VT *-',— 32>": mp. r3.- “ ‘23-- M” “£11.”! 1 h ‘1|||"||‘~‘1.4‘7 '1 1"“Jlr-5‘5‘J ‘ lay MWY'VWY JfiJ‘l‘a'q" INT" )1. VDTT. 1 1-H!“ ' 'i“ -1' [:1 1" ' ‘ 11'3“ Jr} . ..1":.' y 11“.” 1 .1" .‘,| 1.. 1- ’1 . $1 '11:.1't‘m I ’3‘. 1 ,1 10'1.‘ , 1': ,‘ ‘ " “:4" LY |"'V§ n:-‘I1|’1"'I:.;|‘\ (’14:)‘3‘1 FL“ "'2 “J"? I: ‘54! U‘ .511.» . He-‘! 1:31:1"I‘ 'X‘I‘r’ Milt). ‘\: l ’ :1 J 1149' "V 10 > 4.. ' h 4,11?!“ [H.thl .. h ".7 .11 4,“ .|.' 1. j. . ' .- 1 I . 2 t" '_ 3| |i 14"” ‘ ' LI. '1 ”“4"” '1 1 34.4.4.4..siej‘iig'ifmm 1:741 ”WM 1‘111‘2'1 9|“ ‘ 4 | ' 4' " f, . . . ‘ c l ' ‘ I I'é'v‘HH' 4 $4 1"‘414‘51'1 ' |M| . 44411-3441 5 1‘ .1: 3.1% . \"l u‘ I.’ I 11. .tvh": 1 L‘s. 11.1,“: '1Vpl4' “1' 1| "1' . é! -. :-‘-" . ', ‘ ', _ . ,1 néj. 1.11.14 11 11.:‘.1~'1"i“"1“.?:11' ‘ ‘"‘"-"‘~ |"“"| ‘ | ,7 .1 "|14“fb .4 .44. 1'1'04‘ . “,4"! ' ' .1011: :1 1:1“? ‘ "1 -‘|‘ l | .v‘- i—‘" f; 5:31;; g. n .3 .. .4111 g} ' ' ‘tfifi‘ ' d. . 1 g? .. 1 '. - . 4“ ‘ 1d: g 43‘ V ":55 I||3i11.g$|i|'fi:l t5; V y "3‘: :2 1“ kt . K'Ffiz‘fig} ffg1q-l11...1‘~" I. 5 1. I! I | 6-; Juifl h1tim fi"“ '.4 5‘; t'956 P1441~1 " :4 4 - .11.;5412“ 1 113-1 ‘- ‘- L .4311 ‘ | ‘ .| 51-14; 1 than Q1334; 4M 3" £111 .141. £411 1W'9 rugrrfih :1:§ “ 11""14“W'1 “ 111h . Hy; .l..4'--._.. f... .. .-.._'-- -: .Vj‘l r - “k _-, . mar—J o . no... . ..~V‘44 . v-v ' _»..- ”sli.|'h:3 1"- i‘ : .l, I L i , ‘1' l I. 1‘:Ji3_ 1; IE" '13: E | 1';‘:|’"'§|\“>"h 14"." .r h ‘ L 1 ' _ -. .' 4- ‘31 ’1 ‘ 4 i 4 ‘ l-L - .1133"; ,A'i- '1|y.I‘-?‘- ' -:zi.' . 141.11 '14 . -- . IL‘K3";« 11W?! 1,4, ,1 , 4: .41 41111-0 frmJ-r 241 ‘ '1; " «:h - .1 u' l “1...... 4|: ~ ‘1111 .. g 1. 1-. ' ’4 ' , 4' l 4 L 4 ’4 9m“: 4...-1':’ 4 4. 1 1. Le”. L -. ‘. 7, .5, 4 .1 4 1:4,“. 1. -i' ‘ ' 4 3‘ ii l L' ' 4‘ ‘ 'L .. h ' ' 1""LE' . .1",|.“'(" ' ‘I :"h o :12;ng hL 1:.-vl_",r.y,!...r"1‘ J- . ‘ :. ‘ ' 1 - ' ' -_. c-. -4 ,gv-oov’v-N' , a ... . VA 7’ - . w 1. 4; . “'3‘ |‘a‘1" N 0': n r.— ~rov~ ‘1 ‘gfl ,x‘ :44 4.4411 :"'::; - I. _ y . ." I: \. ' " .11l‘1 t ’1‘-1 ' n 111.41 4 - «'1'-1111...". 12‘1" ~ 4144-; .1 ,1: .~; '-'11!'.' a. 4» ~ .111 "134' '43 4511131141 .1 24-441”. 11"”. Nov-1.21: 1 11.1471411111hu‘: - ‘J‘Hl 'l‘v': “it 'L'I.'1V1‘I"‘"I.: I |1||":||.1;‘1'1 I... ‘ ‘ ' ‘ 1%wanyflwLwM‘hl'm“uM ” ‘ i; 'qu . ’£|;|.||( 1.1“ 3|; ||~,7':” ”"0”1“ ’.1 0 ' .1" 11,11!- . 13%|? ,g.-_!-;~)1fi1 'Io;.|1'_4 “1 "Wu; .11.. Wm 4,. l. 1.‘ ”$1413 _ . d ||L|||d1lfldusi ”4:44" ’| .31 - ‘n' '1'1'4 . ‘ ' H l v 4111.4 ”(flit 4.311;”) l .7 gi. 1, . m".-, . '1" .‘" l‘ ‘ '- 4:3'111444'1-"1:1,1 141-?“ 2’3: 111:1.- 11:. 1,,44 ;; 4. ‘. 1429““ .'-4' ’1'. ",1 “nus“ .} " .u..'I.." ..":§‘ ’,‘4-,"1.:_fh 3‘ 1" 1 ":ILW 1;: “313175 1;. I ";nvqahfi”"”’-u{'1euhu1ut ~11u1 H'LgLU11 31tn43$$11nnm .1: 1 ‘r 4. 4:“ t “5" ' M 3,1 1'4]; L 4,. “1 IF“ ,1;?1?5.|§;-,i‘.5 0;, ghfiéhfi§4u1 Lguu'thfinfifn}EFu 4 10“fl|'| 'M! ' L“““4$hz*‘V?— . ‘ “.1, 1 . . - .. , 4:1 ||||||§L||1|||i|||||1':,|||=i.|."éé|| 44414424”: 11:? 1'11, 11111147?! . ‘ . .‘ . l' 51“.“! :‘o- ‘| ".‘V , a“ 5.1 "it: ‘ I |1 v {1 , . 4 4134U|44f ‘nfi'u1uhfifl4fi;44514‘ ' 4 “:4 5‘ EF‘U L ‘ H 1'13. '-1 3:11 11411414 11. '= «NI 1'4?“ 141 ” g“1.1.3 ‘ 4 n 1”Jufl1t"%“w:hr14-|~N”JI. 4 .| i’ V141 ‘ ”'11“??? L’i ”1|” 1‘ ‘2'”, ’ 441% 41““ F“ 1 1:1; ' 1|; 1 4.24 1 :_ ‘| i 1' 44 111141} 7.1"?" ' :‘5" 23:4 £11161“! "It 1 13 ‘1}le 41411}; 513.11% ”1.1 mg figl,‘ $41! 311% 44M44L4MWWfl4 b; 1:1' - ti . I \\_" “ f i Q Ixnll'lllllmlmmllllil 3 00908 ‘A--~ - .-.__._-—-“-" This is to certify that the dissertation entitled QUANHT-n‘rwe TQAIT wcu TubER TRAITS see) ANALYSIS OF L 1M b|PLDID POM“) (50“U‘W“ ‘ 1 presented by Rosa-ma Chew RE has been accepted towards fulfillment , of the requirements for ' degree in MEDIMG- i AM!) 5945:1195 MCD/C: Major prdfessor Date _§/?/¢3 i Pkb MS U is an Affirmative Action/Equal Opportunity Institution 0- 12771 LIBRARY Michigan State Unwerelty PLACE IN RETURN BOX to remove this checkout from your record. TO AVOID FINES return on or before date due. DATE DUE DATE DUE DATE DUE MSU Is An Affirmative Action/Equal Opportunity Institution cMmmunS—pj . 11 T1113?" 7F. , QUANTITATIVE TRAIT LOCI ANALYSIS OF TUBER TRAITS IN DIPLOID POTATO (Solanum spp) By Rosanna Freyre A DISSERTATION Submitted to Michigan state University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Plant Breeding and Genetics Program Department of Crop and Soil Sciences 1993 ABSTRACT QUANTITATIVE TRAI'I‘ LOCI ANALYSIS OF TUBER TRAITS IN DIPLOID POTATO (Solanum spp.) By Rosanna Freyre One breeding method for potato (Solanum tuberosum subspp. tuberosum) is using wild species. This method could be more efficient if the introgression of genes from these species were monitored with molecular markers. Furthermore, the use of molecular markers allows the dissection of quantitative traits into discrete genetic factors. The objective of this research was to perform quantitative trait loci (QTL) analysis on two tuber traits in potato: specific gravity and dormancy. Two diploid populations were constructed from heterozygous self-incompatible parents. These two populations, TRP132 (127 individuals) and TRP133 (110 individuals) have a common maternal parent and combine genomes of Solanum tuberosum (haploid), S. chacoense, and S. phureja. A preliminary analysis using isozymes was performed. QTLs were determined by one-way analyses of variances for each locus by trait combination (P < 0.05). Epistatic interactions were detected through two-way analyses of variance. Further studies focused on TRP133, which was characterized for 10 isozyme loci, 44 RFLPs and 63 RAPDs. Eighty-seven loci segregating from the female parent were utilized to construct a linkage map comprising 10 of the 12 chromosomes in the genome. For dormancy, 6 QTLs were identified that explained 57.5% of the phenotypic variation for the trait. Specific gravity was evaluated in 3 environments. QTLs were mapped separately for each location and in combination. A total of 10 QTLs on six chromosomes were identified. The numbers and effects of QTLs detected varied across environments, and they explained from 39% to 45% of the phenotypic variation for the trait. Using the average data a multilocus model was developed. This gives consistent results when tested across environments, and may be valuable for marker-assisted selection. This research developed the basic methodology for QTL analysis in potato which is now available for future studies with other traits and germplasms. To my parents, Ana and Alfredo, for all their love and support. ACKNOWLEDGMENTS I would like to acknowledge the many people that have helped me during the course of my doctoral studies and this research. I am very grateful to my advisor, Dr. Dave Douches for his support, confidence and friendship during these years. My sincere appreciation goes to the other members of my committee, Drs. Jim Kelly, Jim Hancock and Mike Thomashow. Every step along this research was totally new for me so I needed much help and advice along the way. I would like to thank friends in the Potato Group that have helped me in many ways, and particularly during field work: Donna Kells, Karen Hokanson, Dave Maas, Maywa Blanco, Drs. Kazmierz Jastrzebski and Dick Chase. My appreciation also goes to Theresa Woods and Dr. Kazmierz Jastrzebski for their assistance during the evaluation of traits. Thanks go to Drs. Steve Tanksley and Christianne Gebhardt for supplying the tomato and potato RFLP probes, respectively. Many thanks to Bryon Sosinski for optimizing the PCR protocol and screening of the parents with primers. The helpful advice of Jose Barbosa, Jorge da Silva and Susan McCouch at Cornell University during the construction of the map are greatly appreciated. My thanks go to Dr. Rodomiro Ortiz, Scott Warnke, Paul Fisher and Jeff Smeenk for their valuable help and advice during the statistical analysis and computer work. I am particularly grateful to Brian Diers for thoughtful discussion, letting me borrow his computer many times, and answering many questions during the last phase of my research. The financial support for this research by the Michigan Agricultural Experimental Station, the Michigan Potato Industry Commission, the National Potato Council and a specific cooperative agreement between Michigan State University and the USDA/Agricultural Research Service are greatly appreciated. Special thanks go to my parents, Alfredo and Ana, to Samira Daroub for her special friendship, and to Paul Fisher and many other family members and friends for all their love and support. vi TABLE OF CONTENTS LIST OF TABLES ..................................... LIST OF FIGURES ..................................... LIST OF ABREVIATIONS ................................. GENERAL INTRODUCTION . . .‘ ........................... List of references .................................. CHAPTER 1. Isoenzymatic identification of quantitative traits in crosses between heterozygous parents: mapping tuber traits in diploid potato (Solanum SPP.) Abstract ....................................... Introduction ..................................... Materials and methods ............................... Results ........................................ Discussion ...................................... List of references .................................. CHAPTER 2. Quantitative trait loci analysis of tuber dOrmancy in diploid potato (Solanum spp.) _ Abstract ....................................... Introduction ..................................... Materials and methods ............... ' ................ Results ........................................ List of references .................................. CHAPTER 3. Quantitative trait loci analysis of specific gravity in diploid potato (solanum spp.) Over environments: development of a model for marker- assisted selection Abstract ....................................... Introduction ..................................... Materials and methods ............................... Results ........................................ Discussion ...................................... List of references .................................. CONCLUSIONS ............................... . ....... vii O O x ..xi 5 LIST OF TABLES Chapter 1. Table 1.1. Isozyme genotypes for the parents and two populations ......... 17 Table 1.2. Values of specific gravity obtained for populations TRP133 and TRP132, and the two parents ........................... 19 Table 1.3. Significant association between specific gravity and isozymes for populations TRP133 and TRP132 ......................... 22 Table 1.4. Values obtained from the multiple analyses of variance for specific gravity for populations TRP133 and TRP132 .................. 23 Table 1.5. Significant association between tuber dormancy and isozymes for populations TRP133 and TRP132 ......................... 27 Table 1.6. Values obtained from the multiple analyses of variance for tuber dormancy for populations TRP133 and TRP132 ................ 28 Chapter 2. Table 2.1. Significant association between markers and tuber dormancy ...... 45 Table 2.2. Significant epistatic interactions between significant markers ...... 47 Table 2.3. Models used to determine the amount of phenotypic variation for tuber dormancy explained by QTLs and epistatic interactions ............ 48 Chapter 3. Table 3.1. Values of Specific gravity for parents and population TRP133 at each one of the three environments and the average ................. 60 Table 3.2. Mean squares (MS) from the combined analysis of variance of specific gravity for the three environments ........................ 62 Table 3.3. Significant loci, chromosome locations and R2 values for the three environments and the average .......................... 63 Table 3.4. Multilocus models developed at each one of the environments and with the average data .................................... 66 viii Table 3.5. R2 values obtained from the multiple analyses of variance using the multilocus models with data from the different environments ........ 67 LIST OF FIGURES Chapter 1. Figure 1.1. Frequency distribution of specific gravity values for TRP132 grown in Clarksville, Michigan 1990 ........................... Figure 1.2. Regression on the means for each genotypic class of 6-Pgdh-3 in family TRP132 (averaged over both locations) ................. Figure 1.3. Frequency distribution of dormancy (logmtransformed) values for TRP132 ----------------------------------------- Chapter 2. Figure 2.1. Frequency distribution of length of dormancy (logw transformed) values in population TRP133. . .......................... Figure 2.2. Molecular linkage map and localization of QTLs for tuber dormancy Chapter 3. Figure 3.1. Frequency distributions of specific gravity values at the three environments ..................................... Figure 3.2. Molecular linkage map and localization of QTLs for specific gravity for each environment and AVE .......................... 20 25 TRP132 TRP133 MES90 MES91 CHES90 AVE Isozymes: Dia-l Est-l Got- 1 Got-2 Idh-l Mdh-l 6-Pgdh-3 Pgi—l Pgm—l Pgm-2 Prx-3 RFLPs TG.. CD.. GP.. CP.. RAPDs LIST OF ABREVIATIONS 2x population derived from cross of 84SD22 x 84811 2x population derived from cross of 84SD22 x 84810 Montcalm Experimental Station, 1990 field trial Montcalm Experimental Station, 1991 field trial Clarksville Horticulture Experimental Station, 1990 field trial Average data, combined over 3 environments Diaphorase, allele 1 Esterase, allele 1 Gluconate Oxaloacetate Transaminase, allele 1 Gluconate Oxaloacetate Transaminase, allele 2 lsocitrate Dehydrogenase, allele 1 Malate Dehydrogenase, allele 1 6-Phosphogluconate Dehydrogenase, allele 1 Phosphogluconate Isomerase, allele 1 Phosphoglucomutase, allele 1 Phosphoglucomutase, allele 2 Peroxidase, allele 3 Restriction Fragment Length Polymorphisms Tomato genomic probes, Cornell University Tomato cDNA probes, Cornell University Potato genomic probes, Max Planck Institut Potato cDNA probes, Max Planck Institut Random Amplified Polymorphic DNA xi GENERAL INTRODUCTION The cultivated potato, Solanum tuberosum subsp. tuberosum, is one of the most important world food crops. It is cultivated in 130 countries and ranks fourth in volume of production in the world with approximately 290 million tons annually (FAO, 1992). This crop is tetraploid (2n=4x=48) with tetrasomic inheritance. One methodology utilized to simplify its genetic system is to breed at the diploid level which takes advantage of simple disomic inheritance. Over 70% of the wild and cultivated tuber- bearing Solanum species are diploids (Hawkes, 1990). These species have agronomically important attributes and can be easily crossed with haploids extracted from the cultivated species (Solanum tuberosum subsp. tuberosum or andigena). The improved 2x germplasm is then transferred to the 4x level using sexual polyploidization through 2n gametes (Chase, 1968; Iwanaga, 1983, Peloquin et al., 1989). This ploidy manipulation approach has been applied in several institutions in the world and has led to new cultivars released in USA and the International Potato Center (CIP) (Peloquin et al., 1989; Ortiz, 1991). One drawback to the use of 2x germplasm in potato breeding is the slow progress due to linkage drag of undesirable traits from the wild species. However, it has been suggested that the efficiency of breeding could be largely increased if the introgression of genes from the wild species could be closely monitored with molecular markers (Tanksley et al., 1989). Furthermore, the use of molecular markers for the analysis of quantitative traits has been described (Tanksley et al., 1982; Beckmann and Soller, 1988; v Paterson et al., 1988; lander and Botstein, 1989). In this study, two diploid populations combining genomes of haploid S. tuberosum and the wild species S. chacoense and S. phureja were utilized for quantitative trait loci (QTL) analysis of tuber traits using molecular markers. The traits studied were specific gravity, which is an indirect measurement for dry matter content, and tuber dormancy. Both these traits have importance for the potato industry in Michigan, and moreover they are relatively easily measured. These characteristics made them adequate to study the feasibility of applying QTL analysis in the Potato Breeding Program at Michigan State University. The first molecular markers available for study were isozymes. QTL analysis using these markers has been reported in maize (Stuber et al., 1980; Stuber et al., 1982; Pollack et al., 1984; Frei et al., 1986a; Kahler and Wehrhahn, 1986; Stuber et al., 1987) and tomato (Tanksley et al., 1982; Weller, 1987; Weller et al., 1988). However, their use as markers is limited due to the small numbers available. In potato, 15 enzyme-coding loci are presently known to segregate (Douches and Quiros, 1988a) and they had been previously utilized mostly for variety fingerprinting (Douches and Ludlam, 1991b), half- tetrad analysis (Werner et al., 1992) and systematic studies (Spooner et al., 1992). The first chapter of this study, Isoenzymatic identification of quantitative traits in crosses between heterozygous parents: Mapping tuber traits in diploid potato (Solanum spp), refers to QTL analysis with isozymes utilizing two populations, evaluating specific gravity in two environments, and also tuber dormancy. Another set of molecular markers with greater potential based on restriction fragment length polymorphisms (RFLPs) have the advantage that the number available is virtually unlimited (Beckmann and Soller, 1983; Helentjaris et al. , 1985). RFLPs maps have been 3 constructed for many crops including potato. The first map in potato was developed using tomato RFLP probes on an interspecific 2x population involving three diploid species: S. phureja, haploid S. tuberosum, and S. chacoense (Bonierbale et al., 1988). This map was further saturated with markers using a population involving haploid S. tuberosum and S. berthaultii (Tanksley et al. , 1992). Independently, another map was developed using a 2x S. tuberosum population and potato RFLP probes (Gebhardt et al. , 1989b). This map was also aligned with the homoeologous tomato genome (Gebhardt et al., 1991). RFLP markers have since been used to fingerprint potato lines (Gebhardt et al., 1989a; Douches et al., 1991a), to determine the phylogeny of wild and cultivated species (Debener et al., 1990) and to determine the extent of genetic variability in cultivars (Powell et al. , 1991). Linkage with two major genes conferring resistance to PVX (Ritter et al., 1991), a gene conferring resistance to cyst nematode (Barone et al., 1990; Gebhardt et al., 1993; Pineda et al., 1993), and three flower color loci (van Eek et al., 1993) have also been reported. Research involving genetic mapping of quantitative trichome-mediated insect resistance (Bonierbale et al. , 1992) and loci affecting tuberization (van der Berg et al. , 1992) are in progress. Recently, polymerase chain reaction (PCR)-based genetic markers have become available. A novel technique that rapidly generates and screens random DNA segments for polymorphisms between different genotypes was developed simultaneously by Welsh and McClelland (1990) and Williams et a1. (1990). This technique offers the advantages of less time and labor involved as compared with RFLP analysis. The randomly amplified polymorphic DNA (RAPD) markers have since been incorporated in linkage maps in tomato (Klein-Lankhorst et al., 1991) and Brassica (Quiros et al., 1991); have evolutj 1992) and t‘ ave, been utilized for characterization and cultivar identification in Brassica (Hu and Quiros, 1991; Boury et al., 1992; Kresovich et al., 1992), rice (Fukuoka et al., 1992), cocoa (Wilde et al., 1992), papaya (Stiles et a1, 1993) and apple (Koller et al., 1993; Harada et al., 1993). Linkage to resistance genes have been identified in tomato (Martin et al., 1991), lettuce (Michelmore et al., 1991; Paran et al., 1991) bean (Miklas et al., 1993) and oats (Penner et al., 1993). In potato, RAPD markers have been utilized in evolutionary studies (Cisneros et al., 1991); to detect gene introgression (Waugh et al., 1992); their segregations in 2x and 4x families have been studied (Quiros et al., 1993); and they have been utilized to screen somatic hybrids (Xu et al., 1993). The RFLP and subsequent RAPD analyses in this research focused on one population. Chapter 2, Quantitative trait loci analysis of tuber dormancy in diploid potato (Solanum spp), refers to the development of a linkage map including isozyme, RFLP and RAPD loci and the identification of QTLs associated with dormancy. The third chapter, Quantitative trait loci analysis of specific gravity in diploid potato (Solanum spp) over environments: development of a model for marker-assisted selection, refers to QTL analysis performed on this trait separately for three environments, and the use of the average data to develop a multilocus model to be used for marker-assisted selection. LIST OF REFERENCES Barone A, Ritter E, Schachtschabel U, Debener T, Salamini F, Gebhardt C (1990) Localization by restriction fragment length polymorphism mapping in potato of a major dominant gene conferring resistance to the potato cyst nematode Globodera rostochiensis. Mol Gen Genet 224: 177-182 Beckmann J, Soller M (1983) Restriction fragment length polymorphisms in genetic improvement: Methodologies, mapping and costs. Theor Appl Genet 67:35-43 Beckmann J, Soller M (1988) Detection of linkage between marker loci and loci affecting quantitative traits in crosses between segregating populations. Theor Appl Genet 76:228- 236 Bonierbale M, Plaisted R, Tanksley SD (1988) RFLP maps based on a common set of clones reveal modes of chromosomal evolution in potato and tomato. Genetics 120: 1095- 1 103 Bonierbale M, Plaisted R, Tanksley S (1992) Genetic mapping and utilization of quantitative trichome mediated insect resistance in potato. Neth J Plant Path 98:211-214 Boury S, Lutz I, Gavalda MC, Guidet F, Schlesser A (1992) Genetic fingerprinting in cauliflower by the RAPD method and determination of the level of inbreeding in a set of Fl hybrid seeds. Agronomic 122669-681 Chase SA (1968) Analytical breeding in S. tuberosum L. A scheme utilizing parthenotes and other diploid stocks. Can J Genet Cytol 5:359-363 Cisneros PL, Quiros CF (1991) Evolutionary studies of potatoes using RAPD markers. 2nd Intl Potato Mol Biol Symp, St Andrews, Scotland, August 11-15 Debener T, Salamini F, Gebhardt C (1990) Phylogeny of wild and cultivated Solanum species based on nuclear restriction fragment length polymorphisms (RFLPs). Theor Appl Genet 79:360-368 Douches DS, Quiros CF (1988) Additional isozyme loci in tuber-bearing Solanums: Inheritance and linkage relationships. J Hered 79:377-384 Douches D, Freyre R, Hicks K (1991a) Use of RFLPs to fingerprint North American Cultivars. In: Application of Molecular Techniques to Potato Germplasm Enhancement. International Potato Center (CIP). March 5-9, 1990. Douches DS, Ludlam K (1991b) Electrophoretic characterization of North American potato cultivars. Am Potato J 68:767-780 Food and Agriculture Organisation (1992). Production Yearbook. Vol 46. FAQ, Rome. 330 p Frei OM, Stuber CW, Wendel JF (1986a) Use of allozymes as genetic markers for predicting performance in maize single cross hybrids. Crop Sci 26:37-42 Fukuoka S, Hosaka K, Kamijima O (1992) Use of random amplified polymorphic DNAs (RAPDs) for identification of rice accessions. Jap J Genet 67:243-252 Gebhardt C, Blomendahl C, Schachtschabel U, Debener T, Salamini F, Ritter E (1989a) Identification of Zn breeding lines and 4n varieties of potato (Solanum tuberosum ssp tuberosum) with RFLP fingerprints. Theor Appl Genet 78:16-22 Gebhardt C, Ritter E, Debener T, Schachtschabel U, Walkemeier B, Uhrig H, Salamini F (1989b) RFLP analysis and linkage mapping in Solanum tuberosum. Theor Appl Genet 78:65-75 Gebhardt C, Ritter E, Barone A, Debener T, Walkemeier B, Schachtschabel U, Kaufmann H, Thompson RD, Bonierbale MW, Ganal MW, Tanksley SD, Salamini F (1991) RFLP maps of potato and their alignment with the homoeologous tomato genome. Theor Appl Genet 83:49-57 Gebhardt C, Mugniery D, Ritter E, Salamini F, Bonnel E (1993) Identification of RFLP markers closely linked to the H] gene conferring resistance to Globodera rostochiensis in potato. Theor Appl Genet 85:541-544 Harada T, Matsukawa K, Sato T, Ishikawa R, Niizeki M, Saito K (1993) DNA-RAPDs detect genetic variation and paternity in Malus. Euphytica 65:87-91 Hawkes JG (1990) The Potato. Evolution, biodiversity and genetic resources. Smithsonian Institution Press. Washington, DC. 259 p Helentjaris T, King G, Slocum M, Siedenstrang C, Wegman S (1985) Restriction fragment polymorphisms as probes for plant diversity and their development as tools for applied plant brwding. Plant Mol Biol 5: 109-118 Hu J, Quiros CF (1991) Identification of broccoli and cauliflower cultivars with RAPD markers. Plant Cell Rep 10:505-511 Iwanaga, M (1983) Ploidy level manipulation approach: development of diploid populations with specific resistance and FDR 2n pollen production. In: Present and future strategies for potato breeding and improvement. Report of the 26'“ Planning Conference, CIP. Dec. 1983, Lima, Peru. Kahler AL, Wehrhahn CF (1986) Associations between quantitative traits and enzyme loci in the F2 population of a maize hybrid. Theor Appl Genet 72:15-26 I ll—.. 1 ] Klein-lankhorst RM, Vermunt A, Weide R, Liharska T, Zabel P (1991) Isolation of molecular markers for tomato (Lycopersicon esculentum) using random amplified polymorphic DNA (RAPD). Theor Appl Genet 83:108-114 Koller B, Lehmann A, McDermott J M, Gessler C (1993) Identification of apple cultivars using RAPD markers. Theor Appl Genet 85 :897-900 Kresovich S, Williams JGK, McFerson JR, Routman EJ, Schaal BA (1992) Characterization of genetic identities and relationships of Brassica oleracea L. via a random amplified polymorphic DNA assay. Theor Appl Genet 85:190-196 lander E, Botstein D (1989) Mapping Mendelian factors underlying quantitative traits using RFLP linkage maps. Genetics 121:185-199 Martin GB, Williams JGK, Tanksley SD (1991) Rapid identification of markers linked to Pseudomonas resistance gene in tomato by using random primers and near isogenic lines. Proc Natl Acad Sci USA 88:2336—2340 Michelmore RW, Paran I, Kesseli RV (1991) Identification of markers linked to resistance genes by bulked segregant analysis: a rapid method to detect markers in specific genomic regions using segregating populations. Proc Natl Acad Sci USA 88:9828-9832 Mildas PN, Stavely JR, Kelly JD (1993) Identification and potential use of a molecular marker for rust resistance in common bean. Theop Appl Genet 852745-749 Ortiz, R (1991) Efficiency of potato breeding using 2n gametes; male sterility and Zn pollen in 4x potato. PhD thesis. University of Wisconsin, Madison. 291 p Paran I, Kesseli RV, Michelmore RW (1991) Identification of RFLP and RAPD markers linked to downy mildew resistance genes in lettuce by using near-isogenic lines. Genome 34:1021-1027 Paterson AH, Lander ES, Hewitt JD, Peterson S, Lincoln ES, Tanksley SD (1988) Resolution of quantitative traits into Mendelian factors, using a complete linkage map of restriction fragment length polymorphisms. Nature 335:721-726 Peloquin SJ, Yerk GL, Werner JE, Darmo E (1989) Potato breeding with haploids and Zn gametes. Genome 31:1000-1004 Penner GA, Chong J, Levesquelemay M, Molnar SJ, Fedak G (1993) Identification of a RAPD marker linked to the oat stem rust gene Pg3. Theor Appl Genet 85 2702-705 Pineda O, Bonierbale MW, Plaisted RL, Brodie BB, Tanksley SD (1993) Identification of RFLP markers linked to the H1 gene conferring resistance to the potato cyst nematode Globodera rostochiensis. Genome 36:152-156 Pollac loci an 1 179 Powell estimat Appl B Quiros 36netic Ritter l Chrom Genet SDI] e] Appl $13001 Sect. Stiles ampli. eor Stuber ChangE Geneti. Pollack LM, Gardner CO, Parkhurst AM (1984) Relationships between enzyme marker loci and morphological traits in two mass selected maize populations. Crop Sci 24: l 174- 1 179 Powell W, Phillips MS, McNicol JW, Waugh R (1991) The use of DNA markers to estimate the extent and nature of genetic variability in Solanum tuberosum cultivars. Ann Appl Biol 118:423-432 Quiros CF, Ceada A, Georgescu A, Hu J (1993) Use of RAPD markers in potato genetics - segregations in diploid and tetraploid families. Am Pot J 70:35-42 Ritter E, Debener T, Barone A, Salamini F, Gebhardt C (1991) RFLP mapping on potato chromosomes of two genes conferring resistance to potato virus X (PVX). Mol Gen Genet 227:81-85 Soller M, Brody T (1976) On the power of experimental designs for the detection of linkage between marker loci and quantitative loci in crosses between inbred lines. Theor Appl Genet 47 :35-39 Spooner DM, Douches DS, Contreras AM (1992) Allozyme variation within Solanum sect. Petota, ser. Etuberosa (Solanaceae) Am J Bot 792467-471 Stiles JI, Lemme C, Sondur S, Morshidi MB, Manshardt R (1993) Using randomly amplified polymorphic DNA for evaluating genetic relationships among papaya cultivars. Theor Appl Genet 85:697-701 Stuber CW, Moll RH, Goodman MM, Schaffer HE, Weir BS (1980) Allozyme frequency changes associated with selection for increased grain yield in maize (Zea mays L.). Genetics 95 :225-236 Stuber CW, Goodman MM, Moll RH (1982) Improvement of yield and ear number resulting from selection at allozyme loci in a maize population. Crop Sci 22:737-740 Tanksley SD, Medina-Filho H, Rick CM (1982) Use of naturally occurring enzyme variation to detect and map genes controlling quantitative traits in an interspecific backcross of tomato. Heredity 49:11-25 Tanksley SD, Young ND, Paterson AH Bonierbale MW (1989) RFLP mapping in plant breeding: new tools for an old science. Biotechnology 7:257-264 Tanksley SD, Ganal MW, Prince JP, de Vicente MC, Bonierbale MW, Broun P, Fulton TM, Giovannoni JJ, Grandillo S, Martin GB, Messeguer R, Miller JC, Miller L, Paterson AH, Pineda O, Roder MS, Wing RA, Wu W, Young ND (1992) High density molecular linkage maps of the tomato and potato genomes. Genetics 132:1141-1160 van der Berg JH, Bonierbale MW, Ewing EE, Plaisted RL, Tanksley SD (1992).Use of RFLP - linkage to detect loci affecting tuberization. Am Potato J 69:612 van Eek HJ, Jacobs JME, van Dijk J, Stiekema WJ, Jacobsen E (1993) Identification and mapping of 3 flower colour loci of potato (S. tuberosum L) by RFLP analysis. Theor Appl Genet 86:329-332 Waugh R, Baird E, Powell W (1992) The use of RAPD markers for the detection of gene introgression in potato. Plant Cell Rep 112466-469 Weller II (1987) Mapping and analysis of quantitative trait loci in Lycopersicon (tomato) with the aid of genetic markers using approximate maximum likelihood methods. Heredity 59:413-421 Weller J1, Soller M, Brody T (1988) Linkage analysis of quantitative traits in an interspecific cross of tomato (Lycopersicon esculentum x Lycopersicon pimpinellifolium) by means of genetic markers. Genetics 118:329-339 Welsh J, McClelland M (1990) Fingerprinting genomes using PCR with arbitrary primers. Nucleic Acids Res 18:7213-7218 Werner JE, Douches DS, Freyre R (1992) Use of half-tetrad analysis to discriminate between two types of Zn egg formation in a potato haploid. Genome 35 :741—745 Wilde J, Waugh R, Powell W (1992) Genetic fingerprinting of Theobroma clones using randomly amplified polymorphic DNA markers. Theor Appl Genet 83:871-877 Williams JGK, Kubelik AR, Livak KJ, Rafalski JA, Tingey SV (1990) DNA polymorphisms amplified by arbitrary primers are useful as genetic markers. Nucleic Acids Res 18:6531-6535 Xu YS, Clark MS, Pehu E (1993) Use of RAPD markers to screen somatic hybrids between Solanum tuberosum and S. brevidens. Plant Cell Rep 12:107-109 anal} chror genor diploi CHAPTER ONE ISOENZYMATIC IDENTIFICATION OF QUANTITATIVE TRAITS IN CROSSES BETWEEN HETEROZYGOUS PARENTS: MAPPING TUBER TRAITS IN DIPLOID POTATO (Solanum spp.) ABSTRACT. Eleven isozyme markers were utilized for Quantitative Trait Loci (QTL) analysis in diploid potato. These markers are distributed among seven of the twelve chromosomes and therefore give a representative, though sparse, survey of the potato genome. Tuber specific gravity and tuber dormancy were studied. Two segregating diploid populations were constructed from heterozygous self-incompatible parents. These two populations, TRP132 (127 individuals) and TRP133 (110 individuals) have a common maternal parent and combine genomes of Solanwn tuberosum (haploid), S. chacoense, and S. phureja. The populations were planted at two locations in Michigan in 1990 using a randomized complete block design with three replications per location. After harvest they were characterized with the isozymes and evaluated for specific gravity and tuber dormancy. To test for QTLs, one-way analyses of variance were conducted for each locus by trait combination. Significant associations between markers and quantitative trait variation were identified, which accounted for a range from 4% to 15% of the phenotypic variation for specific gravity, and from 4.5% to 20.4% for tuber dormancy. Two-way analyses of variance between significant markers were used to identify epistatic interactions between markers. Multiple regression analyses were used to estimate the overall effect of the significant markers on the phenotypic variation for 11 these traits. These values ranged from 15.3% and 32.3% for specific gravity. For dormancy, the significant loci accounted for 8.5% and 36.9% of the total phenotypic variation for each of the populations. Isozyme analysis has proved to be a useful tool for preliminary QTL studies in potato. ~ INTRODUCTION The concept of applying marker-assisted selection to the process of plant breeding has long been considered. Sax (1923) proposed identifying and selecting for "minor genes" of interest by linkage with "major genes", which could be scored more easily. Traditionally, the genetic markers used to develop maps in plants have been those affecting morphological characters. However, during recent years the use of isozymes and RFLPs in plant breeding and their advantage over morphological markers has been reported (Tanksley and Rick, 1980; Tanksley, 1983; Beckmann and Soller, 1983; Helentjaris et al., 1985; Tanksley et al., 1989). Genetic maps based upon these biochemical markers have been developed for a number of species such as maize, tomato, pepper, potato, lettuce, rice and Brassica (Helentjaris et a1. , 1986; Bematsky and Tanksley 1986; Tanksley et al., 1988; Landry et al., 1987; Bonierbale et al., 1988; Gebhardt et al., 1989; McCouch etal., 1988; Slocum et al., 1990). In addition to their use for the study and identification of monogenic traits, saturated genetic maps provide a means to estimate the number and genomic distribution of quantitative trait loci (QTLs) and examine them as discrete Mendelian factors (Zn: nuclez genera (Mend Potatoj and cu 86mph sl’eCific “Sing 21 of this a; Species c Wager al., 1990’- I'deutified 12 (Tanksley etal., 1982; Beckmann and Soller, 1988; lander and Botstein, 1989; Stuber, 1989b; Paterson et al., 1991). The utilization of isozymes for the study of quantitative traits has been reported in maize (Stuber et al., 1980; Stuber et al., 1982; Pollack et al., 1984; Frei et al., 1986; Kahler and Wehrhahn, 1986; Stuber et al., 1987) and tomato (Tanksley et al., 1982; Weller, 1987; Weller et al., 1988). Breeding of the cultivated potato, Solanum tuberosum subsp. tuberosum (2n=4x=48), is complicated by tetrasomic inheritance, the presence of cytoplasmic and nuclear sterilities (Grun et al., 1977), and inbreeding depression. In addition, it is generally acknowledged that the genetic base of cultivated tetraploid potato is narrow (Mendoza and Haynes, 1974). One approach utilized to simplify the genetic system in potato is to breed at the diploid level using haploids of cultivated species and diploid wild and cultivated tuber-bearing species. These represent a large source of valuable germplasm, which can broaden the genetic base of the cultivated potato and also provide specific desirable traits. The improved 2x germplasm is then transferred into the 4x level using 2n gametes (Chase, 1968; Iwanaga, 1983; Peloquin et al., 1989). The efficiency of this approach could be greatly increased if the introgression of genes from the wild species could be closely monitored with molecular markers (Tanksley et a1. , 1989). Linkage of RFLPs with a major gene conferring resistance to cyst nematode (Barone et al., 1990) and two genes controlling resistance to PVX (Ritter et al., 1991) have been identified in diploid potato; however, linkages to quantitative traits have not yet been reported in this crop. Two tuber traits of economic importance in potato are dry matter content and tuber dormancy. High dry matter content is a particularly important trait in potato y .. . _' L . . . . I x * y I- l3 cultivars used in the potato chip industry because of its association with increased chip yield and lower oil absorption (Owings, 1979). Tuber dormancy is the obligate period of non-sprouting after harvest even under conditions favorable for sprouting (Thompson et al., 1980), and is critical because long-terrn storage without sprout growth is an important aspect of potato marketing. Previous genetic studies have reported that these two tuber traits are polygenic (Ruttencutter et a1. , 1979; landeo, 1979; Thompson et al. , 1980), and they have been identified in selections made from South American diploid tuber-bearing relatives of the potato. In potato, fifteen enzyme-coding loci are presently known to segregate (Douches and Quiros, 1987; 1988). Some of them have been mapped onto several chromosomes on existing potato RFLP maps (Bonierbale et al. , 1988; Gebhardt et al. , 1989; Gebhardt et al., 1991). Both isozyme and RFLP markers can be used for QTL analysis in potato. Isozymes can be used for a preliminary study, and subsequently, a more detailed genome survey using RFLP markers would be conducted. The objectives of this study were to characterize two diploid populations with isozymes, to conduct field and storage studies to characterize them for two polygenic tuber traits, and to identify associations between the markers and quantitative trait variation. matt COlllt 1989 from 15 84810) (RCBD) beplantl Worm the Mon Honicultu Flanagan 14 MATERIALS AND METHODS Blanunamial Two Fl 2x populations designated as TRP132 and TRP133 were utilized in this study. Clone 848D22, which is a hybrid between haploid s. tuberosum (2x) and s. chacoense was the common female parent. The males used were the S. phureja clones 84811 in the case of TRP132, and 84810 for TRP133. These parents were chosen because of their isozyme diversity and divergent characteristics: 84SD22 has a high dry matter content and long dormancy, while the S. phureja clones have low dry matter content and short dormancy. A total of 127 and 110 genotypes were used in TRP132 and TRP133, respectively. The sad tubers for the 1990 field studies were obtained from 1989 field plots. W Parents and progenies were evaluated for dry matter content and tuber dormancy from field-grown tubers. Both populations along with two of the parents (848D22 and 84810) were planted in 1990 at two locations using a randomized complete block design (RCBD) with three replications per location. There were not enough tubers of 84811 to be planted in the field. Each plot consisted of eight plants with a within row distance of approximately 0.3 m and between row distance of 0.9 m. The two locations used were the Montcalm Research Farm, Edmore, Michigan (MES), and the Clarksville Horticultural Experiment Station, Clarksville, Michigan (CHES). MES location was planted on May 14, 1990 and harvested after 119 days, while CHES was planted on May of; 4a days to 8pmUtil Way AN 24, 1990, and harvested after 131 days. Following harvest, specific gravity was determined for each genotype for both locations using the weight in air/weight in water method: [air wt./(air wt. ~water wt.)]. This value is used to estimate the dry matter content of the tubers (Wilson and Lindsay, 1959). A digital scale with a i1 g. accuracy and a minimum sample size of 1 kg/plot were used. The value of specific gravity for each genotype was obtained from the mean of the three values from each of the replications in the field. For dormancy, a total of 4 tubers per genotype were placed on trays in storage at 10°C following harvest and evaluated weekly. The length of dormancy was determined as the average number of days required for 2 mm long sprouts to be evident for each genotype. I z m si The progenies and the parents were characterized for 11 segregating isozyme loci (Dia—I, Est-I, Got-1, Got-2, Idh-I, Mdh-I, 6-Pgdh-3, Pgi-I, Pgm-I, Pgm—Z, Pix-3) using both leaf and tuber tissue. Electrophoretic and enzyme staining procedures have been described elsewhere (Douches and Ludlam, 1991). The yellow flesh locus (Y) segregating in TRP133 was also scored. Statistical analysis Statistical analyses were carried out for a RCBD at each location for both traits. In the case of tuber dormancy, the logIo transformation for the average number of days to sprouting was used in all the analyses to improve normality. For specific gravity, two- way ANOVAs combined over locations were conducted for each population and broad dc gem stall". signi of Val Were CXplai; 0017:3133 (”Pres RESUL- T ant18481 and 84s1 l6 sense heritability values were estimated by the variance component method. In the case of dormancy, two-way AN OVA over replications and genotypes at the one location was used for the estimation of heritability. Single factor ANOVAs were conducted for each pairwise combination of quantitative trait and marker locus (GLM, Statistical Analysis Systems, Cary, NC). To detect linkage of a marker locus with a QTL, the segregation data was divided into genotypic classes (backcross, F2 or triallelic segregations). F-tests were used to statistically test if the means of the genotypic classes were different (P < 0.05). A significant difference in means was interpreted as linkage of QTL to the marker locus. Epistatic interactions between significant markers were tested by two-way analyses of variance (PROC GLM, SAS). The significant main effects and significant interactions were combined in a multivariate linear regression model to predict the total variation explained with the markers (Keim et al., 1990). To study the effect of heterozygosity, correlation analyses were performed between the number of heterozygous isozyme loci (expressed as percentage) and quantitative trait value for each genotype. RESULTS The populations TRP133 and TRP132 were characterized for 11 and 10 segregating isozyme loci, respectively. The genotypes for the parents (848D22, 84810 and 84811) are shown in Table l. 1. 848D22 was heterozygous for nine loci, while 84810 and 84811 were heterozygous for three and two loci, respectively. Segregation patterns Table 1.1. Isozyme genotypes for the parents and two populations 17 Segregation | | i I l 1 1 l l l l I 1 l 5 I 1 I l l I i i 1 l 1 I Isozymes 848D22 84810 Segregation 84811 . (9) (d)- in TRP133 (d)- 1___ ; 1 Did-1 12. 11 12:11 (BC)° 11 12:11 (BC) 1 1 Est-I BSC 53 FS:SS (BC) 88 FS:SS (BC) = ' Got-1 35 33 35:33 (BC) 33 35:33 (BC) 5 Got-2 15 55 15:55 (BC) 55 15:55 (Be) 1 11112-1 12 11 12:11 (BC) 11 12:11 (BC) : Mdh-I 22 12 22:12 (BC) 12 22:12 (BC) { 6Pgdh-3 12 11 12:11 (BC) 12 11:12:22(F,)° ? Pgi-I 22 12 22: 12 (BC) 22 22 (no leg.) 2 Pgm—I 13 33 13:33 (B0 33 13:33 (BC) . Pgm-Z 23 12 12:22:13:23 22 23:22 (no) 1 (air 1 Pix-3 13 11 13:11 (BC) 11 13:11 (no) _ . ‘ 84810 and 84811 are male parents for populations TRP133 and TRP132, respectively " 12 refers to the allelic designation Dia-l'l2 ° FS refers to Fast and Slow alleles, respectively (Douches and Quiros, 1988) " BC, tli and F2 refer to backcross, triallelic and Fz-type segregations, respectively of 5 hate: distr heter ‘ o 18 in the populations could be of three types: testcross, F2 or triallelic (Table 1.1). Chi- square analyses indicated that segregation of all the isozyme loci fit the expected ratios in both populations (data not shown). Based upon combined data from both populations (237 individuals), a linkage between Est-I and Got-I (12.8 +/- 3.0 map units) was identified. Heterozygosity was estimated as the percentage of heterozygous isozyme loci for each individual. In TRP133 heterozygosity values ranged from 9% to 100% with a mean of 53%, and the distribution of these values in the population was normal. For TRP132 heterozygosity values ranged from 0% to 90%, with a mean of 52%. Nevertheless, the distribution of values in this population was skewed since 42% of the individuals had heterozygosity values between 60 and 70% (data not shown). MES and CHES locations were harvested after 119 and 131 days from planting, respectively. The range of values and means for specific gravity for both populations and parents in both locations is shown in Table 1.2. A range of specific gravity values between 1.062 and 1.107 corresponds to a dry matter content between 17.43% and 24.98% in the tubers. The distribution pattern for both populations and locations is represented by population TRP132 at CHES (Figure 1.1). Broad sense heritability estimates for specific gravity combined over both locations were 89.2% and 86.1% for TRP133 and TRP132, respectively. Values of specific gravity were averaged over each genotypic class for each isozyme locus, and one-way AN OVAs were conducted to test for significant differences . ”A“ 4 -vV‘ .~ 3‘. . it”? «.5 ._ ‘ ‘ '. .I f I I. . .9 1 _d‘- 01...; 9., 21.; :5“ is real. . *3:— 19311.13 pl _, ., . . ..., _ . .3. . .1 H. . 0 5. ‘11. .'_‘. f "’ g ‘7' I . y) ‘93. rial“ '1- - ' a ’ ”w -4rar‘Lsefmihu-f genot) 110m 1 Table 1.2. Values of specific gravity obtained for populations TRP133 and TRP132, and the two parents‘ 19 MONTCALM CLARKSVILLE Range Mean i SE Range Mean 1- SE TRP133 1.046 - 1.099 1.079 : 0.001 1.057 - 1.110 1.083 :t 0.001 TRP132 1.061 - 1.105 1.086 i 0.001 1.062 — 1.115 1.086 d: 0.001 84SD22 1.078 - 1.082 1.080 :1; 0.003 1.076 - 1.092 1.084 -I_- 0.011 84810 1.065 - 1.068 1.067 :1; 0.002 1.062 - 1.065 1.064 i 0.002 ' For the populations, the range and mean are obtained from the mean values for each genotype from the three replications. For the parents, the range and mean are obtained from the mean value from the three replications in each of the population’s fields. Figure 1 Clarksvil l"i‘lllfictiw 20 NUMBER OF INDIVIDUALS 30 8 S 22 ’/ 4 D L062- t069- t075- t082- L089- L095- IJOZ- LOSE t074 t081 LOBB L094 1A01 1107 SPECIFIC GRAVITY Figure 1.1. Frequency distribution of specific gravity values for TRP132 grown in Clarksville, Michigan 1990 (84SD22 and 84810 are the female and male parents, respectively.) 21 among these classes. In population TRP133, significant differences for genotypes were found for three unlinked loci: 6-Pgdh-3, Got-2 and Pgm-I, and results were consistent over both locations. The amount of phenotypic variation (R2) for specific gravity explained by individual markers ranged from 4.5% to 7% for MES, and from 8.2% to 15% for CHES (Table 1.3). In TRP132, significant differences were also found for 6- Pgdh-3 and Got-2 over both locations, however Pgm-I and Dia-I were significant only at the MES location. The amount of phenotypic variation for specific gravity explained by the markers at MES ranged from 4% to 6.8%, while for CHES it ranged from 6.6% to 10% (Table 3.1). No correlation was found between the number of heterozygous loci per genotype and specific gravity for either population grown at either location. Two-way combinations of significant markers were tested for epistatic interactions at the 0.05 level. The only significant interaction found was 6-Pgdh-3*Gor—2 for TRP133 at CHES. Multiple analysis of variance estimated that 16.7% of the total phenotypic variation for specific gravity could be explained by the effect of the three significant loci for TRP133 at MES. For CHES location, this value was 32.3% with the three significant loci and 35.7% when the significant epistatic interaction was included in the model. For population TRP132, 19.6% of the phenotypic variation could be explained by the effect of the four significant loci at MES. For CHES, 17.5% of the variation could be explained by the two significant loci (Table 1.4). Table 1. TRP133 ljl Table 1.3. SigIan nificanzt association between specific gravity and isozymes for populations 22 TRP133 and Genotype Mean Specific Pr > F R2 Gravit ty TRP133 MES 6—Pgdh-3 l 1 1.081 0.019 0.050 12 1.077 Got-2 15 1.081 0.027 0.045 55 1.077 Pgm-I 13 1.081 0.005 0.070 33 1.076 CHES 6-Pgdh-3 l 1 1.086 0.000 0.150 12 1.080 Got-2 15 1.086 0.001 0.100 55 1.081 Pgm-I 13 1.085 0.003 0.082 33 1.080 TRP132 MES 6-Pgdh-3 l 1 1.090 0.013 0.068 12 1.085 22 1.084 Got-2 15 1.088 0.003 0.068 55 1.084 Pgm-I 13 1.088 0.024 0.040 33 1.084 Dia-I 1 1 1.088 0.003 0.068 12 1.084 CHES 6-Pgdh-3 11 1.089 0.015 0.066 12 1.086 22 1.083 Got-2 15 1.089 0.000 0.100 55 1.083 Pgm-I 13 1.088 0 077 33 1.085 ns‘ Dia-I 11 1.088 0.099 12 1.085 ns ' ns = not significant at the 0.05 level Table l. populati "—1 IF Table 1.4. Values obtained from the multiple analyses of variance for specific gravity for populations TRP133 and TRP132 23 6-Pgdh-3 Got-2 Pgm-I 6—Pgdh-3 Got-2 Pgm-I 6-Pgdh-3 Gar-2 Pgm-I 6—Pgdh-3*Got-2 Pr>F 0.000 0.000 R2 0.196 A D 24 In population TRP132, the 6-Pgdh-3 locus segregated in a F; manner and was found to have a significant association with specific gravity. This locus provided the only opportunity to examine gene action in this study (Edwards et al., 1987 ; Nienhuis et al., 1987). The homozygous class 6-Pgdh-3’3’ had higher values than either of the heterozygote 6-ll’gdh-3’32 or the other homozygote 6-Pgdh-323’. Regression analysis was performed using the means for each genotypic class averaged over both locations. The data fit an additive model for gene action, and the effect of allele substitution in this locus could be determined (Figure 1.2). My The average number of days to sprouting at 10°C for the parents was 10 and 80 days for 84810 and 84SD22, respectively. In population TRP133, the average number of days to sprouting ranged from 10 to 110, with a mean of 18 days. The range of days to sprouting for population TRP132 was 10 to 120, with a mean of 34. The distributions of both populations were found to be highly skewed towards lack of dormancy imparted by the S. phureja parent. The transformation logm of the average number of days to sprouting was used in all the analyses. The distribution of transformed values is shown for TRP132 (Figure 1.3). Broad sense heritability estimates for dormancy with data from the one location were 93.8% for TRP133, and 92.6% for TRP132. Correlation between the two tuber traits for each location and population was found to be significant only for TRP133 and specific gravity data from CHES. This showed a weak correlation of r = 0.236 (P = 0.013). The effect of number of heterozygous isozyme loci per genotype and the length of dormancy was also studied in each population and no correlation was found. One-way AN OVAs were conducted between the tuber dormancy data and isozyme locus genotypes. Significant differences were found for 6-Pgdh-3, Got-I, Got-2, Pgm-Z, Prat-3 and Est-I in TRP133. The amount of phenotypic variation for this trait explained by each marker ranged from 5.2% to 20.4%. In population TRP132, significant differences were found for Est-I and Got-I which explained 8.5% and 4.5% of the phenotypic variation, respectively (Table 1.5). 25 1 .090 _, $6 = 1.083 + 0.003X (r2 = 0.95) SPECIFIC GRAVITY 1.080 4‘ . r: 0 1 2 NIMBER OF CDPIIS OF AI'LEICE GPGI‘I-Gl (X) Fi ure 1.2. Regression on the means for each genotypic class of 6-Pgdh-3 in family 'I'IEP132 (averaged over both locations) 26 35 NUMBER OF INDIVIDUALS 30* 25— 114s1322 / 0 1.00-1.12 1.13-1.28 1.27-1.39 1.40-1.52 1.53-1.65 1.66-1.79 1.80-1.92 1.93-2.05 LOGINUMBER 0F DAYS TO SPROUTING) Fi ure 1.3. Fr uenqy distribution of length of dormancy (logmtransformed) values for 'I'IEP132 (84SD2 an 84810 are the female and male parents, respectively). Table 1.5. Si TRP133 and 27 1.164 LogwMean Pr > F Days to __S_2r_ostl_ng 1.267 0.006 1.155 1.286 0.001 1.155 1.325 0.000 1.128 1.265 0.006 1.164 1.264 0.017 1.165 1.269 0.011 R 0.068 0.090 0.204 0.068 0.052 0.058 gnificlafit association between tuber dormancy and isozymes for populations 4w 1 D 28 No significant epistatic interactions between significant markers were found for this trait by two-way analysis of variance in either population. Multiple analysis of variance estimated that 36.9% of the total phenotypic variation for tuber dormancy was due to the effect of the six significant loci for TRP133, while for TRP132, 8.5% of the phenotypic variation could be explained by the effect of the two significant loci (Table 1.6). ' Table 1.6. Values obtained from the multiple analyses of variance for tuber dormancy for populations TRP133 and TRP132 Pr > F R2 0.001 0.369 I l 1 0.004 0.085 1 _l DISCUSSION Ten and eleven isozyme loci were segregating in the two populations studied. Previous linkage analyses with RFLP markers indicate that these isozyme markers are distributed among seven of the twelve potato chromosomes (Bonierbale et al., 1987). In addition, gene-centromere map distance estimates (Douches and Quiros, 1987; 1988) indicate random distribution of these markers along the chromosome arms. This is confirmed by our current linkage analyses of data where only Est-I and Got-I were 4.! 29 found to be linked. These facts lead us to believe that the isozyme markers used in this study give a representative, though sparse, survey of the potato genome. Two quantitative traits were examined in each of the two populations, and one of the traits was examined in two locations. Data for these traits in the two populations was continuous as expected for a polygenic trait. F-tests for each pairwise combination of quantitative trait and isozyme loc'us were used to determine whether significant differences in trait expression were associated with genotypes at each of the segregating isozyme loci. Significant (P < 0.05) associations were found for 12 of 33 comparisons in TRP133 (36%), and 8 of 30 comparisons in TRP132 (27%). These values are low as compared with results obtained in similar studies with tomato where 56% of the comparisons were found to be significant (Tanksley et al., 1982) and in maize where these values ranged from 60% to 66% (Edwards et al., 1987). This may be due to the fact that in these two cases larger population sizes were used which should detect smaller phenotypic effects, and also some of the markers were linked thus reflecting the effect of common quantitative trait loci. Significant association was found between three isozyme loci (6-Pgdh-3, Got-2 and Pgm-I) and specific gravity in population TRP133, and results were consistent in both locations. In TRP132, significant differences were also found for 6-Pgdh-3 and Got- 2 over both locations, and Pgm-I and Dia-I were significant at only one location. We conclude that isozyme loci 6-Pgdh-3 and Got-2 show a strong, stable association with this trait whereas Pgm-I and Dia-I may have a G x E interaction such as found with QTLs for fruit traits in tomato (Paterson et al., 1991). In the case of dormancy, the distribution of the average number of days to sprouting for both populations was highly skewed towards lack of dormancy. This could be explained by dominance effects from the S. phureja parents. For TRP133, significant association was found with six isozyme loci and dormancy. Two of these, Est-I and Got-1 were also significant in TRP132, showing a stable association with this trait. Loci 6-Pgdh-3 and Got-2 were found to be associated both with specific gravity and dormancy in TRP133. Nevertheless, only a weak correlation between both traits was found with specific gravity data from one of the be estimated by the R2 value obtained in the analyses of variance. This study detected 30 effecm as small as 4% of the total phenotypic variation, while in maize factors contributing as little as 0.2% of the phenotypic variation in yield related traits could be detected using isozyme markers and large populations (more than 1500 plants) (Stuber et al., 1987). For specific gravity, the phenotypic variation explained by individual markers ranged between 4.0% and 15%, whereas for tuber dormancy it was between 4.5% and 20.4%. The cumulative effects of all significant marker loci on the traits was estimated through multiple analyses of variance. In this case, the amount of phenotypic variation that was explained by significant markers ranged between 16.7% and 32.3% for specific gravity, and for dormancy it was 8.5 % and 36.9% for TRP132 and TRP133, respectively. At present it is not known whether the isozymes per se have a direct influence on the trait or are associated only through linkage. It is generally assumed that these enzymes are nearly-neutral genetic markers and alleles at most isozyme loci probably do not directly affect the phenotypic expression of the quantitative trait evaluated (Stuber, 1989a). In studies in maize, Pollack et al. (1984) indicate that the Acp- I locus may be associated with yield either directly or through linkage; in tomato, Tanksley et a1. (1982) and Weller et al. (1988) indicate that the effect of significant enzyme loci is due to linkage to the QTLs. The level of variation explained by individual marker loci is thus affected by its genetic linkage to the QTL. In our study the effect could be underestimated due to loose linkages. Subsequent RFLP analysis to survey the whole potato genome should identify more and tighter linkages and therefore a greater percentage of variation for the trait may be explained. All two-way combinations between significant markers were tested to detect significant epistatic interactions affecting the traits. For Specific gravity, the only significant interaction found was between 6—Pgdh-3 and Gar-2 for TRP133 at CHES. These two markers have been previously located in chromosomes 5 and 7, respectively (Bonierbale et al., 1987). The inclusion of this interaction in the multiple analysis of variance resulted in an R2 of 35.7%, which represents a gain of 3.4% from the main- effects model. For dormancy, no significant epistatic interactions between markers were found. This is similar to results found in tomato, where several traits did not show any significant interactions (Weller et al., 1988). . 31 There is no apparent effect of heterozygosity in either of the traits studied as demonstrated by the lack of correlation between the number of heterozygous loci and value of the trait for each individual. This contrasts with results found in maize where the level of heterozygosity plays a very large role in the expression of grain yield (Edwards et al., 1987). Also, there is no association between the highest value for the quantitative trait and the heterozygous genotype of the isozyme loci showing significant linkage. Therefore, an additive model for the traits has been assumed. This is supported by the regression analysis with 6-Pgdh-3 in TRP132, which provided the only opportunity to examine gene action at a locus in this study. The effect of allele substitution can be estimated and specific gravity can be explained by the regression formula Specific Gravity = 1.083 + 0.003X where X equals the number of copies of the allele 6-Pgdh-3’ which corresponds to the favorable allele coming from 84SD22. The potato poses challenges to QTL analysis. Generation of inbred potato lines is not practical due to self-incompatibility and inbreeding depression at the diploid level. Fl populations constructed from heterozygous diploid parents can have backcross, F2, and multiple-allelic segregation patterns occurring in a single cross, thus limiting our ability to examine intralocus effects of QTLs. However, we have identified associations between the isozyme markers and the traits by using one-way ANOVAs and F—tests, using a significance level of P < 0.05. This level of significance has been considered to give great risk of identifying false positives (lander and Botstein, 1989). Nevertheless, in this study, QTL analysis is strengthened by basing the results on two populations over two locations. Since significant linkages were determined across genetic backgrounds and locations, it can be more confidently stated that a QTL has been correctly identified. One advantage of potato over other crops is that since it is clonally propagated, enough seed tubers can be available of the same genotypes to conduct replicated studies, which is not feasible in most seed propagated crops such as maize or tomato. Isozyme analysis has proved to be a useful tool for QTL studies in potato. Isozyme characterization can be quickly completed in a large number of individuals, providing a preliminary identification of putative linkages to quantitative traits. RFLP analysis is being used to further localize and fine-map QTLs with markers and to 32 strategically survey the potato genome for other QTLs not revealed with isozymes. LIST OF REFERENCES Barone 'A, Ritter E,_ Schachtschabel U, Debener T .Salamini_F, Gebhardt C (1990) Localizatlon by restriction fragment length polymorphism mapping in potato of a major domlnant gene conferrln resrstance to the potato cyst nematode Globodera rostochiensis. Mol Gen enet 224:17 -182 , Beckmann J, Soller M (1983) Restriction fragment length pol morphisms in genetic improvement: Methodologies, mapping and costs. Theor Appl enet 67:35-43 Beckmann J, Soller M (1988 Detection of linkage between marker loci and loci affectin aggntltatlve traits ln crosses etween segregating populatlons. Theor Appl Genet 76:22 - Bematsky R, Tanksley SD (1986) Toward a saturated linkage map of tomato based on isozymes and random cDNA sequences. Genetics 120:1095-1103 Bonierbale M, Plaisted R, Tanksley SD (1988) RFLP maps based on a common set of tlzlpélses reveal modes of chromosomal evolution in potato and tomato. Genetics 120:1095- Chase SA ((1968) Analytical breeding in S. tuberosum L. A scheme utilizing parthenotes and other iploid stocks. Can J Genet Cytol 5 :359-363 Douches DS, Quiros CF .(1987) Use of 4x-2x crosses to determine gene-centromere distances of isozyme 1001 in Solanum species. Genome 29:519-527 Douches DS, Quiros CF (1988) Additional isozyme loci in tuber-bearing Solanums: Inheritance and linkage relationships. J Hered 79:377-384 Douches DS, Ludlam K (1991 Electmphoretic characterization of North American potato cultivars. Am Potato 1 6 :767-780 Edwards MD, Stuber CW, Wendel J F 1987) Molecular-marker-facilitated investigations of quantitative-Halt 10Cl in malze. I. umbers, genomic dlstrlbution and types of gene action. Genetics 1 16:1 13- 125 FreiOM, Stuber CW, .Wendel JF (1986) Use of allozymesas enetic markers for predicting performance ln maize srng e cross hybrlds. Crop Scr 26: 7—42 Gebhardt C Ritter E, Debener T, Schachtschabel U, Walkemeier B, Uhrig H, Salamini 58(16g89gRFLP analysrs and llnkage mapplng in Solanum tuberosum. Theor Appl Genet Gebhardt C Ritter E, Barone A, Debener T, Walkemeier B, Schachtschabel U Kaufmann H, Thom son RD, Bonierbale MW, Ganal MW, Tanksley SD, Salamini F 1991) RFLP maps 0 potato and their alignment with the homoeologous tomato genome. eor Appl Genet 83:49-57 Grun P, Ochoa C, Ca ana e D (1977) Evolution of cytoplasmic factors in tetraploid potatoes. Amer J Bot :41 -420. Helentjaris T, King G, Slocum M, Siedenstrang C, Wegman S (1985) Restriction \Q [1‘ 1 c. .. mm- --..s~— _ 33 {Rent lpolymorphisms as probes for plant diversity and their development as tools for app ed p ant breeding. Plant Mol Biol 5:109-118 Helentjaris T, Slocum M_, Wright S, Schaefer A, Nienhuis J. (1986) Construction of genetic hnkage maps m maize and tomato usmg restnction fragment length polymorphisms. Genetics 118:353-363 Iwanaga, M . 81983). Ploidy level mani ulation approach: development of di loid populations wr specific resrstance and F R 2n llen production. In: resent and uture strate res for gotato breedmg and improvement. eport of the 26"1 Planmng Conference, CIP. ec. 19 3, Lima, Peru. Kahler AL, Wehrhahn CF (1986) Associations between uantitative traits and enzyme loc1 in the F2 population of a maze hybrld. Theor Appl enet 72: 15-26 Landeo} (1979).Brwd.ing potential of Group Andigena haploid potatoes. Ph.D. Thesis, Umversrty of Wisconsm, Madison. 124 p Lander E, Botstein D (1989 Mappingz Mendelian factors underlying quantitative traits usmg RFLP linkage maps. enetrcs 1 1:185-199 Landry BS, Kesseli RY Farrara. B, Michelmore RW (1987) A genetic map of lettuce (Lactuca sativa L.) wrth restrrctlon fra ment len th polymorphism, isozyme, (11863.56 resistance and morphological markers. enetlcs 1 6:3 1- 37 ‘ McCouch SR, Kochert G, Yu ZH, Wang ZY, Kush GS, Coffman WR Tanksley SD (1988) Molecular mapping of rice chromosomes. Theor Appl Genet 76:315-829 Mendoza H, Ha nes F (1974) Genetic relationships among potato cultivars grown in the United States. ortscience 9:328-330 Nienhuis J, Helentjaris T, Slocum M, Ruggero B, Schaefer A. (1987) Restriction fragment length.polymo hism analysis of ocn assocrated wrth msect resrstance in tomato. Crop Scr 27:797- 03 Owings T (1979) Cultural practices which influence the specific gravity of Russet Burbank. Proc. 18'” Ann. Washington Potato Conf. Trade Fair. 41-4 Paterson AH, Damon S, Hewitt JD, Zamir D, Rabinowitch HD, Lincoln SE, Lander ES, Tanksley SD (1991) Mendelian .factors underlying quantitative traits 1n tomato: companson across specres, generations and envrronments. Genetics 127:181-197 Peloquin SJ, Yerk GL Werner IE, Darmo E (1989) Potato breeding with haploids and Zn gametes. Genome 31:1000-1004 Pollack LM, Gardner CO, Parkhurst AM (1984) Relationships between enz me marker llolcggand morphological traits in two mass selected maize populations. Crop ci 24:1174- Ritter E, Debener T, Barone A, Salamini F ,.Gebhardt C (1991) RFLP ma ing on tato (éhromgszo7m8els g; two genes confemng resrstance to potato virus X (P ). Mo Gen enet : - Ruttencutter _G, .Ha .nes. F, Moll R 9979) Estimation of narrow-sense heritability for spastic] gram £1531plord potatoes ( . tuberosum subsp. phureja and stenotomum) Am otato : Sax K (1923) The association of size differences with seed coat pattern and pigmentation in Phaseolus vulgaris. Genetics 120:597-604 ‘J # 34 Slocum MK, Figdore. SS, Kennard WC, Suzuki J_Y, storn TC (1990) Linkage arrangement of restrrctron fragment length polymorphrsm 1n Brassica oleracea. Theor Appl Genet 80:57-67 ' Stuber CW, Moll RH, Goodman MM, Schaffer HE, Weir BS (1980) Alloz me frequency chan es associated with selection for increased rain ield in maize ( a s L. , Genegtics 95 :225-236 g y may ) Stuber CW, Goodman MM, Moll RH_(1982) Improvement of ield and ear number resulting from selection at allozymelocr in a marze populatron. rop Scr 22:737-740 Stuber CW, Edwards MD, Wendel J F (1987) Molecular marker-facilitated investigations of quantitative trait loci in maize. 11. Factors influencrng yreld and rts component traits. Crop Sci 22:737-740 Stuber CW 1989a) Comparative studies .using RFLPs and isozymes as molecular markers for e study and analyses of multrgenrc trarts 1n maize. In: Develo ment and A lication of Molecular Markers to Problems 1n Plant Genetrcs. Helentjarrs , Burr B s.). Cold Spring Harbor Laboratory. pp. 103-1067 Stuber CW (1989b) Isozymes as markers for studyin and manipulatin quantitative 21816382216: Isozymes in Plant Biology. Soltrs DE, Soltrs S (Eds.) Droscorr es Press. pp. Tanksley SD Rick CM (1980) Isozyme linkage map of the tomato: Applications in genetics and breeding. Theor Appl Genet 57:1 1-170 Tanksley SD, Medina-Filho H, Rick CM 1982) Use of naturally_occurring enzyme variation to detect and map genes control ing quantrtatrve trarts rn an rnterspecrfic backcross of tomato. Heredity 49: 1 1—25 Tanksley SD (1983) Molecular markers in plant breeding. Plant Mol Biol Rep 1:3-8 Tanksley SD, Bematzky R,.Lapitan NL Prince JP (1988) Conservation of 6gene rangtorre but not gene order 1n pepper and tomato. Proc Natl Acad Scr USA 85: 19- Tanksley SD, Young ND, Paterson AH Bonierbale MW 1989) RFLP mapping in plant breedrng: new tools for an old science. Biotechnology 7: 57-264 Thom son P, Ha nes F, Moll R (1980). Estimation of genetic variance components and herrta rhty for tu er dormancy 1n diplord potatoes. Am Potato J 57:39-46 Weller JI (1987) Mapping and analysis of quantitative trait loci in Lycopersicon (tomato) wr the ard of Eenetrc markers using approximate maximum likelihood methods. Heredity 59:413-4 1 Weller JI Soller M, Brody T (1988) Linkage analysis of quantitative traits in an mterspecrfic cross of tomato (Lécopersicpigegzug efigm x Lycopersicon pimpinellifolrwn) enehcs : — Wilson JH, Lindsay AM (1969) The relationshi between specific gravity and dry matter content of potato tubers. Am Potato] 46:323- 28 by means of genetic markers. CHAPTER TWO QUANTITATIVE TRAIT LOCI ANALYSIS OF TUBER DORMANCY IN DIPLOID POTATO (Solanum spp.) ABSTRACT. Quantitative trait loci (QTL) analysis for tuber dormancy was performed in a diploid population of potato (TRP133) consisting of 110 individuals. This population was derived from the cross of a hybrid between haploid S. tuberosum (2x) and S. chacoense, with a S. phureja clone. The population was characterized for 10 isozyme loci, 44 RFLPs and 63 RAPDs. Eighty-seven of the loci segregating from the female parent were utilized to develop a linkage map that comprises 10 of the 12 chromosomes in the genome. The length of dormancy in the p0pulation ranged from 10 to 90 days to sprouting, with a mean of 19 days. QTLs for this trait were determined by conducting one-way analyses of variance for each locus by trait combination. Twenty-two markers had a significant association with dormancy, identifying 6 QTLs localized on each of chromosomes 2, 3, 4, 5, 7 and 8. The QTL with the Strongest effect on the trait was detected on chromosome 7. A multilocus model was developed using the locus with highest R2 value in each QTL. This model explained 57.5% of the phenotypic variation for dormancy. Seven per cent of the possible epistatic interactions tested through two-way analyses of variance were significant. When these were included in the main effects model, it explained 72.5% of the phenotypic variation for dormancy. QTL analysis in potato, the methodology to transfer traits and interactions into the 4x level, and QTLs of value for marker-assisted selection are discussed. 35 36 INTRODUCTION Tuber dormancy is defined as the obligate period of non-sprouting after harvest even under conditions favorable for sprouting (Simmonds, 1964). Dormancy release involves changes in respiration and levels of enzymes, sucrose, nitrogenous compounds and endogenous hormones in the tuber, as reviewed by Hemberg (1985). Current literature supports an "inhibitor/promoter" hypothesis, with critical events associated with dormancy release involving a shift in the growth regulator ratio in favor of promoters, and subsequent establishment of positive feedback between the bud and mobilized food reserves (Coleman, 1987). The length of dormancy is characteristic of different potato varieties (Burton, 1963; Simmonds, 1964; Bogucki and Nelson, 1980; Jeoung et al., 1983). This is an important trait in potato production, since long-term storage without sprout growth is critical for tuber marketing. One method used to prolong the length of tuber dormancy is storage under low temperatures (4°C), however this causes a conversion of non- reducing to reducing sugars which is undesirable for the processing industry. Various dormancy-inducing chemicals have also been tried (Burton, 1966), but questions concerning their toxicologies were raised (Vaughn and Spencer, 1991) and some have been banned from use. An alternative approach is to increase the length of dormancy through genetic means. Long dormant periods have been identified in selections made from South American diploid tuber-bearing relatives of the potato (Thompson et al., 1980; Hermundstad and Peloquin, 1985 ; Hermundstad, 1986) and this characteristic can be introgressed to the cultivated gene pool. Tuber dormancy is under polygenic control (Simmonds, 1964), and more than three genes are involved (Flewelling, 1987). Quantitative trait loci (QTL) analysis using molecular markers (Lander and Botstein, 1988) provides a tool for a more detailed study of this trait. The numbers and genomic distribution of quantitative trait loci and their contribution to the variation of the trait can be estimated. This knowledge is necessary to be able to monitor the introgression of these genes and provide a framework for a more analytical breeding of the potato using wild relatives. Isozymes and RFLPs have 37 been used for dissecting quantitative traits in maize (Edwards et al., 1987; Stuber etal., 1987), tomato (Tanksley et al., 1982; Weller et al., 1988; Tanksley and Hewitt, 1988; Paterson et al., 1988), soybean (Keim et a1, 1990; Diers etal., 1992), wheat (Miura et al., 1992) and barley (Hayes et al., 1992; Heun, 1992; Hackett et al., 1992). In a previous study we reported on the use of isozymes to identify QTLs for specific gravity and dormancy in potato (Chapter 1). In this study, QTL analysis of tuber dormancy has been complemented through the use of previously-mapped RFLPs and unmapped RAPD markers. MATERIALS AND METHODS ElaaLMatszrial One of the two populations previously utilized in Chapter 1 was chosen for this study. This population named TRP133, is a diploid F, consisting of 110 genotypes. The female parent used in the cross was clone 84SD22, a hybrid between haploid S. tuberosum (2x) and S. chacoense, while the male parent was S. phureja clone 84S10. The parents were chosen because of long and very short dormancy periods, respectively, and previously known isozyme diversity. WM! Four tubers of 3 to 5 cm diameter per genotype were selected after harvest from 1989 field plots at Montcalm Research Farm, Edmore, MI. These were placed on trays in storage at 10°C in the dark, which are the common storage conditions, and evaluated weekly. The length of dormancy was determined as the average number of days required for 2 mm long sprouts to be evident for each genotype. mm: The parents were characterized for the morphological marker yellow flesh (Y), isozymes, RFLPs, and RAPDs. Markers that were heterozygous in one of the parents 38 and homozygous in the other were used to characterize the progeny, where a BC,-type segregation (1:1) was expected. For all markers, the presence of the unique allele from the heterozygous parent was scored as 1 and the homozygous as 0 in the segregating progeny. 21- W3 10 segregating isozyme loci (Dia-I, Est-I, Got-I, Got-2, Idh-I, 6-Pgdh-3, Pgi-I, Pgm-I Pgm-Z, Pix-3) were utilized as described in Chapter 1. b. Wt Tomato genomic and cDNA probes utilized were kindly provided by S. Tanksley, Cornell University, and potato genomic and cDNA probes by C. Gebhardt, Max Planck Institut, Germany. At least four markers per chromosome were selected based on their position on previous potato maps (Bonierbale et al., 1988; Tanksley et al., 1992; Gebhardt et al., 1991). DNA was extracted from leaf tissue for all genotypes following a procedure by Saghai-Maroof et al. (1984). The concentration was quantified using a fluorometer (I-Ioefer Scientific Instruments, Model TKO 100). Seven ugs of DNA was digested with the following endonucleases using 2 units of enzyme per pg of DNA: EcoRI, HindIII, XbaI, Dral, EcoRV, BamHI. The digested DNA samples were separated on 0.8% agarose gels. Southern transfer on Nytran nylon membranes, oligolabelling of probes with 32P, filter hybridization using a Robbins Scientific Incubator (Model 310), and filter washes were all performed according to Sisco et al. (1990). Filters were wrapped in plastic wrap and placed in X-ray cassettes at -80°C for 2-10 days. 0. was: The PCR protocol followed Williams et al. (1990) with minor modifications to Optimize for potato DNA. Each reaction was composed of: 1x buffer (100 mM KCl, 100 mM Tris-HCl pH 8.3), 0.8 mM dNTPs, 5 mM MgCl,, 1 U Stoeffel enzyme (Perkin Elmer Cetus), 12.5 ng of primer and 12.5 ng of potato DNA brought up to a final volume of 12.5 #1 with ddH20. The thermocycler (Perkin Elmer Cetus DNA Thermocycler 480) was programmed for 3 cycles of l min at 94°C, 1 min at 35°C, and 2 rrrin at 72°C; followed by 34 cycles of 1 min at 94°C, 1 min at 40°C, and 2 min at 72°C; and 5 min extension at 72°C. On completion, the amplification products were saw-M I 39 separated by electrophoresis on a 1.6% agarose gel in 1x TAE buffer. Lambda DNA cut with EcoRI and Hindlll was used as fragment size marker. Primers were commercial 10-mers from Operon Technologies (Alameda, California), specifically from Kits A, F, G, H and 1. Primers were selected when they generated bands in one parent and not in the other. Because of complete dominance in these markers, the heterozygous and' homozygous forms for the presence of an allele in a parent could not be distinguished until the band was observed either to segregate or be present in all the progeny, respectively. Wm Markers which were in heterozygous form in the female parent 84SD22 were utilized to develop the linkage map. The LINKAGE-1 program (Suiter et al., 1983) was used to determine fit to expected Mendelian ratios for each marker, and the linkage phase between linked markers. Segregating data for markers linked in repulsion were reversed to allow adequate estimation of recombination distances. The map was then constructed with MAPMAKER (Lander et al., 1987) v.01 for Macintosh, using LOD scores exceeding 3.0. Linkage groups were assigned to specific chromosomes based on the isozyme and RFLP loci previously mapped (Bonierbale et al., 1988; Tanksley et al., 1992); RAPD markers were subsequently added. SEW These have been previously described in Chapter 1. Briefly, the segregation data for each marker in the population was divided into the two genotypic classes. Single factor ANOVAs between each pairwise combination of dormancy data and marker locus were conducted (PROC GLM, Statistical Analysis Systems, Cary, NC). Markers with distorted segregation ratios were not used in the analyses. F-tests were used to statistically test if the means of the genotypic marker classes were different (P < 0.05). A significant difference in dormancy means was interpreted as linkage of a QTL to the marker locus. For linked markers, the phenotypic effect of the marker allele was estimated by the difference between the trait means of its genotypic classes. QTLs were W 40 localized based on the position of the marker loci on the map. Significant markers in the same chromosome were considered as one QTL if the distance between them did not exceed 50 cM (Paterson et al., 1991). The loci with the highest R2 value per QTL were then combined in a multiple analysis of variance model to predict the total variation for dormancy explained by the identified QTLs (Keim et al., 1990). Epistatic interactions between 'significant loci were tested by two-way analyses of variance. Significant interactions were then included in the multiple analysis of variance to determine their contribution in the phenotypic variation for dormancy. When there were several interactions between the same pairs of QTLs, the one with the highest R2 value was utilized. The main effect of the loci in the interactions were also included in the model if not already present. RESULTS The average number of days to sprouting for the parents was 80 and 10 days for the female and male parents, respectively. The distribution of values in the population was continuous and highly skewed towards short dormancy, having a range from 10 to 90 days and mean of 19. Therefore, the transformation logo of the number of days to sprouting was used to improve normality. The frequency distribution of these values is shown in Figure 2.1. Two genotypes (designated as TRP133-l and TRP133-215) showed transgressive segregation, with lengths of dormancy of 90 and 87 days, respectively (logw of 1.954 and 1.940). All isozyme loci fit the expected segregation ratio, as previously described in Chapter 1. From all RFLP probes evaluated in the parents, only one (TG83) was found to be heterozygous in both parents thus having an Fz-type segregation. Data from this probe was not included in the analysis. All other probes resolved loci heterozygous in one of the parents, and homozygous in the other. Thirty-four RFLP probes that segregated in a BC, fashion were scored. Eight of these resolved two loci and were designated with the addition of a T or B (for top and bottom locus), respectively. frequency 84810 848022 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8 1.9 2 distribution of length of dormancy (log 0 transformed) values in SD22 and 84S10 are the female and ' e parents, respectively). 42 Additionally, loci TGIZ2T and TGISZT showed triallelic segregations: both parents had one unique allele and the other in common, resulting in 4 genotypic classes in theprogeny. In these cases the presence or absence of the unique allele from each parent was scored independently. The total number of RFLP loci that were scored was 44. The expected segregation ratio was found for all loci except TG14l, TG18, TG53 and CD31. The female and male parents were heterozygous for 32 and 12 loci, respectively. A total of 29 random primers were utilized in this study. These produced a range of one to five scorable loci, resulting in a total of 63 RAPDs scored in the progeny. Eight of the RAPDs had distorted segregation ratios (data not shown). The female and male parents were heterozygous for 50 and 13 loci, respectively. The linkage map developed with isozyme, RFLP and RAPD loci which were in heterozygous form in the female parent is shown (Fig. 2.2). None of the RFLP probes selected by their known location on chromosomes 9 and 12 showed polymorphism or could be scored successfully, so no markers could be assigned to these chromosomes. Some RAPD loci could not be mapped, either because they showed linkage only to other RAPD loci, or no linkage to any other locus. One-way ANOVAs were conducted between the tuber dormancy data and the two genotypic classes for each of the markers used. In addition to significant association found with the 6 isozyme loci previously described in Chapter 1, significant QTLs were found with l RFLP loci and 15 RAPDs (Fable 2.1). Genotype TRP133-1 with a dormant period of 90 days and TRP-215 with 87 days, had 82% and 95% of the favorable alleles for the 22 significant markers, respectively. The significant RFLP locus (TG31B) and 4 of the RAPD loci were heterozygous in the male parent. They showed no linkage between each other and their chromosomal location has not been established. The position of the significant loci segregating from the female parent is shown (Fig. 2.2). These markers identify 6 QTLs, one on each of chromosomes 2, 3, 4, 5, 7 and 8. 43 Figure 2.2. Molecular linkage map and localization of QTLs for tuber dormancy ‘. ' significant markers are indicated b asterisks on the right side of their name; *, *, *** rndrcate 0.05, 0.01 and .001 probabrlrty levels, respectively. «.« 2a.... soon I 8.0. 5.9 x... * .5. J j x... .3... i o««.o» * . HHH «.2 i So» 38 «.to «.8: /.fl_. o««.o» *** «.8< 89. «.20 . * * :6 HHH awn...“ mm «a... . .20 xx... :8 .32.. «.8 3.9 J. a «.2. 39. a. a. x «a... L L r r N F _. _. op . m m h «a: 4 j p.99 4 58» 80h . * 05m .5 * «.80 «8.. .8: .88. Ir . .. ..«.o 3.. J 1.3 a. u. 23.... .I-.. 02.0» .7 «.«6 .62 1 8.5 . «.8. L. 98.. .. moo * «Eon. ”U . r a. 3.0 .....< 59. «.22 38» 3... 3.: . .29 a... /. r . ..«9. ..«.o . Re» I: ..«.< Wk». 99 so» lfin .39 ..«.< .53. II-.. .3“. :8» is . G r L c m m e m N _. ° indicates p to sprouting. F Marker‘ l PIX-3 G03.2 Gl9.1 Pgm-Z 6-P dh-3 A1 .2 Gar-2 A01.3 111.2 A08.2 Gl7.2 A01.2 A04.l 117.1 Got-I Est-I G05.l eteroz. in 9: Heteroz. in d: FOl.2 F05.1 612.2 613.1 TG31B lyfl—WW “—'__—V_—‘_—'—_—“——— l Chrom.” WOOOOQQQQQQQQQUIADJNN 45 Table 2.1. Significant association between markers and tuber dormancy R2 (%) Phenot ic effect (dng 5.2 * 3.8 5.4 * 4.3 4.3 * 3.4 5.5 * 3.8 6.9 ** 4.2 14.4 *** 6.3 20.4 *** 7.7 14.6 *** 6.5 15.5 *** 6.9 13.4 *** 5.9 15.7 *** 6.7 15.0 *** 6.4 12.4 ** 5.8 13.7 *** 6.1 9.0 *** 5.0 5.8 ** 4.0 4.2 * 3.6 | 10.5 ** 5.7 8.4 ** 4.8 5.2 * 3.7 4.6 * 3.5 7.2 ** 4.6 tivel ' isozyme loci are italicized; TG31B is a RFLP locus; others are RAPD markers. b ccllrromfpes‘omal location for markers heterozygous 1n the male parent have not been 1 entr r . *, **, *** indicate significance at the 0.05, 0.01 and 0.001 probability levels, genotypic difference between the trait means of the markers classes, in days 46 The amount of phenotypic variation for tuber dormancy explained by each significant marker, as determined by its R2 value, ranged from 4.2% to 20.4%, This represents a difference of 3.6 and 7.7 days to sprouting between the means of the genotypic classes of the markers, respectively (Table 2.1). The most frequent R2 values were between 4% and 6% (36% of all markers) and between 14% to 16% (22%). On average the loci on the QTL on chromosome 7 had the highest R2 values, all of them being above 12%. This represented a difference of more than 5 days to sprouting between the means of the genotypic classes. The isozyme marker Got-2 on this same QTL had markedly the highest effect of all loci, explaining 20.4% of the phenotypic variation. Seventeen out of 231 possible epistatic interactions were significant (7%) (Table 2.2). Most of these interactions involve one marker that was heterozygous in the male parent with others heterozygous in the female parent, so the two chromosomal locations involved could not be identified. In the cases where the interaction was between two loci segregating from the female parent, it involved loci on chromosomes 3 and 7 (G19. 1*Gor-2), and 5 and 7 (G05.l with others). The phenotypic variation of dormancy explained by each of these interactions ranged between 3.5% and 7.1%, and more than half of them (53%) explained between 4% and 6% of the variation. The marker with the highest R2 value per QTL was chosen to develop a multilocus model. All significant markers heterozygous in the male parent were also included. Accordingly, a model with the markers Pgm-Z, 6-Pgdh-3, Got-2, Got-I, Est-I, F01.2, F05.1, G03.2, G12.2, Gl3.1, G191 and TG31B was developed, which explained 57.5% of the phenotypic variation for dormancy. This value increased to 72.1% when the significant interactions were included (Table 2.3). Table 2.2. Significant epistatic interactions between significant markers I Interaction Heterozygous in 9: 605.1 * 617.2 605.1 * 111.2 . 605.1 * 117.1 605.1 * A01.3 605.1 * A01.2 605.1 * A04.1 605.1 * A08.2 619.1 * Got-2 Heterozygous in 6': F01.2 * F05.1 F01.2 * G12.2 F01.2 * TGBIB F05.1 * TG31B F05.1 * Prat-3 G12.2 * G19.l 612.2 * 6—Pgdh-3 TG3lB * Pix-3 R2 (%) l 3.9 * 5.5 * 3.8 * 6.4 ** 4.0 * 3.5 * 3.6 * 4.2 * 4.6 * 5.4 * 7.1 ** 6.2 * 5.7 * 5.5 * 5.5 * 6.4 ** ' one or both of the loci in the interaction are heterozygous in the male parent. *, ** indicate significance at the 0.05, and 0.01 probability levels, respectively. I .- '11: 48 Table 2.3. Models used to determine the amount of phenotypic variation for tuber dormancy explained by QTLs and epistatic interactions R2 (%) l Main Effects: Pgm-Z 6-Pgdh-3 Gar-2 Got-I 603.2 G 19.1 F01.2 F05.1 612.2 613.1 TG31B 57.5 *** With Interactions: Pix-3 Pgm—2 6-Pgdh-3 Got-2 Gar-I 603.2 619.1 605.1 A01.3 F01.2 F05.1 612.2 613.1 T631B F01.2 * F05.1 F01.2 * 612.2 F01.2 * TG31B F05.1 * T631B F05.1 * Prx-3 605.1 * A01.3 612.2 * 619.1 612.2 * 6-Pgdh-3 619.1 * Got-2 T631B * Pix-3 72.1 *** ____l| *** indicates significance at the 0.001 probability level 49 DISCUSSION The distribution of number of days to sprouting for population TRP133 was highly skewed towards short dormancy, which may be explained by dominant genes coming from the S. phureja male parent. Although a normal distribution of values is preferred for QTL analysis, studies have been previously performed in traits with skewed distributions in tomato (Nienhuis et al., 1987; Paterson et al., 1991). Some degree of transgressive segregation was also observed in the population, since two genotypes have even longer dormant periods than the female parent. In previous cases of mapping with heterozygous parents in potato, loci segregating from both parents were combined in one map by linkage to markers for which both of them were heterozygous (Bonierbale et al., 1988), and formation of what has been designed as "allelic bridges" (Ritter et al., 1990). In the present case, all isozyme and RFLP markers were heterozygous in either one of the parents, with the exception of only one RFLP locus, heterozygous in both. Most markers were heterozygous in the female parent, which is an interspecific hybrid between S. tuberosum and S. chacoense. Therefore, a linkage map with the markers segregating from this parent was constructed. This map consisted of a total of 87 loci when the RAPDs were included. The linear order of the isozyme and RFLP markers the same as in the previous potato maps (Bonierbale et al., 1988, Tanksley et al., 1992), although recombination distances differ. This may be due to the utilization of different species in the mapping population. In the map developed in this study, a total of 46 RAPDs were incorporated. These markers show high polymorphism and repeatable results in potato. While the number of isozyme loci is limited, there is an immense number of random primers available for use, many of which resolve several segregating loci. The PCR technique is relatively simple, and the time necessary to obtain results is short as compared with RFLP analysis. These characteristics make RAPD markers a valuable addition to QTL analysis and marker- assisted techniques, particularly when using backcross—type populations where there are only two genotypic classes. 50 Significant association was found between tuber dormancy and 22 markers. Seventeen of these markers were heterozygous in the female parent and identified 6 QTLs, one on each of chromosomes 2, 3, 4, 5, 7 and 8. Additionally, 5 markers segregating from the male parent also showed significant association with the trait. These loci were not linked to each other, and their chromosomal location could not be established. The two genotypes with longest dormant periods in the population had 82% and 95% of the favorable alleles of the significant markers, thus corroborating their effects on the trait. On each of chromosomes 3, 4 and 5, only one significant marker was identified. On chromosome 5, this may be due to the fact that 6-Pgdh-3 is not closely linked to any other marker. For Pgm-Z on chromosome 4, the effect of this QTL might be too small to be detected with other linked markers, while 619.1 on chromosome 3 has a high P value (P=0.049) and might be a false positive. The highest number of significant markers was identified on chromosome 7. On this chromosome nine markers are clustered on a region spanning 49 cM, eight of which are RAPDs. The values of R2 for individual markers ranged between 4.2% and 20.4% which represents a difference of 3.3 and 7.7 days to sprouting between the means of the genotypic classes for the markers, respectively. On average, the highest effect on the trait is by the markers on chromosome 7, each of which contributes with more than 12% of the phenotypic variation of the trait. The isozyme marker Got-2 has markedly the highest effect, explaining 20.4% of the variation. All significant loci on this QTL were segregating from the female parent. This QTL constitutes an important region of the genome associated with long dormancy and could be tagged in future generations utilizing the markers identified. A total of 7% of the possible epistatic interactions between significant markers were also significant in this study, as compared with 3% and 1% in two different studies in maize (Edwards et al., 1987; Stuber et al., 1992). When these were included in the multiple analysis model, the amount of phenotypic variation of dormancy explained by the markers was 72.1%, giving an increase of 15% over the main effects model. Therefore, epistatic interactions seem to be contibuting significantly on the variation of at” 51 this trait. To maintain these interactions in future generations, transfer of intact portions of the genome might be necessary. In potato breeding improved 2x germplasm is uansferred into the cultivated 4x level using 2n gametes (Chase, 1968; Iwanaga 1983; Peloquin et al., 1989). These gametes are produced either by genetically equivalent FDR (First Division Restitution) or SDR (Second Division Restitution) mechanisms. 4x-2x crosses with FDR gametes might be more appropriate in breeding for this trait since they transfer approximately 80% of the genome intact from parent to offspring, thus maintaining a large fraction of the epistatic interactions (Peloquin and Ortiz, 1991). This study in potato differs from traditional QTL studies performed on inbreeding crops. 1n potato the generation of inbred parents is not practical due to self- .F‘T-“El incompatibility and inbreeding depression at the 2x level. Therefore, heterozygous parents must be used to develop the mapping population, and both parents can contribute markers associated to the trait. Secondly, utilization of 0.01 or 0.001 significance levels have been suggested to avoid identification of false positives (lander and Botstein, 1989). In this first study on this trait, the less stringent level of 0.05 was utilized as indicated by Soller and Brody (1976). RAPD markers have also been incorporated in the QTL analysis, which has not been previously reported. The markers utilized are not evenly spaced in the map and information for two chromosomes is missing. Although it is recognized that the potato genome was not completely surveyed, 6 QTLs were identified which explain 57.5% of the phenotypic variation for tuber dormancy, a value that is similar to that found in studies with quantitative traits in inbreeding crops. Furthermore, the QTL identified on chromosome 7 which has a significant effect upon tuber dormancy may indicate major gene control of this complex trait, and could be used in marker- assisted selection. 52 LIST OF REFERENCES Bogucki S, Nelson DC (1980) Len th of dormancy and sprouting characteristics of ten potato cultivars. Am Potato] 57:1 l-157 Bonierbale M, Plaisted R, Tanksley SD (1988) RFLP maps based on a common set of 1110‘? reveal modes of chromosomal evolution rn potato and tomato. Genetics 120: 1095- Burton W6 (1963) Concepts and mechanisms of dormancy. In: Ivins JD, Milthorp FL (eds) Growth of the Potato. Butterworth, London. pp 189-284 Burton WG (1966) The Potato. A survey of its historyand factors influencing its yield, nutritive value, quality and storage. Veenman, Wagenrngen Chase SA ((1968) Analytical breeding in S. tuberosum L. A scheme utilizing parthenotes and other iploid stocks. Can J Genet Cytol 5:359-363 Coleman WK (1987) Dormancy release in potato tubers: a review. Am Potato J 64:57-68 Diers BW, Cianzio SR Shoemaker RC (1992 Possible identification of uantitative trait loci affecting iron defficiency rn soybean. J lant Nutntron 15:2127-21 6 Edwards MD, Stuber CW, Wendel J F (1987) Molecular;marker-fa_cilitated investigations of quantitative-trait locr rn marze. 1. Numbers, genomrc distributron and types of gene action. Genetics 116:113-125 . Flewelling HS (1987) Use of haploid Tuberosum-wild Solanum species Fl hybrids to study the relationship between the cultivated and Wild Potatoes, and to analyze the genetrc control of tuber dormancy. MS Thesis. University 0 Wisconsrn-Madrson. Gebhardt C, Ritter E, Barone A, Debener T, Walkemeier B, Schachtschabel U Kaufmann H, Thom son RD, Bonierbale MW, Ganal MW, Tanksley SD, Salamini F 1991) RFLP maps 0 potato and their alignment with the homoeologous tomato genome. eor Appl Genet 83:49-57 Hackett CA, Ellis RP, Forster BP, McNicol JW, Macaulay M (1992) Statistical analysis of a lrnkage. expenment rn barley involving uantitative trait loci for hei ht and ear- erzn6ergence trme and two genetrc markers on c romosome 4. Theor Appl ent 85 :120- Hayes PM Blake T, Chen THH Chen F, Pan A, Liu B (1992) Quantitative trait loci on bar ey (Hordeum vulgare L. chromosome 7 associated with components of winterhardiness. Genome 36:66-71 Hemberg T (1985) Potato Rest. In: Li PH (ed) Potato Phisiology. Academic Press Inc. pp 353- 88 Hermundstad SA (1986) Ha loid-wild s ies hybrids in potato breeding, genetics, and germplasm enhancement. P D Thesis. niversrty of Wisconsin-Madison Hermundstad SA, Peloquin SJ (1985) Germplasm enhancement with potato haploids. J of Hered 76:463-467 Heun M (1992) Ma pin quantitative powdery mildew resistance of barley using a restriction fragment eng polymorphism map. Genome 35:1019-1025 Iwanaga, M. 1983). Ploidy level manipulation approach: develogment of diploid populations wrt specrfic resrstance and FDR 2n pollen production. In: resent and uture 53 stratc ies for tato brwdin and im rovement. Re rt of the 26“I Plannin Conference, CIP- cc. 1983, Lima, Perfi. p p0 g Jeoung LC, Iritani WM, Martin MW 9983) Comparison of methods for measuring dormancy of potatoes. Heredity 19:489- 04 Keim P, Diers BW, Shoemaker RC (1990) Genetic analysis of soybean hard seedeness wrth molecular markers. Theor Appl Genet 79:465-469 Lander ES, Green P, Abrahamson J, Barlow A, Daly MJ, Lincoln SE, Newburg L (1987) MAPMAKER: .an rnteractrve com uter aekage for constructrn primary genetic linkage maps of experimental and natu popu atrons. Genomrcs 1:1 4-181 lander E, Botstein D (198% Mapping Mendelian factors underlying quantitative traits using RFLP linkage maps. enetrcs 1 1:185-199 Miura H, Parker BB, Snape JW (1992) The location of major genes and associated quantitative trait loci on chromosome arm 5BL of wheat. Theor Appl Genet 85 : 197-204 Nienhuis J, Helentjaris T, .Slocum M, Ruegero B, Schaefer A. (1987) Restriction fragment length. lymo hrsm analysis of ocr assocrated wrth insect resrstance rn tomato. Crop Scr 7:797- 03 Paterson AH lander ES, Hewitt JD, Peterson S, Lincoln ES, Tanksley SD (1988) Resolution of quantitative trarts rnto Mendelian factors, usrng a complete lrnkage map of l’eStI'ICthl‘l fragment length polymorphisms. Nature 335:721-726 Paterson AH, DeVerna JW, lanini B, Tanksley SD (1990) Fine mapping of quantitative trait loci usrng selected overla fing recombinant chromosomes, m an rnterspecres cross of tomato. Genetrcs 124:735- 2 Paterson AH, Damon S, Hewitt JD, Zamir D, Rabinowitch HD, Lincoln ES, Lander ES, Tanksley SD (1991) Mendelian .factors underlying quantrtatrve traits in tomato: companson across specres, generations and envrronments. Genetrcs 127:181-197 Peloquin SJ, Yerk GL Werner JE, Darmo E (1989) Potato brwding with haploids and 2n gametes. Genome 31:1000-1004 Pel uin SJ, Ortiz R (1991) Techniques for introgressin unadapted germplasm to br mg populations. In: Stalker HT, MurphydP (eds) P ant Breedrng m the 1990s. Proceedings of the SKlrgposrum on Plant Breedrn 1n the 1990s. North Carolrna State University. Raleigh, March 1991. pp 485-50 Ritter E, Gebhardt C, Salamini F (1990) Estimation of recombination frequencies and construction of RFLP linkage maps rn plants from crosses between heterozygous parents. Genetics 125:645-654 Ruttencutter 6, Ha nes F, Moll R 9979) Estimation of narrow-sense heritability for ific gravit in iploid potatoes ( tuberosum subsp. phureja and stenotomum) Am otato J 6: -453 Saghai-Maroof MA, Soliman KM, Jorgesen RA, Allard RW (1984) Ribosomal DNA spacer length polymorphrsms in barley: Mendelian rnheritance chromosomal location, and population dynamics. Proc Natl Acad Scr USA 81:8014-8018 Simmonds NW. (1964) The 0genetics of seed and tuber dormancy in the cultivated potatoes. Heredity 19:489-5 Sisco PH, Senior ML, Rhyne DC (1990) RFLP techniques. laboratory Manual. USDA- ARS, North Carolrna State Unrversity, Raleigh NC. 54 Soller M, Brody T (1976) On the power of exo’erimantal desi ns for the linka e between T‘la'r3KSCr39OCi ans quantitative loci in crosses etween inbr lines. Theor ppl Genet Stuber CW, Edwards MD, Wendel J F (1987) Molecular marker-facilitated investigations of quantitative trait locr rn marze. 11. Factors rnfluencrng yield and its component traits. Crop Scr 22:737-740 SuiterKS, Wendel JF, Case JS (1983) LINKAGE-1: a Pascal computer program for the detection and analysis of genetic linkage. J Hered 74:203-204 Tanksley SD, Medina-Filho H, Rick CM (1982) Use of naturally-ocurring enzyme varratron to detect and map genes controlling quantrtatrve trarts rn an interspecific backcross of tomato. Heredity 49:11-25 Tanksley SD, Hewitt J (1988) Use of molecular markers in breeding for soluble solids content in tomato - a re-examrnatron. Teor Appl Genet 75 :81 1-823 Tanksley SD, Ganal MW, Prince JP, de Vicente MC, Bonierbale MW, Broun P, Fulton TM, Grovannoni JJ, Grandillo S, Martin GB, Messe uer R, Miller JC, .Mrller L, Paterson AH, Pineda O, Roder MS, Wrn RA, Wu W, ougg ND (1992) Hrgh densrty molecular linkage maps of the tomato an potato genomes. enetrcs 132:114 -1160 'Ihom son P, Ha nes F, Moll R (1980) Estimation of genetic variance components and henta rlity for tu er dormancy in diploid potatoes. Am Potato J 57:39-46 Vau hn SF, S ncer GF (1991) Volatile monote nes inhibit tato tuber s routin . Am Potaio J 68:851e 831 ”’6 9° P 8 Weller J1 Soller M, Brody T (1988) Linkage analysis of quantitative traits in an rnterspecrfic cross of tomato (Lécopersicon escu entum x Lycopersicon pimpinellifolium) by means of genetrc markers. enetrcs 118:329-339 Williams JCK, Kubelik AR, Livak KJ, Rafalski JA, Tingey SV (1990) DNA golymo hisms amplified by arbitrary primers are useful as genetrc markers. Nucl Acids es 18: 531-6535 CHAPTER THREE 8UANTITATIVE TRAIT LOCI ANALYSIS OF SPECIFIC GRAVITY IN DIPLOID P TATO (Solanum spp.) OVER ENVIRONMENTS: DEVELOPMENT OF A MODEL FOR MARKER—ASSISTED SELECTION ABSTRACT. Dry matter content in potato, which is an important factor in potato processing, is estimated through specific gravity. In this study we performed quantitative trait loci (QTL) analysis for this trait in 2x potato over 3 environments. The population used in this study consisted of 110 individuals and was derived from the cross of a hybrid of haploid S. tuberosum (2x) and S. chacoense, with a S. phureja clone. This population was characterized for 10 isozyme loci, 44 RFLPs and 63 RAPDs, and 87 of these loci segregating from the female parent used for mapping. Field trials were conducted in two locations in Michigan in 1990 using 3 replications, and a third field trial was conducted in 1991 with 90 individuals and 2 replications. At each location, specific gravity was determined through the weight in air/weight in water method. QTLs were mapped separately for each location and for their average over environments by one-way analyses of variance for each marker locus by trait combination. A total of 10 QTLs were identified over environments and they were localized on chromosomes 1, 2, 3, 4, 7, and 11. The numbers and effects of QTLs detected varied across environments. The locus with highest R2 value per QTL in each location was chosen to develop multilocus models. Each of these was used in a multiple analysis of variance with data from the location it was developed. The models explained from 39% to 45% of the phenotypic variation for the trait. Each model was also tested with data from the other environments, and in general the predictive value across locations was low. Meanwhile, using the average data a model that gave consistent results when tested across environments was developed, which is comparable to the best model developed for each case. This model may be valuable for marker-assisted selection at the seedling stage in a potato breeding program. 55 56 WODUCHON The proportion of dry matter and water in potato tubers determines to a great extent its food value and culinary quality. A good mealy potato will consist of about 25% dry matter and 75% water, and as dry matter decreases potato sogginess will increase (Chase et al., 1990). Dry matter content can be measured directly by oven drying but this is time consuming and involves a loss of sampling material. A more common practice is the estimation of dry matter from specific gravity. These two characters have a simple linear relationship which is highly correlated; regression equations have been developed for 4x and 2x potatoes (Wilson and Lindsay, 1969; Schippers, 1976; Simmonds, 1977; Wannamaker et al., 1992). High specific gravity is particularly important in the potato chip industry because it is associated with increased chip yield and superior quality product. Chips produced from high specific gravity potatoes absorb less oil during the frying process and are therefore more desirable and cheaper to produce. A specific gravity greater than 1.080 which is equivalent to 21.2% dry matter content is preferred by the chip industry (Gould, 1989). The cultivated potato, Solanum tuberosum spp. tuberosum is tetraploid (2n=4x=48), however, over 70% of the tuber-bearing Solanums are diploid (Hawkes, 1990). These species represent a valuable source of germplasm that can be used to broaden the genetic base of the potato, and provide specific desirable traits. For example, high specific gravity levels have been found in selections of South American diploid species, and furthermore, progress in selection for high specific gravity among 2x populations has been attained (Ruttencutter et al., 1979). These diploid wild and cultivated tuber-bearing species can be used in potato breeding at the diploid level using haploids of cultivated species, and then the improved 2x germplasm can be transferred to the cultivated 4x level through 2n gametes (Chase, 1968; Iwanaga, 1983; Peloquin et al., 1989). Little is known about the genetic control of specific gravity but it is generally treated as a quantitative character in breeding (Haynes and Haynes, 1983). Other quantitative traits in crops such as maize (Edwards et al., 1987; Stuber et al., 1987), 57 tam-‘0 (Weller et al., 1982; Tanksley et al., 1982, Tanksley and Hewitt, 1988; Paterson ct al., 1988), soybean (Keim et al, 1990; Diers et al., 1992), wheat (Miura et al., 1992) and barley (Hayes et al., 1992; Heun, 1992; Hackett et al., 1992) have been studied using molecular markers. The availability of saturated linkage maps makes it possible to dissect quantitative traits into discrete genetic factors (QTLs) and their phenotypic effects and genetic position can be estimated "(Paterson et al., 1988; Lander and Botstein, 1989). Recently, the effect of environment on QTLs was studied in F2 and F3 populations in tomato (Paterson et al. , 1991) and in F3 lines backcrossed to the parents in maize (Stuber et al., 1992). The potato is a clonally propagated crop, and therefore offers a particular advantage for this type of analysis. Genotyping with molecular markers and evaluation of traits can be performed on exactly the same plant material in the first generation, M" allowing the study of effects of QTLs across different environments and years. The use of isozymes for QTL analysis of specific gravity was described in Chapter 1. Here that research is complemented through the addition of RFLP and RAPD markers. Furthermore, the effect of environment on QTL expression for this trait at three locations has been analized. Multilocus models with markers representing the QTLs were developed for each location to determine the contribution of the QTLs to the phenotypic variation of the trait. Additionally, the data from the average of the three locations was used to develop a more stable model than the ones developed at each one of the environments. This model could be used for marker-assisted selection in future generations using this material. MATERIALS AND METHODS BlaaLMatsLial A diploid F, population named TRP133 and consisting of 110 genotypes was utilized in this study (as in Chapters 1 and 2). This population is derived from the cross of clones 84SD22 (a hybrid between haploid Solanum tuberosum and S. chacoense), and 84810 (S. phureja) as female and male parents, respectively. 58 ' \ Tri The clonal material was planted in three environments during two years: at Montcalm Experiment Station, Edmore, MI in 1990 and 1991 (MES90 and MES91), and at Clarksville Horticulture Experiment Station, Clarksville, Michigan, in 1990 (CHES90). A randomized complete block design (RCBD) was utilized, with 8 plants per plot, and spacing of approximately 0.3 m and 0.9 m within and between rows, respectively. In 1990, the 110 genotypes were planted with three replications, while in 1991 due to availability of material, only two replications and 90 of the genotypes were used. WM! After harvest at each location, specific gravity was determined for all genotypes using the weight in air/weight in water method: [air wt./(air wt. - water wt.)]. A minimum sample size of l kg/plot was used. The value of specific gravity for each genotype was obtained from the mean of either the three or two values from each of the replications in the field. n rn The genotypes in the population were characterized for the morphological marker yellow flesh (Y), 10 isozyme loci, 44 RFLPs, and 63 RAPDs. RFLP probes were kindly provided by S. Tanksley at Cornell University, and C. Gebhardt at Max Planck Institut, Germany. RAPDs were resolved using commercial 10—mer primers (Operon Technologies). All the markers used for the analysis were heterozygous in one of the parents and homozygous in the other, thus segregating as a BC, (1:1) in the progeny. Most of the markers were segregating from the female parent 84SD22, and these were used for construction of the linkage map with MAPMAKER (lander et al., 1987) v.01 for Macintosh as described in Chapter 2. Stat-mm Data from the three locations was tested through an ANOVA which partitioned 59 (he effects of location, replications/locations, genotypes, and genotype x location. The memoclology utilized for QTL analyses has been previously described for another trait, tuber dormancy (Chapter 2). Briefly, linkage of QTL to a marker locus was determined with F-tests in single factor ANOVAs between each pairwise combination of specific gravity data and genotypic classes for the marker locus (PROC GLM, Statistical Analysis Systems, Cary, NC). A significant difference in means (P < 0.05) was interpreted as linkage of the QTL to the marker locus. When two or more significant markers were found on the same linkage group, they were considered to be linked to independent QTLs if they were separated by more than 50 cM (Paterson et al., 1991). In this study, QTLs were identified for each environment, and also for the mean of specific gravity from the three environments (from now on described as AVE). AVE data was obtained from the 90 individuals for which there were values at the 3 locations. Epistatic interactions between significant markers at each location were tested by two-way analyses of variance. For each environment, a model with the markers with the highest R2 value per QTL was developed. This was used in a multiple analysis of variance to predict the total variation for specific gravity explained by the identified QTLs. As a matter of comparison, the model obtained at each location was tested with data from the other environments. A model was also developed with data from AVE and tested separately in each one of the locations, using 110 individuals for MES90 and CHES90, and 90 individuals for MES91 and AVE. Furthermore, the significant epistatic interactions at each location were included in the main effects models to determine their contribution to the phenotypic variation for the trait. When there were several interactions between markers linked to the same pairs of QTLs, the interaction with the highest R2 value was utilized. The main effects of the markers in the interactions were also included in the model if not already present. 60 RESULTS The dates of harvest were 119, 131 and 120 days from planting for MES90, CI-ES90 and MES91, respectively. The frequency distributions (Figure 3.1), and the range of values and mean for the genotypes and parents at the three locations are shown (Table 3.1). On average, values of specific gravity were higher at CHES90, and MES90 had higher values than MES91. Both MES90 and CHES90 had ranges of 0.053 specific gravity units, equivalent to 10.3 % dry matter content, and at MES91 the range was 0.043 or 8.4% dry matter. For the AVE data, the mean was in between that of CHES90 and MES90, and the range was 0.037 or 7.2% dry matter. Phenotypic correlation of specific gravity data between MES90 and CHES90 was 0.81; between MES90 and MES91 it was 0.72; and between CHES90 and MES91 it was 0.71. The results from the combined analysis of variance over the three locations are shown in Table 3.2. Locations, genotype, and genotype x locations effects were significant. Table 3.1. Values of specific gravity for parents and population TRP133 at each one of the three environments and the average Environment Range Mean i SE 90: Pogulation 1.046 - 1.099 1.079 1 0.001 84 D22 1.080 i 0.003 481 1.067 i 0.002 CHES90: Pogulation 1.057 - 1.110 1.083 :1: 0.001 84 D22 1.084 :t 0.011 84S10 1.064 t 0.002 MES91: Pogulation 1.052 — 1.095 1.075 :t 0.001 84 D22 1.077 :1; 0.003 481 1.060 -_l- 0.001 VE: Population 1.063 - 1.100 1.080 :t 0.001 61 30 25 6‘ C 20 0) 3 8' 15 .1: 10 c n 4 1.045 1.055 1.065 1.075 1.085 1.095 1.105 1.05 1.06 1.07 1.08 1.09 1.1 1.11 specific gravity [:1 MES 90 - CHES 90 ME891 Figure 3.1. Frequency distributions of specific gravity values at the three environments 62 {£516}: &}£misrq)olara?ng/IS) from the combined analysis of variance of specific gravity \ Source (if MS x 10° H R ltion (Loca ' ) 5 21398‘? * rcatrons trons Gigotype . 109 1310 ** genotype x Location 532 2(2): *** or Total 838 *, **,. *** indicate significance at the 0.05, 0.01 and 0.001 probability levels, respectively. Results from the one-way ANOVAs between marker loci and specific gravity data for each location and AVE are shown (Table 3.3). A total of 28 loci were significant in at least one environment or AVE. Twenty-four loci were segregating from the female parent identifying 10 QTLs, and their positions were localized on chromosomes 1, 2, 3, 5, 7 and 11 (Figure 3.2). Since the loci segregating from the male parent were not mapped, the position of the three loci segregating from this parent was not identified. Two loci which were not significant in any of the 3 locations ('TGZ4T and T614) were significant in AVE. None of the RFLP probes selected by their previously known positions on chromosomes 9 and 12 (Bonierbale et al., 1988; Tanksley et al., 1992) could be scored successfully due to lack of polymorphism in this population or technical problems, so no loci could be assigned to these chromosomes. The number of significant loci identified in each one of the locations was: 19 loci in MES90, 18 in CHES90, 16 in MES91, and 19 in AVE, representing 7, 7, 5 and 7 QTLs, respectively. Two of 10 QTLs were identified in the three locations and also in AVE; 5 were identified in 2 locations, and 3 of these also in AVE; other 4 were identified in only one of the locations, and 2 of these also in AVE. The number of QTLs in common between locations was: 3 QTLs between MES90 and CHES90; 4 QTLs between MES90 and MES91; and 3 QTLs between CHES90 and MES91. From the QTLs identified in two of the locations: 2 were at both MES90 and MES91; 1 at MES90 and CHES90; and 1 at CHES90 and MES91. The values of phenotypic variation for the 63 Table 3.3. Significant loci, chromosome locations and R2 values for the three environments and the average \V Locus Chrom Pl R2 (%) t MES90 CHES90 MES91 AVE I Bet. in 9': T T627 1 ns" 7.5 ** 5.7 * 6.7 * . ll F04.1 2 5.9 * ns ns 5.7 * 1 612.1 2 5.8 * ns 5.5 * 6.6 * , ! F02.l 3 ns 4.8 * ns ns y , P m-I 3 7.0 * 8.2 ** ns 5.8 * l l T 152C 3 4.6 * ns ns ns .1 i TG24T 5 ns ns ns 6.0 * A15.1 5 ns 4.5 * ns ns 1 1 6(1)?th 5 5.0 * 15.0 *** 6.7 * 12.2 *** i l H .1 5 ns 4.0 * ns ns 1 a A15.2 7 4.7 * 7.8 * 6.6 * 7.7 * L I Got-2 7 4.5 * 10.1 ** 10.1 ** 13.9 *** i , A01.3 7 6.1 * 10.5 ** ns 7.3 * l 111.2 7 5.8 * 13.4 *** 5.4 * 10.3 ** I A082 7 7.7 * 15.8 *** 6.8 * 13.1 *** 1 617.2 7 6.0 * 14.9 *** 4.9 * 10.7 ** l A01.2 7 5.8 * 11.3 ** 5.4 * 9.1 ** g , A04.l 7 6.2 * 11.7 ** 6.5 * 10.2 ** g | 117.1 7 9.7 ** 14.8 *** 5.8 * 12.7 *** '1 T613T 7 ns 5.5 * » ns 4.9 * i T613B 7 5.2 * 5.2 * ns ns 1 11 6.3 * ns ns ns 1 l 11 7.2 ** ns 6.9 "‘ ns . l 11 5.7 * ns 5.3 "‘ ns 1 i 11 9.5 ** ns 9.6 * ns ‘ 1r ——— __ w I I ns 7.1 * 8.1 ** 10.4 ** l I ns ns 6.4 * 5.9 * , ns ns ns 5.0 * l ‘ indicates the loci that are heterozygous in the female and male parents, respectively “ ns indicates not significant ° the positions of markers segregatin from the male parent have not been identified *, **, *** indicate significance at 03) 5, 0.01, 0.001 probability levels, respectively Figure 3.2. Molecular linkage map and localization of QTLs for specific gravity for each environment and AVE ' ' QTLs are indicated by bars which define the position on the chromosome, not necessarily the srgnificant markers 65 w>< ”V/fufl pm mm: I om mg I On mm: H Soon I Q36 v.m8 noun: hnnmwh '10» NF p .58... i 30h i .31. «~91- p.20 ll “-52:81 —.m—< . / —.N—< I .38... J . p .T J] In Gun—0h Gum—0... 0.0.0 .6—0 «.0: .62 New. —.—o< €32 avmvh .N —< 2 ea... or rah: antics \\\\\\\\‘ .68 Saw :00 90h Gabi :90... i - on... i «.80 i. .38... i- 0.9: i v.—o< II. «.33 i - ..«.e .L E 2.8 L r. I mn—G» .220.— “K: «.32 «.26 «.22 N... 2 n.—c< «.00 ~.m—< - 1:1 - 3 < \\\\\\\\\\\\\\\\ \\\\ I - - .8. 30h .58h E F 66 trait explained by individual loci, determined by their R2 value, ranged from 4% to 15.8%. Most loci had R2 values between 4% and 8%. The highest values were identified in CHES90. Mean R2 values per location were 6.2 for MES90; 9.56 for CHES90; 6.6 for MES91; and 8.64 for AVE. The locus with highest R2 value per QTL in each location was chosen to develop multilocus models. All loci segregating from the male were also included since their position was unknown and it could not be determined whether they were identifying different QTLs. The models developed are shown in Table 3.4. Seven loci were selected in MES90 (7 QTLs), eight loci in CHES90 (7 QTLs, 1 unmapped), seven loci in MES91 (5 QTLs, 2 unmapped), and 9 loci for the AVE model (7 QTLs, 2 unmapped). The amount of the phenotypic variation for the trait explained by each model its own data and that from the other environments is shown (Table 3.5). Table 3. 4. Multilocus models developed at each one of the environments and with the average data MES90 I CHES90 MES91 AVE Main Effects F04 1 T627 T627 T627 P m-I Pgm- 612.1 612 1 T 152C 6.6?‘1’1-3 6-Pgdh-3 P m-I 6—Pgdh-3 H . Got-2 6-P§dh-3 1 A15.l T626 T613B A082 120. 1 Got-2 T626 T6 13T T6 1523 T6 13T 120.1 T6152B T614 Significant Interactions F04.1 * T630 117.1 * 6-P dh-3 120.1 *117.1 120.1 * P m-I 612.1*111.2 T627 * T6 3T 120.1 * 6-P dh- 3 111.2 * T 1523 117. 1 * 6-P dh-3117.1 * T6 7 117.1 * 6-Pgdh-3 F13. 2 * T 152C A15. 2 * T6152B 67 Table 3.5: R2 values obtained from the multiple analyses of variance using the multilocus models wrth data from the different environments . Model Developed Environment‘ MES90 CHES90 MES91 AVE MES90 39.0b . 27.3 32.7 37.4 56.5 - - 41.1 CHES90 44.0 44.9 36.8 54.9 ll - 53.9 - 61.2 MES91 26.2 23.9 42.6 - - 64.2 AVE - - - ' indicates environment used. to test the models _ . . _ " upper value rs R2 from marn effects model; lower value rs wrth rnclusron of the srgnrficant rnteractrons. A total of 15, 10, 9 and 10 epistatic interactions were significant at MES90, CHES90, MES91 and AVE (data not shown). These correspond to 8.8%, 6.5%, 7.5% and 5.8% of all possible interactions between significant markers at each one of the locations. The interaction with highest R2 was utilized when there were several that showed significance between the same pairs of QTLs. The interactions used for each one of the locations and AVE are shown (Table 3.4). The resulting R2 values when the interactions were included in each of the main effects models are shown (Table 3.5). DISCUSSION Specific gravity is influenced by a number of environmental factors such as temperature, rainfall, day length, etc. (Stevenson et al., 1954). Genotype x environment interactions were found to be significant for this trait in studies with 4x potatoes (Johansen et al. , 1967); however, inherent differences among varieties are apparent over a wide range of environmental conditions (lana et al., 1970). large genotype x 68 environmental effects were also found in diploid populations of S. phureja and S. stenotomum (Ruttencutter et al., 1979), and this is also confirmed in our results using S. tuberosum, S. chacoense and S. phureja material. This fact raises interesting questions that have been adressed in this study: a) what is the effect of environment on the QTLs detected; and b) is it possible to develop a model that will best explain the phenotypic variation for the trait across different environments and thus have predictive value. The utilization of a significance level of 0.01 or 0.001 has been recommended in QTL analysis to reduce the risk of accepting false positives (lander and Botstein, 1989). However in this study we chose to use the less stringent level of 0.05 as indicated by Soller and Brody (1976) for the individual locations and then judge the consistency of significant markers across locations. The total number of significant loci detected was 28, 25 of which were segregating from the female parent and were mapped. This hybrid parent had more heterozygous loci and higher specific gravity than the male consistently across all locations, therefore its larger contribution of loci associated with the trait is not surprising. Nevertheless, three loci segregating from the male parent were also identified. There were differences in the significant loci identified at each environment, even though the total number of loci in each one of them was very similar. These loci identified a total of 10 QTLs on 6 chromosomes. One QTL was identified on each of chromosomes 1, 2 and 11; 2 distinct QTLs were identified on chromosomes 3 and 7; and 3 QTLs on chromosome 5. The marker loci were not evenly spaced across the genome and in some cases there was a cluster of significant loci identifying the same QTL. For example, on chromosome 7, nine significant loci were mapped to a chromosome region spanning 49 cM. The locus with the highest R2 value at a given QTL was not always consistent across environments, for example on chromosome 7, A08.2 was highest in MES90, 117.1 in CHES90, and Got-2 in MES91 and AVE. Also, on all cases except for MES91, there are multiple peaks on this QTL as indicated by the individual R2 values. Nevertheless, due to the closeness between the loci, this would not necessary indicate multiple QTLs (Paterson et al. , 1991). Two of the ten QTLs on chromosomes 5 and 7 were identified in all environments, therefore showing a strong and stable association with specific gravity. 69 Interestingly, these two QTLs also showed association with another quantitative trait in potato, tuber dormancy (Chapter 2), even though there is no correlation between the two tuber traits in this material. The similarity of results for the two traits suggest either pleiotropic effects of single QTLs, or clustering of different QTLs into closely linked groups as explained by Paterson et al. (1991) in tomato. Five other QTLs (50%) were identified in two of the environments and 2 of them were significant also with AVE. The other 3 QTLs (30%) were specific for only one of the environments. One of these was also significant with AVE, while the other two, T6152C in MES90 and H04.1 in CHES90 have low significance levels (4.6% and 4%, respectively) and therefore could possibly be false positives. There is more consistency among the tagged QTLs across environments in this study than in the similar study in tomato (Paterson et al., 1991), where 14%, 34% and 52% of the QTLs detected were identified in 3, 2 and 1 environment, respectively. This could be due to the fact that the environments they used (California and Israel) were more different than the ones used in this study, and as noted in their article, their comparison across environments was confounded by the use of different generations and methods of trait evaluation. ’ The results from the comparison of QTLs across environments, seem to indicate that the two MES trials were more similar to each other than to CHES, even though they were performed in different years. This seems contradictory to the fact that the correlation coefficient was highest between MES90 and CHES90. Nevertheless, this value indicates a similar response of genotypes in different environments and not the overall response of the population in these environments. Also, lower correlation values with MES91 data could be confounded by the use of the smaller population size compared to 1990 (90 versus 110 individuals). The values of phenotypic variation for the trait estimated by the R2 values of individual loci ranged from 4% to 15.8%. These values correspond to a difference of 0.7% and 1.3% dry matter between the means of the marker classes, respectively, as estimated by different methods (Schippers, 1976; Simmonds, 1977; Wannamaker et al., 1992). These are important differences when considering that a difference between 1.075 and 1.080 specific gravity, that can determine acceptance of processing potato varieties, 70 represents a difference of only 1% dry matter. Most loci have only small effects on the trait, as indicated by the fact that most R2 values are between 4% and 8%. On average, R2 values were highest at CHES90, which was the only environment with R2 values higher than 14%. This can be due to the longer growing season for this trial (11 more days to harvest) as compared with both trials at MES. An important part of this study was the development of multilocus models to estimate the phenotypic variation for the trait explained by the QTLs at each one of the locations, make comparisons across them, and develop a model with the best predictive value. However, it is important to note that the analyses are affected by different factors: a) the use of 110 individuals in tests using 1990 data versus only 90 individuals for 1991 and AVE; b) the models were tested only on individuals that had complete sets of data for all loci involved, and due to missing values, in some extreme cases this was limited to numbers as small as 64 individuals; c) the different number of loci utilized in each one of the models. There were differences in the models developed for each of the locations due to variations in the locus with the highest R2 value per QTL, and in the loci segregating from the male parent that were also included. Nevertheless, the portion of the phenotypic variation of specific gravity explained by the respective models is quite similar: 39% in MES90, 44.9% in CHES90, and 42.6% in MES91. The poor predictive values of each of these models is demonstrated by the lower R2 values when they are tested with data from the other environments: 44% and 26.2% for the MES90 model tested on CHES90 and MES91, respectively; 27.3% and 23.9% for the CHES90 model tested on MES90 and MES91; and 32.7% and 36.8% for the MES91 model tested on MES90 and CHES90. On the other hand, the model developed with data from the average of the three environments (AVE) explains a distinctively higher portion of the variation of specific gravity when tested with its own data (57.1%). Moreover, when tested with data from each of the locations it gives consistent results, which are comparable to using the best model for each location: 37.4% in MES90, 54.9% in CHES90, and 41.2% in MES91. The AVE model consists of 7 loci (the same number as MES90 and CHES90 models) plus 2 unmapped loci (one more than the CHES90 model), so the results can not 71 be attributed to the use of a vastly larger number of loci. CHES90 has high values throughout the analysis which could be due to the stronger effect of the loci as indicated by their R2 values, or by being influenced by the confounding factors already mentioned. The numbers of epistatic interactions that were significant represent 8.8%, 6.5 %, 7.5% and 5.8% of all possible interactions for MES90, CHES90, MES91 and AVE, respectively. These values, according to Paterson et al. (1991), would represent only minimal evidence of epistasis. Nevertheless, their effects can not be underestimated, since they give increments of 17.5%, 9% and 21.6% with respect to the R2 values from the main effects models for each one of the locations. Here too, a difference between environments can be detected on the numbers of significant interactions identified, the QTLs involved, and their effects on the trait. However, when included in the AVE main effects model the effects are not as dramatic and they give increments of 5.3%, 3.7%, 6.3% and 8% when tested with AVE, MES90, CHES90 and MES91 data, respectively. This study has unequivocally demonstrated the influence of environmental effects on QTLs associated with specific gravity in potato, by testing the same plant material across three different environments. For breeding purposes, it is important to note that the predictive value of multilocus models developed with data from individual locations is not always effective when used on other environments. On the other hand, the best multilocus model was developed when the data was averaged over the three environments. The value of this model is that the loci involved could be tested in marker- assisted selection in future generations using this material. Moreover, it is of interest to investigate the consistency of these markers in germplasm involving other wild relatives of potato, and the transfer of the QTLS from the 2x to the 4x level. Specific gravity is relatively easily evaluated and traditional potato breeders may argue that there is no value in attempting an indirect selection method. Nevertheless, in potato breeding, both at 2x and at 4x levels, determination of specific gravity is usually not performed at earlier stages of selection but up to 4 years after the initial cross was made. Even though a selection method for high dry matter in seedling generation was reported (Lam and Grenard, 1976), the estimations of specific gravity based on only one plant, and the utilization of a small sample size can not be accurate. In addition, 72 greenhouse-grown 2x seedling tubers in particular are usually very small and have therefore low specific gravity values and high variability (Cole, 1975). Even in field- grown plants, the small sample size from one seedling plant generally resuls in inaccurate estimates of specific gravity. An indirect selection method based upon tagged QTLs associated to the trait is feasible at the seedling stage, and may prove adequate and time- saving to improve the dry matter content in potato. LIST OF REFERENCES Bonierbale M, Plaisted R, T anksley SD (1988) RFLP maps based on a common set of (111001368 reveal modes of chromosomal evolution in potato and tomato. Genetics 120: 1095- Chase SA 31968) Analytical breeding in S. ruberosum L. A scheme utilizing parthenotes and other iploid stocks. Can J Genet Cytol 5:359-363 Chase R, Silva 6, Douches D, Hammerschmidt R (1990) Selecting potato varieties for Miehigan. Best Management Practices. for Potatoes bulletin senes. Mrchrgan State Unrversrty, Cooperative Extensron Servrce. August 1990. 8 p 12328997 (1975) Variation in dry matter between and within potato tubers. Potato Res Diers BW, Cianzio SR Shoemaker RC (1992 Possible identification of guantitative trait loci affecting rron defficrency in soybean. J lant Nutrrtron 15:2127-21 6 Edwards_MD, Stuber CW, Wendel JF (1987) Molecular-marker-faeilitated investigations of quantrtative-trart locr rn marze. 1. Numbers, genomic distributron and types of gene action. Genetics 116:1 13-125 Gould WA (1989) S ific ravit - its measurement and use. In: Chi in tato handbook. The Snackplgcood Agssoc.ypp 18-29 pp g p0 Hackett CA, Ellis RP, Forster BP, McNicol JW, Macaulay M (1992) Statistical analysis of a linkage experiment 1n barley involving quantrtative trart loci for height and ear- erznéergence trme and two genetrc markers on chromosome 4. Theor Appl Genet 85 :120- Hawkes _JG (1990) The Potato. Evolution, biodiversity and genetic resources. Smrthsonran Instrtutron Press. Washington, DC. 259 p Hayes PM, Blake T, Chen THH, Chen F, Pan A, Liu B (1992) Quantitative trait loci on bar ey (Hordeum vulgare L.) chromosome 7 assocrated wrth components of winterhardiness. Genome 36:66-71 HaynesKG, Haynes FL (19831) Stabilitgoof high specific gravity genotypes of potatoes un er hr h tem ratures. Am otatoJ :17-26 . . Heun (199 Mapping quantitative powdery mildew resrstance of barley usrng a restriction fragment eng polymorphism map. Genome 35:1019-1025 Iwanaga, M (1983) Ploidy level manipulation approach: development of diploid 73 populations with specific resistance and FDR 2n eellen production. In: Present and future strate res for tato breedrng and improvement. eport of the 26“I Planning Conference, CIP. ec. 19 3, era, Peru. Johansen RtH, nh/Iiller It}, NewsomIDW, FontenotI JF (l9e7) The influence of envrronmen on e 3 11¢ ravrt , ant maturrt an ' tat ' . Potato J 44:107-122 pec g y p y vrgor o po 0 progenres Am Keim P, Diers BW, Shoemaker RC (1990) Genetic analysis of soybean hard seedeness wrth molecular markers. Theor Appl Genet 79:465-469 lam SL, Grenard R 51976) Potato selection for high dry-matter in seedling generation. Am Potato J 53:285- 91 lana EP, Johansen RH, Nelson DC (1970) Variation in specific gravity of potato tubers. Am Potato J 47:9-12 lander ES, Green P, Abtahamson J, Barlow A, Daly M], Lincoln SE, Newburg L (1987) MAPMAKER: _An rnteractrve computer naekage for constructin primary genetic lrnkage maps of expenmantal and natural popu atrons. Genomics 1:1 4-181 lander E, Botstein D (1989 Mapping Mendelian factors underlying quantitative traits usrng RFLP lrnkage maps. enetrcs l 1:185-199 Miura I-I, Parker BB, Snape JW (1992) The location of major genes and associated quantrtative trait loci on chromosome arm SBL of wheat. Theor Appl Genet 85 : 197-204 Paterson AH Lander ES, Hewitt JD, Peterson S, Lincoln ES, Tanksley SD (1988) Resolution of quantitative traits into Mendelian factors, usrng a complete lrnkage map of restriction fragment length polymorphrsms. Nature 335:721-726 Paterson AH, Damon S, Hewitt JD, Zamir D, Rabinewitch HD, Lincoln ES, lander ES, Tanksley SD (1991) Mendelian factors underlying quantrtative trarts in tomato: comparrson across species, generations and envrronments. Genetrcs 127:181-197 Peloquin SJ, Yerk GL Werner JE, Darmo E (1989) Potato breeding with haploids and 2n gametes. Genome 31:1000-1004 Ruttencutter 6, Ha nes F, Moll R £1979) Estimation of narrow-sense heritability for ificJ r6a\‘/it in53iploid potatoes ( . tuberosum subsp. phureja and stenotomum) Am otato : -4 Schippers PA (1976) The relationship between specific gravity and percentage dry matter in potato tubers. Am Potato] 53:111-122 Simmonds NW (1977) Relations between sgecific gravity, dry matter content and starch content of potatoes. Potato Res 202137-14 Soller M, Brody T (1976) .On the power of experimental designs for the detection of linkage between marker locr and quantrtative locr in crosses between inbred lines. Theor Appl Genet 47:35—39 Stevenson F] Akeley RV, McLean JG (1954 Potato utilization in relation to variety (heredity) and environment. Am Potato J 31:3 7-340 Stuber CW, Edwards MD, Wendel J F (1987) Molecular marker-facilitated investigations of quantitative trait loci in maize. 11. Factors rnfluencrng yreld and rts component trarts. Crop Sci 22:737-740 Stuber CW, Lincoln SE, Wolff DW, Helentjaris T, Lander ES (1992) Identification of 74 genetic factors contributing; to heterosis in a h brid from two elite maize inbred lines usrng molecular markers. enetrcs 132:823-839 Tanksley SD, Medina-Filho H, Rick CM (1982) Use of naturally-ocurring enzyme varratron to detect and map genes controllrng quantitative traits 1n an interspecific backcross of tomato. Heredity 49:11-25 Tanksley SD, Hewitt J (1988) Use of molecular markers in breeding for soluble solids content 1n tomato - a re-examrnatron. Theor Appl Genet 75 :81 1-823 Tanksley SD, Ganal MW, Prince JP, de Vicente MC, Bonierbale MW Broun P, Fulton TM, Grovannorn J] Grandillo S, Martin GB, Messe uer R, Miller JC Miller L Paterson AH, Pm a o, Roder MS, Win RA, Wu , Yougg ND (1992) High density molecular hnkage maps of the tomato an potato genomes. enetics 132:114 -1160 Thom son P, HaKnes F, Moll R (1980) Estimation of genetic variance components and herita ility for tu er dormancy in diploid potatoes. Am Potato 1 57:39-46 Wannamaker M], Coolins WW, Wolters P (1992) Simpletlinear relationship between dry matter, specific gravitg, and tissue specific gravrty 1n a drplord potato brwdrng population. Potato Res 5:157-160 Weller J1 Soller M, Brody T (1988) Linka e analysis of quantitative traits in an interspecific cross of tomato (Lécopersicon escu entum x Lycopersrcon prmpmelltfolrum) by means of genetic markers. enetrcs 118:329-339 Wilson JH, Lindsay AM (1969) The relation between specific gravity and dry matter content of potato tubers. Am Potato J 46:323-328 tr 1 iiiiirII‘l’lIi [iii-I I! CONCLUSIONS This research is one of the first quantitative trait loci studies using molecular markers in potato. Two diploid populations of Solanum spp. were characterized with isozyme loci, and QTL analysis was performed for two traits. The consistency of significant markers across environments and genetic backgrounds was evaluated. One of the populations was further characterized with RFLPs and subsequently RAPDs when this technology became available and was optimized for potato. This resulted in the utilization of 127 marker loci. Eighty-seven of these loci were used to construct a linkage map. Six QTLs with significant effect on tuber dormancy were identified. One of the QTLs clustered several loci and had the highest effect on the trait, and could possibly be of value for tagging this trait in future generations. For specific gravity, a total of 10 QTLs were identified over three environments. Utilizing the average data from the three environments a model with 9 loci was deve10ped, which was consistent when tested separately with data from each one of the environments. This model has potential value for marker-assisted selection for this trait. This study has thus provided the initial information on two tuber traits with polygenic inheritance. There are many directions in which future studies can be directed. The population could be further characterized with markers on chromosomes 9 and 11 to complete and saturate the linkage map and see whether additional QTLs affecting the traits can be identified. Furthermore, with tagged traits the feasibility of using marker- assisted selection in potato breeding can be tested. Also, the importance of these markers at the tetraploid level after 4x x 2x crosses could be studied. These could be used to monitor the introgression of the desirable traits from the diploid wild species, and therefore increase the efficiency of their use in potato breeding. For specific gravity, it would be possible to study the correlation between QTLs tagged for specific gravity and cloned genes involved in the starch production pathway. For tuber dormancy, the same type of QTL study could be performed in haploid S. tuberosum populations which do not have dominance effects from wild Species and therefore not have skewed frequency 75 76 distributions. In this research, the basic methodology for QTL analysis was developed and incorporated into the Potato Breeding Program at MSU and is now available for future studies with other traits and Solanum germplasm. "I11111111111110I“