Mapping Black Spot Disease Resistance and Cold Hardiness in Garden Roses (Rosa X Hybrida) by Cindy Rouet

Mapping black spot disease resistance and cold hardiness in garden roses (Rosa x hybrida) by Cindy Rouet

A Thesis Presented to The University of Guelph

In partial fulfilment of requirements for the degree of Doctor of Philosophy in Plant Agriculture

Guelph, Ontario, Canada

ABSTRACT

MAPPING BLACK SPOT DISEASE RESISTANCE AND COLD HARDINESS IN GARDEN ROSES (ROSA X HYBRIDA)

Cindy Rouet Advisor: University of Guelph, 2021 Professor E. Lee

This thesis aims to contribute to the modernization of Canada’s national rose breeding program by developing molecular markers associated with traits of major interest. Roses are one of the most economically important ornamental crops. Consumers’ preferences are continuously evolving, meaning that the rose industry needs to release new varieties at a fast pace to remain competitive. As a consensus, consumers want low maintenance roses. To meet consumers’ demand, Canada’s national rose breeding program, which is hosted in the Niagara Region, focuses on breeding for black spot disease resistance and winter hardiness. In this regard,

Canadian Explorer roses are known worldwide for their exceptional hardiness. The objective of this thesis was to identify quantitative trait loci (QTL) associated with resistance to black spot disease (Diplocarpon rosae) and winter hardiness in polyploid bi-parental populations derived from Explorer roses, and to set up a framework for the development of molecular markers and the implementation of marker-assisted selection. Molecular tools not only have the potential to improve the accuracy and speed of the selection process, but they also promise to reduce the cost and the labor associated with recording complex phenotypic traits. This research identified a major QTL associated with the resistance to several races of Diplocarpon rosae under natural and artificial conditions on linkage group 1 of the rose genome for which a diagnostic marker

was designed and validated in the genetic background of the parental donor. Furthermore, this research identified several QTLs associated with winter damage, spring regrowth and freezing tolerance measured by electrolyte leakage in artificial conditions. While no diagnostic markers were developed, this research highlighted the limited utility of electrolyte leakage as a proxy for field winter hardiness and the complexity of this phenotypic trait. Together, these results demonstrate the potential for implementing marker-assisted selection for black spot disease resistance in a rose breeding program, and provide a starting point for the molecular characterization of the genetic and molecular basis of winter hardiness in roses.

ACKNOWLEDGEMENTS

I would like to thank the members of my advisory committee Dr. Elizabeth Lee, Dr.

Annette Nassuth and Dr. David Wolyn for their incredible support and advice. I am particularly thankful to Dr. Daryl Somers who was my thesis advisor from Vineland Research and

Innovation Centre until September 2019. His guidance, patience, trust, support and optimism have always been a source of motivation to me. Special thanks to Dr. Jayasankar Subramanian for accepting to join my advisory committee one year before the completion of my research.

I would like to thank the past and current members of Vineland Research and Innovation who contributed to this research. Thanks to Mr. Travis Banks and thanks to Ms. Rachael

Leblanc, Dr. Michael Pautler and Dr. Parminder Sandhu who have shared their scientific expertise. Thanks to the past and current members of the rose breeding lab Mr. Ernie Morimoto,

Mr. Francesco Pacelli, Mrs. Jacqueline Bantenburg and Dr. Ashok Ghosh for their technical support and friendship. Thanks to Dr. Annissa Poleatewich and Ms. Irina Perez-Valdez for sharing their expertise in plant pathology. Thanks to Ms. Jaclyn Prystupa and Mr. Joseph O’Neill for taking the time to train me on genomic laboratory work and bioinformatics and for their friendship. Thanks as well to Mrs. Rachel Smith for her time and expertise with genomic laboratory work. Special thanks to Mr. Michael Josiak and his team, and to Mr. James Toivonen,

Mrs. Cathy Gray and Mr. Kevin Meester for their help with rose cultivation both in the field and in the greenhouse. I also would like to express my sincere gratitude to Dr. Rumen Conev who initially believed in my abilities to study plant breeding and genetics. I would like to thank Dr.

Karen Tanino, Ms. Jacqueline Bantle and Ms. Liu Rensong for their inestimable support with the

Saskatoon cold hardiness field trial.

I would like to thank Dr. Beatrice Amyotte and Mrs. Ingrid Giles for their unconditional friendship and professional support. I also would like to acknowledge the group of graduate students and young scientists from the National Association of Plant Breeders who have been a unique source of inspiration, with special thanks to Dr. Kevin Falk.

In addition, I would like to acknowledge the funding sources that have supported this research. I would like to thank the donors of the Plant Agriculture and Ontario Agricultural

College awards including Mrs. June Laver and the estate of May Ball, the University of Guelph, and NSERC, OMAFRA and AAFC (Growing Forward 2 partnership and Canadian Agricultural partnership) for their generous support.

Last but not least, I would like to thank my family — my mother Christine, my father

Jean-Luc, my brother Matthis, my partner Igor and his parents Olga and Sergei— for their unconditional love and support.

TABLE OF CONTENTS

ABSTRACT ……………………………………………………………………………………..ii ACKNOWLEDGEMENTS …………………………………………………………………....iv

TABLE OF CONTENTS …………………………………………………………………...... vi LIST OF TABLES …………………………………………………………………...... xi LIST OF FIGURES…………………………………………………………………...... xv LIST OF ABBREVIATIONS ………………………………………………………………...xix

CHAPTER 1: INTRODUCTION AND LITERATURE REVIEW ...... 1 1.1 Economical and Sociological Significance of Roses ...... 1 1.2 Roses Origin, Classification and Phylogeny ...... 2 1.3 Rose Biology and Genetics ...... 4 1.4 Rose Breeding ...... 7 1.4.1 Overview ...... 7 1.4.2 Canada’s National Rose Breeding Program ...... 9 1.4.3 Genetic Diversity ...... 9 1.5 Advances in Applied Genomics and Opportunities for Rose Breeding ...... 12 1.5.1 Genetic Maps as a Framework for the Identification of Marker-Trait Association 12 1.5.2 Marker-Assisted Selection : Challenges and Opportunities for Ornamental Crops 18 1.5.3 Genome Database for Rosaceae and RosBREED ...... 20 1.6 Black Spot Disease ...... 21 1.6.1 Biology, Disease Cycle and Ecology of D. rosae ...... 21 1.6.2 Interaction between the Host and the Pathogen ...... 23 1.6.3 Disease Management ...... 24 1.6.4 Diversity of Diplocarpon rosae and Race Characterization ...... 26 1.6.5 Breeding for Resistance ...... 27 1.6.6 Genetically Transformed Roses for Black Spot Resistance ...... 29 1.7 Cold Hardiness ...... 30 1.7.1 Physiology of Cold Hardiness ...... 30 1.7.1.1 Plant Strategies to Survive Low Temperatures during the Winter ...... 30 1.7.1.2 Acclimation and Deacclimation ...... 31 1.7.2 Genetic Control of Cold Hardiness: the CBF Pathway ...... 33

1.7.3 Phenotyping ...... 36 1.7.4 QTL Mapping ...... 39 1.8 Thesis Introduction ...... 40 1.8.1 Background ...... 40 1.8.2 Rationale ...... 41 1.8.3 Implication ...... 42 1.8.4 Objectives and Hypothesis of each Research Chapter ...... 43 CHAPTER 2: IDENTIFICATION OF A POLYMORPHISM WITHIN THE ROSA MULTIFLORA MURDR1A GENE LINKED TO RESISTANCE TO MULTIPLE RACES OF DIPLOCARPON ROSAE W. IN TETRAPLOID GARDEN ROSES (ROSA X HYBRIDA)…….………………………………………………………………………………..45 2.1 Introduction ...... 47 2.2 Materials and Methods ...... 49 2.2.1 Genetic materials ...... 49 2.2.2 D. rosae single spore isolates ...... 52 2.2.3 Detached leaf assay for race specific D. rosae resistance ...... 52 2.2.4 Natural field disease pressure screening for BS resistance ...... 54 2.2.5 Single Nucleotide Polymorphism calling ...... 54 2.2.6 Phenotypic data ...... 55 2.2.7 Construction of genetic maps...... 56 2.2.8 QTL analysis ...... 57 2.2.9 Analysis of muRdr1A sequence ...... 57 2.3 Results ...... 59 2.3.1 Race diversity used in this study ...... 59 2.3.2 Segregation for black spot resistance within the mapping population ‘CA60’ x ‘SITR’ ……………………………………………………………………………………..61 2.3.3 Parent-specific linkage maps and QTL analysis for BS resistance...... 61 2.3.4 Candidate genes and development of ‘CA60’ Rdr1A allele-specific markers ...... 69 2.4 Discussion ...... 76 2.4.1 Race characterization and identification of a new race of D. rosae ...... 76 2.4.2 Development of molecular markers linked to BS resistance ...... 76 2.4.3 Utility and source of the ‘CA60’ Rdr1A BS resistance allele ...... 79 2.5 Conclusions ...... 80 CHAPTER 3: MAPPING COLD HARDINESS IN GARDEN ROSES (ROSA x HYBRIDA)…….………………………………………………………………………………..81 3.1 Introduction ...... 83

vii

3.2 Material and Methods...... 87 3.2.1 Genetic material ...... 87 3.2.1.1 Experiment 1_ Electrolyte leakage as a proxy for winter hardiness ...... 87 3.2.1.2 Experiment 2_ QTL mapping of electrolyte leakage ...... 91 3.2.1.3 Experiment 3_ QTL mapping of field-based winter hardiness ...... 91 3.2.2 Experimental design and growing conditions ...... 92 3.2.2.1 Experiment 1_ Electrolyte leakage as a proxy for winter hardiness ...... 92 3.2.2.2 Experiment 2_ QTL mapping of electrolyte leakage ...... 92 3.2.2.3 Experiment 3_ QTL mapping of field-based winter hardiness ...... 95 3.2.3 Phenotyping ...... 96 3.2.3.1 Experiment 1_ Electrolyte leakage as a proxy for winter hardiness ...... 96 3.2.3.2 Experiment 2_ QTL mapping of electrolyte leakage ...... 97 3.2.3.3 Experiment 3_ QTL mapping of field-based winter hardiness ...... 98 3.2.4 Data Analysis ...... 99 3.2.4.1 Experiment 1_ Electrolyte leakage as a proxy for winter hardiness ...... 99 3.2.4.2 Experiment 2_ QTL mapping of electrolyte leakage ...... 100 3.2.4.3 Experiment 3_ QTL mapping of field-based winter hardiness ...... 101 3.3 Results ...... 104 3.3.1 Experiment 1_ Electrolyte leakage as a proxy for winter hardiness ...... 104 3.3.2 Linkage maps ...... 109 3.3.3 Experiment 2_ QTL mapping of electrolyte leakage ...... 110 3.3.3.1 Calibration with parental lines ...... 110 3.3.3.2 ‘CA60’x’SITR’ population...... 110 3.3.3.3 ‘EDI’x’GV’ population ...... 117 3.3.4 Experiment 3_ QTL mapping of field-based winter hardiness ...... 126 3.3.4.1 ‘CA60’ x ‘SITR’ population...... 126 3.3.4.2 ‘EDI’ x ‘GV’ population ...... 138 3.3.5 Relationship between electrolyte leakage and field winter hardiness ...... 139 3.4 Discussion ...... 143 3.4.1 Relationship between electrolyte leakage and field winter damage ...... 143 3.4.2 Distinct genetic basis of freezing tolerance under artificial stress and field winter hardiness ...... 145 3.4.3 QTL for winter damage ...... 146 3.4.4 QTL for regrowth ...... 147 3.4.5 Genetics of winter hardiness and regrowth ...... 147 viii

3.5 Conclusion ...... 152 CHAPTER 4: GENERAL DISCUSSION: TOWARD A TECHNOLOGY AND DATA DRIVEN ROSE BREEDING PROGRAM ...... 154 4.1 Breeding for black spot disease resistance in garden roses ...... 154 4.1.1 Summary of the findings ...... 154 4.1.2 Implementation of marker-assisted selection: promises and impediments ...... 156 4.1.3 Future directions for the identification and characterization of novel sources of black spot resistance ...... 158 4.2 Breeding for cold hardiness in garden roses ...... 160 4.2.1 Summary of findings...... 160 4.2.2 Future directions for the characterization of the genetic basis of cold hardiness in roses ……………………………………………………………………………………162 4.3 Suggested selections ...... 163 4.4 Tools for polyploids ...... 164 REFERENCES ………………………………………………………………………………166 APPENDICES ………………………………………………………………………………193 APPENDIX 1: Supplementary Materials and Methods for Chapter 2: Identification of a polymorphism within the Rosa multiflora muRdr1A gene linked to resistance to multiple races of Diplocarpon rosae w. in tetraploid garden roses (rosa x hybrida) ...... 193 A1.1 Race characterization and race diversity at VRIC experimental farm ...... 193 A1.2 DNA isolation and sequencing of Genotyping-by-Sequencing (GBS) libraries ...... 193 A1.3 Analysis of muRdr1A sequence ...... 194 A1.4 Statistical Analysis ...... 194 APPENDIX 2: Standard host differential infection patterns to characterize races of D. rosae and list of isolates examined ...... 196 APPENDIX 3: QTL associated with resistance to four Diplocarpon rosae isolates and natural field infection. ………………………………………………………………………………197 APPENDIX 4: Disease rating scale used to screen rose hybrids for black spot disease (Diplocarpon rosae) resistance ...... 198 APPENDIX 5: Genetic map generated for the parental genotype ‘SITR’ in the context of the black spot research project ...... 199 APPENDIX 6: Supplementary Materials and Methods for Chapter 3: Mapping cold hardiness in garden roses (Rosa x hybrida)...... 200 A6.1 Data Analysis of electrolyte leakage data ...... 200 A6.1.1 Parental lines ...... 200 A6.1.2 ‘CA60’ x ‘SITR’ Population ...... 201 A6.1.3 ‘EDI’ x ‘GV’ Population ...... 201 APPENDIX 7: Evidence for the CBF/DREB and ICE1 transcription factors in the reference genome R. chinensis...... 203

APPENDIX 8: Field and electrolyte leakage data for 19 commercial cultivars and eight rose selections used in Experiment 1 ...... 204 APPENDIX 9: Comparison of magnitude of electrolyte leakage between the four replications of Experiment 2 ………………………………………………………………………………206 APPENDIX 10: Range of LT50s in the 'EDI'x’GV' population ...... 207 APPENDIX 11: Key genetic factors involved in the acquisition of cold hardiness within the CBF pathway ………………………………………………………………………………208

LIST OF TABLES Table 1-1. Ploidy level of rose species belonging to the genus Rosa (Adapted from Yokoya et al.

2000) ...... 6

Table 1-2. Desired aesthetic features in garden roses and their genetic control ...... 8

Table 1-3. Rose species that contributed to the genetic background of modern roses ...... 11

Table 1-4. Suggestions of trait introgressions from rose species into modern germplasm ...... 11

Table 1-5. Genotypic and phenotypic segregation for a trait controlled by a single bi-allelic gene, with domainant allele A, under random tetrasomic inheritance and complete dominance reveal the allelic state of the parental lines...... 15

Table 2-1. Plant Material used in the study...... 51

Table 2-2. Comparison of the host range of seven isolates of D. rosae with the published infection patterns for known D. rosae races using a set of differential cultivars ...... 60

Table 2-3. Mixed model variance of the effect of Isolate and Genotype on the disease score attributed in detached leaf assay using single spore inoculum of D.rosae ...... 63

Table 2-4. Segregation of phenotypes among the mapping population 'CA60' x 'SITR' in black spot detached leaf assay conducted with four D.rosae isolates and under field disease pressure 65

Table 2-5. Pearson correlation test between phenotypic data collected in detached leaf assay with four D.rosae isolates and phenotypes under field disease pressure (R version 3.5.3) ...... 66

Table 2-6. QTL associated with resistance to four D. rosae isolates and with resistance to black spot in the field under natural infection (QTL identification conducted by performing a genome scan by Haley-Knott regression, R version 3.5.3, R/qtl2) ...... 67

Table 2-7. Frequency of the different phenotypic groups for different isolates of D. rosae and field resistance under natural infection across the mapping population for the presence/absence and dosage of the ‘CA60’ Rdr1A allele...... 75

Table 3-1. Genetic material used in this multi-experiment study ...... 88

‘GV’ and ‘SITR’...... 112

Table 3-4. LT50 of the parental lines 'CA60', 'EDI', 'GV', 'SITR' estimated from non-linear dosage response curves from the index of injury...... 113

Table 3-5. General linear mixed model (GLMM) of the effect of Temperature and Genotype on the electrolyte leakage measured in controlled conditions after freezing treatments with temperature-tests ranging from -10 to -20˚C for the population ‘CA60’x’SITR’ and from -15 to -

40 ˚C for the population ‘EDI’x’GV’...... 115

xii

Table 3-6. Evidence for transgressive segregation for electrolyte leakage, field winter damage and field spring regrowth in the mapping populations ‘CA60’x’SITR’ and ‘EDI’x’GV’ ...... 119

Table 3-7. QTLs associated with electrolyte leakage (EL), field winter damage (WD), and field spring regrowth (RG) in two mapping populations ‘CA60’ x ‘SITR’ and ‘EDI’ x ‘GV’...... 120

Table 3-8. Generalized Linear Mixed Model (GLMM) of the effect of genotype and genotype- by-environment on winter damage for the mapping population 'CA60' x 'SITR' across three environments (Elora 2019, Elora 2020 and Sask 2019) and across individual environments, and for the mapping population ‘EDI’ x ‘GV’ across one environment (Elora 2020)...... 129

Table 3-9. Correlation between best linear unbiased prediction estimates (BLUPs) of winter damage (WD) in three environments (Elora2019, Elora 2020 and Sask 2019), defoliation and

BLUPs of regrowth (RG) for four environments (Elora2019, Elora 2020, Sask 2019 and Sask

2020) for the mapping population ‘CA60’ x ‘SITR’, and between BLUPs of WG and RG for

‘EDI’ x ‘GV’ population...... 132

Table 3-10. Heritability and variance estimates for field winter damage and regrowth for the

‘CA60’ x ‘SITR’ and ‘EDI’ x ‘GV’ populations...... 133

Table 3-11. Generalized Linear Mixed Model (GLMM) of the effect of genotype and genotype- by-environment on regrowth for the mapping population 'CA60' x 'SITR' across four environments (Elora 2019, Elora 2020, Sask 2019 and Sask 2020) and across individual environments, and for the mapping population ‘EDI’ x ‘GV’ across one environment (Elora

2020)...... 135

xiii

xiv

LIST OF FIGURES Figure 1-1. Typical black spot symptoms on an infected and almost entirely defoliated rose bush

(Rosa x hybrid) (Credit Rouet C 2019, Vineland Research and Innovation Centre) ...... 22

Figure 1-2. Polycyclic disease cycle of Diplocarpon rosae Wolf ...... 22

Figure 1-3. Regulation of the CBF pathway in Arabidopsis (Adapted from Wisniewski et al.

2018) ...... 35

Figure 1-4. Traditional rating scale from 0 to 5 used to assess winter damage in the rose field

(Credit Rouet C. 2014, Vineland Research and Innovation Centre)...... 37

Figure 2-2. Sequence alignment between the published sequence of Rdr1 locus and Rosa chinensis physically locates Rdr1 on Chr.1 in the Rosa chinensis genome...... 68

Figure 2-3. Discovery of a polymorphism in ‘CA60’ that is unique to muRdr1Agene in the

Rdr1locus and Primer Design...... 71

Figure 2-4. Multiple sequence alignment of all nine paralogues of the Rdr1locus around the

‘CA60’ unique INDEL (underlined). The ‘CA60’ consensus sequence was obtained by aligning the ‘CA60’ Whole Genome Sequence reads to the entire Rdr1locus...... 72

Figure 2-5. High resolution melt profiles of markers amplified in segregating populations based around the 32bp INDEL polymorphism present in the muRdr1Agene of ‘CA60’ (Rrrr), SITR

(rrrr) and ‘S13-10’ (Rrrr)...... 73

Figure 3-1. Correlations between USDA cold hardiness zones, winter hardiness (WH) as recorded on a scale from 0 to 5 at Vineland’s experimental farm in 2018 and electrolyte leakage

(EL) measured at different temperatures in artificial freezing experiments for 17 rose cultivars

...... 106

Figure 3-4. Response of four parental genotypes, ‘CA60’, ‘SITR’, ‘EDI’, and ‘GV’, to freezing treatment from -10 to -50°C. Sensitivity to freezing is displayed as the BLUE estimates of the index of injury (I), estimated from two replications of electrolyte leakage (EL) experiments .. 114

xvi

Figure 3-5. Distribution of Best Linear Unbiased Predictors (BLUPs) of electrolyte leakage (EL)

A) in the ‘CA60’ x ‘SITR’ population at -10˚C, -15˚C and -20˚C and B) in the ‘EDI’ x ‘GV’ population at -15˚C, -20˚C, -25˚C, -30˚C, -35˚C and -40˚C ...... 116

Figure 3-6. QTL mapping of Best Linear Unbiased Predictors (BLUPs) of electrolyte leakage

(EL) at -10˚C and -20˚C, field winter damage (WD) and field regrowth (RG) in four environments (Elora 2019, Elora 2020, Sask 2019 and Sask 2020) in the ‘CA60’ female map..

...... 122

Figure 3-7. QTL mapping of Best Linear Unbiased Predictors (BLUPs) of electrolyte leakage

(EL) at -10˚C and -20˚C, field winter damage (WD) and field regrowth (RG) in four environments (Elora 2019, Elora 2020, Sask 2019 and Sask 2020) in the male map ‘SITR’ .... 123

Figure 3-8. QTL mapping of Best Linear Unbiased Predictors (BLUPs) of electrolyte leakage

(EL) at -10˚C and -20˚C, field winter damage (WD) and field regrowth (RG) in one environment

(Elora 2020) in the female map ‘EDI’...... 124

Figure 3-9. QTL mapping of Best Linear Unbiased Predictors (BLUPs) of electrolyte leakage

(EL) at -10˚C, -20˚C and -35˚C and LT50, field winter damage (WD) and field regrowth (RG) in one environment (Elora 2020) in the male map ‘GV’...... 125

Figure 3-10 Distribution of field winter damage (WD) in the ‘CA60’ x ‘SITR’ mapping and among the parental and control genotypes population in four environments: ...... 127

xvii

Figure 3-12. Distribution of field regrowth (RG) in four environments (Elora 2019, Elora 2020,

Sask 2019 and Sask 2020) in A) the ‘CA60’ x ‘SITR’ population and B) among the parental and control genotypes...... 133

Figure 3-13. Distribution of of A) field winter damage (WD%) and B) regrowth (RG%) in the

‘EDI’ x ‘GV’ population and in the parental and control genotypes in Elora 2020...... 136

Figure 3-14. Relationship between Best Linear Unbiased Predictors (BLUPs) of electrolyte leakage and BLUPs of winter damage for A) the ‘CA60’ x ‘SITR’ population and B) the ‘EDI’ x

‘GV’ population...... 142

Figure 3-15. Climatic conditions in Saskatoon (SK) and Elora from June 2018 to July 2020. .. 149

Figure 4-1. Five roses with increased disease resistance, exceptional hardiness and stunning aesthetic features from the 'CA60' x 'Singing in the Rain' population...... 165

xviii

LIST OF ABBREVIATIONS BLUP Best Unbiased Linear Predictor

BLUE Best Unbiased Linear Estimate

BS Black spot

DREB dehydration-responsive element-binding protein

CBF C- repeat binding factor

EDI ‘Easy Does It’

GLMM General Linear Mixed Model

ICE Inducer of CBF expression

LMM Linear Mixed Model

LG Linkage group

LOD Logarithm of odds

LT50 Lethal Temperature for 50% of the plants

MAS Marker-assisted selection

QTL Quantitative Trait Loci

SITR ‘Singing in the Rain’

Sask Saskatoon

SNP Single Nucleotide Polymorphism

RG Regrowth

WD Winter damage

VRIC Vineland Research and Innovation Centre

xix

CHAPTER 1: INTRODUCTION AND LITERATURE REVIEW

1.1 Economical and Sociological Significance of Roses

Roses (Rosa) belong to the genus Rosa (L) and to the Rosoideae subfamily within the

Rosaceae family. The Rosaceae family also includes almonds, apples, apricots, cherries, sour cherries, peaches and nectarines, pears, plums, quinces, raspberries and strawberries and is an important part of human diet, health and well-being (Shulaev et al. 2008). It is the 19th largest plant family (Hummer and Janick 2009) and includes over 100 genera and 3,000 species (Judd

2008). In 2017, the Food and Agriculture Organization of the United Nations (FAO) estimated the worldwide production of edible fruits and nuts from this plant family at 165 million tonnes

(http://www.fao.org/faostat), with a value of approximately $65 billion US (Hummer and Janick

2009). In comparison, the North American rose industry was estimated to have a $1 billion US value in 2013 (Vineland 2013).

Roses are grown as garden and landscape plants, potted plants, cut flowers or rootstocks, and their blooms are collected for the extraction of essential oils. In the Middle East, they are also valued for their nutritious and medicinal fruits, called hips, which are high in antioxidants and vitamin C (Patel 2013). Today, roses are one of the most important ornamental crops, economically and symbolically (Bendahmane et al. 2013). However, although the rose market is important in North America, the sales of landscape roses have decreased by 30% in 20 years.

New varieties must be disease resistant, hardy, fragrant, and vigorous with an abundance of recurrent blooms in order to gain consumers’ enthusiasm (Waliczek, et al. 2015). Therefore, rose breeding programs in North America should be targeting those key features, with disease resistance being the first priority, followed by fragrance and cold hardiness. Moreover,

1 consumers’ preferences evolve continuously, and it is important to release new roses in a timely manner to maintain nursery sector competitiveness (Debener and Byrne 2014). While the implementation of efficient new technologies in rose breeding programs remains limited, marker-assisted selection (MAS) could greatly contribute to improving the accuracy of the selection and to expediting the release of new varieties.

1.2 Roses Origin, Classification and Phylogeny

Roses are native to the Northern hemisphere, and rose fossils were found in Colorado

(USA) dating as far back as 37 million years (Debener and Linde 2009). While south western

Asia and central Asia, in particular Turkey and China, are important centers of diversity of the genus Rosa (Debener and Linde 2009), roses are now cultivated worldwide (Zlesak 2006;

Debener and Linde 2009). Modern cultivars are mainly the result of hybridizations between

Chinese, European and Middle Eastern roses that occurred after the introduction of the Chinese rose in Europe by missionaries during the fourteenth century (Bendahmane et al. 2013). The domestication of roses was then long and complex, making rose taxonomy challenging to establish (Bendahmane et al. 2013).

The first insights into rose taxonomy date back to the beginning of the 19th century

(Tomljenovic and Pejić 2018). Roses belonging to horticultural classes that existed before 1867 are considered old garden roses. The introduction of the first Hybrid Tea 'La France' in 1867 by

Jean-Baptiste André Guillot (1827-1893) marked the beginning of a new class of intentionally bred roses known as modern roses, which includes Hybrid Tea, Polyantha and Floribunda roses

(Zlesak 2006). Hybrid Tea roses resulted from crosses between Tea roses and recurrent

2 bloomers. Their architecture, recurrent blooms and elongated buds distinguished them from the old roses. Polyantha roses were obtained in the 1860s from the cross between R. multiflora and

‘Old Blush China’. Polyantha roses are known for being hardy and recurrent blooming.

Floribundas resulted from the cross between Hybrid Teas and Polyanthas and showcase flower clusters and recurrent blooms (Barrau et al. 2006).

Alfred Rehder (1869-1949) established the foundations of the current rose taxonomy in

1940, based on a large number of morphological traits (Rehder 1940; Fofana et al. 2013;

Tomljenovic and Pejić 2018) and Wissemann updated Rehder’s classification in 2003. The genus

Rosa is commonly divided into four subgenera. The subgenus Rosa is itself divided into ten sections and one section, also named Rosa, contains most of the cultivated varieties (Debener and Linde 2009). Close to 200 rose species from the genus Rosa are recognized today, with 13 to

22 species in North America (Zlesak 2006).

Molecular techniques can bring new insights into the phylogenetic relationships between plant species and their evolution. Studies using chloroplast sequences, RAPD (random amplified polymorphic DNA) markers, FLP (amplified fragment length polymorphism) markers and

UPGMA (Unweighted pair group method with arithmetic mean) clustering proposed to revise the rose classification (Matsumoto et al. 1998; Jan et al. 1999; Bruneau et al. 2007; Koopman et al. 2008; Fougère-Danezan et al. 2015; Zhu et al. 2015). However, even with the use of new molecular technology, phylogenetic relationships within the genus Rosa have not yet been fully clarified (Tomljenovic and Pejić 2018).

1.3 Rose Biology and Genetics

Roses have a small genome of only 500Mb, which makes them practical for genomic studies. They have a short breeding cycle and produce progeny that can be easily propagated from cuttings (Debener and Linde 2009). In addition, roses are highly syntenic to woodland strawberries (Fragaria vesca) and genetically close to other crops from the Rosaceae family

(Debener and Linde 2009; Vukosavljev et al. 2016). For those reasons, the rose could represent a model plant for the development of genetic resources that could be broadly shared between species belonging to the Rosaceae family (Hibrand-Saint Oyant et al. 2007; Debener and Linde

2009).

Although roses have a small genome, which offers practical advantages to the researcher, rose genetics remains complex. Like all Rosoideae crops, roses have a basic chromosome number of seven (n=x=7). Yokoya and colleagues (2000) conducted the first extensive study on the ploidy of species belonging to each subgenera of the genus Rosa and each section of the subgenera Rosa (Error! Reference source not found.). The ploidy level of roses varies from d iploid (2n=14), to octoploid (2n=8x=56) (Yokoya et al. 2000; Roberts et al. 2009; Longhi et al.

2014). Wild roses are both diploid and polyploid. In fact, it has been reported that more than fifty percent of the wild species are polyploid (Bendahmane et al. 2013). Commercial cultivars are mainly tetraploid (2n=4x=28) but can also be triploid (2n=3x=21), combining the advantageous traits associated with both diploidy and tetraploidy (Zlesak, 2009; Zlesak et al. 2015).

This wide range of ploidy is the result of a vast history of hybridization and polyploidisation events and incomplete speciation (Smulders et al. 2019). While tetraploidy is

4 predominant among modern cultivars, most genetic and genomic studies on roses focus on diploid roses to overcome the difficulty arising from higher ploidy levels. However, the results published from research conducted on diploid roses are not directly translatable to tetraploid roses and advancement in rose genomics depends upon our understanding of rose meiosis at the polyploid level.

Table 1-1. Ploidy level of rose species belonging to the genus Rosa (Adapted from Yokoya et al. 2000)

Subgenera Section Ploidy

Hulthemia 2n=2x

Hesperhodos 2n=2x

Plathyrodon 2n=2x

Rosa

Pimpinellifolia 2n=2x

Rosa/Gallicanae 2n=4x

Caninae 2n=4x, 2n=5x and 2n=6x

Carolinae 2n=2x

Cinnamomeae 2n=4x, 2n=5x, 2n=6x and

2n=8x

Synstylae 2n=2x

Indicae 2n=2x

Banksianae 2n=2x

Laevigatae 2n=2x

Bracteatae 2n=2x

1.4 Rose Breeding

1.4.1 Overview

Roses are divided into four major horticultural groups that include cut roses, garden roses, pot roses and rootstocks (Vukosavljev et al. 2013). Each group belongs to different gene pools and is expected to possess different traits. For example, new varieties of cut roses are selected for the quality and the color of blooms and stems, yield and post-harvest quality, while roses used as rootstock, such as R. multiflora, need to be uniform, vigorous, winter hardy, disease and nematode resistant, with a fibrous root system and a broad graft compatibility (Cock

2007).

Garden and landscape roses are selected for specific plant architecture such as miniature, shrubs or climbers. They are expected to display several desirable aesthetic features (Table 1-2) and to be low maintenance due to high levels of disease resistance and cold hardiness. Modern breeding of garden roses is mainly achieved through the traditional method of manual hybridizations between highly heterozygous parental lines. Most of the selection is conducted on the F1 generation, after assessing the performance of the hybrids in the field for several seasons.

Large selection gains can be achieved quickly (Smulders et al. 2019) and the selected genotypes are fixed via vegetative propagation. Although the reproductive cycle of roses is short in comparison with other woody perennials, it takes between 8 and 10 years to commercialize a new variety. A rose breeder can face many challenges such as pollen sterility, poor hip set, poor seed germination, complex genetic inheritance arising from polyploidy and tedious assessment of hybrid performance in the field.

Table 1-2. Desired aesthetic features in garden roses and their genetic control

Trait Description Inheritance Gene of interest for breeding and implementation of MAS Fragrance Over 400 compounds contribute to rose scent and Polygenic The Nudix hydrolase RhNUDX1, a hydrolytic enzyme able to cleave nucleoside (Flament et al. 1993) belong mainly to three diphosphates, is associated with monoterpenes biosynthesis (Magnard et al. 2005). groups: terpenes (the most abundant), benzenoids/phenylpropanoids, and fatty acid derivatives (Bendahmane et al. 2013; Cherri- Martin et al. 2007). Flower Single flowers display five petals, while double Oligogenic One major locus on LG* 3 is involved in the control of the double flower phenotype morphogenesis flowers have over ten petals (Bendahmane et al. (Dubois et al. 2010). A 1,426bp transposable element insertion in intron eight of the 2013). rose APETALA2/TOE homologue (Rc3G0243000) would be associated with RhAG Floral development is under the control of four restriction expression linked to the double flower phenotype (Hibrand Saint-Oyant et al. classes of homeotic genes A, B, C and E. 2018). The C-function gene AGAMOUS (AG) regulates sexual organ identity (Bowman et al.1989; Two additional QTLs on LGs 2 and 5 are associated with the degree of flower Dennis and Peacock 2019). doubleness (Debener and Mattiesch 1999; Crespel et al. 2002; Koning-Boucoiran et al. 2012; Roman et al., 2015).

Flower colour Rose petals exhibit a wide range of colours due to Polygenic Three genes encoding flavonoid 3-glycosyltransferases, RhGT1, RhGT2 and RhGT3, the production of red anthocyanin and are involved in the production of anthocyanin pigments (Kitahara et al. 2001). carotenoids. Blue rose petals can only be achieved through The locus Bfla associated with pink flower colour mapped on LG2 (Debener and genetic transformation, because roses are not able Mattiesch 1999; Yan et al. 2005; Spiller et al. 2011). to produce delphinidin, which is at the origin of the blue colour (Katsumoto et al. 2007). Five genomic regions associated with the production of anthocyanin including and two genomic regions associated with the production of carotenoids were identified in GWAS** (Schultz et al., 2016).

Recurrent Most rose species bloom only once a year during Oligogenic RB is genetically controlled by a single locus mapped on LG3 and co-localized with blooming (RB) the spring. RB rose cultivars bloom continuously RoKSN, an homolog of the TERMINAL FLOWER 1 (TFL1) family (Iwata et al. 2012). from early spring to the first frost, and this The activation of TFL1 by long photoperiods after blooming represses the production of phenotype is highly desirable for modern roses further blooms (Koskela et al. 2012). The insertion of the transposable element Copia in (Bendahmane et al. 2013). RoKSN is a major determinant of recurrent blooming (i.e. allele RoKSNcopia) (Iwata et al. 2012), while the null allele RoKSNnull was also discovered in RB ‘Old Blush’ (Hibrand Saint-Oyant et al. 2018).

Two putative homologs of the transcription factors SPT and DOG1, which are respectively known to control flowering in Arabidopsis thaliana and to modify flowering by acting on miR156 would also be promising candidate genes in the determination of recurrent blooming in roses (Raymond et al. 2018).

* LG = Linkage group, ** GWAS = Genome Wide Association Studies

Rose breeding programs for food, perfume and medicinal use are slowly emerging. The

Dutch Rose breeding company Pheno Geno Roses based in Serbia introduced its collection of

‘Whole Edible roses’, presenting roses in many different colours and flavours. The petals of the cultivar ‘Renée van Wegberg’™ are described as sweet and fragrant, with a raspberry-like taste

(phenogenoroses.com).

1.4.2 Canada’s National Rose Breeding Program

Canada is home to a fruitful rose breeding program that was initiated decades ago.

William Saunders founded Canada’s national rose breeding program in Ottawa in the 1920’s and

Felicitas Svejda bred hardy roses in Ontario and Manitoba from the 1970’s to the mid 1990’s

(Ogilvie 1993). Roses from the Parkland, Explorer and Canadian Artist series were released through this program and are known worldwide for their exceptional winter hardiness. In 2010,

Canada’s national rose breeding program moved to Vineland Research and Innovation Centre

(VRIC) in Ontario’s Niagara Region.

1.4.3 Genetic Diversity

Roses have a large genetic diversity and close to 60,000 rose varieties have been released to date (Bendahmane et al. 2013). Several rose species brought key traits into modern roses through hybridizations. For example, R. gallica brought cold hardiness, R. foetida brought yellow colour and R. chinensis brought recurrent blooming (Koopman et al. 2008; Bendahmane et al. 2013). However, only a limited number of rose species have contributed so far to the gene pool of modern roses, thereby leaving a vast untapped genetic resource (Cock 2007) (Table 1-3).

Breeding strategies that involve a limited number of existing cultivars constrains the gene pool of modern roses (Bruneau et al. 2007). Increasing the genetic diversity in rose breeding programs is

9 necessary to prevent the accumulation of deleterious alleles resulting from inbreeding, but also to introduce new beneficial traits for commercial release.

The introgression of new traits into modern germplasm can be achieved via interspecific crosses. There are a number of unexploited rose species that would be of great interest for rose breeders (Table 1-4). However, exploiting exotic germplasm can be highly challenging due to ploidy levels, fertility and compatibility issues and linkage drag. Crosses between wild diploid roses and modern tetraploid roses usually produce sterile F1 triploid progeny, which becomes a bottleneck for breeders. A lack of reproductive organs or a disturbed meiosis that produces unbalanced gametes could explain the lack of fertility of the triploid progenies (personal communication, David Zlesak). Unsynchronized reception of stigma or stop in pollen tube growth, aborting embryos and aneuploidy, are all factors leading to unsuccessful interspecific crosses.

Table 1-3. Rose species that contributed to the genetic background of modern roses

Rose species Section Ploidy Major contributions R. chinensis Chinensis 2n=2x R. damascenae Gallicanae 2n=4x R. foetida Pimpinellifolia 2n=2x and 4x R. gallica Gallicanae 2n=4x R. gigantea Chinensis 2n=2x R. moschata Sysnstylae 2n=2x R. multiflora Synstylae 2n=2x R. wichuraina Synstylae 2n=2x Minor contributions R. rugosa Cinnamomeae 2n=2x R. cinnamomea Cinnamomeae 2n=2x R. pimpinellifolia Pimpinellifolia 2n=2x R. phoenica Synstylae 2n=2x R.sempervirens Synstylae 2n=2x R. arvensis Synstylae 2n=2x R.rubiginosa Caninae 2n=5x

Table 1-4. Suggestions of trait introgressions from rose species into modern germplasm

Rose species Section Ploidy Traits of interest R. banksiae Banksianae 2n=2x Vigour, Dense and evergreen foliage, Thornless stems Resistance to most insects and pathogens R. spinossissima Pimpinellifolia 2n=4x Cold hardiness, Disease and drought resistance R. rubiginosa Caninae 2n=5x Apple-scent foliage R. bractea Bracteatae 2n=2x Increased vigour

The success of the introgression of a trait from wild to modern germplasm relies on fertile F1 progenies and on the negative selection of undesirable traits inherited from the wild species. Various methods can be employed to make interspecific crosses successful; those methods include manipulating the temperature during hybridizations (Bendahmane et al. 2013), exploiting triploid hybrids as a breeding bridge in the formation of tetraploids (Gallais, 2003), chromosome doubling (i.e. amphidiploidy) using colchicine or Trifuralin (Feng et al. 2017;

Zlesak et al. 2005), pollen sorting (Van Huylenbroeck et al. 2007), early embryo rescue

(Mohapatra and Rout 2005) and protoplast fusion (Pati et al. 2008). Equally important, increased knowledge of rose phylogeny can provide guidance in adopting the most efficient breeding strategies that effectively use the unexploited rose germplasm (Cock 2007; Koopman et al. 2008,

Tomljenovic and Pejić 2018), thereby making a practical contribution to breeding schemes, as long as this knowledge and associated technologies, such as markers and gene bank, are directly accessible to the breeder (Smulders et al. 2019).

1.5 Advances in Applied Genomics and Opportunities for Rose Breeding

1.5.1 Genetic Maps as a Framework for the Identification of Marker-Trait Association

Several maps for diploid roses have been published during the last twenty years with the ultimate goal of identifying quantitative trait loci (QTL) and unraveling the genetic control of major traits. For example, the first Random Amplified Polymorphic DNA (RAPD) and

Amplified fragment length polymorphism (AFLP)-based map for diploid roses, derived from R. multiflora, was used to map petal number and flower colour (Debener and Mattiesh, 1999). This map was further improved with additional plants and marker types (Debener et al. 2001; Yan et al. 2005). The second map created for diploid roses was derived from R. wichurana and AFLP

12 markers and was used as a framework for the linkage mapping of number of prickles, double corolla and recurrent blooming (Crespel et al. 2002). Maps derived from diploid roses were also created by Dugo and colleagues using RAPD, morphological markers and Single Sequence

Repeat (SSR) markers (Dugo et al. 2005). Spiller and collaborators (2011) published the first integrated consensus map for roses using four diploid populations and more than one thousand initial markers. This map is still used as a reference to identify and name rose linkage groups and includes 10 phenotypic single loci, QTL for seven traits and 51 Expressed Sequence Tags (EST) or gene-based molecular markers (Spiller et al. 2011). The maps constructed for diploid roses represented a major step toward the discovery of important genes for ornamental breeding and a framework for linkage mapping. Nevertheless, the extent of the knowledge obtained from those maps that is applicable to modern roses is limited due to differences in ploidy level and the specificity of meiosis of tetraploids.

Tetraploid species have an increased number of alleles at one locus compared to diploid species and five different genotypes are possible at one bi-allelic (A, a) locus: nulliplex (aaaa), simplex (Aaaa), duplex (AAaa), triplex (AAAa) and quadruplex (AAAA). While autotetraploids possess four times the basic number of chromosomes from the duplication of one genome, allotetraploid possess two diploid genomes (Galais, 2003). There is a high degree of preferential pairing during the meiosis of allopolyploids, and as a result, only two homologous chromosomes pair together, leading exclusively to the formation of bivalents and disomic inheritance. In contrast, in autopolyploids, the four homologous chromosomes can form either random bivalents or tetravalents, for which recombinations can occur between the eight chromatids. Moreover, double reduction is specific to autopolyploids and can occur after the formation of tetravalents

13 during a meiosis called “pseudo-equational”, where both sister chromatids migrate to the same pole, resulting in the formation of gametes with multiple copies of the same genes and in inbreeding (Galais, 2003). Under tetrasomic inheritance and the formation of random bivalents at meiosis, the genotypic segregation of a bi-allelic locus in a tetraploid bi-parental population reveals the allelic constitution of the gametes contributed by the parents. For instance, under this genetic model and no double reduction, the cross between two parents Aaaa and Aaaa at a given locus would lead to the segregation ratio 1/4 AAaa: 1/2 Aaaa: 1/4 aaaa within the progeny, and a phenotypic segregation of 3:1 for the corresponding trait, with A being the dominant allele (Table

1-5). However, the examination of the genotypic segregation from all possible allelic states of the parents of the cross is complex. As a result, the creation of genetic maps under tetrasomic inheritance, is more challenging compared to disomic inheritance since it is more difficult to interpret the genetic segregation in the progeny (Ma et al. 2002). The type of meiosis found in tetraploid roses is unique and its understanding comes with an even greater challenge since roses are neither fully allopolyploid nor fully autopolyploid but rather segmental allopolyploid

(Debener and Linde 2009; Bourke et al. 2017). In fact, during meiosis, each of the seven rose chromosomes have different degrees of preferential pairing, and rose populations can display either tetrasomic inheritance without preferential pairing (Vukosavljev et al. 2016; Zurn et al.

2020) or a mixture of both tetrasomic and disomic inheritance (Koning-Boucoiran et al., 2012;

Bourke et al., 2017).

Genotype Terminology

AAAA Monogenic quardriplex

AAAa Digenic triplex

AAaa Digenic duplex

Aaaa Digenic simplex

aaaa Monogenic nulliplex

Allelic state of the Genotypic segregations in tetraploids under tetrasomic Phenotypic segregation

parents of the cross inheritance (‘A’ is the dominant

allele, ‘a’ is the recessive

allele)

AAAA x AAAa ½ AAAA : ½ AAAa 100% A

AAAa x AAAa ¼ AAAA : ½ AAAa : ¼ AAaa 100% A

AAAa x AAaa 3/36 AAAA : 15/36 AAAa : 15/36 AAaa : 3/36 Aaaa 100% A

AAAa x Aaaa ¼ AAAa : ½ AAaa : ¼ Aaaa 100% A

AAaa x AAaa 1/36 AAAA : 2/9 AAAa : 2/3 AAaa : 2/9 Aaaa : 1/36 aaaa 35:1

AAAa x aaaa ½ AAaa : ½ Aaaa 100% A

AAaa x Aaaa 1/12 AAAa : 5/12 AAaa : 5/12 Aaaa : 1/12 aaaa 11:1

AAaa x aaaa 1/6 AAaa : 2/3 Aaaa : 1/6 aaaa 5:1

Aaaa x Aaaa ¼ AAaa : ½ Aaaa : ¼ aaaa 3:1

Aaaa x aaaa ½ Aaaa : ½ aaaa 1:1

The first map for tetraploid roses was published in 2001 (Rajapakse et al. 2001) and was further improved using SSR and EST SSR markers (Hibrand-Saint Oyant et al. 2007). An autotetraploid linkage map of roses was also published by Gar and colleagues in 2011 using sequence-based markers, AFLP and morphological markers. In 2012, Koning-Boucoiran and colleagues studied the inheritance of traits in tetraploid cut roses and used markers from literature to build the third tetraploid rose linkage map. Today, Single Nucleotide Polymorphism

(SNP) markers represent the most effective marker choice for genetic research, including linkage mapping (Rafalski 2002). SNP are single base differences in DNA sequences and are the most common type of genetic variation in genomes. Their detection can be automated and high throughput (Rafalski 2002). SNP array platforms are developed from SNPs discovered by sequence alignment to a reference genome and they are available for Rosaceae crops including apple (Chagné et al. 2012; Bianco et al. 2014, 2016) and peach (Verde et al. 2012), allopolyploid sour cherry (Peace et al. 2012), allo-octoploid strawberry (Bassil et al. 2015) and tetraploid roses

(Koning-Boucoiran et al. 2015). The WagRhSNP Axiom rose SNP Array includes 68,893 SNPs and can be broadly utilized because it was developed from Expressed Sequence Tags (ESTs) from diverse germplasm including tetraploid cut roses and both diploid and tetraploid garden roses (Koning-Boucoiran et al. 2015). The use of SNP arrays for linkage mapping in autotetraploid species makes the estimation of allele dosage possible, thereby allowing the construction of high-density linkage maps from markers in various configurations.

Several methods have been developed to infer allele dosage from an array and exploit marker information. For example, Hackett and colleagues (2013) inferred SNP dosage from allele intensity ratios using normal mixture models to map increased numbers of markers in various configurations in a tetraploid potato mapping population. Vukosavljev and colleagues

(2016) constructed a genetic map for tetraploid roses using the 68K WagRhSNP Axiom array and determined SNP allele dosage with software FitTetra-FitPoly (Vukosavljev et al. 2016; Zych et al. 2019). Hibrand Saint-Oyant and colleagues (2018) generated genetic maps for three populations of roses including a tetraploid population using the 68K WagRhSNP Axiom array with the purpose of evaluating the quality of the whole genome assembly of ‘Old Blush’. More recently, Zurn and colleagues (2020) created a consensus map from two parental maps of

‘Modern Blush’ and ‘George Vancouver’ again using the 68K WagRhSNP and the software

Polymap R in order to map black spot resistance. Even though those examples demonstrate the successful implementation of a SNP array in tetraploid crops, it could also be said that the array is not always accessible due to its high cost. In addition, marker discovery in a population will be limited to the markers present on the array.

Considering the limitations arrays due to cost and coverage, genotyping-by-sequencing

(GBS) appears to be a competitive Next Generation Sequencing (NGS) procedure to develop

SNP markers for mapping. GBS protocols reduce genome complexity thanks to the use of restriction enzymes and provide an efficient barcoding system (Elshire et al. 2011). GBS sequences many individuals at the same loci and generates thousands of SNP markers (He et al.

2014; Li et al. 2014; Elshire et al. 2011). This technology has proved itself to be an affordable key technology to assist with the creation of high-density genetic maps in tetraploid crops. Li et

17 al. (2014) used GBS data to build genetic maps for autotetraploid alfalfa (Li et al. 2014). The authors followed a two-way pseudo-testcross mapping strategy in which the single dose markers are heterozygous in one parent and homozygous in the other (Aaaa x aaaa or aaaa x Aaaaa), thereby overcoming the complexity of the genetic inheritance among tetraploid species due to the presence of markers in various configurations, and producing one genetic map for each parent of the cross (He et al. 2014; Grattapaglia and Sederoff 1994). Under this scenario, software designed for diploid species such as JoinMap or ASMAP package on R (van Ooijen et al. 2006; Taylor and Butler 2017) can be used, because the estimates for the markers, which show the segregation pattern simplex x nulliplex, are identical to those of coupling phase markers in a diploid species (Brouwer and Osborn 1999; Koning-Boucoiran et al. 2012).

1.5.2 Marker-Assisted Selection : Challenges and Opportunities for Ornamental Crops

Genomic tools and the accumulation of genomic resources offer new possibilities for functional genomic research and greatly contribute to increasing our knowledge on the genetic control of key traits for breeding. Research on EST aim to identify, characterize and make available important genes such as the ones involved in the rose scent biosynthesis pathway or flowering behaviour (Channelière et al. 2002; Guterman et al. 2002). Koning-Boucoiran and colleagues (2015) provided the most comprehensive rose transcriptome resource with 13,390 identified full-length genes (Koning-Boucoiran et al. 2015). In addition, a well-annotated whole genome reference sequence is now available for Rosa (R. chinensis Old Blush Homozygous

Genome v2.0 (Raymond et al. 2018)). When combined with high throughput sequencing methods and bioinformatics analysis, the use of the whole reference genome allows the investigation of candidate genes and the development of gene sequence-based molecular markers. Marker-Assisted Selection (MAS) relies on strong associations between molecular

18 markers and candidate genes to move forward the selection of elite varieties. Therefore, it is expected that the development of genomic ressources would facilitate the implementation of

MAS.

The value of MAS has been demonstrated in various ornamental crops to address breeding challenges. MAS was successfully applied in carnation breeding to select for resistance to carnation bacterial wilt (Yagi 2015; Yagi 2018), and in roses to assess shoot organogenesis

(Nguyen et al. 2019). MAS strategies were also developed to design optimal interspecific breeding schemes in Hydrangea (Granados Mendoza et al. 2013) and to introgress traits from wild rose species into modern cultivars (Debener 2003). Despite these success stories and the fact that traditional breeding strategies would greatly benefit from MAS by reducing the costs and the duration from breeding to commercial release and improving the accuracy of the selection (Rout and Mohapatra 2006; Smulders et al. 2019), overall the implementation of MAS has been slow for ornamental crops. This is mainly due to the lack of accessible technologies and the complex genetic nature of ornamentals with higher ploidy and high heterozygosity, which complicate genetic studies in comparison with diploid species (Rout and Mohapatra 2006; Yagi

2018). More specifically, regardless of the advances in fundamental research, the adoption of

MAS in tetraploid rose breeding programs remains limited (Koning-Boucoiran et al., 2012;

Debener and Byrne, 2014, Smulders et al. 2019), indicating that the findings are not directly accessible to rose breeders and that the transition toward the implementation of MAS in the rose sector is not sufficiently supported (Smulders et al. 2019).

1.5.3 Genome Database for Rosaceae and RosBREED

The Genome Database for Rosaceae (GDR) is the main repository for genomic data of

Rosaceae crops, and it represents a major step toward the successful implementation of MAS in roses because it aims to make fundamental research directly accessible to breeders and useful for a breeding program. The GDR is an online platform with an user-friendly embedded engine search that provides a straightforward access to all the genetic resources gathered to date from

Rosaceae crops (www.rosaceae.org). More specifically, information on roses include genomes, lists of markers and genetic maps, SNP arrays, quantitative trait loci and transcripts. The GDR is home to the genomic resources contributed by RosBREED.

RosBREED is a Rosaceae crop breeding consortium designed to develop and implement genomic tools in U.S. Rosaceae crop breeding programs to improve the efficacy and effectiveness of germplasm development with a focus on increased disease resistance and superior horticultural traits (Iezzoni et al. 2020). More specifically, RosBREED’ objectives for roses were to enable DNA-informed breeding for black spot disease resistance (Iezzoni et al.

2020).

1.6 Black Spot Disease

1.6.1 Biology, Disease Cycle and Ecology of D. rosae

Roses grown in the field are challenged by several pests and pathogens, but black spot is the most devastating foliar disease (von Malek and Debener 1998; de Vries and Dubois 2001;

Hattendorf et al. 2004, Whitaker et al. 2010). Diplocarpon rosae Wolf. is the teleomorph stage of the pathogen of black spot disease in roses. It is an ascomycete fungus obligate to the genus Rosa

(von Malek and Debener 1998; Blechert and Debener 2005). Specifically, Marssonina rosae is the imperfect, or asexual, stage of D. rosae (Carlson-Nilsson and Davidson 2000). While some scientists suggest that the sexual stage of D. rosae rarely occurs (Lühmann et al. 2010), some believe, to the contrary, that the genotypic and phenotypic diversity among North American isolates implies that the rate of sexual reproduction of D. rosae might be higher than reported

(Whitaker et al. 2007).

The typical symptoms of black spot are clearly distinguishable and can be described as tiny brown spots growing close to each other on the rose leaflet, surrounded by chlorotic yellow halos. The diseased leaves turn yellow and drop (Gachomo et al. 2006). The plant could be entirely defoliated by a severe infection and no longer able to perform photosynthesis. This causes the vigour of the plant to decline, leading to its eventual death (Figure 1-1). Diplocarpon rosae Wolf. overwinters in fallen leaves and in lesions on infected stems. The nonmotile and microscopic asexual spores, also known as conidia, produced in diseased tissue are splashed by water from infected leaves to the opening leaves in the spring and are the source of primary inoculum.

Figure 1-1. Typical black spot symptoms on an infected and almost entirely defoliated rose bush (Rosa x hybrida) (Credit Rouet C 2019, Vineland Research and Innovation Centre)

Figure 1-2. Polycyclic disease cycle of Diplocarpon rosae Wolf

The germination of the conidia on the surface of the leaves occurs in optimal conditions after the foliage has remained wet for over seven hours with atmospheric humidity above 95%.

The conidia can produce one or more germ tubes that penetrate directly the leaf tissue and grow over the plant cells or via an appressorium that pushes into the plant cuticle via a penetration peg

(Gachomo 2005). The penetration into the host might be facilitated by the secretion of extracellular fungal enzymes that contribute to modifying the properties of the host cell wall and cell membrane (Gachomo 2005). The disease progresses rapidly, following a logarithmic pattern of development (Carlson-Nilsson 2000). After the penetration of the cuticle, hyphae grow radially from the infection site in the intercellular space and invade the epidermal cells to form haustoria. Acervuli are formed at the intersection of the hyphae and rupture the cuticle as they release new conidia as a white viscous mass; the secondary infection occurs in a polycyclic disease cycle (Figure 1-2). In the development of hemibiotrophic D. rosae, the biotrophic phase is associated with the formation of haustoria that absorb nutrients and water from the host cell and the necrotrophic phase is associated with the development and thickening of the intracellular hyphae, host cell wall dissolution and potential secretion of cell wall-degrading enzymes

(Gachomo and Kotchoni 2007).

1.6.2 Interaction between the Host and the Pathogen

The infection of the plant host by D. rosae induces the accumulation of Reactive Oxygen

Species (ROS) in plant tissue, which are involved in the recognition of PAMP (pathogen associated molecular patterns) by the membrane receptors in the PAMP-Triggered Immunity

(PTI) and in phytohormones crosstalk. Genes from the phenylpropanoid and flavonoid pathways become upregulated in the plant host, as well as genetic factors involved in the downstream

23 signalling of plant defense response, including mitogen activated protein kinases (MAP kinases),

Ca2+ and hormone signalling factors (salicylic acid, jasmonic acid and ethylene), Pathogenesis

Related proteins (PR) and WRKY transcription factors, associated with the Effector-Triggered

Immunity (ETI) (Gachomo and Kotchoni, 2010; Neu et al. 2017). In the case of an incompatible reaction with the pathogen, the ETI-mediated response leads to an Hypersensitive Response

(HR), or programmed cell death, since the pathogen’s effector is recognized by plant receptors encoded by resistance genes (R-genes), and to the production of salicylic acid (SA) inducing

Systemic Acquired Resistance (SAR) and the production of PR .

The publication of the first draft genome of D. rosae, contributes to deepening the understanding of the interaction between D. rosae and Rosa. Notably, recent research reveals that among over 14,000 genes predicted in D. rosae, close to 90% are expressed during the early stage of the infection (Neu et al. 2017), and that most of the D. rosae secretome qualifies as virulence factors and contains hydrolases, cellulases and highly expressed pectin-degrading enzymes that are involved in the degradation of components of the host cell wall and further penetration into the host (Neu and Debener 2019). An improved understanding of the interaction of the D. rosae/Rosa pathosystem would help with identifying breeding targets.

1.6.3 Disease Management

Black spot is managed through different practices. The primary source of inoculum can be controlled via soil fumigation. Synthetic chemicals, such as copper oxychloride, tiophenate methyl and cardenbazim, are used to contain the spread of the disease (Yasin and Ahmed 2016).

Black spot disease is usually controlled in the field by preventive fungicide applications every seven to fourteen days when the weather allows the pathogen to develop, resulting in at least 20

24 fungicide applications per year (Debener and Byrne 2014). However, while black spot can be treated with chemicals, they are expensive and unsafe for the environment and their use should be limited (Gachomo 2005). Pest resistance to chemicals, economical pressure, the environmental awareness and changing legislation about pesticide use have led us to require new and efficient disease management strategies that combine cultural practices and minimal fungicide applications.

D. rosae primary inoculum and rate of the disease progression can be controlled through physical practices such as pruning of diseased stems and collection of diseased and fallen leaves.

Non-fungicidal solutions such as sodium bicarbonate or oil are a safer alternative to fungicides

(Horst and Cloyd 2007). Preventive application of ROS (H2O2) on the rose leaves could also be part of an environmentally-friendly black spot disease management (Gachomo and Kotchoni

2010). In addition, rhizosphere bacteria can have a role in activating plant Induced Systemic

Resistance (ISR) to pathogens in many crops including roses (Yasin and Ahmed 2016). Yasin et al. (2016) demonstrated that bacterial strains of Pseudomonas fluorescens and Bacillus subtilis reduced the disease severity by more than 60% compared to the control treatment by inducing the production of chemicals in the plant such as peroxidases, phenolic compounds, ascorbic acid or soluble proteins leading to higher levels of plant’s defense. Therefore, bacterial strains could be included in black spot disease management practices in an Integrated Pest Management (IPM) system. However, the only truly effective way to control black spot to date is through planting of resistant varieties.

1.6.4 Diversity of Diplocarpon rosae and Race Characterization

The existence of multiple pathogenic races of D. rosae in the landscape challenges breeders to release low maintenance roses. In addition, a large number of isolates of D. rosae that are genetically diverse can represent one unique pathogenic race due to the presence of identical avirulence genes (Whitaker et al. 2007). Several research projects have been undertaken to address the challenge of characterizing pathogenic races of D. rosae. The various authors often used different sets of plant hosts and adopted different nomenclature systems. Whitaker and collaborators (2010) conducted the first extensive study on D. rosae race characterization.

The authors studied the unique host range of 15 isolates using the first standard set of differential genotypes composed of nine rose cultivars that can differentiate all known races. As a result, eleven unique D. rosae races are described to date and named according to a newly adopted international nomenclature system (Whitaker et al. 2010). The existence of both a 12th and a 13th race was suggested (Whitaker et al. 2010; Zurn et al. 2018). However, the ability of the standard set of differential genotypes to capture the genetic diversity of the fungus is questioned and it is possible that the number of existing D. rosae races is larger than reported (Menz et al 2018).

Nonetheless, the continued identification and characterization of new races would improve the accuracy of the phenotyping and help identifying R genes (Whitaker et al. 2007). In addition, the creation of an international black spot isolates collection was supported by the Universities of

Wisconsin and Minnesota for safekeeping and easy access to all races worldwide, thereby encouraging efforts toward a global collaboration to fight black spot disease in roses (personal communication, D. Zlesak).

1.6.5 Breeding for Resistance

Different levels of resistance to black spot exist among rose cultivars, but overall, there exists a lack of resistance in modern rose cultivars (Debener and Mattiesch 1999). In comparison, the resistance is thought to be widespread in the old rose species (Palmer et al.

1966a and b). Releasing black spot resistant and low maintenance rose cultivars is challenging due to the complex genetic nature of roses (Gachomo, 2005; Debener, 2019) and the labour, cost and experimental error associated with field screening. Phenotyping for black spot resistance requires multiple years and multiple locations trials and can still be inaccurate due to the lack of disease pressure and spatial heterogeneity of fungal races.

Genetic control of black spot resistance in roses follows a gene-for-gene type of interaction (Neu and Debener 2019). To date, four major resistance (R) genes have been reported

(Rdr1, Rdr2, Rdr3, Rdr4) (von Malek and Debener 1998; Hattendorf et al. 2004; Whitaker et al.

2010; Zurn et al. 2018; Zurn et al. 2020). Rdr1 (muRdr1), was introgressed from the diploid hybrid R. multiflora 88/124-46 (von Malek and Debener 1998), and is located on linkage group 1 of the integrated consensus rose map (Kaufmann et al. 2003; Biber et al. 2010; Terefe-Ayana et al. 2011; Spiller et al. 2011). Rdr1 is a complex locus, conferring resistance to multiple D. rosae races and consisting of a cluster of nine TIR-NBS-LRR (Toll/interleukin-1 receptor-nucleotide binding site-leucine rich repeat; TNL) genes (Terefe-Ayana et al. 2011; Menz et al. 2018). Rdr2 is located on chromosome 1 and is tightly linked to Rdr1, but has not been characterized

(Hattendorf et al. 2004). Rdr3 was the first R gene for black spot resistance to be discovered in tetraploid roses (Whitaker et al. 2010) and was mapped to chromosome 6 (Zurn et al. 2020).

Rdr4 was also discovered in tetraploid roses, has been mapped to the long arm of chromosome 5, and like Rdr1, confers resistance to multiple black spot races (Zurn et al. 2018).

Quantitative resistance to black spot exists and recent research investigated the presence of a QTL associated with it in genetic backgrounds different from R. multiflora. In fact, two major resistance QTLs were mapped on linkage groups 3 and 5 of the rose genome exploiting mapping populations derived from black spot resistant R. wichurana (Soufflet et al. 2019). A major QTL explaining 13% of the total phenotypic variation was also mapped on linkage groups

3 in a mapping project using fifteen diploid rose populations derived from ‘Basye’s Thornless’

(R. carolina hybrid x hybrid Perpetual) (Yan et al. 2019). Yet, due to its monogenic nature and its high expression at the phenotypic level, vertical resistance is often more easily selected than horizontal resistance. However, the former could be quickly overcome by new biotypes (von

Malek and Debener 1998; Xue and Davidson 1998; Zlesak 2006). To address the challenge of breeding for durable vertical resistance, breeders could aim to pyramid multiple race-specific resistance genes into elite genotypes via MAS (Marone et al. 2013). The existence of several multi-races resistance genes, such as Rdr1 and Rdr5, represent an opportunity for pyramiding various sources of black spot resistance (Debener, 2019). However, most available markers associated with black spot resistance genes are germplasm-specific and cannot consistently predict the phenotype of resistance (Debener and Byrne, 2014). Zurn and colleagues (2020) proposed to exploit full haplotype information to increase the accuracy of the diagnostic. The authors identified three SSR marker alleles associated with the Rdr3 gene and the markers were used in a multiplexed reaction to capture the Rdr3 haplotype. In this study, among 63 cultivars tested, 12 showed black spot resistance but did not possess the Rdr3 diagnostic haplotype, suggesting the presence of additional R genes (Zurn et al. 2020). It remains that additional efforts are needed to identify and develop practical markers associated with various sources of black spot resistance across wide collections of germplasm.

1.6.6 Genetically Transformed Roses for Black Spot Resistance

Breeding pest and disease free roses is crucial for the nursery sector and conventional breeding for black spot resistant roses has proved to be difficult. Consequently, some attempts were conducted to create black spot disease resistant transgenic plants using different protocols, leading to successful and promising results (Debener and Byrne 2014). Genetically transformed black spot susceptible cultivars expressing a chitinase gene from rice, which was introgressed by particle bombardment, showed a 13 to 43% decrease in disease severity (Marchant et al. 1998).

Genetically transformed black spot susceptible cultivars expressing a ribosome from barley, which was introgressed via Agrobacterium tumefaciens, showed a 60% decrease in disease severity (Dohm et al. 2002). However, the efficiency of transgenic transformation in roses is estimated at 3% only, which remains too low to be an effective technology (Dohm et al. 2002).

CRISPR technology might be of great interest for rose breeding in the future (Smulders et al.

2019).

1.7 Cold Hardiness

Plants have developed elaborate strategies in order to survive and thrive under extreme environmental conditions. Plant abiotic stresses such as cold, salinity and extreme heat reduce the distribution of species in certain geographic areas (Mahajan and Tuteja 2005). The lack of cold hardy modern roses restricts the cultivation of landscape roses in regions with temperate climate and particularly harsh winters; therefore, winter hardiness is a necessary trait to extend the geographical scope of this species (Svejda 1979). Winter hardiness has been described as the ability of a plant to adapt its metabolism as a response to suboptimal temperatures, survive low temperatures and regrow during the spring (Vukosavljev 2014). Freezing tolerance is the main component of winter hardiness, and it has been described as the ability of a plant to survive subfreezing temperatures (Li and Sakai 1978) without irreversible damage to the cells due to ice formation in extracellular tissues (Vukosavljev 2014).

1.7.1 Physiology of Cold Hardiness

1.7.1.1 Plant Strategies to Survive Low Temperatures during the Winter

Cold stress can be divided into two main types: chilling stress and freezing stress. The former occurs at temperatures between 0°C and 15°C and affects photosynthesis, membrane fluidity, energy metabolism, ROS and homeostasis. The later occurs at temperatures below 0°C and results in the formation of ice within the cells, leading to irreversible damage (Shi et al.

2018). Plants surviving low temperatures are divided into three categories based on their level of cold hardiness. They are chill susceptible if they do not withstand temperatures below 12°C, chill tolerant if they can withstand low temperatures but cannot survive a freezing event, or freeze tolerant if they are able to withstand sub-zero temperatures and survive potential frost damage

30 thanks to two distinct strategies known as tolerance and avoidance (Pearce 1999; Wisniewski et al. 2018). The tolerance strategy involves major transcriptomic changes that result in biochemical alterations, which allow the plant to tolerate the presence of ice in the extracellular space. More specifically, the osmotic pressure increases with the formation of extracellular ice, pushing the water out of the cell, which decreases the freezing point of the cytoplasm. However, in the process, the dehydration of the cells can cause indirect damage to the plants. Conversely, the avoidance strategy prevents the freezing of sensitive tissues by physical attributes that allow pockets of water to remain undercooled down to -40°C. Deep supercooling is a complex mechanism of avoidance from woody perennials from temperate regions that prevents damage due to cell dehydration and protects xylem and dormant buds from freezing injury by suppressing ice formation in specific tissues. In this strategy, the water is not lost toward the sites of extracellular ice, but remains inside the cell in a metastable condition prone to flashing freezing events (Wisniewski et al. 2004; Wisniewski et al. 2014; Wisniewski et al. 2018).

1.7.1.2 Acclimation and Deacclimation

Winter hardiness is a complex process that can be divided into three stages: acclimation, period of freezing tolerance, and deacclimation. The acquisition of freezing tolerance depends upon a successful cold acclimation in the fall, during which the exposure to decreasing temperatures induces a cascade of biochemical, physiological and metabolic changes in plant tissues (Smallwood and Bowles 2002; Wisniewski et al. 2003; Vukosavljev 2014; Panjtandoust and Wolyn 2016). For instance, cold-induced metabolic changes in roses include a decrease in starch content, an increase in oligosaccharides and an increase in sucrose content in the cells

(Ouyang et al. 2019).

Low temperatures and photoperiod are considered to be the first stimuli of acclimation; however, other factors such as high winds or water stress can initiate acclimation and result in certain levels of freezing tolerance. (Kurepin et al. 2013). Acclimation can be divided into two main phases. The first stage of acclimation is induced by short days and decreased photoperiod, leads to the cessation of plant growth, and is accompanied by the initiation of metabolic changes

(Weiser 1970). During the second phase of acclimation, modifications in protein synthesis and configuration, membrane properties and concentration of hormones such as ABA (Abscisic acid), lead to the upregulation of multiple genes involved in cold hardiness that have cryoprotective effects, such as cold regulated genes (COR-genes) (Weiser 1970). The temperature at which the acclimation is induced and the magnitude of the changes that the plant undergoes during the process are genotype-specific (Săulescu and Braun 2001; Kim and Wolyn

2015). For example, cold hardy roses acclimate faster after only one freezing event and are less prone to dehydration caused by the formation of extracellular ice, compared to less cold hardy roses, which acclimate slower and require multiple freezing episodes before reaching their maximum hardiness level (Ouyang et al. 2019).

Cold acclimation in the fall is critical for winter survival; yet, the maintenance of dormancy is also an essential component for plant survivability (Panjtandoust and Wolyn, 2016).

Deacclimation in woody perennials usually occurs in response to warming temperatures during the spring and timing of deacclimation is critical as premature deacclimation results in irreversible damage in the xylem and the buds (Wisniewski et al. 2018). The temperature at which the deacclimation is induced is genotype-specific (Wisniewski et al. 2018) and hardy plants deacclimate more slowly than cold susceptible genotypes (Săulescu and Braun 2001).

1.7.2 Genetic Control of Cold Hardiness: the CBF Pathway

The genetic control of cold tolerance is polygenic and the metabolic pathway leading to hardiness is highly complex. In wheat, up to twenty-one chromosomes carry genetic factors involved in cold tolerance (Săulescu and Braun 2001). Several cold-upregulated transcription factors, genes and miRNAs have been identified and are also known to be involved in other stresses such as drought and salinity. The expression of genes involved in cold response is regulated via ABA-dependent and non-dependent pathways. Although many genetic factors are involved in the acquisition of cold tolerance, some cold-responsive genes are thought to have greater effects than others (Pearce, 1999). For instance, the C-repeat binding factors

(CBF), also known as Dehydration-Responsive Element Binding factors (DREB1), are key regulators of acclimation and response to cold stress in an ABA-independent pathway

(Novillo et al. 2007; Barros et al. 2012; Wisniewski et al. 2018). In Arabidopsis, three cold- induced CBF/DREB1 genes were identified (CBF1/DREB1B, CBF2/DREB1C and

CBF3/DREB1A) (Gilmour et al. 1998) and the expression of those genes is greater in cold hardy genotypes compared to freezing sensitive genotypes (Gery et al. 2011). The CBF genes are conserved across distant species including woody perennials (Shi et al. 2018). Although the information on the CBF regulation in woody plants remains limited, CBF were also identified in important Rosaceae crops (Wisniewski et al. 2014; Wisniewski et al. 2018). Five

CBF genes were reported in peach (Prunus persica) (Wisniewski et al. 2011), three in sweet cherry (Prunus avium) (Kitashiba et al. 2004), two in almond (Prunus dulcis) (Barros et al.

2012) and six in Japanese plum (Prunus mume) (Zhao et al. 2018). Functional characterization via genetic transformation of cold sensitive species confirmed that the overexpression of the CBFs from Rosaceae crops is associated with increased levels of cold tolerance (Kitashiba et al. 2004; Wisniewski et al. 2011; Artlip et al. 2014).

The transcription factors CBF/DREB1 recognize a conserved motif CCGAC (aka

CRT/DRE cis-element) present in the promoter of several COld-Regulated (COR) genes, leading to the activation of their expression (Shi et al. 2018). While the CBF genes expression increases within fifteen minutes after the perception of the low temperature signal (Gery et al.

2011), the expression of the COR-genes is induced several hours after (Kurepin et al. 2013).

This cascade of modifications in gene expression via CBF/DREB1 TFs is known as the CBF pathway (Figure 1-3). The expression of CBF is itself regulated, and in the absence of cold stress, the expression of CBF is downregulated to maintain plant fitness and normal plant growth (Wisniewski et al. 2014).

The COR-genes were first described by Guy and collaborators in 1985 (Guy et al.

1985). They include low-temperature induced (LTI), responsive to desiccation (RD), and early dehydration-inducible (ERD) genes. They code for antioxidants, osmolytes and membrane proteins (Mahajan and Tuteja, 2005). The COR-genes also encode dehydrin proteins, which are a sub-group of the late embryogenesis abundant (Group II LEA) proteins, that are also induced by drought, heat, embryogenesis, and plant hormones such as ABA, salicylic acid and methyl jasmonate (Banerjee and Roychoudhury 2016; Rihan et al. 2017).

Although little is known about their precise functions, cold-regulated dehydrins are thought to have cryoprotective and antifreeze properties, to maintain the integrity of the membrane and to protect the cell from dehydration (Close 1997; Wisniewski et al. 1999; Banerjee and

Roychoudhury 2016; Haimi et al. 2017). The existence of cold responsive dehydrins has been confirmed in several Rosaceae crops, such as apple (Malus x domestica) (Haimi et al. 2017), peach (Prunus persica) (Artlip and Wisniewski 1997; Wisniewski et al. 1999) and rose

(Ouyang et al. 2019). In Arabidopsis, it has been shown that the expression of two thirds of the COR-genes is regulated via the CBF pathway (Wisniewski et al. 2018); however, alternate splicing (AS) of COR-genes associated with chromatin binding, transcriptional regulation,

Figure 1-3. Regulation of the CBF pathway in Arabidopsis (Adapted from Wisniewski et al. 2018)

The cold signal is first perceived by the plasma membrane, which triggers the influx of Ca2+ and then activate a calcium-induced receptor-like kinase (CRLK ½), which triggers the

MEKK1-MMK2-MPK4 kinase cascade (MPK). MPK cascade increases the activity of the transcription factor ICE by inhibiting its phosphorylation, ubiquitination and degradation, which would have occurred under the activity of MPK 3/6. The OST1 (Open Stomata 1), a Ser/Thr protein kinase regulates the activity of ICE, alongside SIZ1. The active ICE activates CBF expression, leading to the expression of downstream COR genes. OST1-direct phosphorylation of BTF3s, which promotes its binding to CBFs, prevents degradation of CBFs. Photosynthesis and circadian clock also play a key role in the activation of CBF (Ding et al. 2015; Shi et al. 2018; Wisniewski et al. 2018)

signal transduction and effects on the plasma membrane has been suggested to play an important role as well in the immediate response to cold stress and the acquisition of hardiness (Calixto et al. 2018; Gallegos 2018).

1.7.3 Phenotyping

A winter hardiness zone index is attributed to plant varieties and indicates to consumers and growers the most suited varieties to grow in a particular geographic location.

The index relies on the United States Department of Agriculture (USDA) Plant Hardiness

Zone Map (https://planthardiness.ars.usda.gov/PHZMWeb/). The map is based on the average annual minimum winter temperature, divided into 10-degree F zones and relies on the correlation between climatological variables by geographic zones and pattern of plant survival

(Vukosavljev, 2014). The winter hardiness zone index of countless rose cultivars can be found on the “help me find” online platform (www.helpmefind.com).

Rose winter hardiness is traditionally evaluated in the field on a scale from 0 to 5 based on the level of damage observed after winter (Svejda 1979) (Figure 1-4). Vukosavljev

(2014) proposed to measure rose winter hardiness in the field as proportion of dieback compared to whole branch length (%) and percentage of regrowth as proportion of new shoots to whole branch length (%). Winter hardiness is a complex trait involving multiple factors such as soil quality, soil moisture during the fall and presence of pathogens. For this reason, winter hardiness can be evaluated only under natural conditions in the field during multiple seasons at multiple locations. However, field evaluation of winter hardiness is labor intensive, often subjective and depends upon weather conditions.

0 : No winter damage 1: Damage on tips only 2: Snow line

3: Damage with good 4: Damage with poor 5: Winter kill

spring regrowth spring regrowth

Figure 1-4. Traditional rating scale from 0 to 5 used to assess winter damage in the rose field (Credit Rouet C. 2014, Vineland Research and Innovation Centre)

Freezing tolerance is one of the main components of winter hardiness and is straightforward to screen for during artificial freezing tests using electrolyte leakage as an index of cold injury of the plant cells. Dexter and collaborators first described the measurement of electrolyte leakage to estimate the degree of freezing tolerance of different varieties of alfalfa after an adequate period of acclimation (Dexter et al. 1930, 1932). The percentage of electrolyte leakage is commonly converted into an arbitrary and comparative index for cold hardiness such as LT50, which is defined as the temperature at which 50% death occurs (Lindén et al. 2000). The electrolyte leakage can be used on various plant organs, as an indicator of cold hardiness and the method has already been applied in a wide variety of crops such as winter wheat (Săulescu and Braun, 2001), durum wheat (Bajji et al.

2002), walnut stem (Poirier et al. 2010), peas (Dumont et al. 2009), red raspberry (Linden et al., 2000), switchgrass and Miscanthus (Peixoto and Sage 2016) and roses (Karam and

Sullivan 1991; Le et al. 2012; Ouyang et al. 2019). The high positive correlation between

LT50 estimated via EL in roses and field winter damage data (Le at al. 2012) makes electrolyte leakage a promising method to estimate rose hybrids’ winter hardiness in the laboratory, while reducing the costs associated with multi-year multi-location field trials.

Nonetheless, the success of cold hardiness research under artificial conditions will depend upon the success of artificial acclimation. Acclimation in growth chambers under controlled temperature and artificial light can achieve an adequate level of acclimation of the plants before conducting freezing experiments (Săulescu and Braun, 2001). However, it is important to consider that cyclic endogenous factors can prevent full acclimation during the spring regardless of the conditions of photoperiod and temperature provided in controlled conditions

(Weiser 1970).

1.7.4 QTL Mapping

Cold hardiness in roses is highly heritable, from 51 to 92% (Svejda 1979), suggesting that large selection gain can be made when combining hardy parental lines (Zlesak 2007).

Major QTLs can be detected in segregating populations depending on the experimental design

(Vukosavljev, 2014). Vukosavljev (2014) mapped frost damage data from field trials and cold chamber experiments from two rose populations from an Explorer genetic background and identified four QTLs on chromosomes 4, 5 and 6, but only one QTL associated with field damage and located on chromosome 4 was identified across populations and this apparent stable QTL was not validated (Vukosavljev, 2014). While no candidate genes came out of

Vukosavljev’s mapping project in roses, most of the QTLs involved in freezing tolerance in

Arabidopsis mapped to CBF gene regions or nearby (Gery et al. 2011). These findings highlight the important role of the CBF pathway in the acquisition of cold hardiness and provide potential direction to identify cold-regulated candidate genes in roses.

1.8 Thesis Introduction

1.8.1 Background

Roses are one of the most important ornamental crops. Breeding consists of manual hybridizations between elite parental lines, commercial cultivars or genetically diverse accessions such as wild plants. Hybridizations can be made within or even between species.

Low maintenance, colourful and full-flower roses are preferred over single flowers present in wild type roses. Most modern roses are tetraploid (2n=4x=28) while half of the wild roses are diploid (2n=2x=14) but the range of ploidy among roses can vary between diploid and octoploid (Yokoya et al. 2000). Canadian landscape roses are predominantly tetraploids hybrids between Rosa kordesii, Rosa acicularis, Rosa amblyotis, Rosa laxa, Rosa spinossisima and Rosa rugosa. The complex genetic nature of roses and the different ploidy levels present a challenge for compatibility, fertility and seed production, and have also contributed to slowing down the research on rose genetics. Currently rose breeding is achieved with limited involvement of molecular technology, which could accelerate the breeding process in a market where high variety turnover is critical to maintaining nursery industry competitiveness. Advances in understanding the genetic architecture of rose traits would benefit the implementation of marker-assisted selection (MAS). Fundamental and applied research is on-going to unravel the genetic mechanisms for traits of importance to rose breeders and consumers.

Black spot is the most devastating disease of landscape roses and low winter temperatures limit the distribution of roses in temperate regions. Therefore, black spot disease resistance and cold hardiness are major breeding targets. While wild roses represent a gene pool for strong disease resistance and exceptional hardiness that is often under-exploited, it is those same qualities that modern roses lack. Unfortunately, introgressions from wild roses are tedious due to ploidy incompatibility and linkage drag which increase the number of

generations needed to breed out undesirable traits. Additionally, breeding for both traits based on field evaluation is challenging as it depends upon the environmental conditions and the observer. It is also costly and labour-intensive since it often requires multiple sites and multiple years of field trials. Moreover, the existence of at least 13 different races of

Diplocarpon rosae worldwide is a challenge for breeding black spot resistant varieties

(Whitaker et al. 2010). Genetic resistance to black spot does exist and is mainly qualitative.

Four resistance genes are known: Rdr1, Rdr2, Rdr3 and Rdr4 (von Malek and Debener 1998;

Hattendorf et al. 2004; Whitaker et al. 2010; Zurn et al. 2018; (Zurn et al. 2020). Cold hardiness in roses has high heritability and might be controlled by a few major genetic factors.

Hardy Canadian roses exist and are known for their exceptional winter survivability down to -

40ºC. The hardiest roses belong to the Explorer series bred by Dr. Felicita Svejda in the 1980s and represent a valuable gene pool for cold hardiness QTL mapping (Svejda 1979;

Vukosavljev 2016). Consequently, both black spot disease resistance and cold hardiness are good candidates for developing molecular markers for MAS implementation.

1.8.2 Rationale

Initiatives to implement MAS in rose breeding will greatly benefit from genomics technology such as genotyping-by-sequencing (GBS) and the Genome Database for Rosaceae

(https://www.rosaceae.org). GBS is a low-cost sequencing technology that can generate thousands of single nucleotide polymorphism (SNP) markers through genome complexity reduction (Elshire et al. 2011). The large number of SNPs generated by GBS is ideal to generate high-density SNP based map. Genetic maps can be generated following a two-way pseudo-test cross strategy using only markers that segregate in simplex configuration and are homozygous in one parent and heterozygous in the other, thereby overcoming the challenges arising from increased ploidy (Grattapaglia and Sederoff 1994; Li et al. 2014). High-density

genetic maps can be used as a framework in QTL studies when combined with accurate phenotypic data. Field phenotyping can be combined with laboratory experiments conducted in controlled conditions with improved accuracy. Detached leaf assays conducted in a controlled environment with multiple single-spore isolates representative of the field pathogen diversity offer an alternative approach to field trials allowing accurate phenotyping of rose disease resistance. Winter hardiness data as measured in the field can be complemented by freezing experiments in artificial conditions. Freezing experiments can be conducted after an artificial acclimation period and electrolyte leakage can be used as an index of freezing tolerance. The reference genome sequence of Rosa chinensis (Raymond et al. 2018) is well annotated and represents a framework for candidate gene identification and molecular marker development for both disease resistance and cold hardiness.

This thesis represents a new investigation of mapping two critical traits for landscape roses: black spot disease resistance and cold hardiness. It examines populations derived from the Explorer roses and focuses on implementing MAS in order to optimize breeding schemes.

This thesis represents the first incentive to modernize Canada’s rose breeding program by bridging the gap between conventional breeding and new molecular technology.

1.8.3 Implication

Research chapters 2 and 3 of this thesis describe the use of bi-parental tetraploid populations derived from a Canadian Explorer rose genetic background to map black spot disease resistance and cold hardiness following a two way-pseudo testcross strategy from

GBS data generated at low-cost. GBS technology and accurate phenotyping permitted the identification of a polymorphism diagnostic of black spot disease resistance against multiple races of D. rosae in the muRdr1A gene. In addition, multi-site and multi-year field trial permitted the exploration of the genetic basis for spring regrowth and winter damage as a 42

measurement of cane die back, and experiments under artificial conditions permitted the exploration of rose freezing tolerance. These experiments set up a framework for future research on gene expression analysis of key genetic factors of the CBF-pathway in roses.

1.8.4 Objectives and Hypothesis of each Research Chapter

Chapter 2 Mapping black spot disease resistance in garden roses (Rosa x hybrida)

Hypothesis:

 Black spot disease resistance is controlled by major R genes and can be mapped onto

the rose genome using a bi-parental segregating population.

 Detached leaf assay (DLA) can be conducted with multiple isolates of D. rosae

representative of the field black spot pathogen diversity to detect QTLs also associated

with field-level resistance.

Objectives:

 To conduct QTL analysis on field and DLA data with genotypic data generated by

GBS.

 To identify candidate genes and polymorphism and develop markers diagnostic of

black spot disease resistance.

Chapter 3 Mapping cold hardiness in garden roses (Rosa x hybrida)

Hypothesis:

 Winter hardiness is heritable in roses and can be mapped onto the rose genome using a

bi-parental segregating population.

 Freezing tolerance can be assessed in artificial conditions using a high-throughput

phenotypic assay based on electrolyte leakage to detect QTLs also associated with

field-level winter hardiness.

Objectives:

 To conduct QTL analysis on field and laboratory data with genotypic data generated

by GBS.

 To identify candidate genes involved in cold hardiness.

CHAPTER 2: IDENTIFICATION OF A POLYMORPHISM WITHIN THE ROSA

MULTIFLORA MURDR1A GENE LINKED TO RESISTANCE TO MULTIPLE

RACES OF DIPLOCARPON ROSAE W. IN TETRAPLOID GARDEN ROSES

(ROSA X HYBRIDA)

Cindy Rouet1,2*, Elizabeth A. Lee2, Travis Banks1, Joseph O’Neill 1, Rachael LeBlanc1, Daryl

J. Somers1

Address: 1 Vineland Research and Innovation Centre, 4890 Victoria Avenue North, Box 4000,

Vineland Station, ON L0R 2E0, Canada.2 Department of Plant Agriculture, University of

Guelph, 50 Stone Rd E, Guelph, ON N1G 2W1Canada*Correspondence: Cindy Rouet. Email: [email protected]

Theoretical and Applied Genetics International Journal of Plant Breeding Research

ISSN 0040-5752

Theor Appl Genet DOI 10.1007/s00122-019-03443-9

Key message

A QTL for resistance to several races of black spot co-located with the known Rrd1 locus in

Rosa. A polymorphism in muRdr1A linked to black spot resistance was identified and molecular markers were designed.

Abstract Black spot, caused by Diplocarpon rosae, is one of the most serious foliar diseases of landscape roses that reduces the marketability and weakens the plants against winter survival. Genetic resistance to black spot (BS) exists and race specific resistance is a good target to implement marker-assisted selection (MAS). High-density Single Nucleotide

Polymorphism (SNP) based genetic maps were created for the female parent of a tetraploid cross between ‘CA60’ and ‘Singing in the Rain’ using Genotyping-By-Sequencing (GBS)

following a two-way pseudo-testcross strategy. The female linkage map was generated based on 227 individuals and included 31 linkage groups, 1055 markers, with a length of 1980 cM.

Race specific resistance to four D. rosae races (5, 7, 10, 14) was evaluated using a detached leaf assay. Black spot resistance was also evaluated under natural infection in the field.

Resistance to races 5, 10 and 14 of D. rosae and field resistance co-located on chromosome 1.

A unique sequence of 32 bp in exon 4 of the muRdr1A gene was identified in ‘CA60’ that co- segregates with D. rosae resistance. Two diagnostic markers, a presence/absence marker and an INDEL marker, specific to this sequence were designed and validated in the mapping population and a backcross population derived from ‘CA60’. Resistance to D. rosae race 7 mapped to a different location on chromosome 1.

Key Words

Tetraploid, Genotyping-by-sequencing (GBS), Single Nucleotide Polymorphism (SNP),

High-density genetic map, Marker-Assisted Breeding, Disease resistance.

Author Contribution Statement

CR and DS designed the experiments. CR implemented the experiments, conducted the phenotyping, mapping and data analysis. TB and JO contributed to GBS library preparation and bioinformatics support. RL contributed to marker development. CR authored the manuscript. DS, TB and EL edited the manuscript. All authors read and approved the final manuscript.

Acknowledgments and Funding

This work was supported by funding from the Vineland Research and Innovation Centre,

Agriculture and Agri-Food Canada Growing Forward 2 project #AIP-P013, the Canadian

Nursery Landscape Association, Landscape Manitoba and Landscape Alberta.

Conflict of Interest

The authors declare that they have no conflict of interest. 46

2.1 Introduction

Black spot (BS) (Diplocarpon rosae Wolf) is one of the most economically important foliar diseases of outdoor grown roses (Rosa x hybrida), as it reduces both marketability and survival of the plant (de Vries and Dubois 2001; Whitaker et al. 2010a; b). BS infection is characterized by black lesions with fringed margins appearing on the leaf surface that senesce and eventually lead to defoliation of the plant (Gachomo et al. 2006). While chemical options for controlling BS are available, they require weekly applications throughout the season to be effective (Jeliazkova 2012). Environmental awareness and increased demand for low maintenance and easy to grow rose cultivars have elevated the importance of breeding roses for genetic-based BS resistance (Yokoya et al. 2000; de Vries and Dubois 2001; Blechert and

Debener 2005; Waliczek et al. 2015). Rose breeders worldwide are addressing the challenge of creating and testing BS resistant roses displaying suitable ornamental features. Research programs such as Earth-Kind Rose Trialing, started at Texas A&M University, are identifying geographically adapted genetically-superior roses (Zlesak et al. 2015). The Knock Out® rose family is an example of successful cultivars with increased BS resistance.

Genetic control of BS resistance in roses follows the classic gene-for-gene type of interaction in the pathosystem of D. rosae/Rosa (Neu and Debener 2019). To date, four major resistance (R) genes have been reported (Rdr1, Rdr2, Rdr3, Rdr4) (von Malek and Debener

1998; Hattendorf et al. 2004; Whitaker et al. 2010a; Zurn et al. 2018). Eleven different pathogenic races of D. rosae have been identified and characterized (Whitaker et al. 2010) and two additional races have been recently reported (Whitaker et al. 2010a, b; Zurn et al.

2018). The first R gene, Rdr1 (muRdr1), was introgressed into modern roses from the diploid hybrid R. multiflora 88/124-46 (von Malek and Debener 1998). Rdr1 is located on chromosome 1 (Kaufmann et al. 2003; Biber et al. 2010; Terefe-Ayana et al. 2011; Spiller et al. 2011) and is a complex locus, consisting of a cluster of nine tandemly repeated R genes

with closely related sequences (Terefe-Ayana et al. 2011). The muRdr1 genes belong to the

TIR-NBS-LRR (Toll/interleukin-1 receptor-nucleotide binding site-leucine rich repeat; TNL) family of R genes (Terefe and Debener 2011). In R. rugosa, the Rdr1 locus (ruRdr1) consists of 11 tandemly repeated TNLs exhibiting a high degree of sequence similarity to muRdr1

(Terefe-Ayana et al. 2012). Specifically, it is the muRdr1A gene within the Rdr1 locus that confers resistance to five D. rosae pathogenic races (races 2, 4, 5, 6 and 10) (Menz et al.

2018). The remaining three black spot R genes are less characterized. Rdr2, like Rdr1, is located on chromosome 1 and is tightly linked to Rdr1 (Hattendorf et al. 2004). Rdr3 was the first R gene for BS resistance to be discovered in tetraploid roses, but to date has not been mapped (Whitaker et al. 2010a, b). Rdr4 was also discovered in tetraploid roses, has been mapped to the long arm of chromosome 5, and confers resistance to multiple D. rosae races

(races 2, 3, 8, 9, 10, 11, and 13) (Zurn et al. 2018).

Breeding roses for field-level BS resistance is challenging and costly, as it requires multi- year and multi-location testing, the ratings are often subjective, and field-level resistance generally involves resistance to multiple races of D. rosae. However, given the recent advances in understanding the genetics of BS resistance it is now possible to utilize genetic markers to aid in eliminating germplasm from testing. Marker-assisted selection (MAS) relies on tight genetic associations between DNA markers and the trait of interest and can be applied to both single-gene as well as polygenic traits. Genotyping of the breeding populations is done and individuals are eliminated or retained for further evaluation, based solely on the marker genotype. MAS has already been successfully used in the breeding of various ornamental crops such as greenhouse carnations against bacterial wilt (Yagi 2013), roses for genetic background selection and shoot organogenesis (Debener 2003; Nguyen et al. 2019) and hydrangea to direct interspecific crosses (Granados Mendoza et al. 2013). The implementation of MAS in garden roses for field-level BS resistance would allow for elimination of

susceptible individuals, thereby greatly reducing development costs and increasing genetic gain as larger breeding populations could be effectively evaluated without an increase in resources.

There have been a number of rose breeding programs focused on lower maintenance style roses, characterized by excellent winter hardiness and disease resistance. Perhaps one of the best known of these programs was the Agriculture and Agri-food Canada (AAFC) program based in Ottawa, ON Canada that developed the Explorer Roses. The program was aimed at developing winter hardy, disease resistant rose cultivars for northern climates. To achieve these goals, the Explorer rose breeding program relied quite extensively on germplasm from

R. laxa Retzius, R. spinosissima, and R. kordesii Wulff for both cold hardiness and disease resistance (Svejda and Mcgee 2006). In this research we characterize and exploit a source of field-level BS resistance from the AAFC program and develop a genetic marker system that would facilitate MAS for field-level BS resistance. The objectives of this study are 1) to perform high throughput phenotyping to assess the level of BS resistance in 167 progeny derived from a tetraploid bi-parental rose population, 2) to identify the genetic control of BS resistance using QTL analysis and re-sequencing technology and 3) to develop diagnostic markers paired with the donor genotype in order to implement MAS in the rose breeding program.

2.2 Materials and Methods

2.2.1 Genetic materials

The mapping population for BS resistance consisted of 365 tetraploid F1 progeny resulting from the cross between the BS resistant female ‘CA60’ and the BS susceptible male

‘Singing in the Rain’ (‘SITR’). ‘CA60’ (23104FR2) was an experimental rose developed at the Agricultural and Agri-Food Canada’s (AAFC) Morden Manitoba research station from the

cross ‘RSM 104’ × ‘Frontenac’. ‘Frontenac’ was developed by Dr. Felicita Svejda for both cold hardiness and BS resistance and released in 1983 by AAFC L’Assomption Quebec research station as part of the Explorer Series Collection (Ogilvie et al. 1993). ‘CA60’ has exhibited excellent field-level resistance to BS under high disease pressure at Vineland

Research and Innovation Centre (Vineland, ON). ‘Singing in the rain’ (‘SITR’) (MACivy) was a commercial floribunda rose cultivar developed by McGredy from the cross Sexy

Rexy® x ’Pot of Gold’ (U.S. Patent PP8,362). ‘SITR’ has exhibited extremely poor field- level resistance to BS under high disease pressure at Vineland (Ontario, Canada). The mapping population was created over a 2-year period (2014-2016) to get sufficient numbers of progeny. The marker validation population consisted of 88 individuals from a backcross to

‘CA60’: ‘CA60’ x ‘S13-10’. ‘S13-10’ was derived from the cross (‘Basye’s Legacy’ x

‘Frontenac’) x ‘CA60’ and was BS resistant under natural field disease pressure. Differential cultivars for pathogenic races of D. rosae consisted of clones of cv. ‘Mermaid’, cv. ‘Surrey’,

Honey Bee™, Sexy Rexy®, Love and Peace™, cv. ‘George Vancouver’, cv. ‘Mrs Doreen

Pike’, Knock Out® and Baby Love™ (Table 2-1). In the greenhouse, only sulfur applications were applied to control powdery mildew on the plants, no additional chemicals were used.

Biocontrol was used to control insects’ pressure. All plant material was maintained in the greenhouse and field using standard propagation and cultivation techniques and all plants were grown on their own roots.

Table 2-1. Plant Material used in the study.

Plant material Type of material Use in the study Size

Differentials a Cultivars D. rosae race 9 characterization ‘CA60’ x ‘SITR’ F1 bi-parental Mapping population 365 including 167 cross used in phenotyping

‘CA60’ x ‘S13-10’ Backcross Genetic model and 88 including 59 marker validation with known BS population phenotype a Whitaker et al., (2010 a, b)

2.2.2 D. rosae single spore isolates

Four different single spore isolates were used for mapping BS disease resistance in the segregating ‘CA60’ x ’SITR’ population. Isolates VHy-12.4, VSKO4 and VOTB17-1 originated from the Vineland experimental field (Vineland, ON, 43°11'30.9"N

79°23'45.7"W). Infected rose leaves from experimental hybrids and commercial cultivars were collected during the summer and fall in 2014 and the single spore isolates were obtained through conidia transfer onto a young detached rose leaf or with the agar methods from infected leaves as described by Debener et al. (1998). Isolates VSKO4 and VOTB17-1 were collected from Sunny Knock Out ® and cv. 'Out of The Blue' respectively. Isolate VHy-12.4 was isolated from an infected leaf of an experimental rose hybrid. Isolates VHy-12.4, VSKO4 and VOTB17-1 represented the most prevalent D.rosae races in the Vineland experimental field (Supplementary Material – Appendix 1, Supplementary Table 2-1 – Appendix 2). Isolate

B005, originating from Belgium, was obtained from the University of Minnesota D. rosae collection. The universally BS susceptible cultivar ‘Mermaid’ (part of the differential cultivar set) was used to increase inoculum of the races as described in Debener et al. (1998).

2.2.3 Detached leaf assay for race specific D. rosae resistance

Detached leaf assays (DLA) with the four isolates of D. rosae were performed in order to phenotype a subset (n=167) of the 365 progeny from the ‘CA60’ x ‘SITR’ mapping population, the two parents (‘CA60’ and ‘SITR’), as well as the susceptible (cv. ‘Mermaid’) and the resistant (cv. ‘Mrs. Doreen Pike’) controls. The experimental design was a split-plot design with three replications. The experimental unit consisted of a young but fully expanded greenhouse grown rose leaf comprised of five leaflets, with the main plot being D. rosae isolate and sub-plot being genotype.

The DLA were performed by adapting a protocol described by Debener at al. (1998).

Briefly, each assay consisted of two young and fully expanded detached rose leaves per genotype; each leaf was comprised of five leaflets. The leaves were collected, carefully washed under running tap water at room temperature and blotted dry. Each leaf was placed in a 15 cm diameter Petri dish on wet filter paper and inoculated with 10 droplets (two droplets of 10 μL per leaflet) of D. rosae single spore inoculum adjusted to a concentration of 2 to

2.5x105 conidia mL-1. The Petri dishes were incubated in a controlled environment at 23°C with 12 hours artificial daylight, and 80% humidity to promote the infection. The droplets on the surface of the leaves were blotted away two days after the inoculation to prevent secondary infections and the Petri dishes were sealed with Parafilm®. After 14 days of incubation at 23°C with 12 hours artificial daylight, BS symptoms were recorded. Using a dissecting microscope each lesion was observed to detect the presence of acervuli and a compound microscope was used to observe conidia in suspension. The visual rating scale used in the detached leaf assay ranged from 0 to 5 (Supplementary Figure 2-1 - Appendix 4):

0 = no symptom

1 = presence of spots but no acervuli and no conidia

2 = presence of spots, no conidia and presence of acervuli

3 = presence of lesions with a few conidia

4 = lesions with a high number of conidia and the presence of fungal mycelium

5 = high infection level with lesions expanding beyond the droplet area, conidia,

acervuli

One single disease score was given to each leaf based on the presence of up to ten lesions.

2.2.4 Natural field disease pressure screening for BS resistance

Field screening for D. rosae under natural inoculation conditions was done by planting a subset (n=157) of the mapping population, the two parents and an additional control Knock

Out® at the Vineland station during the summer of 2015 using a completely randomized design (CRD) with 6 rows. The experimental unit was a plant, with six replications of 157 progeny of the ‘CA60’ x ‘SITR’ mapping population, 51 reps of ‘CA60’, 61 reps of ‘SITR’, and 12 reps of Knock Out®, in each row. All roses were grown on their own roots. Data on

BS resistance under natural field disease pressure was recorded in October 2016 when the disease pressure was high. BS resistance for each progeny was rated on a continuous scale from 0 to 100% based on visual assessment of the total leaf surface area of the plant showing symptoms attributed to BS. Six clones or less were available in the field per progeny in fall

2016 and the highest disease score (i.e., most severe) among the clones was retained for each progeny because only one season of data was available. Field assessment of BS resistance at high disease pressure was made on 59 progeny from the ‘CA60’ x ‘S13-10’ backcross population in September 2018 based on only one plant per genotype and using.the visual rating scale of 0% visible lesions = resistant to BS, <5% visible lesions = tolerant to BS,

>15% visible lesions = susceptibile to BS. Data on BS resistance in the field for the 156 elite genotypes was recorded during three periods, in August 2017, September 2017 and October

2017 with 1 to 10 clones per genotype using a continuous scale from 0 to 100%. The highest score across clone and period of data collection was recorded as the disease score of each the genotypes.

2.2.5 Single Nucleotide Polymorphism calling

DNA isolation and sequencing of Genotyping-by-Sequencing (GBS) were conducted as described in supplementary material (Appendix 1). The Illumina reads for each GBS

sample were aligned to the R. chinensis genome V2 (Raymond et al. 2018) using the bwa mem algorithm (Li 2013) and the alignments were sorted using samtools (Li et al. 2009; Li

2013). For each sample, the Genome-Analysis-Tool-Kit (GATK v3.7) (McKenna et al. 2010) was used to call high-confidence Single Nucleotide Polymorphism (SNP) variant genotypes relative to the R. chinensis genome V2. Homozygote calls supported by less than 11 reads, heterozygote calls supported by less than 2 reads for one or both alleles, and heterozygote calls with minor allelic read frequency <0.1 were eliminated. Simplex-by-nulliplex segregating SNPs were obtained by selecting SNPs where one parent was heterozygous, one parent was homozygous, and the p-value for a chi-sq. test of 1:1 het:hom segregation ratio in the progeny was > 0.01. Products of self hybridization were then identified and eliminated by calculating the fixed female fraction and fixed male fraction for each individual using the simplex-by-nulliplex segregating SNPs. F1 progeny resulting from selfing were visualized by plotting the fraction of fixation of the progeny to parental alleles. Fifteen progeny from selfing and uncontrolled crosses for which the fixed female fraction or fixed male fraction was either greater than 0.75 or less than 0.25 were removed from this analysis. After product-of-selfing individuals had been removed from analysis, the simplex-by-nulliplex segregating SNPs were re-obtained. The GBS sequencing data is accessible at NCBI Sequence Read Archive under

SRA accession PRJNA544718.

2.2.6 Phenotypic data

Three replications of DLA were conducted on the same subset of 167 individuals from the mapping population using four single-spore isolates including VHy-12.4, VSKO4,

VOTB17-1 and B005. Each replication was conducted with two leaves per rose genotype placed in individual Petri dishes and individual disease scores were rated for each leaf. A minimum of 4 disease scores recorded in the DLA for each genotype were averaged from the

three replicate assays to calculate the average disease score for each genotype. A genotype with an averaged disease score of 0 or 1 indicated an incompatible reaction with the pathogen and was rated resistant. A genotype with an averaged disease score of 3 to 5 indicated a compatible reaction with the pathogen characterized by the production of conidia and was rated susceptible. There were 34 individuals removed from the analysis since a BS rating could not be reliably determined. Chi-Square goodness-of-fit tests were performed on DLA data and data from the field for the mapping population and the backcross population, using R v.3.5.3 (R Core Team 2019) to study the segregation between resistant and susceptible genotypes among the progeny. In addition, a correlation matrix was created for DLA and field data using the Pearson correlation test in R v.3.5.3.

A linear mixed effects model approach was used to analyze the DLA split-plot design using the SAS software (SAS ver. 9.4, SAS Institute Inc., Cary, NC, USA) as described in supplementary material (Appendix 1). Type III tests of fixed effects tested the significance of the fixed effects specified in the model (Isolate, Genotype and Isolate * Genotype interaction), and random effects (Block and Block * Isolate interaction) were tested with a likelihood test for covariance parameters.

2.2.7 Construction of genetic maps

In the linkage analysis prior to the creation of the map, both the SNP markers and individuals were tested against the expected segregation ratio of 1:1 het:hom. Markers showing significant segregation distortion (p-value <0.01, χ2 test) were removed. Markers with more than 30% of missing individuals were deleted and individuals with more than 40% of missing data were removed from the analysis. The genetic linkage map for ‘CA60’ was constructed with the simplex markers following a two-way pseudo-testcross strategy

(Grattapaglia and Sederoff 1994) using R-package ASMap (Taylor & Butler 2017) with the

Kosambi mapping function, based on the corrected and imputed SNP data (R 3.5.3). A p- value of 1x10-6 was given to the MSTmap algorithm (Wu et al. 2008) to use a conservative minimum threshold of 30-35 cM before linking markers between clusters. The four homologous chromosomes from each of the seven groups were identified based on the physical position (in Mb) of each SNP marker in the R. chinensis reference genome. Linkage groups were visualized using Mapchart (Voorrips 2002).

2.2.8 QTL analysis

A total of 110, 96, 101, 97 progeny were used to identify QTL associated with resistance to isolates VHy-12.4, VSKO4, VOTB17-1 and BOO5, respectively. QTLs associated with BS resistance under field disease pressure were identified based on 99 progeny. QTLs for BS resistance were detected by performing a genome scan using the Haley and Knott algorithm

(Haley and Knott 1992), in R/qtl2 (Broman et al. 2019) (R v.3.5.3). The significance cut-off was determined by performing 1000 permutations. The fitqtl function from the package R/qtl

(Broman et al. 2003) was applied to estimate the percentage of variation explained by the main QTL. Multiple QTL analysis (MQM) was conducted using the main QTL previously detected as a cofactor to search for additional QTLs using the R/qtl implementation of Ritsert

Jansen's MQM method (Arends et al. 2010) (R v.3.5.3).

2.2.9 Analysis of muRdr1A sequence

The published sequence of the 265,477 bp Rdr1 locus (Terefe-Ayana et al. 2011) was aligned with the publicly available whole reference genome of R. chinensis V2 using

BLASTN (Altschul et al. 1990) to investigate the presence of candidate resistance genes around the BS QTLs identified in ‘CA60’. The presence of BS candidate genes was further investigated in the parental genotypes. Leaf tissue was collected from ‘CA60’ and ‘SITR’,

frozen with liquid nitrogen, and ground to a powder using a mortar and pestle. High molecular weight genomic DNA was extracted from the ground tissue using a MagAttract

HMW DNA kit (Qiagen) following the manufacturers instructions. The two DNA samples were sequenced using 10X Genomics’ Chromium Genome Library Kit v2 and each library was sequenced on an Illumina HiSeq-X instrument. For each parent, the first 23 bases of the forward reads were trimmed off to remove the 10X Genomics barcode using the software bbmap (Bushnell B.). The sets of reads were then mapped to the muRdr1A gene (Genbank

Accession HQ455834) (Terefe-Ayana et al. 2011; Menz et al. 2018) using bwa mem (Li

2013). The resulting alignments were visualized using Tablet (Milne et al. 2013). A unique region of muRdr1A covered by ‘CA60’ reads but not ‘SITR’ reads was identified and primers to amplify this region were designed using Primer3 (forward primer -

ACAAGTGTTTCCAGATCCACAAG; reverse primer 1-

CCAAATTTGAGCACACCATGG; reverse primer 2- AGTGCGGTCGGTAATCAAGA)

(Figure 2-3) (Untergasser et al. 2012). The resequencing data for ‘CA60’ and ‘SITR’ is accessible at NCBI Sequence Read Archive under SRA accession PRJNA544718. muRdr1A specific genotyping was done by PCR amplification and High Resolution Melting Analysis

(HRM) (Smith, Lu, and Bremer 2010) as described in supplementary material (Appendix 1).

The presence of the unique INDEL was investigated in the other paralogues of the Rdr1 locus through multiple sequence alignment using CLUSTAL Omega (1.2.4) (Sievers et al. 2011).

Additionally, de novo assembly of the Rdr1 locus in ‘CA60’ was executed using IDBA_UD

(Peng et al. 2012).

2.3 Results

2.3.1 Race diversity used in this study

Isolate VHy-12.4 was characterized as race 10 based on the infection pattern on differential rose cultivars (Table 2-1). Isolate VSKO4 was able to infect all differential cultivars with the exception of cv. ‘Mrs. Doreen Pike’. This pattern of resistance and susceptibility did not correspond to any infection pattern published to date and is assumed to be a newly identified race (tentatively referred to as race 14). The mapping population parents

‘CA60’ and ‘SITR’ differed in their response to this newly identified race with ‘CA60’ being the resistant parent. The differential cultivars cv. ‘Mermaid’, Honey Bee™, Sexy Rexy®, cv.

‘Surrey’, Love and Peace™, and Knock out® were susceptible to isolate VOTB17-1 (Table

2-2), which was consistent with the pattern identified as race 7 (Whitaker et al. 2010). The fourth isolate used in this study was BOO5, obtained from the collection at the University of

Minnesota, and identified as race 5 (Whitaker et al. 2010), however, the pattern of infection on the set of differential cultivars differed from the published literature showing infection on the cultivar Love and Peace™ (Table 2-1).

Table 2-2. Comparison of the host range of seven isolates of D. rosae with the published infection patterns for known D. rosae races using a set of differential cultivars

Current Study Whitaker et al. (2010)

Isolates Name HyVicH-4b VHy- VSKO4 b VOTB17-1 KOMN DA-1 R6 B005 b KOMN DA-1 R6 B005 b 12.4 b b

Country of Canada Canada Canada Canada North UK Germany Belgium North UK Germany Belgiu origin America America m

Host genotype CA33 x 3 (93 x Sunny cv. Out of Knock Unknown 91 ⁄ 100- Unknown Knock Out® Unknown 91 ⁄ 100- Unkno 174) Knock The Blue Out® Hybrid Tea 5c Hybrid Tea 5 c wn Out® Differentials cv. Mermaid + + + + + + + + + + + + Honeybee™ - + + + + + + + + - + + Sexy Rexy® + + + + + + + + + + + + cv. Surrey + + + + + + + - + + + - Love and + + + + + + + + + + + - Peace™ cv. George - + + - + - - - + - - - Vancouver Knock out® + + + + + + + + + - - + Baby Love™ - - + - + ------cv. Mrs Doreen ------Pike Additional CA60 a ------cv. Singing In + + + + + + The Rain a

Race 8* 10 UD 7* 10* 8* 7* UD 10 8 7 5 characterization

+ Susceptible; - Resistant; UD Undescribed Race; a Mapping population parents; b Isolates used in mapping; c von Malek and Debener (1998) *Potential race characterization

2.3.2 Segregation for black spot resistance within the mapping population ‘CA60’ x ‘SITR’

Variance analysis clearly demonstrated that the level of BS resistance was genotype dependent (Table 2-3). The segregation pattern between resistant and susceptible phenotypes appeared to be bimodal for all sources of infection (Figure 2-1). Chi-square tests confirmed this for three of the sources of infection (VHy-12.4, BOO5, and field) (Table 2-4). The field

BS infection data showed there were 14 out of 157 intermediate phenotypes ranging from

10%-70% infection, which were removed from this analysis (Figure 2-1). The number of resistant individuals was greater than expected for isolates VSKO4 and VOTB17-1 (Table 2-4) and did not segregate 1:1. All individuals resistant to isolate VHy-12.4 (race 10) were resistant to the three other isolates, however some individuals that were susceptible to VHy-

12.4 (race 10) were resistant to VOTB17-1 (race 7) (Table 2-4). The relationships between individual isolate DLA scores and field disease pressure were all highly correlated ranging from R2=0.96 to 0.87, with the lowest correlation involving isolate VOTB17-1 (Table 2-5).

Additionally, segregation for field-level BS resistance in a backcross population involving

‘CA60’ as the recurrent parent (44 resistant: 15 susceptible) fit a 3:1 ratio (χ2= 0.005, p-value

= 0.94).

2.3.3 Parent-specific linkage maps and QTL analysis for BS resistance

Parent specific linkage maps were generated for the mapping population following a two- way pseudo-testcross strategy. A total of 4,055 simplex SNP markers were identified that were heterozygous in ‘CA60’ and homozygous in ‘SITR’ and 3,385 simplex SNPs were identified that were homozygous in ‘CA60’ and heterozygous in ‘SITR’. The ‘CA60’ female map included 227 individuals and spanned 31 linkage groups with 1055 simplex SNP markers across 1980 cM (Supplementary Figure 2-2 – Appendix 5). The length of linkage groups varied from 7.1 cM (LG1.H4) to 152.7 cM (LG5.H1) with a mean interval distance between

markers of 1.88 cM. The ‘SITR’ male map included 200 individuals and spanned 31 linkage groups across 1937cM with 801 simplex SNP markers and a mean interval distance between markers of 2.6 cM. The relative position of the SNPs in R. chinensis genome were used to assign homologous linkage groups to the seven rose chromosomes. The genetic map from the

BS resistant female parent ‘CA60’ was used in further BS QTL analysis.

QTLs for resistance to three isolates (VHy-12.4, VSKO4, BOO5) and field-level resistance were co-localized on chromosome 1 (LG1.H3), explaining between 63 and 88% of the phenotypic variation (Table 2-6). Additionally, a second distinct QTL for resistance to isolate VOTB17-1 was also mapped to chromosome 1 (LG1.H3), 22 cM proximal from the multi-isolate QTL and explained 42% of the phenotypic variation. No other QTLs were detected for BS resistance in this population, when the major QTL was set as the co-factor.

These results were confirmed by conducting MQM, which identified the same QTLs with slightly wider confidence intervals (Supplementary Table 2-2 – Appendix 3). Using the SNPs that defined the two QTLs and their positions in the R. chinensis genome, the multi-isolate

QTL was physically located within the reported Rdr1 locus (Figure 2-2). However, the confidence interval for the QTL involved in BS resistance to isolate VOTB17-1 (race 7) did not include the Rdr1 locus.

Table 2-3. Mixed model variance of the effect of Isolate and Genotype on the disease score attributed in detached leaf assay using single spore inoculum of D.rosae

Covariance Subject Estimate

parameters

Block Genotype x 0.08569

Petri_dish

Block x Isolate Genotype x 2.32*10-13

Petri_dish

Fixed effects Numerator df Denominator df F value Pr>F

Isolate 3 1798 0.02 0.9960

Genotype 162 801 142.28 <.0001

Isolate x Genotype 380 1798 55.95 <.0001

Figure 2-1. Phenotypic segregation within the mapping population ‘CA60’ x SITR’ for BS resistance to isolates VHy-12.4, VSKO4, VOTB17-1 and BOO5 based on disease scores averaged across three replicates of DLA and for field disease resistance under natural disease pressure recorded in October 2016 (measured as percentage of diseased leaf surface at the whole plant scale). The frequency of resistant (grey bars) and susceptible (black bars) is displayed. Intermediate phenotypes ranging from 10%-70% infection in the field were removed for QTL mapping.

Table 2-4. Segregation of phenotypes among the mapping population 'CA60' x 'SITR' in black spot detached leaf assay conducted with four D.rosae isolates and under field disease pressure

Source of infection Number Ratio Chi square 1 Number of genotypes with different phenotypes

individuals in Resistant : between pair of isolates

DLA Susceptible

VHy-12.4 VSKO4 VOTB17- BOO5

Isolate

VHy-12.4 149 73 : 76 0.06 (0.80) - 1 15 5

VSKO4 133 79 : 54 4.69 (0.03) - 1 1

VOTB17-1 135 88 : 47 12.45 (0.0004) - 6

BOO5 130 69 : 61 0.49 (0.48) -

Field 61 : 80 2.56 (0.10)

1 Chi square test for 1:1 segregation ration (P Value)

Table 2-5. Pearson correlation test between phenotypic data collected in detached leaf assay with four D.rosae isolates and phenotypes under field disease pressure (R version 3.5.3)

VHy- VSKO4 VOTB17- BOO5 Field

12.4 1

VHy-12.4 - 0.95 0.90 0.89 0.96

VSKO4 - 0.95 0.91 0.92

VOTB17- - 0.92 0.87

BOO5 - 0.86

Field -

Source Number Linkage Position Closest Marker QTL size PVE1 (%) LOD of individuals in group (cM) (cM) value infection QTL mapping

Isolate VHy- 110 LG1.H3 15 SChr1_6705113 3.1 88.0 52.6 12.4 9 VSKO4 96 LG1.H3 15 SChr1_6705113 5.3 84.8 40.0 9 VOTB17 101 LG1.H3 37 SChr1_6617011 14.2 42.0 21.1 -1 5 BOO5 97 LG1.H3 15 SChr1_6705113 5.3 62.8 21.0 9 Field 99 LG1.H3 15 SChr1_6705113 3.1 87.7 46.7 9 1Percent of variance explained

Figure 2-2. Sequence alignment between the published sequence of Rdr1 locus and Rosa chinensis physically locates Rdr1on Chr.1 in the Rosa chinensis genome. The red line represents the muRdr1gene locus in Rosa chinensis. QTL for resistance to isolates VHy-12.4, VSKO4 and BOO5 and field resistance on ‘CA60’ LG1.H3 is anchored to the Rosa chinensis physical map with SNP marker sequences derived from GBS. QTLs confidence intervals and field resistance are indicated on the right of ‘CA60’ LG1.H3. The Rdr1locus is physically located within the QTL confidence intervals.

2.3.4 Candidate genes and development of ‘CA60’ Rdr1A allele-specific markers

Given that the SNPs that defined the multi-isolate QTL are located within the Rdr1 locus, one of the muRdr1 genes is a likely candidate resistance gene. De novo assembly of the Rdr1 locus in ‘CA60’ identified a 3739 bp scaffold identical at 99.97% similarity to the published muRdr1A genomic sequence (Terefe-Ayana et al. 2011). A 32 bp sequence

(TTCCAGAACCACCAAATTTGAGCACACCATGG) unique to muRdr1A was present in

‘CA60’ and not present in ‘SITR’ (Figure 2-3). The polymorphism was found in the fourth exonic region of muRdr1A between positions 3645 to 3685. The specificity of this INDEL to the muRdr1A gene was confirmed through multiple sequence alignment of all nine Rdr1 locus paralogous genes (Figure 2-4.). Two diagnostic markers, a presence/absence marker and an

INDEL marker, were designed based on this polymorphism that could identify the ‘CA60’

Rdr1A allele (Figure 2-3). The presence/absence marker was able to distinguish between progeny carrying the ‘CA60’ Rdr1A allele and progeny without the allele for the mapping population and the backcross population using HRM (Figure 2-5 A). The mapping population segregated genotypically 98 presence of the ‘CA60’ Rdr1A allele: 90 absence of the ‘CA60’

Rdr1A allele (χ2 1:1 = 0.34, p = 0.56). The backcross population segregated genotypically 52 presence of the ‘CA60’ Rdr1A allele: 17 absence of the ‘CA60’ Rdr1A allele (χ2 3:1 = 0.01, p

= 0.94). In the same manner, the INDEL marker was able to distinguish between progeny carrying the ‘CA60’ Rdr1A allele and progeny without the allele for the mapping population.

The individuals that carried the ‘CA60’ Rdr1A allele (Rrrr) displayed a unique HRM profile with two peaks in contrast to a single peak for the individuals that did not carry the allele

(rrrr) (Figure 2-5 B). The INDEL marker further allowed an estimation of the number of copies of the ‘CA60’ Rdr1A allele present in the progeny of the backcross population. The melt curves with one peak at 79.5⁰C corresponded to the nulliplex genotype rrrr and the melt curves with two peaks at 79.5⁰C and 81⁰C corresponded to the simplex genotype Rrrr. The

duplex genotype RRrr was associated with one melting curve peak at 81⁰C (Figure 2-5 B).

The genotypic results obtained from the presence/absence marker in the mapping population and backcross population correlated 100% with the genotypic data obtained from the INDEL marker. The ‘CA60’Rdr1A marker mapped to LG1.H3 at 17.6 cM, between markers

SChr1_67051139 and SChr1_65811939 (Figure 2-6). QTL analysis was performed on the new map created and QTL associated with resistance to the four sources of BS infection

(VHy12-4, VSKO4, BOO5 and field disease pressure) co-located on chromosome 1. The

QTL confidence intervals included the ‘CA60’Rdr1A marker (Figure 2-6). There was evidence of recombination between the ‘CA60’Rdr1A marker genotype and the resistance phenotype for all four sources of BS infection (Table 2-7).

Figure 2-3. Discovery of a polymorphism in ‘CA60’ that is unique to muRdr1Agene in the Rdr1locus and Primer Design. This image shows the region of muRdr1A unique to BS resistant ‘CA60’ when compared to ‘SITR’. Cultivar ‘SITR’ has no reads covering the unique region, while hybrid ‘CA60’ does. The primers Forward Primer and Reverse Primer 1 are used as a presence/absence PCR marker designed around the polymorphism identified in ‘CA60’. The pair of primers Forward Primer and Reverse Primer 2 are used as an INDEL marker.

Figure 2-4. Multiple sequence alignment of all nine paralogues of the Rdr1 locus around the ‘CA60’ unique INDEL (underlined). The ‘CA60’ consensus sequence was obtained by aligning the ‘CA60’ Whole Genome Sequence reads to the entire Rdr1 locus.

Figure 2-5. High resolution melt profiles of markers amplified in segregating populations based around the 32bp INDEL polymorphism present in the muRdr1A gene of ‘CA60’ (Rrrr), SITR (rrrr) and ‘S13-10’ (Rrrr). A primer within the INDEL resulted in the presence or absence of a PCR product in the A) ‘CA60’ x ‘SITR’ population and C) ‘CA60’ x ‘S13-10’ population. Primers flanking the INDEL amplify all alleles in the B) ‘CA60’ x ‘SITR’ populations and in the D) ‘CA60’ x ‘S13-10’ populations.

Figure 2-6. A genetic map of ‘CA60’ LG1.H3 arising from the ‘CA60’ x SITR population that includes QTL for infection to four isolates of black spot and natural black spot infection under field conditions. The genetic map includes a marker derived from the polymorphism between black spot segregating parents ‘CA60’ and ‘SITR’ at the muRdr1A gene and shows alignment of the ‘CA60’Rdr1A allele-specific marker with the co-locating QTL for different infections of black spot

Population ‘CA60’ Rdr1A allele Dosage Phenotype Sources of infection

VHy-12.4 VSKO4 BOO5 Field ‘CA60’ x ‘SITR + R 60 54 56 57 - S 73 44 43 72 + S 1 1 5 - R 1 5 10 ‘CA60’ x ‘S13-10’ + Rrrr R 33 + RRrr R 11 _ rrrr S 13 _ rrrr R 1 + Rrrr S 1 Resistant allele + presence – absence ; Phenotype: (R) Resistant (S) Susceptible

2.4 Discussion

2.4.1 Race characterization and identification of a new race of D. rosae

This research used four D.rosae isolates, including three isolates from Vineland (VHy-

12.4, VSKO4 and VOTB17-1) that represented the diversity of D.rosae in the Vineland experimental field. The isolate VSKO4 presented a new and repeatable infection pattern on the set of differentials (Table 2-2). Multiple single-spore isolates were isolated from infected roses in

Vineland experimental farm and up to 7 isolates shared this infection pattern on the set of differentials (Supplementary Table 2-1 – Appendix 2). Therefore, we propose to name isolate

VSKO4 as race 14 following the initial 13 races (Whitaker et al. 2010 a, b; Zurn et al. 2018).

Whittaker et al. (2010) described the infection pattern of race 7 without infection on Knock

Out®. However, the isolate R6 obtained from the University of Minnesota previously described as race 7 infected Knock Out® in the current study and otherwise showed the same infection pattern on the set of differentials. Isolate VOTB17-1 showed the same infection pattern on differentials as R6 in the current study, therefore we conclude that isolate VOTB17-1 is race 7, and we suggest that the infection pattern on the set of differentials may be updated.

2.4.2 Development of molecular markers linked to BS resistance

The segregation ratio between field BS resistant and susceptible genotypes in the mapping population fit a 1:1 ratio and a 3:1 ratio in the backcross population. The 1:1 segregation ratio is expected under the presence of a single dominant gene in simplex configuration in one of the parents, under tetrasomic inheritance (Rrrr x rrrr). The 3:1 segregation ratio is in agreement with the presence of one gene in simplex configuration present in both parents of the backcross population ‘CA60’ x ‘S13-10’ (Rrrr x Rrrr) (Koning-Boucoiran et al. 2012). Therefore, phenotypic segregation within the mapping population and the backcross

population supported a single R gene model for BS resistance (Table 2-4). Furthermore, a multi- isolate QTL involved in resistance to D. rosae races 5, 10 and 14 and field-level resistance and a

QTL involved in resistance to D. rosae race 7 were both identified on chromosome 1. Given that the inheritance pattern and the QTL results both implicated single major genes; the previously identified Rdr genes from chromosome 1 were all candidates for the genes underlying these two

QTLs. Both Rdr1 and Rdr2 are located on chromosome 1 in the rose genome (Kaufmann et al.

2003; Biber et al. 2010), while Rdr4 is located on chromosome 5 (Zurn et al. 2018) and Rdr3 is not mapped. Although little is known about Rdr2, the Rdr1 gene locus has been sequenced and well characterized and confers broad spectrum resistance to multiple BS races including races 10 and 5 (Menz et al. 2018). QTL analysis and sequence alignment between the Rdr1 locus (Terefe-

Ayana et al. 2011) and the whole genome of R. chinensis suggested that the candidate genes involved in resistance to BS races 5, 10 and 14 and field-level resistance were located within the

Rdr1 locus (Figure 2-2) (von Malek and Debener 1998; Kaufmann et al. 2003; Biber et al. 2010;

Terefe-Ayana et al. 2011; Menz et al. 2018). The resistance to BS race 7 was located outside the

Rdr1 locus which is consistent with previous research (Menz et al. 2018).

The Rdr1 locus contains nine TIR-NBS-LRR (Toll/interleukin-1 receptor-nucleotide binding site-leucine rich repeat) (TNL) resistance genes, muRdr1A-I, ranging between 4085 and

5920 bp, each with four exons and three introns. The sequence similarities between those nine paralogs vary between 87.8% and 99.5% (Kaufmann et al. 2010; Terefe-Ayana et al. 2011).

Terefe-Ayana and colleagues (2011) investigated the expression of the Rdr1 gene family members in different rose organs and concluded that only five genes (muRdr1A, C, G, H and I) were expressed in leaves and petals and their functions were studied in transient infiltration experiments. The expression of muRdr1H was associated with a significant reduction in the

number of D. rosae colonies at the surface of the rose leaves in all experiments and muRdr1H was proposed to be the active gene in the Rdr1 locus. The expression of muRdr1A was associated with a decrease in fungal infection in a limited number of experiments only (Terefe-Ayana et al.

2011). Menz and colleagues further analyzed the muRdr1A and muRdr1H genes in transgenic rose plants and their sexual progeny. Incompatible interactions between the host and multiple races of D. rosae were associated with the presence of the muRdr1A transgene in genetically transformed BS susceptible rose cultivar. Genetically transformed BS susceptible rose cultivar containing the muRdr1H transgene remained BS susceptible. Therefore, the muRdr1A gene was suggested to be the gene that confers BS resistance (Menz et al. 2018). In this study, we re- sequenced the current mapping parents in order to search for polymorphisms in the muRdr1A gene. The parental genotype ‘CA60’ showed an INDEL sequence of 32bp in exon 4 of muRdr1A that was not found in the susceptible cultivar ‘SITR’ (Figure 2-3). This region was unique to muRdr1A based on a multiple sequence alignment of all of the nine TIR-NBS-LRR genes in the

Rdr1 locus (Figure 2-4). The identification of an INDEL in the muRdr1A gene in the resistant parent that is absent in the susceptible parent was exploited to develop diagnostic markers linked to resistance to multiple isolates of black spot. The presence of limited recombinants between the

‘CA60’ Rdr1A allele-specific marker and the resistance phenotype did not support that muRdr1A was the active gene conferring resistance to multiple isolates as suggested by Menz and colleagues (2018) (Table 2-7). The presence of recombinants for the different isolates suggested that additional R-genes other than muRdr1A, are likely functional within the ‘CA60’ Rdr1 locus and confer resistance to different D. rosae isolates.

2.4.3 Utility and source of the ‘CA60’ Rdr1A BS resistance allele

In this research, the polymorphic region of the muRdr1A gene was a good target to develop

PCR markers to distinguish BS susceptible from BS resistant progeny. The high correlation between field resistance and marker data is a good indicator of how robust the markers are. The markers show very limited amounts of recombination between the INDEL sequence and the BS resistance/susceptibility phenotype for multiple BS races (Table 2-7).

The ‘CA60’Rdr1A allele-specific diagnostic markers are robust and can be used to select accurately for BS resistant progeny derived from ‘CA60’. ‘CA60’ is a key parent in the Vineland breeding program as it possesses a number of favourable traits such as good hip setting and a desirable level of disease resistance and cold hardiness. Multiple BS resistant parental genotypes used in rose breeding at Vineland are derived from ‘CA60’ and crosses between those elite lines and BS susceptible rose cultivars with desirable ornamental features are productive for rose improvement. Therefore, markers developed in this study in the muRdr1A gene can be used efficiently in the progeny derived from those crosses to eliminate BS susceptible progeny from further testing and assure the breeder that the progeny posses field-level resistance to multiple

BS races.

‘CA60’ was an experimental hybrid from the Morden Research Station, Canada and was created from a cross between ‘RSM 104’ and an Explorer rose ‘Frontenac’ that was bred by Dr.

Felicitas Svejda (Canada, 1981). Explorer roses are known for their increased levels of cold hardiness and resistance to pathogens inherited from wild rose species (Vukosavljev et al. 2013).

‘Frontenac’ is a hybrid kordesii that would contain both R. rugosa and R. wichurana in its genetic background, two rose species known for their exceptional vigor and disease resistance.

The diagnostic markers designed in this study were applied to ‘Frontenac’ DNA and both

markers revealed that ‘Frontenac’ did not carry the ‘CA60’Rdr1A allele (data not shown). The de novo assembly of ‘CA60’ at the Rdr1 locus (Terefe-Ayana et al., 2011) did not provide a robust comparison with the published muRdr1 locus. Additional molecular data provided evidence that the Rdr1A gene present in ‘CA60’ was identical to the R. multiflora muRdr1A gene. Therefore, those results suggested that the parental genotype ‘RSM 104’ might be carrying the source of the

‘CA60’ Rdr1A allele. However, no additional information on ‘RSM 104’ is available except that it was a German selection whose other name is known as ‘German selection 91-104-1’

(www.helpmefind.com).

2.5 Conclusions

QTL mapping for BS resistance in a bi-parental tetraploid rose populations identified a major multi isolate resistance QTL that co-localized with the known Rrd1 locus. Evidence was obtained for a 32bp INDEL region in the muRdr1A gene that was linked to BS resistance. Two markers were developed around the INDEL specific to the muRdr1A gene. The ‘CA60’ Rdr1A allele-specific markers developed in this study were tightly linked to BS disease resistance to isolates BOO5 (race 5), VHy-12.4 (race 10) and VSKO4 (newly identified race 14). Resistance to isolate VOTB17-1 (race 7) did not co-locate on the genetic map with resistance to the other four sources of BS infection, suggesting that another R-gene outside the Rdr1 locus is involved.

The markers developed will be useful to implement MAS for the ‘CA60’ Rdr1A allele. This work represents an important step toward implementation of MAS for disease resistance in rose breeding programs.

CHAPTER 3: MAPPING COLD HARDINESS IN GARDEN ROSES (ROSA x

HYBRIDA)

Rouet Cindy*1,2, Joseph O’Neill1, Travis Banks1, Karen Tanino3, Elodie Derivry4, Daryl

Somers5, Elizabeth Lee2

1 Vineland Research and Innovation Centre, 4890 Victoria Ave N, Lincoln, ON L0R 2E0,

Canada; 2 University of Guelph, Plant Agriculture Department, 50 Stone Rd E, Guelph, ON N1G

2W1, Canada; 3 University of Saskatchewan, Department of Plant Sciences, 51 Campus Dr,

Saskatoon, SK S7N 5A8, Canada; 4 APREL, Association Provencal de Recherche et d’Expérimentation Légumière, Saint-Rémy-de-Provence, France; 5 Somers Consulting, Ontario,

Canada

Authors’ contributions: CR wrote the manuscript, CR, DS and EL conceived the experiments;

CR, ED and KT collected data; CR analysed and interpreted the data; JO and TB brought bioinformatics support. All the authors reviewed the manuscript.

Abstract Field winter hardiness is a critical trait in new rose varieties for northern climates.

While the molecular basis of cold hardiness has been well documented in model organisms such as Arabidopsis, little is known about the genetics and mechanisms underlying winter hardiness in roses. This chapter aims to explore the genetic control of winter hardiness for application in breeding programs using QTL analysis in two bi-parental rose populations derived from exceptionally cold hardy Canadian Explorer roses. Field winter hardiness was assessed as a complex trait with winter damage and regrowth recorded in multi-year and multi-location trials in Ontario and Saskatchewan, Canada. In addition, this research explored the relationship between field measurements and electrolyte leakage recorded under artificial conditions.

Electrolyte leakage had limited utility for application in rose breeding programs as a substitute

for field evaluation but did enable identification of QTLs associated with potential cold hardiness candidate genes. Due to the complexity of field winter hardiness and its direct reliance on intertwined factors, such as overall plant health, moisture status, snow cover, period of prolonged sub-zero temperatures, field trials are the ultimate measurement of field winter hardiness.

Transgressive segregation was observed for all traits, and it was most likely due to complementary gene action. Field winter damage and regrowth were highly heritable in single environments, but they were subject to genotype-by-environment interaction with severe climatic conditions in Saskatchewan and pest pressure.

Additional Index Words: Cold hardiness; Freezing tolerance; Electrolyte leakage; Heritability;

BLUP; QTL; Woody Perennials; black spot

3.1 Introduction

Cold winters limit the cultivation of roses in northern climates and require rose varieties that are able to survive prolonged periods of sub-zero temperatures. Different factors influence perennial crops’ survivability such as snow cover, moisture status, disease tolerance, insect damage, pathogen-induced defoliation, overall plant health and environmental adaptation

(Bélanger et al. 2006). Environmental adaptation includes dormancy and acquisition of maximum cold tolerance, which relies on the timing of acclimation and de-acclimation

(Wisniewski et al. 2014). Plant breeders can advance both disease tolerance and environmental adaptation by exploiting existing genetic diversity. Most existing cultivars' climatic adaptability ranges are indicated by their USDA winter hardiness zone. While most tender roses such as R. chinensis, Hybrid Teas and Floribunda are not able to withstand temperatures below -20˚C

(USDA zones 7 and warmer - Southern US), many shrub roses are able to withstand -45˚C

(USDA zones 2 and warmer - northern US and into the Canadian Prairies). Most particularly,

Canadian roses are known worldwide for their exceptional hardiness. They were created between

1960 and 2016 as part of three cold hardy rose collections: the Explorer Series Collection, The

‘Parkland’ Series and the ‘Canadian Artist’ Series. The Explorer Series Collection was created in the 1960’s in Ottawa (Ontario) via interspecific crosses from rose species R. wichurana, R. rugosa, R. kordesii, R. laxa, R. spinossisima and R. acicularis. The roses from this collection are truly winter hardy as they can survive -45°C with minimal damage to the canes and show good spring regrowth. Although they display exceptional cold hardiness, the Explorer roses are morphologically similar to the wild ancestors, thereby lacking desirable aesthetic features. The

‘Parkland’ roses were bred from the 1960’s to the 1990’s in the Canadian prairies at Morden

(Manitoba). They possess desirable flowering behaviours, such as recurrent blooming; however,

they are genetically distinct from the Explorer roses and not as hardy (Vukosavljev et al. 2013).

In fact, they genetically resemble modern roses rather than the rose species R. spinossisima and

R. wichurana whose genetic contribution faded away after many generations of hybridizations.

The ‘Parkland’ roses are able to survive -35˚C, provided that they are protected by a snow cover, but they die-back to the snow line or to the crown, expressing different strategies of winter survival and environmental adaptation in comparison to the Explorer roses. Although cold hardiness is overall considered as a multigenic trait (Guy 1990; Wisniewski et al. 2014), the trait in roses is thought to be controlled by a few major genetic factors and to be highly heritable

(Svejda 1979). As a result, large gains from selection can be made when hybridizing hardy parental genotypes (Zlesak 2007). A recent Quantitative Trait Loci (QTL) mapping study suggested the potential existence of a major QTL for cold hardiness in an Explorer genetic background on linkage group 5, which also displayed the highest genetic differentiation between

Canadian and European roses (Vukosavljev, 2014). However, this QTL was not validated.

Further efforts are needed towards the identification of robust QTLs for winter hardiness in roses to improve the efficiency of breeding schemes through the development of molecular markers and the pyramiding of these QTLs in elite roses.

The power and robustness of QTL detection and the development of new genomic tools rely upon the experimental design and the accuracy of the phenotyping; however, winter hardiness is a complex trait that is challenging to define and assess. Our ability to accurately assess a genotype’s true winter hardiness is challenging as it is affected by many abiotic and biotic factors and a successful plant acclimation process. While winter hardiness in roses is traditionally evaluated in the field on a scale from 0 to 5 based on the overall level of damage

observed after several winters across multiple locations, cane die-back, freezing tolerance and spring regrowth are all components of field winter hardiness that need to be considered together and recorded separately (Zlesak et al. 2017). In fact, freezing tolerance is a trait of important agronomic interest (Gery et al., 2011), and screening methods under artificial conditions for freezing tolerance would complement field-screening approaches for QTL detection. Plant electrolyte leakage, which is a measurement of membrane damage after subjecting a plant organ to freezing treatment, has been successfully used as an index of freezing tolerance for various crops, such as alfalfa (Dexter et al. 1930), winter wheat (Săulescu and Braun 2002; Willick et al.

2019), durum wheat (Bajji et al. 2002), walnut (Poirier et al. 2010), peas (Dumont et al. 2009), red raspberry (Linden et al., 2000), switch grass and Miscanthus (Peixoto and Sage 2016), hybrid poplar (Kalcsits et al. 2009) and roses (Karam and Sullivan 1991; Le et al. 2012; Ouyang et al.

2019). Therefore, electrolyte leakage has the potential to be used in QTL mapping as a quantitative and objective measurement of frost damage for application in a rose breeding program to select for increased freezing tolerance and minimal cane die-back without the need for field screening.

Robust QTL detection also leads to the identification of candidate genes. The genetics of cold hardiness has been well documented in model organisms such as Arabidopsis. In temperate regions, the perception of the cold signal at the early stage of acclimation induces a cascade of changes in gene expression and metabolism pathways that leads to the acquisition and the support of cold hardiness (Akhtar et al. 2012). For instance, cold-induced metabolic changes in roses include a decrease in starch content, an increase in oligosaccharides and an increase in sucrose content in the cells (Ouyang et al. 2019). At the genetic level, the CBF pathway is known

to play a critical role in the acquisition and regulation of plant cold hardiness in Arabidopsis thaliana (Novillo et al. 2007). Part of the CBF pathway is thought to be conserved among plant species, but the regulation of CBF genes is more complex in woody perennials that in herbaceous plants (Wisniewski et al. 2014). CBF and additional transcription factors, such as ICE1, are involved in the activation of a subset of COld-Regulated (COR) genes. Although the expression of a small number of proteins involved in cold hardiness, such as dehydrins, has been studied in roses (Ouyang et al. 2019), little is known about the genetics and molecular basis underlying winter hardiness in roses. The expression of few CBF/DREB-like proteins have been studied in various Rosaceae crops, such as peaches (Wisniewski et al. 2011), almonds (Barros et al. 2012), sweet cherries (Kitashiba et al. 2004), and Japanese plums (Zhao et al. 2018), so it is reasonable to assume that these proteins are present in roses as well. In fact, sequences for CBF/DREB and

ICE1-like transcription factors can be identified in the R. chinensis reference genome

(Supplementary Table 3-1 – Appendix 7) (Raymond et al. 2018), but their role in rose winter hardiness remains unknown. Most of the QTLs involved in freezing tolerance in Arabidopsis are located in the CBF gene regions or nearby (Gery et al. 2011), highlighting the important role of the CBF-pathway in the acquisition of cold hardiness and setting up a framework for the identification of candidate genes in linkage mapping research.

The aim of this research is to provide rose breeders with the necessary tools to most efficiently develop winter hardy roses. The objectives are to: 1) investigate the utility of electrolyte leakage as a measure of cold hardiness in rose breeding material, 2) measure electrolyte leakage in two mapping populations with different levels of winter hardiness and identify QTLs associated with electrolyte leakage, 3) conduct multi-year multi-location field

trials to measure winter damage and regrowth in the same two mapping populations and identify

QTLs for regrowth and winter damage, and 4) compare both winter hardiness inferred by electrolyte leakage and by field screening in the breeding material.

3.2 Material and Methods

This study comprised three experiments: 1) electrolyte leakage experiments on a small set of commercial and elite roses (Experiment 1); 2) electrolyte leakage experiments on the parental genotypes and large-scale electrolyte leakage experiments on two mapping populations with

QTL mapping of electrolyte leakage (Experiment 2); and 3) multi-year, multi-site winter hardiness field trials with QTL mapping for field winter damage and spring regrowth

(Experiment 3).

3.2.1 Genetic material

3.2.1.1 Experiment 1_ Electrolyte leakage as a proxy for winter hardiness

Experiment 1 was conducted on a set of 28 roses consisting of eight elite roses from the

Vineland Research and Innovation Centre’s (VRIC) breeding program (Ontario, Canada), one breeding line and 19 commercial rose cultivars that were a mixture of Explorer roses,

Floribunda, Hybrid Tea and Grandiflora and that varied in their level of winter hardiness (Table

3-1). All the roses were grown on their own roots.

Table 3-1. Genetic material used in this multi-experiment study

Genotype USDA Ploidy Type zone Experiment 1_Electrolyte leakage as a proxy for winter hardiness ‘Cardinal Song’ 7b * Grandiflora ‘Caroline de Monaco’ 7b * Hybrid Tea ‘Desert Peace’ 7b * Hybrid Tea ‘Gentle Giant’ 6b * Hybrid Tea ‘George Vancouver’ (‘GV’) 3b 4x Hybrid Kordesii, Shrub. Explorer Series Collection ‘John Cabot’ 2b 4x Hybrid Kordesii. Explorer Series Collection ‘John Davis’ 2b 3x Hybrid Kordesii. Explorer Series Collection ‘John Franklin’ 3b 4x Hybrid Spinosissima, Shrub. Explorer Series Collection Knock Out® 4b 3x Shrub rose. Earth Kind ™ ‘Lambert Closse’ 2b 4x Shrub (Floribunda X Hybrid Kordesii). Explorer Series Collection ‘Martin Frobisher’ 2b 2x Hybrid Rugosa. Explorer Series Collection ‘Peace’ 5b 4x Hybrid Tea ‘Poseidon’ 5b * Floribunda ‘Quadra’ 3b * Hybrid Kordesii, Shrub. Explorer Series Collection ‘Salmon Vigorosa’ 5b * Floribunda, Shrub, County Series Collection, Vigorosa® Collection ’Singing in the Rain’ (‘SITR’) 6b * Floribunda

‘William Baffin’ 2b 4x Climber, Hybrid Kordesii, Shrub. Explorer Series Collection ‘William Booth’ 2b 4x Hybrid Kordesii, Shrub. Explorer Series Collection ‘Yellow Submarine’ 4b * Climber, Shrub. Garden Art ® Collection S13-10 * Elite rose from VRIC S13-29 * Elite rose from VRIC S13-3 * Elite rose from VRIC S13-32 * Elite rose from VRIC S13-35 * Elite rose from VRIC S13-6 * Elite rose from VRIC S13-7 * Elite rose from VRIC S13-8 * Elite rose from VRIC ‘CA60’ * Hybrid derived from Explorer ‘Frontenac’ Experiment 2_QTL mapping of electrolyte leakage ‘CA60’ * Hybrid derived from Explorer ‘Frontenac’ ‘SITR’ 6b * Floribunda ‘Easy Does It’ (‘EDI’) 5 * Floribunda and Hybrid Tea ‘GV’ 3b 4x Hybrid Kordesii, Shrub. Explorer Series Collection. Mapping population ‘CA60’ x ‘SITR’ * Mapping population ‘EDI’ x ‘GV’ * ‘CA60’ * Hybrid derived from Explorer ‘Frontenac’ ‘SITR’ 6b * Floribunda

Experiment 3_ QTL mapping of field-based winter hardiness Mapping population ‘CA60’ x ‘SITR’ * Mapping population ‘EDI’ x ‘GV’ * ‘Caroline de Monaco’ 7b * Hybrid Tea ‘Frontenac’ 2b 4x Hybrid Kordesii, Shrub. Explorer Series Collection ‘Gentle Giant’ 6b * Hybrid Tea ’Nicolas’ 3b 4x Shrub. Explorer Series Collection ‘EDI’ 5 * Floribunda and Hybrid Tea ‘GV’ 3b 4x Hybrid Kordesii, Shrub. Explorer Series Collection ‘CA60’ * Hybrid derived from Explorer ‘Frontenac’ ‘SITR’ 6b * Floribunda * ploidy unknown

3.2.1.2 Experiment 2_ QTL mapping of electrolyte leakage

Two mapping populations and their parental genotypes grown on their own roots were used in this experiment (Table 3-1). The first mapping population was a tetraploid rose population of 365 F1 progeny created between 2015 and 2016 from the cross between the cold hardy female parent ‘CA60’ and the cold susceptible male parent ‘Singing in The Rain’

(‘SITR’). The cold hardy female ‘CA60’ originated from the Morden rose breeding program and was bred from the exceptionally cold hardy Canadian Explorer rose ‘Frontenac’ (Ogilvie and

Canada. Agriculture Canada. Communications Branch 1993). ‘CA60’ has demonstrated exceptional winter hardiness in the field at Vineland, ON, Canada (Vineland, ON, 43°11'30.9"N

79°23'45.7"W). A subset of this large population was used in the experiments. The second mapping population consisted of 107 individuals and resulted from the 2016-cross between the cold susceptible female parent Easy Does It® ('HARpageant') (‘EDI’) and the cold hardy male parent ‘George Vancouver’ (‘GV’). ‘GV’ was bred by Dr. Felicitas Svejda and belongs to the

Explorer Rose Collection.

3.2.1.3 Experiment 3_ QTL mapping of field-based winter hardiness

The two mapping populations described above, ‘CA60’ x ‘SITR’ and ‘EDI’ x ‘GV’, and their parental lines were used in this experiment. Additionally, four cultivars that were a mixture of Hybrid Teas and Explorer roses were used as controls in this experiment: cold sensitive

‘Caroline de Monaco’ and ‘Gentle Giant’, and cold hardy ‘Frontenac’ and ’Nicolas’ (Table 3-1).

The mapping populations were grown on their own roots. The parental lines ‘SITR’ and ‘EDI’, and the controls were obtained from a local nursery and grown on R. multiflora rootstock. The parental line ‘GV’ was grown both on its own roots and grafted on R. multiflora rootstock (Table

3-1).

3.2.2 Experimental design and growing conditions

3.2.2.1 Experiment 1_ Electrolyte leakage as a proxy for winter hardiness

The plants used in this experiment were propagated at once from cuttings in early spring of 2016 using Oasis® strips in the mist house, potted into two gallon pots in a soilless potting mix (Sungro Sunshine® Mix #1) and grown outdoors. Roses were manually watered daily and fertilized two to three times per week with 20-8-20 and 14-0-14 fertilizers (Plant-Prod®). In brief, the electrolyte leakage assay consisted of three main steps: two-weeks whole-plant acclimation at 4˚C, freezing treatment of stem segment from acclimated plants, and measurement of conductivity before and after autoclaving of the rose tissues subjected to a range of sub-zero temperatures. Experiment 1 was replicated between one and three times depending on the genotype (i.e. replications in time) using unique sets of clones for each replication (i.e. true biological replications) (Supplementary Table 3-2 – Appendix 8), and it was conducted from

May to August 2016. In addition, the 19 rose cultivars were planted in 2017 in the field at the

Vineland experimental farm (VRIC) (Vineland, ON, 43°11'30.9"N 79°23'45.7"W) in a sandy loam soil — Gleyed Brunisolic Gray Brown Luvisol soil — using a completely random design

(CRD) with three replications. Plants were spaced by 50cm within rows, and the rows were spaced by 1,50m. All genotypes were grown on a R. multiflora rootstock. The field data used on the eight elite roses were historical data collected in 2013 in Saskatoon (SK) and Olds Alberta

(OA) as part of the Pan Canadian testing, which can be described as a network of growers and university partners across Canada who evaluate VRIC’s selected roses.

3.2.2.2 Experiment 2_ QTL mapping of electrolyte leakage

The plants used in this experiment were propagated from cuttings between August 2017 and January 2018 — ‘CA60’ x ‘SITR’ population and its parents — and from January to March

2019 — ‘EDI’ x ‘GV’ population and its parents —. The propagation was staggered over a few months to manage a variable rooting success rate, but the age of each cutting was not recorded.

The cuttings were propagated in the mist house directly into a soilless potting mix (Sungro

Sunshine® Mix #1) or using Oasis® strips. The mist house was heated at 22°C; the temperature varied between 20°C and 25°C but rose to 35°C during the summer months. The relative humidity was comprised between 60-75% and the misting system was activated every 5 minutes during a 16h/ day photoperiod. High Pressure Sodium supplemental light was provided to maintain the photoperiod at 16h/ day. Cuttings were transferred into four inch pots between two to four weeks after propagation before being transplanted into three gallon pots. Plants were grown on their own roots and maintained in the greenhouse from September 2017 to October

2018 — ‘CA60’ x ‘SITR’ population — or outdoors from May 2019 to October 2019 — ‘EDI’ x

‘GV’ population —. For the ‘CA60’ x ‘SITR’ population, temperature in the greenhouse was maintained at an average of 24°C, but diurnal variation of temperature occurred with a maximum–minimum temperature range of 30/18°C; 400W High Pressure Sodium supplemental light was provided from October 2017 to May 2018 to maintain the photoperiod at 16h/ day. For the ‘EDI’ x ‘GV’ population, average outdoor temperature in 2019 varied from 12°C in May,

23°C in the hottest summer months and 12°C in October. All plants were watered with the use of irrigation strips and fertilized two to three times per week with 20-8-20 and 14-0-14 fertilizers

(Plant-Prod®). Only sulfur was applied to control powdery mildew on the plants, no additional chemicals were used. Biocontrol was used to control insects’ pressure.

A total of 100 genotypes from the population ‘CA60’ x ‘SITR’ and 88 from the population ‘EDI’ x ‘GV’, in addition of the parental genotypes ‘CA60’, ‘SITR’, ‘EDI’ and ‘GV’, were used in electrolyte leakage assays. In preparation of the electrolyte leakage experiments, the

plants from each mapping population were moved at once to a greenhouse maintained at an average temperature of 10˚C (i.e. pre-acclimation phase) where diurnal variation of temperature occurred with a maximum–minimum temperature range of 25/8°C with 16h dark/ 24h and no supplemental light. Plants were then subjected to a two-week artificial acclimation period in a walk-in cooler at 4°C (KeepRite®) in which two LED tripods (Husky – Ultra Products Corp –

Model DE007) and two 100-feet metal cage string lights with LED bulbs (E253217 Model LS-

100) delivered 10 µmol photons s-1 m-2 of light at canopy level for 6h/day. Light measurements were obtained using a light meter with a Quantum sensor (LI-COR® LI-250A). The F1 progeny from the two mapping populations were randomized in the cooler using the online plant breeding platform Phenome (Phenome Networks Ltd ©). Stem sections (i.e. experimental unit) from acclimated plants were then subjected to freezing treatments in a programmable freezer. This experiment used a split-plot design in which temperature represented the main-plot factor and genotypes were assigned to the subplots.

Electrolyte leakage experiments on the parental lines, the population ‘CA60’ x ‘SITR’ and the population ‘EDI’ x ‘GV’ were conducted separately. Experiments on the parental lines were conducted in November 2019 and replicated three times with unique sets of clones (i.e. true biological replications) with three stem sections per genotype and temperature (i.e. technical replications) and one week between each replication (i.e. replications in time). A fourth replication was conducted on ‘CA60’ and ‘GV’ only, using a distinct set of clones two weeks after the third replication. Clones of the parental genotypes were moved to the pre-acclimation house in October 2019. Experiments on the population ‘CA60’x’SITR’ were conducted from

November 2018 to January 2019 and replicated five times (i.e. replications in time) using five stem sections per temperature and per genotype (i.e. technical replications) and three unique sets

of clones (i.e. true biological replications), two of which were used in two replications and were allowed to de-acclimate in a greenhouse at room temperature for two weeks before being reused.

There were two weeks between each replication, at the exception of three weeks between replications 3 and 4. Plants from the ‘CA60’ x ‘SITR’ population were moved to the pre- acclimation house in October 2018. Eventually, experiments on the population ‘EDI’ x ’GV’ were conducted from January 2020 to February 2020 and replicated three times (i.e. replications in time) with unique sets of clones (i.e. true biological replications), with two weeks between each replication. Only one stem section per temperature and genotype was used in the experiment on the ‘EDI’ x ‘GV’ population. Plants from the ‘EDI’ x ‘GV’ population were moved to the pre-acclimation house in October 2019.

3.2.2.3 Experiment 3_ QTL mapping of field-based winter hardiness

The population ‘CA60’ x ‘SITR’ was propagated from cuttings using Oasis® strips in the mist house between August 2017 and January 2018 at the same time as the cuttings intended for electrolyte leakage experiments. As mentioned above, the propagation was staggered over a few months to manage a variable rooting success rate, but the age of each cutting was not recorded.

A total of 101 F1 progeny and nine control and parental genotypes were planted in two locations in Canada in early June 2018: Elora (ON) (43°38'27.4"N 80°24'03.2"W) (USDA zone 5b — cultivated Brunisol soil with silt loam) and Saskatoon (SK) (52°07'21.5"N 106°36'41.8"W)

(USDA zone 3b — Dark Brown Chernozemic soil). Roses were planted at each location using a randomized complete block design (RCBD) with five replications designed with the online plant breeding platform Phenome (Phenome Networks Ltd ©), the experimental unit being the rose bush. Plants were spaced by 60cm within rows, and the rows were spaced by 1,50m. Black heavy duty woven polypropylene landscape fabric and 130 yards of light coloured mulch, which was

applied on top of the landscape fabric, were used to control weeds in Elora, while only mulch was used in Saskatoon. Plants were gently pruned in the summer of 2019 to remove all the dead canes. The population ‘EDI’ x ‘GV’ was propagated from cuttings into a soilless potting mix in the mist house from January to March 2019 at the same time as the cuttings intended for electrolyte leakage experiments. A total of 96 F1 progeny and four parental genotypes were planted in late July 2019 in Elora using a randomized complete block design (RCBD) with three replications. Plants were spaced by 1m within rows to accommodate their spreading architecture, and the rows were spaced by 1,50m. Mulch was used to control weeds. The population ‘CA60’ x

‘SITR’ was assessed over two seasons (winters 2018-2019 and 2019-2020), while the population

‘EDI’ x ‘GV’ was assessed over only one season (winter 2019-2020). In addition, climate data for Saskatoon were retrieved from the Saskatchewan Research Council Website at the Saskatoon

Climate Reference Station (www.src.sk.ca/labs/climate-reference-stations), and climate data collected at the Elora Research Station were retrieved from the Environment Canada Website

(https://climate.weather.gc.ca/historical_data).

3.2.3 Phenotyping

3.2.3.1 Experiment 1_ Electrolyte leakage as a proxy for winter hardiness

Experiment 1 was conducted to assess rose freezing tolerance under artificial conditions according to Dexter et al. (1930) with various modifications to the authors’ original protocol.

After the acclimation period, rose stems from acclimated plants were collected, sprayed with

MilliQ water to nucleate ice formation and placed into Ziploc bags. The bags were moved to a programmable freezer in which the temperature dropped from 0°C to -20°C, at a rate of -

2.5°C/hour. One stem per genotype was removed from the freezer at 0°C, -5°C, -10°C, -15°C and -20°C. The stem removed from the freezer was then cut into ten 1cm-long stem sections and

each section placed into individual 20mL centrifuge tubes filled with milliQ water (i.e. technical replications). The tubes were shaken overnight before conductivity was measured and recorded

TM as ‘R0’ using a traditional electrical conductivity (EC) meter with a probe (Oakton CON 450).

The tubes were then autoclaved for 1h at 120°C to promote cell rupture and shaken again overnight. Conductivity was measured again and recorded as ‘Rt’. The percentage of electrolyte

푅표 leakage was calculated as: 퐸퐿 (%) = 100 ∗ . Additionally, field winter hardiness of the 28 푅푡 roses was recorded on a scale from 0 (no damage) to 5 (dead plant) in the spring of 2018 (VRIC)

— cultivars — and the springs of 2014 and 2015 (Alberta and Saskatchewan) — elite roses.

3.2.3.2 Experiment 2_ QTL mapping of electrolyte leakage

Experiment 2 was conducted as described in the previous section (3.2.3.1 Experiment 1_

Electrolyte leakage as a proxy for winter hardiness) with some modifications. The bags containing the rose stems were moved to a programmable freezer in which the temperature dropped from 0°C to -50°C (experiments on the parental lines), from 0°C to -20°C (experiments on the population ‘CA60’ x ‘SITR’) and from 0°C to -40°C (experiments on the population

‘EDI’ x ‘GV’), at a rate of -2.5°C/hour, with a 12-hour nucleation step at -4°C. The initial stem from the Ziploc bag that was taken out of the freezer was then cut into either three, five or one

1cm-long stem sections in the experiments conducted on the parental lines, the population

‘CA60’ x ‘SITR’ and the population ‘EDI’ x ‘GV’, respectively. Each section was placed into individual 5mL centrifuge tubes filled with milliQ water (i.e. technical replications). ‘R0’ and

‘Rt’ were recorded before and after autoclave as described above using a compact electrical conductivity meter (LAQUAtwin-EC-11) with 120 μL of solution. Data on the parental lines were recorded at 4˚C (i.e. control), -10°C, -15°C, -20°C, -25°C, -30°C, -35°C, -40°C, -45°C and

-50°C; data on the population ‘CA60’ x ‘SITR’ were recorded at -10°C, -15°C and -20°C; and data on the population ‘EDI’ x ‘GV’ were recorded at 4˚C (i.e. control), -10°C, -15°C, -20°C, -

25°C, -30°C, -35°C and -40°C. The percentage of electrolyte leakage was calculated as described above. In addition, an index of injury corrected for electrolyte leakage not due to freezing — measured at 4˚C — was estimated for the parental genotypes and the population ‘EDI’ x ‘GV’ by

푅0 푅표푟푒푓 − 푅푡 푅푡푟푒푓 the formula 퐼 (%) = 푅표푟푒푓 as described by Ouyang and colleagues (2019). 1− 푅푡푟푒푓

3.2.3.3 Experiment 3_ QTL mapping of field-based winter hardiness

Field winter hardiness was assessed as two components: winter damage and spring regrowth as described by Vukosavljev (2016). Winter damage was a measurement of cane die-

푙푑 back using the following formula: 푊퐷 (%) = 100 ∗ with WD being the percentage of winter 푙푡 damage, ld the length of dieback and lt the length of the whole cane. Three stems per plant were measured and the average value for each plant was used as the rating for winter damage.

푙푛 Regrowth was estimated with the following formula 푅퐺 (%) = 100 ∗ with RG being the 푙푡 percentage of regrowth, ln the length of the new shoot and lt the initial length of the whole cane.

Winter damage was measured in early spring when the leaf buds started to grow actively. Spring regrowth was measured during the second week of June in 2019 and first week of July in 2020 in

Elora and during the last week of July in 2019 and 2020 in Saskatoon. Regrowth value for each genotype was calculated as the average regrowth of three branches. For both winter damage and regrowth, the three longest stems from the crown were chosen. Plants that died through the winter of 2019 were assigned 100% winter damage for winter 2019 and missing data for winter damage 2020, regrowth 2019 and regrowth 2020. Plants that died in the spring 2019 as a result

of planting, drought or pest pressure were assigned missing data for winter damage and regrowth across years. Plants that died through the winter of 2020 were assigned 100% winter damage in

2020 and missing data for regrowth 2020. This was done to ensure that the timing of a plants’ death was accurately captured and that the ability of the meristem to resume growth in the spring even on highly damaged plants was well represented. Winter damage and regrowth data were collected across four environments (Elora 2019, Elora 2020, Sask 2019 and Sask 2020) for the population ‘CA60’x’SITR’ and one environment (Elora 2020) for the population ‘EDI’ x ‘GV’.

Pest — black spot (Diplocarpon rosae), rose slugs and adult sawflies (Tenthredinidae)— - induced defoliation was also recorded in September 2019 in Elora on a scale from 0 (no defoliation) to 5 (severe defoliation).

3.2.4 Data Analysis

3.2.4.1 Experiment 1_ Electrolyte leakage as a proxy for winter hardiness

The percentage of electrolyte leakage was averaged for up to three replications for each cultivar (Supplementary Table 3-2 – Appendix 8). The lethal temperature at which 50% of leakage occurred (LT50) was estimated using a logistic regression on electrolyte leakage data

푃 according to the formula: ln ( ) = 푎 + 푏푋 where P was the proportion of electrolyte leakage 1−푃 corrected for leakage not due to freezing (i.e. control) and X was the logarithm to base ten of the temperature. Correlations between electrolyte leakage, LT50, field data and USDA cold hardiness zone were computed using the Pearson method in the R studio software (R Studio

Team, 2020).

3.2.4.2 Experiment 2_ QTL mapping of electrolyte leakage

The analysis was conducted separately for the parental lines, the population ‘CA60’ x

‘SITR’ and the population ‘EDI’ x ‘GV’. Generalized linear mixed models (GLMM) were fitted to the electrolyte leakage data using SAS software (SAS ver. 9.4, SAS Institute Inc., Cary, NC,

USA), specifying a beta distribution and the logit link under the GLIMMIX procedure

(Supplementary Material – Appendix 6). Temperature was a fixed effect. Genotypes were considered as fixed effects when analysing only the parental lines, and the Best Linear Unbiased

Estimates (BLUE) were computed. Genotypes were considered as random effects in the analysis of the mapping populations, and the Best Linear Unbiased Predictors (BLUP) were computed. In addition, LT50 was estimated for the parental lines ‘CA60’, ‘SITR’ and ‘EDI’, ‘GV’ by fitting a non-linear regression dosage curve response on the index of injury with nine temperatures (-

10°C, -15°C, -20°C, -25°C, -30°C, -35°C, -40°C, -45°C and -50°C), using the nlmixed procedure in the SAS software. In the same way, LT50 was estimated individually for each sibling of the

‘EDI’ x ‘GV’ population, from the index of injury obtained from six temperatures (-15°C, -20°C,

-25°C, -30°C, -35°C and -40°C). Correlations between the BLUEs and the BLUPs of electrolyte leakage and index of injury, LT50, and field data were computed using the Pearson method in the R studio software. The index of injury was used to compute the LT50s; the BLUPs of electrolyte leakage and the LT50 were used in QTL mapping.

The female and male genetic maps of ‘CA60’ and ‘SITR’ were created from previously generated GBS data (Rouet et al. 2019) in order to include all individuals for which cold hardiness data were available. The female and male maps of ‘EDI’ and ‘GV’ respectively, were created from newly generated GBS data following a two-way pseudo-test cross mapping strategy, using R-package ASMap (Taylor & Butler 2017), as described by Li et al. (2014) and

100

applied on roses in Rouet al. (2019). The four parental genetic maps were used as a framework for QTL mapping. A total of 83 and 80 individuals were used to identify QTLs associated with electrolyte leakage from the ‘CA60’ x ‘SITR’ and ‘EDI’ x ‘GV’ populations, respectively. For the ‘EDI’ x ‘GV’ population, 76 individuals were also used to identify QTLs associated with

LT50.

QTLs were initially detected by interval mapping (IM) using the R/qtl2 package (Broman et al. 2019) and performing a genome scan by Haley-Knott regression (Haley and Knott 1992).

Then, results from IM were compared to a Multiple QTL Mapping (MQM) approach using the R implementation of Ritsert Jansen’s MQM method (Arends et al. 2010) (R v.3.5.3), and additional

QTLs were detected. Both forward stepwise and backward elimination strategies were employed to identify QTLs underlying the trait in MQM. The forward stepwise approach was conducted by setting the major QTLs detected in IM as cofactors. Variance components used in QTL modeling were estimated by the Restricted Maximum Likelihood (REML) approach. Significance cut-off was determined by performing 1000 permutations and the fitqtl function from the package R/qtl

(Broman et al. 2003) was applied to estimate the percentage of variation explained by the main

QTLs. Bayes interval were obtained with 95% confidence for the QTLs. Only the results from

MQM were reported. The relative position of the SNP markers in the reference genome R. chinensis ‘Old Blush’ V2 were used to identify QTL intervals.

3.2.4.3 Experiment 3_ QTL mapping of field-based winter hardiness

Descriptive statistical analyses were first conducted on field data for the two mapping populations using the R Studio software. The data were then modeled with a 3-D productivity contour map using SAS software (SAS ver. 9.4, SAS Institute Inc., Cary, NC, USA) and the spline method in the G3GRID procedure to investigate the presence of non-random error

101

distribution. A GLMM was fitted to field data using the SAS software, and the GLIMMIX procedure was used in order to conduct a genotype-by-environment interaction analysis with up to four environments (Elora 2019, Elora 2020, Sask 2019, Sask 2020). The variances of regrowth and winter damage for the field trial ‘CA60’ x ‘SITR’ with multiple environments were partitioned into random effects Genotype, Environment, Genotype * Environment and Block

(Environment). When environments were analyzed separately, the variances of regrowth and winter damage for the field trial ‘EDI’ x ‘GV’ and the field trial ‘CA60’ x ‘SITR’were partitioned into random effects: Block and Genotype. The linear predictors were respectively: ƞij

= ƞ + Ei + aj +(Ea)ij + b(E)ki where ƞ was the intercept, Ei the environment effect, aj the genotype effect, (Ea)ij the genotype by environment interaction and b(E)ki the block effect nested within environment, and ƞij = ƞ + bi + aj where ƞ was the intercept, aj the genotype effect, and bi the bloc effect. A type I error of 0.05 was used to determine the significance of tests in this analysis.

Random effects were assessed with a likelihood test for covariance parameters. Winter damage data across environments and from single environments Elora 2020 and Sask 2019 were fitted to a beta distribution with a logit link function and Laplace interval estimation method. Winter damage data from single environments Elora 2019 were fitted to a normal distribution with the identity link function. Regrowth data were fitted to a lognormal distribution with the identity link function. The best fitted model was chosen based on the analysis of residuals and the fit statistics such as the AIC. The models were surveyed to determine if setting a heterogeneous covariance structure was necessary. BLUP estimates were computed for each genotype in each environment.

Heritability estimates were obtained for each trait in single environments by estimating the genetic variance components from the completely random mixed model. In the case where the data fitted a normal or lognormal distribution in a completely random model, the Restricted

102

Maximum Likelihood (REML) was used to compute and retrieve the variance estimates. Broad

2 푉푔 2 sense heritability used the following equation: 퐻 = 푉푟 where H was the broad sense 푉푔+( ) 푟 heritability, Vg the genetic variance, Vr the residual variance, and r the number of replicates or blocs in the RCBD trial (Gitonga et al. 2014). Broad sense heritability was obtained from

REML-derived estimates whenever the data were approximately normally distributed as advised by Nakagawa and Schielzeth (2010). However, if the data fitted a beta distribution with a logit link function, the latent-scale heritability was obtained from the GLMM-based estimates

(Nakagawa and Schielzeth 2010). Under this scenario, the residual variance was calculated as:

휋2 푉푟 = ∅ + where ∅ was the scale parameter computed by the GLMM and represented the 3

휋2 over dispersion of the data under the latent scale, and represented the variance for the logistic 3 distribution. The latent-scale broad sense heritability was calculated by the formula: 퐻2 =

푉푔 푉푔푒 . In addition, the ratio was calculated to quantify the size of the genotype-by- 휋2 ∅ + 푉푔 푉푔+( 3 ) 푟 environment interaction relative to the genetic variance in mixed models using multiple environments (Gitonga et al. 2014).

Environment-metric preserving GGE (genotype plus genotype-by-environment interaction) biplots were generated using the BLUPs of field winter damage and regrowth in order to visually characterize simultaneously genotype by environment relationships using R

Studio and the packages GGEBiplot GUI (Frutos et al. 2014). The GGE biplot model uses the principle of singular value decomposition (SVD) to decompose genotype (G) and genotype-by- environment (GE) effects into two or more components represented on the graph axis. GGE biplots were generated using the column metric preserving option (SVP = 2), tester-centered

103

(Center = 2) and unscaled G + GE (Ling et al. 2021). BLUPs of winter damage and regrowth were also used in QTL mapping. A total of 83 and 86 individuals were used to identify QTLs in the population ‘CA60’ x ‘SITR’ and in the population ‘EDI’ x ‘GV’, respectively. The presence of QTLs was investigated and reported as previously described. Finally, pathogen-induced defoliation data recorded in Elora in 2019 were averaged for each genotype across five blocs, and used in correlation analysis with BLUPs of winter damage and regrowth and in mapping.

Climate data were visualized in R Studio using the package ggplot2 (Wickham 2016). The monthly number of freeze-thaw cycles were calculated as the number of days where the maximum temperature was higher than 0˚C and the minimum temperature was below -1˚C.

3.3 Results

3.3.1 Experiment 1_ Electrolyte leakage as a proxy for winter hardiness

Electrolyte leakage experiments conducted on 28 commercial cultivars and elite genotypes to investigate the relationship between electrolyte leakage and field winter hardiness on a small scale showed positive correlations between the different variables. Two of the genotypes (‘Martin Frobisher’ and ‘William Booth’) were inconsistent in their blooming patterns and were removed from the analysis due to concerns about correct identity. The USDA cold hardiness zone of the remaining cultivars was highly correlated with field winter hardiness

(Figure 3-1). In addition, the percentage of electrolyte leakage at -15°C and -20°C was strongly correlated with field winter hardiness and with the USDA cold hardiness zone (Figure 3-1). The

LT50 was also correlated with field winter hardiness and the USDA cold hardiness zone (Figure

3-2). LT50s of the commercial cultivars ranged from -20˚C to -10˚C, and as expected, the hardiest cultivars, such as ‘GV’, had the lowest LT50s (Supplementary Table 3-2 – Appendix 8).

104

When focusing on the elite genotypes, the percentage of electrolyte leakage at -15°C and at -

20°C was highly correlated with winter hardiness recorded in Olds Alberta and Saskatoon respectively (Figure 3-3). The LT50s estimated for the elite roses was strongly correlated with winter hardiness recorded in Saskatoon (Figure 3-3).

105

Figure 3-1. Correlations between USDA cold hardiness zones, winter hardiness (WH) as recorded on a scale from 0 to 5 at Vineland’s experimental farm in 2018 and electrolyte leakage (EL) measured at different temperatures in artificial freezing experiments for 17 rose cultivars. Correlations were computed using Pearson method. Only significant correlations are given, with * indicating significant correlation at p=0.05 and ** significant at p=0.01.

106

Figure 3-2. Correlations between USDA cold hardiness zones, winter hardiness (WH) as recorded on a scale from 0 to 5 at Vineland’s experimental farm in 2018 and LT50 estimated using a logit model approach for 17 rose cultivars. Correlations were computed using Pearson method. Only significant correlations are given, with * indicating significant correlation at p=0.05 and ** significant at p=0.01.

107

Figure 3-3. Correlations between field winter hardiness (WH) as recorded on a scale from 0 to 5 in two different Canadian locations (Olds Alberta, OA; Saskatchewan, SK) and electrolyte leakage (EL) measured at different temperatures in artificial freezing experiments for eight elite genotypes selected at Vineland Research and Innovation Centre, ON, Canada. Correlations between LT50 estimated using a logit model approach and field data were also given. Correlations were computed using Pearson method. Only significant correlations are given, with * indicating significant correlation at p=0.05

108

3.3.2 Linkage maps

The parental maps for ‘CA60’ and ‘SITR’ were slightly modified from the previous published maps (Rouet et al. 2019) in order to include all individuals for which electrolyte leakage and winter hardiness data were available. The newly created genetic map for ‘CA60’ included 140 individuals and spanned 31 linkage groups with 814 simplex SNP markers across

1,863cM. The length of the linkage groups varied from 6cM to 150cM with a mean interval distance between markers of 2.3cM. The newly created genetic map for ‘SITR’ included 130 individuals and spanned 29 linkage groups with 660 simplex SNP markers across 1,773 cM. The length of linkage groups varied from 8cM to 161cM with a mean interval distance between markers of 2.7cM. Some homologs were missing, and others were fragmented and were represented by more than one fragment. In addition, rearrangements in marker order were observed in comparison to R. chinensis. The parental maps for ‘EDI’ and ‘GV’ were created from newly generated GBS data. A total of 5,211 simplex SNP markers that were heterozygous in ‘EDI’ and homozygous in ‘GV’ were identified, and 2,940 simplex SNPs that were homozygous in ‘EDI’ and heterozygous in ‘GV’ were identified. The ‘EDI’ female map included 97 individuals and spanned 29 linkage groups with 685 simplex SNP markers across

2,093cM. The length of linkage groups varied from 6cM to 139cM with a mean interval distance between markers of 3cM. The ‘GV’ male map included 97 individuals and spanned 24 linkage groups across 1,799cM with 489 simplex SNP markers and a mean interval distance between markers of 3.7cM. The length of linkage groups varied from 6cM to 149.5cM.

109

3.3.3 Experiment 2_ QTL mapping of electrolyte leakage

3.3.3.1 Calibration with parental lines

Electrolyte leakage experiments on the four parental genotypes, conducted to establish the range of freezing temperatures that would be further implemented to screen their segregating progeny, were highly repeatable (Table 3-2). However, the electrolyte leakage did not increase from -30˚C to -50˚C in the third replication (data not shown). A fourth replication was conducted with new plants for ‘CA60’ and ‘GV’, and the new data were in the same range as the first and second replications (Supplementary Figure 3-1 – Appendix 9). Therefore, the third replication was removed from the analysis most likely because of a technical issue with the freezer. Only the first two replications were retained for the statistical analysis. The mixed-model analysis indicated that the electrolyte leakage and the index of injury were genotype-dependent (Table 3-

3). In addition, there were significant differences between the amount of electrolyte leakage and the LT50 of the parental genotypes (Figure 3-4, Table 3-4).

3.3.3.2 ‘CA60’x’SITR’ population

Electrolyte leakage experiments conducted on 100 F1 progeny and their two parental genotypes over three temperatures (-10˚C, -15˚C and -20˚C) were highly repeatable (Table 3-2), and the BLUPs of electrolyte leakage were obtained from the mixed model analysis (Table 3-5).

Although the LT50 could not be calculated in this experiment without the index of injury, the amount of electrolyte leakage for ‘CA60’ and ‘SITR’ at -20˚C (59% and 66% respectively) were comparable to the values found in the experiment on the parental lines (58% and 71% respectively) (Figure 3-4). Within the population, as the experimental temperature decreased, the amount of electrolyte leakage increased from 28% (-10˚C) to 54% (-20˚C) (Figure 3-5).

110

Table 3-2. Correlation between the replications of electrolyte leakage experiments conducted separately on the parental lines ‘CA60’, ‘EDI’, ‘GV’ and ‘SITR’, the population ‘CA60’x’SITR’ and the population ‘EDI’x’GV’. Two variables, the electrolyte leakage and the index of injury, were recorded for the parental lines and the population ‘EDI’ x ‘GV’. Pearson coefficients of correlation are given in the table and significant correlation at α=0.01 are given by *.

1. Parental lines Electrolyte leakage Rep 1 Rep 2 Rep 3 Rep 1 0.95* 0.94* Rep 2 0.95* Rep 3

Index of Injury Rep 1 Rep 2 Rep 3 Rep 1 0.93* 0.88* Rep 2 0.88* Rep 3 2. ‘CA60’ x ‘SITR’ Electrolyte leakage Rep 1 Rep 2 Rep 3 Rep 4 Rep 5 Rep 1 0.77* 0.77* 0.79* 0.73* Rep 2 0.76* 0.78* 0.72* Rep 3 0.76* 0.75* Rep 4 0.80* Rep 5 3. ‘EDI’ x ‘GV’ Electrolyte leakage Rep 1 Rep 2 Rep 3 Rep 1 0.72* 0.72* Rep 2 0.75* Rep 3

Index of injury Rep 1 Rep 2 Rep 3 Rep 1 0.70* 0.71* Rep 2 0.75* Rep 3

111

Table 3-3. General linear mixed model (GLMM) of the effect of temperature and genotype on the electrolyte leakage and index of injury measured in controlled conditions after freezing treatments with temperature-tests ranging from -10 to -50˚C for the parental lines ‘CA60’, ‘EDI’, ‘GV’ and ‘SITR’. The data were analyzed with a beta distribution and a logit link function, and the model was used to compute the best linear unbiased estimators (BLUE) of electrolyte leakage.

1. Electrolyte Leakage Covariance parameters Subject Estimate Temperature*Rep 0 Genotype * Temperature Stem_Section 0.05406 *Rep Fixed effects Numerator df Denominator df F value Pr>F Temperature 3 9 210.15 <.0001 Genotype 1 171 1903.61 <.0001 Temperature x Genotype 1 171 2.37 0.1259

2. Index of Injury Covariance parameters Subject Estimate Temperature*Rep 0.004033 Genotype* Temperature * Stem_Section 0.08258 Rep Fixed effects Numerator df Denominator df F value Pr>F Temperature 8 9 82.16 <.0001 Genotype 3 167 84.98 <.0001 Temperature x Genotype 24 167 2.76 <.0001

112

Table 3-4. LT50 of the parental lines 'CA60', 'EDI', 'GV', 'SITR' estimated from non-linear dosage response curves from the index of injury

Label LT50 Standard DF t Value Pr>|t| Lower Higher

Error Interval Interval

LT50 ‘CA60’ -17 1.25 70 -20 -15

LT50 ‘SITR’ -14 0.69 70 -16 -13

Difference in |LT50 ‘CA60’| and |LT50 ‘SITR’| 3 1.08 70 2.67 0.0094

LT50 ‘EDI’ -17 0.55 70 -18 -16

LT50 ‘GV’ -28 1.97 70 -31 -24

Difference in |LT50 ‘EDI’| and |LT50 ‘GV’| 11 175 70 6.21 <.0001

Difference in |LT50 ‘SITR’| and |LT50 ‘EDI’| 2 0.69 70 3.51 0.0008

Difference in |LT50 ‘CA60’| and |LT50 ‘GV’| 10 1.52 70 6.98 <.0001

113

114

1. ‘CA60’x’SITR’ Covariance parameters1 Subject Estimate Intercept Genotype 0.02857 Temperature Genotype 0.008490 Intercept Rep 0.002808 Temperature Rep 0.004504 Temperature x Genotype Rep 0.08122 Fixed effects Numerator df Denominator df F value Pr>F Temperature 2 15 251.51 <.0001

2. ‘EDI’x’GV’ Covariance parameters Subject Estimate Intercept Rep 0.03838 Temperature Rep 0.006010 Intercept Genotype 0.02795 Temperature Genotype 1.11*10-12 Fixed effects Numerator df Denominator df F value Pr>F Temperature 5 10 136.40 <.0001 1The data was analyzed with a beta distribution and a logit link function, and the model was used to compute the best linear unbiased prediction (BLUP) of electrolyte leakage.

115

Figure 3-5. Distribution of Best Linear Unbiased Predictors (BLUPs) of electrolyte leakage (EL) A) in the ‘CA60’ x ‘SITR’ population at -10˚C, - 15˚C and -20˚C and B) in the ‘EDI’ x ‘GV’ population at -15˚C, -20˚C, -25˚C, -30˚C, -35˚C and -40˚C

116

The amount of electrolyte leakage ranged from 41% to 74% at -20˚C, suggesting the existence of large unidirectional transgressive segregation (Table 3-6). Three QTLs associated with electrolyte leakage were identified (Table 3-7; Figure 3-6; Figure 3-7). A QTL for electrolyte leakage at -10˚C explaining 22% of the phenotypic variance mapped to LG3.H1 at 15cM in the

‘CA60’ female map. A QTL for electrolyte leakage at -20˚C explaining 15% of the phenotypic variance mapped to LG2.H1 at 99cM in the ‘CA60’ female map. A QTL for electrolyte leakage at -20˚C explaining 13% of the phenotypic variance also mapped to LG2.H5 at 15cM in the

‘SITR’ male map.

3.3.3.3 ‘EDI’x’GV’ population

Electrolyte leakage experiments were conducted on 88 F1 progeny over a wide range of sub-zero temperatures from -15˚C down to -45˚C with a measurement taken at 4˚C for a control.

The experiments were highly repeatable (Table 3-2), the BLUPs of electrolyte leakage were obtained from the mixed model analysis (Table 3-5), and the LT50s were obtained for each genotype (Supplementary Table 3-3 – Appendix 10). Although the parental genotypes ‘EDI’ and

‘GV’ were not included in this experiment, inferences on their LT50s could be made based on the experiments conducted on the parental genotypes (Table 3-4). Within the population, the amount of electrolyte leakage increased from 47% (-15˚C) to 78% (-45˚C) as the experimental temperature decreased (Figure 3-5) without evidence for transgressive segregation (Table 3-6).

Four QTLs for electrolyte leakage and LT50 were identified (Table 3-7; Figure 3-8; Figure 3-9).

A QTL for electrolyte leakage at -20˚C explaining 18% of the phenotypic variance mapped to

LG6.H1 at position 45cM in the ‘EDI’ female map. QTLs for electrolyte leakage at -20˚C and for LT50 mapped to the same position on LG7.H1 at 135cM in the ‘GV’ male map. The two

QTLs explained 14 and 16% of the phenotypic variance respectively. A QTL for electrolyte

117

leakage at -35˚C mapped to LG7.H3 at 20cM in the ‘GV’ male map, and it explained 22% of the phenotypic variance.

118

Table 3-6. Evidence for transgressive segregation for electrolyte leakage, field winter damage and field spring regrowth in the mapping populations ‘CA60’x’SITR’ and ‘EDI’x’GV’

‘CA60’ x ‘SITR’ ‘EDI’ x ‘GV’ EL -20˚C EL -25˚C Parental genotypes ‘CA60’ 58% 65% ‘SITR’ 71% 72% ‘EDI’ 71% 83% ‘GV’ 43% 55% Population Mean 54% 71% Max. 74% 78% Min. 41% 62% Winter Elora 2019 Elora 2020 Sask 2019 Sask 2020 Elora 2020 damage Parental genotypes ‘CA60’ 33% 64% 83% 14% ‘SITR’ 55% 94% 83% 91% ‘EDI’ 71% 94% 95% 85% ‘GV’ 19% 55% 82% 34% Population Mean 51% 74% 81% 58% Max. 71% 97% 93% 92% Min. 24% 21% 64% 12% Regrowth Elora 2019 Elora 2020 Sask 2019 Sask 2020 Elora 2020 Parental genotypes ‘CA60’ 87% 103% 126% 76% 96% ‘SITR’ 90% 67% 81% 69% 65% ‘EDI’ 81% 71% 70% ‘GV’ 47% 61% 110% 80% 96% Population Mean 94% 79% 103% 77% 89% Max. 140% 102% 149% 92% 148% Min. 60% 41% 73% 56% 53%

119

Table 3-7. QTLs associated with electrolyte leakage (EL), field winter damage (WD), and field spring regrowth (RG) in two mapping populations

‘CA60’ x ‘SITR’ and ‘EDI’ x ‘GV’. The QTL analysis was conducted using MQM with the software R studio using field data collected from four environments (Elora 2019, Elora 2020, Sask 2019 and Sask 2020). Mapping was conducted separately for each parental map ‘CA60’, ‘SITR’,

‘EDI’ and ‘GV’. Phenotypic means are given for each genotypic group a single putative QTL position. All QTLs for which the LOD score was above the LOD threshold were reported.

Trait Map LG1 LOD Marker peak Flanking markers % PVE2 Genotype3

AB AA

EL4 -10˚C CA60 LG3.H1 5.59 29705230 16726330…38175338 22% 29.55 26.85

EL -20˚C CA60 LG2.H1 2.99 86652613 81996645…86652613 15% 52.29 56.75

SITR LG2.H5 2.37 60933255 67472600…81894926 13% 52.42 56.36

EDI LG6.H1 3.68 38291929 32920513…38292014 18% 60.25 62.84

GV LG7.H1 2.77 57129108 50501062…67941547 14% 60.21 62.61

EL -35˚C GV LG7H3 4.8 176793 176793…1997550 22% 78.57 80.41

LT50 GV LG7H1 3.08 57129108 50501062…62247052 16% -21.08 -18.7

WD5 Elora 2019 CA60 LG2.H1 3.20 65756692 69386276…81071479 16% 49.24 56.63

SITR LG2.H5 4.93 81894926 71623330…81894926 21% 48.59 56.45

LG5.H1 3.11 65876381 35540595…75948827 11% 54.74 48.86

WD Elora 2020 CA60 LG1.H3 5.61 67169791 66170115…67492819 24% 66.40 83.11

EDI LG6H2 3.30 68093598 67121912…69256534 15% 50.27 64.72

120

GV LG5H2 2.55 28914028 25970959…53663853 14% 51.13 64.84

WD Sask 2019 SITR LG6.H1 3.26 54366335 40154937…62419734 17% 82.72 77.64

RG6 Elora 2019 CA60 LG7.H1 2.88 4530588 23986267…37107077 14% 97.11 88.43

RG Elora 2020 CA60 LG1.H3 10.18 66233855 66170116…66233855 37% 85.54 72.02

SITR LG6.H4 3.06 66596604 55947087…62762841 15% 73.98 82.51

EDI LG5.H2 3.05 10621719 4837677…17237029 14% 97.48 82.05

GV LG2.H2 3.03 68182818 59928853…79522994 14% 83.18 99.55

RG Sask 2019 CA60 LG6.H2 4.19 47563687 27091872…63202691 16% 108.43 97.42

SITR LG6.H1 4.03 54366335 53244936…55947116 18% 97.10 108.74

Defoliation CA60 LG1.H3 10.17 67169791 66170116…66233855 36% 2.551 3.681

1LG Linkage group; 2%PVE Percentage of variance explained; 3AA homozygous allele and AB heterozygous allele; 4EL Electrolyte leakage; 5WD Winter damage; 6RG Regrowth

121

Figure 3-6. QTL mapping of Best Linear Unbiased Predictors (BLUPs) of electrolyte leakage (EL) at -10˚C and -20˚C, field winter damage (WD) and field regrowth (RG) in four environments (Elora 2019, Elora 2020, Sask 2019 and Sask 2020) in the ‘CA60’ female map. LG1.H3 also display a QTL for defoliation in Elora 2019. The markers were named based on their relative position in the reference genome Rosa chinensis (Raymond et al. 2018), and the nomenclature used for the linkage groups followed Spiller et al. (2011).

122

Figure 3-7. QTL mapping of Best Linear Unbiased Predictors (BLUPs) of electrolyte leakage (EL) at -10˚C and -20˚C, field winter damage (WD) and field regrowth (RG) in four environments (Elora 2019, Elora 2020, Sask 2019 and Sask 2020) in the male map ‘SITR’. The markers were named based on their relative position in the reference genome Rosa chinensis (Raymond et al. 2018), and the nomenclature used for the linkage groups followed Spiller et al. (2011).

123

Figure 3-8. QTL mapping of Best Linear Unbiased Predictors (BLUPs) of electrolyte leakage (EL) at -10˚C and -20˚C, field winter damage (WD) and field regrowth (RG) in one environment (Elora 2020) in the female map ‘EDI’. The markers were named based on their relative position in the reference genome Rosa chinensis (Raymond et al. 2018), and the nomenclature used for the linkage groups followed Spiller et al. (2011).

124

Figure 3-9. QTL mapping of Best Linear Unbiased Predictors (BLUPs) of electrolyte leakage (EL) at -10˚C, -20˚C and -35˚C and LT50, field winter damage (WD) and field regrowth (RG) in one environment (Elora 2020) in the male map ‘GV’. The markers were named based on their relative position in the reference genome Rosa chinensis (Raymond et al. 2018), and the nomenclature used for the linkage groups followed Spiller et al. (2011).

125

3.3.4 Experiment 3_ QTL mapping of field-based winter hardiness

3.3.4.1 ‘CA60’ x ‘SITR’ population

Winter damage, which was the first aspect of field winter hardiness, varied across environments (Figure 3-10 A) to D)). As expected, cold susceptible genotypes suffered higher damage than the hardy control and parental genotypes (Figure 3-10 E) to H)). The data from

Sask 2020 were highly skewed towards extreme winter damage and not informative as 90% of the population exhibited over 95% winter damage (Figure 3-10 D)); therefore, Sask 2020 was dropped from the rest of the winter damage analysis. Genotype-by-environment interaction was significant for the three remaining environments with a high Vge/Vg ratio (1.33); therefore, the

BLUPs of winter damage were obtained separately for each of these environment (Table 3-8).

GGE-biplots were generated using the BLUPs of winter damage in order to further visualize the relationships between the environments and characterize the nature of genotype-by-environment

(GE) interaction. The two principal components PC1 and PC2 of the GGE-biplot explained

71.51% and 20.8% of the GGE variation respectively, meaning that the two sources of variation genotype (G) plus genotype-by-environment (GE) explained 92.31% of the total phenotypic variance (Figure 3-11). The angle between the environment vectors associated with Elora 2019 and Sask 2019 was acute, meaning that the environments were positively correlated without crossovers GE patterns; however, the vector associated with Elora 2019 was longer than that of

Sask 2019. Therefore, Elora 2019 was more discriminative than Sask 2019, and the nature of the genotype-by-environment (GE) interaction was mainly a change in magnitude due to the severity of the climatic conditions in Sask.

126

Figure 3-10 A) to D). Distribution of field winter damage (WD) in the ‘CA60’ x ‘SITR’ mapping population in four environments: A) Elora 2019, B) Elora 2020, C) Sask 2019 and D) Sask 2020 and distribution of field winter damage (WD) among the parental and control genotypes used in E) Elora 2019, F) Elora 2020, G) Sask 2019 and H) Sask 2020

127

Figure 3-10 E) to H). Distribution of field winter damage (WD) in the ‘CA60’ x ‘SITR’ mapping population in four environments: A) Elora 2019, B) Elora 2020, C) Sask 2019 and D) Sask 2020 and distribution of field winter damage (WD) among the parental and control genotypes used in E) Elora 2019, F) Elora 2020, G) Sask 2019 and H) Sask 2020. The parental and control genotypes are divided into cold hardy (‘Frontenac’ (Fr), ‘George Vancouver’ (GV), ‘George Vancouver’ grown on its own roots (GV1), ‘Nicolas’ (Ni) and ‘CA60’ (60)) and non cold hardy genotypes (‘Gentle Giant’ (GG), ‘Caroline de Monaco’ (CM), ‘Easy Does It’ (EDI), and ‘SITR’) relatively to their USDA cold hardiness zones.

128

Table 3-8. Generalized Linear Mixed Model (GLMM) of the effect of genotype and genotype-by-environment on winter damage for the mapping population 'CA60' x 'SITR' across three environments (Elora 2019, Elora 2020 and Sask 2019) and across individual environments, and for the mapping population ‘EDI’ x ‘GV’ across one environment (Elora 2020).

Covariance parameters1 Estimate Tests of Covariance Parameters Ratio Vge/Vg ChiSq Pr > ChiSq 1. ‘CA60’x’SITR’ Elora2019, Elora 2020, Sask 2019 Environment 0.4264 <.0001 1.33 Bloc(Environment) 0.08588 <.0001 Genotype 0.2390 <.0001 Genotype*Environment 0.3188 <.0001

Elora 2019 Bloc 66.51 Genotype 158.09 175.22 <.0001 Residual 160.73

Elora 2020 Bloc 12.6980 Genotype 1.0654 316.37 <.0001

Sask 2019

Bloc 0.01650 Genotype 0.3156 84.73 <.0001 2. ‘EDI’x’GV’ Bloc 0 Genotype 1.2053 97.54 <.0001

1 The data was analyzed separately for each population. At the exception of the data for the individual environment Elora 2019, the data were fitted to a beta distribution and a logit link function. Elora 2019 data were fitted to a Gaussian distribution. The models from individual locations were used to compute the best linear unbiased prediction (BLUP) of winter damage. The Vge/Vg ratio was given to estimate the amount of variation due to the genotype-by-environment interaction relatively to the genotypic variance, with Vge the variance associated with the genotype-by- environment interaction and Vg the genotypic variance.

129

Figure 3-11. Environment-metric preserving GGE-Biplot representing genotype plus genotype-by-environment interaction and generated from Best Linear Unbiased Predictors (BLUPs) of A) field winter damage and B) regrowth in the ‘CA60’ x ‘SITR’ population. The length of the environment vectors provides information on the ability of an environment to discriminate among genotype (i.e. discriminativeness), while the distance between environment markers and the angles between the environment vectors are associated with the correlation between environments. * Environment ** Genotype

130

Furthermore, Elora 2019 and Elora 2020 were distant on the GGE-biplot, meaning that the two environments discriminated the genotypes differently on the basis of winter damage with minor crossovers GE patterns (Figure 3-11). These observations were supported by the Pearson’s correlations (Table 3-9). The heritability of winter damage varied by location and year; it was high in Elora 2019, but it was low in Sask 2019 and Elora 2020, ranging from 0.13 to 0.83

(Table 3-10). Noticeably, pest-induced defoliation that occured in Elora 2019 was positively and highly correlated with winter damage in Elora 2020 (Table 3-9). Overall, Elora 2019 was the least severe environment, with a population mean of 51% winter damage, while Sask 2019 was the most severe environment with a population mean of 81% winter damage (Table 3-6). There was evidence for transgressive segregation beyond both parental values (Table 3-6), with 4% of individuals that were as hardy as the hardiest parent ‘CA60’ and 30% that were as sensitive or more sensitive than the tender parent ‘SITR’ in Elora 2019.

The second aspect of field winter hardiness in this study was spring regrowth. Regrowth in the population and the control and parental genotypes did not vary across environments as winter damage did (Figure 3-12 A) and B)). While ‘SITR’ and ‘CA60’ did not differ in Elora

2019, ‘SITR’ had inferior potential for regrowth in the other environments compared to ‘CA60’

(Figure 3-12 B)). The mixed model analysis of the four environments indicated significant genotype-by-environment interaction and high Vge/Vg ratio (3.09) (Table 3-11). The BLUPs for regrowth were obtained separately for each environment (Table 3-11) and visualized with a

GGE-biplot (Figure 3-11). The two principal components PC1 and PC2 explained 44.41% and

28.9% of the GGE variation respectively, meaning that genotype (G) plus genotype-by- environment (GE) explained 73.31% of the total phenotypic variance (Figure 3-11).

131

Table 3-9. Correlation between best linear unbiased prediction estimates (BLUPs) of winter damage (WD) in three environments (Elora2019, Elora 2020 and Sask 2019), defoliation and BLUPs of regrowth (RG) for four environments (Elora2019, Elora 2020, Sask 2019 and Sask 2020) for the mapping population ‘CA60’ x ‘SITR’, and between BLUPs of WG and RG for ‘EDI’ x ‘GV’ population. Significant correlations are given by *, ** and ***, indicating significant correlation at p=0.05, p=0.01 and p=0.001 respectively. Defoliation 2019 data were recorded in Elora 2019 exclusively.

WD Elora WD Elora 2020 WD Sask 2019 Defoliation Elora RG Elora 2020 RG Sask 2019 RG Sask 2020

2019 2019

WD Elora 2019 0.41*** 0.28*** 0.15

WD Elora 2020 0.15 0.65***

WD Sask 2019 0.06

RG Elora 2019 -0.06 0.09 0.04 0.13 0.16 0.15 0.07

RG Elora 2020 -0.03 -0.44*** 0.06 -0.56*** 0.04 0.02

RG Sask 2019 -0.19 0.04 -0.25* 0.02 0.29**

RG Sask 2020 -0.15 0 -0.18 -0.11

‘EDI’x’GV’

RG Elora 2020 -0.18

132

Table 3-10. Heritability and variance estimates for field winter damage and regrowth for the ‘CA60’ x ‘SITR’ and ‘EDI’ x ‘GV’ populations. Heritability of winter damage and regrowth in ‘CA60’ x ‘SITR’ population was estimated for single environments.

Population Trait Environment Method1 Broad sense Estimates heritability ‘CA60’ x ‘SITR’ Winter damage Elora 2019 LMM Gaussian 0.83 Vg = 158.09 Vresidual = 160.73 r = 5 Elora 2020 GLMM, dist = Beta link = logit 0.29 Vg = 1.0654 r=5 Φ = 9.9512 Sask 2019 GLMM, dist = Beta link = logit 0.13 Vg = 0.3156 r = 5 Φ = 7.5348 ‘EDI’ x ‘GV’ Winter damage Elora 2020 LMM Approx. Gaussian 0.81 Vg = 0.04228 Vresidual = 0.03022 r = 3 ‘CA60’ x ‘SITR’ Regrowth Elora 2019 GLMM dist = lognormal link = 0.82 Vg = 0.04041 identity Vresidual = 0.04457 r = 5 Elora 2020 GLMM dist = lognormal link = 0.77 Vg = 0.03012 identity Vresidual = 0.04377 r = 5 Sask 2019 GLMM dist = lognormal link = 0.59 Vg = 0.03274 identity Vresidual = 0.1129 r = 5 Sask 2020 GLMM dist = lognormal link = 0.41 Vg = 0.02542 identity Vresidual = 0.1435 r = 4 ‘EDI’ x ‘GV’ Regrowth Elora 2020 GLMM dist = lognormal link = 0.53 Vg = 0.05231 identity Vresidual = 0.14 r = 3 1 Data were fitted either to a linear mixed model (LMM) with a Gaussian distribution or to a general linear mixed model (GLMM) with a beta distribution and logit link function or lognormal distribution and identity link function. Vg corresponded to the genetic variance, Vresidual corresponded to the residual variance and Φ was the scale parameter, they were retrieved from the mixed linear model; r corresponded to the number of blocks in the field trial.

133

Figure 3-12. Distribution of field regrowth (RG) in four environments (Elora 2019, Elora 2020, Sask 2019 and Sask 2020) in A) the ‘CA60’ x ‘SITR’ population and B) among the parental and control genotypes ‘CA60’, ‘Caroline de Monaco’ (Cdm), ‘Easy Does It’ (EDI), ‘Frontenac’ (Fr), ‘Gentle Giant’ (GG), ‘George Vancouver’ (GV), ‘George Vanvouver’ grown on its own roots (GV1), ‘Nicolas’, and ‘SITR’.

134

Table 3-11. Generalized Linear Mixed Model (GLMM) of the effect of genotype and genotype-by- environment on regrowth for the mapping population 'CA60' x 'SITR' across four environments (Elora 2019, Elora 2020, Sask 2019 and Sask 2020) and across individual environments, and for the mapping population ‘EDI’ x ‘GV’ across one environment (Elora 2020).

Covariance parameters1 Estimate Tests of Covariance Parameters Vge/Vg Ratio ChiSq Pr > ChiSq 1. ‘CA60’x’SITR’ Elora2019, Elora 2020, Sask 2019 and Sask 2020 Environment 113 0.0013 3.09 Bloc(Environment) 52.6643 <.0001 Genotype 55.3091 0.0015 Genotype*Environment 170.67 <.0001 Residual 839.70

Elora 2019 Bloc 0.001476 Genotype 0.04041 <.0001 Residual 0.04457

Elora 2020 Bloc 0.000745 Genotype 0.03012 <.0001 Residual 0.04377

Sask 2019 Bloc 0.006495 Genotype 0.03274 <.0001 Residual 0.1129

Sask 2020 Bloc 0.002719 Genotype 0.02542 0.003 Residual 0.1435 2. ‘EDI’ x ‘GV’ Fixed effect FValue Pr > F Cov_spline 6.98 <.0001 Covariance parameters Estimate Tests of Covariance Parameters ChiSq Pr > ChiSq Genotype 0.05231 12.90 <.0001 Residual 0.14

1 The data was analyzed separately for each population. The data were fitted to a lognormal distribution and to an identity link function. The data for the ‘EDI’ x ‘GV’ population were corrected for spatial variability using radial smoothing. The models from individual locations were used to compute the best linear unbiased prediction (BLUP) of regrowth. The Vge/Vg ratio was given to estimate the amount of variation due to the genotype-by-environment interaction relatively to the genotypic variance, with Vge the variance associated with the genotype-by-environment interaction and Vg the genotypic variance.

135

Figure 3-13. Distribution of of A) field winter damage (WD%) and B) regrowth (RG%) in the ‘EDI’ x ‘GV’ population and in the parental and control genotypes ‘CA60’, ‘EDI’, ‘GV’ and ‘SITR’ in Elora 2020.

136

The right angle between the vectors associated with Sask and the vectors associated with

Elora indicates that Sask and Elora environments were not correlated for regrowth. Moreover, the angles between the vectors associated with Sask 2019 and Sask 2020 and between Elora

2019 and Elora 2020 were acute, indicating that regrowth was positively correlated at each location across years. This was corroborated by the Pearson’s correlations (Table 3-9). Elora

2019 and Sask 2019 had the longest vectors, meaning that they were more capable of discriminating among genotypes than Elora 2020 and Sask 2020 (Figure 3-11). The heritability of regrowth was moderate to high with the highest estimate obtained for Elora 2019 (Table 3-

10). Overall, regrowth was higher in 2019 than in 2020 at both locations, and there was evidence for transgressive segregation beyond both parents (Table 3-6). Winter damage and regrowth were not correlated in most environments, which suggested that regrowth and winter damage be inherited separately. However, regrowth and winter damage were highly correlated in Elora

2020 (Table 3-9). Noticeably, pest-induced defoliation recorded in Elora 2019 was highly correlated with regrowth in Elora 2020 (Table 3-9).

QTL analysis for winter damage and regrowth was conducted separately for each environment because of the existence of significant genotype-by-environment interaction (Table

3-7; Figure 3-6; Figure 3-7). A QTL for winter damage in Elora 2019 mapped to LG2.H1 at

55cM in the ‘CA60’ female map. This QTL alone explained 16% of the total phenotypic variance. A major QTL for winter damage in Elora 2019 mapped to LG2.H5 at 17cM in the

‘SITR’ male map. A minor QTL for winter damage in Elora 2019 mapped to a fragment of

LG5.H1 at 10cM in the ‘SITR’ male map. The two-QTL model explained 30% of the phenotypic variance. A QTL for winter damage in Sask 2019 mapped to LG6.H1 at 20cM in the ‘SITR’ male map, and it explained 17% of the phenotypic variance. A QTL for winter damage in Elora

137

2020 mapped to LG1.H3 at 10cM in the ‘CA60’ female map, explaining 24% of the phenotypic variation. No QTL for winter damage in Sask 2020 were detected. In addition, a QTL for regrowth in Elora 2019 mapped to LG7.H1 at 40cM in the ‘CA60’ female map; it explained

14% of the phenotypic variance. A QTL for regrowth in Elora 2020 mapped to LG1.H3 at 20cM in ‘CA60’ — 10cM away from the QTL for winter damage in the same environment — and explained 37% of the phenotypic variance. A QTL for regrowth in Elora 2020 mapped to

LG6.H4 at 60cM in the ‘SITR’ male map, and it explained 15% of the phenotypic variance. A

QTL for regrowth in Sask 2019 mapped to LG6.H2 at 20cM in the ‘CA60’ female map, and it explained 18% of the phenotypic variance. A QTL for regrowth in Sask 2019 mapped to LG6.H1 at 20cM in the ‘SITR’ male map, and it explained 18% of the phenotypic variance. No QTL for regrowth in Sask 2020 were detected. A QTL for defoliation that occurred in Elora in 2019 mapped to LG1.H3 at 15cM in the ‘CA60’ female map and explained 36% of the phenotypic variance.

3.3.4.2 ‘EDI’ x ‘GV’ population

Winter damage and regrowth were recorded in one environment (Elora 2020). Winter damage was distributed around a mean of 58% with minimum damage of 12% and maximum damage of 92%, and regrowth was distributed around a mean of 89% with minimum regrowth of

53% and maximum regrowth of 148% (Figure 3-13; Table 3-6). The control genotypes ‘CA60’ and ‘SITR’ and the parental lines ‘EDI’ and ‘GV’ differed, but their respective ratings were similar to the ‘CA60’ x ‘SITR’ multi-site field trial (Table 3-6). The female parent ‘EDI’ exhibited substantial winter damage (85%), close to total dieback to the ground, while the male parent ‘GV’ presented minimal winter damage (34%) (Figure 3-13). ‘GV’’s regrowth appeared greater than ‘EDI’ (96% and 70% respectively) (Table 3-6). There was evidence for

138

transgressive segregation beyond both parents for both winter damage and regrowth (Table 3-6), with 7% of individuals that were as hardy as the hardiest parent ‘GV’ and 11% that were as sensitive or more sensitive than the tender parent ‘EDI’. The genotypic variance was significant for both winter damage and regrowth (Table 3-8; Table 3-11). Both winter damage and regrowth were heritable (Table 3-10), but regrowth and winter damage were not correlated for the ‘EDI’ x

‘GV’ population (Table 3-9). QTLs for winter damage and regrowth mapped to different LGs of the rose genome (Table 3-7; Figure 3-8; Figure 3-9). A QTL for winter damage in Elora 2020 mapped to LG6.H2 at 10cM in the ‘EDI’ female map; this QTL explained 15% of the variability observed. A QTL for winter damage in Elora 2020 also mapped to LG5.H2 at 65cM in ‘GV’.

This QTL explained 14% of the phenotypic variance. In addition, a QTL for regrowth mapped to

LG2.H2 at 30cM in the ‘GV’ male map and explained 14% of the observed phenotypic variation.

A QTL for regrowth mapped to LG5.H2 at 50cM in the ‘EDI’ female map and explained 14% of the phenotypic variance.

3.3.5 Relationship between electrolyte leakage and field winter hardiness

Electrolyte leakage measured at -20˚C under artificial conditions was positively but moderately correlated with winter damage in all environments for the ‘CA60’ x ‘SITR’ population, with the highest correlation being in the environment Elora 2019 and the weakest being in Elora 2020 (Table 3-12). Electrolyte leakage measured at 25˚C under artificial conditions was poorly correlated with winter damage for the ‘EDI’ x ‘GV’ population, and the

LT50 did not correlate with winter damage (Table 3-12). Different levels of freezing tolerance under artificial conditions were identified for each individual in both mapping populations based on the distribution of electrolyte leakage data; individuals were grouped into three categories of electrolyte leakage according to their affiliation to either the 1st quartile of the distribution, the

139

2nd and 3rd quartiles or the 4th quartile, and their level of freezing tolerance was compared to their level of field winter hardiness. For both populations, while the individuals from the 1st quartile had the least electrolyte leakage, they did not consistently show the least winter damage in the field. Likewise, while the individuals from the 4th quartile had the most electrolyte leakage, they did not consistently show the most winter damage in the field. Consequently, electrolyte leakage did not align with field winter hardiness (Figure 3-14).

140

Table 3-12. Correlation between best linear unbiased prediction estimates (BLUPs) of winter damage (WD), electrolyte leakage (EL) and LT50 for the mapping populations ‘CA60’ x ‘SITR’ and ‘EDI’ x ‘GV’. Pearson correlation coefficients are indicated, and significant correlation at p=0.05, p=0.01 and p=0.001 are indicated by *, ** and *** respectively.

EL - EL - EL - EL - EL - EL - EL - LT50 10˚C 15˚C 20˚C 25˚C 30˚C 35˚C 40˚C 1. ‘CA60’ x ‘SITR’ WD Elora 0.08 0.26** 0.35*** 2019 WD Elora -0.03 0.08 0.19* 2020 WD Sask 0.26** 0.26** 0.31*** 2019 2. ‘EDI’ x ‘GV’ WD Elora 0.13 0.20 0.25* 0.14 0.15 0.14 0.09 2020

141

Figure 3-14. Relationship between Best Linear Unbiased Predictors (BLUPs) of electrolyte leakage and BLUPs of winter damage for A) the ‘CA60’ x ‘SITR’ population and B) the ‘EDI’ x ‘GV’ population.

1 Different levels of freezing tolerance under artificial conditions were identified for each individual of the ‘CA60’ x ‘SITR’ population based on the distribution of electrolyte leakage data at -20˚C. Different levels of freezing tolerance under artificial conditions were identified for each individual of the ‘EDI’ x ‘GV’ population based on the distribution of electrolyte leakage data at -25˚C. For both populations, individuals were grouped into three categories of electrolyte leakage according to their affiliation to either the 1st quartile of the distribution, the 2nd and 3rd quartiles or the 4th quartile, and their level of freezing tolerance under artificial conditions was plotted against their level of field winter damage in Elora 2019 for the ‘CA60’ x ‘SITR’ population (A) and Elora 2020 for the ‘EDI’ x ‘GV’ population (B).

142

3.4 Discussion

Electrolyte leakage appeared as a good proxy for field winter hardiness in a small panel of commercial cultivars; however, it did not align with field winter damage in the breeding material. Field winter damage was recorded in two locations — Elora and Saskatoon — over two years, but the climatic conditions in Saskatoon during the second year were too extreme to provide meaningful information. QTLs were detected for electrolyte leakage, winter damage and spring regrowth, and no stable QTLs were detected for winter damage and spring regrowth across environments.

3.4.1 Relationship between electrolyte leakage and field winter damage

Electrolyte leakage measured under artificial conditions at -15°C and at -20°C, and the

LT50 were strong indicators of field winter hardiness in a panel of 28 phenotypically distinct commercial cultivars and elite roses (Figure 3-1 to 3-3). These results were supported by the literature (Ouyang et al. 2019). Ouyang et al. (2019) investigated the usefulness of the electrolyte leakage method in 17 rose cultivars, which were naturally acclimated in the field. The authors demonstrated that the genotypes with the highest and lowest levels of cold hardiness relative to their USDA zone could be identified using electrolyte leakage. Our results suggested that the implementation of electrolyte leakage assays for screening winter hardiness in a timely manner in roses could be a strong asset for a rose breeding program.

However, while electrolyte leakage was a strong indicator of field winter hardiness in cultivars, it did not align with field winter damage for the two mapping populations (Figure 3-

14), as supported by the poor correlation between electrolyte leakage and winter damage in the mapping populations (r = 0.35) (Table 3-12). Perhaps, there were larger differences in cold hardiness among the rose cultivars that were specifically chosen for their extreme responses to

143

cold stress than the breeding populations that displayed intermediate phenotypes. These results suggested that electrolyte leakage assays present limited utility for application on breeding material to develop winter hardy roses adapted to Elora and Sask. Gusta et al. (1997) studied winter hardiness in winter wheat and suggested that the nature of the stress in artificial freeze- tests — quick exposure to low temperatures with plant organs prone to flash freeze — differ from natural conditions where winter kill occurs due to prolonged exposure to sub-zero temperatures with severe freeze-induced dehydration stress (Gusta et al. 1997) or changing global climate change winter patterns (Willick et al., 2021). In addition, electrolyte leakage experiments are independent from intertwined factors that occur in nature and greatly impact field winter damage ratings, such as overall plant health, pest-pressure and soil status, acclimation, and carbohydrate allocation. Consequently, the multifactorial complex nature of field winter survival makes field trials the ultimate approach to measure field winter hardiness in roses, and only multi-year field testing can estimate winter survival (Karam and Sullivan 1991).

As a result, despite the initial observations of electrolyte leakage being highly correlated with the USDA hardiness zones and field winter hardiness in a set of 28 commercial cultivars and elite genotypes, electrolyte leakage has limited utility to be used as a tool to select for hardy roses in a breeding program without the need for further field evaluation. While electrolyte leakage and field winter hardiness data did not align for the breeding material, electrolyte leakage experiments are independent from the complex network of abiotic and biotic factors that occur in the field and could be relevant for the identification of candidate genes associated with freezing tolerance in roses.

144

3.4.2 Distinct genetic basis of freezing tolerance under artificial stress and field winter

hardiness

Given the lack of relationship between electrolyte leakage and winter hardiness, it was not surprising that the QTLs for electrolyte leakage and field winter hardiness generally did not overlap. In addition, QTLs for electrolyte leakage mapped to different genomic regions in the different parental maps; therefore, the action of complementary genes could explain the transgressive segregation. A total of seven QTLs were identified for electrolyte leakage across mapping populations. QTLs for electrolyte leakage at -20˚C were located on LG2 in the ‘CA60’ x ‘SITR’ mapping population, and both parents contributed favourable alleles to freezing tolerance. Using the relative position of the molecular markers in the R. chinensis genome to survey the genome for potential candidate genes in this region, the QTL for electrolyte leakage at

-20˚C on LG2 mapped nearby a key regulator of winter hardiness: the ICE1-transcription factor

(Supplementary Table 3-1 – Appendix 7). ICE is an upstream transcription factor of CBF-genes with a major regulatory role in the acquisition of cold tolerance within the CBF pathway

(Chinnusamy et al. 2003) (Supplementary Table 3-4 – Appendix 11). A QTL for electrolyte leakage at -10˚C mapped to a different linkage group in the ‘CA60’ female map, suggesting that temperature-specific QTLs for electrolyte leakage may exist, and this could be directly related to the cascade of activation and degradation of various genetic factors involved in the acquisition of freezing tolerance. The QTL mapped to a genomic region that harbors a CBF1-like transcription factor (Table 3-7; Supplementary Table 3-1 – Appendix 7). CBF-transcription factors are considered master regulators of cold hardiness (Wisniewski et al. 2014). They are overexpressed shortly after the perception of the cold signal before being degraded, and they activate

145

downstream the expression of several COR-genes. QTLs for electrolyte leakage at -20˚C mapped to LG6 in the ‘EDI’ female map, while it mapped to LG7 in the ‘GV’ male map.

3.4.3 QTL for winter damage

QTLs were mapped for each environment separately. No stable QTLs were identified.

While a QTL for winter damage in Elora 2019 was detected on LG2 in ‘CA60’, the QTL for winter damage in Elora 2020 mapped to LG1. Furthermore, although the QTLs for electrolyte leakage at -20°C and for winter damage in Elora 2019 did not collocate, they overlapped on

LG2. A minor QTL for winter damage in Elora 2019 was identified in ‘SITR’ on LG5. In addition, a QTL for winter damage in Elora 2020 mapped to LG5 in the ‘GV’ male map. While cold hardy Explorer rose ‘GV’ contributed a favorable allele to winter hardiness, cold sensitive

Floribunda ‘SITR’ carried a deleterious allele (Table 3-7). These results were consistent with expectations from a recent study conducted on genetic diversity between cold hardy Canadian roses and non-cold hardy European roses and between cold hardy European roses and non-cold hardy Canadian roses that suggested that LG5 would be a potential location for a QTL associated with winter hardiness (Vukosavljev 2014). Similarly, QTLs for winter damage in Elora 2020 in

‘EDI’ mapped to LG6, which was in agreement with Vukosavljev (2014) who indicated the presence of QTLs associated with cold hardiness on LG6 of the non-cold hardy Large Flowered

Climber ‘Red New Dawn’.

146

3.4.4 QTL for regrowth

Regrowth was inherited independently from winter damage, as the QTLs for regrowth generally did not overlap with the QTLs for winter hardiness. QTLs for regrowth mapped to LG6 in the ‘CA60’ x ‘SITR’ population, while they mapped to LG2 and LG5 in the ‘EDI’ x ‘GV’ population. Exploration of the QTL intervals revealed the presence of numerous genes with GO terms associated with cellular components and biological processes. GO terms associated with cellular components included plant-type cell wall organization, cell wall modification, and plasma membrane. GO terms associated with biological processes related to plant growth included carbohydrate metabolic process, phloem and xylem histogenesis, cell wall macromolecule catabolic process, cellulose biosynthetic process, seed development, and flower development. Further research is needed to examine patterns of gene expression in roses- development stages across environments (Walls et al. 2019).

3.4.5 Genetics of winter hardiness and regrowth

Winter damage is a quantitative trait under polygenic control. It is highly heritable but subject to genotype-by-environment interaction. The heritability estimates were comparable for

‘CA60’ x ‘SITR’ population in Elora 2019 and the ‘EDI’ x ‘GV’ population in Elora 2020, and they were high (Table 3-10). These heritability estimates were also comparable to those found by

Svejda (Svejda 1979). However, the heritability of winter damage was low in Elora 2020 and

Sask 2019.

The ‘CA60’ x ‘SITR’ population was evaluated for field winter damage at two different locations (Elora and Sask) featuring different climates (USDA zones 3b and 5b). Overall, winters in Sask were far colder than winters in Elora with extended periods of sub-zero

147

temperatures, and they did not offer much insulating snow cover during the colder months, making the bush more vulnerable to freezing wind chill and desiccation (Figure 3-15). The Sask environment in 2018 had an unusually cold spring, an extremely hot and dry summer at the time of planting, followed by a particularly cold fall. The coolest temperature in the fall reached -

2.6˚C, and the depth of the snow did not exceed 5cm deep in average by the end of December.

February 2019 was one of the coldest on records with extreme minimal temperatures close to -

39˚C, and February-March had the longest period with sustained sub-zero temperatures of 38 days. In comparison, extreme minimal temperatures in Elora did not go below -26˚C, and the longest period with sustained sub-zero temperatures was of 13 days. The following summer in

Sask did not get particularly hot and the temperatures dropped near freezing at the end of the summer. Under these extreme conditions, the progeny expressed lower genetic potential in Sask than in Elora (Mathew et al. 2018). The extreme conditions contributed to cause severe damage to most plants, reducing the genetic variance and the heritability of the trait from 0.83 in Elora

2019. to 0.13 in Sask 2019. Sask 2020 was the most severe environment in which most plants suffered extensive damage. Consequently, winter temperatures are not the only driver of the genotype-by-environment interaction, but prolonged periods of sub-zero temperatures, snow cover, late fall and early spring temperatures that impact acclimation and de-acclimation, and precipitation are also involved.

148

Figure 3-15. Climatic conditions in Saskatoon (SK) and Elora from June 2018 to July 2020. The longest period of sustained sub-zero temperatures is given has a count of consecutive days with sustained sub- zero temperatures. Climatic data for SK were retrieved from the Saskatchewan Research Council Website at the Saskatoon Climate Reference Station (www.src.sk.ca/labs/climate-reference-stations), and climate data collected at the Elora Research Station were retrieved from the Environment Canada Website (https://climate.weather.gc.ca/historical_data).

149

The extent of winter damage is directly affected by various climatic factors; however, it also relies heavily upon the overall health of the plant. The discrepancy between Elora 2019 and

Elora 2020 was most likely due to a severe defoliation mainly caused by black spot

(Diplocarpon rosae) that occurred at the end of August 2019 and that impaired the assessment of true winter hardiness in 2020, as indicated by the high positive correlations between winter damage, regrowth and defoliation (Table 3-9). In addition, the QTLs associated with winter damage and regrowth in Elora 2020, and defoliation mapped to a region with a major known locus for black spot disease resistance, Rdr1 (von Malek and Debener 1998). In a complementary study on black spot disease resistance in the ‘CA60’ x ‘SITR’ population, susceptibility ratings were collected on the progeny in detached leaf assays using single spore inoculum, and a marker linked to the black spot resistance ‘CA60’ Rdr1A allele was developed (Rouet et al. 2019). The

QTLs associated with winter damage and regrowth in Elora 2020, and defoliation, mapped to the same location as the QTLs of resistance to black spot and the ‘CA60’Rdr1A allele (Fig. 3.6; Fig.

2.6). Black spot has been found to be correlated with overall plant performance, horticultural value, vigour, winter survival and winter injury in several studies (Mackay et al. 2008; Carlson-

Nilsson and Davidson 2009; Zlesak et al. 2017). Black spot can be responsible for much of the field winter injury because it severely weakens the plants as they enter dormancy, so it is not surprising to have detected a QTL for black spot resistance in the heavily infected Elora 2020 trial. Severe defoliation may promote the growth of new shoots late in the season that will not be mature enough to acclimate properly and will have not have enough carbohydrate reserves to allocate to overwintering (Carlson-Nilsson and Davidson 2009). Dhont et al. (2006) also showed that early fall defoliation reduces accumulation of carbohydrate reserves in perennial alfalfa associated with a reduction in spring regrowth. Consequently, the ability to create exceptionally

150

winter hardy roses depends on the ability to simultaneously breed for increased black spot disease resistance as well as winter hardiness and regrowth.

The hardy Explorer roses are able to survive winters with minimum damage limited to the tips of the canes; however, the hardy ‘Parkland’ roses die back to the snow line or to the crown, before growing back from the crown in the spring. Therefore, different strategies exist for winter survival that complicate how rose breeders define and evaluate field winter hardiness across genotypes and breeding material. Winter survival does not only depend on the degree of damage that the plant suffered in the winter but also on its ability to resume meristematic activity in the spring. In this research, regrowth was highly heritable in both populations in single environments (Table 3-10); however, the heritability decreased in higher stress environments associated with severe pest-pressure or severe climatic conditions. Regrowth was in the same range across Elora and Sask (Table 3-6), but the plants grew much bigger in Elora than in Sask.

The average shoot length of a rose bush in Elora was 55cm, with a maximum of one meter; however, the average shoot length of a rose bush in Sask was 20cm only, up to 50cm (data not shown). While QTLs for winter damage and regrowth mapped to different linkage groups for

Elora 2019, they co-localized on LG6 for Sask 2019 (Figure 3-7). While winter damage and regrowth were not correlated in Elora 2019 (r = -0.06) and were inherited separately, they were negatively correlated in Sask 2019 (r = -0.25) (Table 3-9). Although the correlation was small, these results suggested that the more damaged the plants were after the winter in Sask 2019, the less regrowth occurred in the spring. With extensive winter damage on the canes, the new growth from the crown may not have benefited from newly fixed carbon by emerging leaflets performing photosynthesis but rather relied solely on stored carbohydrate reserves. Notably, the extremely hot and dry summer at the time of planting in Sask, followed by a particularly cold

151

fall, might have 1) compromised the establishment of the root system and slowed the overall growth after planting; 2) contributed to a reduction in the accumulation of carbohydrate reserves in the same year; 3) compromised the successful entry into dormancy and into the first winter; and 4) compromised the regrowth during the following growing season due to limited carbohydrate reserves. These results highlight the inability of the roses in this study to express their full genetic potential in extremely severe environments that impose extreme selective pressure.

While Elora (USDA hardiness zone 5b) was an appropriate environment to evaluate the progeny of a cross between a hardy and a tender rose as long as that pest-pressure remained under control,

Sask (USDA hardiness zone 3a) would be an optimum environment to evaluate the adaptability of seedlings originating from the cross between a hardy rose and a semi-hardy rose or between two hardy roses (Svejda 1979).

3.5 Conclusion

While electrolyte leakage was used as a proxy for field winter hardiness in a set of extremely differing commercial and elite roses, it had limited utility in the breeding material and the target environments of this study. Winter damage is a quantitative trait under polygenic control, with parental genotypes contributing different alleles and mechanisms for winter hardiness. It was highly heritable, yet subjected to genotype-by-environment interactions, and no

QTLs for winter damage stable across locations were detected. The climate in Saskatoon — particularly in 2020— was too extreme for the roses to express their full genetic potential; therefore, Saskatoon had limited ability to discriminate the genotypes of this study on the basis of their true hardiness levels. In addition, both winter damage and spring regrowth were impacted by disease pressure. Breeding roses with increased winter hardiness clearly requires the breeder to simultaneously breed for disease resistance.

152

Acknowledgements

This work was supported by funding from the Vineland Research and Innovation Centre,

Agriculture and Agri-Food Canada Growing Forward 2 project #AIP-P013, Canadian

Agricultural partnership, NSERC, OMAFRA, the Canadian Nursery Landscape Association,

Landscape Manitoba and Landscape Alberta. The authors are also grateful to the support of the

International Tuition Scholarship from Ontario Agriculture College (OAC) and the technical support of Rensong Liu, Jackie Bantle, Ruobin Liu and Gowribai Valsala (University of

Saskatchewan).

153

CHAPTER 4: GENERAL DISCUSSION: TOWARD A TECHNOLOGY AND DATA

DRIVEN ROSE BREEDING PROGRAM

Vineland Research and Innovation Centre’s (VRIC) garden rose breeding program is part of Canada’s history and heritage. VRIC holds many roses that originated from the breeding efforts of Canadian rosarians and scientists who aimed to expand the geographical scope of rose cultivation across the country thanks to an understanding of the genetic inheritance of winter hardiness. While roses for commercial release are being selected based on many important aesthetic traits, such as continuous blooming, plant architecture, flower colour and petal number, exceptional cold hardiness and black spot disease resistance remain the primary targets of the breeding program. The roses are mainly created through traditional breeding methods — with manual hybridizations in the greenhouse and field selections — and the currently available molecular technologies have been so far underutilized. The intent of this thesis was to bridge the gap that exists in establishing a framework for developing molecular technologies to effectively breed roses for black spot disease resistance (Chapter 2) and winter hardiness (Chapter 3), and to modernize Canada’s national rose breeding program; therefore, this work represents a step forward towards the implementation of marker-assisted selection in rose breeding.

4.1 Breeding for black spot disease resistance in garden roses

4.1.1 Summary of the findings

Black spot disease is caused by the asexual stage of the fungus Diplocarpon rosae, and it is the most devastating foliar disease of field-grown roses. In addition of reducing the marketability of the cultivars, black spot is expensive and extremely challenging to assess in the field. In fact, a cultivar could be planted in the field for up to three years before showing its true

154

susceptibility to the pathogen, misleading the selection of best-performing cultivars in early years and making necessary the implementation of multi-year field trials. Therefore, it appears critical to develop a molecular assay to reliably screen roses for black spot resistance at early stages in the breeding program, reducing the cost associated with the phenotyping and accelerating the selection of high quality hybrids. The second chapter of this thesis focused on mapping black spot resistance in garden roses by exploiting field data and data collected under controlled conditions in a bi-parental rose population ‘CA60’ x ‘Singing in the Rain’ derived from the cultivar ‘Frontenac’ of the Explorer series, with black spot-resistant ‘CA60’ being a direct offspring of ‘Frontenac’.

Black spot disease resistance is known to be vertical with the existence of race-specific resistance genes. Preliminary pathology work helped to understand the fungal race diversity in

VRIC’s experimental farm and allowed the identification of predominant races which to breed resistance (races 5, 7, 10, and newly identified race 14). While the set of differential cultivars is necessary to identify and characterize the diversity of Diplocarpon rosae races, some inconsistencies in the infection patterns were reported and were supported by the literature

(Zlesak et al. 2020). Another challenge of this research was to address the complex genetic inheritance of segmental allopolyploid roses when generating high-density linkage maps, a precursor of QTL detection. To overcome the complexity associated with roses’ meiosis without compromising on marker density and map coverage, single nucleotide polymorphism-based genetics maps were created for each parental genotypes using the cost-effective GBS sequencing from over 300 F1 progeny and following a two-way pseudo-testcross strategy (Grattapaglia and

Sederoff 1994). Black spot resistance being a simply inherited trait, this approach was effective at detecting robust QTLs for black spot resistance despite the small sample size used in

155

phenotyping (n < 100). In fact, black spot resistance to races 5, 10 and 14 assessed under an improved detached leaf assay, and field resistance co-located on LG1 and mapped to a cluster of nine TIR-NBS-LRR paralogues known from the literature. A unique sequence of 32 bp in exon 4 of the active gene of this cluster, muRdr1A, that co-segregated with the resistant phenotypes was identified in the BS-resistant female parental genotype. Two diagnostic markers, a presence/absence marker and an INDEL marker, specific to this sequence were designed and validated in the mapping population (Aaaa x aaaa) and a backcross population derived from

‘CA60’ (Aaaa x Aaaa). The existence of a few recombinants between the markers and the phenotypes and between the races suggested that the unique muRdr1A sequence not be the causal polymorphism. Nevertheless, the unique polymorphism associated with black spot resistance found in muRdr1A was successfully used as a template to develop markers.

4.1.2 Implementation of marker-assisted selection: promises and impediments

The genotypes at the ‘CA60’Rdr1A marker were compared to the field performance of the individuals from the mapping population (Aaaa x aaaa) and the validation population (Aaaa x

Aaaa) (Chapter 2). In the former, 57 individuals had the ‘CA60’ resistance allele and were free of symptoms in the field, while 72 lacked the resistant allele and were black spot-susceptible.

Only five individuals had the ‘CA60’ resistance allele and were black spot -susceptible in the field (false positives), yielding a 92% success rate for marker-based selection. In the latter, 44 individuals had the ‘CA60’ resistance allele and were free of symptoms in the field, while 13 lacked the resistance allele and were black spot -susceptible in the field. Only one individual had the ‘CA60’ resistant allele and was BS-susceptible in the field (false positive), leading to a 98% success rate for marker-based selection.

156

Additional screening in a panel of 500 VRIC’s elite genotypes revealed that the ‘CA60’Rdr1A marker was specific to the donor ‘CA60’, suggesting that its use be restricted to segregating bi- parental populations derived from it (data not shown). Nonetheless, because ‘CA60’ has been extensively used as a breeding line at VRIC, its genetic is widely spread among VRIC’s funding parents and elite genotypes. Thus, the use of the marker was implemented in Canada’s national rose breeding program in 2020 to screen over 2,000 seedlings from 37 bi-parental segregating populations including only six populations unrelated to ‘CA60’ for which one of the two parental lines was clear of symptoms in the field and presented the ‘CA60’ allele. Marker assisted- selection (MAS) was then used to discard close to a thousand of undesirable material that lacked the ‘CA60’ allele and to retain over 1,200 hybrids that tested positive for the marker, increasing genetic gain for black spot resistance in comparison with phenotypic selection. In order to reduce the cost of the genomic laboratory work required for MAS and justify the implementation of

MAS for thousands of seedlings over field planting and field selection, a cost-effective and automation friendly DNA extraction protocol was required. The MagAttract 96 DNA Plant Core

Kit, which is based on magnetic particle technology, was chosen over the DNeasy Plant Mini

Kit, which includes a system of filtration and homogenization column (qiagen). However, the

DNA extraction failed to yield high quality DNA for 50% of the samples from the initial pool of

5,000 seedlings, making screening of these samples with the marker impossible and reducing the scope of MAS. Further efforts will be needed to optimize low-cost DNA extraction protocols for recalcitrant rose tissues that contain high levels of interfering metabolites such as phenols.

157

4.1.3 Future directions for the identification and characterization of novel sources of black spot

resistance

VRIC inherited the genotype ‘CA60’ from Canada’s public research; ‘CA60’s hardiness, increased disease resistance and good hip setting made it a very popular parental line in the hybridization schemes. More specifically, ‘CA60’ was created at the rose breeding station of

Morden, Manitoba, in an effort to improve roses winter hardiness, and originated from the cross between ‘RSM 104’ and ‘Frontenac’. Although little information was available on ‘RSM 104’ at the time of conducting this research, ‘CA60’-black spot resistance was initially believed to have been inherited from its Canadian Explorer hardy male parent ‘Frontenac’, and the first chapter originally aimed to characterize new sources of black spot resistance from Explorer roses for the development of broad-range molecular markers. However, high resolution melting (HRM) revealed that ‘Frontenac’ did not carry the muRdr1’CA60’ allele linked to resistance to races 5,

10 and 14, suggesting that ‘RSM 104’ be the source of the resistance for which we developed a marker. After the publication of the black spot research chapter in a peer-reviewed journal, further knowledge was gathered on ‘RSM 104’ (Peter Harris, personal communication). ‘RSM

104’, also known as German selection 91/104-1, is a BS-resistant tetraploid R. multiflora hybrid and resulted from the cross between a colchicine-doubled R. multiflora used as a female parent and BS-susceptible ‘Caramba’ as the pollen parent, with black spot resistance most likely originating from R. multiflora (Carlson-Nilsson and Davidson 2006). The genotype 91/100-5 was derived from the cross between the colchicine-doubled R. multiflora 88/124-46 and

‘Caramba’, and 91/100-5 was the genotype in which the first black spot resistance gene Rdr1 was described (von Malek and Debener 1998). In the latter research, the phenotypic segregation ratios from various populations such as F2 and BC derived from 91/100-5 supported the presence

158

of a single resistance gene in the duplex configuration (AAaa), with the dominant resistance allele A, in 91/100-5. ‘RSM 104’ was derived from the cross between a different colchicine- doubled R. multiflora and ‘Caramba’ (Peter Harris, personal communication). The complete sequence of the muRdr1 gene locus used in sequence alignment in the second chapter was obtained from the diploid homozygous 88/124-46. HRM conducted with the presence/absence diagnostic marker further revealed that ‘RSM 104’ (91/104-1) carried the muRdr1’CA60’ allele

(data not shown), but further research would be needed to determine the allele dosage of the gene. In addition, although the phylogenetic relatedness between ‘CA60’ and 88/124-46 is questionable, the sequencing and assembly of ‘CA60’Rdr1 would be necessary for comparison with muRdr1. While this task has been attempted in this research with 150bp reads from resequencing, it has been challenged by the tetraploid nature of ‘CA60’, high heterozygosity, short reads and high sequence similarities between the different paralogues of this locus.

Although this research did not pursue the characterization of the source of resistance against race 7 (isolate VOTB17-1) after QTL detection, the QTL for resistance to race 7 mapped to a different location on LG1 than the QTLs for resistance to races 5, 10 and 14. To our knowledge, no sources of resistance against Diplocarpon rosae race 7 have been identified to date. While 88/124-46 and its hybrids, including 91/100-5, are infected by race 7 and that muRdr1A does not confer resistance to it (Menz et al. 2018), ‘CA60’ was not infected by race 7 and the population ‘CA60’ x ‘SITR’ segregated for resistance to race 7. These observations could be regarded as a framework for the identification of a new source of resistance to race 7 potentially originating from the Explorer roses, enlarging the scope of the research.

159

4.2 Breeding for cold hardiness in garden roses

4.2.1 Summary of findings

Canadian roses are genetically distinct from European roses, are more closely related to rootstocks (Vukosavljev), and they are known for their superior winter hardiness. Cold hardiness is a quantitative trait but has been reported to be under the control of a few genes only in roses

(Svejda). The highly heritable nature of cold hardiness is an advantage for plant breeders to achieve quick genetic gain. The third chapter of this thesis aimed to study the genetics of rose cold tolerance, both in the field under multiple environments (Elora, Ontario and Saskatoon,

Saskatchewan) and under artificial conditions using electrolyte leakage as a quantitative and objective measurement of freezing tolerance using a QTL mapping approach. This third chapter also aimed to explore whether labour and cost-intensive field trials could be substituted by an electrolyte leakage assay to estimate the winter hardiness of breeding material. Two segregating bi-parental rose populations derived from Explorer roses were investigated: ‘CA60’ x ‘Singing in the Rain’ derived from Canadian hardy Explorer roses ‘Frontenac’ and ‘George Vancouver’ x

‘Easy Does It’ with ‘George Vancouver’ being an exceptionally hardy Explorer rose.

While electrolyte leakage was not able to predict field winter hardiness in the breeding material used in the study, it can be a valuable approach to identify potential candidate genes for freezing tolerance to be used in future studies on the genetic characterization of freezing tolerance in garden roses. Field trial is the ultimate way to measure field winter hardiness because winter hardiness in the field depends upon so many biotic and abiotic natural factors. In this research, winter hardiness was highly heritable. Yet, it was subjected to genotype-by- environment interaction. The climatic conditions in Saskatoon were so extreme that the plants, which were under extreme selective pressure, were not able to express their full genetic potential.

160

Since both genotypic effect and genotype-by-environment interaction need to be considered together to make useful selection decision, the data collected in Saskatoon on this material had limited utility to inform breeding decisions and selections. Therefore, this location (USDA zone

3b) would not be recommended to evaluate the level of winter hardiness of the progeny from a cross between a hardy rose and a non-hardy rose. It is reasonable to suggest that, perhaps,

Ottawa with its less extreme environment (zone 5a) than Saskatoon would have been a more informative environment for this research and that Saskatoon can remain a location-test for breeding material derived from crosses between hardy or semi-hardy roses. Furthermore, winter hardiness relies upon the overall health of the plant and severe disease pressure negatively impacts winter survival as observed in Elora 2020. Therefore, it is critical to breed for pest- resistance and winter hardiness simultaneously. Trialed roses are rarely sprayed for pests, and under these circumstances, it is reasonable to assume that diseases contributed to most of the winter damage observed. Overall, this research emphasized two strategies of winter survival. The first strategy involves the successful entry into dormancy and into the first winter with the acquisition of winter hardiness. It relies upon the accumulation of carbohydrate reserves and the timing of dormancy and growth cessation, and it is directly impacted by the overall health of the plant, meaning that disease resistance is a critical feature. The second strategy corresponds to the regrowth after winter. It relies upon the ability of the rose bush to not depend entirely on stored carbon reserves but also on newly fixed carbon by new leaflets performing photosynthesis, and it is dependent on bud survival on the canes.

161

4.2.2 Future directions for the characterization of the genetic basis of cold hardiness in roses

This research presented the advantage to use two bi-parental populations and a multi- experiment approach to investigate the genetic basis of winter hardiness in roses. However, due to biological reasons (i.e. hip setting and germination rate) and practical reasons (i.e. feasibility of the experiments and limited resources), the size of the populations remains small and the QTL intervals fairly large, which made difficult the identification of candidate genes. In addition, one of the challenges of this research was to identify candidate genes using the cold-sensitive R. chinensis as a reference genome. The roses used in this study were genetically different from R. chinensis, especially R. kordessii-derived ‘George Vancouver’, implying that we can miss information on key genes involved in winter hardiness in hardy roses and that it is difficult to compare genes between species due to rearrangements, inversions and deletions across genomes.

Further research on bioinformatics and computational biology would be needed to sequence wild rose species, such as R. kordessii, that directly contributed to the exceptionally hardy Explorer roses and to use these sequences as references to identify variants from sequencing data of mapping populations derived from Explorer roses for mapping purposes and search for candidate genes. Further research will also be needed to identify SNPs and insertion/deletion between parents for the candidate genes from resequencing data. Nevertheless, this research suggested that QTLs for electrolyte leakage mapped to genomic regions that harbour CBF- and ICE1- homolgs. The expression profiles of these genes could be further characterized using targeted

RT-qPCR on roses subjected to cold stress either under natural or artificial conditions. It will be relevant to determine which housekeeping genes to use as reference in the qPCR with rose tissue

(Klie and Debener 2011). In addition, because the response to cold stress is tissue-dependent, preliminary experiments would have to be conducted in order to determine which tissue to use

162

(stems, leaves, bark or buds). In this research, the electrolyte leakage experiments were conducted on the stems in order to investigate cane freezing tolerance under artificial conditions and compare it to field cane die-back. However, since the response to cold stress is tissue- dependent, future experiments could focus on buds using either electrolyte leakage or digital thermal analysis as done by Kovaleski et al. (2018). Buds could be collected from plants artificially acclimated or dormant under natural conditions during mid-winter.

While a molecular marker associated with the major QTLs for freezing tolerance and winter hardiness in Elora 2019 (i.e. winter damage) could be developed for MAS, another approach would be to develop genomic selection. Genomic selection is particularly valuable when the phenotype is complex with the implication of numerous QTLs with small effects that are difficult to identify in linkage mapping. Genomic selection has proven to be useful in breeding polyploid Rosaceous crops (Gezan et al. 2017). In addition, genomic selection for winter hardiness has proven to be a valuable tool to improve winter wheat (Michel et al. 2019), and the authors suggested to combine measures of field winter hardiness and frost tolerance measured under artificial conditions in a genomic selection index. Future research on roses could focus on developing genomic selection for winter hardiness and combine measures of field winter damage, field spring regrowth and measurements of electrolyte leakage in a genomic selection index.

4.3 Suggested selections

On the basis of the findings from the second and third chapters, five roses from the cross

‘CA60’ x ‘Singing in the Rain’ were notable. The five roses stood out in detached leaf assay and field trials as black spot disease resistant, and they possessed the ‘CA60’Rdr1A resistance allele.

In addition, they had less winter damage in average in Elora 2019 that the other hybrids. Only

163

the fifth selection showed above-average defoliation in the summer of 2019 in Elora, and further field screening would be necessary to assess the genotype’ sensitivity to sawflies. All five genotypes had good spring regrowth. The five roses also had attractive aesthetic features, such as full flowers, vibrant colours and glossy foliage. Moreover, one rose genotype attracted attention because of its stunning bi-colour and striped flowers. This genotype was susceptible to black spot, but further backcrosses could be considered to introduce hardiness and disease resistance

(Figure 4-1).

4.4 Tools for polyploids

The genetic inheritance in roses is complex, partly due to the polyploidy. In order to overcome the challenges associated with polyploidy, we performed a pseudo-test cross strategy in order to generate the linkage maps of the parental genotypes using only the markers that segregated in simplex configuration, homozygous in one parent and heterozygous in the other

(Grattapaglia and Sederoff 1994). While this approach allowed us to generate quality genetic maps with good coverage and perform QTL analysis using common computational tools designed for diploid species, we lost genetic information carried by additional markers in other configurations and we could not generate a consensus map. In the last few years, collaborative efforts have been undertaken by the breeding community under the leadership of Dr. David

Byrne to develop computational tools specifically for polyploid crops

(https://www.polyploids.org/). From genotype calling, linkage analysis and haplotype reconstruction to QTL analysis, future research on polyploid roses will greatly benefit from these novel tools.

164

Figure 4-1. Five roses with increased disease resistance, exceptional hardiness and stunning aesthetic features from the 'CA60' x 'Singing in the Rain' population. The sixth rose (bottom right corner) does not possess black spot resistance, but its beautiful flowers would make it a good candidate for a backcrossing program (Credit Rouet C. Elora 2019).

165

REFERENCES

Akhtar, M., A. Jaiswal, G. Taj, J. P. Jaiswal, M. I. Qureshi, and N. K. Singh (2012) “DREB1/CBF Transcription Factors: Their Structure, Function and Role in Abiotic Stress Tolerance in Plants.” Journal of Genetics 91: 385–95. https://doi.org/10.1007/s12041-012-0201- 3.

Altschul SF, Gish W, Miller W, Myers EW, Lipman DJ (1990) Basic local alignment search tool. J Mol Biol. 215:403-10.

Arends D, Prins P, Jansen RC, and Broman KW (2010) R/Qtl: High-Throughput Multiple QTL Mapping. Bioinformatics 26:2990–92. https://doi.org/10.1093/bioinformatics/btq565.

Artlip T S, Wisniewski M E, and Norelli J L (2014) Field Evaluation of Apple Overexpressing a Peach CBF Gene Confirms Its Effect on Cold Hardiness, Dormancy, and Growth. Environmental and Experimental Botany 106:79–86. https://doi.org/10.1016/j.envexpbot.2013.12.008.

Artlip T S, Wisniewski M E (1997) Tissue-Specific Expression of a Dehydrin Gene in One- Year-Old `Rio Oso Gem’ Peach Trees. Journal of the American Society for Horticultural Science 122: 784–87. https://doi.org/10.21273/JASHS.122.6.784.

Bajji M, Kinet J M, Lutts S (2002) The Use of the Electrolyte Leakage Method for Assessing Cell Membrane Stability as a Water Stress Tolerance Test in Durum Wheat. Plant Growth Regulation 36: 61–70. https://doi.org/10.1023/A:1014732714549.

Banerjee A, Roychoudhury A (2016) Group II Late Embryogenesis Abundant (LEA) Proteins: Structural and Functional Aspects in Plant Abiotic Stress. Plant Growth Regulation; Dordrecht 79: 1–17. http://dx.doi.org.subzero.lib.uoguelph.ca/10.1007/s10725-015-0113-3.

Barrau J, Bonduel P, A. E. (2006) L’ABCdaire des Roses. Edition number FA 178009. Editions Flammarion, Paris.

Barros P, M., Gonçalves N, Saibo N J M, Oliveira M M (2012) Functional Characterization of Two Almond C-Repeat-Binding Factors Involved in Cold Response. Tree Physiology 32: 1113– 28. https://doi.org/10.1093/treephys/tps067.

166

Bassil NV, Davis T M, Zhang H, Ficklin S, Mittmann M, Webster T, Mahoney L, et al. (2015) Development and Preliminary Evaluation of a 90 K Axiom[R] SNP Array for the Allo-Octoploid Cultivated Strawberry Fragaria x Ananassa.” BMC Genomics 16. http://link.galegroup.com/apps/doc/A541364613/AONE?u=guel77241&sid=AONE&xid=e7cd8 520.

Bélanger, G., Y. Castonguay, A. Bertrand, C. Dhont, P. Rochette, L. Couture, R. Drapeau, D. Mongrain, F.-P. Chalifour, and R. Michaud (2006) “Winter Damage to Perennial Forage Crops in Eastern Canada: Causes, Mitigation, and Prediction.” Canadian Journal of Plant Science 86: 33–47. https://doi.org/10.4141/P04-171.

Bendahmane M, Dubois A, Raymond O, Le Bris M (2013) Genetics and Genomics of Flower Initiation and Development in Roses. Journal of Experimental Botany 64: 847–57. https://doi.org/10.1093/jxb/ers387.

Bernardo, R (2002) Breeding for Quantitative Traits in Plants. Woodbury, Minn: Stemma Press.

Bianco L, Cestaro A, Linsmith G, Muranty H, Denancé C, Théron A, Poncet C, et al. (2016) Development and Validation of the Axiom®Apple480K SNP Genotyping Array. The Plant Journal 86: 62–74. https://doi.org/10.1111/tpj.13145.

Bianco L, Cestaro A, Sargent D J, Banchi E, Derdak S, Guardo M D, Salvi S, et al. (2014) Development and Validation of a 20K Single Nucleotide Polymorphism (SNP) Whole Genome Genotyping Array for Apple (Malus × Domestica Borkh). PloS One 9: e110377. https://doi.org/10.1371/journal.pone.0110377.

Biber A, Kaufmann H, Linde M, Spiller M, Terefe D, Debener T (2010) Molecular Markers from a BAC Contig Spanning the Rdr1 Locus: A Tool for Marker-Assisted Selection in Roses. Theoretical and Applied Genetics 120: 765–73. https://doi.org/10.1007/s00122-009-1197-9.

Blechert O, Debener T (2005) Morphological Characterization of the Interaction between Diplocarpon Rosae and Various Rose Species. Plant Pathology 54: 82–90. https://doi.org/10.1111/j.1365-3059.2005.01118.x.

167

Bourke P M, Arens P, Voorrips R, Esselink G D, Koning‐Boucoiran C F S, van’t Westende W P C, Leonardo T S, et al. (2017) Partial Preferential Chromosome Pairing Is Genotype Dependent in Tetraploid Rose. The Plant Journal 90: 330–43. https://doi.org/10.1111/tpj.13496.

Bourke P M, van Geest G, Voorrips R E, Jansen J, Kranenburg T, Shahin A, Visser R G F, Arens P, Smulders M J M, Maliepaard C (2018) PolymapR-Linkage Analysis and Genetic Map Construction from F1 Populations of Outcrossing Polyploids. Bioinformatics (Oxford, England) 34: 3496–3502. https://doi.org/10.1093/bioinformatics/bty371.

Bowman J L, Smyth D R, Meyerowitz E M (1989) Genes Directing Flower Development in Arabidopsis. The Plant Cell 1: 37–52. https://doi.org/10.2307/3869060.

Bozovic, V. J., J. Schmitt, and C.Kohn (2013) “Dehydrins (LTI29, LTI30, COR47) from Arabidopsis Thaliana Expressed in Escherichia Coli Protect Thylakoid Membrane during Freezing.” Journal of the Serbian Chemical Society 78: 1149–60. https://doi.org/10.2298/JSC121127017B.

Broman KW, Gatti DM, Simecek P, Furlotte NA, Prins P, Sen Ś, Yandell BS, Churchill GA (2019) R/Qtl2: Software for Mapping Quantitative Trait Loci with High-Dimensional Data and Multiparent Populations. Genetics 211:495–502. https://doi.org/10.1534/genetics.118.301595.

Broman KW, Wu H, Sen Ś, Churchill GA (2003) R/Qtl: QTL Mapping in Experimental Crosses. Bioinformatics 19:889–90. https://doi.org/10.1093/bioinformatics/btg112.

Brouwer D J, Osborn T C (1999) A Molecular Marker Linkage Map of Tetraploid Alfalfa (Medicago Sativa L.). Theoretical and Applied Genetics 99: 1194–1200. https://doi.org/10.1007/s001220051324.

Bruneau A, Starr J R, Joly S (2007) Phylogenetic Relationships in the Genus Rosa: New Evidence from Chloroplast DNA Sequences and an Appraisal of Current Knowledge. Systematic Botany 32: 366–78.

Butler D. (2017) R Package ASMap: Efficient Genetic Linkage Map Construction and Diagnosis. Journal of Statistical Software 79 :1-29. https://doi.org/10.18637/jss.v079.i06

Calixto C P G, Guo W, James A B, Tzioutziou N A, Entizne J C, Panter P E, Knight H, Nimmo H G, Zhang R, Brown J W S (2018) Rapid and Dynamic Alternative Splicing Impacts the

168

Arabidopsis Cold Response Transcriptome. The Plant Cell 30: 1424–44. https://doi.org/10.1105/tpc.18.00177.

Carlson-Nilsson B U, Davidson C G (2000) Research on Blackspot Pathogen Diplocarpon rosae (Marssonina rosae) And Resistance in Rosa. Acta Horticulturae, no. 508: 141–148. https://doi.org/10.17660/ActaHortic.2000.508.18.

Carlson-Nilsson B U (2000) Resistance to Marssonina rosae in Rosa L. Seedlings Obtained from Controlled Crosses Including Germplasm L83. Acta Agriculturae Scandinavica, Section B-Plant Soil Science 50: 176–182.

Carlson-Nilsson, B. U., and C.G. Davidson (2006) “Variation in Resistance to Marssonina Rosae (Lib.) Died. among Different Rosa L. Cultivars and Species Including Three Dogrose Species (Rosa Sect. Caninae).” Scientia Horticulturae 109: 353–60. https://doi.org/10.1016/j.scienta.2006.05.008.

Carlson-Nilsson, B.U., and C. G. Davidson (2009) “A Three Year Study of Winter Hardiness and Blackspot Resistance (Diplocaron rosae Wolf) of Roses in Two Climatically Different Environments." Floriculture and Ornamental Biotechnology 3, 139-151.

Chagné D, Crowhurst R N, Troggio M, Davey M W, Gilmore B, Lawley C, Vanderzande S, et al (2012) Genome-Wide SNP Detection, Validation, and Development of an 8K SNP Array for Apple. PloS One 7: e31745. https://doi.org/10.1371/journal.pone.0031745.

Channelière S, Rivière S, Scalliet G, Szecsi J, Jullien F, Dolle C, Vergne P, et al. (2002) Analysis of Gene Expression in Rose Petals Using Expressed Sequence Tags. FEBS Letters 515: 35–38. https://doi.org/10.1016/S0014-5793(02)02413-4.

Cherri-Martin M, Jullien F, Heizmann P, Baudino S (2007) Fragrance Heritability in Hybrid Tea Roses. Scientia Horticulturae 113: 177–81. https://doi.org/10.1016/j.scienta.2007.03.002.

Chinnusamy, V., M. Ohta, S. Kanrar, B.H. Lee, Xuhui Hong, Manu Agarwal, and Jian-Kang Zhu (2003) “ICE1: A Regulator of Cold-Induced Transcriptome and Freezing Tolerance in Arabidopsis.” Genes & Development 17: 1043–54. https://doi.org/10.1101/gad.1077503.

169

Chung, S., and R.W. Parish (2008) “Combinatorial Interactions of Multiple Cis-Elements Regulating the Induction of the Arabidopsis XERO2 Dehydrin Gene by Abscisic Acid and Cold.” The Plant Journal 54: 15–29. https://doi.org/10.1111/j.1365-313X.2007.03399.x.

Close T J (1997) Dehydrins: A Commonalty in the Response of Plants to Dehydration and Low Temperature. Physiologia Plantarum 100: 291–96. https://doi.org/10.1111/j.1399- 3054.1997.tb04785.x.

Cock K (2007) Understanding Genetic Relationships of Wild and Cultivated Roses and the Use of Species in Breeding. CAB Reviews: Perspectives in Agriculture, Veterinary Science, Nutrition and Natural Resources 2. https://doi.org/10.1079/PAVSNNR20072052.

Crespel L, Chirollet M, Durel C, Zhang D, Meynet J, Gudin S (2002) Mapping of Qualitative and Quantitative Phenotypic Traits in Rosa Using AFLP Markers. Theoretical and Applied Genetics 105: 1207–14. https://doi.org/10.1007/s00122-002-1102-2.

De Vries D P, Dubois L A. M (1978) On the Transmission of the Yellow Flower Colour from Rosa foetida to Recurrent Flowering Hybrid Tea-Roses. Euphytica 27: 205–10. https://doi.org/10.1007/BF00039136.

Debener T (2019) The Beast and the Beauty: What Do we know about Black Spot in Roses?, Critical Reviews in Plant Sciences, https://doi.org/10.1080/07352689.2019.1665778

Debener T (2003) Marker Assisted Background Selection for the Introgression of Black Spot Resistance into Cultivated Roses. Europ.J.Hort.Sci., 68:245–252,

Debener T, Drewes‐Alvarez R, Rockstroh K (1998) Identification of Five Physiological Races of Blackspot, Diplocarpon Rosae, Wolf on Roses. Plant Breeding 117:267–70. https://doi.org/10.1111/j.1439-0523.1998.tb01937.x.

Debener T, Mattiesch L (1999) Construction of a Genetic Linkage Map for Roses Using RAPD and AFLP Markers.” Theoretical and Applied Genetics 99: 891–99. https://doi.org/10.1007/s001220051310.

Debener T, Mattiesch L, Vosman B (2001) A Molecular Marker Map for Roses. Acta Horticulturae, 547: 283–87. https://doi.org/10.17660/ActaHortic.2001.547.33.

170

Debener T, Linde M (2009) Exploring Complex Ornamental Genomes: The Rose as a Model Plant. Critical Reviews in Plant Sciences 28: 267–80. https://doi.org/10.1080/07352680903035481.

Debener T, Byrne D H (2014) Disease Resistance Breeding in Rose: Current Status and Potential of Biotechnological Tools. Plant Science, Disease resistance: molecular mechanisms and biotechnological applications, 228: 107–17. https://doi.org/10.1016/j.plantsci.2014.04.005.

Debener T, Gudin S, Roberts A (2003) Encyclopedia of Rose Science. 1 edition. Amsterdam; Boston: Academic Press.

Dennis L, Peacock J (2019) Genes Directing Flower Development in Arabidopsis.” The Plant Cell 31: 1192–93. https://doi.org/10.1105/tpc.19.00276.

Dexter S T, Tottingham W E, Graber L F (1930) Preliminary Results in Measuring the Hardiness of Plants 1. Plant Physiology 5: 215–23.

Dexter S T, Tottingham W E, Graber L F (1932) Investigations of the hardiness of plants by measurement of electrical conductivity. Plant Physiology 7: 63-78.

Dhont, C., Y. Castonguay, P. Nadeau, G. Bélanger, R. Drapeau, S. Laberge, J.C. Avice, and F.P. Chalifour (2006) “Nitrogen Reserves, Spring Regrowth and Winter Survival of Field-Grown Alfalfa (Medicago sativa) Defoliated in the Autumn.” Annals of Botany 97: 109–20. https://doi.org/10.1093/aob/mcj006.

Doherty, C.J., H. A. Van Buskirk, S. J. Myers, and M. F. Thomashow (2009) “Roles for Arabidopsis CAMTA Transcription Factors in Cold-Regulated Gene Expression and Freezing Tolerance.” The Plant Cell 21: 972–84. http://dx.doi.org.subzero.lib.uoguelph.ca/10.1105/tpc.108.063958.

Dohm A, C. Ludwig, D. Schilling D, T. Debener (2002) Transformation of Roses with Genes for Antifungal Proteins to Reduce their Susceptibility to Fungal Diseases. Acta Horticulturae, no. 572: 105–11. https://doi.org/10.17660/ActaHortic.2002.572.12.

Dong, C.H., M. Agarwal, Y. Zhang, Q. Xie, and J.K. Zhu (2006) “The Negative Regulator of Plant Cold Responses, HOS1, Is a RING E3 Ligase That Mediates the Ubiquitination and

171

Degradation of ICE1.” Proceedings of the National Academy of Sciences 103: 8281–86. https://doi.org/10.1073/pnas.0602874103.

Dugo M L, Satovic Z, Millán T, Cubero J I, Rubiales D, Cabrera A, Torres A M (2005) Genetic Mapping of QTLs Controlling Horticultural Traits in Diploid Roses. Theoretical and Applied Genetics 111: 511–20. https://doi.org/10.1007/s00122-005-2042-4.

Dumont E, Fontaine V, Vuylsteker C, Sellier H, Bodèle S, Voedts N, Devaux R, et al. (2009) Association of Sugar Content QTL and PQL with Physiological Traits Relevant to Frost Damage Resistance in Pea under Field and Controlled Conditions. Theoretical and Applied Genetics 118: 1561–71. https://doi.org/10.1007/s00122-009-1004-7.

Elshire R J, Glaubitz J C, Sun Q, Poland J A, Kawamoto K, Buckler E S, Mitchell S E (2011) A Robust, Simple Genotyping-by-Sequencing (GBS) Approach for High Diversity Species. PLoS ONE 6: e19379. https://doi.org/10.1371/journal.pone.0019379.

Feng H, Wang M L, Cong R C, Dai S L (2017) Colchicine- and Trifluralin-Mediated Polyploidization of Rosa Multiflora Thunb. Var. Inermis and Rosa Roxburghii f. Normalis. The Journal of Horticultural Science and Biotechnology 92: 279–87. https://doi.org/10.1080/14620316.2016.1249964.

Flament I, Debonneville C, Furrer A (1993) Volatile Constituents of Roses. Characterization of Cultivars Based on the Headspace Analysis of Living Flower Emissions. In Bioactive Volatile Compounds from Plants Chapter 19:269-281. ACS Symposium Series Vol. 525 ISBN13: 9780841226395eISBN: 9780841213814. American Chemical Society. https://doi.org/ 10.1021/bk-1993-0525.ch019

Fofana B, Ghose K, Chapman B, Sanderson K (2013) Genetic, Agronomy, and Metabolomics of Prince Edwards Island Wild Rose Collection and Promise for Cultivar Development. Using Old Solutions to New Problems - Natural Drug Discovery in the 21st Century, June. https://doi.org/10.5772/54688.

Fougère-Danezan M, Joly S, Bruneau A, Gao X F, Zhang L B (2015) Phylogeny and Biogeography of Wild Roses with Specific Attention to Polyploids. Annals of Botany 115: 275– 91. https://doi.org/10.1093/aob/mcu245.

172

Frutos, E., Galindo, M.P., Leiva, V (2014) An interactive biplot implementation in R for modeling genotype-by-environment interaction. Stoch. Environ. Res. Risk Assess. 28:1629- 1641.

Gachomo EW, Dehne HW, Steiner U (2006) Microscopic Evidence for the Hemibiotrophic Nature of Diplocarpon Rosae, Cause of Black Spot Disease of Rose. Physiological and Molecular Plant Pathology 69:86–92. https://doi.org/10.1016/j.pmpp.2007.02.002.

Gachomo E W, Kotchoni S O (2007) Detailed Description of Developmental Growth Stages of Diplocarpon rosae in Rosa: A Core Building Block for Efficient Disease Management.” Annals of Applied Biology 151: 233–43. https://doi.org/10.1111/j.1744-7348.2007.00167.x.

Gachomo E W, Kotchoni S O (2010) Microscopic and Biochemical Evidence of Differentially Virulent Field Isolates of Diplocarpon rosae Causing Black Spot Disease of Roses. Plant Physiology and Biochemistry 48: 167–75. https://doi.org/10.1016/j.plaphy.2010.01.003.

Gachomo E W (2005) Studies of the Life Cycle of Diplocarpon rosae Wolf on Roses and the Effectiveness of Fungicides on Pathogenesis. Cuvillier Verlag.

Gachomo E W, Dehne H W, Steiner U (2006) Microscopic Evidence for the Hemibiotrophic Nature of Diplocarpon rosae, Cause of Black Spot Disease of Rose. Physiological and Molecular Plant Pathology 69: 86–92. https://doi.org/10.1016/j.pmpp.2007.02.002.

Gallais A. Quantitative genetics and breeding methods in autopolyploid plants. INRA, Paris, 2003.

Gallegos J E (2018) Alternative Splicing Plays a Major Role in Plant Response to Cold Temperatures. The Plant Cell, June, tpc.00430.2018. https://doi.org/10.1105/tpc.18.00430.

Gery C, Zuther E, Schulz E, Legoupi J, Chauveau A, McKhann H, Hincha D K, Téoulé E (2011) Natural Variation in the Freezing Tolerance of Arabidopsis Thaliana: Effects of RNAi-Induced CBF Depletion and QTL Localisation Vary among Accessions. Plant Science 180: 12–23. https://doi.org/10.1016/j.plantsci.2010.07.010.

Gezan, S. A., L. F. Osorio, S. Verma, and V. M. Whitaker (2017) “An Experimental Validation of Genomic Selection in Octoploid Strawberry.” Horticulture Research 4: 1–9. https://doi.org/10.1038/hortres.2016.70.

173

Gilmour S J, Zarka D G, Stockinger E J, Salazar M P, Houghton J M, Thomashow M F (1998) Low Temperature Regulation of the Arabidopsis CBF Family of AP2 Transcriptional Activators as an Early Step in Cold-Induced COR Gene Expression. The Plant Journal 16: 433–42. https://doi.org/10.1046/j.1365-313x.1998.00310.x.

Gitonga, V.W., C. F.S. Koning-Boucoiran, K. Verlinden, O. Dolstra, R. GF Visser, C. Maliepaard, and F. A. Krens (2014) “Genetic Variation, Heritability and Genotype by Environment Interaction of Morphological Traits in a Tetraploid Rose Population.” BMC Genetics 15: 146. https://doi.org/10.1186/s12863-014-0146-z.

Granados Mendoza C, Wanke S, Goetghebeur P, Samain M S (2013) Facilitating Wide Hybridization in Hydrangea s. l. Cultivars: A Phylogenetic and Marker-Assisted Breeding Approach. Molecular Breeding 32: 233–39. https://doi.org/10.1007/s11032-012-9822-8.

Grattapaglia D, Sederoff R (1994) Genetic Linkage Maps of Eucalyptus Grandis and Eucalyptus Urophylla Using a Pseudo-Testcross: Mapping Strategy and Rapd Markers. Genetics 137: 1121– 37.

Guo, W., R. W. Ward, and M. F. Thomashow (1992) “Characterization of a Cold-Regulated Wheat Gene Related to Arabidopsis Cor47.” Plant Physiology 100: 915–22.

Gusta, L. V., B. J. O’Connor, and M. G. MacHutcheon (1997) “The Selection of Superior Winter-Hardy Genotypes Using a Prolonged Freeze Test.” Canadian Journal of Plant Science 77: 15–21. https://doi.org/10.4141/P95-197.

Guterman I, Shalit M, Menda N, Piestun D, et al. (2002) Rose Scent: Genomics Approach to Discovering Novel Floral Fragrance-Related Genes. Plant Cell; Rockville 14: 2325–38.

Guy, C. L. (1990) “Cold Acclimation and Freezing Stress Tolerance: Role of Protein Metabolism.” Annual Review of Plant Physiology, Plant Molecular Biology 41:187-223.

Guy C L, Niemi K J, Brambl R (1985) Altered Gene Expression during Cold Acclimation of Spinach. Proceedings of the National Academy of Sciences of the United States of America 82: 3673–77.

Hackett C A, McLean K, Bryan G J (2013) Linkage Analysis and QTL Mapping Using SNP Dosage Data in a Tetraploid Potato Mapping Population. PLoS ONE 8: e63939.

174

Haimi P, Vinskienė J, Stepulaitienė I, Baniulis D, Stanienė G, Šikšnianienė J B, Rugienius R (2017) Patterns of Low Temperature Induced Accumulation of Dehydrins in Rosaceae Crops— Evidence for Post-Translational Modification in Apple. Journal of Plant Physiology 218: 175– 81. https://doi.org/10.1016/j.jplph.2017.08.008.

Haley CS, Knott SA (1992) A Simple Regression Method for Mapping Quantitative Trait Loci in Line Crosses Using Flanking Markers. Heredity 69: 315. https://doi.org/10.1038/hdy.1992.131.

Hattendorf A, Linde M, Mattiesch L, Debener T, Kaufmann H (2004) Genetic Analysis Of Rose Resistance Genes And Their Localisation In The Rose Genome. Acta Horticulturae, no. 651: 123–30. https://doi.org/10.17660/ActaHortic.2004.651.14.

He J, Zhao X, Laroche A, Lu Z X, Liu H K, Li Z (2014) Genotyping-by-Sequencing (GBS), an Ultimate Marker-Assisted Selection (MAS) Tool to Accelerate Plant Breeding. Frontiers in Plant Science 5. https://doi.org/10.3389/fpls.2014.00484.

Hibrand-Saint Oyant L, Crespel L, Rajapakse S, Zhang L, Foucher F (2007) Genetic Linkage Maps of Rose Constructed with New Microsatellite Markers and Locating QTL Controlling Flowering Traits. Tree Genetics & Genomes 4: 11. https://doi.org/10.1007/s11295-007-0084-2.

Hibrand Saint-Oyant L, Ruttink T, Hamama L, Kirov I, Lakhwani D, Zhou N N, Bourke P M, et al. (2018) A High-Quality Genome Sequence of Rosa chinensis to Elucidate Ornamental Traits. Nature Plants 4: 473. https://doi.org/10.1038/s41477-018-0166-1.

Horst K R, Cloyd R (2007) Compendium of Rose Diseases and Pests, Second Edition. 2007. http://www.greencure.net/compendium_of_rose_diseases_book_ken_horst.asp.

Hummer K E, Janick J (2009) Rosaceae: Taxonomy, Economic Importance, Genomics. In Genetics and Genomics of Rosaceae. https://doi.org/10.1007/978-0-387-77491-6_1.

Iwata H, Gaston A, Remay A, Thouroude T, Jeauffre J, Kawamura K, Hibrand-Saint Oyant L, Araki T, Denoyes B, Foucher F (2012) “The TFL1 Homologue KSN Is a Regulator of Continuous Flowering in Rose and Strawberry. The Plant Journal: For Cell and Molecular Biology 69: 116–25. https://doi.org/10.1111/j.1365-313X.2011.04776.x.

Jan C H, Byrne D H, Manhart J, Wilson H (1999) Rose Germplasm Analysis with RAPD Markers. HortScience 34: 341–45. https://doi.org/10.21273/HORTSCI.34.2.341.

175

Jeliazkova EA (2012) Evaluating Natural Products for Control of Black Spot Disease on Roses. https://doi.org/10.7275/r58050j3.

Judd W S (2008) Plant Systematics: A Phylogenetic Approach. 3rd ed. Sunderland, Mass.: Sinauer Associates, Inc, Sinauer Associates.

Kalcsits, L. A., Silim, S., Tanino, K. (2009) “Warm temperature accelerates short photoperiod- induced growth cessation and dormancy induction in hybrid poplar (Populus x spp.).” Trees 23:971–979.

Karam F H, Sullivan J A (1991) A Rapid Method for Detection of Cold Hardiness in Roses. HortScience 26: 59–60. https://doi.org/10.21273/HORTSCI.26.1.59.

Katsumoto Y, Fukuchi-Mizutani M, Fukui Y, Brugliera F, Holton T A, Karan M, Nakamura N, et al. (2007) Engineering of the Rose Flavonoid Biosynthetic Pathway Successfully Generated Blue-Hued Flowers Accumulating Delphinidin. Plant and Cell Physiology 48 (11): 1589–1600. https://doi.org/10.1093/pcp/pcm131.

Kaufmann H, Mattiesch L, Lörz H, Debener T (2003) Construction of a BAC Library of Rosa Rugosa Thunb. and Assembly of a Contig Spanning Rdr1, a Gene That Confers Resistance to Blackspot. Molecular Genetics and Genomics 268: 666–74. https://doi.org/10.1007/s00438-002- 0784-0.

Kim SH, Kwon SI, Saha D, Anyanwu NC, Gassmann W (2009) Resistance to the Pseudomonas Syringae Effector HopA1 Is Governed by the TIR-NBS-LRR Protein RPS6 and Is Enhanced by Mutations in SRFR1. Plant Physiology 150:1723–32. https://doi.org/10.1104/pp.109.139238.

Kim J, Wolyn D J. (2015) Freezing Tolerance Assessment for Seedlings of Three Asparagus Cultivars Grown under Controlled Conditions. Canadian Journal of Plant Science 95: 495–504. https://doi.org/10.4141/cjps-2014-359.

Kitashiba H, Ishizaka T, Isuzugawa K, Nishimura K, Suzuki T (2004) Expression of a Sweet Cherry DREB1/CBF Ortholog in Arabidopsis Confers Salt and Freezing Tolerance. Journal of Plant Physiology 161: 1171–76. https://doi.org/10.1016/j.jplph.2004.04.008.

176

Klie, M., and T. Debener (2011) “Identification of Superior Reference Genes for Data Normalisation of Expression Studies via Quantitative PCR in Hybrid Roses (Rosa Hybrida).” BMC Research Notes 4: 518. https://doi.org/10.1186/1756-0500-4-518.

Koning-Boucoiran C F S, Gitonga V W, Yan Z, Dolstra O, van der Linden C G, van der Schoot J, Uenk G E, et al. (2012) The Mode of Inheritance in Tetraploid Cut Roses. Theoretical and Applied Genetics 125: 591–607. https://doi.org/10.1007/s00122-012-1855-1.

Koning-Boucoiran C F S, Esselink G D, Vukosavljev M, van ’t Westende W P C, Gitonga V W, Krens F A, Voorrips R E, et al. (2015) Using RNA-Seq to Assemble a Rose Transcriptome with More than 13,000 Full-Length Expressed Genes and to Develop the WagRhSNP 68k Axiom SNP Array for Rose (Rosa L.). Frontiers in Plant Science 6. https://doi.org/10.3389/fpls.2015.00249.

Koopman W J M, Wissemann V, De Cock K, Van Huylenbroeck J, De Riek J, Sabatino G J H, Visser D, et al. (2008) AFLP Markers as a Tool to Reconstruct Complex Relationships: A Case Study in Rosa (Rosaceae). American Journal of Botany 95: 353–66. https://doi.org/10.3732/ajb.95.3.353.

Koskela E A, Mouhu K, Albani M C, Kurokura T, Rantanen M, Sargent D J, Battey N H, Coupland G, Elomaa P, and Hytönen T (2012) Mutation in TERMINAL FLOWER1 Reverses the Photoperiodic Requirement for Flowering in the Wild Strawberry Fragaria Vesca. Plant Physiology 159: 1043–54. https://doi.org/10.1104/pp.112.196659.

Kovaleski, A. P., B. I. Reisch, and J. P. Londo. 2018. “Kinetics of Winter Deacclimation in Response to Temperature Determines Dormancy Status and Quantifies Chilling Requirement in Grapevines (Vitis Spp.).” BioRxiv, 256362. https://doi.org/10.1101/256362.

Kurepin L V, Dahal K P, Savitch L V, Singh J, Bode R, Ivanov A G, Hurry V, Hüner N P A (2013) Role of CBFs as Integrators of Chloroplast Redox, Phytochrome and Plant Hormone Signaling during Cold Acclimation. International Journal of Molecular Sciences 14: 12729–63. https://doi.org/10.3390/ijms140612729.

Le L, Rui M, Sun X F, Qiong W, Hui L, Pan H T, Zhang Q X (2012) Preliminary Studies on Cold Resistance Evaluation of Modern Roses. Acta Horticulturae, no. 9372: 889–900.

177

Li H (2013) Aligning Sequence Reads, Clone Sequences and Assembly Contigs with BWA- MEM. ArXiv:1303.3997 [q-Bio], March. http://arxiv.org/abs/1303.3997.

Iezzoni, A.F., McFerson, J., Luby, J. et al. (2020) RosBREED: bridging the chasm between discovery and application to enable DNA-informed breeding in rosaceous crops. Hortic Res 7, 177. https://doi.org/10.1038/s41438-020-00398-7

Li, H., Y. Ding, Y. Shi, X. Zhang, S. Zhang, Z. Gong, and S. Yang (2017) “MPK3- and MPK6- Mediated ICE1 Phosphorylation Negatively Regulates ICE1 Stability and Freezing Tolerance in Arabidopsis.” Developmental Cell 43: 630-642.e4. https://doi.org/10.1016/j.devcel.2017.09.025.

Li H, Handsaker B, Wysoker A, Fennell T, Ruan J, Homer N, Marth G, Abecasis G, Durbin R (2009) The Sequence Alignment/Map format and SAMtools. Bioinformatics 25:2078–2079

Li P H, Sakai A (1978) Plant Cold Hardiness and Freezing Stress: Mechanisms and Crop Implications. 1st edition. New York: Academic Press.

Li X, Wei Y, Acharya A, Jiang Q, Kang J, Brummer E C (2014) A Saturated Genetic Linkage Map of Autotetraploid Alfalfa (Medicago Sativa L.) Developed Using Genotyping-by- Sequencing Is Highly Syntenous with the Medicago Truncatula Genome. G3: Genes, Genomes, Genetics 4: 1971–79. https://doi.org/10.1534/g3.114.012245.

Lin, C., and M. F. Thomashow (1992) “DNA Sequence Analysis of a Complementary DNA for Cold-Regulated Arabidopsis Gene Cor15 and Characterization of the COR 15 Polypeptide.” Plant Physiology 99: 519–25. https://doi.org/10.1104/pp.99.2.519.

Lindén L, Palonen P, Lindén M (2000) Relating Freeze-Induced Electrolyte Leakage Measurements to Lethal Temperature in Red Raspberry. Journal of the American Society for Horticultural Science 125: 429–35. https://doi.org/10.21273/JASHS.125.4.429.

Ling, J., Y. Xiao, J. Hu, F. Wang, F. Ouyang, J. Wang, Y. Weng, and H. Zhang (2021) “Genotype by Environment Interaction Analysis of Growth of Picea Koraiensis Families at Different Sites Using BLUP-GGE.” New Forests 52: 113–27. https://doi.org/10.1007/s11056- 020-09785-3.

178

Longhi S, Giongo L, Buti M, Surbanovski N, Viola R, Velasco R, Ward J A, Sargent D J (2014) Molecular Genetics and Genomics of the Rosoideae: State of the Art and Future Perspectives. Horticulture Research 1: 1. https://doi.org/10.1038/hortres.2014.1.

Lühmann A K, Linde M, Debener T (2010) Genetic Diversity of Diplocarpon rosae: Implications on Practical Breeding. Acta Horticulturae, no. 870: 157–62. https://doi.org/10.17660/ActaHortic.2010.870.20.

Ma C X, Casella G, Shen Z J, Osborn T C, Wu R (2002) A Unified Framework for Mapping Quantitative Trait Loci in Bivalent Tetraploids Using Single-Dose Restriction Fragments: A Case Study from Alfalfa. Genome Research 12: 1974–81. https://doi.org/10.1101/gr.320202.

Mackay, W. A., S. W. George, C. McKenney, J. J. Sloan, R. I. Cabrera, J. A. Reinert, P. Colbaugh, L. Lockett, and W. Crow (2008 “Performance of Garden Roses in North-Central Texas under Minimal Input Conditions.” HortTechnology 18: 417–22. https://doi.org/10.21273/HORTTECH.18.3.417.

Magnard J L, Roccia A, Caissard J C, Vergne P, Sun P, Hecquet R, Dubois A, Hibrand-Saint Oyant L, Jullien F, Nicolè F, Raymond O, Huguet S, Baltenweck R, Meyer S, Claudel P, Jeauffre J, Rohmer M, Foucher F, Hugueney P, Bendahmane M, Baudino S (2015) Biosynthesis of monoterpene scent compounds in roses. Science 349-81. https://doi.org/10.1126/science.aab0696

Mahajan S, Tuteja N (2005) Cold, Salinity and Drought Stresses: An Overview. Archives of Biochemistry and Biophysics 444: 139–58. https://doi.org/10.1016/j.abb.2005.10.018.

Malek von B, Debener T (1998) Genetic Analysis of Resistance to Blackspot (Diplocarpon rosae) in Tetraploid Roses. Theoretical and Applied Genetics 96: 228–31. https://doi.org/10.1007/s001220050731.

Mantyla, E., V. Lang, and E. T. Palva (1995) “Role of Abscisic Acid in Drought-Induced Freezing Tolerance, Cold Acclimation, and Accumulation of LT178 and RAB18 Proteins in Arabidopsis Thaliana.” Plant Physiology 107: 141–48.

Marchant R, Davey M R, Lucas J A, Lamb C J, Dixon R A, Power J B (1998) Expression of a Chitinase Transgene in Rose (Rosa Hybrida L.) Reduces Development of Blackspot Disease

179

(Diplocarpon rosae Wolf).” Molecular Breeding 4: 187–94. https://doi.org/10.1023/A:1009642707505.

Marone D, Russo M A, Laidò G, De Vita P, Papa R, Blanco A, Gadaleta A, Rubiales D, Mastrangelo A M (2013) Genetic Basis of Qualitative and Quantitative Resistance to Powdery Mildew in Wheat: From Consensus Regions to Candidate Genes. BMC Genomics 14: 562. https://doi.org/10.1186/1471-2164-14-562.

Mathew, I., H. Shimelis, L. Mwadzingeni, R. Zengeni, M. Mutema, and V. Chaplot (2018) “Variance Components and Heritability of Traits Related to Root: Shoot Biomass Allocation and Drought Tolerance in Wheat.” Euphytica 214: 225. https://doi.org/10.1007/s10681-018-2302-4.

Matsumoto S, Kouchi M, Yabuki J, Kusunoki M, Ueda Y, Fukui H (1998) Phylogenetic Analyses of the Genus Rosa Using the MatK Sequence: Molecular Evidence for the Narrow Genetic Background of Modern Roses. Scientia Horticulturae 77: 73–82. https://doi.org/10.1016/S0304-4238(98)00169-1.

Medina, J., M. Bargues, J. Terol, M. Pérez-Alonso, and J. Salinas (1999) “The Arabidopsis CBF Gene Family Is Composed of Three Genes Encoding AP2 Domain-Containing Proteins Whose Expression Is Regulated by Low Temperature but Not by Abscisic Acid or Dehydration.” Plant Physiology 119: 463–70.

Menz I, Straube J, Linde M, Debener T (2018) The TNL Gene Rdr1 Confers Broad-Spectrum Resistance to Diplocarpon rosae.” Molecular Plant Pathology 19: 1104–13. https://doi.org/10.1111/mpp.12589.

McKenna A, Hanna M, Banks E, Sivachenko A, Cibulskis K, Kernytsky A, Garimella K, Altshuler D, Gabriel S, Daly M, DePristo MA (2010) The Genome Analysis Toolkit: A MapReduce framework for analyzing next-generation DNA sequencing data. Genome Res 20:1297–1303

Michel, S., F. Löschenberger, J. Hellinger, V. Strasser, C. Ametz, B. Pachler, E. Sparry, and H. Bürstmayr. 2019. “Improving and Maintaining Winter Hardiness and Frost Tolerance in Bread Wheat by Genomic Selection.” Frontiers in Plant Science 10. https://doi.org/10.3389/fpls.2019.01195.

180

Milne I, Stephen G, Bayer M, Cock PJA, Pritchard L, Cardle L, Shaw PD, Marshall D (2013) Using Tablet for visual exploration of second-generation sequencing data. Brief Bioinform 14:193–202

Mohapatra A, Rout G R (2005) “Study of Embryo Rescue in Floribunda Rose. Plant Cell, Tissue and Organ Culture 81: 113–17. https://doi.org/10.1007/s11240-004-3128-4.

Nakagawa, S., and H. Schielzeth (2010) “Repeatability for Gaussian and Non-Gaussian Data: A Practical Guide for Biologists.” Biological Reviews 85: 935–56. https://doi.org/10.1111/j.1469- 185X.2010.00141.x.

Neu E, Debener T (2019) Prediction of the Diplocarpon rosae Secretome Reveals Candidate Genes for Effectors and Virulence Factors. Fungal Biology 123: 231–39. https://doi.org/10.1016/j.funbio.2018.12.003.

Neu E, Featherston J, Rees J, Debener T (2017) A Draft Genome Sequence of the Rose Black Spot Fungus Diplocarpon rosae Reveals a High Degree of Genome Duplication. PLOS ONE 12: e0185310. https://doi.org/10.1371/journal.pone.0185310.

Nguyen T H N, Winkelmann T, Debener T (2019) Development of Markers for Shoot Organogenesis in Roses. Acta Horticulturae, no. 1232: 7–14. https://doi.org/10.17660/ActaHortic.2019.1232.2.

Novillo F, Medina J, Salinas J (2007) Arabidopsis CBF1 and CBF3 Have a Different Function than CBF2 in Cold Acclimation and Define Different Gene Classes in the CBF Regulon.” Proceedings of the National Academy of Sciences of the United States of America 104: 21002– 7.

Ogilvie I S, Agriculture Canada. Communications Branch (1993) Winter-Hardy Roses from Agriculture Canada. Publication (Canada. Agriculture Canada).

Ooijen Van J W (2006) JoinMap®4, Software for the Calculation of Genetic Linkage Maps in Experimental Populations. Kyazma BV, Wageningen.

Ouyang L, Leus L, De Keyser E, Van Labeke M C (2019) Seasonal Changes in Cold Hardiness and Carbohydrate Metabolism in Four Garden Rose Cultivars. Journal of Plant Physiology 232: 188–99. https://doi.org/10.1016/j.jplph.2018.12.001.

181

Ouyang L, Leus L, Van Labeke M C (2019) Three-Year Screening for Cold Hardiness of Garden Roses. Scientia Horticulturae 245: 12–18. https://doi.org/10.1016/j.scienta.2018.10.003.

Palmer J G, Semeniuk P, Stewart R N (1966) a. Roses and blackspot. I. Pathogenicity to excised leaflets of Diplocarpon rosae from seven geographic locations. Phytopathology 56: 1277-1282.

Palmer J G, Semeniuk P, Stewart R N (1966) b. Roses and Blackspot. II. Seasonal variation in host susceptibility and decline of virulence in culture of conidia from Diplocarpon rosae. Phytopathology 56: 1283-1286.

Panjtandoust M, Wolyn D J (2016) Asparagus Cultivars with Varying Adaptation to Southern Ontario Differ for Induction of Freezing Tolerance in the Fall. Canadian Journal of Plant Science 96: 252–64. https://doi.org/10.1139/cjps-2015-0242.

Patel S (2013) Rose Hips as Complementary and Alternative Medicine: Overview of the Present Status and Prospects. Mediterranean Journal of Nutrition and Metabolism 6: 89–97. https://doi.org/10.3233/s12349-012-0118-7.

Pati P K, Sharma M, Ahuja PS (2008) Rose Protoplast Isolation and Culture and Heterokaryon Selection by Immobilization in Extra Thin Alginate Film. Protoplasma 233: 165–71. https://doi.org/10.1007/s00709-008-0297-8.

Peace C, Bassil N, Main D, Ficklin S, Rosyara U R, Stegmeir T, Sebolt A, et al. (2012) Development and Evaluation of a Genome-Wide 6K SNP Array for Diploid Sweet Cherry and Tetraploid Sour Cherry. PLoS ONE 7. https://doi.org/10.1371/journal.pone.0048305.

Pearce R S (1999) Molecular Analysis of Acclimation to Cold. Plant Growth Regulation 29: 47– 76. https://doi.org/10.1023/A:1006291330661.

Peixoto M M, Sage R F (2016) Improved Experimental Protocols to Evaluate Cold Tolerance Thresholds in Miscanthus and Switchgrass Rhizomes. GCB Bioenergy 8: 257–268. https://doi.org/10.1111/gcbb.12301.

Peng Y, Leung HCM, Yiu SM, Chin FYL (2012) IDBA-UD: a de novo assembler for single-cell and metagenomic sequencing data with highly uneven depth. Bioinformatics 28, 1420– 1428, https://doi.org/10.1093/bioinformatics/bts174.

182

Poirier M, Lacointe A, Améglio T (2010) A Semi-Physiological Model of Cold Hardening and Dehardening in Walnut Stem. Tree Physiology 30: 1555–69. https://doi.org/10.1093/treephys/tpq087.

Puhakainen, T., M. W. Hess, P. Mäkelä, J. Svensson, P. Heino, and E. Tapio Palva (2004) “Overexpression of Multiple Dehydrin Genes Enhances Tolerance to Freezing Stress in Arabidopsis.” Plant Molecular Biology 54: 743–53. https://doi.org/10.1023/B:PLAN.0000040903.66496.a4.

R Core Team (2019) R: A Language and Environment for Statistical Computing. https ://www.R-proje ct.org/

Rafalski A (2002) Applications of Single Nucleotide Polymorphisms in Crop Genetics. Current Opinion in Plant Biology 5: 94–100.

Rajapakse S, Byrne D H, Zhang L, Anderson N, Arumuganathan K, Ballard R E (2001) Two Genetic Linkage Maps of Tetraploid Roses. Theoretical and Applied Genetics 103: 575–83. https://doi.org/10.1007/PL00002912.

Raymond O, Gouzy J, Just J, Badouin H, Verdenaud M, Lemainque A, Vergne P, et al. (2018) The Rosa Genome Provides New Insights into the Domestication of Modern Roses. Nature Genetics 50: 772–77. https://doi.org/10.1038/s41588-018-0110-3.

Rehder A (1940) Manual of cultivated trees and shrubs hardy in North America, ed. 2. New York: Macmillan.

Rihan H Z, Al-Issawi M, Fuller M P (2017) Advances in Physiological and Molecular Aspects of Plant Cold Tolerance. Journal of Plant Interactions 12: 143–57. https://doi.org/10.1080/17429145.2017.1308568.

Roberts AV, Gladis T, Brumme H (2009) DNA Amounts of Roses (Rosa L.) and Their Use in Attributing Ploidy Levels. Plant Cell Reports 28: 61–71. https://doi.org/10.1007/s00299-008- 0615-9.

Robinson JT, Thorvaldsdóttir H, Winckler W, Guttman M, Lander ES, Getz G, Mesirov JP (2011) Integrative Genomics Viewer. Nature Biotechnology 29:24–26. https://doi.org/10.1038/nbt.1754.

183

Roman H, Rapicault M, Miclot A S, Larenaudie M, Kawamura K, Thouroude T, Chastellier A, et al. (2015) Genetic Analysis of the Flowering Date and Number of Petals in Rose. Tree Genetics & Genomes 11: 85. https://doi.org/10.1007/s11295-015-0906-6.

Rouet, C., E. A. Lee, T. Banks, J. O’Neill, R. LeBlanc, and D.J. Somers (2019) “Identification of a Polymorphism within the Rosa Multiflora MuRdr1A Gene Linked to Resistance to Multiple Races of Diplocarpon rosae W. in Tetraploid Garden Roses (Rosa × Hybrida).” Theoretical and Applied Genetics. https://doi.org/10.1007/s00122-019-03443-9.

Rout G R, Mohapatra A (2006) Use of Molecular Markers in Ornamental Plants: A Critical Reappraisal. European Journal of Horticultural Science 71: 53–68.

Săulescu NN, Braun HJ (2001) Cold tolerance. In: Application of Physiology in Wheat Breeding (eds Reynold MP, Ortiz-Monasterio JI, McNab A), pp. 111–123. Mexico, Cimmyt

Schultz D F, Schott R T, Voorips R E, Smulders M J M, Linde M, Debener T (2016) Genome- Wide Association Analysis of the Anthocyanin and Carotenoid Contents of Rose Petals. Front Plant Sci.; 7: 1798. https://doi.org/10.3389/fpls.2016.01798.

Shi, H., Y. Chen, Y. Qian, and Z. Chan (2015) “Low Temperature-Induced 30 (LTI30) Positively Regulates Drought Stress Resistance in Arabidopsis: Effect on Abscisic Acid Sensitivity and Hydrogen Peroxide Accumulation.” Frontiers in Plant Science 6. https://doi.org/10.3389/fpls.2015.00893.

Shi Y, Ding Y, Yang S (2018) Molecular Regulation of CBF Signaling in Cold Acclimation. Trends in Plant Science 23: 623–37. https://doi.org/10.1016/j.tplants.2018.04.002.

Shulaev V, Korban S S, Sosinski B, Abbott A G, Aldwinckle H S, Folta K M, Iezzoni A, et al. (2008) Multiple Models for Rosaceae Genomics. Plant Physiology 147: 985–1003. https://doi.org/10.1104/pp.107.115618.

Sievers F, Wilm A, Dineen DG, Gibson TJ, Karplus K, Li W, Lopez R, McWilliam H, Remmert M, Söding J, Thompson JD, Higgins DG (2011) Fast, scalable generation of high-quality protein multiple sequence alignments using Clustal Omega. Molecular Systems Biology 7:539 https://doi.org/10.1038/msb.2011.75.

184

Smallwood M, Bowles D J (2002) Plants in a Cold Climate. Philosophical Transactions: Biological Sciences 357: 831–47.

Smith BL, Lu CP, Alvarado Bremer JR (2010) High-Resolution Melting Analysis (HRMA): A Highly Sensitive Inexpensive Genotyping Alternative for Population Studies. Molecular Ecology Resources 10:193–96. https://doi.org/10.1111/j.1755-0998.2009.02726.x.

Smulders M J M, Arens P, Bourke P M, Debener T, Linde M, De Riek J, Leus L, et al. (2019) In The Name of the Rose: A Roadmap for Rose Research in the Genome Era. Horticulture Research 6: 65.

Soufflet-Freslon, V., Marolleau, B., Thouroude, T., Chastellier, A., Pierre, S., Bellanger, M.N., Le Cam, B., Bonneau, C., Porcher, L., Leclere, A., Robert, F., Felix, F., Foucher, F. and Hibrand-Saint Oyant, L. (2019). Development of tools to study rose resistance to black spot. Acta Hortic. 1232, 213-220 https://doi.org/10.17660/ActaHortic.2019.1232.31

Stockinger, E. J., S. J. Gilmour, and M. F. Thomashow (1997) “Arabidopsis Thaliana CBF1 Encodes an AP2 Domain-Containing Transcriptional Activator That Binds to the C-Repeat/DRE, a Cis-Acting DNA Regulatory Element That Stimulates Transcription in Response to Low Temperature and Water Deficit.” Proceedings of the National Academy of Sciences of the United States of America 94: 1035–40.

Spiller M, Linde M, Hibrand-Saint Oyant L, Tsai C J, Byrne D H, Smulders M J M, Foucher F, Debener T (2011) Towards a Unified Genetic Map for Diploid Roses. Theoretical and Applied Genetics 122: 489–500. https://doi.org/10.1007/s00122-010-1463-x.

Svejda F (1979) Inheritance of Winterhardiness in Roses. Euphytica 28 (2): 309–14. https://doi.org/10.1007/BF00056587.

Taylor J, Butler D (2017) R Package ASMap: Efficient Genetic Linkage Map Construction and Diagnosis. Journal of Statistical Software 79. https://doi.org/10.18637/jss.v079.i06.

Teige, M., E. Scheikl, T. Eulgem, R. Dóczi, K. Ichimura, K. Shinozaki, J. L. Dangl, and H. Hirt. (2004). “The MKK2 Pathway Mediates Cold and Salt Stress Signaling in Arabidopsis.” Molecular Cell 15: 141–52. https://doi.org/10.1016/j.molcel.2004.06.023.

185

Terefe-Ayana D, Debener T (2011) An SSR from the Leucine-Rich Repeat Region of the Rose Rdr1 Gene Family Is a Useful Resistance Gene Analogue Marker for Roses and Other Rosaceae: An SSR from the Leucine-Rich Repeat Region. Plant Breeding 130:291–93. https://doi.org/10.1111/j.1439-0523.2010.01780.x.

Terefe-Ayana D, Kaufmann H, Linde M, Debener T (2012) Evolution of the Rdr1 TNL-Cluster in Roses and Other Rosaceous Species. BMC Genomics 13:409. https://doi.org/10.1186/1471- 2164-13-409.

Terefe-Ayana D, Yasmin A, Le T H, Kaufmann H, Biber A, Kühr A, Linde M, Debener T (2011) Mining Disease-Resistance Genes in Roses: Functional and Molecular Characterization of the Rdr1 Locus. Frontiers in Plant Science 2: 35. https://doi.org/10.3389/fpls.2011.00035.

Thalhammer, A., G. Bryant, R. Sulpice, and D. K. Hincha (2014) “Disordered Cold Regulated15 Proteins Protect Chloroplast Membranes during Freezing through Binding and Folding, But Do Not Stabilize Chloroplast Enzymes in Vivo.” Plant Physiology 166: 190–201.

Tomljenovic N, Pejić I (2018) Taxonomic Review of the Genus Rosa. Agriculturae Conspectus Scientificus 83: 139-147–147.

Untergasser A, Cutcutache I, Koressaar T, Ye J, Faircloth BC, Remm M, Rozen SG (2012) Primer3—New Capabilities and Interfaces. Nucleic Acids Research 40:115. https://doi.org/10.1093/nar/gks596.

Van Huylenbroeck J, Eeckhaut T, Leus L, Werlemark J, De Riek J (2007) Introduction of Wild Germplasm in Modern Roses. Acta Horticulturae, no. 751: 285–90. https://doi.org/10.17660/ActaHortic.2007.751.36.

Verde I, Bassil N, Scalabrin S, Gilmore B, Lawley C T, Gasic K, Micheletti D, et al. (2012) Development and Evaluation of a 9K SNP Array for Peach by Internationally Coordinated SNP Detection and Validation in Breeding Germplasm. PloS One 7: e35668. https://doi.org/10.1371/journal.pone.0035668.

Vineland Research and Innovation Centre 2013. The Innovation Report. Vineland Station, Ontario, Canada.Voorrips R E, Maliepaard C A (2012) The Simulation of Meiosis in Diploid and

186

Tetraploid Organisms Using Various Genetic Models. BMC Bioinformatics 13: 248. https://doi.org/10.1186/1471-2105-13-248.

Vogel, J. T., D. G. Zarka, H. A. Van Buskirk, S. G. Fowler, and M. F. Thomashow (2005) “Roles of the CBF2 and ZAT12 Transcription Factors in Configuring the Low Temperature Transcriptome of Arabidopsis.” Plant Journal 41: 195–211. https://doi.org/10.1111/j.1365- 313X.2004.02288.x.

Voorrips RE (2002) MapChart: Software for the Graphical Presentation of Linkage Maps and QTLs. Journal of Heredity 93:77–78. https://doi.org/10.1093/jhered/93.1.77.

Vukosavljev M (2014) Towards Marker Assisted Breeding in Garden Roses: From Marker Development to QTL Detection. Wageningen University. http://edepot.wur.nl/318534.

Vukosavljev M, Zhang J, Esselink GD, van ‘t Westende WPC, Cox P, Visser RGF, Arens P, Smulders MJM (2013) Genetic Diversity and Differentiation in Roses: A Garden Rose Perspective. Scientia Horticulturae 162:320–32. https://doi.org/10.1016/j.scienta.2013.08.015.

Vukosavljev M, Arens P, Voorrips R E, van ‘t Westende W P C, Esselink G D, Bourke P M, Cox P, et al. (2016) High-Density SNP-Based Genetic Maps for the Parents of an Outcrossed and a Selfed Tetraploid Garden Rose Cross, Inferred from Admixed Progeny Using the 68k Rose SNP Array. Horticulture Research 3: 16052. https://doi.org/10.1038/hortres.2016.52.

Waliczek T M, Byrne D H, Holeman D J (2015) Growers’ and Consumers’ Knowledge, Attitudes and Opinions Regarding Roses Available for Purchase.” Acta Horticulturae,1064: 235– 39. https://doi.org/10.17660/ActaHortic.2015.1064.26.

Walls, R. L., L. Cooper, J. Elser, M. Alejandra Gandolfo, C.J. Mungall, B. Smith, D.W. Stevenson, and P. Jaiswal (2019) “The Plant Ontology Facilitates Comparisons of Plant Development Stages Across Species.” Frontiers in Plant Science 10. https://doi.org/10.3389/fpls.2019.00631.

Weiser CJ (1970) Cold Resistance and Injury in Woody Plants. Science 169: 1269–78.

Werlemark G (2000) Evidence of Apomixis in Hemisexual Dogroses, Rosa Section Caninae. Sexual Plant Reproduction 12: 353–59. https://doi.org/10.1007/s004970000028.

187

Whitaker VM, Bradeen JM, Debener T, Biber A, Hokanson SC (2010) Rdr3, a Novel Locus Conferring Black Spot Disease Resistance in Tetraploid Rose: Genetic Analysis, LRR Profiling, and SCAR Marker Development. Theoretical and Applied Genetics 120:573–85. https://doi.org/10.1007/s00122-009-1177-0.

Whitaker VM, Debener T, Roberts AV, Hokanson SC (2010) A Standard Set of Host Differentials and Unified Nomenclature for an International Collection of Diplocarpon Rosae Races. Plant Pathology 59:745–52. https://doi.org/10.1111/j.1365-3059.2010.02281.x.

Whitaker V M, Zuzek K, Hokanson S C (2007) Resistance of 12 Rose Genotypes to 14 Isolates of Diplocarpon rosae Wolf (Rose Blackspot) Collected from Eastern North America.” Plant Breeding 126: 83–88. https://doi.org/10.1111/j.1439-0523.2007.01339.x.

Whitaker V M, Hokanson S C, Bradeen J (2007) Distribution of Rose Black Spot (Diplocarpon rosae) Genetic Diversity in Eastern North America Using Amplified Fragment Length Polymorphism and Implications for Resistance Screening. Journal of the American Society for Horticultural Science 132: 534–40. https://doi.org/10.21273/JASHS.132.4.534.

Wickham H. (2016). ggplot2: Elegant Graphics for Data Analysis. Springer-Verlag New York. ISBN 978-3-319-24277-4, https://ggplot2.tidyverse.org.

Willick, I. R., Tanino, K. K., Gusta, L. V. (2021) “The impact of global climate change on the freezing tolerance of winter cereals in Western Canada.” Journal of Agronomy and Crop Science 201(1): 88-99.

Willick, I.R., Gusta, L.V., Fowler, B.D., Tanino, K.K. (2019) “Ice segregation in the crown of winter cereals: Evidence for extraorgan and extratissue freezing.” Plant, Cell & Environment 42(2): 701-716.

Wisniewski M, Fuller M, Palta J, Carter J, Arora R (2004) Ice Nucleation, Propagation, and Deep Supercooling in Woody Plants. Journal of Crop Improvement 10: 5–16. https://doi.org/10.1300/J411v10n01_02.

Wisniewski M, Bassett C, Gusta L V (2003) An Overview of Cold Hardiness in Woody Plants: Seeing the Forest Through the Trees. HortScience 38: 952–59. https://doi.org/10.21273/HORTSCI.38.5.952.

188

Wisniewski M, Gusta L V, Neuner G (2014) Adaptive Mechanisms of Freeze Avoidance in Plants: A Brief Update. Environmental and Experimental Botany 99: 133–40. https://doi.org/10.1016/j.envexpbot.2013.11.011.

Wisniewski M, Nassuth A, Arora R (2018) Cold Hardiness in Trees: A Mini-Review. Frontiers in Plant Science 9. https://doi.org/10.3389/fpls.2018.01394.

Wisniewski M, Nassuth A, Teulières C, Marque C, Rowland J, Cao P B, Brown A (2014) Genomics of Cold Hardiness in Woody Plants. Critical Reviews in Plant Sciences 33: 92–124. https://doi.org/10.1080/07352689.2014.870408.

Wisniewski M, Norelli J, Bassett C, Artlip T, Macarisin D (2011) Ectopic Expression of a Novel Peach (prunus Persica) CBF Transcription Factor in Apple (malus × Domestica) Results in Short-Day Induced Dormancy and Increased Cold Hardiness.” Planta 233: 971–83. https://doi.org/10.1007/s00425-011-1358-3.

Wisniewski M, Webb R, Balsamo R, Close T J, Yu X M, Griffith M (1999) Purification, Immunolocalization, Cryoprotective, and Antifreeze Activity of PCA60: A Dehydrin from Peach (prunus Persica). Physiologia Plantarum 105: 600–608. https://doi.org/10.1034/j.1399- 3054.1999.105402.x.

Wissemann V, Ritz C M (2005) The Genus Rosa (Rosoideae, Rosaceae) Revisited: Molecular Analysis of NrITS-1 and AtpB-RbcL Intergenic Spacer (IGS) versus Conventional Taxonomy. Botanical Journal of the Linnean Society 147: 275–90. https://doi.org/10.1111/j.1095- 8339.2005.00368.x.

Wu Y, Bhat PR, Close TJ, Lonardi S, (2008) Efficient and accurate construction of genetic linkage maps from the minimum spanning tree of a graph. PLoS Genetics https://doi.org/10.1371/journal.pgen.1000212.

Xin, Z., and J. Browse (2000) “Cold Comfort Farm: The Acclimation of Plants to Freezing Temperatures.” Plant, Cell & Environment 23: 893–902. https://doi.org/10.1046/j.1365- 3040.2000.00611.x.

189

Xue A G, Davidson C G (1998) Components of Partial Resistance to Black Spot Disease (Diplocarpon rosae Wolf) in Garden Roses. HortScience 33: 96–99. https://doi.org/10.21273/HORTSCI.33.1.96.

Yagi M (2013) Application of DNA Markers for Breeding Carnations Resistant to Bacterial Wilt. Japan Agricultural Research Quarterly 47:29–35. https://doi.org/10.6090/jarq.47.29.

Yagi M (2015) Recent Progress in Genomic Analysis of Ornamental Plants, with a Focus on Carnation. The Horticulture Journal 84: 3–13. https://doi.org/10.2503/hortj.MI-IR01.

Yagi M (2018) Recent Progress in Whole Genome Sequencing, High-Density Linkage Maps, and Genomic Databases of Ornamental Plants. Breeding Science 68: 62–70. https://doi.org/10.1270/jsbbs.17080.

Yan, M., Byrne, D.H., Klein, P.E., van de Weg, W.E., Yang, J. and Cai, L. (2019). Black spot partial resistance in diploid roses: QTL discovery and linkage map creation. Acta Hortic. 1232, 135-142 https://doi.org/10.17660/ActaHortic.2019.1232.21

Yan Z, Denneboom C, Hattendorf A, Dolstra O, Debener T, Stam P, Visser P B (2005) Construction of an Integrated Map of Rose with AFLP, SSR, PK, RGA, RFLP, SCAR and Morphological Markers. Theoretical and Applied Genetics 110: 766–77. https://doi.org/10.1007/s00122-004-1903-6.

Yasin, N A, Ahmed S (2016) Induction of Defence-Related Biochemicals by Rhizosphere Bacteria against Black Spot Disease of Rose. Biological Agriculture & Horticulture 32: 34–46. https://doi.org/10.1080/01448765.2015.1017737.

Yokoya, K, Kandasamy KI, Walker S, Mandegaran Z, Roberts AV (2000) Resistance of Roses to Pathotypes of Diplocarpon Rosae. Annals of Applied Biology 136 :15–20. https://doi.org/10.1111/j.1744-7348.2000.tb00003.x.

Yokoya K, Roberts A V, Mottley J, Lewis R, Brandham P E (2000) Nuclear DNA Amounts in Roses. Annals of Botany 85: 557–61. https://doi.org/10.1006/anbo.1999.1102.

Zhao K, Zhou Y, Li Y, Zhuo X, Ahmad S, Han Y, Yong X, Zhang Q (2018) Crosstalk of PmCBFs and PmDAMs Based on the Changes of Phytohormones under Seasonal Cold Stress in

190

the Stem of Prunus Mume.” International Journal of Molecular Sciences 19: 15. https://doi.org/10.3390/ijms19020015.

Zlesak, D. C., D. Ballantyne, M. Holen, A. Clark, S. C. Hokanson, K. Smith, J. D. Zurn, N. V. Bassil, and J. M. Bradeen (2020) “An Updated Host Differential Due to Two Novel Races of Diplocarpon Rosae Wolf, the Causal Agent of Rose Black Spot Disease.” HortScience 1: 1–3. https://doi.org/10.21273/HORTSCI14902-20.

Zhu Z M, Gao X F, Fougère-Danezan M (2015) Phylogeny of Rosa Sections Chinenses and Synstylae (Rosaceae) Based on Chloroplast and Nuclear Markers. Molecular Phylogenetics and Evolution 87: 50–64. https://doi.org/10.1016/j.ympev.2015.03.014.

Zlesak D C (2006) Rose.” In Flower Breeding and Genetics: Issues, Challenges and Opportunities for the 21st Century, edited by Neil O. Anderson, 695–740. Dordrecht: Springer Netherlands. https://doi.org/10.1007/978-1-4020-4428-1_26.

Zlesak D C, Harp D A, Zuzek K, Sloan J J, Owings A, George S W (2015) Earth-Kind® Rose Trialing: An International Model for the Identification of Regionally-Adapted Landscape Roses. VI International Symposium on Rose Research and Cultivation, 123–29. https://doi.org/10.17660/ActaHortic.2015.1064.15.

Zlesak, D.C., R. Nelson, D. Harp, B. Villarreal, N. Howell, J. Griffin, G. Hammond, and S. George (2017) “Performance of Landscape Roses Grown with Minimal Input in the North- Central, Central, and South-Central United States.” HortTechnology 27: 718–30. https://doi.org/10.21273/HORTTECH03681-17.

Zlesak D C, Thill C A, Anderson N O (2005) Trifluralin-Mediated Polyploidization of Rosa chinensis Minima (Sims) Voss Seedlings. Euphytica 141: 281–90. https://doi.org/10.1007/s10681-005-7512-x.

Zlesak D C, Zuzek K, Melcher M, Hokanson S C, Van Schaick J, Varghese A, Carlson C, Ghavami F (2019) SNP Markers Linked to Sex in Dioecious Rosa setigera Michx. Acta Horticulturae 1232: 33–39. https://doi.org/10.17660/ActaHortic.2019.1232.6.

191

Zurn J D, Zlesak D C, Holen M, Bradeen J M, Hokanson S C, Bassil N V (2018) Mapping a Novel Black Spot Resistance Locus in the Climbing Rose Brite EyesTM (‘RADbrite’). Frontiers in Plant Science 9. https://doi.org/10.3389/fpls.2018.01730.

Zurn, J D, Zlesak D C, Holen M, Bradeen J M, Hokanson S C, Bassil N V (2020) Mapping the Black Spot Resistance Locus Rdr3 in the Shrub Rose ‘George Vancouver’ Allows for the Development of Improved Diagnostic Markers for DNA-Informed Breeding. Theoretical and Applied Genetics https://doi.org/10.1007/s00122-020-03574-4.

Zych K, Gort G, Maliepaard C A, Jansen R C, Voorrips R E (2019) FitTetra 2.0 – Improved Genotype Calling for Tetraploids with Multiple Population and Parental Data Support. BMC Bioinformatics 20: 148. https://doi.org/10.1186/s12859-019-2703-y.

192

APPENDICES

APPENDIX 1: Supplementary Materials and Methods for Chapter 2: Identification of a polymorphism within the Rosa multiflora muRdr1A gene linked to resistance to multiple races of

Diplocarpon rosae w. in tetraploid garden roses (rosa x hybrida)

A1.1 Race characterization and race diversity at VRIC experimental farm

Investigation of the diversity of D. rosae and race characterization using detached leaf assay

(DLA) was conducted for a large number of isolates collected from infected roses grown in

Vineland, ON. A total of 18 isolates were divided into three prevalent race groups based on their infection patterns on the differential hosts (Whitaker et al. 2010) (data not shown). Isolates VHy-

12.4, VSKO4 and VOTB17-1 represented each of the prevalent groups of races and were chosen for mapping BS resistance (Supplementary Table 2-1 – Appendix 2).

A1.2 DNA isolation and sequencing of Genotyping-by-Sequencing (GBS) libraries

Leaf tissue was collected from the mapping population and the parents, freeze dried and ground to a powder. The DNA was extracted with Qiagen DNeasy kit (Qiagen Sciences) using the buffer EB to elute the DNA instead of Buffer AE as recommended in the published protocol during the final steps, in order to preserve the DNA for Genotyping-by-Sequencing (GBS) library preparation. DNA was quantified using Quant-iT™ PicoGreen® (Thermo Scientific) and normalized to 10ng/ μL. DNA quality was evaluated for each sample by running 30ng of DNA on a 1% agarose gel stained with GelRed® nucleic acid gel stain (Biotium) for about 45 minutes at 100 volts. The DNA was also examined for the presence of inhibitors by digesting 300ng of

DNA for 32 samples (8 per plate) using 1 μL of HindIII (New England BioLabs). The construction of the GBS libraries was based on the published protocol from Elshire et al. (2011)

193

with minor revisions based on the Rieseberg lab recommendations (Elshire et al. 2011)

(Rieseberg Lab Production Protocol, 2013). Two 96 well plates of library prep DNA were run on an Illumina MiSeq to check for consistency of sequencing data before sending the GBS libraries for sequencing at a deeper coverage on an Illumina HiSeq 2500 using V4 chemistry for 101 cycles.

A1.3 Analysis of muRdr1A sequence

muRdr1A specific genotyping was done by PCR amplification as described and was performed in a high magnesium mix with genomic DNA diluted 20 times in assays containing

1.65 μL sterile water, 1X ABM Buffer, 0.4 mM dNTPs, 0.25 μL Evagreen, 0.25 U ABM Taq

DNA polymerase, 1 mM MgSO4 and 0.2 μM of each forward and reverse PCR primers. The following PCR programme was used: initial denaturation for 3 min and 30 s at 94⁰C, then 33 cycles of 30 s at 94⁰C, 30 s at an annealing temperature of 59⁰C and 30 s at 72⁰C. The two diagnostic markers were validated on the mapping population and the backcross population using high-resolution melting analysis (HRMA) (Smith et al. 2010). Alignments were visualized using

IGV (Robinson et al. 2011).

A1.4 Statistical Analysis

A mixed effects model approach was used on the detached leaf assay (DLA) split-plot design using the SAS software (SAS ver. 9.4, SAS Institute Inc., Cary, NC, USA). Replication was set as the factor Block, Isolate was the main factor and Genotype was allocated to sub-plot.

The disease scores for two Petri dishes per genotype and per replication were used. The disease score was an ordinal measure with six categories and was defined as a non-Gaussian multinomial

194

dependent variable. The cumulative logit function was used as the linear predictor in a GLMM analysis. The variance of the disease score was partitioned into fixed effects (Isolate, Genotype and Isolate * Genotype) and random effects (Block and Block * Isolate). While the individual rose leaf was the experimental unit, the interaction between the Petri dish or individual leaf and the genotype was set as random. A type I error of 0.05 was used to determine the significance of tests in this analysis. Variance analysis was performed using the GLIMMIX procedure with

Laplace interval estimation. Type III tests of fixed effects were used in the variance analysis and random effects were tested with a likelihood test for covariance parameters.

195

APPENDIX 2: Standard host differential infection patterns to characterize races of D. rosae and

list of isolates examined

Supplementary Table 2-1. A) Standard host differential infection patterns to characterize races of D. rosae and B) List of isolates examined, host source and determination of three prevalent D. rosae races based on differential infection patterns. A. Differential Infection Pattern Differential Genotypea Race 10 Unidentified Race 7 cv. Mermaid + + + Honeybee™ + + + Sexy Rexy® + + + cv. Surrey + + + Love and Peace™ + + + cv. George Vancouver + + - Knock out® + + + Baby Love™ - + - cv. Mrs Doreen Pike - - - CA60 - - - cv. Singing In The Rain + + + B. Isolate host source and Race Isolate Host Source Race VHy-12.4b (21ARFROI x Cadenza) x 93 (Basye’s Legacy x Frontenac) 10 VCA29171 CA-29 (MyHero x Frontenac) 10 VFFT171 Floral Fairy Tale 10 VGV171 George Vancouver 10 VHy4171 CA33 x (Prairie Joy x Frontenac) 10 VCA33171 CA-33 (Morden Sunrise x Golden Celebration) 10 VMP3 Melody Parfumée™ 10 VH14171 93 x CA29 10

VSKO4 b Sunny Knock Out® Undescribed VSNKO171 Sunny Knock Out® Undescribed VHVY7171 (21ZZBL11 x Yellow Submarine) x Oscar Peterson Undescribed V33171 ‘33’ ((75252-1 x Morden Sunrise) x Baby Love) Undescribed VHy5171 33 x (Yellow Submarine x PO4) Undescribed VBL25 Baby Love™ Undescribed VBL26 Baby Love™ Undescribed

VOTB17-1 b Out of The Blue 7 Vy1117-1 CA28 x Oscar Peterson 7 VJC17-1 John Cabot 7 aCA60 and Singing In The Rain are mapping parents bD. rosae isolates used in genetic mapping

196

APPENDIX 3: QTL associated with resistance to four Diplocarpon rosae isolates and natural

field infection.

Supplementary Table 2-2. Summary of QTL associated with resistance to four D. rosae isolates and natural field infection. Source of # of Mapping Map Closest Marker Infection Individuals Position Marker Interval VHy-12.4 110 LG1.H3-15cM SChr1_67051139 SChr1_64649660- SChr1_65811939 VSKO4 96 LG1.H3-15cM SChr1_67051139 SChr1_64649660 - SChr1_65811939 VOTB17-1 101 LG1.H3-35cM SChr1_66170115 SChr1_67169791 - SChr1_66169890 BOO5 97 LG1.H3-15cM SChr1_67051139 SChr1_64528752 - SChr1_65811939 Field 99 LG1.H3-15cM SChr1_67051139 SChr1_64819130 - SChr1_65811939

197

APPENDIX 4: Disease rating scale used to screen rose hybrids for black spot disease (Diplocarpon rosae)

resistance

Supplementary Figure 2-1. Photographs of detached rose leaves and lesions from black spot infection 14 days after inoculations, representative of the disease rating scale used in this study to evaluate the range of response in detached leaf assay. Pictures of lesions were obtained under Stereomicroscope (20X)

198

APPENDIX 5: Genetic map generated for the parental genotype ‘SITR’ in the context of the black spot research project

Supplementary Figure 2-2. The high density SNP based genetic map of rose. Genetic length and marker distribution along the 7 rose chromosomes for 31 linkage groups of the female parent of the cross ‘CA60’ x ‘SITR’. The name of each linkage groups is indicated above the linkage groups. Each horizontal bar represents a simplex SNP marker. Chromosomal fragments are observed for homologous groups 2, 5 and 6.

199

APPENDIX 6. Supplementary Materials and Methods for Chapter 3: Mapping cold hardiness in

garden roses (Rosa x hybrida)

A6.1 Data Analysis of electrolyte leakage data

Statistical analysis was conducted separately for the parental lines (‘CA60’, ‘SITR’,

‘EDI’ and ‘GV’) and the two mapping populations ‘CA60’ x ‘SITR’ and ‘EDI’ x ‘GV’. General linear mixed models (GLMM) were fitted on the EL and I data collected in a split-plot design using the SAS software (SAS ver. 9.4, SAS Institute Inc., Cary, NC, USA). Replication was set as the factor Block, Temperature was the main-plot factor and Genotype was allocated to subplot. EL and I data corresponded to a proportion measure and was defined as a non-Gaussian dependent variable with a beta distribution. The logit function was used as the link function in the GLMM analysis.

A6.1.1 Parental lines

The variance of EL and I was partitioned into fixed effects (Temperature and Genotype,

Temperature * Genotype) and random effects (Rep and Rep* Temperature and

Rep*Temperature * Genotype). Thus, the linear predictor was defined as ƞij = ƞ + bi + αj +(ba)ij

+ τk + (bt)ik + (ατ)jk + (bat)ijk, where ƞ is the intercept, bi is the effect of block, αj the effect of main treatment temperature, (ba)ij is the random effect of the whole-plot units involving factor temperature, τk the effect of the genotype, (bt)ik the random effect involving genotype, (ατ)jk the interaction effect between genotype and temperature and (bat)ijk the error associated with the interaction of the effects at the whole-plot unit. A type I error of 0.05 was used to determine the significance of tests in this analysis. Variance analysis was performed using the GLIMMIX procedure with Laplace interval estimation. Type III tests of fixed effects were used in the

200

variance analysis and random effects were tested with a likelihood test for covariance parameters. The fit of the models was investigated based on an analysis of the studentized conditional residuals. BLUE estimates of EL were computed for the parental genotypes ‘CA60’,

‘SITR’, ‘EDI’ and ‘GV’.

A6.1.2 ‘CA60’ x ‘SITR’ Population

The variance of EL was partitioned into fixed effects (Temperature) and random effects

(Rep and Rep* Temperature, Genotype, Temperature * Genotype and Rep*Temperature *

Genotype). Thus, the linear predictor was defined as ƞij = ƞ + bi + αj +(ba)ij + τk + (bt)ik + (ατ)jk +

(bat)ijk, where ƞ is the intercept, bi is the effect of block, αj the effect of main treatment temperature, (ba)ij is the random effect of the whole-plot units involving factor temperature, τk the effect of the genotype, (bt)ik the random effect involving genotype, (ατ)jk the interaction effect between genotype and temperature and (bat)ijk the error associated with the interaction of the effects at the whole-plot unit. BLUP estimates of EL were computed for each temperature.

A6.1.3 ‘EDI’ x ‘GV’ Population

The variance of EL and I was partitioned into fixed effects (Temperature) and random effects (Rep and Rep* Temperature, Genotype, Temperature * Genotype). The random effect

Rep*Temperature * Genotype was not included in this model because only one technical replication was available per biological replication of the EL assay. Thus, the linear predictors were defined as ƞij = ƞ + bi + αj +(ba)ij + τk + (bt)ik + (ατ)jk, where ƞ is the intercept, bi is the effect of block, αj the effect of main treatment temperature, (ba)ij is the random effect of the whole-plot units involving factor temperature, τk the effect of the genotype, (bt)ik the random effect involving genotype, (ατ)jk the interaction effect between genotype and temperature. BLUP

201

estimates of EL and I were computed for ‘EDI’ x ‘GV’. BLUP estimates of EL were computed for each temperature.

202

APPENDIX 7. Evidence for the CBF/DREB and ICE1 transcription factors in the reference genome R. chinensis

Supplementary Table 3-1 Evidence for the CBF/DREB and ICE1- transcription factors in the reference genome R. chinensis Old Blush homozygous v2.0. Gene name Location Description

RcHm_v2.0_Chr1g0376641 RcHm_v2.0_Chr1:63847494..63848870 CBF4/DREB1D

RcHm_v2.0_Chr3g0472361 RcHm_v2.0_Chr3:18244191..18245644 CBF1/DREB1B

RcHm_v2.0_Chr7g0199331 RcHm_v2.0_Chr7:17371406..17372131 CBF4/ DREB1D

RcHm_v2.0_Chr7g0199381 RcHm_v2.0_Chr7:17422407..17424183 CBF3/DREB1A

RcHm_v2.0_Chr2g0176421 RcHm_v2.0_Chr2:88244910..88247805 ICE1

RcHm_v2.0_Chr7g0188921 RcHm_v2.0_Chr7:8295405..8298561 ICE1

203

APPENDIX 8. Field and electrolyte leakage data for 19 commercial cultivars and eight rose

selections used in Experiment 1

Supplementary Table 3-2. Field and electrolyte leakage data for 19 commercial cultivars and eight rose selections, which aimed to determine the correlations between electrolyte leakage (EL) as an index of freezing tolerance and winter hardiness recorded in the field at different locations (VicFarm Ontario, Saskatoon (SK) and Old Alberta (OA)) on a scale from 1 to 5 and estimated as USDA cold hardiness zone. The LT50 was estimated for each genotype using a logit model approach. Genotype Field data Freezing experiments in artificial conditions Cultivars USDA Field Number of EL% - EL% EL% EL% EL% LT50 zone WH replications 20ºC -15ºC -10ºC -5ºC Control (ON) ‘Cardinal 4 3 86.9 82.4 47.9 20.3 14.7 -11 Song’ ‘Caroline 7 3 3 83.5 84 41.4 20 14.3 -11 de Monaco’ ‘Desert 7 3 1 75.2 71.4 20.3 16.2 13 -14 Peace’ ‘Gentle 6 4 1 94.5 91.8 27.7 14.1 13.6 -12 Giant’ ‘George 3 1 1 72.3 63.3 20.2 25.7 19.9 -20 Vancouver’ ‘John 3 0.33 1 63.2 59.7 38.6 12.6 15.9 -15 Cabot’ ‘John 2 0 1 67.2 42.9 21.7 25 13.7 -17 Davis’ ‘John 3 3 77.9 54.8 22.8 24.1 14.8 -15 Franklin’ Knock 4 1 89.6 85.2 15.2 11.9 17.8 Out® ‘Lambert 3 2 3 80.1 75.4 39.4 14.1 11.3 -12 Closse’ ‘Peace’ 5 3 3 80.4 74.5 36.9 15.9 15.4 -13 ‘Poseidon’ 6 2 77 54.2 21.6 23.5 15.2 -16

204

‘Quadra’ 3 0 3 72.3 59.2 28.7 18.4 16.6 -15 ‘Salmon 5 2 76.6 81 53.1 15.8 16.9 -10 Vigorosa’ ‘Singing in 6 4 1 86.4 88.7 82.6 19.2 18.9 -11 the Rain’ ‘William 2 0 3 61.8 43.6 24.6 16.2 13.8 -18 Baffin’ ‘Yellow 4 3 2 95.3 89.9 43.2 17.7 12.9 -10 Submarine’ Control 0.67 3 78 76.2 27.4 13.9 11.9 -13 ‘CA60’ Field Field Number of EL% EL% EL% EL% EL% LT50 WH WH OA replications - -15ºC -10ºC -5ºC Control SK 20ºC Selections ‘S13-10’ 0.2 0 2 58.9 43.9 22.3 18.2 16.9 -19 ‘S13-29’ 0.8 2 2 54.9 36.3 21.4 17.1 14.6 -22 ‘S13-3’ 4.2 3 1 78.3 58.4 15.1 21.8 22.9 -15 ‘S13-32’ 0.2 0 1 64.5 43.2 25.8 18.0 16.2 -18 ‘S13-35’ 3.4 3 2 73.3 56.9 24.7 19.0 14.7 -15 ‘S13-6’ 3.2 5 1 72.2 79.7 19.0 24.1 14.3 -14 ‘S13-7’ 3.6 2 2 70.1 51.9 29.0 19.7 14.7 -16 ‘S13-8’ 2 2 1 77 58.1 25.8 16.4 21.2 -16

205

APPENDIX 9. Comparison of magnitude of electrolyte leakage between the four replications of

Experiment 2

Supplementary Figure 3-1. Comparison of magnitude of electrolyte leakage (EL) between the four replications (Rep) of electrolyte leakage assays conducted on the hardy parental lines ‘CA60’ and ‘GV’. The electrolyte leakage did not increase over 60% in the third replication. Correlations were also computed using Pearson method. Significant correlations are given, with ** significant at p=0.001.

206

APPENDIX 10. Range of LT50s in the 'EDI'x’GV' population

Supplementary Table 3-3. LT50 estimated from non-linear dosage response curve for the progeny of the 'EDI'x’GV' population .Genotype LT50 Genotype LT50 Genotype LT50 Genotype LT50 Above average freezing tolerance Average freezing tolerance Below average freezing tolerance EDI.GV.60 -28 EDI.GV.13 -21 EDI.GV.28 -19 EDI.GV.104 -17 EDI.GV.10 -26 EDI.GV.43 -21 EDI.GV.36 -19 EDI.GV.21 -17 EDI.GV.24 -25 EDI.GV.47 -21 EDI.GV.41 -19 EDI.GV.3 -17 EDI.GV.37 -25 EDI.GV.79 -21 EDI.GV.44 -19 EDI.GV.35 -17 EDI.GV.39 -25 EDI.GV.90 -21 EDI.GV.51 -19 EDI.GV.68 -17 EDI.GV.42 -25 EDI.GV.93 -21 EDI.GV.53 -19 EDI.GV.88 -17 EDI.GV.5 -25 EDI.GV.106 -20 EDI.GV.7 -19 EDI.GV.22 -16 EDI.GV.8 -25 EDI.GV.15 -20 EDI.GV.72 -19 EDI.GV.33 -16 EDI.GV.17 -24 EDI.GV.34 -20 EDI.GV.73 -19 EDI.GV.54 -16 EDI.GV20 -24 EDI.GV.4 -20 EDI.GV.74 -19 EDI.GV.56 -16 EDI.GV.38 -24 EDI.GV.50 -20 EDI.GV.81 -19 EDI.GV.58 -16 EDI.GV.27 -23 EDI.GV.52 -20 EDI.GV.91 -19 EDI.GV.66 -16 EDI.GV.40 -23 EDI.GV.61 -20 EDI.GV.96 -19 EDI.GV.70 -16 EDI.GV.45 -23 EDI.GV.63 -20 EDI.GV.25 -18 EDI.GV.76 -16 EDI.GV.48 -23 EDI.GV.67 -20 EDI.GV.31 -18 EDI.GV.86 -16 EDI.GV.11 -22 EDI.GV.75 -20 EDI.GV.46 -18 EDI.GV.105 -15 EDI.GV.14 -22 EDI.GV.78 -20 EDI.GV.55 -18 EDI.GV.49 -15 EDI.GV.16 -22 EDI.GV.85 -20 EDI.GV.65 -18 EDI.GV.57 -15 EDI.GV.30 -22 EDI.GV.100 -19 EDI.GV.71 -18 EDI.GV.23 -15 EDI.GV.32 -22 EDI.GV.12 -19 EDI.GV.9 -18 EDI.GV.82 -15 EDI.GV.92 -22 EDI.GV.26 -19 EDI.GV.89 -15

207

APPENDIX 11. Key genetic factors involved in the acquisition of cold hardiness within the CBF pathway

Supplementary Table 3-4. List of transcription factors and genes known to be involved in the acquisition of cold hardiness and CBF- pathway in Arabidopsis thaliana. The list is not exhaustive, but can serve as a framework when exploring the molecular basis of freezing tolerance in roses. Search terms Description Reference

DREB/CBF Key regulator of cold hardiness (Gilmour et al. 1998; Medina et al. 1999; Stockinger, et al. 1997; Novillo et al. 2007)

ICE1 Inducer of CBF expression 1; activates DREB1/CBF2 with (Chinnusamy et al. 2003) MYB15

CAMTA3 Calmodulin binding transcription activator of CBF2 (Doherty et al. 2009)

LHY Late Elongated Hypocotyl; activates DREB1C/CBF2 with CCA1 in (Dong et al. 2017) response to the circadian rhythm

CCA1 Circadian Clock Associated 1, activates DREB1C/CBF2 in (Dong et al. 2017) response to the circadian rhythm

MPK Protein kinase; MPK cascade (MPK4) activates ICE while MPK3/6 (Teige et al. 2004; Li et al. 2017) inhibits it by phosphorylation. The cascade MEKK1-MKK2- MPK4/6 transduce cold stress signal

ZAT Zing finger protein implicated in the regulation of DREB/CBF (Vogel et al. 2005) expression

HOS1 Ring finger HOS1 is known to regulate the activity of ICE1 (C.-H. Dong et al. 2006)

BZR Upregulate the activity of CBF alongside CESTA

CESTA Upregulate the activity of CBF alongside BZR

COR15 Cold-induced LEA protein (COR-protein); chloroplast proteins (Lin and Thomashow 1992; Thalhammer et al. 2014) necessary for full cold acclimation in Arabidopsis. They protect

208

cell membranes, as indicated by electrolyte leakage and chlorophyll fluorescence measurements

COR47 Cold-induced LEA protein (COR-protein); with cryoprotective (Guo et al. 1992; Puhakainen et al. 2004; Bozovic et al. activity; protects the membrane during freezing events 2013)

LTI29 Low-Temperature Induced protein; cryoprotective activity on the (Puhakainen et al. 2004; Bozovic et al. 2013) membrane

LTI30 Low-Temperature Induced protein involved in plant freezing stress (Chung and Parish 2008; Shi et al. 2015) resistance; cryoprotective activity on the membrane

RD29A Cold-regulated genes (Mantyla, Lang, and Palva 1995)

KIN1/KIN2 Protein kinase (Dong et al. 2006) * TFs = Transcription factor

209