ABSTRACT
MCNELLIE, JAMES PATRICK. Mapping QTLs for Fruit Quality and Horticultural Traits in Fresh Market Tomato (Solanumlycopersicum) Using SNP Markers. (Under the direction of Dr. Dilip R. Panthee.)
Quantitative trait loci (QTL) mapping has been performed extensively in tomato, but
a majority of mapping populations have been inter-specific. The use of inter-specific
populations has been a necessity due to low genetic diversity within cultivated germplasm,
which limits the availability of polymorphic molecular markers. This research used the
Solanaceae Coordinated Agricultural Project (SolCAP) SNP array to genotype an
intraspecific F2 population for QTL mapping of fruit quality and horticultural traits.
NC 10204 is an intraspecific mapping population derived from two inbred breeding
lines: NC 30P (a plum breeding line) and NC 22-L(2008) (a grape breeding line). In 2013, F2
plants were grown at the Mountain Horticultural Crops Research and Extension Center in
Mills River, NC. F2 plants were individually harvested to create F2:3 families. In 2014, F2:3
families were grown at the Mountain Horticultural Crops Research and Extension Center in
two replications of six plants (when possible) and individual lines were bulk harvested to
obtain F2:4 seed. F2:4 lines were evaluated the following summer at the Mountain Horticultural
Crops Research and Extension Center and at the Mountain Research Station at Waynesville,
NC.
Plant growth habit and inflorescence branching were phenotyped in all three
generations. The following traits were phenotyped in two generations: jointless pedicles (F2
and F3), total soluble solids (F2 and F3), fruit shape (F2 and F3), and days to 50% fruit ripe (F2
Analyzer quantitatively measures forty-four fruit quality traits related to size, shape,
uniformity and color.
Genomic DNA was extracted from F2 plants using a modified cetyltrimethyl
ammonium bromide method and used to genotype 190 lines with the SolCAP Illumina
Infinium Assay. The linkage map for NC 10204 was constructed using ‘R/qtl’, 187 plants,
and contained 415 polymorphic SNPs across 858.9cM with an average spacing of 2.13cM.
Segregation distortion favoring NC 22L-1(2008) was present on chromosome 2, and to a
lesser degree on chromosome 4.
Twenty-six QTLs were mapped for twenty-two fruit quality traits measured by
Tomato Analyzer. Size traits mapped QTLs to chromosome 2 (four QTLs between 78.10cM
and 80.10cM), and chromosome 9 (three QTLs at 55.66cM). Fruit shape traits mapped to
chromosome 3 (two QTL between 3.10cM and 6.10cM), chromosome 9 (four QTLs between
29.10cM and 41.19cM), and chromosome 12 (three QTLs between 5.10cM and 6.10cM).
Uniformity traits were mapped to chromosome 3 (one QTL at 7.10cM), chromosome 9 (three
QTLs between 27.10cM and 41.19cM), chromosome 11 (one QTL at 52.9cM) and
chromosome 12 (one QTL at 6.10cM). Color traits had three QTLs mapped to chromosome 2
between 78.1cM and 81.10cM. Visually phenotyping fruit shape mapped QTLs to roughly
the same location on chromosome 9 (56.10cM) and 12 (5.10cM) as the fruit shape traits
measured with Tomato Analyzer.
Total soluble solids mapped two QTLs to chromosome 2 (78.10cM) and chromosome
17cM away from the QTL for growth habit. Inflorescence branching mapped two QTLs to
chromosome 6 (50.43cM) and 12 (29.21cM).
This research suggests that NC 10204 is segregating for genes controlling fruit size
on chromosome 2 and 9, fruit shape on chromosome 3, 9, and 12, fruit uniformity on
chromosome 3, 9, 11 and 12, and fruit color on chromosome 2. The fruit quality traits on
chromosome 9 and 12, if validated, may have value in marker assisted selection in tomato
© Copyright 2016 James Patrick McNellie
Mapping QTLs for Fruit Quality and Horticultural Traits in Fresh Market Tomato (Solanum lycopersicum) Using SNP Markers
by
James Patrick McNellie
A thesis submitted to the Graduate Faculty of North Carolina State University
in partial fulfillment of the requirements for the degree of
Master of Science
Horticulture Science
Raleigh, North Carolina
2016
APPROVED BY:
_______________________________ _______________________________ Dr. Dilip R. Panthee Dr. Peter Balint-Kurti
Committee Chair
ii
DEDICATION
iii
BIOGRAPHY
James Patrick McNellie was born in Fairfax, Virginia and raised in nearby McLean, VA.
After high school he attended Indiana University and earned a Bachelor of Arts in Economics
and History in 2003. James returned to northern Virginia and worked as a Financial Analyst
and Government Contractor in the Washington, D.C. metropolitan area before return to
Indiana. After working for three years as a Government Contractor in Indianapolis, IN, James
returned to school and in 2013 earned a Bachelor of Science in Plant Genetics and Breeding
from Purdue University. He started his Master of Science research at North Carolina State
University the following fall. His research focused on QTL mapping fruit quality and
iv
ACKNOWLEDGMENTS
First and foremost, Dr. Dilip R. Panthee for giving me the opportunity to earn my Master of
Science in his program. Secondly my committee members, Dr. Todd Wehner and Dr. Peter
Balint-Kurti. Other noteworthy faculty members at North Carolina State University: Dr.
Ralph Dewey, Dr. Christopher Gunter, Dr. Sergei Krasnyanski, Dr. Jeanine Davis, Dr.
Thomas Ranney, and Dr. Randy Gardner.
Everyone at the Mountain Horticultural Crops Research and Extension Center, Mills River,
NC; especially Ann Piotrowski, Jeremy Smith, Ragy Ibrahem, Nathan Lynch, Dr.
v
TABLE OF CONTENTS
TABLE OF CONTENTS ... V LIST OF TABLES ... VII LIST OF FIGURES ... IX
CHAPTER 1 FRUIT QUALITY AND HORTICULTURAL TRAITS IN TOMATO .... 1
INTRODUCTION ... 1
FRUITQUALITYTRAITS ... 3
HORTICULTURALTRAITS ... 4
RESEARCHOBJECTIVES ... 6
REFERENCES ... 8
CHAPTER 2 QTL ANALYSIS FOR FRUIT TRAITS PHENOTYPED USING TOMATO ANALYZER IN THE INTRA-SPECIFIC TOMATO POPULATION NC 10204... 16
INTRODUCTION ... 16
MATERIALSANDMETHODS ... 18
Population Development ... 18
DNA Extraction and Genotyping ... 19
Phenotyping ... 19
Data Analysis and QTL Mapping ... 20
RESULTSANDDISCUSSION ... 23
Segregation Distortion ... 25
QTLs on Chromosome 2 Associated with Multiple Fruit Quality Traits ... 27
QTLs on Chromosome 9 Associated with Multiple Fruit Quality Traits ... 27
QTLs on Chromosome 12 Associated with Multiple Fruit Quality Traits ... 28
QTLs on Chromosome 3 & 11 Associated with Multiple Fruit Quality Traits ... 29
Total soluble solids ... 30
Fruit Shape... 31
Comparison of Phenotypic Data Obtain using Tomato Analyzer and Phenotyping Fruit Shape Visually ... 31
CONCLUSION ... 32
REFERENCES ... 33
CHAPTER 3 QTL ANALYSIS FOR TOMATO HORTICULTURAL TRAITS IN THE INTRA-SPECIFIC POPULATION NC 10204 ... 70
INTRODUCTION ... 70
MATERIALSANDMETHODS ... 73
vi
DNA Extraction and Genotyping ... 73
Phenotyping ... 74
Data Analysis and QTL Mapping ... 74
RESULTSANDDISCUSSION ... 78
Growth Habit ... 78
Jointless Pedicel... 78
Inflorescence branching... 79
Days to 50% Fruit Ripe ... 81
CONCLUSION ... 81
vii
LIST OF TABLES
TABLE 2.0.1NUMBER OF LINES PHENOTYPED IN THE INTRA-SPECIFIC TOMATO POPULATION NC 10204 FOR FRUIT QUALITY TRAITS BY GENERATION AND NUMBER OF LINES WITH SNP MARKER DATA ... 38 TABLE 2.0.2.DESCRIPTION OF TOMATO ANALYZER TRAITS WITH SIGNIFICANT QTLS REPORTED IN THE INTRA-SPECIFIC TOMATO POPULATION NC10204(RODRIGUEZ ET AL.2010) ... 39 TABLE 2.0.3.SUMMARY OF NC10204 LINKAGE MAP BY CHROMOSOME.FOR EACH
CHROMOSOME THE NUMBER OF POLYMORPHIC SNP MARKERS, LENGTH, AVERAGE
MARKER SPACING AND MAXIMUM SPACING IS SHOWN ... 41 TABLE 2.0.4.SOLCAP MARKERS IN THE NC10204 LINKAGE MAP WITH CHROMOSOME NUMBER
AND POSITION IN CENTIMORGAN.MARKER ORDER WAS DETERMINED USING THE PHYSICAL MAP (SL2.40) AND THE DISTANCE BETWEEN MARKERS WAS DETERMINED USING THE KOSAMBI MAPPING FUNCTION IN ‘R/QTL’(KOSAMBI 1943)... 42 TABLE 2.0.5.PEARSON’S CORRELATION COEFFICIENT AMONG FRUIT QUALITY TRAITS IN THE
INTRA-SPECIFIC TOMATO POPULATION NC10204 ... 53 TABLE 2.0.6.CORRELATION BETWEEN REPLICATIONS FOR FRUIT QUALITY TRAITS IN THE INTRA
-SPECIFIC TOMATO POPULATION NC10204 MEASURED WITH TOMATO ANALYZER ... 55 TABLE 2.0.7.ANALYSIS OF VARIANCE FOR TOTAL SOLUBLE SOLIDS AND FRUIT SHAPE USING F2
AND F2:3PHENOTYPIC DATA.MEAN SQUARED IS DERIVED FROM TYPE III SUM OF SQUARES USING PROC GLM IN SAS VERSION 9.4 FOR WINDOWS (SASINSTITUTE INC,CARY,NC) ... 56 TABLE 2.0.8.HERITABILITY ESTIMATES FOR FRUIT QUALITY TRAITS IN THE INTRA-SPECIFIC
TOMATO POPULATION NC10204.THE NARROW-SENSE HERITABILITY IS SHOWN FOR TOTAL SOLUBLE SOLIDS, FRUIT SHAPE AND GROWTH HABIT.THE BOARD-SENSE
HERITABILITY IS SHOWN FOR THE TRAITS MEASURED WITH TOMATO ANALYZER ... 57 TABLE 2.0.9.SUMMARY OF QTL ANALYSIS OF FRUIT QUALITY TRAITS IN THE INTRA-SPECIFIC
TOMATO POPULATION NC10204 USING HALEY-KNOTT REGRESSION.THE QTLS FOR THE
TRAITS MEASURED WITH TOMATO ANALYZER WERE MAPPED USING PHENOTYPIC DATA
FROM THE F3 GENERATION.THE QTLS FOR TOTAL SOLUBLE SOLIDS AND FRUIT SHAPE WERE MAPPED USING THE LEAST SQUARED MEANS OF THE F2 AND F3 GENERATIONS ... 58 TABLE 2.0.10.COMPARISON OF THREE NC10204 LINKAGE MAPS DIFFERING FOR SNP
MARKERS WITH SEGREGATION DISTORTION.MAP 1 IS THE LINKAGE MAP USED FOR QTL MAPPING IN THE INTRA-SPECIFIC TOMATO POPULATION NC10204.MAP 2 HAD SNP MARKERS REMOVED THAT DEVIATED SIGNIFICANTLY (Α =0.05) FROM EXPECTED
MENDELIAN SEGREGATION RATIOS AFTER BONFERRONI CORRECTION FOR MULTIPLE
TESTS.MAP 3 HAD MARKERS REMOVED THAT SIGNIFICANTLY DEVIATED (Α =0.05) FROM
MENDELIAN SEGREGATION RATIOS WITHOUT THE BONFERRONI CORRECTION FOR
viii
TABLE 2.0.11.COMPARISON OF LOD THRESHOLD, MAXIMUM LOD SCORE AND SIGNIFICANCE
OF TRAITS MEASURED WITH TOMATO ANALYZER USING THE THREE LINKAGE MAPS
DESCRIBED IN TABLE 2.10 ... 62 TABLE 2.0.12.COMPARISON OF QTLS MAPPED IN THE INTRA-SPECIFIC TOMATO POPULATION
NC10204 USING PHENOTYPIC DATA OBTAINED USING TOMATO ANALYZER AND QTLS
MAPPED IN FOUR POPULATIONS PREVIOUSLY PHENOTYPED WITH TOMATO ANALYZER.
ONLY CHROMOSOME NUMBER IS COMPARED.THE POPULATION HOWARD GERMAN X
LA1589 WAS PHENOTYPED IN TWO GENERATIONS (F2 AND BC1) ... 64 TABLE 3.0.1.NUMBER OF LINES IN EACH GENERATION PHENOTYPED FOR HORTICULTURAL
TRAITS IN THE INTRA-SPECIFIC TOMATO POPULATION NC10204 ... 87 TABLE 3.0.2.PEARSON’S CORRELATION COEFFICIENT BETWEEN HORTICULTURAL TRAITS IN THE INTRA-SPECIFIC TOMATO POPULATION NC10204 ... 88 TABLE 3.0.3.NARROW SENSE HERITABILITY FOR HORTICULTURAL TRAITS IN THE INTRA
-SPECIFIC TOMATO POPULATION NC10204.THE HERITABILITY FOR GROWTH HABIT,
INFLORESCENCE BRANCHING AND DAYS TO 50% RIPE WAS CALCULATED USING PARENT –
OFFSPRING REGRESSION.THE HERITABILITY FOR JOINTLESS PEDICELS WAS CALCULATED ON A LOGISTIC SCALE ... 89 TABLE 3.0.4.ANALYSIS OF VARIANCE FOR HORTICULTURAL TRAITS IN THE INTRA-SPECIFIC
TOMATO POPULATION NC10204.GROWTH HABIT AND INFLORESCENCE BRANCHING USED
PHENOTYPIC DATA FROM THE F2,F3, AND F4GENERATIONS.JOINTLESS PEDICELS USED PHENOTYPIC DATA FROM THE F2 AND F3 GENERATIONS, AND DAYS TO 50% FRUIT RIPE USED PHENOTYPIC DATA FROM THE F2 AND F4 GENERATIONS.MEAN SQUARED IS DERIVED FROM TYPE III SUM OF SQUARES USING PROC GLM IN SAS VERSION 9.4 FOR WINDOWS (SASINSTITUTE INC,CARY,NC) ... 90 TABLE 3.5.SUMMARY OF QTL ANALYSIS FOR HORTICULTURAL TRAITS IN THE INTRA-SPECIFIC
ix
LIST OF FIGURES
FIGURE 1.1.EXAMPLE OF TOMATO ANALYZER PHENOTYPING SOFTWARE ... 13 FIGURE 1.2.EXAMPLE OF A JOINTED TOMATO PEDICEL ... 14 FIGURE 1.3.EXAMPLE OF A JOINTLESS TOMATO PEDICEL ... 15 FIGURE 2.0.1.THE LINKAGE MAP FOR THE INTRA-SPECIFIC TOMATO POPULATION NC10204
SHOWING SNP MARKER DENSITY ON INDIVIDUAL CHROMOSOMES.CREATED USING THE
KOSAMBI MAPPING FUNCTION AND MARKER ORDER WAS DETERMINED USING PHYSICAL
MAP POSITION (SL2.40)(KOSAMBI 1943) ... 66 FIGURE 2.0.2.SEGREGATION DISTORTION ON CHROMOSOMES 2 AND 4 IN THE INTRA-SPECIFIC
TOMATO POPULATION NC10204.THE TOP PANEL SHOWS THE NEGATIVE LOG OF P-VALUES FROM CHI-SQUARE TESTS FOR MENDELIAN SEGREGATION.THE BOTTOM PANEL
SHOWS ALLELE FREQUENCIES.THE BLUE LINE REPRESENTS THE FREQUENCY OF
HETEROZYGOUS SNPS (AB), THE BLACK LINE HOMOZYGOUS SNPS FOR NC22L-1(2008) (AA) AND THE RED LINE HOMOZYGOUS SNPS FOR NC30P(BB).THE DASHED
HORIZONTAL LINES INDICATE THE PREDICTED MENDELIAN SEGREGATION FREQUENCIES OF
50% FOR HETEROZYGOTES AND 25% HOMOZYGOTES. ... 67 FIGURE 2.0.3.THE GUIDE USED TO PHENOTYPE PREDOMINANT FRUIT SHAPE IN THE INTRA
-SPECIFIC TOMATO POPULATION NC10204 IN THE F2 AND F2:3 GENERATIONS ... 68 FIGURE 2.0.4.EXAMPLE OF FRUIT SHAPE SEGREGATION IN THE INTRA-SPECIFIC TOMATO
POPULATION NC10204.ALL FRUIT ARE FROM A SINGLE F2:3FAMILY.THE FRUIT WITHIN THE YELLOW BOX IS PEAR SHAPED (OBOVOID) ... 69
FIGURE 3.0.1.AVERAGE MONTHLY PRODUCER PRICE INDEX FOR FRESH MARKET TOMATOES,
1947–2010. ... 92 FIGURE 3.0.2.PLOT OF SNP MARKER EFFECT ON GROWTH TYPE FOR THE SNP CLOSEST TO THE
QTL FOR GROWTH TYPE ON CHROMOSOME 1.THE TOP PANEL (A) IS THE F2 GENERATION, MIDDLE PANEL (B) THE F3 GENERATION AND THE BOTTOM PANEL (C) THE F4 GENERATION. THE GENOTYPE EFFECT APPEARS TO BE RANDOM IN A AND C, BUT NOT IN B.THIS IS LIKELY DUE TO THE LOWER NUMBER OF PLANTS PHENOTYPED IN THE F3 GENERATION (TABLE 3.1). ... 93 FIGURE 3.0.3.INTERACTION BETWEEN SNPS ASSOCIATED WITH JOINTLESS PEDICELS AND
GROWTH TYPE ON INFLORESCENCE BRANCHING USING DATA FROM ALL GENERATIONS.
THE X AXIS SHOWS THE SNP GENOTYPES FOR GROWTH TYPE WHERE A/- IS INDETERMINATE AND BB DETERMINATE.THE THREE LINES REPRESENT THE SNP
x
FIGURE 3.0.4.EXAMPLE OF INFLORESCENCE REVERTING TO VEGETATIVE GROWTH (WHITE ARROW) IN AN INDETERMINATE, JOINTLESS PEDICEL CULTIVAR ... 97 FIGURE 3.5.INFLORESCENCE BRANCHING IN THE INTRA-SPECIFIC TOMATO POPULATION NC
10204.THE INFLORESCENCES LABELED P1 AND P2 ARE REPRESENTATIVE INFLORESCENCE FROM THE TWO INBRED PARENTS USED TO CREATED NC10204, WITH P1 CORRESPONDING TO NC30P,P2 TO NC22L-1(2008) AND F1 TO THE NC10204 HYBRID.THE THREE INFLORESCENCES TO THE RIGHT OF F1 ARE REPRESENTATIVE OF THE MULTIPAROUS
1
CHAPTER 1 FRUIT QUALITY AND HORTICULTURAL TRAITS IN TOMATO
INTRODUCTION
The tomato, Solanum lycopersicum, is considered the model organism for fleshy fruit development and is an economically important crop with the 2014 United States of America
fresh market tomato harvest valued at $1.3 billion (USDA National Agricultural Statistics
Service 2015). A small diploid genome and autogamous reproductive system makes tomato
ideal for genetic studies. Quantitative trait loci (QTL) mapping has been performed
extensively in tomato. To give an idea of just how extensively, ‘Genetics, Genomics and
Breeding of Tomato’ lists 102 mapping studies reporting 5,675 QTLs (Grandillo, Termolino,
van der Knaap 2013). Unfortunately, the words of Rex Bernardo from his 2008 review of
QTLs is true that “the vast majority of the favorable alleles at these identified QTL reside in
journals on library shelves rather than in cultivars that have been improved through the
introgression or selection of these favorable QTL alleles” (Bernardo 2008). For tomato, this
can partly be attributed to the use of inter-specific mapping populations.
Inter-specific populations are derived from crossing two different species, S. lycopersicum
crossed to a wild relative. An intra-specific population is a cross between organisms of the
same species, S. lycopersicum crossed with another S. lycopersicum or S. lycopersicum var.
cerasiforme (cherry or grape tomatoes). A majority of S. lycopersicum linkage maps are inter-specific and are “of low-to-moderate density, having an average inter-marker spacing of
around 5cM and each includes between 70 and 400 markers” (Foolad 2007). During
2
(Bai and Lindhout 2007). Inter-specific mapping populations were a necessity because the
low genetic diversity within cultivated germplasm hindered the detection of polymorphic
molecular markers (Foolad, Jones, Rodriguez 1993; García-Martínez et al. 2006; Joshi,
Gardner, Panthee 2012; Miller and Tanksley 1990; Park, West, St. Clair 2004; van Berloo et
al. 2008; Williams and Clair 1993). Twelve years after the first molecular linkage map by
Paterson et al. (1988), the first dense intra-specific linkage map was created in a cross
between a cultivated cherry line (Cervil) and a large fruited variety of (Levovil)
(Saliba-Colombani et al. 2000). The linkage map created by Saliba-(Saliba-Colombani et al. (2000)
contained 376 molecular markers, an improvement over two previous attempts at creating
whole genome intra-specific linkage maps which had at most 67 markers (Danesh et al.
1994; Lindhout et al. 1994). Problems obtaining sufficient polymorphic markers in
intra-specific populations persisted into the 2000s. Of the two intra-intra-specific populations in the
literature created after Saliba-Colombani et al. (2000), one had 37 polymorphic markers
(Yang et al. 2005) and the other 44 polymorphic markers (Wang et al. 2011).
The Solanaceae Coordinated Agricultural Project (SolCAP) 7,720 SNP array is the first
publically available tomato SNP array and was created by sequencing six tomato accessions
generating 62,576 non-redundant SNPs, of which 7,720 became the SolCAP array (Hamilton
et al. 2012; Sim et al. 2012a). Rodriguez et al. (2013) used the SolCAP array to genotype
three intra-specific bi-parental populations and found an average of 625 polymorphic
bi-3
parental QTL mapping (Rodríguez, Kim, van der Knaap 2013). Now that molecular markers
can readily be obtained in intra-specific populations, existing genetic variation within
breeding germplasm can be utilized in mapping experiments.
FRUIT QUALITY TRAITS
The most important traits for fresh market tomatoes are fruit quality traits. Tomato taste and
other organoleptic traits like scent, are complex quantitatively controlled traits. The
intra-specific population developed by Saliba-Colombani et al. (2000) has been used to map at
least 130 organoleptic related QTLs (Causse et al. 2001; Causse et al. 2002;
Saliba-Colombani et al. 2001). Total soluble solids is an estimation of how much total sugar a
tomato contains without measuring the exact amount of glucose, fructose or sucrose
(Grandillo, Termolino, van der Knaap 2013). Increasing total soluble solids in large fruit
varieties is difficult because it is negatively correlated with fruit size (Panthee et al. 2013b).
By 2007 there were 95 QTLs reported for total soluble solids, fructose, glucose and sucrose
(Grandillo, Termolino, van der Knaap 2013).
Fruit color is a polygenically-controlled trait; however, most discussions on the subject focus
on recessive mutations in the carotenoid pathway. For example, plants containing the ogc
gene have a mutation in a -cyclase protein preventing the conversion of lycopene to
-carotene, leading to an accumulation of lycopene which gives the fruit a deep red color
(Ronen et al. 2000). Fruit color can be measured using a spectrophotometer (Ashrafi et al.
2012), colorimeter (Sacks and Francis 2001) or a flatbed scanner (Darrigues et al. 2008).
The fruit phenotyping program Tomato Analyzer was released in 2006 and, at the time, could
4
phenotypes the scanned image of sliced tomato fruit (Figure 1.1). Subsequent versions of
Tomato Analyzer added the ability to phenotype color and additional morphological traits
(Gonzalo et al. 2009). Brewer et al. 2007 used Tomato Analyzer to map 95 QTLs for 15 fruit
traits in two inter-specific populations. A similar study by Gonzalo et al. 2009 mapped an
additional 63 QTLs and identified chromosomal regions that contain genes influencing fruit
shape. Tomato Analyzer has also been used to fine map genes controlling fruit shape
(Rodríguez, Kim, van der Knaap 2013; Sun et al. 2015), estimate fruit lycopene content
based upon color parameters (Panthee et al. 2013a) and to calculate the inheritance of fruit
color (Panthee et al. 2015).
Bauchet et al. (2014) used both the SolCAP array and Tomato Analyzer in a genome wide
association study for fruit size; but only one trait was phenotyped using Tomato Analyzer.
Rodriguez et al. (2013) used Tomato Analyzer to phenotype three intra-specific populations
for three fruit traits, followed by analysis of variance to identify SolCAP markers associated
with fruit traits. Phenotypic data obtained with Tomato Analyzer was used to map loci
influencing fruit color in a bi-parental inter-specific population (Darrigues 2007). To the
authors knowledge, Tomato Analyzer has not been used to map color traits in an
intra-specific bi-parental population.
HORTICULTURAL TRAITS
In tomato, polygenic horticultural traits have not been studied in as much depth as disease
resistance and fruit traits, even though their importance to growers is great. The production of
fresh market tomatoes is labor intensive; requiring hand pruning, stringing and multiple
5
were estimated at roughly $7,000 per hectare, with harvest labor accounting for about half
the cost (Sydorovych et al. 2008). Tomato breeders can help growers by developing cultivars
with reduced production costs and increased marketable yield.
There are two widely deployed genes that tomato breeders use to improve horticultural traits:
the self-pruning and jointless-2 genes. When the self-pruning gene is homozygous recessive, plants have a determinant growth habit (Carmel-Goren et al. 2003). Determinant growth is a
requirement for processing tomatoes and is desirable for fresh market tomatoes as
determinant plants have a shorter harvest window. When the jointless-2 gene is homozygous recessive, inflorescences do not form an abscission zone on the pedicel (Figure 1.2 - Figure
1.3). The lack of an abscission zone makes it easier to separate fruit from stem and is a
required trait for mechanically harvesting processing tomatoes. Fresh market varieties that
are jointed (as most are), have stems that have to be manually removed from fruit. Picking
times can be reduced by 17 to 33% by using jointless varieties (Zahara and Scheuerman
1988). The acceptance of jointless fresh market hybrids has been slow due to undesirable
pleotropic effect on taste (Scott et al. 2013; Scott, Bryan, Ramos 1997).
Much of the research on horticultural traits has focused on finding genes with an
over-dominant effect. For example, one paper by Semel et al. 2006 reported 841 overover-dominant
QTLs for 35 traits. Perhaps the most significant horticultural trait discovery since the self-pruning gene has been the over-dominant effect of the single flower truss gene. When
heterozygous and in a determinant genetic background, single flower truss alters the hormone florigen leading to significant increase in yield (Krieger, Lippman, Zamir 2010). The
6
Determinant plants have a gradual decrease in the number of leafs between inflorescences
before the apical meristem terminates in two inflorescences. The single flower truss gene interacts with the self-pruning gene to delay the termination of the apical meristem (Krieger, Lippman, Zamir 2010; Pnueli et al. 1998). The growth habit of determinant, single flower truss heterozygotes is called semi-indeterminate. The single flower truss increases yield by increasing the number of inflorescences, roughly 20% more per plant with an average of 1
extra flower per inflorescence (Krieger, Lippman, Zamir 2010).
The single flower truss and the jointless-2 genes also effect the differentiation of the inflorescence meristem, though in different ways. When single flower truss is homozygous recessive, many inflorescences have a single flower. When jointless and in an indeterminate
genetic background, many inflorescences revert to vegetative growth (Emergy and Munger
1970; Rick and Sawant 1955; Szymkowiak and Irish 1999). Double mutant studies have
hypothesized that jointless-2 and single flower truss are involved in a regulatory pathway for inflorescence meristem, but the pathway is far from resolved (Périlleux, Lobet, Tocquin
2014; Thouet et al. 2012). QTL mapping studies have not been conducted to identify
additional loci involved in regulating the inflorescence meristem in addition to the jointless-2
and self-pruning genes.
RESEARCH OBJECTIVES
The major objectives of this research were:
1. Perform QTL mapping in the intra-specific tomato population NC 10204 for fruit
size, shape, uniformity and color using traits using phenotypic data obtained with
7
8
REFERENCES
Ashrafi H, Kinkade MP, Merk HL, Foolad MR. 2012. Identification of novel quantitative trait loci for increased lycopene content and other fruit quality traits in a tomato recombinant inbred line population. Molecular Breeding 30(1):549-67.
Bai Y and Lindhout P. 2007. Domestication and breeding of tomatoes: What have we gained and what can we gain in the future? Annals of Botany 100(5):1085-94.
Bauchet G, Munos S, Sauvage C, Bonnet J, Grivet L, Causse M. 2014. Genes involved in floral meristem in tomato exhibit drastically reduced genetic diversity and signature of selection. BMC Plant Biology 14:279,014-0279-2.
Bernardo R. 2008. Molecular markers and selection for complex traits in plants: Learning from the last 20 years. Crop Science 48(5):1649-64.
Blanca J, Montero-Pau J, Sauvage C, Bauchet G, Illa E, Díez MJ, Francis D, Causse M, van der Knaap E, Cañizares J. 2015. Genomic variation in tomato, from wild ancestors to contemporary breeding accessions. BMC Genomics 16(1):257.
Brewer MT, Moyseenko JB, Monforte AJ, van der Knaap E. 2007. Morphological variation in tomato: A comprehensive study of quantitative trait loci controlling fruit shape and development. J Exp Bot 58(6):1339-49.
Brewer MT, Lang L, Fujimura K, Dujmovic N, Gray S, van der Knaap E. 2006.
Development of a controlled vocabulary and software application to analyze fruit shape variation in tomato and other plant species. Plant Physiology 141(1):15-25.
Carmel-Goren L, Liu YS, Lifschitz E, Zamir D. 2003. The SELF-PRUNING gene family in tomato. Plant Molecular Biology 52(6):1215-22.
Causse M, Saliba-Colombani V, Lesschaeve I, Buret M. 2001. Genetic analysis of
organoleptic quality in fresh market tomato. 2. mapping QTLs for sensory attributes. Theoretical and Applied Genetics 102(2-3):273-83.
Causse M, Saliba-Colombani V, Lecomte L, Duffe P, Rousselle P, Buret M. 2002. QTL analysis of fruit quality in fresh market tomato: A few chromosome regions control the variation of sensory and instrumental traits. Journal of Experimental Botany
53(377):2089-98.
9
Darrigues A. 2007. Dissecting variation in tomato fruit color quality through digital phenotyping and genetic mapping. PhD Dissertation. The Ohio State University.
Darrigues A, Hall J, van der Knaap E, Francis DM, Dujmovic N, Gray S. 2008. Tomato analyzer-color test: A new tool for efficient digital phenotyping. Journal of the American Society for Horticultural Science 133(4):579-86.
Emergy GC and Munger HM. 1970. Alteration of growth and flowering in tomatoes by the jointless genotype. Journal of Heredity 61:51-3.
Foolad MR. 2007. Genome mapping and molecular breeding of tomato. International Journal of Plant Genomics 2007.
Foolad MR, Jones RA, Rodriguez RL. 1993. RAPD markers for constructing intraspecific tomato genetic maps. Plant Cell Reports 12(5):293-7.
García-Martínez S, Andreani L, Garcia-Gusano M, Geuna F, Ruiz JJ. 2006. Evaluation of amplified fragment length polymorphism and simple sequence repeats for tomato germplasm fingerprinting: Utility for grouping closely related traditional cultivars. Genome 49(6):648-56.
Gonzalo MJ, Brewer MT, Anderson C, Sullivan D, Gray S, van der Knaap E. 2009. Tomato fruit shape analysis using morphometric and morphology attributes implemented in tomato analyzer software program. Journal of the American Society for Horticultural Science 134(1):77-87.
Grandillo S, Termolino P, van der Knaap E. 2013. Molecular mapping of complex traits in tomato. In: Genetics, genomics and breeding of tomato. B. Liedl, J. Labate, J. Stommel, A. Slade & C. Kole, editor. 1st ed. Boca Raton, FL: CRC Press. 150 p.
Hamilton JP, Sim S, Stoffel K, Van Deynze A, Buell CR, Francis DM. 2012. Single nucleotide polymorphism discovery in cultivated tomato via sequencing by synthesis. The Plant Genome 5(1):17-29.
Joshi BK, Gardner RG, Panthee DR. 2012. Diversity analysis of tomato cultivars based on coefficient of parentage and RAPD molecular markers. Journal of Crop Improvement 26(2):177-96.
10
Lindhout P, Van Heusden S, Pet G, Van Ooijen JW, Sandbrink H, Verkerk R, Vrielink R, Zabel P. 1994. Perspectives of molecular marker assisted breeding for earliness in tomato. Euphytica 79(3):279-86.
Miller J and Tanksley S. 1990. RFLP analysis of phylogenetic relationships and genetic variation in the genus lycopersicon. Theoretical and Applied Genetics 80(4):437-48.
Panthee DR, Perkins-Veazie P, Anderson C, Ibrahem R. 2015. Diallel analysis for lycopene content in the hybrids derived from different colored parents in tomato. American Journal of Plant Sciences 6(09):1483.
Panthee DR, Perkins-Veazie P, Randall D, Brown AF. 2013a. Lycopene estimation in tomato lines using infrared absorbance and tomato analyzer. International Journal of Vegetable Science 19(3):240-55.
Panthee DR, Labate JA, McGrath MT, Breksa III AP, Robertson LD. 2013b. Genotype and environmental interaction for fruit quality traits in vintage tomato varieties. Euphytica 193(2):169-82.
Park YH, West MA, St. Clair DA. 2004. Evaluation of AFLPs for germplasm fingerprinting and assessment of genetic diversity in cultivars of tomato (lycopersicon esculentum L.). Genome 47(3):510-8.
Périlleux C, Lobet G, Tocquin P. 2014. Inflorescence development in tomato: Gene functions within a zigzag model. Frontiers in Plant Science 5,121.
Pnueli L, Carmel-Goren L, Hareven D, Gutfinger T, Alvarez J, Ganal M, Zamir D, Lifschitz E. 1998. The SELF-PRUNING gene of tomato regulates vegetative to reproductive switching of sympodial meristems and is the ortholog of CEN and TFL1. Development 125(11):1979-89.
Rick C and Sawant A. 1955. Factor interactions affecting the phenotypic expression of the jointless character in tomatoes. Journal of the American Society for Horticultural Science 66:354-60.
Rodríguez G, Kim H, van der Knaap E. 2013. Mapping of two suppressors of OVATE (sov) loci in tomato. Heredity 111(3):256-264.
11
Ruggieri V, Francese G, Sacco A, D'Alessandro A, Rigano MM, Parisi M, Milone M, Cardi T, Mennella G, Barone A. 2014. An association mapping approach to identify
favourable alleles for tomato fruit quality breeding. BMC Plant Biology 14:337,014-0337-9.
Sacco A, Ruggieri V, Parisi M, Festa G, Rigano MM, Picarella ME, Mazzucato A, Barone A. 2015. Exploring a tomato landraces collection for fruit-related traits by the aid of a high-throughput genomic platform. PloS One 10(9):e0137139.
Sacks EJ and Francis DM. 2001. Genetic and environmental variation for tomato flesh color in a population of modern breeding lines. Journal of the American Society for
Horticultural Science 126(2):221-6.
Saliba-Colombani V, Philouze J, Gervais L, Causse M. 2000. Efficiency of RFLP, RAPD, and AFLP markers for the construction of an intraspecific map of the tomato genome. Genome 43(1):29-40.
Saliba-Colombani V, Causse M, Langlois D, Philouze J, Buret M. 2001. Genetic analysis of organoleptic quality in fresh market tomato. 1. mapping QTLs for physical and chemical traits. Theoretical and Applied Genetics 102(2-3):259-72.
Scott JW, Myers JR, Boches PS, Nichols CG, Angell FF. 2013. Classical genetics and traditional breeding. In: Genetics, genomics and breeding of tomato. Liedl BE, Labate JA, Stommel JR, et al, editors. 1st ed. New York: Taylor & Francis Group. 37 p.
Scott JW, Bryan HH and Ramos LJ. 1997. High temperature fruit setting ability of large-fruited, jointless pedicel tomato hybrids with various combinations of heat-tolerance. Proceedings-Florida State Horticultural Society. 281 p.
Semel Y, Nissenbaum J, Menda N, Zinder M, Krieger U, Issman N, Pleban T, Lippman Z, Gur A, Zamir D. 2006. Overdominant quantitative trait loci for yield and fitness in tomato. Proceedings of the National Academy of Sciences of the United States of America 103(35):12981-6.
Sim S, Durstewitz G, Plieske J, Wieseke R, Ganal MW, Van Deynze A, Hamilton JP, Buell CR, Causse M, Wijeratne S. 2012a. Development of a large SNP genotyping array and generation of high-density genetic maps in tomato. PLoS One 7(7):e40563.
12
Sun L, Rodriguez GR, Clevenger JP, Illa-Berenguer E, Lin J, Blakeslee JJ, Liu W, Fei Z, Wijeratne A, Meulia T. 2015. Candidate gene selection and detailed morphological evaluations of fs8. 1, a quantitative trait locus controlling tomato fruit shape. Journal of Experimental Botany, 361.
Sydorovych O, Safley CD, Welker RM, Ferguson LM, Monks DW, Jennings K, Driver J, Louws FJ. 2008. Economic evaluation of methyl bromide alternatives for the production of tomatoes in north carolina. HortTechnology 18(4):705-13.
Szymkowiak EJ and Irish EE. 1999. Interactions between jointless and wild-type tomato tissues during development of the pedicel abscission zone and the inflorescence meristem. Plant Cell 11(2):159-75.
Thouet J, Quinet M, Lutts S, Kinet JM, Perilleux C. 2012. Repression of floral meristem fate is crucial in shaping tomato inflorescence. PLoS One 7(2):e31096.
Quick Stats: Fresh Market Tomato Production [Internet]; c2015 [cited 2015 10/15]. Available from: http://quickstats.nass.usda.gov/results/B7893057-F9C2-3244-8836-EBB3A586332D?pivot=short_desc .
van Berloo R, Zhu A, Ursem R, Verbakel H, Gort G, van Eeuwijk FA. 2008. Diversity and linkage disequilibrium analysis within a selected set of cultivated tomatoes. Theoretical and Applied Genetics 117(1):89-101.
Wang H, Hutton SF, Robbins MD, Sim S, Scott JW, Yang W, Jones JB, Francis DM. 2011. Molecular mapping of hypersensitive resistance from tomato 'hawaii 7981' to
xanthomonas perforans race T3. Phytopathology 101(10):1217-23.
Williams CE and Clair DAS. 1993. Phenetic relationships and levels of variability detected by restriction fragment length polymorphism and random amplified polymorphic DNA analysis of cultivated and wild accessions of lycopersicon esculentum. Genome
36(3):619-30.
Yang W, Sacks EJ, Lewis Ivey ML, Miller SA, Francis DM. 2005. Resistance in lycopersicon esculentum intraspecific crosses to race T1 strains of xanthomonas campestris pv. vesicatoria causing bacterial spot of tomato. Phytopathology 95(5):519-27.
16
CHAPTER 2 QTL ANALYSIS FOR FRUIT TRAITS PHENOTYPED USING TOMATO ANALYZER IN THE INTRA-SPECIFIC TOMATO POPULATION
NC 10204
INTRODUCTION
Plant breeders working on fresh market tomato put significant effort into the development of
cultivars with high yield, earliness, and disease resistance needed by growers. Consumers
demand that tomato fruit be large, red color, firm texture, sweet taste, and have no visual
defects (Oltman, Jervis, Drake 2014; Piombino et al. 2013; Stommel et al. 2005). Fruit shape
and size is largely controlled by five genes: ovate, sun, lc, fs8.1,fw2.2 and fas (Grandillo, Termolino, van der Knaap 2013; Tanksley 2004). The improvement of fruit quality is a
priority for breeders. Our research made use of the Solanaceae Coordinated Agricultural
Project (SolCAP) 7720 SNP array and the precision phenotyping software Tomato Analyzer
to quantify and map quantitative trait loci (QTLs) for fruit quality traits in an intra-specific
bi-parental F2 derived population.
Inter-specific mapping populations have been used for the majority of QTL mapping studies
in tomato due to the lack of polymorphic markers in elite populations (Joshi, Gardner,
Panthee 2012; Miller and Tanksley 1990; Park, West, St. Clair 2004). The confounding
effect of different genetic backgrounds and deleterious linkage drag have made the
introgression of potentially beneficial alleles mapped in inter-specific populations difficult
(Lecomte et al. 2004; Robert et al. 2001). The first dense intra-specific linkage map was
created in 2000 using 376 markers (Saliba-Colombani et al. 2000). That linkage map was the
17
populations remained problematic until the release of the SolCAP array (Rodríguez, Kim,
van der Knaap 2013; Wang et al. 2011; Yang et al. 2005). The current research used to the
SolCAP 7,720 SNP array for genotyping to solve the problem of obtaining polymorphic
markers in intra-specific populations.
Mapping of QTLs hinges upon the identification of significant correlations between variation
at the phenotypic and genotypic levels. The SolCAP array increases the quantity of genetic
data, and Tomato Analyzer improves the quality of the measurement of the phenotypic data.
Tomato Analyzer was released in 2006 and with subsequent upgrades can now quantitatively
measure forty-seven fruit shape and color traits (Darrigues et al. 2008; Rodriguez et al.
2010). Data derived from the SolCAP array and Tomato Analyzer were used to map
inhibitors of the ovate gene (Rodríguez, Kim, van der Knaap 2013) and, in a separate study, to map fruit size QTL in a genome wide association study (Bauchet et al. 2014). Tomato
Analyzer has also been used to estimate the amount of lycopene in fruit, which largely
influences how red fruit are (Panthee et al. 2013; Panthee et al. 2015). The improvement of
fruit color is important for consumer preference and health benefits (Oltman, Jervis, Drake
2014; Perveen et al. 2015). A better understanding of the loci influencing fruit color and
nutritional value will help increase those qualities in released hybrids.
The fresh market tomato breeding program at North Carolina State University is an applied
breeding program, and has two overarching objectives: to elucidate the genetic control of
economically important traits and to release improved cultivars. These two goals can be
addressed simultaneously by using intra-specific mapping populations. In the present study
18
data obtained from Tomato Analyzer in the F2:3 generation. Visual fruit shape and total
soluble solids was also phenotyped in two generations (F2 and F3). The main objective of this
research was to identify QTLs segregating within NC 10204 for fruit quality. It is unknown if
using Tomato Analyzer improves QTL mapping over traditional fruit phenotyping methods,
i.e. visual phenotyping.
MATERIALS AND METHODS
Population Development
The hybrid NC 10204 was created in 2010 by crossing the plum tomato breeding line NC
30P to the grape tomato breeding line NC 22L-1(2008). The F1 hybrid was allowed to
self-pollinate to create a segregating F2 population and in the summer of 2013, 284 F2 plants were
grown at the Mountain Horticultural Crops Research and Extension Center in Mills River,
NC. F2 plants were individually harvested to create F2:3 families. During seed cleaning in the
fall of 2013 F2:3 seeds were exposed to deleterious amounts of bleach and 111 lines were
eliminated leaving 173 F2:3 families. Problem with germination vigor reduced the number of
lines phenotyped in the F3 generation (Table 2.1).
Seeds were germinated in 72 cell trays (56 x 28 cm2) in potting mix and grown for 6 weeks
before hand transplanting in the field. Individual plants were grown 45 cm apart within rows,
150 cm apart between rows, grown in plastic mulch, with drip irrigation. Plants were hand
strung and sprayed according to the recommended schedule (Cooperative Extension Services
19 DNA Extraction and Genotyping
Genomic DNA was obtained from F2 plants and parental lines in the summer of 2013 using a
modified cetyltrimethyl ammonium bromide method, and stored at -20C in 10 mM Tris–
HCl pH 8.0, and 1 mM EDTA (Kabelka, Franchino, Francis 2002). Prior to genotyping,
DNA was quantified using a NanoDrop 2000 Spectrophotometer (Thermo Scientific,
Wilmington, Delaware, USA). 190 F2 plants (94 in 2013 and 96 in 2015) were genotyped
using the SolCAP Illumina Infinium Assay. SNP genotypes were determined using
GenomeStudio version 1.0 (Illumina Inc, San Diego, California, USA).
Phenotyping
Fruit shape was visually phenotyped in the F2 and F3 generations using a 1-8 scale according
to the International Plant Genetic Resources Institute’s ‘Descriptors for Tomato’
(International Plant Genetic Resources Institute 1996). Total soluble solids was measured
using an ATAGO Co. Ltd PR-32 refractometer (ATAGO U.S.A., Inc. Bellevue, Washington
USA). In the F2 generation total soluble solids was measured from five to six ripe fruit for
280 plants and in the F3 generation total soluble solids was measured from two repetitions of
twelve fruits from 126 families (Table 2.1).
In 2014, 129 F2:3 lines were phenotyped using Tomato Analyzer version 3.0, but SNP data
exists for only 106 families (Darrigues et al. 2008; Gonzalo et al. 2009a). Twenty fruit were
cut proximal to distal (stem end to blossom end), scanned using a flatbed scanner (CanoScan
8800F, Canon U.S.A. Inc, Melville, NY, USA) and images saved in JPEG format. Images
were analyzed as described by Rodriguez et al. (2010) and the results saved in an Excel
20
Tomato Analyzer with QTLs reported in this study can be found in Table 2.2; see the Tomato
Analyzer Version 3 User Manual for a full description of all traits
(http://oardc.osu.edu/vanderknaap/tomato_analyzer.php).
Data Analysis and QTL Mapping
Data analysis was performed in R 3.2.2 (R Core Team 2015) and SAS software, version 9.4
for Windows (SAS Institute Inc, Cary, NC). Pearson correlation coefficients were calculated
in R using the package ‘Hmisc’ (Table 2.5 - Table 2.6) (Harrell 2015). The broad-sense
heritability (H2) for fruit quality traits measured by Tomato Analyzer were calculated on a
plot basis by using the following formula:
𝐻2 = 𝜎𝑓 2
𝜎𝑓2+ 𝜎𝑒2
Where σ2
f is the family variance and σ2e is the residual variance. The results are shown in
Table 2.8.
The narrow-sense for total soluble solids, fruit shape and growth habit were calculated via
parent-offspring regression (Table 2.8):
ℎ2 = 𝐶𝑜𝑣𝐹2,𝐹3 𝜎𝐹22
Where 𝐶𝑜𝑣𝐹2,𝐹3is the covariance between the F2 and F3 generations, and 𝜎𝐹2
2 is the variance in
the F2 generation (Holland, Nyquist, Cervantes-Martínez 2003). Analysis of variance was
performed in SAS using ‘Proc GLM’ (Table 2.7). The least squared mean for individual
families (lines) in NC 10204 was calculated in R using the packages: ‘lsmeans’, ‘lme4’ and
21
estimate variance and least squared means are the same. Traits measured with Tomato
Analyzer the equation was:
𝐿𝑒𝑎𝑠𝑡 𝑠𝑞𝑢𝑎𝑟𝑒𝑑 𝑚𝑒𝑎𝑛𝑠 = 𝜇 + 𝑙𝑖𝑛𝑒 + 𝑓𝑟𝑢𝑖𝑡(𝑏𝑙𝑜𝑐𝑘) + 𝑒𝑟𝑟𝑜𝑟
where fruits from each line (or family) are nested within blocks and 𝜇 is the population mean.
The equation for total soluble solids and fruit shape was:
𝐿𝑒𝑎𝑠𝑡 𝑠𝑞𝑢𝑎𝑟𝑒𝑑 𝑚𝑒𝑎𝑛𝑠 = 𝜇 + 𝑙𝑖𝑛𝑒 + 𝑦𝑒𝑎𝑟 + 𝑙𝑖𝑛𝑒(𝑦𝑒𝑎𝑟) + 𝑓𝑟𝑢𝑖𝑡(𝑏𝑙𝑜𝑐𝑘 ∗ 𝑦𝑒𝑎𝑟) + 𝑒𝑟𝑟𝑜𝑟
where fruits from each line (or family) are nested within year, and 𝜇 is the population mean.
The NC 10204 linkage map was created using the R package ‘R/qtl’. R/qtl was also used for
QTL mapping and analysis (Broman et al. 2003). The Kosambi mapping function and a
genotyping error probability of 0.001 was used (Kosambi 1943). The final linkage map was
composed of 415 polymorphic SNPs, with an average of 2.1cM between markers across
858.9cM (Table 2.3 – Table 2.4, Figure 2.1). Marker order was determined using the physical
map position of each SNP marker according to SL2.40
(http://solgenomics.net/organism/Solanum_lycopersicum/genome). The final linkage map
contained markers with segregation distortion for twelve of the twenty-six SNPs on
chromosome 2 and for three of the forty-six SNPs on chromosome 4.
Maximum likelihood methods of QTL mapping assume phenotypic data has a normal
distribution, and it is recommend that non-normal phenotypic data be transformed prior to
QTL mapping (Yang, Yi, Xu 2006). Phenotypic data was tested for normality using the
Shapiro-Wilk test and traits with a p-value greater than 0.05 were considered normal. Traits
with a p-value less than 0.05 were transformed using Box-Cox power transformation (Box
22
𝑦𝑡 =(𝑦
𝜆 − 1)
𝜆
where yt is the transformed variable, y is the untransformed variable, and λ is the
transformation exponent determined using the ‘forecast’ package in R (Hyndman and
Khandakar 2008). When λ = 0 the transformation becomes a log transformation.
Traits were initially tested for the presence of a single QTL using the ‘scanone’ function in
R/qtl before a multiple QTL model was explored using the ‘stepwiseqtl’ command (Broman
and Speed 2002). No epistatic interactions were sought for fruit quality traits measured with
Tomato Analyzer because only one generation was phenotyped and marker data was
available for only 106 lines (Table 2.1). The same procedure was used to map QTLs for total
soluble solids and fruit shape but interactions were allowed during ‘stepwiseqtl’
(Manichaikul et al. 2009). Haley-Knott regression was used for QTL mapping (Haley and
Knott 1992) and the step width was set at 1cM. The LOD threshold for declaring a QTL was
determined via 5000 permutations at a significance level of α = 0.05. The 95% Bayesian
confidence intervals for the QTLs was determined in the R/qtl package. The LOD threshold
for epistatic interactions was determined for total soluble solids and fruit shape using the
‘scantwo’ function, but due to computational requirements was calculated at step width of
2.5cM and 1000 permutations.
QTL effects were calculated using ‘R/qtl’. The percent variance explained by the QTL, as
well as the additive and dominance effects were determined using the command ‘fitqtl’. The
additive effects were calculated using the formula:
23
where 𝜇𝐵𝐵 is the average value of the allele from NC30P and 𝜇𝐴𝐴 is the average of allele
from NC 22L-1(2008). Dominance effects were calculated using the formula:
𝐷𝑜𝑚𝑖𝑛𝑎𝑛𝑐𝑒 𝑒𝑓𝑓𝑒𝑐𝑡 = 𝜇𝐴𝐵 −
𝜇𝐴𝐴 + 𝜇𝐵𝐵 2
where 𝜇𝐴𝐵 is the average value of the heterozygote allele. The 95% Bayesian confidence
interval was calculated using the command ‘bayesint’ and the confidence interval was
expanded to the nearest marker (Broman et al. 2003).
RESULTS AND DISCUSSION
The fruit quality traits that Tomato Analyzer measures can be placed into four categories:
shape, size, uniformity and color traits. Only traits with QTLs reported in NC 10204 will be
discussed (Table 2.1). Size traits that will be discussed are: area, curved height, height
mid-width, maximum mid-width, perimeter, and width mid-height. Shape traits with QTLs in NC
10204 were distal fruit blockiness, ovoid, proximal fruit blockiness, proximal indentation
area, rectangular, shoulder height and width widest position. Uniformity traits with QTLs in
NC 10204 were eccentricity, eccentricity area index, and horizontal asymmetry ovoid. Only
three color traits mapped QTLs in NC 10204, average a* (colors ranging from red to green),
average chroma (color saturation) and average green (the green value in the RGB color
system).
A total of twenty-six QTLs were mapped for fruit quality traits (Table 2.9). Twenty-two fruit
quality traits had at least one QTL above the LOD threshold. Those twenty-two traits were
then analyzed using ‘stepwiseqtl’ giving twenty-seven QTLs. Roughly 80% were mapped to
24
mapped to chromosome 3 and one QTL to chromosome 11. Five fruit quality traits had two
QTLs: eccentricity, proximal angle micro, rectangular, shoulder height and horizontal
asymmetry ovoid.
Size traits mapped to chromosome 2 (four QTLs between 78.10cM and 80.10cM), and
chromosome 9 (three QTLs at 55.66cM). Fruit shape QTLs mapped to chromosome 3 (two
QTL between 3.10cM and 6.10cM), chromosome 9 (four QTLs between 29.10cM and
41.19cM), and chromosome 12 (three QTLs between 5.10cM and 6.10cM). Uniformity traits
were mapped to chromosome 3 (one QTL at 7.10cM), chromosome 9 (three QTLs between
27.10cM and 41.19cM), chromosome 11 (one QTL at 52.90cM) and chromosome 12 (one
QTL at 6.10cM). Color traits three QTLs mapped to chromosome 2 between 78.1cM and
81.10cM.
One trait, proximal angel micro, had a different number of significant QTLs when
transformed. The untransformed data for proximal angel micro mapped a single QTL to
chromosome 9 at 31.07cM. After Box-Cox transformation, proximal angel micro mapped
two QTLs to chromosome 9 (29.10cM) and 12 (6.10cM) (Table 2.9). Two traits, curved
height and maximum height, had no significant QTLs when untransformed but a single QTL
when transformed at 55.66cM on chromosome 9.
The correlation between blocks for all fruit quality traits were significant at p < 0.001 (Table
2.6). The broad-sense heritability for fruit quality traits measured with Tomato Analyzer
ranged from 0.22 (rectangular) to 0.52 (area) (Table 2.8). Overall fruit size traits (area,
perimeter, and width mid-height) had a higher heritability than fruit shape traits (rectangular,
25
Two QTLs were mapped for both total soluble solids and fruit shape. There were no
significant interactions for either total soluble solids or fruit shape. Fruit shape mapped
QTLs to roughly the same location on chromosome 9 (56.10cM) and 12 (5.10cM) as the
shape traits measured by Tomato Analyzer.
Segregation Distortion
There was extensive segregation distortion on chromosome 2, and to a lesser degree on
chromosome 4, with segregation distortion favoring NC 22L-1(2008) alleles over NC 30P
(Figure 2.2). On chromosome 2, segregation distortion extended for 52cM and included
twelve of twenty-six SNPs. On chromosome 4, three of the forty-six SNPs had segregation
distortion and two SNPs were adjacent with segregation distortion extending 3cM.
Villalta et al. (2005) reported extensive segregation distortion in two inter-specific tomato
populations. The populations had the same S. lycopersicum cv. cerasiforme accession crossed to different wild relatives (S. pimpinellifolium and S. cheesmanii). Segregation distortion generally favored heterozygotes, but on chromosome 2 S. pimpinellifolium markers were favored in one population, and S. lycopersicum cv. cerasiforme in the other (Villalta et al. 2005). Another study used restriction fragment length polymorphic markers to genotype a S. lycopersicum x S. cheesmanii population and all three markers on chromosome 2 had
segregation distortion favoring the S. lycopersicum allele (Paran et al. 1995).
Since marker order was determined using the physical map positions and not recombination
frequencies, any detrimental effects of segregation distortion on linkage map construction
was limited. Studies have suggested segregation distortion has little impact on the ability to
26
in segregation distortion (Xu 2008; Zhang et al. 2010). In an attempt to estimate the effects of
segregation distortion on QTL detection in NC 10204, the LOD threshold and maximum
LOD score for fruit quality traits was calculated using three different linkage maps.
Map 1 is the NC 10204 linkage map used for QTL mapping, Map 2 had twelve SNPs
showing segregation distortion after Bonferroni correction removed, and Map 3 had an
additional sixty-eight SNPs with segregation distortion without Bonferroni correction
removed (Table 2.10). There is little variation in LOD thresholds between the three linkage
maps, as the greatest difference is only 0.18. Similarly, the maximum LOD scores do not
vary by more than 1.00 for any linkage map. Map 1 had twenty-two fruit quality traits with a
maximum LOD score greater than the threshold, Map 2 had twenty-one, and Map 3 had
eighteen (Table 2.11). Two traits (fruit shape index external II and fruit shape index internal
I) had significant QTLs with Map 3 but not with Map 1 or 2.
The markers with segregation distortion were kept in the final linkage map because of the
effect on chromosome 2 when removed. Chromosome 2 in Map 1 has twenty-six markers
covering 85.26cM. When twelve markers with segregation distortion are removed to create
Map 2, the maximum spacing between SNPs increases from 41.09cM to 109.06cM (Table
2.10). Chromosome 2 in Map 3 only has four markers covering 4.45cM (Table 2.10). The
minimal difference between the three maps to detect QTLs, plus the detrimental effect on
chromosome 2 when segregation distortion SNPs are removed justifies the decision to not
27
QTLs on Chromosome 2 Associated with Multiple Fruit Quality Traits
Four fruit size traits had QTLs on chromosome 2: area, perimeter, maximum width, and
width mid-height. All mapped to a 2cM region between 78.1cM and 80.1cM (Table 2.9). The
NC 30P markers were associated with greater fruit size, as seen by the positive additive
values in Table 2.9. Three color traits (average a*, average chroma and average green)
mapped to a 7.26cM region (78.10cM to 85.36cM) but their 95% confidence intervals range
from 19cM to 66cM.
QTLs for fruit size and color have previously been mapped to chromosome 2. Three genes
important for fruit size are located on chromosome 2 (ovate, fw2.2 and lcn2.1), as well as at least three other putative QTLs (Lin et al. 2014). An intra-specific population mapped QTLs
using the CIELAB color space for a* (where a* represents colors ranging from red to green),
b* (representing blue to yellow) and L* (color lightness ranging from white to black) to
chromosomes 2, 4 and 9 (Saliba-Colombani et al. 2001). (Yang et al. 2004) mapped a QTL
for L* and (Darrigues 2007) a QTL for color intensity (a function of a*, b* and chroma) to
chromosome 2.
QTLs on Chromosome 9 Associated with Multiple Fruit Quality Traits
QTLs for curved height, eccentricity area index, eccentricity, height mid-width, maximum
height, proximal angle micro, proximal fruit blockiness, proximal indentation area,
rectangular, and shoulder height mapped to chromosome 9 (Table 2.9). The allele from NC
22L-1(2008) conferred a higher value for all traits except eccentricity, cured height,
28
and 55.66cM; though confidence intervals vary from 7.44cM (rectangular) to nearly the
entire chromosome 87.17cM (maximum height).
(Chen et al. 1999) mapped QTLs for fruit shape, polar diameter and total soluble solids to
chromosome 9 using an inter-specific (S. lycopersicum x S. pimpinellifolium) BC1S1 population. The S. lycopersicum parent in the aforementioned study was NC 84173 which has common ancestry with NC 30P and NC 22L-1(2008). Researchers have used Tomato
Analyzer to phenotype four populations derived from crossing the same S. pimpinellifolium
accession to different S. lycopersicum accessions. Table 2.12 compares the QTLs mapped using those four inter-specific populations, to NC 10204 (Brewer et al. 2007; Gonzalo and
van der Knaap 2008; Gonzalo et al. 2009b). Two traits in NC 10204 with QTLs on
chromosome 9 mapped to chromosome 9 in other populations: proximal fruit blockiness (one
population) and proximal indentation area (two populations) (Table 2.12).
QTLs on Chromosome 12 Associated with Multiple Fruit Quality Traits
Six fruit quality traits had QTLs mapped to chromosome 12: eccentricity, horizontal
asymmetry ovate, ovoid, shoulder height, width widest position, and proximal angle micro.
All six QTLs were located between 4.10-6.10cM and confidence intervals ranged
from10.65cM (ovoid and width widest position) to 75cM (horizontal asymmetry ovate)
(Table 2.9). The allele from NC 22L-1(2008) conferred a higher phenotypic value for
shoulder height, ovoid and horizontal asymmetry ovate) and the allele from NC 30P
conferred a higher phenotypic value for eccentricity and width widest position.
There are no similarities between traits with QTLs on chromosome 12 in NC 10204 and
29
diameter on chromosome 12 located between 10.00cM and 38.00cM. A QTL for fruit weight
was also mapped to a similar region (0.00-22.00cM) using an inter-specific recombinant
inbred line population (S. lycopersicum x S. pimpinellifolium) (Ashrafi et al. 2012). The S. lycopersicum parent in RIL population described in Ashrafi et al. (2012), NC EBR-1, has common ancestry with both NC 30P and NC 22L-1(2008).
Ovoid is measured when fruit mass is greater in upper half of the fruit. If fruit mass is greater
in the lower half, then the fruit is considered obovoid and can be pear shaped. There are no
significant QTLs at α = 0.05 for obovoid, but after Box-Cox transformation there is a
significant QTL at α = 0.10 on chromosome 12 at 4.1cM. The ovate gene is responsible for
fruit elongation, and fruit are more rounded when ovate is not homozygous recessive (Rodríguez, Kim, van der Knaap 2013). Previous studies have identified putative QTLs on
chromosome 12 that influence ovate (Gonzalo and van der Knaap 2008; Rodríguez, Kim, van der Knaap 2013). The loci identified by Gonzalo and van der Knaap 2008 corresponds the
same end of chromosome 12 as the QTL for obovoid in NC 10204 (α = 0.10). The loci
identified by Rodriguez et al. 2013 is on the other end of chromosome 12.
QTLs on Chromosome 3 & 11 Associated with Multiple Fruit Quality Traits
One QTL for distal fruit blockiness, rectangular, and fruit shape triangle mapped to
chromosome 3. All three QTLs mapped to a 4cM region (3.10cM to 7.10cM) and the
confidence intervals ranged from 14.82cM to 35.16cM. NC 22L-1(2008) conferred a higher
phenotypic value for distal fruit blockiness and rectangular. One population in Table 2.12
30
marker associated with distal fruit blockiness on chromosome 3, but the LOD score was not
above the significance threshold (Table 2.12).
One QTL for horizontal asymmetry ovate and ovoid mapped to the same location (52.90 cM)
on chromosome 11. The four populations in Table 2.12 mapped many fruit quality traits on
chromosome 11, but none reported QTLs for horizontal asymmetry ovate or ovoid.
Total soluble solids
QTLs for total soluble solids were mapped to chromosome 2 (78.1cM) and 9 (19.35cM)
when the least squared means of the F2 and F3 generations was analyzed (Table 2.1). For both
QTLs the allele from the small fruited NC 22L-1(2008) increased total soluble solids. The
position of the QTL on chromosome 2 corresponded to the QTLs for fruit size (area,
perimeter, width mid-height, maximum width) (Table 2.9). The QTL on chromosome 9 is
located between 7.75cM and 36.31cM from the QTLs for fruit quality traits measured with
Tomato Analyzer and the 95% confidence interval extended 46.01cM.
Total soluble solids and fruit size traits (perimeter, maximum width, and area) are negatively
correlated, r ≈ -0.50, p < 0.001 (Table 2.5). When phenotypic data from only the F2
generation was analyzed, two QTLs were mapped to the same chromosome and position as
the QTLs when the least squared means of both generations was analyzed (Table 2.9). No
significant QTLs were mapped in the F3 generation. The F2 and F3 generations are correlated,
31 Fruit Shape
Fruit shape measured visually mapped two QTLs to chromosome 9 (56.10cM) and 12
(5.10cM) when the least squared means from the F2 and F3 generations was used (Table 2.9).
When the generations were analyzed separately, the F2 generation mapped a single QTL to
chromosome 9 (46.10cM) and the F3 generation mapped no QTLs. The correlation between
generations is, r = 0.38, p < 0.001.
Comparison of Phenotypic Data Obtain using Tomato Analyzer and Phenotyping Fruit Shape
Visually
It took the combined phenotypic data from two generations of visual fruit phenotyping to
map QTLs to chromosome 9 and 12. Fruit shape traits measured with Tomato Analyzer
mapped QTLs to the same regions of chromosome 9 and 12 using one generation of
phenotypic data. The 95% confidence interval were generally smaller with data obtained with
Tomato Analzyer. The confidence interval for fruit shape when phenotyped visually was
25.77cM (chromosome 12) to 35.60cM (chromosome 9), while six of the nine fruit quality
traits measured using Tomato Analyzer that mapped to chromosome 9 and 12 had shorter
confidence intervals (Table 2.9). This is not surprising as Tomato Analyzer quantitatively
measures fruit traits and visual fruit shape phenotyping is a categorical scale. The quality of
phenotypic data is only one factor to consider when comparing Tomato Analyzer to visually
phenotyping fruit shape.
Visually phenotyping fruit shape requires one to two people to phenotype ripe fruit in the
field. The speed and accuracy of fruit shape phenotyping depends largely upon the
32
is undoubtedly more accurate than visually phenotyping fruit shape, but requires more labor.
The time required to scan fruit and manually adjust the images prior to analysis cannot be
understated. Fruit scanning occurs as summer field trials are being harvested and the fall
greenhouse season is starting. This is the busiest times of the year for breeding programs and
it is difficult to allocate the labor required to scan fruit.
The multiple generations of visually phenotyping fruit shape that were required equal one
generation of phenotyping with Tomato Analyzer should not be a limitation in most
scenarios since most populations are evaluated in multiple generations and locations anyway.
The amount of time and labor required to use Tomato Analyzer make it impractical for
regular use in an applied breeding program.
CONCLUSION
QTL analysis suggests that NC 10204 is segregating for genes controlling fruit shape on
chromosomes 9 and 12, and for fruit size on chromosome 2. Fruit shape QTLs were detected
at the same locations using both Tomato Analyzer and visual fruit phenotyping. Future
research is needed to validate these QTLs. Even if QTLs are never validated, the loci
identified in this research are still of value to the fresh market tomato breeding program at
33
REFERENCES
Ashrafi H, Kinkade MP, Merk HL, Foolad MR. 2012. Identification of novel quantitative trait loci for increased lycopene content and other fruit quality traits in a tomato recombinant inbred line population. Molecular Breeding 30(1):549-67.
Bates D, Mächler M, Bolker B, Walker S. 2015. Fitting linear mixed-effects models using lme4. Journal of Statistical Software 67(1):1-48.
Bauchet G, Munos S, Sauvage C, Bonnet J, Grivet L, Causse M. 2014. Genes involved in floral meristem in tomato exhibit drastically reduced genetic diversity and signature of selection. BMC Plant Biology 14:279,014-0279-2.
Box GE and Cox DR. 1964. An analysis of transformations. Journal of the Royal Statistical Society.Series B (Methodological) :211-52.
Brewer MT, Moyseenko JB, Monforte AJ, van der Knaap E. 2007. Morphological variation in tomato: A comprehensive study of quantitative trait loci controlling fruit shape and development. Journal of Experimental Botany 58(6):1339-49.
Broman KW and Speed TP. 2002. A model selection approach for the identification of quantitative trait loci in experimental crosses. Journal of the Royal Statistical Society: Series B (Statistical Methodology) 64(4):641-56.
Broman KW, Wu H, Sen S, Churchill GA. 2003. R/qtl: QTL mapping in experimental crosses. Bioinformatics 19(7):889-90.
Causse M, Saliba-Colombani V, Lecomte L, Duffe P, Rousselle P, Buret M. 2002. QTL analysis of fruit quality in fresh market tomato: A few chromosome regions control the variation of sensory and instrumental traits. Journal of Experimental Botany
53(377):2089-98.
Chen F, Foolad M, Hyman J, Clair DS, Beelaman R. 1999. Mapping of QTLs for lycopene and other fruit traits in a lycopersicon esculentum× L. pimpinellifolium cross and comparison of QTLs across tomato species. Molecular Breeding 5(3):283-99.
Cooperative Extension Services. 2015. X. disease control. In: 2015 north carolina agricultural chemical manual. 1st ed. Raleigh, North Carolina: North Carolina State University. 525 p.
34
Darrigues A, Hall J, van der Knaap E, Francis DM, Dujmovic N, Gray S. 2008. Tomato analyzer-color test: A new tool for efficient digital phenotyping. Journal of the American Society for Horticultural Science 133(4):579-86.
Gonzalo MJ and van der Knaap E. 2008. A comparative analysis into the genetic bases of morphology in tomato varieties exhibiting elongated fruit shape. Theoretical and Applied Genetics 116(5):647-56.
Gonzalo MJ, Brewer MT, Anderson C, Sullivan D, Gray S, van der Knaap E. 2009a. Tomato fruit shape analysis using morphometric and morphology attributes implemented in tomato analyzer software program. Journal of the American Society for Horticultural Science 134(1):77-87.
Gonzalo MJ, Brewer MT, Anderson C, Sullivan D, Gray S, van der Knaap E. 2009b. Tomato fruit shape analysis using morphometric and morphology attributes implemented in tomato analyzer software program. Journal of the American Society for Horticultural Science 134(1):77-87.
Grandillo S, Termolino P, van der Knaap E. 2013. Molecular mapping of complex traits in tomato. In: Genetics, genomics and breeding of tomato. B. Liedl, J. Labate, J. Stommel, A. Slade & C. Kole, editor. 1st ed. Boca Raton, FL: CRC Press. 150 p.
Halekoh U and Højsgaard S. 2014. A kenward-roger approximation and parametric bootstrap methods for tests in linear mixed models–the R package pbkrtest. Journal of Statistical Software 59(9):1-32.
Haley CS and Knott SA. 1992. A simple regression method for mapping quantitative trait loci in line crosses using flanking markers. Heredity 69(4):315-24.
Harrell FE. 2015. Hmisc: Harrell Miscellaneous.[computer program]. Version 3.16. http://CRAN.R-project.org/package=Hmisc.
Holland JB, Nyquist WE, Cervantes-Martínez CT. 2003. Estimating and interpreting heritability for plant breeding: An update. Plant Breed Reviews 22:9-112.
Hyndman R and Khandakar Y. 2008. Automatic time series forecasting: The forecast package for R. Journal of Statistical Software 26(3):1-22.