m.mH mm Ame ozv muw:HEHmuwch Apv I I m.v obmh.a m.ma n Amm ozv mumcflfihmumocH ADV I I m.H an.o m.m oe Ammo mum3o any I I I m>N.o v.n mm Ammmanosmunv mumzo Adv ucmHm\mB pcmam\mm .ucmHm\mm ucmHm\uH5Hm 02 wow» Human Hmmlmh sz x HQHlmh sz mmomU mmB 20mm .mmmwfi Bz mmB mom BZdam mmm mmmUZHmwuumu I ma monocwun mumpcoomm I mm monocmun humEHum I mm 89 o.~ w.o m.m omm.H o.o~ m Amev mHHmunsa any I ~.o o.m ooao.~ o.- ma Ame ozv oHHmunea loo h.m m.o R.o omm.o o.e~ mod Amao muocfleumumocH Am. I H.m n.o oov.o o.oH mom Ame ozo muooflsumuoocH .61 I I m.o omn.H o.~H Hm .mw ozo muocflsumumocH Loo I I I ome.o m.o ma Ammmasocmuno mamas Ame unmamxma ucoaoxmm ucmflm\mm unmao\ufloum oz wasp ucoam ommIon omz x HmmImn am: mmomo use 20mm .mmmwe azaqm moon<> was mom az mmB mom BZfiAm mmm mmmozuou Howxwo< HmoH .moo Hmegng mom onaomHmamHo Nozmoommm Nm mamas 103 NN NNH. ooN. N NH NH N mm “How NN oN NoN. NNN. N H N N N NH N H Nu “How Nm HNH NNo. ooN. H NH NN NN Nom Nm x Hm NNH NoN. NNN. N H H H N HH NN NN NN H Hum Ho x Hm NoN NNN. ooNN. H H o H N 3 mm NoH NN HN N.,H NHom Hm NN NoN. omNN. H N N NH NH NH H HN «NINN om: x «NINN om: NN HNH. mNN. N N NN Ho NNINN om: NN NoN. NNN. H N N NH N Hm «NINN om: mocmHm m omoz o.N N.H N.H N.H N.H o.H N. N. N. N. ooHumuocoo moumHomo mo .02 meGH musum>uso HMHpré . NNNH .mmINN mm: x NNINN om: mmomo one mom mmxmozH mmoaN>moo HNmeom mom oneomHmsmHo Nozmoommm mm MHm