AN ABSTRACT OF THE DISSERTATION OF

Transcription

1

2 AN ABSTRACT OF THE DISSERTATION OF Dustin Walker Herb for the degree of Doctor of Philosophy in Crop Science presented on March 9, Title: Developing Flavorful and Sustainable Barley (Hordeum vulgare L.): Integrating Malting Quality and Barley Contributions to Beer Flavor within the Framework of Facultative Growth Habit. Abstract approved: Patrick M. Hayes This dissertation consists of a general introduction, three manuscripts, and general conclusions. The research integrates research on (1) the effects of barley genetics and production environment on the contribution of barley to beer flavor, (2) the effects of degree of malt modification on barley contributions to beer flavor and (3) a genetic analysis of low temperature tolerance (LTT) and vernalization (VRN) sensitivity. The first manuscript reports the relative impacts of barley genotype and environment on sensory descriptors of beers brewed from a subsample of a bi-parental mapping population grown at three locations. The second manuscript addresses the effects of genotype and degree of malt modification on beer flavor based on two experiments involving: a) length of grain storage prior to malting using samples from one of the environments utilized in the first experiment and b) alterations of malting protocol to produce three levels of malt modification in two varieties. The third manuscript describes the results from association mapping of LTT and VRN sensitivity in a large sample of diverse barley accessions that were extensively phenotyped and genotyped. The results of the analysis are applied in the context of facultative growth habit a tool for dealing with the challenges of climate change on barley production. These manuscripts comprise a roadmap for integrating sensory science with contemporary breeding methods for the development of value-added barley varieties.

3 Copyright by Dustin Walker Herb March 9, 2017 All Rights Reserved

4 Developing Flavorful and Sustainable Barley (Hordeum vulgare L.): Integrating Malting Quality and Barley Contributions to Beer Flavor within the Framework of Facultative Growth Habit by Dustin Walker Herb A DISSERTATION submitted to Oregon State University in partial fulfillment of the requirements for the degree of Doctor of Philosophy Presented March 9, 2017 Commencement June 2017

5 Doctor of Philosophy dissertation of Dustin Walker Herb presented on March 9, 2017 APPROVED: Major Professor, representing Crop Science Head of the Department of Crop and Soil Science Dean of the Graduate School I understand that my dissertation will become part of the permanent collection of Oregon State University libraries. My signature below authorizes release of my dissertation to any reader upon request. Dustin Walker Herb, Author

6 ACKNOWLEDGEMENTS I would like to express sincere appreciation to my advisor, Dr. Patrick Hayes you have guided and supported me from the time that I was an ambitious undergraduate in your plant genetics class, to the pessimistic graduate student in your lab. Throughout my education, you ve conducted yourself with integrity, and have stood in solidarity with myself and with others, while consistently providing valuable insight into plant breeding and genetics, as well as life. It is my honor to call you a dear friend. A special thanks to the members of my committee Dr. Alfonso Cuesta-Marcos for our morning greetings and coffee breaks, and for guiding me through the LTT project; Dr. Jeff Leonard for encouraging me to continue on to graduate school and insisting that someone else pay for it, and for the crash course in quantitative genetics as a student worker in your lab; Dr. Andrew Ross for every joke you ve ever told me, and for your patience with me and my frequent questions; Dr. Thomas Shellhammer for not recording my mad karaoke skills, and for introducing me to the joys and complexities of brewing science; and my Graduate Council Representative, Dr. Glenn Howe, for always checking in on my progress. Your willingness to serve on my committee, and your unfailing encouragement of me during my examinations has made me a stronger scientist and individual, thank you! The research in this dissertation would not have been possible without the diligence and pure devotion of the Barleyworld staff. I would like to express my gratitude to Scott Fisk for the many beers and combine races in the name of science; Laura Helgerson for presenting me with the broken tweezer award, and your willingness to always help out; Tanya Filichkin for our exchanges of rock concert stories and for teaching me all there is to know about doubled haploid production; and my fellow barley graduate students, Brigid Meints for partnering with me on the Oregon Promise; Ryan Graebner for the R code training; Dr. Araby Belcher for the GWAS help; and Javier Hernandez for your sunny disposition. A special thanks to Mike

7 Adams, my officemate who made each work day another adventure I ve never laughed harder! It is my privilege to consider all of you friends - next rounds on me! I have had the pleasure of collaborating with a wide-range of scientists, maltsters, and brewers from around the world, to name a few Dr. William Thomas, Dr. Igancio Romagosa, Dr. Kevin Smith, Dr. Shawn Townsend, Dr. Jennifer Kling, Dr. Patricia Aron, Dr. Dominique Vequaud, Dr. Jarislav von Zitzewitz, Jeff Clawson, Rebecca Jennings, Rebecca Newman, Bobby Monsour, Sean Tynan, Juan and Miguel Medina, John Andrews, Krisiti Vinkemier, Tom Nielson, Seth Klann, Aaron MacLeod, Andrea and Chris Stanley, Dan Carey, and many more. This community makes the barley industry so great! Finally, this dissertation would not have been possible without the support of my family and friends my loving wife, Danielle Herb, who has put up with so much for so long, my parents Matt and Rachel Herb, who have given up much on my behalf, my brother Brandon Herb and his fiancée Delani O Connell, my in-laws, Loretta, Jim, and Katie Austin, and my grandparents for their continued support. I would be remiss if I did not acknowledge the support of my closest friends Brandon & Elli Sapp, Ben & Rachel Stackhouse, Jordan & Kasia Whitaker, Rodney Allen, Troy Vanloovan, John Ryan Mullenix, Justin Jackola, Wayne Stearns, Thomas Evans, and Gatlen Richardson. Special thanks to Donald Cowgill, Tom Davis, the Korner Kitchen Crew, and the Tennessee Bottom Boys. Thank you all for your support of me throughout these past three years, cheers!

8 CONTRIBUTION OF AUTHORS All authors played a role in reviewing the manuscript of their respective chapters. Dustin Herb was primarily responsible for all data collection, analysis, interpretation, and writing of the manuscripts. Dr. Patrick Hayes coordinated the development and collection of the germplasm used for these studies and designed the phenotypic and genotypic experiments implemented. Chapter 2 and 3: Since both of these manuscripts stemmed from the Oregon Promise population, many of the authors had key roles for both studies. Dr. Luis Cistue created some of the doubled haploid lines that compose the Oregon Promise population. Laura Helgerson and Tanya Filichkin also creasted doubled haploids that made up the Oregon Promise population and propagated, maintained, and increased seed of all doubled haploids in the greenhouse. Scott Fisk coordinated the planting, maintenance, harvesting, and processing for all the Oregon Promise trials. Brigid Meints assisted in agronomic evaluations of the Oregon Promise population and created the original linkage map. Chis Martens micro-malted the 2014 Oregon Promise grain samples. Rebecca Jennings, Robert Monsour, Sean Tynan, and Kristi Vinkemeier micro-malted, analyzed malt quality, nano-brewed, and performed sensory assessments of the 2015 Oregon Promise grain samples at Rahr Malting Co. Dr. Yueshu Li, Andrew Nuygen, and Aaron Onio pilot-malted, analyzed malt quality, pilot-brewed, and performed sensory assessments on the 2015 CDC Copeland and Full Pint at the Canadian Malting Barley Technical Centre. Dan Carey and Randy Thiel nano-brewed and performed sensory assessments on the 2014 Oregon Promise beer samples and 2015 CDC Copeland and Full Pint beers at New Glarus Brewing Co. Amanda Benson and Veronica Vega performed sensory assessments of the 2015 CDC Copeland and Full Pint beers at the Deschutes Brewery. Dr. Ignacio Romagosa assisted in the analysis of sensory data and interpretation of results. Dr. Matthew Moscou genotyped the Oregon Promise, updated the linkage map, and assisted in the QTL mapping of sensory data and interpretation of results. Dr. William Thomas analyzed and interpreted QTL x Environment interaction of the sensory data.

9 Chapter 4: Alfonso Cuesta-Marcos assisted in the coordination of the LTT trials and analysis of the winter survival data. Scott Fisk coordinated the planting, maintenance, harvesting, and processing of the LTT trials. Dr. William Thomas assisted in the analysis of the 2014 and 2015 winter survival data.

10 TABLE OF CONTENTS Page Chapter 1: General Introduction...1 Chapter 2: The Oregon Promise I: Preliminary Evidence that Barley (Hordeum vulgare L.) Variety and Growing Environment can Contribute to Beer Flavor...7 Abstract... 7 Key Words... 8 Introduction... 8 Materials & Methods Results Discussion Conclusion Acknowledgements Tables Figures Chapter 3: The Oregon Promise II: Effect of Malt Modification on Barley (Hordeum vulgare L.) Contributions to Beer Flavor...34 Abstract Key Words Introduction Material & Methods Experiment I Experiment II Results Experiment I: Experiment II: Discussion Conclusion Acknowledgements Tables Figures... 66

11 TABLE OF CONTENTS (Continued) Page Chapter 4: Support for Facultative Growth Habit Barley as a Tool for Dealing with Climate Change: Results of GWAS of Winter Survival/Low Temperature Tolerance and Vernalization Sensitivity...70 Abstract Key Words..70 Introduction Materials & Methods Results Discussion Conclusion Acknowledgements Tables Figures Chapter 5: General Conclusions...97 Bibliography Appendix Appendix A Appendix B Appendix C

12 LIST OF FIGURES Figure Page Figure 2.1 (a-d): Principle component analysis of flavors consistently significant across three production environments in Oregon, USA (COR, LEB, and MAD). PCA-A is the combined environment Biplot where COR=circle, LEB=cross, and MAD=diamond, PCA-B is the Corvallis environment, PCA-C is the Lebanon environment, and PCA-D is the Madras environment. Parents and check are labelled: Golden Promise = G; Full Pint = F; CDC Copeland = C Figure 2.2: Relative chromosomal positions of beer flavor quantitative trait loci (QTL) and origins of favorable alleles (Golden Promise = Left of chromosome; Full Pint = right of chromosome) detected in the 2015 Oregon Promise subset (OPS) across three environments in Oregon, USA (Corvallis = A; Lebanon = B; Madras = C) Figure 3.1 (a-b): Principal component analysis of malt modification-related traits based on three grain storage intervals and 37 barley genotypes. A) is the score plot and B) is the loading plots of coefficients Figure 3.2 (a-b): Principal component analysis of beer flavor sensory descriptors based on malts made from three grain storage intervals and 37 barley genotypes. A) is the score plot and B) is the loading plots of coefficients Figure 3.3: Genetic map positions of quantitative trait loci (QTLs) for malt-modification related traits and beer flavor sensory descriptors. Beers were brewed from malts of 34 barley genotypes sampled at three storage intervals. Numbers in parentheses indicate the grain storage interval for which the QTL(s) were detected Figure 3.4 (a-b): Principal component analysis of malt modification-related traits and beer flavor traits based on three levels of malt modification of the varieties CDC Copeland and Full Pint. A) is the score plot and B) is the loading plots of coefficient Figure 4.1 (a-b): Frequency distributions of average winter survival (WS) across 12 environments where there was differential winter injury in a panel of 941 accessions (A) and days-to-flowering (DTF) for the same panel under controlled environment conditions without vernalization...93 Figure 4.2 (a-d): Principal Component Analysis (PCA) of winter survival (WS) of 941 barley accessions in 12 environments where there was differential winter injury. A) distribution of the loading coefficients. The biplots were overlaid with B) average WS scores, where grey < 75% and black is >75% LTT, C) days-to-flowering (DTF) under greenhouse conditions without vernalization, where grey > 96 days and black is < 96 days, and D) spike type, where grey is 6-row and black is 2-row...94

13 LIST OF FIGURES (Continued) Figure Page Figure 4.3 (a-b): Manhattan plots of LOD scores from GWAS of winter survival (WS) scores averaged across 12 environments where there was differential winter injury (A) and days-to-flowering (DTF) times under greenhouse conditions without vernalization (B) of 941 barley accession...95 Figure 4.4: Dendrogram of genetic diversity of the 47 accession from the panel of 941 barley accessions with maximum winter survival across 12 environments where there was differential winter injury...96

14 LIST OF TABLES Table Page Table 2.1: P-values from the analysis of variance of beer flavor descriptors from the Oregon Promise subset grown at three locations (Corvallis, OR, Lebanon, OR, and Madras, OR USA) in Table 2.2: Best Linear Unbiased Estimates (BLUEs) for beer sensory descriptors based on replicated beers (Golden Promise; Full Pint; Rahr Pils; Miller High Life) Table 2.3 (a): Best Linear Unbiased Predictors (BLUPs) for beer sensory descriptors based on un-replicated beers: the Oregon Promise subset doubled haploid lines, Golden Promise, Full Pint and CDC Copeland grown in Corvallis, OR Table 2.3 (b): Best Linear Unbiased Predictors (BLUPs) for beer sensory descriptors based on un-replicated beers: the Oregon Promise subset doubled haploid lines, Golden Promise, Full Pint and CDC Copeland grown in Lebanon, OR 29 Table 2.3 (c): Best Linear Unbiased Predictors (BLUPs) for beer sensory descriptors based on un-replicated beers: the Oregon Promise subset doubled haploid lines, Golden Promise, Full Pint and CDC Copeland grown in Madras, OR..30 Table 2.4: Averaged Best Linear Unbiased Predictors (BLUPs) for beer flavor sensory descriptors (Rahr Malting) and free choice descriptors (New Glarus Brewing) on selected doubled haploid lines from the Oregon Promise subset compared to Golden Promise and Full Pint Table 3.1: Loading coefficients from the principal component analysis of malt modification traits and beer flavor descriptors based on three grain storage intervals and 37 barley genotypes Table 3.2: Best linear unbiased estimates for beer flavor descriptors in replicated sensory checks Table 3.3: Key sources of variation and their p-values from the combined analysis of variance of beer flavor descriptors based on malts made at each of three grain storage intervals (Storage Treatment) from 37 barley genotypes (Genotype) Table 3.4: (A) Grain storage duration treatment means for malt modification-related traits for Golden Promise, Full Pint, their 34 progeny, and CDC Copeland and (B) beer flavor descriptor BLUPs that had significant treatment effects in the combined Table 3.5 (a): (A) Grain storage treatment (ST-1) values for malt modification-related traits for Full Pint, Golden Promise, their 34 progeny, and CDC Copeland and (B) beer flavor descriptor BLUPs that had significant treatment x genotype interaction in the combined ANOVA....59

15 LIST OF TABLES (Continued) Table Page Table 3.5 (b): (A) Grain storage treatment (ST-2) values for malt modification-related traits for Full Pint, Golden Promise, their 34 progeny, and CDC Copeland and (B) beer flavor descriptor BLUPs that had significant treatment x genotype interaction in the combined ANOVA 60 Table 3.5 (c): (A) Grain storage treatment (ST-3) values for malt modification-related traits for Full Pint, Golden Promise, their 34 progeny, and CDC Copeland and (B) beer flavor descriptor BLUPs that had significant treatment x genotype interaction in the combined ANOVA 61 Table 3.6: Loading coefficients from the principal component analysis of malt modification traits and beer flavor traits based on three malting treatments and two varieties (CDC Copeland and Full Pint) Table 3.7: Malt modification-related trait values for, and Least Square Means for beer flavor descriptors made from, under- modified, modified, and over-modified malts (averaged over the varieties CDC Copeland and Full Pint) where there was a significant treatment effect in the combined ANOVA Table 3.8: Malt modification-related trait values for, and Least Square Means for beer flavor descriptors made from, under- modified, modified, and over-modified malts of CDC Copeland and Full Pint where there was significant treatment x genotype interaction in the combined ANOVA Table 3.9: Key sources of variation and their p- values from the combined analysis of variance of beer flavor descriptors identified in PC analysis and based on under-modified, modified, and over-modified malts (Malt Treatment) made from CDC Copeland and Full Pint Table 4.1: Loading coefficients from the principal component analysis of winter survival of 941 barley accessions from theaverage of 12 differential environments Table 4.2 (a): Chromosome position, LOD score, most significant SNP, and candidate gene annotation for quantitative trait loci (QTLs) detected in the GWAS of winter survival (LTT) and days-to-flowering (DTF) without vernalization in 941 barley accessions. The GWAS of WS/LTT is based on the average of 12 differential environments Table 4.2 (b): Chromosome position, LOD score, most significant SNP, and candidate gene annotation for quantitative trait loci (QTLs) detected in the GWAS of winter survival (LTT) and days-to-flowering (DTF) without vernalization in 941 barley accessions. The GWAS of WS/LTT is based on the average of 12 differential environments (Continued)....87

16 LIST OF TABLES (Continued) Table Page Table 4.3: Chromosome position, LOD score, most significant SNP, and candidate gene annotation for quantitative trait loci (QTLs) detected in the GWAS of days-to-flowering (DTF) without vernalization in 941 barley accessions. The GWAS of WS/LTT is based on the average of 12 differential environments Table 4.4 (a): Winter survival (WS) of the top 5% (47 accessions) in the LTT panel across 12 environments with differential winter injury, days-to-flowering (DTF) under greenhouse condition without vernalization, and spike type Table 4.4 (b): Winter survival (WS) of the top 5% (47 accessions) in the LTT panel across 12 environments with differential winter injury, days-to-flowering (DTF) under greenhouse condition without vernalization, and spike type (Continued) Table 4.4 (c): Winter survival (WS) of the top 5% (47 accessions) in the LTT panel across 12 environments with differential winter injury, days-to-flowering (DTF) under greenhouse condition without vernalization, and spike type (Continued) Table 4.5: Winter survival (WS) of the ten most winter hardy accessions with potential faculative growth habit and checks across 12 environments with differential winter injury, days-to-flowering (DTF) under greenhouse condition without vernalization, and spike type....92

17 Table LIST OF APPENDIX FIGURES Page Supplementary Figure C2: Frequency distributions of averaged winter survival scores across 12 environments (A) and flowering times in unvernalized conditions (B) for the 941 LTT panel Supplementary Figure C3: Manhattan plots of LOD scores from GWAS of winter survival scores averaged across differential environments of the 941 accession in the LTT panel

18 Table LIST OF APPENDIX TABLES Page Supplementary Table A1.1: Mean values for agronomic traits measured on the Oregon Promise subset (OPS) grown in 2015 Corvallis, OR Supplementary Table A1.2: Mean values for agronomic traits measured on the Oregon Promise subset (OPS) grown in 2015 Corvallis, OR (Continued) Supplementary Table A2.1: Mean values for agronomic traits measured on the Oregon Promise subset (OPS) grown in 2015 Lebanon, OR Supplementary Table A2.2: Mean values for agronomic traits measured on the Oregon Promise subset (OPS) grown in 2015 Lebanon, OR (Continued) Supplementary Table A3.1: Mean values for agronomic traits measured on the Oregon Promise subset (OPS) grown in 2015 Madras, OR Supplementary Table A3.2: Mean values for agronomic traits measured on the Oregon Promise subset (OPS) grown in 2015 Madras, OR (Continued) Supplementary Table A4.1: Malting quality data from the Oregon Promise subset grown in 2015 Corvallis, OR Supplementary Table A4.2: Malting quality data from the Oregon Promise subset grown in 2015 Corvallis, OR (Continued) Supplementary Table A5.1: Malting quality data from the Oregon Promise subset grown in 2015 Lebanon, OR Supplementary Table A5.2: Malting quality data from the Oregon Promise subset grown in 2015 Lebanon, OR (Continued) Supplementary Table A6.1: Malting quality data from the Oregon Promise subset grown in 2015 Madras, OR Supplementary Table A6.2: Malting quality data from the Oregon Promise subset grown in 2015 Madras, OR (Continued) Supplementary Table A7.1: Best linear unbiased predictors (BLUP) for individual doubled haploids in the Oregon Promise subset grown in Corvallis, OR in Supplementary Table A7.2: Best linear unbiased predictors (BLUP) for individual doubled haploids in the Oregon Promise subset grown in Corvallis, OR in 2015 (Continued)

19 Table LIST OF APPENDIX TABLES (Continued) Page Supplementary Table A8.1: Best linear unbiased predictors (BLUP) for individual doubled haploids in the Oregon Promise subset grown in Lebanon, OR in Supplementary Table A8.2: Best linear unbiased predictors (BLUP) for individual doubled haploids in the Oregon Promise subset grown in Lebanon, OR in 2015 (Continued) Supplementary Table A9.1: Best linear unbiased predictors (BLUP) for individual doubled haploids in the Oregon Promise subset grown in Madras, OR in Supplementary Table A9.2: Best linear unbiased predictors (BLUP) for individual doubled haploids in the Oregon Promise subset grown in Madras, OR in 2015 (Continued) Supplementary Table A10: Quantitative traits loci (QTL) results table for the flavor traits measured in the Oregon Promise subset (OPS) grown in three locations (Corvallis, OR; Lebanon, OR; Madras, OR) in Supplementary Table B1.1: Malting quality data from the Oregon Promise subset grain stored for 1 month (ST-1) prior to malting and brewing Supplementary Table B1.2: Malting quality data from the Oregon Promise subset grain stored for 1 month (ST-1) prior to malting and brewing (Continued) Supplementary Table B2.1: Malting quality data from the Oregon Promise subset grain stored for 5 months prior to malting and brewing Supplementary Table B2.1: Malting quality data from the Oregon Promise subset grain stored for 5 months prior to malting and brewing (Continued) Supplementary Table B3.1: Malting quality data from the Oregon Promise subset grain stored for 10 months prior to malting and brewing Supplementary Table B3.2: Malting quality data from the Oregon Promise subset grain stored for 10 months prior to malting and brewing Supplementary Table B4: Malting protocol used to produced three levels of malt modification at the Canadian Malting Barley Technical Centre, Winnipeg, CA Supplementary Table B5.1: Best linear unbiased predictors (BLUP) for individual doubled haploids in the Oregon Promise subset grain stored for 1 month (ST-1) prior to malting and brewing

20 Table LIST OF APPENDIX TABLES (Continued) Page Supplementary Table B5.2: Best linear unbiased predictors (BLUP) for individual doubled haploids in the Oregon Promise subset grain stored for 1 month (ST-1) prior to malting and brewing (Continued) Supplementary Table B6.1: Best linear unbiased predictors (BLUP) for individual doubled haploids in the Oregon Promise subset grain stored for 5 months (ST-2) prior to malting and brewing Supplementary Table B6.2: Best linear unbiased predictors (BLUP) for individual doubled haploids in the Oregon Promise subset grain stored for 5 months (ST-2) prior to malting and brewing (Continued) Supplementary Table B7.1: Best linear unbiased predictors (BLUP) for individual doubled haploids in the Oregon Promise subset grain stored for 10 months (ST-3) prior to malting and brewing Supplementary Table B7.2: Best linear unbiased predictors (BLUP) for individual doubled haploids in the Oregon Promise subset grain stored for 10 months (ST-3) prior to malting and brewing (Continued) Supplementary Table B8.1: QTL results table for the malting quality traits measured in the OPS grown in Madras, Or and stored at three intervals (1 month (A); 5 months (B); 10 months (C)) Supplementary Table B8.2: QTL results table for the malting quality traits measured in the OPS grown in Madras, Or and stored at three intervals (1 month (A); 5 months (B); 10 months (C)) (Continued) Supplementary Table B9.1: QTL results table for the flavor traits measured in the OPS grown in Madras, Or and stored at three intervals (1 month (A); 5 months (B); 10 months (C)) Supplementary Table B9.2: QTL results table for the flavor traits measured in the OPS grown in Madras, Or and stored at three intervals (1 month (A); 5 months (B); 10 months (C)) (Continued) Supplementary Table C1.1: Potential facultative barley accessions based on the phenotypic criteria of >50% WS and <95 DTF under greenhouse conditions without vernalization Supplementary Table C1.2: Potential facultative barley accessions based on the phenotypic criteria of >50% WS and <95 DTF under greenhouse conditions without vernalization

21 Table LIST OF APPENDIX TABLES (Continued) Page Supplementary Table C1.3: Potential facultative barley accessions based on the phenotypic criteria of >50% WS and <95 DTF under greenhouse conditions without vernalization Supplementary Table C1.4: Potential facultative barley accessions based on the phenotypic criteria of >50% WS and <95 DTF under greenhouse conditions without vernalization

22 DEDICATION To Danielle

23 1 Chapter 1: General Introduction Barley is the fourth most important cereal (FAO-STAT, 2014) and is one of the oldest domesticated crops with utility in addressing the challenges of climate change, applications across multiple agricultural industries, and as a model organism for genetic studies. As an ancient crop, barley was a staple in many regions of the world. Currently, food barley accounts for only 2% of the annual global production, with the majority of it being produced for feed (Ullrich, 2011). However, barley sold for malt demands the highest premium and must meet strict quality requirements. Malting barley varieties are selected for traits within approved specifications to ensure high malthouse and brewhouse performance (Schwartz et al., 2011; AMBA, 2014). Oregon State University and other modern breeding programs routinely select for competitive, high quality malting barley varieties, however the persistence of the older heritage varieties such as Golden Promise, Maris Otter, Harrington, and Klages have remained in high demand despite their agronomic and quality deficiencies on the basis of unique and different flavors. Malting: Barley requires advanced processing in order to fully characterize flavor. Malting is a controlled process consisting of the hydration of grain (steeping phase), initiation of growth under moist conditions (germination phase), and the termination of the grain s physiological activities by drying and curing with heat (kilning phase). Fundamentally, the aim of malting is to expose starch granules from the encompassing cell walls and protein matrix so that fermentable sugars can be extracted efficiently from starch during brewing, a process known as modification (Swanston et al. 2014). Malt modification is the uniform degradation of cell walls, small starch granules, and protein matrix throughout the endosperm (Palmer 1993). This requires rapid uptake and distribution of water into the endosperm during the steeping phase, triggering enzymes hydrolyzing protein, starch, and cell wall structures, while coinciding with the initiation of seedling growth in the germination phase (Davies 1989; Brennan et al. 1997).

24 Flavor: Flavor is formed through the interaction of barley and the malting process, and is highly dependent upon the compositional quality of the grain, degree of modification, and kilning parameters (e.g. temperature, moisture, and time). Kernel composition is the foundation of malt flavor and there are numerous traits such as amino acids, peptides, protein, starch, various aroma compounds, and melanoidins precursors inherent in barley that contribute to flavor. Amino acids and peptides activate at least five human taste receptors, allowing for the detection of various flavors (Kohl et al., 2013). Pure amino acids have the following taste properties: neutral/low perceptible taste, sweet taste, bitter taste, and sulfurous taste, however complex flavors form as amino acids collude with other compounds (Solms, 1969; Kirimura et al., 1969; Silva et al., 2008). Dong et al. (2013) reported 41 aroma compounds present across five malting barley varieties consisting of alcohols, aldehydes, ketones, and organic acids. These compounds elicit flavors such as malty, fruity, vegetable, and toasty as well as adds to the overall mouthfeel and basic taste (Jin et al., 2013; Jin et al., 2014). Although total contributions to beer flavor are relatively low, differences in pedigree among barley varieties, production environment, and GxE interaction may significantly impact finished beer. 2 Modification involves the degradation of cell walls and is the conversion of the starchy endosperm and protein into fermentable sugars and amino acids, respectively - making barley more suitable for brewing (Lewis and Young, 2001). This process breaks down structural and soluble carbohydrates releasing high- and low-molecular weight substances bound within hemicellulose, polysaccharides, protein, and starch allowing these compounds to interact independently throughout malting. Modification is classified into three categories: 1) under-modified, 2) well-modified, and 3) over-modified and all have a significant impact on the flavor potential of barley. Modification is controlled primarily through parameter adjustments during steeping and germination stages of malting. Under-modified malt is common of traditional European lager and North American malts (Briggs, 1998), resulting in less starch conversion and adding a sweet, full body characteristic to finished beers. The primary factors affecting under-modified malt are

25 grain dormancy and beta-glucans. Dormancy, a function of germination capacity and energy, is a mechanism which suppresses germination for a set period of time resulting in extended steeping and longer germination. Inability to account for dormancy in the malthouse may lead to under-modified malt caused by insufficient enzymatic degradation, scheduling issues, and even loss of crop (Jones, 2005). The extent of dormancy varies by variety and environment, therefore varieties with low dormancy are desired and germination test are performed to ensure germination capabilities. Betaglucan, a hemicellulosic polysaccharide found in the cell walls of the endosperm is an indicator of under-modified malt. Beta-glucans restrict water uptake into the cell and limits enzymatic degradation of starch. Depending on the variety, undermodified malts typically have high beta-glucan levels which result in lower friability, lower extracts, and higher wort viscosity causing filtration and turbidity issues during brewing. Mitigating high beta-glucans during malting requires increasing kernel moisture and enzyme levels by extending steeping and raising water temperatures, elevating germination temperatures for longer duration, and a lower starting and intermediate kiln temperature profiles to ensure appropriate modification (Jones, 2005; Schmitz et al., 2013; Sato et al., 2009; Ullrich et al., 2009). Due to complications in the brewhouse, under-modified malts are typically undesirable to brewers. Over-modified malts, such as some British ale malt and distillers malt, also give rise to complication. Maximizing cell wall breakdown and conversion leads to damaged kernels during transportation and handling, and loss of available extract. Upon milling, over-modified malts yield high proportions of flour in grist and cause wort separation and filtering issues in mash/lauter tuns during brewing (Briggs, 1998). Beer made from over-modified malt tend to lack correct character and flavor, as well as have a reduced shelf life (Briggs, 1998). Therefore, a balanced wellmodified malt is optimum for desired performance and flavor. 3 Kilning, also known as curing, stops germination by drying the malt and reducing the enzymatic activity. During kilning, complex color, aroma, and flavor are formed and generally increase with time through the prolonged interaction of reducing sugars and

26 amino acids with elevated temperatures (Briggs, 1998). This interaction produces melanoidins and precursors thereof. Melanoidins are among the many complex compounds called Maillard products which are derived from a series of reactions known to produce high- and low-molecular weight substances (Briggs, 1998). High-temperature reactions produce high-molecular weight substances that are colored, such as melanoidins, whereas low-temperature reactions produce low-molecular weight substances which contribute to aroma and flavor, such as alcohols, aldehydes, ketones, esters, and O-, S-, and N-containing heterocyclic substances (Briggs, 1998). These complex substances are highly volatile and are individually present below flavor threshold concentrations, but together create a strong collusion of flavor characteristic of malt. However, many of these volatiles are evaporated off during high temperature kilning, resulting in the loss of low-molecular weight substances in darker malts. Additionally, kilning removes green-grain off-flavors from malt, but produces other offflavors that are undesirable in finished beer including oxidation flavors from fatty acids of lipids and dimethyl-sulfide (DMS). DMS is a heat liable compound formed by the methylation of methionine and is the primary off-flavor associated with malt, and it contributes a sulphur aroma similar to canned corn, rotting vegetables, and tomato juice. In addition to color, aroma, and flavor formation, kilning increases the malt friability, allowing for better utilization and efficiency in the brewhouse as well as increased shelf stability. 4 Facultative: Barley consist of two main growth habits (spring and winter) which expand the range of adaptation. Spring barley does not require vernalization, a period of low temperatures that initiate the transition from vegetative to reproductive state, and is planted when the risk of winter injury is limited. This planting range varies depending on the growing season. Spring barley takes advantage of optimal growing conditions during the early stages of plant development and matures in a shorter period of time, but utilizes more water and has an increased susceptibility to disease due to immature plantlets during high disease sporulation time. Winter barley uses the fall period for establishment,

27 thereby allowing the crop to take advantage of a longer growing season in the spring and early summer. This reduces the risk of yield loss from summer heat, drought, and disease while expanding adaptation. Winter barley consists of two growth habit types: obligate winter and facultative. Low temperature tolerance is essential for both types, however obligate types require vernalization acquiring of substantial cold units to trigger the vegetative-to-reproductive transition - and limits winter types to fall-sowing. Facultative types have the capacity to cold acclimate and achieve levels of low temperature tolerance equal to obligate winter types and do not require vernalization, but rather rely on shortday insensitivity - delaying the vegetative-to-reproductive transition until long-days to ensure the risk of late winter/early spring low temperature injury has passed. Therefore, facultative types give the grower absolute flexibility in planting date, which is a key consideration for sustainable and profitable grain production in the face of an increasingly volatile climate (von Zitzewitz et al., 2011; Lobell et al., 2008). Facultative growth habit is achieved through the combination of alleles at cold temperature tolerance (FR-H1, FR-H2, FR-H3), vernalization (VRN-H1, VRN-H2, VRN-H3), and short-day photoperiod sensitivity (PPD-H2) loci (von Zitzewitz et al., 2011; Comadran et al., 2012; Fisk et al., 2013; Cuesta-Marcos et al., 2015). Facultative growth habit is defined by a deletion of VRN-H2 coupled with complete winter alleles at all other vernalization, cold temperature tolerance, and photoperiod loci (Fisk et al., 2013). The OSU program has been an active contributor to the literature regarding low temperature tolerance, vernalization and photoperiod sensitivity. 5 This thesis investigates the potential for developing flavorful barley varieties within the framework of facultative growth habit through contemporary breeding methods, such as linkage (QTL) mapping, genome-wide association studies (GWAS). In barley, highly structured linkage populations have been created to estimate the bi-allelic mean and variance associated with a specific locus and find markers linked with QTLs for a trait of interest. Linkage mapping utilizes a high power genome wide scan for linkage groups, however resolution is relatively low (Wurschum et al., 2012). GWAS panels have been

28 constructed for the discovery of novel alleles and characterize genetic variation for a trait(s) of interest that is more immediately useful in breeding (Yu et al., 2006; Roy et al., 2010; Cuesta-Marcos et al., 2010; Massman et al., 2011; von Zitzewitz et al., 2011; Bradbury et al., 2011). These panels are typically comprised of accessions with diverse genetic backgrounds, therefore population structure is unknown and statistical power is relatively low. However, GWAS is a high resolution mapping tool that analyzes many alleles allowing for candidate gene testing. 6 Modern breeding programs are limited in size, capacity, and available resources restricting the use of large diverse pools of exotic germplasm and resulting in lower genetic gains for a trait of interest. In order to meet the production demands, breeding tools have been developed to increase gains from selection. Screening barley lines for malting quality and flavor profiling requires advanced processing, analytical instrumentation, and human sensory, which is laborious, time-consuming, and expensive. Therefore, capacity is limited and only select advanced material can be submitted for malt and flavor analysis. Populations consist of a smaller sample size (n=50-100) are utilized to genetically map traits that are more difficult to phenotype (Cuesta-Marcos et al., 2010). The USDA Coordinated Agricultural Project (CAP) projects were established to create large phenotypic and SNP marker data sets of elite breeding lines from U.S. breeding programs with which to conduct genetic mapping studies (Rae et al., 2007; Waugh et al., 2009; Ariyadasa et al., 2014; Blake et al., 2015). The SNP genotyping platform, trait and marker database (The Hordeum Toolbox; and wealth of QTL information generated from the CAP provide a rich framework and set of breeder s tools used to initiate ambitious new breeding efforts for barley flavor and malt quality within desired growth habit, maturity, height, and morphological combinations.

29 Chapter 2: The Oregon Promise I: Preliminary evidence that barley (Hordeum vulgare L.) variety and growing environment can contribute to beer flavor 1 Herb, D.W., 2 Meints, B.M., 3 Jennings, R., 4 Romagosa, I., 5 Moscou, M., 6 Carey, D., 7 Cistue, L., 1 Filichkin, T, 1 Fisk, S.P., 1 Helgerson, L., 8 Martens, C., 3 Monsour, R., 6 Thiel, R., 3 Tynan, S.P., 9 Thomas, W.B., 3 Vinkemeier, K., 1 Hayes, P.M. 7 1 Crop & Soil Science Dept., Oregon State University, Corvallis OR USA; 2 Dept. of Crop & Soil Science, Washington State University, Mt. Vernon, WA USA; 3 Rahr Malting Co., Shakopee, MN USA; 4 Agrotecnio, University of Lleida, Lleida, Spain; 5 The Sainsbury Laboratory, Norwich Research Park, Norwich NR4 7UH UK; 6 New Glarus Brewing Co., New Glarus WI USA; 7 Estación Experimental Aula Dei, CSIC, Zaragoza, Spain; 8 Cereal Crop Research Unit, USDA-ARS, Madison, WI USA; 9 The James Hutton Institute, Invergowrie, Dundee DD2 5DA, Scotland UK Abstract The relationships between barley varieties and beer flavors are not well-understood. Therefore, this research assessed the relative importance of barley genotypes, production environments, and their interactions on beer flavor descriptors. Thirty-four doubled haploid lines from the Oregon Promise population (Golden Promise x Full Pint) were grown in three environments in the USA (Corvallis, OR; Lebanon, OR; Madras, OR) and assessed for flavor differences through micro-malting, nano-brewing, and sensory evaluations. There were significant differences between barley genotypes and environments for specific flavor descriptors and genotype x environment interactions were significant for a subset of descriptors. The most important flavor descriptors were cereal, floral, fruity, grassy, honey, malty, toasted, toffee, and sweet. Notable differences were detected between parents: Golden Promise was significantly higher in fruity, floral, and grassy flavors whereas Full Pint was significantly higher in malty, toffee, and toasted flavors. New combinations of parental flavor attributes were observed in the progeny.

30 Exploratory and conservative QTL analysis revealed regions of the barley genome with significant effects on flavor. Multiple lines of evidence show that (i) barley genotypes can make different contributions to beer flavor, (ii) barley contributions to beer flavor have a genetic basis, and (iii) production environment can affect beer flavor. 8 Key Words Barley, malt, beer, flavor, genetic, variety Introduction Barley, in its malted form, is the principal source of fermentable sugars for most beers and some spirits. Because barley malt, rather than barley grain, is used in these applications, flavor contributions to beer are usually ascribed to the malt rather than to the variety. Indeed, the range of malts available from base (e.g. Pilsen, and Pale) to specialty (e.g. Vienna, Munich, Crystal, Caramel), to roasted (e.g. Chocolate, Coffee, and Roasted grains) types gives the brewer a palette of potential flavors (Briggs, 1998; Mallet, 2014). Fundamental contributors to flavor are Maillard reaction and Strecker degradation products, such as melanoidins, formed during the kilning and roasting stages of malting (Briggs, 1998). Understanding the differences in flavor found in different malts is the subject of extensive ongoing research (Barr, personal communication; Liscomb, personal communication). Since the same variety can be used to make the full spectrum of malts, most emphasis is placed on the suitability of barley cultivars for malting, rather than on the potential favorable contributions of the variety to beer flavor. New barley varieties are rigorously tested through programs set by bodies such as the American Malting Barley Association (AMBA) ( the Brewing and Malting Barley Research Institute (BMBRI) ( the Barley Australia ( the Canadian Malting Barley Technical Centre (CMBTC) ( and the European Brewing

31 Convention (EBC) ( to meet a comprehensive list of agronomic, malting, and brewing metrics prior to release. The approval process, in its final steps, involves elimination of potential varieties based on negative contributions to beer flavor. Commercially perceived negative flavors contributed by the barley variety via its malt include, but are not limited to excessive dimethyl sulfide (DMS), diacetyl, and aldehydes resulting from lipoxygenase activity (Hirota et al., 2006; Zhao et al., 2006; Saerens et al., 2008) 9 Assuming that the malt (or malts) used in a particular brew give desired flavors, aromas, and other sensory attributes, brewers have been free to explore the major contributions of hops, water, yeast and adjuncts to beer flavor. The impacts of these materials can be considerable and account for some obvious flavor differences between different beer styles e.g. between a standard American lager, a saison and an India pale ale. It is not surprising that the barley contribution to beer flavor has not been a high research priority. Nonetheless, certain barley varieties are acknowledged by some brewers to have notable flavor attributes. These notable flavor attributes have been sufficient to ensure the continued production of specific varieties even when newer varieties have superior agronomic and/or malting performance. Notable examples from Europe are Golden Promise and Maris Otter. Notable examples from North America are Harrington, and Klages. Occasionally, newer varieties, such as Full Pint attract the interest of the craft malting, brewing, and/or distilling industries based on their perceived unique contributions to product flavor (Thomas, 2014). Paralleling our limited understanding of the variety contribution to beer flavor, it is unclear how barley cultivar growing region contributes to beer flavor. Barley production is largely driven by a desire to maximize productivity, consistency, and profitability across as extensive a geographic area as possible (Briggs, 1998; Johnson et al., 1999; Mallet, 2014). An underlying assumption of this approach is that by generating as large a pool of barley as possible, at least some will meet malting specifications. This

32 approach contrasts with the terroir concept as applied to fine wines in which growing environment is of paramount importance. Considerable progress has been made in characterizing contributors to terroir in terms of viticultural practices, wine making practices (Kontkanen et al., 2005; Schlosser et al., 2005) and environmental factors (Bodin & Morlet, 2006; Morlet & Bodin, 2006). However, terroir remains elusive in the wine industry and only recently has the term appeared with reference to cereal grains ( 10 To test the hypothesis that barley variety contributes to beer flavor, a genetics-based model, in contrast to a survey-based model, was used. That is, rather than comparing flavor attributes of beer brewed in an identical fashion from malts made in an identical fashion from a small panel of different varieties, crossing was carried out between two varieties reported to contribute unique flavors to beer (Golden Promise and Full Pint). Flavor was assessed, using sensory descriptors, in beers made from malts of these varieties and from a sample of their cross progeny. Due to the complexity and cost of assessing beer flavor, micro-malting was done on grain from research plots, followed by nano-brewing and sensory analysis of the nano-beers. The primary objective of this exploratory research was to determine if there were differences in beer flavor between the two parents, between the parents and commercial beers, and among cross progeny. If there were differences, the research sought to characterize differences in terms of flavor descriptors, to determine if the flavor descriptors have a genetic basis, and to assess the role of production environment on flavor. Materials & Methods Germplasm Two hundred doubled haploid lines (DHLs) were derived from the F1 generation of the cross of Golden Promise x Full Pint following the anther culture protocol described by

33 Cistue et al (2003). The doubled haploids referred to as the Oregon Promise population are available to the research community; instruction to access seed and data are available at ( The DHLs were produced in the L. Cistue lab at Estacion Experimental de Aula Dei, Consejo Superior de Investigaciones Cientificas in Aula Dei, Spain and in the P. Hayes lab at Oregon State University in Corvallis, OR USA. For this study, a subset of 34 DHLs was used. These 34 DHLs were the subject of extensive agronomic trait assessments. 11 Testing Environments The Oregon Promise subset (hereafter referred to as OPS) of 34 DHLs was grown in 2015 at three test environments in Oregon. Two of the environments - Corvallis and Lebanon - are in the high rainfall Willamette Valley with cm and cm of annual precipitation respectively. Therefore, no supplemental irrigation was used at these sites. The Corvallis site (COR) is located at N, W and the Lebanon (LEB) site is located at N, W. The third site - at Madras (MAD) - was irrigated, since it is located in a 25.4 cm annual precipitation zone in central Oregon. The MAD site is located at N, W. Crop management practices seeding rate, fertilization and weed control were in accordance with local best practices. Field Plot Experimental Design The OPS, Golden Promise, Full Pint and a check (CDC Copeland) were grown in a tworeplicate Randomized Complete Block design at COR in 2014 and COR, LEB, and MAD in Plot sizes were 9.2m 2 and were sown with a Wintersteiger Plot Seeder XL grain drill and harvested with a Wintersteiger Classic plot combine. Agronomic data for the OPS is shown in Supplementary Tables (Appendices A1-3). Micro-malting & Malt Quality The 2014 OPS and 2015 OPS samples were malted at the USDA Cereal Crop Research Unit (CCRU) in Madison, WI USA and Rahr Malting Co. in Shakopee, MN USA,

34 respectively. Two hundred fifty grams (250 g) of grain samples from each environment were micro-malted using a Joe White Malting unit. Samples were randomized and malted by environment in groups of 30 entries per malt run. Additional samples of the parents from an alternate source harvested at Corvallis were malted for use as sensory checks and each run included an internal standard (Moravian 69) to track malting consistency. A total of four runs were required to malt all 2015 samples. The base malt protocol was as follows: Steeping 8 hrs wet, 16 hrs couch, 8 hrs wet, 10 hrs couch, 1 hr wet, and 2 hrs dry at 14 C; Germination 96 hrs at 15 C; Kilning 24 hrs at 85 C. Malt quality analyses were performed by the respective institutions following ASBC ( standard methods (ASBC Barley-3, Barley-5, Malt-3, Malt-4, Malt-6, Malt-7, Malt-12, Wort-8, Wort -9, Wort -10, Wort -12, Wort -13, Wort-18) (2). Malt quality data for the OPS is shown in Supplementary Tables (Appendices A4-6). 12 Nano-brewing Nano-brews from OPS malts were produced at two facilities using two different recipes: 1) a commercial beer recipe developed by New Glarus Brewing Co, in New Glarus WI and 2) a research recipe developed by Rahr Malting Co. The commercial recipe was prepared as follows: One hundred sixteen grams (116 g) of 2014 OPS malt produced at CCRU was milled to approximately 2 mm using a Captain Crush Grain Mill (Northern Brewer PN: 41212). Nine hundred twenty-five ml (925 ml) of brewery water were added into 1,200 ml stainless steel beakers and maintained 68 C (strike water). To each beaker 115 g of malt grist were added and mash temperature was held at 65 C for 30 minutes. Mash temperature was increased to 70 C at a rate of 1 C/min and held for 60 minutes. The mash was then cooled from 70 C to room temperature. Mash samples were lautered using 64 oz HDPE funnels (Fisher PN: A) lined with Ahlstrom 562 fluted filter paper (Fisher PN: E). Filtration was done over 2.5 hours, without sparging, with one stir of the lauter bed done at 2 hours to improve run-off. Approximately 800 ml of clean wort per sample was collected and standardized to a total volume of 1,100 ml

35 with brewery water. The resultant wort was boiled in 4 L beakers without stir bars for 1 hr. A single hop addition of Cascade 45-pelletized hops at 30 min was done for a target of 12 BU, assuming 20% utilization. Kettle boils were done for 45 minutes, boiled to a target volume of 500 ml to target a finished beer of 5% alc v/v. Each wort was then standardized up to a volume of 650 ml with brewery water, filtered into 1 L sterile fermentation flask, then cooled to room temperature. Wort samples were pitched at a rate of 14x10 6 cells/ml of New Glarus Brewing yeast. After pitching, wort samples were swirled vigorously for 10 revolutions to aerate and mix the yeast. Samples were fermented at room temperature, using air locks with silicone stoppers to avoid excessive oxidation until fully attenuated. The samples were swirled daily during fermentation to prevent foaming during the bottling phase. After 6 days of fermentation, the flasks were cooled for 2 hours at 4 C to reduce foaming. Approximately 2.2 g of sugar was added to each empty bottle then purged with CO2 before being filled with the beer sample, crowned, and labeled. The bottles were stored at room temperature to bottle condition for 14 days before sensory analysis. 13 The research recipe was prepared as follows: Eighty grams (80 g) of the 2015 OPS malt produced at Rahr Malting Co. was coarse-milled with a Monster Mill MM3 3-roller mill (Northern Brewer PN: 40331) and 75 g was sieved to a 2 mm particle size into mashing cans to which 400 ml of distilled water were added. The mash cans were then placed in a 25-sample CT4 mash bath (Canongate Technology Limited). Two mash cans were prepared per malt sample to obtain the target beer volume 700 ml minimum. The mashing protocol followed a modified EBC standard method: 1) heat to 45 C and hold to 30 minutes, 2) ramp temperature from 45 C to 70 C at a rate of 1 C per min, 3) hold at 70 C for 1 hr, 4) cool from 70 C to 22.5 C, and 5) remove samples. The volume of each mash was standardized to 450 ml with distilled water and poured over 30 μm pore size Ahlstrom 562 fluted filter paper (15 μm particle size) (Fisher PN: E) to separate the clear wort from the grist. The first 100 ml of filtered wort was poured over the grain bed to increase clarity. Each sample was sparged with 100 ml of 78 C distilled water to

36 wash out residual sugars in the grain bed. Once the collected wort volume reached 500mL, the two mash cans per sample were combined and dosed with 10 μl of Isohop, 30% iso-alpha acid concentrate from the Barth-Haas Group, pipetted for a target of BU assuming 80% utilization. The combined wort samples were moved to a hot plate and boiled for 35 minutes. The sample volume at the end of boiling ranged from ml and were thus standardized to target a specific gravity of through the addition of distilled water. The samples were filtered into 1 L centrifuge bottles to removed denatured protein and trub, then placed into a water bath to reduce temperature to 20 C for pitching yeast. WhiteLabs ( Czech Budejovice flavor inert yeast was propagated from media to a slurry 48 hrs. prior to pitching. The wort samples were transferred to ASBC fermentation tubes and pitched with ml of yeast slurry at a rate of 1.0 x 10 7 cells/ml into wort with a target specific gravity The fermentation tubes were placed in a fermentation chamber at 12 C and fermented for 14 days. After fermentation, the wort was transferred to autoclaved bottles, centrifuged, and sterile-filtered before bottling and carbonated to approximately 10 psi using Fermenter s Favorite Fizz Drops (Northern Brewer). The bottles were boxed and placed in a refrigerator at 16 0 C to condition for three weeks. All beers, including parent malt beers and check malt beers, were brewed over 12 days. Eleven OPS samples were brewed in ten days. Replicated beers of the parents (11 Golden Promise and 11 Full Pint) were brewed across two days. All brewing sessions included an internal control (Rahr Pils) brewed at the Rahr Malting Co Tech Center. The Rahr Pils is a light ale made with pilsner base malt and crystal malt and hopped with Amarillo and cascade hops. A total of 145 beers were brewed to an ale style. 14 Sensory Nano-beers were evaluated via an augmented design (Lin & Poushinsky, 1985) where variance was measured on the performance of the replicated checks (Golden Promise, Full Pint, Rahr Pils, and Miller High Life) and used to calculate Best Linear Unbiased Estimators (BLUEs) and the value of un-replicated DHL for a given trait were adjusted

37 15 based on the mean of the checks within a tasting session for calculation of Best Linear Unbiased Predictors (BLUPs). The sensory assessment of the OPS nano-beers was based on a 0-8 scale comparison-toreference descriptive analysis, in which each sample was compared to a reference beer of similar style to characterize flavors and estimate variability between samples. The ballot consisted of 17 flavor descriptors. Twelve trained panelists conducted the sensory assessments. Tasting samples were approximately 60 ml, with a total of 150 beers tasted over 10 tasting sessions. Each session included 11 OPS beers, 2 parental check beers, 1 Rahr Pils, and 1 embedded reference (Miller High Life). Miller High Life was used as an embedded reference because it is a commercially available beer exemplary of an American lager with a consistent flavor profile. Genotyping Genotyping of the Oregon Promise population was performed using a custom Illumina BeadExpress 384-plex based on previously characterized SNPs with a high minor allele frequency (Closer et al., 2009). A total of 171 BeadExpress SNP markers were polymorphic. KASP markers were developed from SNPs in the designs of the POPA/BOPA and OPA 9K to bridge fragmented linkage groups and ensure markers were present at distal positions of chromosome arms (Close et al. 2009; Comadran et al. 2012). Linkage Maps A framework genetic map was initially developed with the BeadExpress SNP markers using JoinMap (version 4), which integrated 168 markers into nine linkage groups with chromosomes 1H and 6H fragmented into two linkage groups. Addition of KASP markers generated a final genetic map with 251 markers, of which 206 are nonredundant, with a genetic distance of 1,311 cm, using the Kosambi function, over eight linkage groups. The majority of intervals between markers are below 20 cm, with only four regions on chromosomes 3H, 6H, and 7H having regions above 20 cm. Despite

38 16 substantial effort, markers could not be developed to bridge the two linkage groups of chromosome 1H. Collinearity was observed for all markers relative to the consensus genetic map of barley (Close et al., 2009). Statistical Analysis Least square means of sensory data were calculated with JMP Pro statistical software (Version 12: SAS Institute Inc.) using Best Linear Unbiased Estimates (BLUEs) as an estimation an individual fixed effect mean based on the lowest variance within a population. Analysis of variance (ANOVA) was performed using BLUE values and a mixed linear model (MLM) as implemented in JMP Pro 12. Student s t-test was used for mean separation. Best Linear Unbiased Predictors (BLUPs) were calculated for the unreplicated data from each environment to estimate the least square means of the random variables. Heritability estimates were calculated with JMP Pro 12 using the following formula: h 2 = σ 2 G /(σ 2 G + σ e ) where σ G represents the genetic variance and σ e the residual variance. Quantitative trait loci (QTL) analyses were performed for the 17 flavor traits for each environment using composite interval mapping (CIM) as implemented in Windows QTL Cartographer 2.5 (Fisk et al., 2013). Eight co-factors were chosen using the CIM standard model with forward and backward elimination, 0.1 probabilities in and out of the model, a walk speed of 1.0 cm, and a scan window of 10 cm. A significant (α = 0.05) likelihood ratio test (LR) threshold for QTL identification were determined with 1,000 permutations and expressed as logarithm of the odds (LOD) score. Results The significance of key terms in the ANOVAs for the 14 of the 17 sensory descriptors where at least one of the terms was significant in at least one environment are shown in Table 2.1. Of the 14 descriptors, the Genotype term was significant in all three environments for nine of the descriptors: cereal, floral, fruit, grass, honey, malt, sweet,

39 17 toasted, and toffee. For four of the descriptors, Genotype was significant in two environments: sulfur, chemical, roasted, and bitter. The descriptor grain was significant in only one environment (LEB). Principal Component Analysis The relationship between flavors was observed in the PC analysis of 14 flavor descriptors with significant a Genotype term. The first 3 PCs accounted for 69.9% of the variability (Figure 2.1). The magnitude of the loading coefficients indicates that 9 of the 17 coefficients contributed the majority of the variation observed for flavor, including cereal, floral, fruit, grass, honey, malt, sweet, toffee, and toasted. The remainder of the paper will be reported on these descriptors. PC1 accounted for 36.5% of the variation and was contributed predominately by fruit, malt, toasted, and toffee. PC2 accounted for 20.3% of the variation and was contributed predominately by floral, grass, and honey. PC3 accounted for 13.1% of the variation and was contributed predominately by cereal and sweet. OPS Flavor Considering the BLUEs for the nine flavor descriptors where Genotype was significant (Table 2.2), there were significant differences between replicated sensory checks for the descriptors floral, fruit, malt, toasted, and toffee. Overall, Rahr Pils had the highest value for cereal, floral, fruit, honey, malt, and sweet. When comparing only the parental genotypes used as checks, Golden Promise had the higher BLUEs for floral and fruit whereas Full Pint was higher for malt, toasted and toffee. Miller High Life was neutral (with an average of 3.71 across descriptors) and relatively consistent (with a standard error of 0.30) across all flavor descriptors. The BLUPs for the nine flavor descriptors where Genotype was significant are shown for the un-replicated Golden Promise and Full Pint parents from each environment in Table 2.3. The individual environment BLUPs for the un-replicated parents are similar to the replicated BLUEs for the parents, where the malt was made from the Corvallis, 2014 crop: Golden Promise had higher

40 values for floral and fruit, whereas Full Pint was higher for malt, toasted, and toffee. Across environments, the Copeland field check came closest to 4 on the 0-8 scale across the three environments and most similar to Golden Promise. 18 Heritability Maximum and minimum descriptor values for the 34 OPS along with the corresponding standard errors and h 2 values are reported in Table 2.3. Most flavor descriptors had the highest values at COR, with notable exceptions being cereal and sweet. BLUPs for each DHL are shown in Supplementary Tables (Appendices A7-9). There were moderate to high (>0.30) h 2 estimates for malt and toffee flavors in all three environments; these were highest at LEB (61.4% and 41.9%, respectively). In contrast, h 2 estimates for honey were very low (<3.2%). Other flavor descriptors had h 2 values falling between those for honey vs. those for malt and toffee. For most traits there were phenotypic transgressive segregants - progeny with values higher and/or lower than the parents (Tables 2.3 and 2.4). QTL mapping Using the OPS BLUPs from each environment, 41 significant marker: trait associations (QTLs) were detected for the 9 flavor descriptors. Twelve QTLs for 7 flavor descriptors (floral, fruit, honey, malt, sweet, toasted, toffee) had LOD scores 5.5 in one environment or were detected in more than one environment but had LOD score 4.5. Due to the limited population size used in this experiment, which can incur a high risk of false positives, these 12 QTLs will be the focus of the remainder of this report. These 12 QTLs are highlighted in Supplementary Table (Appendix A10) and shown graphically in Figure 2.2. Six QTLs were detected on 5H, 4 on 2H and 2 on 1H. Five QTLs were detected at COR, 4 at LEB, and 1 at MAD. In Figure 2.2, descriptors to the left of each chromosome cartoon indicate that the higher value was contributed by Golden Promise while those to the right indicate that the higher value allele was contributed by Full Pint. This QTL tally counts the floral QTL detected in COR and MAD on 2H as two QTLs.

41 This is the only case where the same SNP had the most significant association with the same trait in more than one environment. The only other instance of potential QTL coincidence was on the long arm of 5H where two QTLs for toffee are 21 cm apart and a QTL for honey was within 4 cm of a QTL for toffee. 19 Discussion Multiple lines of evidence generated by this research indicate that there are significant differences in beer flavor due to barley genotype and that these differences have a genetic basis. There is also evidence growing region contributes to beer flavor. Beer sensory experiments are typically conducted using trained panelists, a small number of test beers, and as many replications as possible (Lawless & Heymann, 2010; Peltz et al., submitted). In our experiment, due to the large number of samples involved, an alternative approach was necessary: we used trained panelists, but due to the large number of samples required for the genetic-based approach to assess barley variety and growing region contribution to beer flavor, we used an augmented design that utilized replicated reference beers, checks and 111 un-replicated beers brewed from malts made from 37 different barley varieties grown in three environments. A comparison of the ANOVA results from this experiment with the literature (Vollmer et al., 2016; Sharp et al., 2016) indicates that our sensory approach was successful: significant effects were detected for multiple terms in the ANOVAs of beer flavors. The term Panelist was significant in all but two of the 42 ANOVAs, which indicates that panelist perception was, in most cases, a significant source of variation. Recent papers involving hop contributions to beer flavor also report that Panelist effect was often significant (Vollmer et al., 2016; Sharp et al., 2016). Due to the number of distinct beer samples, and replicated checks, this experiment required separate tasting sessions in order to prevent incurring panelist fatigue. The results of the ANOVAs confirm the importance of partitioning out this source of variation for a number of the flavor descriptors. Most importantly, after accounting for other sources of variation, there were significant

42 Genotype effects for most of the flavor descriptors at each of the three locations. We focus on nine of these descriptors in this report because there were significant Genotype terms at each location, there were significant differences between replicated reference beers, there were significant differences between OPS and these differences were heritable, there were consistent patterns in the Principal Component Analyses and significant QTL effects. 20 Considering the replicated sensory assessments, the flavor descriptor means for Miller High Life attest to its value as the reference beer upon which to perform flavor difference assessments. The Rahr Pils consistently had the highest, or among the highest, scores for all flavor descriptors. It therefore set an upper limit for flavors that could be expected from a flavorful beer brewed using base malt, specialty malts, and hops. Golden Promise and Full Pint checks had unique and contrasting flavor signatures. Golden Promise had flavor scores comparable to the Rahr Pils for fruit and floral and low scores for malt, toasted, and toffee. Full Pint had the reverse pattern, with malt, toasted, and toffee flavors comparable to Rahr Pils and a notably high flavor score for malt. For the remaining flavor descriptors grass, cereal, honey, and sweet Golden Promise and Full Pint were similar to each other and Miller High Life. A similar pattern for the un-replicated Full Pint and Golden Promise beers was observed. In this case, the comparator is CDC Copeland. Golden Promise had significantly higher scores for fruit and floral and lower scores for malt, toasted, and toffee. Full Pint had the reverse pattern. Golden Promise and Full Pint were similar to CDC Copeland for grass, cereal and honey. In these data, Full Pint had the highest scores for sweet. A flavor score near four for key flavors may be an advantage for a raw material that is grown widely, used by multiple processors (e.g. maltsters) to create a range of end products by different brewers. The confirmation of flavor differences between nano-brews made from Golden Promise, Full Pint, and CDC Copeland malts provides crucial evidence that barley variety can contribute to beer flavor. Most immediately, it supports the popular perception that

43 Golden Promise can contribute unique flavors to beer and the accumulating evidence that Full Pint can also make unique and positive contributions (25, 33). The associations of fruit and floral vs. malt, toasted, and toffee are apparent in the PCA. These figures also underscore the environmentally-dependent nature of flavor scores for honey, sweet, grass, and cereal. The clearest separation of these flavors was in beers made from malts produced from Madras grain: at this location, grass was associated with floral and fruity and Golden Promise; the other flavors were associated with Full Pint. 21 A comparison of the significance of flavor descriptors across locations suggests that growing environment can differentially affect the contribution a barley variety can make in terms of specific flavors. While these results are based on one year of data, this suggests that attributes of individual environments may promote specific flavors over others. These environmental factors include, but are not limited to climate, soil type, irrigation, nutrients, pest control, and other management practices. The association of environment with flavor, known as terroir, which was coined by the wine industry to differentiate quality attributes based on production region (Bohmrich, 1996; van Leeuwen et al., 2006) and since, has been used internationally to describe a wide-range of agricultural products, including meat & dairy dairy (Salette et al., 1996; Coulon et al., 2002), tea (Chan, 2012), coffee (Avelino et al., 2005) and tequila (Bowen et al., 2009). Additionally, Murphy et al (2016) reported on the development of technologies, such as the Irish Single Pot Still, that may accentuate terroir-based attributes. However, terroir in barley has not been reported. On one hand, terroir can be a marketing advantage for specialty malts and locally-based maltsters. On the other hand, it can complicate efforts to produce uniform malts from large quantities of barley sourced from multiple environments. Whether or not the differences in flavor descriptors observed across environments in this study are intrinsic effects characteristic of these environments or are due to seasonal variation cannot be answered with the data presented. However, based on these preliminary results a deeper characterization and analysis of the role of environment in barley contributions to beer flavor appears to be warranted.

44 22 In addition to production environment, as defined above, other factors may influence barley contributions to beer flavor. Harvest and storage conditions including moisture content, temperatures, and timing can influence the dormancy and germination capacity of barley which is vital for malting quality (Jacobsen et al., 2002; Gubler et al., 2005). Differences in malting protocols during steeping, germination, and kilning can affect degree of modification and therefore potentially affect malt and beer flavor (Agu et al., 1997; Briggs, 1998; Guido et al., 2007). Herb et al. (In Review) further explored the effects of degree of modification on beer flavor descriptors. Two lines of evidence support a genetic basis to barley contributions to beer flavor: estimates of heritability and significant molecular marker: trait associations (QTLs). Heritability estimates ranged from 0 (indicating a large environmental effect, complex inheritance, and/or un-accounted for sources of variation in measuring the trait) to very high (indicative of simple inheritance, small environmental effects, and/or precise trait assessment). The flavor descriptors associated with the parents (floral and fruity with Golden Promise and malt, toasted, toffee with Full Pint) showed contrasting patterns of heritability: those associated with Golden Promise were lower and those with Full Pint higher. Gains from phenotypic selection for specific flavors will be greater in high h 2 environments compared to lower h 2 environments. For instance, selecting for fruit and malt flavors in LEB may be more effective because more of the phenotypic variance is due to genotype compared to COR and MAD. Likewise, selection gains for sweet flavor in MAD may be greater compared to COR and LEB. However, as noted earlier, a deeper characterization of the effects of environment is warranted. The discovery of significant QTLs associated with specific flavor descriptors further corroborates a genetic basis to barley contributions to beer flavor. QTL results indicate that both parents provided favorable alleles: Golden Promise contributed alleles to floral, fruit, sweet, and toffee flavors, whereas Full Pint contributed to honey, malt, toasted, and

45 toffee flavors. The contribution of favorable alleles for toffee flavor from both parents indicates a genetic basis for transgressive segregation observed in the progeny. However, these QTL results can only be viewed as preliminary due to the small sample size used (n = 34): larger population sizes are required for more accurate estimates of allele effect and gene location (Bernardo, 2016). A QTL mapping experiment is in progress using a larger sample of Oregon Promise doubled haploids. That larger and more robust data set can be used, together with the available barley genome sequence (The International Barley Genome Sequencing Consortium, 2012) and metabolomic data to identify candidate genes and pathways for specific flavor descriptors. Accurate QTL data can also be used as a basis for implementing a range of molecular breeding strategies - from marker assisted selection to genome editing - for specific flavor components. These marker and/or gene-based approaches to developing barley varieties with specific attributes are feasible, but not straightforward. Bernardo (2016), for example, has pointed out the challenges of marker-assisted selection for complex traits, and application of CRISPR genome editing to barley is constrained by the transformability of barley genotypes. In this context, Golden Promise is one of the most transformation-amenable varieties known and doubled haploids from the Oregon Promise population have been identified with Golden Promise levels of transformability (Hisano et al., 2016). 23 Of immediate interest are Oregon Promise doubled haploids with unique flavors and commercial potential. In this regard, there are OPS DHLs with specific flavor attributes higher and lower than either parent, and DHLs with novel combinations of flavors - four examples are shown in Table 2.4 In this table, the average values across the three locations are shown for the four OPS DHLS, the two parents and the CDC Copeland check. DHL was high for the flavors characteristic of Full Pint (malt, toasted, and toffee) and low for the Golden Promise flavors (fruity and floral) whereas DHL was high for the Golden Promise flavors and low for the Full Pint flavors; DHL was a low transgressive segregant for both Golden Promise and Full Pint flavors; and DHL was a high transgressive segregant for both Golden Promise and Full Pint

46 flavors. This same subset of DHLs was nano-brewed by New Glarus Brewing Co. to specifications typical of commercial beer and subjected to free-choice flavor profiling. Interestingly, the objective and replicated sensory assessment coincided with the free choice estimates for Full Pint and but there is no obvious relationship between the two sensory assessments for the other three DHLs. It is important to note that in the case of flavor, positive terms such as malty are not always a case of more is better and negative terms such as astringency are not always best when absent. Rather, a balance of flavors may be the goal. The next challenges will be to engage maltsters and brewers in deciding which flavor combinations to assess in the larger scale pilot malting and brewing trials that will be undertaken to (i) validate the nano-brew results and (ii) move these flavors into commercial beers. 24 Conclusion This multi-faceted exploratory study implemented high-throughput small-scale malting, brewing, and sensory assessment of a large number beers and provided evidence that (i) barley varieties can make different contributions to beer flavor, (ii) that variety contributions to beer flavor have a genetic basis, and (iii) that growing environment of the barley can have an effect on beer flavor. Acknowledgements We would like to thank Seth Klann (Klann Family Farm) and Matt Herb (OreGro Seeds, Inc) for the production of the OPS, Patricia Aron, Paul Kramer, and Xiang Yin - Rahr Malting Co. for the malting, quality analysis, and nano-brewing of the OPS, and Richard Goram at the John Innes Centre genotyping facility for KASP genotyping. This research was supported and funded by the Flavor 7-pack of breweries: John Mallett - Bells Brewing, Veronica Vega - Deschutes Brewery, Matthew Brynildson - Firestone-Walker Brewing Co., Daniel Carey - New Glarus Brewing Co., Mike Gilford and Vinnie Cilurzo - Russian River Brewing Co., Tom Nielsen - Sierra Nevada Brewing Co., and Damian

47 25 McConn - Summit Brewing Co, the Gatsby Charitable Foundation, and the Spanish Ministry of Economy and Competitiveness: project AGL C3.

48 MAD LEB COR 26 Tables Table 2.1: P-values from the analysis of variance of beer flavor descriptors from the Oregon Promise subset grown at three locations (Corvallis, OR, Lebanon, OR, and Madras, OR USA) in Source Env Bitter Cereal Chemical Floral Fruit Grain Grass Honey Malt Roasted Sulfur Sweet Toasted Toffee Panelist *** *** *** *** ** * *** *** *** * *** *** Session * * Genotype * ** *** *** ** ** *** ** ** *** *** ** Panelist *** *** *** *** *** *** *** *** *** *** *** *** *** *** Session * * ** Genotype ** *** ** ** *** *** *** ** *** *** *** *** *** Panelist ** *** *** ** ** * *** *** *** *** *** *** ** * Session ** *** * * Genotype ** ** ** *** *** ** *** *** *** *** *** NS, *, **, & *** for No Sig Diff, P<0.05 >0.01, P<0.01 >.001 and <.001 respectively

49 27 Table 2.2: Best Linear Unbiased Estimates (BLUEs) for beer sensory descriptors based on replicated beers (Golden Promise; Full Pint; Rahr Pils; Miller High Life). Entry Cereal Floral Fruit Grass Honey Malt Sweet Toasted Toffee Golden Promise Full Pint Rahr Pils Miller High Life LSD (0.05)

50 Corvallis 28 Table 2.3a: Best Linear Unbiased Predictors (BLUPs) for beer sensory descriptors based on un-replicated beers: the Oregon Promise subset doubled haploid lines, Golden Promise, Full Pint and CDC Copeland grown in Corvallis, OR (Continued). Entry Env Cereal Floral Fruit Grass Honey Malt Sweet Toasted Toffee Golden Promise Full Pint CDC Copeland OPS Min OPS Max OPS AVG St. Error h LSD (0.05)

51 Lebanon 29 Table 2.4b: Best Linear Unbiased Predictors (BLUPs) for beer sensory descriptors based on un-replicated beers: the Oregon Promise subset doubled haploid lines, Golden Promise, Full Pint and CDC Copeland grown in Lebanon, OR (Continued). Entry Env Cereal Floral Fruit Grass Honey Malt Sweet Toasted Toffee Golden Promise Full Pint CDC Copeland OPS Min OPS Max OPS AVG St. Error h LSD (0.05)

52 Madras 30 Table 2.5c: Best Linear Unbiased Predictors (BLUPs) for beer sensory descriptors based on un-replicated beers: the Oregon Promise subset doubled haploid lines, Golden Promise, Full Pint and CDC Copeland grown in Madras, OR (Continued). Entry Env Cereal Floral Fruit Grass Honey Malt Sweet Toasted Toffee Golden Promise Full Pint CDC Copeland OPS Min OPS Max OPS AVG St. Error h LSD (0.05)

53 31 Table 2.4: Averaged Best Linear Unbiased Predictors (BLUPs) for beer flavor sensory descriptors (Rahr Malting) and free choice descriptors (New Glarus Brewing) on selected doubled haploid lines from the Oregon Promise subset compared to Golden Promise and Full Pint. Entry Cereal Floral Fruit Grass Honey Malt Sweet Toasted Toffee Free-Choice Descriptions Malty, Sweet Bitter, Bland, Clean, Hoppy, Malty American, Balanced, Crisp, Sweet, Thin Nuetral Golden Promise Clean, Grainy, Harsh bitterness, Lean, Sweet Full Pint European, Full, Grainy, Malty, Sulfur LSD (0.05) N/A

54 32 Figures Figure 2.1 (a-d): Principle component analysis of flavors consistently significant across three production environments in Oregon, USA (COR, LEB, and MAD). PCA-A is the combined environment Biplot where COR=circle, LEB=cross, and MAD=diamond, PCA-B is the Corvallis environment, PCA-C is the Lebanon environment, and PCA-D is the Madras environment. Parents and check are labelled: Golden Promise = G; Full Pint = F; CDC Copeland = C.

55 33 1Ha 2H 5H Sweet (A) Toffee (B) Toffee (A) 75 Floral (A) Floral (C) cm Fruit (A) 150 1Hb 175 Honey (A) Toffee (B) Toffee (C) 225 Toasted (C) 25 cm cm cm Malt (B) Honey (B) Figure 2.2: Relative chromosomal positions of beer flavor quantitative trait loci (QTL) and origins of favorable alleles (Golden Promise = Left of chromosome; Full Pint = right of chromosome) detected in the 2015 Oregon Promise subset (OPS) across three environments in Oregon, USA (Corvallis = A; Lebanon = B; Madras = C).

56 34 Chapter 3: The Oregon Promise: Effect of malt modification on barley (Hordeum vulgare L.) contributions to beer flavor 1 Herb, D.W., 2 Meints, B.M., 3 Jennings, R., 4 Romagosa, I., 5 Moscou, M., Yueshu, L., Benson, A., 6 Carey, D., Cistue, L., 1 Filichkin, T, 1 Fisk, S.P., 1 Helgerson, L., 3 Monsour, R., Nuygen, A., Onio, A., 6 Thiel, R., 3 Tynan, S.P., 9 Thomas, W.B., Vega, V., Vinkemeier, K., 1 Hayes, P.M. 1 Crop & Soil Science Dept., Oregon State University, Corvallis OR USA 1 Crop & Soil Science Dept., Oregon State University, Corvallis OR USA; 2 Dept. of Crop & Soil Science, Washington State University, Mt. Vernon, WA USA; 3 Rahr Malting Co., Shakopee, MN USA; 4 Agrotecnio, University of Lleida, Lleida, Spain; 5 The Sainsbury Laboratory, Norwich Research Park, Norwich NR4 7UH UK; 6 New Glarus Brewing Co., New Glarus WI USA; 7 Estación Experimental Aula Dei, CSIC, Zaragoza, Spain; 9 The James Hutton Institute, Invergowrie, Dundee DD2 5DA, Scotland UK Abstract The goal of this research was to assess the relative importance of barley genotype vs. degree of malt modification on beer flavor. Grain from a subset of the Oregon Promise population (Golden Promise x Full Pint) was stored for three post-harvest intervals (1, 5, and 10 months). This grain was micro-malted and nano-beers were brewed and assessed for sensory descriptors. In a second experiment, designed to assess effects of modification using larger grain, malt and beer amounts, two varieties (Full Pint and CDC Copeland) were pilot-malted to three levels of modification (under-, well-, and over-modified). Pilot brews were made from these malts and assessed for flavor differences by three independent sensory panels. Overall, the barley genotype was a greater contributor to the significance of differences in beer flavor descriptors than malt treatment. Flavor differences were more pronounced in nano-brews than in pilot brews. Floral and fruit values were highest in Golden Promise and in under-modified malts and decreased with increasing malt modification whereas malt and toffee flavors were highest in Full Pint and overmodified malts and increased with malt modification. Principal component analyses,

57 35 correlations, heritability estimates, and QTL mapping confirm the genetic basis for malting quality and flavor traits. Key Words Barley, malt, beer, flavor, genetic, modification Introduction Barley is the primary ingredient in beer and provides a source of fermentable sugars, however its role in brewing and distilling is dictated by the malt which it produces. Malt must meet specifications designated by the industry before being commercially produced. For example, in the United States (U.S.) the American Malting Barley Association (AMBA) ( has a set of quality guidelines for breeders to ensure variety malting performance. Similar agencies found in other countries include Barley Australia ( Barley, Malting, and Brewing Research Institution (BMBRI) ( the Canadian Malting Barley Technical Centre (CMBTC) ( and the European Brewing Convention (EBC) ( However, the acceptable ranges for malt quality are ultimately determined by the maltster and the brewer. Malt quality arises from three distinct factors: 1) the genetics of the barley, 2) the production environment, and 3) the malting protocol. Modern barley varieties may differ for key malting quality (MQ) traits. For example, varieties have intrinsic levels of specific traits - e.g. high or low grain protein (Dailey et al. 1988) regardless of how they are malted. Differences between varieties may be derived from various allele combinations at loci controlling MQ traits (Mohammadi et al. 2015). Those genes may interact with the production environment, further accentuating MQ differences (Therrien et al. 1994). An otherwise excellent malting variety may make poor malt when grown in a particular environment due to factors ranging from the obvious (e.g. excessive nitrogen fertilization, preharvesting sprouting, and/or Fusarium infection) to the less obvious (e.g. water sensitivity or induced dormancy). Barley genotypes may respond differently to variations in steeping, germination, and kilning parameters. Therefore, a protocol for a contemporary malting variety

58 36 may not be suitable for a variety developed half a century ago (Gothard et al. 1978; Psota et al. 2010). Malting quality parameters can be grouped by measures of malt modification, malt enzymes, and congress wort traits (ambainc.org/media/amba_pdfs/news/guidelines_for_breeders.pdf). According to these criteria, modification traits include: beta-glucan (BG), fine/coarse (FC) difference, soluble/total protein (S/T), turbidity (T), and viscosity (VC). These traits are relative measures of the cell wall degradation and hydrolysis of starch and protein, and estimate extract availability (Briggs, 1998; Ullrich, 2010). Modification affects the wort-related traits such as soluble protein (SP), free amino nitrogen (FAN), malt extract (ME), and color (C) used by brewers to design recipes (Edney et al. 2007). Beta-glucan is a key trait for determining modification as it is the principal fiber of cell walls binding starch granules within the protein matrix (Jamar et al. 2011; Thomas, 2014). If beta-glucan is not properly degraded, the total extract available for brewers is severely diminished (Thomas, 2014). High levels of beta-glucan can result from dormancy, where water uptake and germination is slowed (Fiserova et al. 2015). This can affect malt enzyme-related traits including diastatic power (DP), alpha- amylase (AA) and beta-amylase, proteases, and beta-glucanase. Dormancy can be overcome by prolonged storage of the grain, or lengthening the time spent in the steeping and germination stages of malting (Woonton et al. 2005). However, the latter is challenging in commercial malt production as it leads to downstream logistic issues. Dormancy and water sensitivity are physiological mechanisms within barley that are highly influenced by genotype and environment (Gong et al. 2014). This study is part of ongoing research using a genetics approach to testing the hypothesis that barley can contribute to beer flavor. In a previous report (Herb et al., submitted) we documented significant differences in beer sensory descriptors between two varieties (Golden Promise and Full Pint) and used beer sensory data from their progeny (a subset of the Oregon Promise population) to show that beer sensory attributes have a genetic basis. In this report, we focus on the effects of malt modification on beer sensory descriptors using two approaches: (1) sensory assessment of beers made from malt produced from Oregon Promise parent and progeny grain

59 37 stored for three time periods, and (2) sensory assessment of beers of made from intentionally under-properly-and over-modified malts of Full Pint and CDC Copeland. Material & Methods Experiment I A subset of 34 Oregon Promise population (OPS) doubled haploid lines and the two parents of the population (Golden Promise and Full Pint) were used for this research. The population, the grain production, and details of genetic analysis are reported in Herb et al. (submitted). In this report, we focus on a single production environment (the Klann Family Farm near Madras, Oregon USA), the procedures/analyses used for storing and micro-malting this grain, and the procedures/analyses used for nano-brewing and sensory assessment of the nano-beers made from these micro-malts. The Klann Family Farm is located at N, W. The farm receives 25.4 cm of annual precipitation -therefore barley is grown with supplemental irrigation. The OPS, Golden Promise, Full Pint, and a check (CDC Copeland) were grown in a two-replicate Randomized Complete Block design. The plot size was 9.2 m 2. Plots were seeded at a rate of 250 seeds per m 2 using a Plotseed XL (Wintersteiger AG) grain drill. Plots were harvested at maturity using a Classic Plot (Wintersteiger AG) combine. Grain samples were cleaned using a Sample Cleaner SLN3 (Rationel Kornservice) prior to micro-malting. Micro-malting 250g grain samples were micro-malted using a Joe White Malting unit at Rahr Malting Co. in Shakopee, MN at three different time points. These three time points (hereafter referred to as Storage Treatments (ST) and a numerical suffix indicating the order of grain sampling) were: 1 mo. post-harvest (ST-1), 5 mo. post-harvest (ST-2), and 10 mo. post-harvest (ST-3). The three time points were selected to assess the effects of grain storage on malt modification. All micromalt quality data are shown in Supplementary Tables (Appendices B1-3); selected data are presented in the body of this report. Nano-brewing

60 Malt samples were nano-brewed to a pilsner style with a target specific gravity of and a final volume of 800 ml. The brewing protocol is described in detail in Herb et al. (submitted). A total of 145 beers were brewed for this experiment. One hundred and eleven un-replicated beers were brewed from the 34 OPS samples plus the two parents and the CDC Copeland check from each of the three storage treatments. In addition, malt produced from grain of each of the parents (2014 harvest, Corvallis, OR USA) was used to brew replicated beers of Golden Promise and Full Pint (11 beers each). Twelve replicates of the control beer (Rahr Pils) was also brewed. All beers were brewed over a 12-day period. 38 Sensory Assessments of the nano-beers were based on a ballot consisting of 17 robust flavor descriptors. The sensory assessment of the nano-beers used a comparison-to-reference descriptive analysis, in which each sample was compared to an industry reference beer (Miller High Life) to describe flavors. Each flavor descriptor was ranked based on the magnitude of differences from Miller High Life using a 0-8 scale - where 0-3 is less than the reference, 4 is no difference, and 5-8 is greater than the reference. An augmented design (Lin & Poushinsky, 1985) was used for the sensory assessments. In this design, four beers were tasted at each of the three storage treatment intervals: Golden Promise and Full Pint (both brewed from malt made from 2014 Corvallis, OR grain since there was not sufficient grain available from Madras for replicated brewing); Rahr Pils, and Miller High Life. The Rahr Pils - brewed at the Rahr Malting Co Tech Center - is a light ale made with pilsner base malt and crystal malt and hopped with Amarillo and Cascade hops. These replicated beers provide the variance estimates used to calculate Best Linear Unbiased Estimators (BLUEs). The 34 OPS, Full Pint, Golden Promise and CDC Copeland (all varieties sourced from the Madras experiment) beers from each storage treatment were unreplicated and the value for a given sensory descriptor for each un-replicated experimental unit was adjusted based on the mean of the replicated checks for calculation of Best Linear Unbiased Predictors (BLUPs). Across the three STs, there were 10 sensory sessions during which a total of 151 beers were tasted (111 un-replicated from the OPS, Golden Promise, Full Pint, and CDC Copeland; and 10 replicates of each of the four control beers (Golden Promise, Full Pint, Rahr

61 39 Pils, and Miller High Life). During 9/10 sessions, 15 beers were tasted. In the last session 16 beers were tasted. Statistical analysis Least square means of sensory data for each treatment were calculated with JMP Pro 12 (SAS Institute Inc. 2006) using Best Linear Unbiased Estimates (BLUEs) as an estimation an individual fixed effect means based on the lowest variance within a population. Analysis of variance (ANOVA) was performed using BLUE values and a mixed linear model (MLM) as implemented in JMP Pro 12. Student s t-test was used for mean separation. Best Linear Unbiased Predictors (BLUPs) were calculated for the un-replicated data from each environment to estimate the least square means of the random variables. Narrow-sense heritability estimates were calculated with JMP Pro 12 using the following formula: h 2 = σ 2 G /(σ 2 G + σ 2 2 e ) where σ G 2 represents the genetic variance and σ e the residual variance. Genotyping, linkage map construction, and QTL analysis These processes are described in detail in Herb et al. (submitted). Briefly, there are 206 DNAbased marker loci in the linkage map, which totals 1,311 cm. Each of the 7 chromosomes of barley is represented by a single linkage group, except for chromosome 1, where there is a gap necessitating labelling of this linkage group as 1Ha and 1Hb. Collinearity was observed for all markers relative to the consensus genetic map of barley (Close et al. 2009). QTL analyses were performed for the 17 flavor traits for each treatment using composite interval mapping (CIM) as implemented in Windows QTL Cartographer 2.5 (Wang et al. 2010). A significant (α = 0.05) False Discovery Rate (FDR) likelihood ratio (LR) test threshold for QTL identification were determined with 1,000 permutations and expressed as logarithm of odds (LOD) score. Experiment II Two varieties were used for this experiment (Full Pint and CDC Copeland). Full Pint is a parent of the Oregon Promise population and a check for experiment I. CDC Copeland (WM861-5 / TR118) was developed by Crop Development Centre, University of Saskatchewan and was

62 40 released in This variety was a check in experiment I. The grain for both varieties was sourced from the same trial harvested in 2015 grown at Lebanon, OR USA. The grain of each variety had the same grain protein (11%) and was plump (>95% on a 2.4 mm screen). Malting Treatments & Malt Quality Grain samples were malted at the Canadian Malting Barley Technical Center (CMBTC) in Winnipeg, CA. Each variety was malted three times, with a batch size of 5 kg of cleaned barley. The malting conditions employed were specially designed to generate malts with different degrees of modification (e.g. MT-1 = under-, MT-2 = well-, and MT-3 = over-modified). The malting conditions are described in detail in Supplementary Table (Appendix B4). Quality analyses were performed by the CMBTC on barley (water sensitivity, dormancy/germination capacity and energy, protein, plump, and moisture) and malt (fine- and coarse-extract, F/C difference, soluble protein, total protein, Kolbach Index, free amino nitrogen, alpha-amylase, diastatic power, wort color, beta-glucan, wort ph, viscosity, and friability) following ASBC procedures (ASBC: Barley 3, Barley 5, Malt 3, Malt 4, Malt 6, Malt 7, Malt 12, Wort 8, Wort 9, Wort 10, Wort 12, Wort 13, Wort 18). Brewing Parameters The six malt samples were each brewed to an all-malt ale specification. Each beer was brewed with 1.5 kg of malt and had a water: malt ratio of 3.75:1 during mashing. The mash profile was as follows: 1) 48 C for 30 minutes, 2) raise 1.5 C per minute to 65 C for 30 minutes, and 3) raise 1.5 C per minute to 77 C for 1 minute. The mash was transferred for lautering then sat for 5 minutes allowing the grain bed to settle before vorlaufing for 1 minute until the wort ran clear. Lautering rakes ran at slow speeds while sparging 6 L of water to wash out residual sugars. The wort was boiled for 60 minutes then adjusted to 12 P with a water addition then chilled to 20 C and pitched with Lallamand Nottingham Ale yeast and fermented at 14 C until fully attenuated. The beer was then lagered at 1 C for 7 days then bottled using a HDP counter pressure bottle filler. The bottles were then packaged and used on site for sensory analysis or shipped to cooperating breweries (Deschutes Brewing and New Glarus Brewing Co.) for sensory analysis.

63 41 Sensory The bottled beers were evaluated for flavor at the CMBTC, Deschutes Brewing in Bend, OR, and the New Glarus Brewing Co. in New Glarus, WI. Each sensory assessment utilized a trained panel varying in size and tasting unit: At the CMBTC, 12 panelists tasted 60 ml beers samples; Deschutes Brewing used 6 panelists tasting 120 ml beer samples and New Glarus had 3 panelists tasting 240 ml beer samples. The sensory assessment of the six beer samples used a comparisonto-reference descriptive analysis, in which each sample was compared to an industry reference beer (Budweiser) to describe flavors at each of the modification levels. The ballots consisted of the same 20 robust flavor descriptors as experiment I as well as included chocolate-coffee, dimethyl sulfide (DMS), and diacetyl. All beer samples were tasted in a single session at each location. Statistical Analysis Analyses of variance (ANOVA) of the sensory data were performed using a mixed linear model (MLM) implemented in JMP Pro 12 (SAS Institute Inc. 2006). Student s t-test was calculated for mean separation analysis. Principal component analysis was implemented in JMP Pro 12 to assess the distribution of variation among ST and varieties. Results Experiment I: In the principal component (PC) analysis of fifteen malt quality traits over three grain storage treatments, the first two PCs accounted for 66% of the variability (Figure 3.1). Eight loading coefficients (AA, BG, DP, FAN, ME, MC, S/T, and VC) accounted for most of the variation in modification (Table 3.1). PC1 accounted for 52.9% of the variation; the major contributors were AA, DP, FAN, ME, and S/T. PC2 accounted for 13.1% of the variation, with by MC the major contributor (Figure 3.1). The distribution of loading coefficients revealed positive correlations between AA, DP, FAN, and S/T and negative correlations of BG with AA, DP, FAN, ME, and S/T. MC was only correlated with ME. The distribution of grain storage treatment values along

64 42 PC1 shows that malts made from ST-3 grain were better-modified than those made from ST-2, and ST-1 (Figure 3.1). Malts made from ST-1 grain had the highest BG and VC levels and both parameters were lowest at ST-3. AA, DP, FAN, ME, and S/T all increased from ST-1 to ST-3. These results suggest that there was likely some level of dormancy and/or water sensitivity in the grain and that this was overcome with extended grain storage. Because the standard measures of grain suitability for malting (e.g. germinative energy, germinative capacity, and water sensitivity) were not taken on these samples it is not possible to determine the cause of change in capacity for modification over time. In general, these micro-malts could be considered under-modified compared to commercial criteria. Across treatments, these micro-malts represent a tremendous range of malt modification, which is advantageous for assessing the effects of malt modification on beer flavor. OPS Beer Flavor The relationships between 17 flavor descriptors were assessed by PC analysis. The first three PCs accounted for 66.5% of the variability. Eleven of the seventeen coefficients (body, chemical, color, floral, fruit, grass, honey, malt, sweet, toasted, and toffee) contributed most of the variation observed for flavor. Therefore, the remainder of this report will focus on these 11 descriptors. PC1 accounted for 32.8% of the variation and the principal contributors were body, honey, malt, sweet, toasted, and toffee. PC2 accounted for 15.6% of the variation: predominant contributors were color, fruit, and floral (Figure 3.2). PC3 accounted for 11.8% of the variation and the primary contributors were chemical and grass (Table 3.1). The distribution of loading coefficients revealed positive correlations among malt, toffee, honey, toasted, and sweet, and between fruit and floral coefficients. Negative correlations were detected between fruit and floral with malt, toasted, toffee, honey, and chemical. Clustering of flavor descriptors across grain storage treatments was not as distinct as it was for malt modification traits. However, the distributions of flavor descriptors for genotypes along PC1 shows that the ST-3 malts made were

65 43 generally higher in honey, malt, sweet, toasted, and toffee flavors than ST-2 or ST-1 malts (Figure 3.2). Grain storage treatment was not a consideration for the replicated sensory check beers (Miller High Life, Rahr Pils, Golden Promise, and Full Pint). Therefore, it is not possible to assess the effects of modification on flavor with these beers. The sensory data from these beers does, however, have important implications for the effects of modification on beer flavor in the unreplicated beers (BLUPs), as discussed in the next section. In the ANOVA of the replicated sensory checks (BLUEs)for the eleven flavor descriptors (Table 3.2), the Genotype term was the most consistent source of variation with significance for 10 descriptors (Table 3.3). The Rahr Pils - made with specialty malts and hopped - was significantly higher for color, honey, malt, sweet, toasted, and toffee than Golden Promise, Full Pint, and Miller Higher Life. Rahr Pils was significantly higher for body than Golden Promise and Full Pint but not Miller High Life. Golden Promise was significantly higher for floral, fruit, and grass than Full Pint or Miller High Life, and higher than Rahr Pils, but not significantly so. Miller High Life was relatively neutral (with an average of 3.91 across descriptors) and consistent (with a standard error of 0.3) across all flavors. There was no significant difference between sensory checks for chemical or grass (Table 3.3). Comparing the parental genotypes, Golden Promise had significantly higher values for floral and fruit, whereas Full Pint was significantly higher for honey, malt, sweet, toasted, and toffee descriptors. Key results from the ANOVA of the un-replicated OPS (BLUPs) for the eleven flavor descriptors identified in the PC analysis are shown in Tables 3.4 and 3.5. The significance of main effects and interactions thereof varied across flavor descriptors. Therefore, the descriptors without significant genotype x storage treatment interaction are addressed as grain storage treatment main effects (genotypes averaged for each grain storage treatment) and genotype main effects (grain storage treatments averaged for genotypes). Descriptors with significant interactions are addressed separately for each grain storage treatment. The Treatment term was significant for three flavors descriptors (chemical, floral, and grass) indicating that these flavors

66 were significantly affected by the length of grain storage. Beer made from ST-1 malt were significantly higher for floral and grass than those made from ST-3, whereas beer made from ST- 3 malt were significantly higher for chemical. The Genotype term was significant for all descriptors except grass - indicating significant genetic differences for 10 of the flavor descriptors (body, color, chemical, floral, fruit, honey, malt, sweet, toasted, and toffee) as shown in Supplementary Tables (Appendices B5-7). The parental BLUPs were similar to the parental BLUEs across flavor descriptors. Golden Promise had values of 4.4 and 5.1 for floral and fruit, whereas the OPS ranged from 3.5 to 5.0 and 3.0 to 5.5 respectively. Similarly, Full Pint had values of 4.5 (honey), 5.2 (malt), 4.4 (toasted), 4.5 (toffee), and 4.5 (sweet). The OPS ranged from 3.1 to 4.7 (honey), 3.0 to 5.2 (malt), 3.1 to 4.9 (toasted), 3.0 to 5.0 (toffee), and 3.1 to 5.0 (sweet). The field check CDC Copeland was consistently closest to the reference across flavors descriptors. 44 The Genotype x Treatment interaction term was significant for five of the descriptors (color, chemical, floral, fruit, and sweet) indicating that genotypes responded differentially for these descriptors across treatments. Color increased with modification and was significantly highest in ST-3 compared to ST-1 and ST-2 for Golden Promise, Full Pint, and CDC Copeland, whereas there was no significant difference across the mean of the OPS. Chemical decreased from ST-1 to ST-2 and then increased significantly in ST-3 for Full Pint and CDC Copeland. Floral was significantly higher in ST-1 and decreased with greater malt modification in ST-3 for Golden Promise. This pattern was not observed for Full Pint, CDC Copeland, nor the OPS where there was no significant difference in floral between ST-1 and ST-3. Fruit was significantly higher in ST-1 than in ST-3 for Golden Promise and CDC Copeland there was no significant difference between storage treatments for Full Pint or the OPS. Sweet was significantly higher in ST-3 compared to ST-1 and ST-2 for Golden Promise, Full Pint, and CDC Copeland. The OPS increased in sweet slightly, but not significantly so. There were moderate to high (>0.30) h 2 estimates for malt and toffee for all three storage treatments: malt was highest in ST-1 (51%) and toffee in ST-2 (36%). In contrast, h 2 estimates

67 for chemical and honey were very low (<5%). Other flavor descriptors had h 2 values falling between those for chemical and honey vs. those for malt and toffee. For most traits, there were phenotypic transgressive segregants - progeny with values higher and/or lower than the parents and these differences were most notable for malt, toasted, and toffee flavors in ST Modification & Flavor Correlation Analysis Across storage treatments, correlations between malt modification and beer flavor traits were both positive (AA with malt (0.36), toasted (0.32), sweet (0.26), toffee (0.27); ME with malt (0.33), toffee (0.29), sweet (0.25); S/T-malt (0.52), toffee (0.19), sweet (0.24), color (0.21); BGfruit (0.20), floral (0.07); MC-honey (0.24), color (0.18) and negative correlations (AA-fruit (- 0.11); ME-floral (-0.28); S/T-fruit (-0.29) were detected. Within individual storage treatments, the magnitudes of the correlations were higher in ST-1 and ST-2 compared to ST-3. This may be a result of genes that drive malt modification having a larger effect in under-modified malts compared to over-modified malts. For example, AA and DP are the principal enzymes in starch hydrolysis and therefore shows a positive correlation with malting traits that increase with modification (FAN, ME, MC, and S/T), and negative correlations with those that decrease with modification (BG and VC). However, as modification increased with storage treatment, overall correlations with AA and DP decrease from ST-1 to ST-3. Similar patterns were observed between AA and DP with flavors traits, where AA-malt and AA-sweet decrease from ST-1 (0.36 and 0.38) to ST-3 (0.24 and 0.18). Inversely, positive correlations of BG and VC increased from ST-1 (0.61) to ST-3 (0.83), whereas negative correlations were more consistent. This pattern was also observed in the correlations between BG and flavor traits, where BG-floral and BG-fruit increased from ST-1 (0.09 and 0.05) to ST-3 (0.26 and 0.24), where remaining flavors showed negligible changes in magnitude. These overall low to modest correlations suggest that factors other than degree of modification are driving flavor. Modification & Flavor QTL Analysis Forty-six marker: trait associations (QTLs) were detected for the 7 modification traits and 76 were detected for eleven flavor descriptors using a standard 0.05 FDR and data from each of the

68 three storage treatments. Due to the small population size, however, more stringent thresholds were used to reduce the risk of false positives: LOD 7.5 if the QTL was identified only in one storage treatment and >5.5 if the QTL was detected in more than one storage treatment. These QTLs are the focus in the remainder of the paper and are highlighted in Supplementary Tables (Appendices B8-9) and shown graphically in Figure 3.3 where descriptors to the left of each chromosome cartoon indicate that the higher value allele was contributed by Golden Promise while those to the right indicate that the higher value allele was contributed by Full Pint. Multiple QTLs were detected in all linkage groups. Alpha-amylase on 5H and toasted on 7H had the largest effects and were detected in ST-2 and ST-3, respectively. Full Pint had the larger allele value at each of these QTL. Similar numbers of QTLs were detected in ST-1, ST-2, and ST-3 (22, 26, and 26, respectively). QTLs were more often detected at the same chromosome position for modification-related traits than for flavor traits. The LOD scores for BG, FAN, and ME decreased from ST-1 to ST-3 and AA, DP, MC, and S/T increased. This change in significance corresponds to a change in phenotypic variation between ST-1 and ST-3, indicating that detection of BG, FAN, and ME QTLs is generally greater in under-modified malts and detection of AA, DP, MC, and S/T QTLs increase with malt modification. This pattern of storage treatment effects on QTL significance was not observed for flavor traits. There are comparable numbers of examples of overlap and lack of overlap between modification-related and flavor QTLs. This may be the basis for the modest correlations observed between the two classes of traits and suggests that malt modification is not the sole, or principal, cause of flavor. 46 Experiment II: The relationships between modification and flavor traits were identified using PC analysis of 15 malt quality traits and 20 sensory descriptors. The first two PCs accounted for 77.8% of the variability. Nine modification and 11 flavor traits contributed most of the variation (Table 3.6). Therefore, the remainder of this experiment will focus on these traits. PC1 accounted for 46% of the variation: principal contributors were BG, color, FAN, F/C, floral, FR, fruit, grass, honey, malt, MC, S/T, and VC (Figure 3.4). PC2 accounted for 31.8% of the variation: predominant contributors were AA, chemical, DP, toasted, and toffee (Figure 3.4). Body and sweet explained

69 little of the variation in the first two PCs, however these were greater sources of variability in PC3. Distribution of treatments along PC1 indicate that treatment 1 (under-modified) was higher in BG, F/C, floral, fruit, honey, sweet, and VC whereas treatment 3 (over-modified) had higher values for color, FAN, grass, malty, and MC. Clustering of varieties along PC2 indicates that Full Pint had higher values for AA, DP, toasted, and toffee whereas CDC Copeland had higher values for chemical. 47 Malt Modification The changes to malting protocol - notably changes wet and dry periods duration in steeping - produced malts differing in degree of modification. Based on the CMBTC guidelines, both CDC Copeland and Full Pint malts produced in Treatment 1 were under-modified as evidenced by low FR, low SP, and high BG. Treatment 2 produced better-modified malts based on FR, F/C, and S/T values and a reduction in BG. BG, however, remained slightly higher than the target (Table 3.7). Treatment 3 produced modified to over-modified malt with the highest values for FR, F/C, S/T, and FAN (Table 3.7). The varieties differed in that Full Pint had consistently lower ME than CDC Copeland and higher DP and AA levels (Table 3.8). These results reflect commercial results: CDC Copeland is an industry standard used for making a range of malts destined for a range of beer styles whereas Full Pint is currently a niche variety whose malts may contribute unique flavor profiles to finished beers. The results of the PC analysis of the malt data are presented together with beer sensory data in the next section. Beer Flavor In the combined ANOVA (across breweries) of the 11 sensory descriptors identified in PC analysis, the Malt Treatment term was significant for two descriptors (color and grass), the Variety term was significant for two descriptors (sweet and toffee), and the Malt Treatment x Variety interaction was significant for one descriptors (fruit) - however the Brewery and Panelist terms were the most significant effects and the largest sources of variation (Table 3.9). Therefore, individual ANOVAs (by brewery) of sensory descriptors were performed. The CMBTC dataset had the greatest number of significant terms for the largest number of sensory

70 descriptors. Due to the larger panel size at CMBTC and number of significant effects, we focus on these data in this report and highlight key results from the other two panels. The Malt Treatment term was only significant in the CMBTC dataset for three descriptors (body, color, and malt) indicating that these flavors were significantly affected by malt modification. Body increased with malt modification initially from MT-1 to MT-2 than decreased from MT-2 to MT- 3. Color increased with malt modification from MT-1 to MT-3. Malt increased with malt modification and was significantly higher in MT-3 than MT-1 (Table 3.7). The Variety term was significant for seven descriptors (body, color, floral, fruit, grass, toasted, and toffee) in the CMBTC dataset, and four descriptors (body, color, honey, and toffee) in the DB dataset indicating variation in these flavors between CDC Copeland and Full Pint. CDC Copeland was significantly higher in fruit and grass flavors, whereas Full Pint was significantly higher in color, floral, toasted, and toffee flavors. There was no significant difference between varieties for body or honey (Table 3.8). The Malt Treatment x Variety interaction was only significant in the CMBTC dataset for three flavor descriptors (floral, fruit, and toffee) indicating that these flavors responded differentially across malt treatments. Full Pint was significantly higher for floral in MT-2 and CDC Copeland was significantly higher in MT-3 there was no difference in floral between the two varieties in MT-1. A similar pattern was observed for CDC Copeland and Full Pint for fruit. Full Pint was significantly higher for toffee in all three malt treatments (Table 3.8). Although not included in the 11 flavor descriptors identified by PC analysis, variety effects were detected for astringency, grain, and vegetable in the NGB dataset, vegetable in the DB dataset, and malt treatment and variety effects for astringency, diacetyl, cereal, grain, and vegetable in the CMBTC dataset. 48 Modification & Flavor Correlation Analysis Averaged over treatments, there were both positive and negative correlations notably between modification traits (AA-DP (0.97), DP-FR (0.72), BG-FR (-0.89), BG-MC (-0.85), MC- FR (0.67), S/T-MC (0.77), S/T-FR (0.98), and VC-BG (0.97)), between flavor traits (color-malt (0.98), floral-fruit (0.57), fruit-grass (0.68), honey-sweet (0.91), malt-fruit (-0.74), malt-floral (- 0.73), toasted-chemical (-0.82), toasted-toffee (0.93), toffee-malt (0.67), and toffee-chemical (-

71 0.76)), and between modification and flavor traits (AA-fruit (-0.20), AA-floral (0.07), AA-malt (0.49), AA-toasted (0.82), AA-toffee (0.84), BG-fruit (0.67), BG-floral (0.79), BG-malt (-0.49), BG-color (-0.49), DP-fruit (-0.19), DP-malt (0.40), DP-toasted (0.85), DP-toffee (0.82), FANmalt (0.86), FAN-toasted (0.63), FAN-toffee (0.76), FR-malt (0.19), MC-malt (0.84), and S/Tmalt (0.35)). These correlations correspond to the clustering of loading coefficients in the PC analysis, and validate the relationship specific modification and flavor traits. 49 Discussion We generated malting quality and beer sensory data in two different and complementary experiments designed to answer the question does the degree of malt modification affect beer flavor? The first experiment involved a relatively large sample of experimental samples: 37 different barleys and three grain-storage regimes, leading to 151 beers assessed for 17 sensory descriptors. To assess this number of barley grain samples, micro-malts, nano-beers, we had to employ an augmented design (Lin and Poushinsky, 1983) a novel experimental design for sensory analysis employing a subset of replicated standards and a large number of un-replicated samples. The second experiment used a more typical approach to beer sensory focusing on beers made from two varieties and each of three types of malts: intentionally under-modified, modified, and over-modified, for a total of 6 beers assessed for 20 sensory descriptors. Together, the two experiments provide evidence that barley varieties can differ for their contributions to beer flavor, that there is a genetic basis to barley contributions to beer flavor and that the degree of malt modification can affect sensory descriptor ratings. Each of the experiments provided distinct and additional insights with implications for further research. In Experiment 1 we took a genetic approach to asking the question driving this research. A genetic approach means that rather than assessing a panel of barley varieties, we chose two parents reported in the literature to make significant contributions to beer flavor (Thomas, 2014), and we produced doubled haploid progeny from these two varieties (Hisano et al. 2016). 250g micro-malts were made from these progenies and 800 ml nano-beers were made from these malts. Neither Full Pint or Golden Promise meet current standard malting quality and brewing

72 criteria, and their progeny represent a spectrum of unfavorable and favorable malt quality attributes. There is much more variation for malting quality traits in the progeny of this cross than is typically encountered in a sample of approved malting barley varieties. For example, BG ranged from a low of 112 ppm to a high of 1,220 ppm and ME ranged from 73% to a high of 78%. Despite this variation in malting quality, as described in Herb et al. (submitted) we found significant differences in beer flavor sensory attributes between the parents and between progeny and we demonstrated that barley contributions to beer flavor have a genetic basis via estimation of heritability and exploratory QTL mapping. Excitement over these results was, however, tempered by concerns that observed differences in flavor could be the consequence of a causative difference in malting quality. In other words, could significant differences in specific flavor descriptors, such as fruit or malt simply be a consequence of something as simple high vs. low wort beta glucan? 50 Addressing modification directly in an experiment involving 111 barley samples is not feasible given the design of the micro-malting apparatus. In such a device, it is not possible to tailor malting regimes and thus optimize the degree of modification for every sample. Rather, every sample is treated alike, and a protocol based on a relevant contemporary variety is employed so that merit of every sample can be compared to the standard. In this experiment, the standard was CDC Copeland. Therefore, in order to increase the likelihood of malting all samples during every batch to different levels of modification, we stored the grain and sampled at three intervals: 1 month, 5 months, and 10 months. Even in the absence of obvious dormancy, extended grain storage under proper conditions, can enhance performance in the malt house (Reuss et al. 2003). In this respect, we were successful in increasing the overall degree of modification over time. For example, BG values for Golden Promise, Full Pint, and CDC Copeland were 491, 402, and 253 ppm in ST-1 and 174, 141, 65 ppm in ST-3. Regardless of grain storage interval, however, genotype effects were consistently more significant than grain storage treatment effects. As in our previous report (Herb et al. submitted), in these experiments seven sensory attributes had heritability estimates > than 30%, and exploratory QTL analysis identified regions of the barley

73 genome where there are loci controlling these traits. The validity of the sensory descriptor QTLs cannot be assessed because there are no QTL reports in the literature. There are, however, extensive malting quality QTL reports, most recently reviewed by Mohammadi et al (2015). Nine of the malting quality QTL we found are in the same general regions of the barley genome as in prior reports, thus confirming that, despite the small sample size of 34 DHL, there is validity to the results of the exploratory QTL analysis. 51 The genotype effect on flavor was greater than the treatment effect for body, color, fruit, honey, malt, sweet, toasted, and toffee. The degree of malt modification did affect some flavor descriptors in that the magnitude of significant flavor effects changed, even when the genotype effect was significant at each of the grain storage intervals, for example - chemical and floral. Significant differences in grass were attributable to grain storage, and there were significant genotype x grain storage interactions for a further subset of sensory descriptors. For example, Golden Promise contributed floral and fruit flavors (and the alleles that determine these flavors) whereas Full Pint contributed malt and toasted flavors (and the alleles that determine these flavors). However, the degree of modification can affect the intensity of these flavors, with a general trend towards less floral and fruit and more malty and toasted with increasing degree of modification. The PC analyses and correlations further support the contention that degree of malt modification is associated with sensory descriptors but it is neither causative nor predictive (Altman et al. 2015). There is a precedence for this flavor difference in the literature, where Dong et al. (2013 and 2014) reported 41 aroma compounds consisting of alcohols, aldehydes, ketones, and esters in malting barley, and showed differential response during malting due to individual molecular-weights and volatilization properties and Maillard reactions. A key implication of Experiment 1 is that the exploratory QTL analyses need to be extended to a larger population, and this is indeed underway using micro-malts, nano-brewing, and sensory analysis of nano-brews from the larger Oregon Promise population. Furthermore, genetic data on sensory attributes need to be integrated into the larger picture of metabolomics, as described for tomato (Tieman et al., 2017). This research line is also underway, building on the results of

74 52 Experiment 1. However, a concern remains: could the observed modification and flavor differences be artifacts of micro-malting, nano-brewing, and/or the augmented design? To address this question, we conducted Experiment 2 where we employed larger scale malting and more standardized sensory assessment procedures. To achieve the goals of the experiment with available resources, we limited the experiment to two varieties (Full Pint and CDC Copeland) and three malting regimes. In this experiment, the malting protocol was adjusted to intentionally affect degree of modification, but due to technical limitations, malting regimes could not be developed specifically for each variety. Since the standard malting protocol at the CMBTC is designed for CDC Copeland, the three regimes did achieve the goals of creating under-modified, modified, and over-modified malts of this variety. For Full Pint, however, the three malt regimes generated malts that could be described as very under-modified, undermodified, and modified. As with Experiment 1, optimization of malting quality specifications is not possible for every barley variety. Nonetheless, the modified and over-modified treatments did produce CDC Copeland and Full Pint malts closer to commercial specifications than did the micro-malts. For example, mean BG and ME values were 89 ppm and 81% and 38 ppm and 83% for Full Pint and CDC Copeland, respectively in the CMBTC MT-3. In Experiment 1, the corresponding values were 141 ppm and 78% and 65 ppm and 76% at ST-3. Golden Promise was not included in the CMBTC experiment and therefore its malting quality and sensory descriptors cannot be directly compared with the results of the current study. It is interesting, however, that the variety continues to be highly popular in craft brewing due to flavor attributes and its malting quality profile, as available in a commercial certificate of analysis ( is quite different from that of Full Pint and CDC Copeland. Overall the results of Experiment 2 parallel those of Experiment 1 in that Variety had consistently the most significant effects on sensory attributes, Modification Treatment was significant for a subset of descriptors and the Variety x Modification Treatment interaction was significant for a further subset of descriptors. Experiment 2 further corroborates Experiment

75 1 in that the variety effect on flavor is greater than the treatment effect. Again, Golden Promise contributed floral and fruit flavors, whereas Full Pint contributed malt and toasted flavors. The caveat is that the degree of modification can affect the intensity of these flavors, with a general trend towards less floral and fruity and more malty and toasted with increasing modification. The degree of difference between varieties for the flavor descriptors was not as great as in Experiment 1, underscoring the differences between nano-beers brewed specifically for assessing the contributions of barley to beer flavor vs. a lightly hopped and balanced beer with a higher alcohol level. The PC analysis, and the correlations, in Experiment 2 indicate a much stronger relationship between malt modification and beer flavor descriptors than in Experiment 1. It remains to be seen, in future studies involving pilot malting and brewing of more and different varieties, if the results of Experiment 2 are limited to the two varieties tested, are a consequence of small sample size, and/or if modification will play a larger role in barley contributions to beer flavor in finished beers than in nano-brews. 53 Conclusion In summary, the results of these two experiments, and those of our prior report (Herb et al. submitted) show that barley varieties can contribute to differences in beer flavor and that modification is not the only driver of these differences: there is, in fact, a genetic component. The nature and contributions of this genetic component are the target of ongoing research. More broadly, Full Pint, Golden Promise and their Oregon Promise progeny are a small sample of the total genetic diversity in barley. Based on the results of this study and the inevitable challenges of malt-based research, genetic characterization of, and selection for, barley contributions to beer flavor may be optimized in under-modified malts. Under these conditions, genotypic variation for most flavor descriptors is greatest, whereas the commercial flavor profiling of a variety is best conducted using properly modified malts. A next step is to develop a standardized methodology for assessing contributions to beer flavor in large numbers of barley samples and to apply this methodology to the abundant genetic resources available. For example, the USDA World Core consists of 33,176 accessions (Munoz-Amatrain et al. 2014). Of these, we are targeting sensory analysis of 29 grown in three environments (Oregon, Minnesota, and

76 Saskatchewan). In the final analysis, the contributions of barley variety to beer flavor - be it a contemporary variety, an heirloom variety, or an exotic land race - will likely be modest compared to the effects achieved by manipulating Maillard reactions in malting (Briggs, 1998), using copious amounts of hops with intense aromas, and brewing with different yeasts. Nonetheless, in certain beer styles and for some maltsters, brewers and consumers, the barley contributions to beer flavor will worth pursuing. 54 Acknowledgements We would like to thank Seth Klann (Klann Family Farm) for the field experiment facilities at Madras; Patricia Aaron, and John Andrews, Paul Kramer, and Xiang Yin - Rahr Malting Co. - for critical review of the manuscript; and Richard Goram at the John Innes Centre genotyping facility for KASP genotyping. This research was supported and funded by the Flavor 7-pack of breweries: John Mallett - Bells Brewing, Veronica Vega - Deschutes Brewery, Matthew Brynildson - Firestone-Walker Brewing Co., Dan Carey - New Glarus Brewing Co., Mike Gilford and Vinnie Cilurzo- Russian River Brewing Co., Tom Nielsen - Sierra Nevada Brewing Co., and Damian McConn - Summit Brewing Co, the Gatsby Charitable Foundation, and the Spanish Ministry of Economy and Competitiveness: project AGL C3.

77 55 Tables Table 3.1: Loading coefficients from the principal component analysis of malt modification traits and beer flavor descriptors based on three grain storage intervals and 37 barley genotypes. Malt Modification Traits Beer Flavor Traits PC1 (48.8%) PC2 (15.4%) PC1 (32.8%) PC2 (15.6%) PC3 (11.8%) Trait Loading Coefficient Trait Loading Coefficient Trait Loading Coefficient Trait Loading Coefficient Trait Loading Coefficient AA 0.87 MC 0.45 Body 0.54 Color 0.52 Chemical 0.60 BG VC 0.76 Honey 0.55 Floral 0.64 Grass 0.52 DP 0.65 Malt 0.74 Fruit 0.71 FAN 0.72 Sweet 0.72 ME 0.66 Toasted 0.84 S/T 0.82 Toffee 0.87 Malt Modification Traits: AA= alpha-amylase; BG = beta-glucan; DP = diastatic power; FAN = free amino nitrogen; MC = malt color; ME = malt extract; S/T = Kolbach Index; VC = viscosity

78 56 Table 3.2: Best linear unbiased estimates for beer flavor descriptors in replicated sensory checks. DHL Body Chemical Color Floral Fruit Grass Honey Malt Sweet Toasted Toffee Golden Promise Full Pint Rahr Pils Miller High Life LSD (0.05)

79 57 Table 3.3: Key sources of variation and their p-values from the combined analysis of variance of beer flavor descriptors based on malts made at each of three grain storage intervals (Storage Treatment) from 37 barley genotypes (Genotype). Source Body Chemical Color Floral Fruit Grass Honey Malt Sweet Toasted Toffee Treatment * * *** Genotype *** *** *** *** *** *** *** *** *** *** Treatment x Genotype * *** ** * * NS, *, **, & *** for No Sig Diff, P<0.05 >0.01, P<0.01 >.001 and <.001 respectively

80 58 Table 3.4: (A) Grain storage duration treatment means for malt modification-related traits for Golden Promise, Full Pint, their 34 progeny, and CDC Copeland and (B) beer flavor descriptor BLUPs that had significant treatment effects in the combined ANOVA (Table 3). Source Malt Modification Traits Beer Flavor Traits TRT AA BG DP FAN MC ME S/T VC Chemical Floral Grass ST ST ST LSD (0.05) Source: TRT = storage treatment. Malt Modification Traits: AA= alpha-amylase; BG = beta-glucan; DP = diastatic power; FAN = free amino nitrogen; MC = malt color; ME = malt extract; S/T = Kolbach Index; VC = viscosity

81 ST-1 Table 3.5a: (A) Grain storage treatment (ST-1) values for malt modification-related traits for Full Pint, Golden Promise, their 34 progeny, and CDC Copeland and (B) beer flavor descriptor BLUPs that had significant treatment x genotype interaction in the combined ANOVA. 59 Source Malt Modification Traits Beer Flavor Traits DHL TRT AA BG DP FAN MC ME S/T VC Chemical Color Floral Fruit Sweet Golden Promise Full Pint CDC Copeland OPS Mean OPS MIN OPS MAX LSD (0.05) Source: DHL = Doubled Haploid Line; TRT = storage treatment. Malt Modification Traits: AA= alpha-amylase; BG = beta-glucan; DP = diastatic power; FAN = free amino nitrogen; MC = malt color; ME = malt extract; S/T = Kolbach Index; VC = viscosity

82 ST-2 Table 3.5b: (A) Grain storage treatment (ST-2) values for malt modification-related traits for Full Pint, Golden Promise, their 34 progeny, and CDC Copeland and (B) beer flavor descriptor BLUPs that had significant treatment x genotype interaction in the combined ANOVA (Continued). 60 Source Malt Modification Traits Beer Flavor Traits DHL TRT AA BG DP FAN MC ME S/T VC Chemical Color Floral Fruit Sweet Golden Promise Full Pint CDC Copeland OPS Mean OPS MIN OPS MAX LSD (0.05) Source: DHL = Doubled Haploid Line; TRT = storage treatment. Malt Modification Traits: AA= alpha-amylase; BG = beta-glucan; DP = diastatic power; FAN = free amino nitrogen; MC = malt color; ME = malt extract; S/T = Kolbach Index; VC = viscosity

83 ST-3 Table 3.5c: (A) Grain storage treatment (ST-3) values for malt modification-related traits for Full Pint, Golden Promise, their 34 progeny, and CDC Copeland and (B) beer flavor descriptor BLUPs that had significant treatment x genotype interaction in the combined ANOVA (Continued). 61 Source Malt Modification Traits Beer Flavor Traits DHL TRT AA BG DP FAN MC ME S/T VC Chemical Color Floral Fruit Sweet Golden Promise Full Pint CDC Copeland OPS Mean OPS MIN OPS MAX LSD (0.05) Source: DHL = Doubled Haploid Line; TRT = storage treatment. Malt Modification Traits: AA= alpha-amylase; BG = beta-glucan; DP = diastatic power; FAN = free amino nitrogen; MC = malt color; ME = malt extract; S/T = Kolbach Index; VC = viscosity

84 Table 6: Loading coefficients from the principal component analysis of malt modification traits and beer flavor traits based on three malting treatments and two varieties (CDC Copeland and Full Pint). 62 Trait Malt Modification Traits Beer Flavor Traits PC1 (48.8%) PC2 (15.4%) PC1 (32.8%) PC2 (15.6%) PC3 (11.8%) Loading Coefficient Trait Loading Coefficient Trait Loading Coefficient Trait Loading Coefficient BG AA 0.98 Color 0.74 Chemical Body FAN 0.81 DP 0.94 Floral Toasted 0.90 Sweet 0.63 F/C Fruit Toffee 0.90 FR 0.74 Grass 0.78 MC 0.97 Malt 0.75 S/T 0.83 Honey 0.75 VC Trait Loading Coefficient Malt Modification Traits: AA= alpha-amylase; BG = beta-glucan; DP = diastatic power; FAN = free amino nitrogen; MC = malt color; ME = malt extract; S/T = Kolbach Index; VC = viscosity

85 63 Table 3.7: Malt modification-related trait values for, and Least Square Means for beer flavor descriptors made from, undermodified, modified, and over-modified malts (averaged over the varieties CDC Copeland and Full Pint) where there was a significant treatment effect in the combined ANOVA (Table 7). Source Malt Modification Traits Beer Flavor Traits Treatment AA BG DP FAN F/C FR MC ME S/T VC Body Color Malty MT MT MT LSD (0.05) Source: Malt Modification Traits: AA= alpha-amylase; BG = beta-glucan; DP = diastatic power; FAN = free amino nitrogen; MC = malt color; ME = malt extract; S/T = Kolbach Index; VC = viscosity

86 Table 3.8: Malt modification-related trait values for, and Least Square Means for beer flavor descriptors made from, undermodified, modified, and over-modified malts of CDC Copeland and Full Pint where there was significant treatment x genotype interaction in the combined ANOVA (Table 7). 64 Variety Treatment Malt Modification Traits Beer Flavor Traits AA BG DP FAN F/C FR ME MC S/T VC Floral Fruit Toffee CDC Copeland Full Pint MT LSD (0.05) CDC Copeland Full Pint MT LSD (0.05) CDC Copeland Full Pint MT LSD (0.05) Malt Modification Traits: AA= alpha-amylase; BG = beta-glucan; DP = diastatic power; FAN = free amino nitrogen; MC = malt color; ME = malt extract; S/T = Kolbach Index; VC = viscosity.

87 Table 3.9: Key sources of variation and their p- values from the combined analysis of variance of beer flavor descriptors identified in PC analysis and based on under-modified, modified, and over-modified malts (Malt Treatment) made from CDC Copeland and Full Pint. 65 Source Chemical Color Body Floral Fruit Grass Honey Malt Sweet Toasted Toffee Treatment ** * * Variety ** *** * *** * * * ** Treatment x Variety ** * ** NS, *, **, & *** for No Sig Diff, P<0.05 >0.01, P<0.01 >.001 and <.001 respectively

88 66 Figures Figure 3.1a-b: Principal component analysis of malt modification-related traits based on three grain storage intervals and 37 barley genotypes. A) is the score plot and B) is the loading plots of coefficients.

89 67 Figure 3.2a-b: Principal component analysis of beer flavor sensory descriptors based on malts made from three grain storage intervals and 37 barley genotypes. A) is the score plot and B) is the loading plots of coefficients.

90 68 FAN (2) Chemical (3) 1Ha 1Hb 2H 3H AA (2) 0 0 Toasted 0 DP (3) Body (3) (1) Body Floral 25 DP (2) (1) 50 Ext (2) cm FAN (3) Body AA (1) 100 cm 125 DP (1) Chemical (3) Toasted (3) 125 FAN (2) 150 Color (2) Malt (3) 150 Body (2,3) 175 Sweet (3) 175 Body (2) FAN (3) 225 cm cm Honey (3) BG (3) 4H 5H 6H AA (1,2) 7H Color (2) DP (1,3) ME (3) S/T (1,3) 25 DP (2) Malt (2) FAN (1,3) 50 ME (2) BG (2) DP (1,2) 75 DP (2) Color Color (1) Color 100 ME (1) 100 (2,3) 100 Body (1) (1) ME (2) 125 Malt (2) Toffee (3) AA (2) 150 MC (2) Malt (2) 150 BG (1) 175 Toasted (3) cm 200 Floral cm Toasted (3) 200 Toffee (1,3) FAN (1,2) ME (3) 250 cm S/T (1,2,3) cm AA (1,2,3) BG (1,2,3) MC (1,3) Figure 3.3: Genetic map positions of quantitative trait loci (QTLs) for malt-modification related traits and beer flavor sensory descriptors. Beers were brewed from malts of 34 barley genotypes sampled at three storage intervals. Numbers in parentheses indicate the grain storage interval for which the QTL(s) were detected.

91 69 Figure 3.4a-b: Principal component analysis of malt modification-related traits and beer flavor traits based on three levels of malt modification of the varieties CDC Copeland and Full Pint. A) is the score plot and B) is the loading plots of coefficient.

92 Chapter 4: Support for facultative growth habit barley as a tool for dealing with climate change: results of GWAS of winter survival/low temperature tolerance and vernalization sensitivity (To be submitted to the Journal of Theoretical and Applied Genetics) Dustin Herb 1, Alfonso Cuesta-Marcos 2, Scott Fisk 1, Bill Thomas 3, Patrick Hayes 1 and the barley LTT collaborative 1 Oregon State University, Crop & Soil Science dept. Corvallis, OR 97331, 2 Seminis, Monsanto, Inc., Woodland, CA, 3 James Hutton Institute, Dundee, UK 70 Abstract Low temperature tolerance (LTT), a lack of vernalization (VRN) sensitivity, and sensitivity to short day (PPD) are the determinants of facultative growth habit in barley. Facultative barley could be an important tool in dealing with effects of climate change by providing a crop that can be fall-sown or spring-sown. A diverse panel consisting of 941 accessions from 21 breeding programs was assembled for the extensive phenotyping of winter survival (WS) in 26 fall-sown field experiments and vernalization sensitivity under controlled environment conditions. Based on phenotypic criteria, 73 accessions in the panel are potentially facultative and these comprise 19% of the accessions with the highest WS. A genome-wide association studies (GWAS) approach implemented using these phenotypic data and genome-wide SNP data identified known and novel QTLs and genes associated with LTT and VRN sensitivity. These results establish a foundation for the systematic pyramiding of favorable alleles into facultative and winter growth habit backgrounds. Key Words Barley, facultative, winter, cold, vernalization, photoperiod

93 Introduction To meet the challenges of climate change, there has been increasing interest in the agronomic potential of fall-sown barley in northern latitudes (Fisk et al., 2013). Although fall-sown barley in regions with winter precipitation patterns can have significant yield advantages over - and water use efficiency compared to - spring-sown barley, there is a heightened risk of low temperature-induced crop injury. This risk is problematic at two stages of development: at the vegetative to reproductive transition, which is the focus of this research, and at flowering. For a recent review of low temperature injury at flowering in cereals, please see (Trevaskis et al., 2003; 2006; 2007). Low temperature tolerance (LTT) at the vegetative to reproductive transition stage is an inducible trait that involves a complex gene regulon (reviewed by Cuesta-Marcos et al., 2015) and it is a key component of the mega-phenotype known as winter hardiness. Additional traits contributing to winter hardiness are sensitivity to vernalization (VRN) and sensitivity to short-day photoperiod (sd-ppd) (Hayes et al. 1993). Understanding the genetics and physiology of LTT, VRN, and sd-ppd will allow for efficient improvement of winter hardiness in barley and utilization of the growth habit type referred to as facultative. Facultative growth habit, as defined by von Zitzewitz et al. (2011) is the case of maximum LTT coupled with sd-ppd sensitivity but no VRN sensitivity. The advantage of facultative growth habit in the context of climate change and volatility is that it will give growers and processors maximum flexibility: the same variety can be planted in fall or in the spring. If planted in the fall, the variety is capable of achieving maximum LTT and sd-ppd ensures a timely vegetative to reproductive transition once day-length reaches a critical threshold. If planted in the spring, the inducible LTT regulon is not triggered and thus there is no cost in reproductive fitness to the variety. Likewise, daylength is sufficient that sd-ppd is not invoked. The loss-of-function deletion of VRN-H2 makes this facultative growth habit scenario possible. 71 There is an extensive literature on the association of VRN sensitivity with LTT (reviewed by Cuesta-Marcos et al., 2015). VRN sensitivity will delay the vegetative to reproductive

94 transition and is therefore a potential mechanism for ensuring maximum LTT. However, the problem with VRN sensitivity and climate change is that VRN sensitivity cannot be relied upon to delay the vegetative to reproductive transition: the VRN requirement can be met long before the risk LTT injury is past. In contrast, sd-ppd sensitivity is a much better insurance against precocious transition to reproductive growth because climate change is not altering daylength. There is no known cost to VRN sensitivity under fallsown conditions, and there are reported yield advantages under such conditions whose physiological basis is not clear (Casao et al., 2011). However, if VRN sensitivity is eliminated within an overall genome architecture of maximum LTT and sd-ppd sensitivity, variety development is streamlined and commercial production is facilitated. The breeder can accelerate cycle time; the grower can plant the same variety whenever field and/or market conditions are optimum; and the end-user has the assurance of the same variety and expectations of key quality/processing attributes. There is accumulating evidence that facultative growth habit, as defined, is feasible and viable. Bi-parental (Fisk et al., 2013) and GWAS (von Zitzewitz et al., 2011) studies confirm that maximum LTT is possible with the complete deletion of VRN-H2 and recently Cuesta-Marcos et al. (2015) used near-isogenic lines and a suite of genomics tools to make the same point. A brief review of the key genes discovered in these studies provides essential perspective and framework for the research described in this report. Briefly, QTLs and genes (when known) related to LTT are: FR-H1 (HvBM5a) (Fu et al., 2005; von Zitzewitz et al., 2005; Dhillon et al., 2010); FR-H2 (a cluster of CBF transcription factors) (Skinner et al., 2005, 2006); and FR-H3 (Fisk et al, 2013). The VRN genes are VRN-H1 (HvBM5a) (Danyluk et al., 2003; Trevaskis et al., 2003; Yan et al., 2003; von Zitzewitz et al., 2005; Cockram et al., 2007; Dhillon et al., 2013); VRN-H2 (ZCCT-Ha,b,c) (Yan et al., 2004); and VRN-H3 (HvFt1)(Turner et al., 2005; Yan et al., 2006; Faure et al., 2007). Sd-PPD is determined by PPD-H2 (Pan et al., 1994; Fowler et al., 2001; Faure et al., 2007; Cuesta-Marcos et al., 2008; Kikuchi et al., 2009; 2012; Casao et al., 2011) and PPD-H1 (Karsai et al., 1999; Doktovgrades et al., 2014). These studies cited were based on a relatively narrow sample of barley germplasm. 72

95 73 In order to build on this foundation and determine if there is additional genetic variation for LTT, we developed a large (n = 941) panel of diverse barley germplasm and assessed it for winter survival (WS) under field conditions in a large number of field tests (26 locations over a two-year period). We also characterized this panel for VRN sensitivity under controlled environment conditions in order to identify potential facultative types. Materials & Methods The LTT array is a set of 941 accessions composed of modern varieties, landraces, and advanced generation experimental lines contributed by twenty-one barley breeding and/or genetics programs from around the world. With the goal of excluding spring accessions that would serve only to confirm the value of alleles at the known major LTT QTLs, the criterion for inclusion in the array was the expectation of some degree of LTT. The largest contributors to the panel were (1) Oregon State University and the University of Minnesota (USA) via the Facwin-6 (Belcher et al., 2015) (2) the James Hutton Institute and the University of Dundee (Scotland, UK) via the AGOUEB and ExBarDiv projects (Tondelli, et al., 2014). Phenotyping The array was phenotyped for LTT under field conditions at fourteen locations in (Canada, France, Germany (3 sites), Hungary, Japan, Scotland, Spain, and the USA (5 sites) and twelve locations in (Canada, France, Germany (2 sites), Hungary, Scotland, Spain, and the USA (5 sites). At some locations only a subset of the total panel was assessed. A Type II modified augmented design (Lin and Pouhsinsky, 1983; 1985) was used at each location. In this design, a single replicate of the accessions was distributed among blocks, each containing five replicated checks. The primary check was Alba (winter) and the secondary checks were Maja (facultative), Full Pint (spring), a local winter wheat, and a local barley. Entry values were adjusted based on the relative efficiency of the primary and secondary checks. Each entry was grown in single row, 1

96 m-long plot and LTT was assessed visually as the percentage of winter survival (WS) on a plot basis. The array was phenotyped for VRN sensitivity under greenhouse conditions in 2015 at Oregon State University, Corvallis, Oregon (USA). A single plant of each accession was grown in a six cell-pack, where each plant had a total soil volume of 85 cm 3. The Alba, Maja, and Full Pint checks were each replicated ten times. The greenhouse was maintained at 18 ± 2 C. Natural daylight was supplemented with high intensity lighting to ensure a photoperiod regime of 16 h light/24 h. Days-to-flowering (DTF) was recorded for each plant when the first inflorescence on each plant was 50% emerged from the boot. The trial was ended at 154 days after planting and any accession that had not flowered was assigned a DTF value of 154. Inflorescence type (2-row, 6- row) was recorded for each accession. 74 Genotyping The AGOUEB and ExBarDiv germplasm was genotyped with 7864 gene-based SNPs with a single Illumina TM iselect assay conducted by TraitGenetics GmbH in Gatersleben, Germany (Tondelli et al. 2013). The remaining germplasm was genotyped at the USDA- ARS Small Grains Genotyping Center in Fargo, ND with the Illumina 9K SNP chip (Comadran et al., 2012). After combining all available data and filtering for data quality, there were 6683 SNP loci assayed on each of 941 accessions. The estimated positions of the SNP loci are based on the consensus map of Close et al. (2009). All genotypic and phenotypic data are available from the T3 database ( Analysis Phenotypes were adjusted using two block methods: i) column and row averages of the primary check, or ii) whole plot averages of the primary and secondary checks. The adjustment formula was as follows: a i = X i X, where X i is the block check mean and X is the overall check mean. Based on the relative efficiency (RE) of the adjustment methods, calculated as RE = Intra Whole Plot Error without adjustments ) x 100, the Intra Whole Plot Error with adjustments individual accessions were adjusted as follows: Y = Y i a i, where Y i is the unadjusted

97 75 phenotype. The highest RE was interpreted to indicate the greatest reduction in spatial variation and experimental error. Least square (LS) means of the WS data from each location were calculated with JMP Pro statistical software (Version 12: SAS Institute Inc.). Best Linear Unbiased Estimates (BLUEs) were used to estimate individual fixed effect means based on the lowest variance within the blocks. Analysis of variance (ANOVA) was performed using BLUE values and a mixed linear model (MLM) as implemented in JMP Pro 12. Student s t-test was used for mean separation. Best Linear Unbiased Predictors (BLUPs) were calculated for the individual accessions from each location to estimate the LS means of the random variables. The linear mixed model approach used in the association mapping analysis, including the estimation of multiple levels of relatedness between accessions, was previously described by Yu et al. (2006) and Kang et al. (2008). The vector of phenotypes, y, is modeled as: y = Xβ + Pν + Zu + e, in which X contains the marker data, β is a vector of marker allele effects to be estimated, P contains the population co-variates from PCA within TASSEL 5.0, ν is a vector of subpopulation effects, Z is an identity matrix, u is the random variance due to genome-wide relatedness, and e is the random variance due to error. The phenotypic covariance matrix is assumed to have the following form: var (y) = 2Kσ 2 g + 2σ 2 e, in which K the matrix of kinship coefficients, σ 2 g is the genetic variance from the genome-wide effects, and σ 2 e is the residual variance. Genetic distance analysis and dendrogram generation was performed in TASSEL 5.0 using Archaeopteryx, an evolutionary tree visualization and analysis software, where SNPs across the genome were used to compared the genetic relatedness of the LTT panel. Results In this report we focus on the phenotypic assessments of the full array, as sample size is essential for greater detection power in association mapping studies (Yu et al., 2006;

98 Zhao et al., 2007; Myles et al., 2009). Differential WS data were obtained from 12 of the 26 environments assessed over a two-year period. The greatest magnitudes of differential survival were observed in data from: Idaho (ID), Minnesota (MN), and Ohio (OH) (USA) and Alberta (AB) (Canada) in both years; 2014 Germany (DE); and 2015 Nebraska (NE) (USA), Scotland (UK), and Spain (ES). All other environments had complete WS or minimal differential WS, except 2014 Nebraska where there was zero WS (including the local barley check). 76 There is extensive phenotypic variation for WS in this germplasm (Figure 4.1). The mean performance of the replicated checks across environments was as expected for the winter and spring checks (Alba and Full Pint, respectively). The facultative check (Maja) was significantly lower in WS (44%) than Alba (61%) but significantly higher in WS than Full Pint (3%). Maja had a maximum WS of 80% and was often comparable in WS to Alba and the local check. Low WS at 14AB, 15MN, 15NB, and 15SP led to the overall lower WS for this variety. There was a normal distribution of average WS across the 12 differential environments. Individual phenotypic frequencies are shown in Supplementary Figures (Appendices C2). The top 5% (47 accessions) ranged in WS from 65-84%. Evaluation of the array for days-to-flowering (DTF) under greenhouse conditions (without vernalization) allowed for the assessment of VRN sensitivity. A continuum of phenotypes was observed, ranging from insensitive to sensitive and there were transgressive segregants beyond the ranges of the spring and winter checks (Figure 4.1). Maja was not significantly different from Full Pint in DTF. The frequency distribution of DTF was bimodal, indicating differentiation for vernalization response and clear separation of spring and facultative types from those with winter growth habit. The vernalization insensitive group contained 276 accessions (50 to 105 DTF). The vernalization sensitive group contained 627 accessions (105 to 154 DTF). Within this group, Alba had a DTF value of 130 days, a 68-day difference from Maja. Thirty-five

99 accessions had not flowered when the experiment was terminated at took longer than 154 days. 77 Principal Component Analysis The relationship between WS and environment was assessed using PC analysis. The first four PCs accounted for 54.2% of the variation for WS (Table 4.1 and Figure 4.2). PC1 accounted for 23.6% of the variation and was contributed predominately by 14ID, 14MN, 14OH, 15AB, 15ID, and 15OH. PC2 accounted for 14.8% of the variation and was contributed predominately by 14BL, 15MN, and 15NB. PC3 accounted for 8.5% of the variation and was contributed predominately by 14AB and 15SL. PC4 accounted for 7.3% of the variation and was contributed predominately by 15SP. The clustering of accessions based on WS relative to the checks was as expected (Figure 4.2). The overlay of VRN sensitivity from the greenhouse identifies spring types (low DTF and low WS), potential facultative types (low DTF and high WS) and winter types (high DTF and high WS) (Figure 4.2). The distributions of DTF and WS overlapped in 14AB, 14DE, 14ID, 14MN, 15AB, 15ID, 15OH, 15UK, and 15ES indicate that potentially facultative and winter types had similar levels of WS. There were distinct patterns of DTF and WS association at 14OH, 15MN, and 15NE, where higher WS associated with higher DTF indicating that winter genotypes had greater LTT in these environments. The overlay of spike type indicates no association between inflorescence type (2- row vs. 6-row) and WS (Figure 4.2). Genome-wide association (GWAS) Marker: trait association for WS were explored using GWAS of all 26 environments. However, in this report we focus on the GWAS results from only those environments with differential WS and associations above a FDR threshold of α = These environments were 14AB, 14DE, 14ID, 14MN, 14OH, 15AB, 15ID, 15MN, 15NE, 15OH, 15UK, and 15ES and are shown in Supplementary Figures (Appendices C3). Illustrated in Figure 4.3 is the GWAS results for WS averaged across these environments.

100 Genome scans validated the detection of known LTT QTLs: FR-H1/VRN-H1, FR-H2, FR-H3, VRN-H2, PPD-H1, and PPD-H2. The most significant marker: reported LTT QTL associations were BOPA2_12_30930 with FR-H1; BOPA2_12_30852 with FR-H2 and BOPA2_12_11498 with FR-H3 (Table 4.2). In the separate analysis of each of the 12 environments averaged for Figure 4.3 and Table 4.2, FR-H1 was detected in five environments (14ID, 14MN, 15AB, 15ID, and 15NE), FR-H2 in five environments (14ID, 15ID, 15MN, 15NE, and 15OH) and FR-H3 were detected in three (14ID, 14OH, and 15NE). VRN and sd-ppd sensitivity loci were also determinants of WS. In the GWAS of the combined WS data, BOPA1_11_10610 is at the inferred position of VRN- H2 (Figure 4.3 and Table 4.2). VRN-H2 was significantly associated with WS in four of the 12 environments (14ID, 15MN, 15NE, and 15OH). There were no significant associations at the predicted position of VRN-H3 (Figure 4.3). A significant association was detected with SCRI_RS_ at the predicted position of PPH-H2 in 14ID (Table 4.2). An additional 27 significant marker: WS associations with no reported LTT-related gene/qtl were detected in the combined analysis and/or in individual environments. 78 The GWAS of DTF revealed highly significant effects for VRN-H1 and VRN-H2 and an almost significant association near the position of VRN-H3 (Figure 4.3 and Table 4.3). There were significant associations of DTF with markers on the long arm of chromosome 7H in a region where no growth habit genes/qtls are reported. Frequency of potential facultative growth habit types Facultative growth habit was defined by von Zitzewitz et al. (2011) as a genome-wide haplotype comprised of the deletion of VRN-H2, winter alleles at FR-H1, FR-H2, FR-H3, and the short-day sensitive allele at PPD-H2. Phenotypically, facultative types in this panel should have a minimum average WS of 50% and a DTF value < 95 days. Seventythree accessions meet these criteria and shown in Supplementary Tables (Appendices C1). Within the top 5% for WS, there are six accessions meeting these criteria (Table 4.4

101 79 and 4.5). Currently there are no phenotypic data on short-day sensitivity, or allelespecific genotypes for PPD-H2 in these accessions. Discussion Phenotyping LTT poses challenges. Controlled environment assays may not reflect the complex and interacting environmental signals that, together with plant genotype, determine winter survival. Furthermore, assessment of large numbers of accessions may not be economically feasible. Field trials, on the other hand, will often result in no differential injury and can be subject to within field-variation. Within-field variation can, to some extent, be addressed by use of the appropriate experimental design. Predicting what geographic locations are the most likely to provide differential WS can be to some extent predicted based on historical data, but the effects of climate change are making selection of optimum test environments increasingly problematic. In this project, we were fortunate to have differential WS in 12 out of the 26 environments. All of which were useful for GWAS. Forty-seven accessions were in the top 5% in terms of average WS (Table 4.4). These trace to multiple breeding programs/germplasm collections. The top accession, MO_B475, was developed by J. Poehlman at the University of Missouri (USA) in the 1950s and features historical Midwestern land races in its pedigree: Admire and Missouri Early Beardless. Nine of the top accessions were landraces and originated from Armenia (Leninakanskij), the Czech Republic (PI ), Russia (PI , PI , PI , PI , and PI ), and Korea (PI & MO_B969). The clustering of accessions in the phylogenetic analysis of the top 5% reveals distinct germplasm groups. Accessions from the Oregon State University and University of Minnesota breeding program form a group (top of Figure 4.4 from 06OR-20 clockwise to OBADV11-2). An accession from the Ackermann breeding program (29613/2591) is most closely related to Ayana (UK), Diadem (UK) and Jet (UK). MO_B75 and its parent Admire form a clade with Cetin (Italy) and Herefordia (UK). There were two accessions

102 identified as Admire in the panel: these are genetically distinct and are identified as Admire-1 and Admire-2. They have contrasting alleles for SNPs predominately on chromosome 1H. The next clade includes germplasm from multiple sources. This is followed by a clade including germplasm from the University of Nebraska, a MO line from the same Missouri program that developed MO_B475, five Plant Introduction (PI) accessions from the USDA Barley Core Collection, and the aforementioned Leninakanskij. This is followed by a large cluster of 17 related accessions from the University of Nebraska breeding program and one accession from Texas (USA). Overall, it is encouraging that germplasm of diverse pedigree and origin had the best WS over environments and provides evidence that exotic, un-adapted material may serve as a new source of favorable alleles for LTT. Fortuitously, some of these alleles may already be available in the more recent germplasm/varieties. Of the top accessions six are facultative. The Oregon State University the University of Minnesota programs has specifically targeted this growth habit: in other cases, the presence of facultative types may not be recognized in programs/germplasm collections that only fall-plant germplasm that is expected to have LTT. 80 Determinants of Low Temperature Tolerance GWAS analysis of combined WS data validated the effects of known LTT QTL and the large sample size (n=941) of the array allowed for the clear separation of FR loci (FR-H1, FR-H2, and FR-H3). FR-H1 co-segregates with VRN-H1 and this QTL/gene has been reported as the most significant determinant of LTT (von Zitzewitz et al., 2011). There are diagnostic SNPs in this gene (e.g. BOPA2_12_30930). However, SNPs with higher significance than those in predicted functional domains of HvBM5a have been reported (von Zitzewitz et al. 2005; Szucs et al., 2007; Dhillon et al., 2010) and were found in this research. For example, BOPA1_11_10783 is annotated as a protein linker H1 and H5 family protein. These proteins have been reported to be down-regulated in rice after a period of cold stress (Neilson et al., 2011). Cuesta-Marcos et al. (2015) also reported the presence of putatively LTT-related genes in the HvBM5a region and hypothesized that

103 they could be members of a super-gene complex associated with LTT. Since a criterion for inclusion in the panel was winter hardiness, we had not expected to see significant diversity in FR-H1 region. This could lend support to the super-gene argument. However, it is possible that significant marker: trait associations are simply a consequence of LD with HvBM5a and the unanticipated inclusion of accessions with unexpectedly low level of WS. The most significant SNP in association with FR-H2, BOPA2_12_30852 is annotated as a CBF, and FR-H2 is reported to be due to a cluster of CBF genes (Francia et al., 2007; Stockinger et al., 2007; Knox et al., 2010). FR-H3 was mapped as a QTL to a large interval on 1HS using bi-parental mapping populations where one parent was an accession from the University of Nebraska program (Fisk et al. 2013). In this GWAS, there were three significant SNPs at the same position as the peak of the FRH-3 QTL on the consensus map. One of these, BOPA2_12_10314, is annotated as a glutamate-cysteine ligase, chloroplast precursor. Intriguingly, this class of proteins is reported to affect the rate of protein import into the chloroplast at low temperatures and is mediated by photoperiod conditions (Leheny et al. 1994; Dutta et al. 2009). 81 Although the most significant LTT association involve SNPs and FRH loci, there were four significant markers that coincided with known genes and/or QTLs related to the timing of vegetative-to-reproductive transition. The role of VRN-H2 (on 4HL) has been extensively studied in the context of LTT and growth habit (Yan et al., 2004). Allelic variation at VRN-H2 is due to a deletion and therefore not directly accessible via the Illumina 9K platform. SNP 11_10610 is approximately 3 cm from the predicted position of VRN-H2, and would therefore be expected to be in LD with the ZCCTa,b, and c family of genes, one or more of which is responsible for VRN-H2. The significance of VRN-H2 can be explained by the fact that most accessions with a functional copy of the gene had high WS, whereas those with the deletion included accessions with high and low WS. In the same region, upstream for VRN-H2, we found a significant association with BOPA2_12_10271, annotated as an SNF2P protein with a role in transcriptional regulation. SNF2P proteins are reported to be linked, but not responsible for VRN-H2

104 (Yan et al., 2004; Karsai et al., 2005; Nitcher et al., 2013). VRN-H3 was a major driver of LTT in the near-isogenic lines assayed by Cuesta-Marcos et al. (2015). However, allelic variation at this locus is reported to be limited in most cultivated barley germplasm and that may well be the case in this germplasm array since no significant associations of SNPs with WS were found in the region of VRN-H3. 82 The discovery of 27 novel maker associations across the genome offers the prospects for deeper insights into the genetics of LTT (Table 4.2). Annotation of significant markers were determined using the T3 and BarleyMap databases for alignment within five cm of the consensus map (Close et al., 2016; Blake et al., 2015). While potentially biologically meaningful annotations of significant SNPs are not grounds for candidacy, it is interesting to note the frequency of annotations relating to abiotic stress resistance (Table 4.2). For example, SNP 12_11353 is annotated as a FK506 binding protein, a class of cold-shock proteins with a role in adaption to cold temperatures (Budiman et al., 2011). SNP 11_10383 is annotated as a member of the SAUR protein family comprised of auxin-related genes which are down-regulated in response to cold temperatures (Ren et al., 2015). SNPs 12_30540 and 11_10943 are annotated RING Finger and PIP2 proteins, respectively with roles in responses to abiotic stresses including cold and drought (Guerra et al., 2012; Mori et al., 2014; Alavilli et al., 2016). SNP 12_31402 is annotated as a SPATULA protein, which controls the growth response under cool temperatures during (Heisler et al., 2001; Asakura et al., 2008; Chaudhary et al., 2015). These annotations provide leads for further genetic dissection and assessment of gene candidacy. They also provide immediate targets for marker assisted selection for LTT. Determinants of Flowering Time The DTF data under unvernalized conditions confirms the roles of VRN-H1 and VRN-H2 in this trait. Both loci were detected, with VRN-H2 having the larger effect. This is due to the presence of a spring allele (deletion) at VRN-H2 in the spring and facultative accessions in the panel and they delayed flowering time, or failure to flower, of

105 83 vernalization sensitive accessions. One novel DTF QTL was detected 11_21178 (Table 4.3), which is annotated as a NAM protein involved in the development of apical meristems, floral organs, shoots, and biotic defense mechanisms (Aida et al., 1997). Facultative growth habit and climate change Facultative growth habit is determined by a VRN-H2 deletion and winter alleles at FR- H1, FR-H2, and FR-H3 and PPD-H2. In this study, multiple lines of evidence support the contention that VRN sensitivity is not a requirement for maximum LTT (Table 4.5). The only intentionally bred facultative types were those from the Oregon State University and University of Minnesota programs. In general, breeding program targeting fall-sown plantings at risk of low temperatures never plant under springs-sown conditions. Therefore, there is no opportunity to distinguish between winter and facultative growth habit. The potentially facultative accessions that make up 7% of the panel and that comprise 13% of the top 47 for LTT have yet to be assayed for sd-ppd sensitivity at the genotypic, and ideally, phenotypic levels. Deeper analysis is warranted to relate PPD-H2 allele state with WS in those environments where there was a clear demarcation in WS between the potentially facultative and the VRN-sensitive groups. Given the apparent novel alleles for LTT in the germplasm array, and the advantage of facultative growth habit for producers and end-users, it appears warranted to pursue accumulation of favorable LTT alleles in a VRN-H2 deletion background in order to assist in dealing with the challenges of climate change. Manipulation of other alleles related to growth habit could further tailor varieties to specific environments. For example, the long-day insensitivity allele at PPD-H1 could increase yield in environments with longer growing seasons while the sensitive allele could assist in avoiding heat and water stress and even permit double cropping.

106 84 Conclusion In summary, the GWAS results of this large germplasm panel provide favorable prospects for facultative growth habit and are grounds for optimism that there can be continued progress in improving the LTT of barley. A key to such progress will be collaboration and exchange of genetic resources. Acknowledgements This study was funded by the USDA-NIFA Triticeae Cooperative Agriculture Project (TCAP) project No We would also like to thank the breeding programs who donated germplasm for inclusion in this study and the members of the Barley LTT collaborative who grew and phenotyped the panel - Canada (Flavio Capettini), France (Amelie Genty), Germany (Claus Einfeldt, Markus Herz), Hungary (Ildiko Karsai), Japan (Toshiki Nakamura, Kazuhiro Sato), Scotland (Bill Thomas), Spain (Ernesto Igartua), USA (Stephen Baenziger, Gongshe Hu, Kevin Smith, Eric Stockinger).

107 85 Tables Table 4.1: Loading coefficients from the principal component analysis of winter survival of 941 barley accessions from the average of 12 differential environments. Low Temperature Tolerance PC1 (23.6%) PC2 (14.8%) PC3 (8.48%) PC4 (7.3%) Env* LC** Env LC Env LC Env LC 14ID DE AB ES MN MN UK OH NE ID AB OH 0.50 *Environment = 2014 & 2015 Alberta, Canada (AB), Germany (DE), Spain, (ES), Idaho, USA (ID), Minnesota, USA (MN), Nebraska, USA (NE), Ohio, USA (OH), Scotland (UK); **LC = loading coefficient.

108 Winter Survival 86 Table 4.2: Chromosome position, LOD score, most significant SNP, and candidate gene annotation for quantitative trait loci (QTLs) detected in the GWAS of winter survival (LTT) and days-to-flowering (DTF) without vernalization in 941 barley accessions. The GWAS of WS/LTT is based on the average of 12 differential environments. TRAIT QTL CHR* POS** LOD MARKER GENE ID ANNOTATION 1H BOPA2_12_30919 MLOC Drought induced HYP1 protein FR-H3 1H BOPA2_12_11498 AK Pathogenesis-related transcriptional factor/erf MLOC Ribosome biogenesis regulatory protein FR-H3 1H SCRI_RS_ MLOC Protein splicing factor, arginine/serine-rich FR-H3 1H BOPA2_12_10314 AK Glutamate--cysteine ligase, chloroplast precursor protein 1H BOPA2_12_10938 AK CRS1 / YhbY protein 1H BOPA1_11_10302 MLOC Heat shock protein 1H SCRI_RS_ AK Predicted protein 2H BOPA1_11_10943 AK Aquaporin PIP2 protein 2H BOPA2_12_20368 AK Galactinol synthase 2H SCRI_RS_ AK NUDIX hydrolase protein 2H BOPA2_12_31402 AK SPATULA protein 2H BOPA1_11_10383 MLOC Auxin responsive SAUR protein 2H SCRI_RS_ AK Heat shock protein 2H SCRI_RS_ AK Predicted protein 3H SCRI_RS_67208 MLOC Heat shock protein 3H SCRI_RS_ MLOC Root cap protein 4H BOPA2_12_30540 AK RING finger protein 4H SCRI_RS_ AK Pathogenesis-related transcriptional factor/erf 4H BOPA2_12_20207 N/A 60S ribosomal protein *Chromosome number; **Chromosome position (cm)

109 Winter Survival 87 Table 4.2: Chromosome position, LOD score, most significant SNP, and candidate gene annotation for quantitative trait loci (QTLs) detected in the GWAS of winter survival (LTT). The GWAS of WS/LTT is based on the average of 12 differential environments (Continued). TRAIT QTL CHR* POS** LOD MARKER GENE ID ANNOTATION 4H BOPA2_12_10271 AK SNF2-related protein 4H BOPA1_11_10610 AK Cortical cell-delineating precursor protein 4H BOPA1_11_20668 AK Heat shock protein Chlorophyll a-b binding protein, chloroplast 5H BOPA1_11_10127 AK precursor FR-H2 5H BOPA2_12_30852 AK CBF6 protein FR-H1 5H BOPA1_11_10783 AK H1 and H5 linker protein FR-H1 5H BOPA2_12_30930 AK MADS-box transcription factor 5H SCRI_RS_ AK Pathogenesis-related transcriptional factor/erf 5H SCRI_RS_ AK Heat shock protein 6H BOPA2_12_11353 AK FK506 binding protein MLOC 6H BOPA2_12_ Mlo-related protein 7H BOPA1_11_10983 AK No apical meristem (NAM) protein 7H BOPA2_12_30565 AK HCF106C precursor protein 7H BOPA1_11_21363 AK Ribosome protein *Chromosome number; **Chromosome position (cm)

110 Days-to-Flowering 88 Table 4.3: Chromosome position, LOD score, most significant SNP, and candidate gene annotation for quantitative trait loci (QTLs) detected in the GWAS of days-to-flowering (DTF) without vernalization in 941 barley accessions. The GWAS of WS/LTT is based on the average of 12 differential environments. TRAIT QTL CHR* POS** LOD MARKER GENE ID ANNOTATION 4H BOPA1_11_11513 AK Multi-vascular body protein VRN-H2 4H BOPA1_11_20668 AK Multi-vascular body protein 5H BOPA1_11_11249 AK pnfl-2 protein VRN-H1 5H BOPA1_11_20188 AK MADS-box transcription factor VRN-H1 5H BOPA2_12_30930 AK MADS-box transcription factor 7H BOPA1_11_11222 AK Transcription factor GRAS 7H BOPA1_11_21178 MLOC No apical meristem (NAM) protein 7H SCRI_RS_ MLOC S ribosomal protein *Chromosome number; **Chromosome position (cm)

111 89 Table 4.4 (a): Winter survival (WS) of the top 5% (47 accessions) in the LTT panel across 12 environments with differential winter injury, days-to-flowering (DTF) under greenhouse condition without vernalization, and spike type. Accession ST 14AB* 14DE 14ID 14MN 14OH 15AB 15ID 15MN 15NE 15OH 15UK 15ES DTF MO_B PI PI PI NB NB NB F PI NB Admire Admire MO_B NB Diadem MW09S *Environment: 2014 & 2015 Alberta, Canada (AB), Germany (DE), Spain, (ES), Idaho, USA (ID), Minnesota, USA (MN), Nebraska, USA (NE), Ohio, USA (OH), Scotland (UK).

112 90 Table 4.4 (b): Winter survival (WS) of the top 5% (47 accessions) in the LTT panel across 12 environments with differential winter injury, days-to-flowering (DTF) under greenhouse condition without vernalization, and spike type (Continued). Accession ST 14AB* 14DE 14ID 14MN 14OH 15AB 15ID 15MN 15NE 15OH 15UK 15ES DTF NB NB NB OR NB PI NB F Cetin OBADV P VA08B NB JET NB Leninakanskij *Environment: 2014 & 2015 Alberta, Canada (AB), Germany (DE), Spain, (ES), Idaho, USA (ID), Minnesota, USA (MN), Nebraska, USA (NE), Ohio, USA (OH), Scotland (UK).

113 91 Table 4.4 (c): Winter survival (WS) of the top 5% (47 accessions) in the LTT panel across 12 environments with differential winter injury, days-to-flowering (DTF) under greenhouse condition without vernalization, and spike type (Continued). Accession ST 14AB* 14DE 14ID 14MN 14OH 15AB 15ID 15MN 15NE 15OH 15UK 15ES DTF NB F MW09S NB TAMBAR_ OR NB / McGregor NB Ayana Herefordia NB OBADV F *Environment: 2014 & 2015 Alberta, Canada (AB), Germany (DE), Spain, (ES), Idaho, USA (ID), Minnesota, USA (MN), Nebraska, USA (NE), Ohio, USA (OH), Scotland (UK).

114 Table 4.5: Winter survival (WS) of the ten most winter hardy accessions with potential facultative growth habit across 12 environments with differential winter injury, days-to-flowering (DTF) under greenhouse conditions without vernalization, and spike type. 92 Accession ST 14AB 14DE 14ID 14MN 14OH 15AB 15ID 15MN 15NE 15OH 15UK 15ES DTF MW09S OR OBADV MW09S OR OBADV P Grete OR F Dicktoo Maja Alba Winter Wheat NA *Environment: 2014 & 2015 Alberta, Canada (AB), Germany (DE), Spain, (ES), Idaho, USA (ID), Minnesota, USA (MN), Nebraska, USA (NE), Ohio, USA (OH), Scotland (UK).

115 93 Figures MA AB FP MA AB FP WW Figure 4.1 (a-b): Frequency distributions of average winter survival (WS) across 12 environments where there was differential winter injury in a panel of 941 barley accessions (A) and days-to-flowering (DTF) for the same panel under controlled environment conditions without vernalization.

116 94 AB, MA, LC FP WW AB, MA, LC AB, MA, LC FP FP WW WW Figure 4.2 (a-d): Principal Component Analysis (PCA) of winter survival (WS) of 941 barley accessions in 12 environments where there was differential winter injury. A) distribution of the loading coefficients. The biplots were overlaid with B) average WS scores, where grey < 75% and black is >75% LTT, C) days-to-flowering (DTF) under greenhouse conditions without vernalization, where grey > 96 days and black is < 96 days, and D) spike type, where grey is 6-row and black is 2-row.

117 95 A B Figure 4.3 (a-b): Manhattan plots of LOD scores from GWAS of winter survival (WS) scores averaged across 12 environments where there was differential winter injury (A) and days-to-flowering (DTF) times under greenhouse conditions without vernalization (B) of 941 barley accession.

118 Figure 4.4: Dendrogram of genetic diversity of the 47 accession from the panel of 941 barley accessions with maximum winter survival across 12 environments where there was differential winter injury. 96

119 97 Chapter 5: General Conclusions Overall, the QTL mapping of these studies provide us with the foundation for introgressing favorable alleles for malting quality and flavor into a facultative framework. In chapter 2, we demonstrated that barley variety can make significant contributions to beer flavor and the study provided evidence that these contributions have both a genetic and environmental basis. The heritability estimates ranged from low (indicating a large environmental effect, complex inheritance, and/or un-accounted for sources of variation in measuring the trait) to high (indicative of simple inheritance, small environmental effects, and/or precise trait assessment) across the sensory descriptors, and this suggests that some flavors are more heritable in barley than others. Since these estimates varied across environments, the gains from phenotypic selection for a specific flavor(s) will be greater in environments that contribute less variation compared to those that contribute more to barley terroir. The discovery of 12 QTLs further corroborates a genetic basis for flavor. Both parents provided favorable alleles: Golden Promise contributed alleles to floral, fruit, sweet, and toffee flavors, whereas Full Pint contributed to honey, malt, toasted, and toffee flavors. Since the parents were distinctly different in their flavor profiles and both contributed favorable alleles for toffee, there is a basis for transgressive segregants within the progeny and an opportunity to develop valued-added varieties. The results of this study validates the use of high-throughput small-scale malting and brewing, and an augmented design as effective tools for screening large number of samples for flavor an inevitable dilemma in breeding-based research. In chapter 3, we determined that malt modification is not the driving source of variation for barley contributions to beer flavor and our study further validates the results in chapter 2 indicating a genetic component. In both Experiment I and II, there were

120 contrasting trends for various flavor descriptors across levels of malt modification floral and fruit decreased in more modified malts whereas malt, toasted, and toffee increased and both positive and negative correlations were observed between malt modification and flavor traits. However, these correlations decreased in magnitude and corresponded with a decrease in h 2 estimates with each subsequent level of malt modification. This indicates that the genes driving malt modification are having a larger effect in under-modified malts compared to over-modified malts. 98 In Experiment I, 74 QTLs were detected 46 malt modification QTLs and 28 flavor QTLs across all linkage groups. The distribution and varied overlapping of chromosome positions of these QTLs suggest association between specific traits. The significance of malt modification QTLs, however, varied across modification level, whereas there was no change in the significance for flavor QTLs suggesting that more flavor is independent of malt modification. Therefore, based on the results of this study and the inevitable challenges of malt-based research, genetic characterization of, and selection for, barley contributions to beer flavor may be optimized in under-modified malts where genotypic variation for most flavor descriptors is greatest. In chapter 4, we discovered 18 WS QTLs, including three QTLs near FR loci, two QTLs near VRN loci, two QTLs near PPD loci, and at least 11 QTLs that appear to be novel. GWAS of DTF under greenhouse conditions without vernalization detected seven QTLs including two QTLs near VRN loci and at least five QTLs that appear to be novel. The detection of the novel QTLs within this germplasm suggest that the large sample size of the panel provided the statistical power required to detect both large and small effect QTLs and further corroborates the complexity of LTT. Furthermore, since the top LT tolerant accession included eight landrace varieties, the diverse composition of the panel may serve to identify new sources of LT tolerant alleles for introgression into contemporary germplasm. Additionally, six of the top accessions

121 were facultative indicating that VRN sensitivity is not a requirement for maximum LTT, and that facultative barley has the potential to be as LT tolerant as winter barley. Throughout the panel, 73 accessions meet the requirements for facultative growth habit, therefore barley breeding programs have been indirectly selecting for facultative. However, the inconsistent performance of both winter and facultative varieties across years may be challenging in the most extreme environments indicating that continued breeding efforts are needed to identify and introgress new sources of favorable LTTrelated alleles into a facultative background to meet the demands of an increasingly volatile climate. 99 Our closing conclusions regard these chapters contributions to the ability to develop barley as a value-added crop. Since barley is predominately grown for feed, it is frequently out produced by crops that demand a higher price. However, the potential to develop and produce high quality varieties with new and unique flavors lend the opportunity to expand the marketability of barley and therefore demand a substantial price premium. More importantly, new varieties must be able to adapt to changing climates. Fall-sown barley is in general more sustainable than spring-sown barley as it can exploit winter precipitation patterns and better resist spring disease infection, therefore increasing the profit margins growers by requiring fewer applications of irrigation, nitrogen, and pesticides. Currently, the majority of recommended malting barley varieties are of spring growth habit, thus indicating the importance of developing sustainable winter and facultative malting barley varieties that will be useful in diversifying the prevailing monoculture, providing crop security and flexibility to the grower, and quality that will benefit maltsters and brewers alike.

122 100 Bibliography 1. Agu RC, Palmer GH (1997) The effect of temperature on the modification of sorghum and barley during malting. Process Biochem. 32: Aida M, Ishida T, Fukaki H, Fujisawa H, Tasaka M (1997) Genes Involved in Organ Separation in Arabidopsis: An Analysis of the Cup-Shaped Cotyledon Mutant. THE PLANT CELL ONLINE 9 (6): Alavilli H, Awasthi JP, Rout GR, Sahoo L, Lee BH, Panda SK (2016) Overexpression of a Barley Aquaporin Gene, HvPIP2: Confers Salt and Osmotic Stress Tolerance in Yeast and Plants. Frontiers in Plant Science 7: Altman N, Krzywinski M (2015) Points of Significance: Association, correlation and causation. Nat. Methods 12: American Malting Barley Association (2014) Malting Barley Breeding Guidelines and Ideal Commercial Malt Criteria. Available at: 6. American Society of Brewing Chemists (1992) Methods of Analysis, 8th ed. Barley-3 germination, -5 moisture, Malt- 3 moisture, -4 Extract, -6 Diastatic Power, -7 Alpha-amylase, -12 Malt modification by friability, Wort-8 ph, -9 Wort color and sample preparation, -10 Protein, -12 Free Amino Nitrogen, -13 Viscosity. ASBC, St. Paul, MN 7. Ariyadasa R, Mascher M, Nussbaumer T, Schulte D, Frenkel Z, Poursarebani N, Zhou R, Steuernagel B, Gundlach H, Taudien S, Felder M, Matthias P, Himmelbach A, Schmutzer T, Hedley PE, Muehlbauer GJ, Scholz U, Korol A, Mayer KFX, Waugh R, Langridge P, Graner A, Stein N (2014) A sequence-ready physical map of barley anchored genetically by two million single-nucleotide polymorphisms. Plant Physiol. 164(1): Asakura Y, Bayraktar OA, Barkan A (2008) Two CRM Protein Subfamilies Cooperate in the Splicing of Group IIB Introns in Chloroplasts. RNA (New York, N.Y.) 14(11): Avelino J, Barboza B, Araya JC, Fonseca C, Davrieux F, Guyot B, Cilas C (2005) Effects of slope exposure, altitude and yield on coffee quality in two altitude terroirs of Costa Rica, Orosi and Santa María de Dota. Journal of the Science of Food and Agriculture. 85(11):

123 Bernardo R (2016) Bandwagons I, too, have known. Theor. Appl. Genet. 129: Blake VC, Birkett C, Matthews DE, Hane D, Bradbury P, Jannink JL (2015) The Triticeae Toolbox: combining phenotype and genotype data to advance smallgrains breeding. Plant Genome. Available at: Bodin F, Morlat R (2006) Characterization of Viticultural Terroirs using a Simple Field Model Based on Soil Depth I. Validation of the Water Supply Regime, Phenology and Vine Vigour, in the Anjou Vineyard (France). Plant Soil. 281: Bohmrich R (1996) Terroir: competing perspectives on the roles of soil, climate and people. J. Wine Res. 7: Bowen S, Zapata AZ (2009) Geographical indications, terroir, and socioeconomic and ecological sustainability: The case of tequila. Journal of rural studies 25.1: Bradbury P, Parker T, Hamblin MT, Jannink JL (2011) Assessment of power and false discovery rate in genome-wide association studies using the Barley CAP germplasm. Crop Sci. 51(1): Briggs DE (1998) Malts and Malting. Blackie Academic & Professional, London. pp Budiman C, Koga Y, Takano K, Kanaya S (2011) FK506-Binding Protein 22 from a Psychrophilic Bacterium, a Cold Shock-Inducible Peptidyl Prolyl Isomerase with the Ability to Assist in Protein Folding. International Journal of Molecular Sciences 12(12): Casao MC, Igartua E, Karsai I, Bhat PR, Cuadrado N, Gracia MP, Lasa JM, Casas AM (2011) Introgression of an Intermediate VRNH1 Allele in Barley (Hordeum Vulgare L.) Leads to Reduced Vernalization Requirement without Affecting Freezing Tolerance. Molecular Breeding 28(4):475 84

124 19. Casao MC, Karsai I, Ernesto Igartua E, Gracia MP, Veisz O, Casas AM (2011) Adaptation of Barley to Mild Winters: A Role for PPDH2. BMC Plant Biology 11(1): Chan SC (2012) Terroir and Green Tea in China: The Case of Meijiawu Dragon Well (Longjing) Tea. In: Geographical Indications and International Agricultural Trade. Palgrave Macmillan, London. pp Chaudhary S, Sharma PC (2015) DeepSAGE Based Differential Gene Expression Analysis under Cold and Freeze Stress in Seabuckthorn (Hippophae Rhamnoides L.). PloS One 10(3) 22. Chutimanitsakun Y, Cuesta-Marcos A, Chao S, Corey A, Filichkin T, Fisk S, Kolding M, Meints B, Ong YL, Rey JI, Ross AS, Hayes PM (2013) Application of marker-assisted selection and genome-wide association scanning to the development of winter food barley germplasm resources. Plant Breeding 132: Cistué L, Vallés MP, Echávarri B, Sanz JM, Castillo A (2003) In: Doubled Haploid Production in Crop Plants: A Manual. Maluszynski, M., Kasha, K. J., Forster, B. P. and Szarejko, I., eds. Springer, Netherlands. pp Close TJ, Prasanna R, Bhat R, Lonardi S, Wu Y, Rostoks N, Ramsay L, Druka A, Stein N, Svensson JT, Wanamaker S, Bozdag S, Roose ML, Moscou MJ, Chao S, Varshney RK, Szuecs P, Sato K, Hayes PM, Matthews DE, Kleinhofs A, Muehlbauer GJ, DeYoung J, Marshall DF, Madishetty K, Fenton RD, Condaime P, Graner A, Waugh R (2009) Development and Implementation of High- Throughput SNP Genotyping in Barley. BMC Genomics 10(1): Cockram J, Chiapparino E, Taylor SA, Stamati K, Donini P, Laurie DA, O Sullivan DM (2007) Haplotype Analysis of Vernalization Loci in European Barley Germplasm Reveals Novel VRN-H1 Alleles and a Predominant Winter VRN-H1/VRN-H2 Multi-Locus Haplotype. Theor Appl Genetics 115(7): Comadran J, Kilian B, Russell J, Ramsay L, Stein N, Ganal M, Shaw P (2012) Natural Variation in a Homolog of Antirrhinum CENTRORADIALIS Contributed to Spring Growth Habit and Environmental Adaptation in Cultivated Barley. Nature Genetics 44(12):

125 27. Coulon JB, Priolo A, Durand JL, Emile JC, Huyghe C, Lemaire G (2002) Influence of forage feeding on the composition and organoleptic properties of meat and dairy products: bases for a" terroir" effect. In: Multi-function grasslands: quality forages, animal products and landscapes. Proceedings of the 19th General Meeting of the European Grassland Federation, La Rochelle, France. pp Cuesta-Marcos A, Igartua E, Ciudad FJ, Codesal P, Russell JR, Molina-Cano JL, Moralejo M, Szucs P, Gracia MP, Lasa JM, Casas AM (2008) Heading Date QTL in a Spring Winter Barley Cross Evaluated in Mediterranean Environments. Molecular Breeding 21(4): Cuesta-Marcos A, Szucs P, Close TJ, Filichkin T, Muehlbauer GJ, Smith KP, Hayes PM (2010) Genome-wide SNPs and re-sequencing of growth habit and inflorescence genes in barley: Implications for association mapping in germplasm arrays varying in size and structure. BMC Genomics 11: Cuesta-Marcos A, Muñoz-Amatriaín M, Filichkin T, IKarsai I, Trevaskis B, Yasuda S, Hayes P, Sato K (2015) The Relationships between Development and Low Temperature Tolerance in Barley Near Isogenic Lines Differing for Flowering Behavior. Plant and Cell Physiology 56(12): Danyluk J, Kane NA, Breton G, Limin AE, Fowler DB, Sarhan F (2003) TaVRT- 1, a putative transcription factor associated with vegetative to reproductive transition in cereals. Plant Physiol. 132: Dhillon, T, Pearce SP, Stockinger EJ, Distelfeld A, Li C, Knox AK, Vashegyi I, Vágújfalvi A, Galiba G, Dubcovsky J (2010) Regulation of Freezing Tolerance and Flowering in Temperate Cereals: The VRN-1 Connection. Plant Physiology 153(4) Dhillon, T, Stockinger EJ (2013) Cbf14 Copy Number Variation in the A, B, and D Genomes of Diploid and Polyploid Wheat. Theor Appl Genetics 126(11): Doktorgrades E (2014) The effect of natural genetic variation at Ppd-H1 on the regulation of pre-anthesis development in barley ( Hordeum vulgare L.) in response to the photoperiod.

126 35. Dong L, Piao Y, Zhang C, Hou Y, Shi Z (2013) Analysis of volatile compounds from a malting process using headspace solid-phase micro-extraction and GC- MS. Food research International. 51: Dutta, S, Mohanty S, Tripathy BC (2009) Role of Temperature Stress on Chloroplast Biogenesis and Protein Import in Pea Edney MJ, Eglinton JK, Collins HM, Barr AR, Legge WG, Rossnagel BG (2007) Importance of Endosperm Modification for Malt Wort Fermentability. J. Inst. Brew. 113: Faure S, Higgins J, Turner A, Laurie DA (2007) The FLOWERING LOCUS T- Like Gene Family in Barley (Hordeum Vulgare). Genetics 176(1): Fišerová H, Hartman I, Prokeš J (2015) The effect of weather and the term of malting on malt quality. Plant Soil Environ. 61: Fisk SP, Cuesta-Marcos A, Cistué L, Russell J, Smith KP, Baenziger S, Bedo Z, Corey A, Filichkin T, Karsai I, Waugh R, Hayes PM (2013) FR-H3: A New QTL to Assist in the Development of Fall-Sown Barley with Superior Low Temperature Tolerance. Theor Appl Genetics 126(2): Fowler DB, Breton G, Limin AE, Mahfoozi S, Sarhan F (2001) Photoperiod and Temperature Interactions Regulate Low-Temperature-Induced Gene Expression in Barley. Plant Physiology 127(4): Francia E, Barabaschi D, Tondelli A, Laidò G, Rizza F, Stanca AM, Busconi M, Fogher C, Stockinger EJ, Pecchioni N (2007) Fine Mapping of a HvCBF Gene Cluster at the Frost Resistance Locus Fr-H2 in Barley. Theor Appl Genetics 115(8): Fu D, Szűcs P, Yan L, Helguera M, Skinner JS, von Zitzewitz J, Hayes PM, Dubcovsky J (2005) Large Deletions within the First Intron in VRN-1 Are Associated with Spring Growth Habit in Barley and Wheat. Molecular Genetics and Genomics 273(1): Gong X, Li C, Zhou M, Bonnardeaux Y, Yan G (2014) Seed dormancy in barley is dictated by genetics, environments and their interactions. Euphytica 197:

127 45. Gothard PG, Jenkins G, Morgan AG (1978) Comparative Malting Performance of Old and New Barley Varieties. J. Inst. Brew. 84: Gubler F, Millar AA, Jacobsen JV (2005) Dormancy release, ABA and preharvest sprouting. Curr. Opin. Plant Biol. 8: Guerra D, Mastrangelo AM, Lopez-Torrejon G, Marzin S, Schweizer P, Stanca AM, del Pozo JC, Cattivelli L, Mazzucotelli E (2012) Identification of a Protein Network Interacting with TdRF1, a Wheat RING Ubiquitin Ligase with a Protective Role against Cellular Dehydration. Plant Physiology 158(2): Guido LF, Curto AF, Boivin P, Benismail N, Gonçalves CR, Barros AA (2007) Correlation of Malt Quality Parameters and Beer Flavor Stability: Multivariate Analysis. J. Agric. Food Chem. 55: Hayes PM, Liu BH, Knapp SJ, Chen F, Jones B, Blake T, Franckowiak J (1993) Quantitative Trait Locus Effects and Environmental Interaction in a Sample of North American Barley Germ Plasm. Theor Appl Genetics 87(3): Heisler MG, Atkinson A, Bylstra YH, Walsh R, Smyth DR (2001) SPATULA, a Gene That Controls Development of Carpel Margin Tissues in Arabidopsis, Encodes a bhlh Protein. Development (Cambridge, England) 128(7): Hirota N, Kuroda H, Takoi K, Kaneko T, Kaneda H, Yoshida I, Takashio M, Ito K, Takeda K (2006) Brewing Performance of Malted Lipoxygenase-1 Null Barley and Effect on the Flavor Stability of Beer. Cereal Chem. J. 83: Hisano H, Sato K (2016) Genomic regions responsible for amenability to Agrobacterium-mediated transformation in barley. Sci. Rep. 6: International Barley Genome Sequencing Consortium (2012) A physical, genetic and functional sequence assembly of the barley genome. Nature : Jacobsen JV, Pearce DW, Poole AT, Pharis RP, Mander LN (2002) Abscisic acid, phaseic acid and gibberellin contents associated with dormancy and germination in barley. Physiol. Plant. 115: Jamar C, Jardin P, Fauconnier M (2011) Cell wall polysaccharides hydrolysis of malting barley (Hordeum vulgare L.): a review. Biotechnol. Agron. Société Environnement; Gembloux 15:

128 Jin Z, Mu YW, Sun JY, Li XM, Gao XL, Lu J (2013) Proteome analysis of metabolic proteins (pi 4-7) in barley (Hordeum vulgare) malts and initial application in malt quality discrimination. J. Agric. Food Chem. 61: Jin Z, Cai GL, Li XM, Gao F, Yang JJ, Lu J, Dong J (2014) Comparative proteomic analysis of green malts between barley (Hordeum vulgare) cultivars. Food Chem. 151: Johnson DD, Janzen EL (1999) Barley Production Costs: A Cross-Border Comparison. North Dakota State University, Fargo. Agri. Econ. Report Jones BL (2005) The endogenous endoproteinase inhibitors of barley and malt and their roles in malting and brewing. J. Cereal Sci. 42: Kang HM, Zaitlen NA, Wade CM, Kirby A, Heckerman D, Daly MJ, Eskin E (2008) Efficient control of population structure in model organism association mapping. Genetics 178: Karsai I, Meszaros K, Szucs P, Hayes PM, Lang L, Bed Z(1999) Effects of Loci Determining Photoperiod Sensitivity (Ppd-H1) and Vernalization Response (Sh2) on Agronomic Traits in the Dicktoo X Morex Barley Mapping Population. Plant Breeding 118(5) Karsai I, Szűcs P, Mészáros K, Filichkin T, Hayes PM, Skinner JS, Láng L, Bedő Z (2005) The Vrn-H2 Locus Is a Major Determinant of Flowering Time in a Facultative Winter Growth Habit Barley (Hordeum Vulgare L.) Mapping Population. Theor Appl Genetics 110(8): Kikuchi R, Handa H (2009) Photoperiodic Control of Flowering in Barley. Breeding Science 59(5): Kikuchi R, Kawahigashi H, Ando T, Tonooka T, Handa H (2009) Molecular and Functional Characterization of PEBP Genes in Barley Reveal the Diversification of Their Roles in Flowering. Plant Physiology 149(3): Kikuchi R, Kawahigashi H, Oshima M, Ando T, Handa H. (2012) The Differential Expression of HvCO9, a Member of the CONSTANS-like Gene Family, Contributes to the Control of Flowering under Short-Day Conditions in Barley. Journal of Experimental Botany 63(2): Kirimura J, Shimizu A, Kimizuka A, Ninomiya T, Katsuya N (1969) The Contributions of Pepties and Amino Aicds to the Taste of Foodstuffs. J. Agr. Food Chem. 17(4)

129 67. Kohl S, Behrens M, Dunkel A, Hofmann J, Meyerhof W (2013) Amino Acids and Peptides Activate at Least Five Members of the Human Bitter Taste Receptor Family. J. Agr. Food Chem. 61: Knox AK, Dhillon T, Cheng H, Tondelli A, Pecchioni N, Stockinger EJ. (2010) CBF Gene Copy Number Variation at Frost Resistance-2 Is Associated with Levels of Freezing Tolerance in Temperate-Climate Cereals. Theor App Genetics 121(1): Kontkanen D, Reynolds AG, Cliff MA, King M (2005) Canadian terroir: sensory characterization of Bordeaux-style red wine varieties in the Niagara Peninsula. Food Res. Int. 38: Lawless HT, Heymann H (2010) Sensory Evaluations of Food: Principles and Practice. 2 nd ed. Springer, New York. pp Leheny EA, Teter SA, Theg SM (1998) Identification of a Role for an Azide- Sensitive Factor in the Thylakoid Transport of the 17-Kilodalton Subunit of the Photosynthetic Oxygen-Evolving Complex. Plant Physiology 116(2): Lewis M, Young TW (2001) Brewing. New York City. Aspen Publishers, Inc. 73. Lin CS, Poushinsky G A (1985) Modified augmented design (type 2) for rectangular plots. Canadian Journal of Plant Science. 65: Lin CS, Poushinsky GA (1983) Modified augmented design for an early stage of plant selection involving a large number of test lines without replication. Biometrics 39: Lobell DB, Burke MB, Tebaldi C, Mastrandrea MD, Falcon WP, Naylor RL (2008) Prioritizing climate change adaptation needs for food security in Science 319(5863): Mallet J (2014) Malt: A practical Guide from Field to Brewhouse. Brewers Publication, Colorado. pp Massman J, Cooper B, Horsley R, Neate S, Macky RD, Chao S, Dong Y, Schwarz P, Muehlbauer GJ, Smith K P (2011) Genome-wide association mapping of Fusarium head blight resistance in contemporary barley breeding germplasm. Mol Breeding 7: Mohammadi M, Blake TK, Budde AD, Chao S, Hayes PM, Horsley RD, Obert DE, Ullrich SE, Smith KP (2015) A genome-wide association study of malting 107

130 108 quality across eight U.S. barley breeding programs. Theor. Appl. Genet. 128: Mori IC, Rhee J, Shibasaka M, Sasano S, Kaneko T, Horie T, Katsuhara M (2014) CO2 Transport by PIP2 Aquaporins of Barley. Plant and Cell Physiology 55 (2): Morlat R, Bodin F (2006) Characterization of Viticultural Terroirs using a Simple Field Model Based on Soil Depth -- II. Validation of the Grape Yield and Berry Quality in the Anjou Vineyard (France). Plant Soil. 281: Muñoz-Amatriaín M, Cuesta-Marcos A, Endelman JB, Comadran J, Bonman JM, Bockelman HE, Chao S, Russell J, Waugh R, Hayes PM, Muehlbauer GJ (2014) The USDA Barley Core Collection: Genetic Diversity, Population Structure, and Potential for Genome-Wide Association Studies. PLoS One 9: e Murphy B (2016) Advertising Gastronomic Identity in an Epicurean World: The Case for Irish Single Pot Whiskey. Irish Communication Review. 15.1:7 83. Myles S, Peiffer J, Brown PJ, Ersoz ES, Zhang ZW, Costich DE, Buckler ES (2009) Association mapping: Critical considerations shift from genotyping to experimental design. Plant Cell 21: Neilson KA, Mariani M, Haynes PA (2011) Quantitative Proteomic Analysis of Cold-Responsive Proteins in Rice. PROTEOMICS 11(9): Nitcher R, Distelfeld A, Tan C, Yan L, Dubcovsky J (2013) Increased Copy Number at the HvFT1 Locus Is Associated with Accelerated Flowering Time in Barley. Molecular Genetics and Genomics 288(5-6): Pan A, Hayes PM, Chen F, Chen THH, Blake T, Wright S, Karsai I, Bed Z (1994) Genetic Analysis of the Components of Winterhardiness in Barley (Hordeum Vulgare L.). Theor App Genetics 89-89(7-8): Peltz ML (2015) The Role of Alcohol Content on Sensory Aroma Detection Thresholds in Beer. Oregon State University 88. Rae SJ, Macaulay M, Ramsay L, Leigh F, Matthews D, O Sullivan DM, Donini P, Morris PC, Powell W, Marshall DF, Waugh R, Thomas WTB (2007) Molecular barley breeding. Euphytica 158(3): Ren H, Gray WM, (2015) SAUR Proteins as Effectors of Hormonal and Environmental Signals in Plant Growth. Molecular Plant 8(8):

131 90. Roy JK, Smith KP, Muehlbauer GJ, Chao S, Close TJ, Steffenson BJ (2010) Association mapping of spot blotch resistance in wild barley. Mol Breed 26: Saerens SMG, Verbelen PJ, Vanbeneden N, Thevelein JM, Delvaux FR (2008) Monitoring the influence of high-gravity brewing and fermentation temperature on flavour formation by analysis of gene expression levels in brewing yeast. Appl. Microbiol. Biotechnol. 80: Salette J, Asselin C, Morlat R (1998) The relationship between terroir and product: an analysis of the terroir-vine-wine system and its analogous application to other products. Sci. des Aliment Schlosser J, Reynolds AG, King M, Cliff M (2005) Canadian terroir: sensory characterization of Chardonnay in the Niagara Peninsula. Food Research International. 38.1: Schmitt MR, Skadsen RW, Budde AD (2013) Protein mobilization and maltingspecific proteinase expression during barley germination. J. Cereal Sci. 58: Schwarz P, Li Y (2011) Malting and brewing uses of barley. In: S. Ullrich, editor, Barley: production, improvement, and uses. Blackwell Publishing, Ltd., Chichester, West Sussex, UK. P Sharp D, Qian YP, Clawson JC, Shellhammer TH (2016) An exploratory study toward describing hop aroma in beer made with American and European Hop Cultivars. Brew. Sci. 69: Silva F, Nogueira LC, Goncalves C, Ferreira AA, Ferreira I, Teixeira N (2008) Electrophoretic and HPLC methods for comparative study of the protein fractions of malts, worts and beers produced from Scarlett and Prestige barley (Hordeum vulgare L.) varieties. Food Chem. 106: Simpsons Malt Finest Pale Ale Golden Promise TM. Available at: Skinner JS, Szucs P, von Zitzewitz J, Marquez-Cedillo L, Filichkin T, Stockinger EJ, Thomashow MF, Chen THH, Hayes PM (2006) Mapping of Barley Homologs to Genes That Regulate Low Temperature Tolerance in Arabidopsis. Theor App Genetics 112(5): Skinner JS, von Zitzewitz J, Szucs P, Marquez-Cedillo L, Filichkin T, Amundsen K, Stockinger EJ, Thomashow MF, Chen THH, Hayes PM (2005) 109

132 110 Structural, functional, and phylogenetic characterization of a large CBF gene family in barley. Plant Mol Biol 59: Solms J (1969) The Taste of Amino Acids, Peptides, and Proteins. J. Agr. Food Chem. 17(4): Stockinger EJ, Skinner JS, Gardner KG, Francia E, Pecchioni N (2007) Expression Levels of Barley Cbf Genes at the Frost Resistance - H2 Locus Are Dependent upon Alleles at Fr-H1 and Fr-H2. The Plant Journal 51(2): Szucs P, Skinner JS, Karsai I, Cuesta-Marcos A, Haggard KG, Corey AE, Chen THH, Hayes PM (2007) Validation of the VRN-H2/VRN-H1 Epistatic Model in Barley Reveals That Intron Length Variation in VRN-H1 May Account for a Continuum of Vernalization Sensitivity. Molecular Genetics and Genomics 277(3): Therrien MC, Carmichael CA, Noll JS, Grant CA (1994) Effect of fertilizer management, genotype, and environmental factors on some malting quality characteristics in barley. J. Plant Sci. 74: Thomas D (2014) The Craft Maltsters Handbook. White Mule Pess, California. pp Tieman D, Zhu G, Resende MFR, Lin T, Nguyen C, Bies D, Rambla JL, Beltran KSO, Taylor M, Zhang B, Ikeda H, Liu Z, Fisher J, Zemach I, Monforte A, Zamir D, Granell A, Kirst M, Huang S, Klee H (2017) A chemical genetic roadmap to improved tomato flavor. Science. 355: Tondelli A, Xu X, Moragues M, Sharma R, Schnaithmann F, Ingvardsen C, Manninen O, Comadran J, Russell J, Waugh R, Schulman AH, Pillen K, Rasmussen SK, Kilian B, Cattivelli L, Thomas WTB, Flavell AJ (2013) Structural and temporal variation in genetic diversity of European spring two-row barley cultivars and association mapping of quantitative traits. Plant Genome 6(2) 108. Tondelli A, Pagani D, Ghafoori IN, Rahimi M, Ataei R, Rizza F, Flavell AJ, Cattivelli L (2014) Allelic Variation at Fr-H1/Vrn-H1 and Fr-H2 Loci Is the Main Determinant of Frost Tolerance in Spring Barley. Environmental and Experimental Botany 106: Trevaskis B, Bagnall DJ, Ellis MH, Peacock WJ, Dennis ES (2003) MADS Box Genes Control Vernalization-Induced Flowering in Cereals. Proc Natl Acad Sci USA 100(22):

133 110. Trevaskis B, Hemming MN, Dennis ES, Peacock WJ (2007) The Molecular Basis of Vernalization-Induced Flowering in Cereals. Trends in Plant Science 12(8): Trevaskis B, Hemming MN, Peacock WJ, Dennis ES (2006) HvVRN2 Responds to Daylength, Whereas HvVRN1 Is Regulated by Vernalization and Developmental Status. Plant Physiology 140(4): Turner A, Beales J, Faure S, Dunford RP, Laurie DA (2005) The pseudoresponse regulator Ppd-H1 provieds adaption to photoperiod in barley. Science 310: Ullrich SE, Lee H, Clancy JA, del Blanco IA, Jitkov VA, Kleinhofs A, Han F, Prada D, Romagosa I, Molina-Cano JL (2009) Genetic relationships between preharvest sprouting and dormancy in barley. Euphytica 168: Ullrich SE (2011) Barley: production, improvement, and uses. Wiley- Blackwell 115. Van Leeuwen C, Seguin G (2006) The concept of terroir in viticulture. J. Wine Res. 17: Vollmer D (2016) Factors that influence the sensory characteristics in dryhopped beers. Oregon State University 117. von Zitzewitz J, Cuesta-Marcos A, Condon F, Castro AJ, Chao S, Corey A, Filichkin T, Fisk SP, Gutierrez L, Haggard K, Karsai I, Muehlbauer GJ, Smith KP, Veisz O, Hayes PM (2011) The Genetics of Winterhardiness in Barley: Perspectives from Genome-Wide Association Mapping. Plant Genome J. 4: von Zitzewitz J, Szucs P, Dubcovsky J, Yan LL, Francia E, Pecchioni N, Casas A, Chen THH, Hayes PM, Skinner JS (2005) Molecular and Structural Characterization of Barley Vernalization Genes. Plant Mol. Biol. 59(3): Waugh R, Jannink JL, Muehlbauer GJ, Ramsay L (2009) The emergence of whole genome association scans in barley. Curr Opin Plant Biol 12: Woonton BW, Jacobsen JV, Sherkat F, Stuart IM (2005) Changes in Germination and Malting Quality During Storage of Barley. J. Inst. Brew. 111: Wurschum T, Liu W, Gowda M, Maurer HP, Fischer S, Schechert A, Reif JC 2012 Comparison of biometrical models for joint linkage association mapping. Heredity 108(3),

134 Yan L, Fu D, Li C, Blechl A, Tranquilli G, Bonafede M, Sanchez A, Valarik M, Yasuda S, Dubcovsky J (2006) The Wheat and Barley Vernalization Gene VRN3 Is an Orthologue of FT. Proc Natl Acad Sci USA 103: Yan L, Loukoianov A, Tranquilli G, Helguera M, Fahima T, Dubcovsky J (2003) Positional cloning of the wheat vernalization gene VRN1. Proc. Natl. Acad. Sci. USA 100: Yan L, Louloianov A, Blechl A, Tranquilli G, Ramakrishna W, SanMiguel P, Bennetzen JL, Echenique V, Dubcovsky J (2004) The Wheat VRN2 Gene Is a Flowering Repressor Down-Regulated by Vernalization. Science 303(5664): Yu JM, Pressoir G, Briggs WH, Bi IV, Yamasaki M, Doebley JF, McMullen MD, Gaut BS, Nielsen DM, Holland JB, Kresovich S, Buckler ES (2006) A unified mixed-model method for association mapping that accounts for multiple levels of relatedness. Nat. Genet. 38: Zhang, Z.W., E. Ersoz, C.Q. Lai, R.J. Todhunter, H.K. Tiwari, M.A. Gore, P.J. Bradbury, J.M. Yu, D.K. Arnett, J.M. Ordovas, and E.S. Buckler Mixed linear model approach adapted for genome-wide association studies. Nat. Genet. 42: Zhao KY, Aranzana MJ, Kim S, Lister C, Shindo C, Tang CL, Toomajian C, Zheng HG, Dean C, Marjoram P, Nordborg M (2007) An Arabidopsis example of association mapping in structured samples. PLoS Genet. 3: Zhao FJ, Fortune S, Barbosa VL, McGrath SP, Stobart R, Bilsborrow PE, Booth EJ, Brown A, Robson P (2006) Effects of sulphur on yield and malting quality of barley. J. Cereal Sci. 43:

135 113 Appendix Appendix A Supplementary Table A1.1: Mean values for agronomic traits measured on the Oregon Promise subset (OPS) grown in 2015 Corvallis, OR. Genotype BSR (%) DTF HT (cm) Plump (6/64) Protein (%) TWT (g/l) Yield (Kg/A) LSD (0.05) BSR = Barley Stripe Rust; DTF = Days-to-flowering; HT = plant height; Plump = kernel plumpness; TWT = test weight.

136 Supplementary Table A1.2: Mean values for agronomic traits measured on the Oregon Promise subset (OPS) grown in 2015 Corvallis, OR (Continued). 114 Genotype BSR (%) DTF HT (cm) Plump (6/64) Protein (%) TWT (g/l) Yield (Kg/A) Golden Promise Full Pint CDC Copeland LSD (0.05) BSR = Barley Stripe Rust; DTF = Days-to-flowering; HT = plant height; Plump = kernel plumpness; TWT = test weight.

137 Supplementary Table A2.1: Mean values for agronomic traits measured on the Oregon Promise subset (OPS) grown in 2015 Lebanon, OR. 115 Entry BSR (%) DTF HT (cm) Plump (6/64) Protein (%) TWT (g/l) Yield (Kg/A) LSD (0.05) BSR = Barley Stripe Rust; DTF = Days-to-flowering; HT = plant height; Plump = kernel plumpness; TWT = test weight.

138 Supplementary Table A2.2: Mean values for agronomic traits measured on the Oregon Promise subset (OPS) grown in 2015 Lebanon, OR (Continued). 116 Entry BSR (%) DTF HT (cm) Plump (6/64) Protein (%) TWT (g/l) Yield (Kg/A) Golden Promise Full Pint CDC Copeland LSD (0.05) BSR = Barley Stripe Rust; DTF = Days-to-flowering; HT = plant height; Plump = kernel plumpness; TWT = test weight.

139 Supplementary Table A3.1: Mean values for agronomic traits measured on the Oregon Promise subset (OPS) grown in 2015 Madras, OR. 117 Genotype HT (cm) Lodging (%) Plump (6/64) Protein (%) TWT (g/l) Yield (Kg/A) LSD (0.05) BSR = Barley Stripe Rust; DTF = Days-to-flowering; HT = plant height; Plump = kernel plumpness; TWT = test weight.

140 Supplementary Table A3.2: Mean values for agronomic traits measured on the Oregon Promise subset (OPS) grown in 2015 Madras, OR (Continued). 118 Genotype HT (cm) Lodging (%) Plump (6/64) Protein (%) TWT (g/l) Yield (Kg/A) Golden Promise Full Pint CDC Copeland LSD (0.05) BSR = Barley Stripe Rust; DTF = Days-to-flowering; HT = plant height; Plump = kernel plumpness; TWT = test weight.

141 Supplementary Table A4.1: Malting quality data from the Oregon Promise subset grown in 2015 Corvallis, OR. 119 Genotype AA BG DP FAN MC ME SP S/T TP VC AA: alpha-amylase; BG: beta-glucan; DP: diastatic power; ME: malt extract; SP: soluble protein; S/T: Kolbach Index; TP: total protein; FAN: free amino nitrogen; VC: viscosity

142 Supplementary Table A4.2: Malting quality data from the Oregon Promise subset grown in 2015 Corvallis, OR (Continued). 120 Genotype AA BG DP FAN MC ME SP S/T TP VC Golden Promise Full Pint CDC Copeland AA: alpha-amylase; BG: beta-glucan; DP: diastatic power; ME: malt extract; SP: soluble protein; S/T: Kolbach Index; TP: total protein; FAN: free amino nitrogen; VC: viscosity

143 Supplementary Table A5.1: Malting quality data from the Oregon Promise subset grown in 2015 Lebanon, OR. 121 Genotype AA BG DP FAN MC ME SP S/T TP VC AA: alpha-amylase; BG: beta-glucan; DP: diastatic power; ME: malt extract; SP: soluble protein; S/T: Kolbach Index; TP: total protein; FAN: free amino nitrogen; VC: viscosity

144 Supplementary Table A5.2: Malting quality data from the Oregon Promise subset grown in 2015 Lebanon, OR (Continued). 122 Genotype AA BG DP FAN MC ME SP S/T TP VC Golden Promise Full Pint CDC Copeland AA: alpha-amylase; BG: beta-glucan; DP: diastatic power; ME: malt extract; SP: soluble protein; S/T: Kolbach Index; TP: total protein; FAN: free amino nitrogen; VC: viscosity

145 Supplementary Table A6.1: Malting quality data from the Oregon Promise subset grown in 2015 Madras, OR. 123 Entry AA BG DP FAN MC ME S/T SP TP VC AA: alpha-amylase; BG: beta-glucan; DP: diastatic power; ME: malt extract; SP: soluble protein; S/T: Kolbach Index; TP: total protein; FAN: free amino nitrogen; VC: viscosity

146 Supplementary Table A6.2: Malting quality data from the Oregon Promise subset grown in 2015 Madras, OR (Continued). 124 Entry AA BG DP FAN MC ME S/T SP TP VC Golden Promise Full Pint CDC Copeland AA: alpha-amylase; BG: beta-glucan; DP: diastatic power; ME: malt extract; SP: soluble protein; S/T: Kolbach Index; TP: total protein; FAN: free amino nitrogen; VC: viscosity

147 125 Supplementary Table A7.1: Best linear unbiased predictors (BLUP) for individual doubled haploids in the Oregon Promise subset grown in Corvallis, OR in Genotype BT CH CL FL FR GR GS HN MT RO SF ST TD TF LSD (0.05) BT=bitter; CH=chemical; CL=cereal; FL=floral; FR=fruit; GR=grain; GS=grass; HN=honey; MT=malt; RO=roasted; SF=sulfur; ST=sweet; TD=toasted; TF=toffee

148 126 Supplementary Table A7.2: Best linear unbiased predictors (BLUP) for individual doubled haploids in the Oregon Promise subset grown in Corvallis, OR in 2015 (Continued). Genotype BT CH CL FL FR GR GS HN MT RO SF ST TD TF Golden Promise Full Pint CDC Copeland LSD (0.05) BT=bitter; CH=chemical; CL=cereal; FL=floral; FR=fruit; GR=grain; GS=grass; HN=honey; MT=malt; RO=roasted; SF=sulfur; ST=sweet; TD=toasted; TF=toffee

149 127 Supplementary Table A8.1: Best linear unbiased predictors (BLUP) for individual doubled haploids in the Oregon Promise subset grown in Lebanon, OR in Genotype BT CH CL FL FR GR GS HN MT RO SF ST TD TF LSD (0.05) BT=bitter; CH=chemical; CL=cereal; FL=floral; FR=fruit; GR=grain; GS=grass; HN=honey; MT=malt; RO=roasted; SF=sulfur; ST=sweet; TD=toasted; TF=toffee

150 128 Supplementary Table A8.2: Best linear unbiased predictors (BLUP) for individual doubled haploids in the Oregon Promise subset grown in Lebanon, OR in 2015 (Continued). Genotype BT CH CL FL FR GR GS HN MT RO SF ST TD TF Golden Promise Full Pint CDC Copeland LSD (0.05) BT=bitter; CH=chemical; CL=cereal; FL=floral; FR=fruit; GR=grain; GS=grass; HN=honey; MT=malt; RO=roasted; SF=sulfur; ST=sweet; TD=toasted; TF=toffee

151 129 Supplementary Table A9.1: Best linear unbiased predictors (BLUP) for individual doubled haploids in the Oregon Promise subset grown in Madras, OR in Genotype BT CH CL FL FR GR GS HN MT RO SF ST TD TF LSD (0.05) BT=bitter; CH=chemical; CL=cereal; FL=floral; FR=fruit; GR=grain; GS=grass; HN=honey; MT=malt; RO=roasted; SF=sulfur; ST=sweet; TD=toasted; TF=toffee

152 130 Supplementary Table A9.2: Best linear unbiased predictors (BLUP) for individual doubled haploids in the Oregon Promise subset grown in Madras, OR in 2015 (Continued). Genotype BT CH CL FL FR GR GS HN MT RO SF ST TD TF Golden Promise Full Pint CDC Copeland LSD (0.05) BT=bitter; CH=chemical; CL=cereal; FL=floral; FR=fruit; GR=grain; GS=grass; HN=honey; MT=malt; RO=roasted; SF=sulfur; ST=sweet; TD=toasted; TF=toffee

153 131 Supplementary Table A10: Quantitative traits loci (QTL) results table for the flavor traits measured in the Oregon Promise subset (OPS) grown in three locations (Corvallis, OR; Lebanon, OR; Madras, OR) in Trait Chromosome Peak Position (cm) LOD threshold LOD score A B C A B C A B C A B C Floral 2H 2H Fruit 5H Honey 2H 5H Malt H Sweet 1Ha Toasted 1Hb Toffee 5H 2H, 5H 5H , , Closest Marker Marker Name R2 Additive Effect A B C A B C A B C A B C Floral _1166 2_ Fruit 32 2_ Honey _0183 3_ Malt 15 3_ Sweet 10 2_ Toasted 1 1_ Toffee 23 3, MC_1598 2_0112_120, 2_1141 MC_ , , Additive Effect: Golden Promise (-); Full Pint (+)

154 132 Appendix B Supplementary Table B1.1: Malting quality data from the Oregon Promise subset grain stored for 1 month (ST-1) prior to malting and brewing. Genotype AA BG MC DP ME S/T SP TP FAN VC AA: alpha-amylase; BG: beta-glucan; DP: diastatic power; ME: malt extract; SP: soluble protein; S/T: Kolbach Index; TP: total protein; FAN: free amino nitrogen; VC: viscosity

155 Supplementary Table B1.2: Malting quality data from the Oregon Promise subset grain stored for 1 month (ST-1) prior to malting and brewing (Continued). 133 Genotype AA BG MC DP ME S/T SP TP FAN VC Golden Promise Full Pint CDC Copeland AA: alpha-amylase; BG: beta-glucan; DP: diastatic power; ME: malt extract; SP: soluble protein; S/T: Kolbach Index; TP: total protein; FAN: free amino nitrogen; VC: viscosity

156 Supplementary Table B2.1: Malting quality data from the Oregon Promise subset grain stored for 5 months (ST-2) prior to malting and brewing. 134 Genotype AA BG MC DP ME S/T SP TP FAN VC AA: alpha-amylase; BG: beta-glucan; DP: diastatic power; ME: malt extract; SP: soluble protein; S/T: Kolbach Index; TP: total protein; FAN: free amino nitrogen; VC: viscosity

157 135 Supplementary Table B2.2: Malting quality data from the Oregon Promise subset grain stored for 5 months (ST-2) prior to malting and brewing (Continued). Genotype AA BG MC DP ME S/T SP TP FAN VC Golden Promise Full Pint CDC Copeland AA: alpha-amylase; BG: beta-glucan; DP: diastatic power; ME: malt extract; SP: soluble protein; S/T: Kolbach Index; TP: total protein; FAN: free amino nitrogen; VC: viscosity

158 Supplementary Table B3.1: Malting quality data from the Oregon Promise subset grain stored for 10 months (ST-3) prior to malting and brewing. 136 Genotype AA BG MC DP ME S/T SP TP FAN VC AA: alpha-amylase; BG: beta-glucan; DP: diastatic power; ME: malt extract; SP: soluble protein; S/T: Kolbach Index; TP: total protein; FAN: free amino nitrogen; VC: viscosity

159 Supplementary Table B3.2: Malting quality data from the Oregon Promise subset grain stored for 10 months (ST-3) prior to malting and brewing (Continued). 137 Genotype AA BG MC DP ME S/T SP TP FAN VC Golden Promise Full Pint CDC Copeland AA: alpha-amylase; BG: beta-glucan; DP: diastatic power; ME: malt extract; SP: soluble protein; S/T: Kolbach Index; TP: total protein; FAN: free amino nitrogen; VC: viscosity

160 Supplementary Table B4: Malting protocol used to produced three levels of malt modification at the Canadian Malting Barley Technical Centre, Winnipeg, CA. 138 Treatment Steeping Germination Kilning MT-1 14 C 96 C 24 Hour with 4 85 MT-2 8W-16D-8W-10D-1W-2D@ 14 C 96 C 24 Hour with 4 85 MT-3 8W-16D-10W-8D-3W-2D@ 14 C 96 C 24 Hour with 4 85

161 139 Supplementary Table B5.1: Best linear unbiased predictors (BLUP) for individual doubled haploids in the Oregon Promise subset grain stored for 1 month (ST-1) prior to malting and brewing. Genotype BD BT CL CH CO FL FR GR GS HN MT RO SF ST TO TF LSD (0.05) BD: body; BT: bitter; CL: cereal; CH: chemical; CO: color; FL: floral; FR: fruit; GR: grain; GS: grass; HN: honey; MT: malt; RO: roasted; SF: sulfur; ST: sweet; TO: toasted; TF: toffee.

162 140 Supplementary Table B5.2: Best linear unbiased predictors (BLUP) for individual doubled haploids in the Oregon Promise subset grain stored for 1 month (ST-1) prior to malting and brewing (Continued). Genotype BD BT CL CH CO FL FR GR GS HN MT RO SF ST TO TF Golden Promise Full Pint CDC Copeland LSD (0.05) BD: body; BT: bitter; CL: cereal; CH: chemical; CO: color; FL: floral; FR: fruit; GR: grain; GS: grass; HN: honey; MT: malt; RO: roasted; SF: sulfur; ST: sweet; TO: toasted; TF: toffee.

163 141 Supplementary Table B6.1: Best linear unbiased predictors (BLUP) for individual doubled haploids in the Oregon Promise subset grain stored for 5 months (ST-2) prior to malting and brewing. Genotype BD BT CL CH CO FL FR GR GS HN MT RO SF ST TO TF LSD (0.05) BD: body; BT: bitter; CL: cereal; CH: chemical; CO: color; FL: floral; FR: fruit; GR: grain; GS: grass; HN: honey; MT: malt; RO: roasted; SF: sulfur; ST: sweet; TO: toasted; TF: toffee.

164 142 Supplementary Table B6.2: Best linear unbiased predictors (BLUP) for individual doubled haploids in the Oregon Promise subset grain stored for 5 months (ST-2) prior to malting and brewing (Continued). Genotype BD BT CL CH CO FL FR GR GS HN MT RO SF ST TO TF Golden Promise Full Pint CDC Copeland LSD (0.05) BD: body; BT: bitter; CL: cereal; CH: chemical; CO: color; FL: floral; FR: fruit; GR: grain; GS: grass; HN: honey; MT: malt; RO: roasted; SF: sulfur; ST: sweet; TO: toasted; TF: toffee.

165 143 Supplementary Table B7.1: Best linear unbiased predictors (BLUP) for individual doubled haploids in the Oregon Promise subset grain stored for 10 months (ST-3) prior to malting and brewing. Genotype BD BT CL CH CO FL FR GR GS HN MT RO SF ST TO TF LSD (0.05) BD: body; BT: bitter; CL: cereal; CH: chemical; CO: color; FL: floral; FR: fruit; GR: grain; GS: grass; HN: honey; MT: malt; RO: roasted; SF: sulfur; ST: sweet; TO: toasted; TF: toffee.

166 144 Supplementary Table B7.2: Best linear unbiased predictors (BLUP) for individual doubled haploids in the Oregon Promise subset grain stored for 10 months (ST-3) prior to malting and brewing (Continued). Genotype BD BT CL CH CO FL FR GR GS HN MT RO SF ST TO TF Golden Promise Full Pint CDC Copeland LSD (0.05) BD: body; BT: bitter; CL: cereal; CH: chemical; CO: color; FL: floral; FR: fruit; GR: grain; GS: grass; HN: honey; MT: malt; RO: roasted; SF: sulfur; ST: sweet; TO: toasted; TF: toffee.

167 145 Supplementary Table B8.1: QTL results table for the malting quality traits measured in the OPS grown in Madras, Or and stored at three intervals (1 month (A); 5 months (B); 10 months (C)). Trait Chromosome Peak Position (cm) Closest Marker Marker Name A B C A B C A B C A B C AA 2H, 5H, 7H 5H, 6H, 7H 2H, 5H 117.3, 236.3, , 133.6, , , 53, 1 52, 19, 2 6, 52 SC_116694, MC_3360, 1_0682 MC_4933, 2_0005, 1_0121_120 SC_131218, MC_4933 BG 5H, 7H 6H, 7H 4H, 7H 167.2, , , , 3 7, 2 18, 6 2_1355, 3_1350_60 1_0427, 3_1350_60 MC_36830_732, 3_1350_60 MC 7H 4H 7H _1305 2_0272 3_1305 DP ME 2H, 6H, 7H 5H 2H, 5H, 5H, 6H 2H, 5H, 5H, 7H 2H, 7H 5H, 7H 117.3, 76.33, , 41.4, , 73.8, 234.3, , , , 14, , 6, 52, 14 8, 15, 52, 11 8, 7 SC_116694, 2_1256, 1_ , 1 MC_1598 MC_21545, 1_0688, 2_0206, 2_1256 2_1261, 3_1427, MC_4933, 1_0451 2_1261, 1_0406_60 MC_1598, 1_0682 FAN 5H, 7H 1Ha, 2H, 3H 2H, 5H, 7H 234.3, , 137.0, , 234.3, , 7 2, 32, 24 24, 52, 2 MC_4933, 1_0406_60 2_1174, 1_1323, 1_1411 1_0859, MC_4933, 1_0121_120 S/T 5H, 7H 5H 5H, 7H 234.3, , , , 2 MC_4933, 3_1350_60 MC_4933 MC_4933, 3_1350_60

168 146 Supplementary Table B8.2: QTL results table for the malting quality traits measured in the OPS grown in Madras, Or and stored at three intervals (1 month (A); 5 months (B); 10 months (C)) (Continued). Trait LOD LOD score R2 Additive Effect* A B C A B C A B C A B C AA , 23.9, , 18.52, , , 0.19, , 0.05, , , , , , , BG , , , , , , , , , MC DP , 16.1, , 22.4, 30.01, , , 0.31, , 0.35, 0.35, , , 14.95, , 0.12, , , , ME , 16.71, 53.15, , , 0.08, 0.47, , , 0.42, , , FAN , , 14.17, , 15.88, , , 0.18, , 0.14, , , 15.54, , , S/T , , , , , , *Additive Effect: Golden Promise (-); Full Pint (+)

169 Supplementary Table B9.1: QTL results table for the flavor traits measured in the OPS grown in Madras, Or and stored at three intervals (1 month (A); 5 months (B); 10 months (C)). 147 Trait Color Body 6H, 7H 3H, 3H, 6H Chromosome Peak Position (cm) Closest Marker Marker Name A B C A B C A B C A B C 3H, 5H, 7H 2H, 3H 5H 79, 54 2H, 3H 16, 82, , 101.7, , , , 151 3, 11, 17 19, 7, 1 39, , 18 1_0331, 1_0406 2_0529, 3_0754, 3_1235 1_0584, 3_0654, 1_0841 3_1461, 2_1212 Floral 2H 5H _1166 2_1141 Fruit 3_0654 3_1461, 2_1212 Grass 3H _0584 Honey 3H _0953 Malt 4H, 6H, 7H 2H 148.1, 47.8, , 6, _0610, MC_1277_F, 3_0368 1_2016 Sweet 2H _1323 Toasted 1Hb 2H, 5H, 7H , 202, , 36, 27 1_0522 1_0876, 2_1141, 3_1325 Toffee 5H 2H MC_4933 2_1166

170 Supplementary Table B9.2: QTL results table for the flavor traits measured in the OPS grown in Madras, Or and stored at three intervals (1 month (A); 5 months (B); 10 months (C)) (Continued). 148 Trait LOD threshold LOD score R2 Additive Effect* A B C A B C A B C A B C Color Body , , 3.45, , 11.6, , , , , 0.16, , 0.49, , , , , , 0.03 Fruit Floral -0.67, 0.76, , Grass Honey , 0.67, - 9.3, Malt , , , Sweet Toasted , 10.5, , 0.33, 0.41 Toffee *Additive Effect: Golden Promise (-); Full Pint (+) , , ,

171 149 Appendix C Supplementary Table C1.1: Potential facultative barley accessions based on the phenotypic criteria of >50% WS and <95 DTF under greenhouse conditions without vernalization. Accession ST WS DTF 14AB 14DE 14ID 14MN 14OH 15AB 15ID 15MN 15NE 15OH 15UK 15ES GH FD MW09S OR OBADV MW09S OR OBADV P Grete OR F PYT Ab08-X03W OBADV OR F Scio Sancheong_Covered_ OR OR

172 150 Supplementary Table C1.2: Potential facultative barley accessions based on the phenotypic criteria of >50% WS and <95 DTF under greenhouse conditions without vernalization (Continued). Accession ST WS DTF 14AB 14DE 14ID 14MN 14OH 15AB 15ID 15MN 15NE 15OH 15UK 15ES GH FD 06OR OR MW09S OR Dea Dicktoo OR OBADV Kentucky OR Gaiano OBADV F OBADV PO71DH OR F OR OR

173 151 Supplementary Table C1.3: Potential facultative barley accessions based on the phenotypic criteria of >50% WS and <95 DTF under greenhouse conditions without vernalization (Continued). Accession ST WS DTF 14AB 14DE 14ID 14MN 14OH 15AB 15ID 15MN 15NE 15OH 15UK 15ES GH FD Bulbul Tarm Maja p OR OR F OR OR OR Ab08-X03W F OR F Bereke_ F F OR Ab08-X03W

174 152 Supplementary Table C1.4: Potential facultative barley accessions based on the phenotypic criteria of >50% WS and <95 DTF under greenhouse conditions without vernalization (Continued). Accession ST WS DTF 14AB 14DE 14ID 14MN 14OH 15AB 15ID 15MN 15NE 15OH 15UK 15ES GH FD OR OR OR OBADV F OR F OR Zeynelaga Amb GK_Stramm OR MW09S OR F OR

175 Supplementary Figure C2: Frequency distributions of averaged winter survival scores across 12 environments (A) and flowering times in unvernalized conditions (B) for the 941 LTT panel. 153

176 Supplementary Figure C3: Manhattan plots of LOD scores from GWAS of winter survival scores averaged across differential environments of the 941 accession in the LTT panel. 154