The Pennsylvania State University

The Graduate School

EXAMINING REGIONAL TRENDS IN PENNSYLVANIA GRÜNER VELTLINER

WINES USING INSTRUMENTAL AND HUMAN SENSORY METHODS

A Thesis in

Food Science

by

Stephanie Keller

Ó 2020 Stephanie Keller

Submitted in Partial Fulfillment of the Requirements for the Degree of

Master of Science

August 2020 ii The thesis of Stephanie Keller was reviewed and approved by the following:

Helene Hopfer Assistant Professor of Food Science Thesis Co-Advisor

Ryan J. Elias Professor of Food Science Thesis Co-Advisor

Michela Centinari Associate Professor of Viticulture

Robert F. Roberts Professor of Food Science Head of the Department of Food Science

iii ABSTRACT

It is often said that high quality grapes must be used in order to create high quality wines.

This production begins in the vineyard and is impacted by viticultural and environmental conditions that may or may not be able to be controlled. Weather conditions are among these uncontrollable factors, and the influence of weather conditions on final grape and wine quality has been the subject of investigation in both research and industry for many years. Many studies have determined that factors such as rainfall, sunlight exposure, and temperature play an important role in the development of phenolic and aromatic compounds and their precursors in berries, which ultimately affects wine aroma, taste, and flavor.

Examination of weather conditions and climate in wine regions have been the subject of studies not only to understand impacts on wine quality attributes, but also to determine if regional trends exist for particular areas. The concept of regionality, or the particular style of wine that a growing region produces, is a new area of study for the Eastern United States, including

Pennsylvania, which is the focus of this study. Determining regional characteristics can assist growers in predicting the quality of grapes before harvest and can also identify viticultural and sensory factors that can improve marketing strategies to wine consumers.

Grüner Veltliner (Vitis vinifera; GV), an Austrian grape variety, is a relatively new grape to wine growers and producers in the northeast United States including Pennsylvania (PA). While climatic conditions are favorable to its growth, the PA wine industry is still becoming familiar with the varietal characteristics of GV grown and produced across the state. The aim of this study was to characterize the chemical and sensory factors that drive regional differences in PA GV wines through a controlled winemaking study.

GV grapes were harvested from experimental sites in four PA growing regions over two vintages. A total of 9 sites were chosen in order to capture climate variability of the growing

iv regions and to be representative of the various regions that currently grow GV in PA – two in the

Northwest (NW), one in the North Central region (NC), one in the Northeast (NE), and five in the

Southeast region (SE). Weather data was collected from each site during both growing seasons as well. Wines were vinified using a standard vinification method to eliminate the influence of winemaking style on samples. Chemical analysis included headspace solid phase microextraction gas chromatography mass spectrometry (HS-SPME-GC-MS), high performance liquid chromatography (HPLC), and color analysis using CIE-Lab. Descriptive analysis was used to evaluate aroma, taste, mouthfeel, and flavor intensities in wines vinified from each experimental site. Finally, Pearson’s correlation was used to examine if any relationships existed between sensory and instrumental variables.

Two descriptive analysis panels (one for each year of study) found significant differences

(p < 0.05) in multiple aroma, taste, and flavor attributes. In Year 1, regional differences were found in yellow color and haziness, with the NW region produced wines rated highest in yellow color and lowest in haziness. In Year 2, regional differences were found again in yellow color and haziness, as well as in green apple, citrus, and thiol flavor; thiol aroma; sour, sweet, and salty taste; and astringent mouthfeel. NW wines were again highest in yellow color, while the SE region produced wines highest in thiol aroma and flavor in both years of study. In Year 2, NW wines were highest in citrus and green apple flavors, as well as sour taste, which may be due to the higher acidity in grapes at harvest from this region.

Aroma analysis found regional differences in the concentration of a number of aromatic precursors in both juice and wine samples. Some differences in wine aroma profiles were likely due to higher concentrations of aroma precursors, such as hexyl acetate in NC wines.

Significantly different volatile compounds did not necessarily align with the significantly different aroma attributes found in descriptive analysis.

v There were not regional differences in concentrations of phenolic compounds, namely

(+)-catechin, (-)-epicatechin, and gallic acid, except in Year 1 where NC wines were lower in (-)- epicatechin content than the NW and SE regions. CIE-Lab color analysis showed similar results to panelist appearance ratings; NW wines were higher in b*, indicating the wines were more yellow in color than those from other regions.

A number of significant positive and negative linear correlations (p < 0.05) were found when examining sensory and chemical variables together. In both years of study, yellow color ratings were positively correlated with b* and negatively with a*. Sour taste was positively correlated with titratable acidity and negatively correlated with pH. Other correlations between aroma compounds and aroma attributes were found, and while these correlations do not prove causality, these can give insight into which relationships should be further studied.

Overall, this study successfully identified regional differences in PA GV wines using both sensory and instrumental methods. Since winemaking was controlled in this study, thus eliminating variation in winemaking style that can be found in commercial GV, differences in appearance, aroma, and flavor are likely due to differences in weather conditions across growing regions. Further work is necessary to relate climatic conditions to final wine composition, and a third year of study is planned to provide further insight into regional climatic differences and drivers of regionality in GV. Results of this study unveiled key sensory attributes of PA GV, which will be useful for wineries and their tasting room staff to market GV wines to consumers and increase recognition of the variety as it becomes more popular with producers in the state.

vi TABLE OF CONTENTS

LIST OF FIGURES ...... viii

LIST OF TABLES ...... ix

ACKNOWLEDGEMENTS ...... xi

Chapter 1 Literature Review ...... 1

1.1 Viticultural Factors that Influence Wine Quality ...... 1 1.1.1 Sunlight Exposure ...... 1 1.1.2 Water Status and Rainfall ...... 5 1.2 Wine Regionality ...... 8 1.2.1 Regionality Studies ...... 8 1.3 Grüner Veltliner ...... 10 1.4 Comparison of Austrian and Pennsylvanian Growing Conditions ...... 13 1.5 Purpose and Significance ...... 14 1.6 Hypothesis and Objectives ...... 15

Chapter 2 Materials and Methods ...... 17

2.1 Site Selection ...... 17 2.2 Vinification ...... 18 2.3 Descriptive Analysis ...... 20 2.4 Chemical Analysis ...... 25 2.4.1 Juice and Wine Chemical Analysis ...... 25 2.4.2 Aroma Analysis of Juice via HS-SPME-GC-MS ...... 25 2.4.3 Aroma Analysis of Wine via HS-SPME-GC-MS ...... 26 2.4.4 Phenolic Analysis of Wine via HPLC ...... 27 2.4.5 Color Analysis of Wines via CIE-Lab Color Measurements ...... 28 2.5 Statistical Analysis ...... 28

Chapter 3 Results and Discussion ...... 30

3.1 Descriptive Analysis ...... 30 3.1.1 Year 1 ...... 30 3.1.2 Year 2 ...... 34 3.2 Aroma Analysis of GV Juices by HS-SPME-GC-MS ...... 40 3.2.1 Year 1 ...... 40 3.2.2 Year 2 ...... 42 3.3 Basic Wine Chemistry Results ...... 44 3.4 Aroma Analysis of GV Wiens by HS-SPME-GC-MS ...... 45 3.4.1 Year 1 ...... 45 3.4.2 Year 2 ...... 48 3.5 Phenolic Analysis of GV Wines by HPLC ...... 50 3.5.1 Year 1 ...... 50

vii 3.5.2 Year 2 ...... 51 3.6 Color Analysis of GV Wines ...... 55 3.6.1 Year 1 ...... 55 3.6.2 Year 2 ...... 56 3.7 Correlation of Sensory and Instrumental Data ...... 58

Chapter 4 Conclusions and Future Work ...... 62

4.1 Overall Conclusions ...... 62 4.2 Future Work ...... 63

References ...... 65

Appendix ...... 77

Chapter 2 Supplementary Data ...... 78 Chapter 3 Supplementary Data ...... 82

viii LIST OF FIGURES

Figure 2.1 Location of experimental sites within PA wine regions recognized by the PA Winery Association (PWA; pennsylvaniawine.com)...... 18

Figure 3.1 PCA biplots for Year 1 GV wines A) with the haziness and B) without the haziness attribute. Wines are shown with their 95% confidence ellipses and colored by region. Attributes ending in A indicate aroma, F indicate flavor, MF indicate mouthfeel, and T indicate taste attributes ...... 34

Figure 3.2 PCA biplot for Year 2 GV wines. Wines are shown with their 95% confidence ellipses and colored by region. Attributes ending in A indicate aroma, F indicate flavor, MF indicate mouthfeel, and T indicate taste attributes...... 38

Figure 3.3 PCA biplot with volatile compounds that were significantly different by juice for Year 1 GV juice samples...... 41

Figure 3.4 PCA biplot for Year 2 samples with volatile compounds that were significantly different by juice...... 43

Figure 3.5 PCA biplot for Year 1 GV wines with volatile compounds that were significantly different by wine sample. Wines are shown with 95% confidence ellipses...... 47

Figure 3.6 PCA biplot for Year 2 GV wines with volatile compounds that were significantly different by wine sample. Wines are shown with 95% confidence ellipses...... 49

Figure 3.7 ANOVA by wine and Tukey’s post-hoc comparison results for relative abundance of catechin (A,D), epicatechin (B,E), and gallic acid (C,F) in GV wines. Results for Year 1 wines are shown in A, B, and C, while Year 2 wine results are shown in D, E, and F...... 53

Figure 3.8 Results for ANOVA by region with Tukey’s post-hoc comparison for relative abundance of catechin (A,D), epicatechin (B,E), and gallic acid (C,F). Year 1 results are shown in A, B, and C, while Year 2 results are in D, E, and F...... 54

Figure 3.9 Comparison of L*, a*, and b* color coordinates for Year 1 (A,B,C) and Year 2 (D,E,F) wines. Measurements shown for each fermentation replicate are the average of analytical replicate values...... 57

Figure 3.10 Significant (p < 0.05) correlation of sensory and chemistry parameters for Year 1 (A) and Year 2 (B) GV wines. Positive correlations are shown in red shades while negative correlations are shown in blue...... 61

ix LIST OF TABLES

Table 2.1A Appearance, aroma, and flavor reference standards used for training of sensory attributes for Year 1 and Year 2 DA. All standards were created in Bota Box Pinot Grigio (Bota Box Vineyards, Manteca, CA) base wine unless otherwise noted .... 23

Table 2.1B Taste and mouthfeel reference standards used for training of sensory attributes for Year 1 and Year 2 DA. All standards were created in Bota Box Pinot Grigio (Bota Box Vineyards, Manteca, CA) base wine unless otherwise noted ...... 24

Table 3.1 Significant differences between PA regions for Year 1 GV wines. Values that share the same letter within column are not significantly different according to Tukey’s post-hoc comparison (p < 0.05) ...... 33

Table 3.2 Significant differences between PA regions for Year 2 GV wines. Values that share the same letter within column are not significantly different according to Tukey’s post-hoc comparison (p < 0.05). Attributes that end in F indicate flavor, A indicate aroma, T indicate taste, and MF indicate mouthfeel attributes ...... 37

Table 3.3 Significant differences between regions for Year 1 GV juice samples. Values that share a letter in column are not significantly different according to Tukey’s post-hoc comparison (p < 0.05)...... 42

Table 3.4 Significant differences between regions for Year 2 GV juice samples. Values that share a letter in column are not significantly different according to Tukey’s post-hoc comparison (p < 0.05)...... 44

Table 3.5 Significant differences between regions for Year 1 GV wine chemistry parameters...... 45

Table 3.6 Significant difference between regions for Year 2 GV wine chemistry parameters...... 45

Table 3.7 Significant differences between regions for Year 1 GV wines. Internal standard equivalent values that share a letter in column are not significantly different according to Tukey’s post-hoc comparison (p < 0.05)...... 47

Table 3.8 Significant differences between regions for Year 2 GV wines. Internal standard equivalent values that share a letter in column are not significantly different according to Tukey’s post-hoc comparison (p < 0.05)...... 50

Appendix Table 1 ANOVA for Year 1 sensory attributes showing the F-values for the factors Wine (W), Fermentation Replicate (FR), Judge (J), and their interactions. Bolded F-values are significant (p < 0.05) ...... 82

Appendix Table 2 ANOVA for Year 2 sensory attributes showing the F-values for the factors Wine (W), Fermentation Replicate (FR), Judge (J), and their interactions. Bolded F-values are significant (p < 0.05) ...... 84

x Appendix Table 3 Wine chemistry data for Year 1 wines ...... 85

Appendix Table 4 Wine chemistry data for Year 2 wines ...... 86

Appendix Table 5 Harvest and juice chemistry results for Year 1 experimental sites ...... 87

Appendix Table 6 Harvest and juice chemistry results for Year 2 experimental sites ...... 87

Appendix Table 7 Volatile compounds that significantly differed by juice or wine samples for both years of study, retention time, and calculated and reported retention indices (RI) ...... 88

xi ACKNOWLEDGEMENTS

I must first thank my advisors, Dr. Ryan Elias and Dr. Helene Hopfer, for giving me the opportunity to continue my education with the Department of Food Science. I greatly appreciate the support and encouragement you have given these last two years. Thank you for trusting and believing in me from the beginning, and giving me confidence to tackle a project that took me from the vineyards to the pilot plant, lab, and SEC.

Thank you as well to my third committee member, Dr. Michela Centinari, who first encouraged me to consider graduate school and supported me through my entire journey. I have been able to have such a well-rounded graduate experience because of your support. I loved being able to see this project from vine to wine, and I was pleased to learn so much about viticultural practices along the way.

The “Grüner Project” was especially rewarding to work on because there were so many people involved in it. Thank you to Drew Harner for being a great teammate on this project, driving the lab van at 7 in the morning when I was still half-asleep, and being willing to help with all aspects of the project, even if they were on my side of Curtin Road. It would have been impossible to make so much wine on my own, and I am grateful for everyone who assisted in harvesting and winemaking, including Don Smith, Colton Craig, Bryan Hed, and Dr. Molly

Kelly. Thank you also to Dr. Alyssa Bakke for her guidance in descriptive analysis.

I cannot thank my lab mates enough for their support, encouragement, assistance in the pilot plant, enthusiasm for creative Halloween costumes, and friendship over the past two years. I always knew I could go to any one of you for help, advice, or a coffee break when I needed it, and I am very grateful for that. Thank you Ben and Conor for assistance in HPLC method development for this project as well. Drew, Ben, Conor, Magda, and Andrew – thanks for being such wonderful lab mates and friends. These crazy times will be hard to forget.

xii There are many people at Penn State that have been amazing friends and supporters to me during my time here. Thank you to Kayleigh, Molly, Kate, Terianne, Weslie, and Lisa for so many fun times. Thank you to Denise Gardner for your mentorship since my sophomore year of undergrad. It’s because of you that I caught the wine bug and I’m so grateful for it. Your guidance, advice, and wise words are always appreciated, and I value our friendship so much.

Jen, thank you so much for being there for me since day one (literally!). I am so fortunate to have a friend like you who always listens, never judges, and is there for me every step of the way. Allison, I am incredibly thankful that our time running descriptive analysis led to a supportive and great friendship. Thank you for listening when I needed to vent, giving advice when I felt overwhelmed, always checking in, and of course, all the social distance walks over the last few months!

I would like to thank the faculty and staff of the Penn State Department of Food Science for not only their support over the last two years, but also the four years I was an undergraduate in this department. The sense of belonging I felt the moment I started graduate school has stuck with me all this time, and it has made all the difference. I honestly can’t picture myself going to school anywhere else, and I believe that’s because of the kindness and care the department has shown me. In this department I’ve found a place where I am appreciated, valued, and respected, and for that I am incredibly grateful.

Finally, I would like to thank my parents for their love and support. And of course, thank you Natalie for being my best friend, listening to me complain, and not being mad that I stayed at

Penn State for two more years while you had to go be a “real adult.”

1

Chapter 1

Introduction

1.1 Viticultural Factors that Influence Wine Quality

It is often said that producing high quality wine begins in the vineyard, with the careful and controlled growth of quality wine grapes. Viticulturists use various canopy management practices throughout the growing season as a way to influence final grape quality; however, there are still factors that influence final quality at harvest that are unable to be controlled. These factors are most significantly determined by weather conditions, such as temperature, rainfall, and solar radiation. The influence of weather on final wine quality has been the subject of investigation in research and the industry for many years as a means to gain an understanding of how these conditions enhance or diminish quality attributes that may impact a consumer’s purchasing decision. These studies have also been important to the industry in guiding producers’ decisions made in the vineyard to optimize production of high quality wine grapes.

1.1.1 Sunlight Exposure and Temperature

Grape cluster and leaf exposure to sunlight during veraison and ripening is key to the accumulation of sugars and degradation of organic acids in berries. Throughout the growing season, this exposure is generally controlled by manual or mechanical leaf removal within the canopy to ensure clusters receive adequate sunlight without resulting in sunburnt berries.

Leaf removal can also increase air flow into the canopy and therefore reduce disease pressure, which is especially important in areas with humid and wet climates such as

2 Pennsylvania. Fungal species such as Botrytis cinerea, which causes bunch rot, thrive in humid conditions with frequent rainfall. Specifically, B. cinerea produces laccases that can oxidize phenolic compounds in grape berries, causing browning in both juice and finished wines. Downy mildew (Plasmopara viticola) is also a serious threat to berry quality in humid and wet regions, and can reduce photosynthesis in leaves and eventually negatively impact yield and fruit quality

(König et al., 2017). Surface water on berries, as a result from rainfall, can increase spread of these fungal diseases, which makes leaf removal especially important in humid and wet growing areas to improve airflow and increase berry surface temperature, facilitating evaporation and drying on the berry surface (Keller, 2010). Resulting sun exposure from leaf removal aids in increasing berry surface temperature and eventual drying of the berry surface.

Many studies have examined the role of sunlight exposure with respect to the accumulation of volatile aroma compounds and precursors in berries throughout the growing season. Levels of monoterpenes (i.e. linalool and hotrienol) and C13-norisoprenoids (i.e. β- damascenone and β-ionone), which can impart aromas that range from citrus to floral in grapes and wine, have been found to increase with greater sunlight exposure (Reynolds & Wardle,

1989). Timing of sun exposure during the ripening stage is of importance as well, as accumulation of aromatic compounds can vary depending on the timing of leaf removal after berry set (Kwasniewski et al., 2010).

While some volatile compounds may accumulate in response to increasing sunlight exposure, others may decline. Compounds such as methoxypyrazines, which provide green/vegetal aromas compared to the fruity aromas of terpenes, are found in higher concentrations in shaded clusters that receive less sunlight compared to unshaded clusters (Marais et al., 1999). Similar results have been found in studies that did not employ additional leaf removal to examine this phenomenon, but instead relied on the natural shading or exposure of clusters based on row orientation (Martin et al., 2016).

3 Sunlight exposure not only impacts volatile aroma composition in berries and wines, but also affects accumulation and profile of nonvolatile compounds. Sunlight exposure has been found to increase total phenolic concentration in many red varieties; however, it is likely that effects differ by cultivar and season (Downey et al., 2006). Formation of phenolic compounds is dependent on the enzyme phenylalanine ammonia lyase (PAL), which is involved in the first steps of the shikimate pathway and converts phenylalanine to cinnamic acid. This enzymatic pathway eventually leads to formation of flavonoids and results in darker berry color, and PAL activity can increase with greater exposure to light (Roubelakis-Angelakis & Kliewer, 1986;

Teixeira et al., 2013).

Again, timing of sunlight exposure or sunlight exclusion is of importance when considering development of these compounds in berries. Li et al. (2013) found that sunlight exclusion from fruit set to 1 week pre-veraison increased anthocyanin content at maturity compared to a control, and these clusters also developed deeper color than control and other treatment clusters. However, sunlight exclusion 1 week pre- and post-veraison decreased anthocyanin content in berries. Findings by Hickey et al. (2018) showed contrasting results, in which wine color intensity was rated higher and total phenolics increased for wines that had grapes exposed to sunlight pre-bloom, but total anthocyanin concentration was not affected by sunlight exposure. Shading of berries has been found to significantly decrease synthesis of flavonols in grape skin of Shiraz (Downey et al., 2008), but this has not been frequently studied in white varieties.

The previously mentioned studies used red grape varieties to study the effects of sunlight exclusion. Compared to red wines, the influence of sunlight exposure on phenolic compounds in white wines is not well studied, in part due to the role of sunlight exposure in anthocyanin development in red varieties that contribute to red wine color. However, some research has shown that sunlight exposure can impact the sensory aspects of white wines, even though white wines

4 undergo less phenolic extraction due to limited skin contact in fermentation. Allegro et al. (2019) found that white wines vinified with sun exposed clusters had more intense color, less vegetal aroma, and were perceived as more bitter and astringent. Sun exposure has also been found to increase phenolic content in the white variety Sauvignon blanc (Martin et al., 2016). Because of limited extraction from skin contact, typically hydroxycinnamates from the pulp of grapes are the most abundant phenolic compounds in white grape juice, while flavanols from grape skin and seeds are still present in white varieties (Adams, 2006; Waterhouse et al., 2016).

Generally, it is difficult for researchers to decouple the effects of sunlight exposure and increased berry temperature in field studies (Bergqvist et al., 2001; Tarara et al., 2008). As previously mentioned, leaf removal will result in an increase in sunlight exposure for clusters, but it will also result in an increase in berry surface temperature. When examining results from leaf removal studies, it is also important to consider increased berry surface temperature may have affected results.

Temperature is an important factor for grape growers to consider when planting vines, as proper temperature ranges are needed for berry development as well as minimize likelihood of freeze injury that can increase bud mortality and therefore impact yields for the following season

(Kurtural, 2003). In general, higher temperatures are known to increase total soluble solids (TSS) and decrease organic acid levels, indicating higher temperatures greatly impact ripening

(Kabinett, 1993). Studies have shown that both day and night temperature variation can impact both acid degradation and anthocyanin synthesis in berries (Sweetman et al., 2014; Kliewer &

Torres, 1972).

It is likely that aroma development in response to warmer or cooler temperatures in the growing season is cultivar and season dependent. In Nebbiolo grapes, increased temperatures in the vineyard decreased norisoprenoid concentration (Asproudi et al., 2016). In contrast, higher temperatures in South African growing regions increased levels of some norisoprenoids in

5 Weisser Riesling wine, namely 1,1,6-trimethyl-1,2-dihydronaphthalene (TDN), trans-vitispirane, and trans-1,8-terpin compared to cooler regions in Germany and Northern Italy (Marais et al.,

1992). When evaluated by a trained panel, Semillon wines were more tropical in flavor and had less green and citrus aroma if they were made with grapes grown in higher temperature vineyards

(Sadras et al., 2013a). Generally, green aroma and flavor is believed to decline with increased growing temperatures, but compounds that impart green aroma, namely isobutyl-2- methoxypyrazine (IBMP) and C6 compounds, can be found in higher concentrations with warmer temperatures in some instances. Again, this may be dependent on growing season (Martin P.

Mendez-Costabel et al., 2013).

Higher temperatures have also been associated with more advanced berry color and higher concentrations of anthocyanins in red grapes; however, if temperatures become too elevated, anthocyanin concentration can decrease regardless of sun exposure (Sadras et al.,

2013b; Tarara et al., 2008). Work using Shiraz found that clusters exposed to temperature maxima of 35°C and 46°C generally did not differ in concentration and profile of flavonoids, however, berries that were exposed to high temperatures of 54°C had rapid degradation of anthocyanins and flavonols in addition to berry shriveling (Gouot et al., 2019). In Merlot berries, loss of anthocyanins has been observed at 35°C (Spayd et al., 2002). It is possible that the decline of anthocyanin concentration at high temperatures is caused by both anthocyanin degradation and inhibition of anthocyanin biosynthesis (Mori et al., 2007).

1.1.2 Water Status and Rainfall

The amount of rainfall received during fruit growing and ripening processes will have varying effects on berry and wine quality. This phenomenon is typically studied through examination of vine water status and leaf water potential. Many studies have examined the effect

6 of water deficit on berry sugar accumulation, a key indicator of ripeness and quality. Mild water stress was found to be most beneficial for sugar accumulation in berries, while severe water stress decreased sugar content (Peyrot des Gachons et al., 2005; Van Leeuwen et al., 2009; Zsófi et al.,

2011). Under moderate water stress, photosynthetic activity is reduced, which can result in a decline in shoot growth (Coombe, 1989; Chaves et al., 2010). If photosynthesis is reduced post- veraison, sugar accumulation in berries can decline and negatively impact flavor development in berries (M. Centinari, personal communcation, July 3, 2020). The hormone abscisic acid (ABA) promotes ripening, and its increase in response to water deficit may also promote sugar accumulation in berries under these conditions (Kuhn et al., 2014). However, ABA concentration in response to water deficit is cultivar dependent (Deluc et al., 2009). Berry weight and size is also reduced by water stress and less rainfall (Peyrot des Gachons et al., 2005; Van Leeuwen et al., 2009; Mendez-Costabel et al., 2014), with deficits between veraison and harvest having the most drastic effect on berry weight (Ojeda et al., 2002). Balance of water deficit is important in determining fruit quality, as mild stress can enhance sugar accumulation, but severe water stress will ultimately restrict photosynthesis, reducing yield and negatively affecting fruit maturity (Van

Leeuwen et al., 2009).

Concentrations of nonvolatile compounds are also influenced by water status and rainfall.

Van Leeuwen and colleagues found that mild water-stress between veraison and ripening caused increased anthocyanin accumulation in berries (Van Leeuwen et al., 2009). Deluc and colleagues found water deficit to increase the transcriptional expression of PAL in Cabernet Sauvignon, promoting anthocyanin production, but did not find this in Chardonnay (Deluc et al., 2009).

Water deficit has also been found to increase expression of genes that control anthocyanin biosynthesis in Cabernet Sauvignon when deficit occurred both before onset of veraison and after it (Castellarin et al., 2007a,; Castellarin et al., 2007b). It is possible that accumulation of anthocyanins and phenolics in water-deprived berries may be due to reduction in berry size

7 (Ojeda et al., 2002). With higher concentrations of anthocyanins, color intensity ratings for wines that were made from grapes that received less rainfall in a growing season were higher compared to a control (Mendez-Costabel et al., 2014).

The impacts of water status on berry and wine aroma have been examined in both red and white wine grape varieties. In Sauvignon blanc, mild water stress can enhance aroma and concentration of the varietal thiol 4-mercapto-4-methylpentan-2-one (4MMP) (Peyrot des

Gachons et al., 2005). In Riesling, aroma and flavor intensities and their profiles were altered by vine water status. Low water status vines resulted in wines that were rated low in flavor and aroma intensity, but characterized by citrus and apple/pear flavors, while medium water status resulted in wines that were higher in stone fruit aroma (Marciniak et al., 2013). Kotsaki et al.

(2020) found that Riesling wines with higher water potential had lower concentrations of monoterpenes. In Chardonnay, isoprenoid and carotenoid metabolism was found to be enhanced by water deficit, while flavonol metabolism was enhanced by water deficit in both Charonnday and Cabernet Sauvignon (Deluc et al., 2009). Research examining Muscat aroma and rainfall found that rainfall before veraison may negatively impact terpene precursor development, as well as result in less characteristic aroma for the variety (Crespo et al., 2018).

Higher rainfall has also been associated with lower wine quality ratings in both Italian and French varieties (D. Grifoni et al., 2006; C. Van Leeuwen et al., 2009). In a study in which the effect of rainfall was removed during the dormant season by covering the ground, Merlot wines that did not receive rainfall in the dormant season were more intense in color, mouthfeel, and red berry aroma ratings, as well as less intense in vegetable and capsicum aromas (Mendez-

Costabel et al., 2014). However, this study’s authors reported reduced growth and therefore increased sun exposure in rain-removed vines, which likely contributed to lower concentrations of IBMP and reduced green aroma intensity.

8 1.2 Wine Regionality

It is important to study the effects of environmental conditions such as sunlight exposure, temperature, and rainfall in the context of wine grapes because these conditions, in combination with many other factors, are known to affect final wine quality. It is also believed that consistent weather patterns, along with viticultural factors such as soil type and cultivar, impart specific characteristics to wines (Van Leeuwen et al., 2004). This concept has been popularized in Old

World wine regions as terroir, a term that refers to the geographical influence a place has on wine produced there (“The Geography of Wine,” 2012). While regional characteristics in wines are well recognized in Old World countries, New World wine regionality is less commonly recognized but still an important concept as wines produced in these regions become more recognized among the industry and consumers (Easingwood et al., 2011).

Regionality is a term used in the industry to describe the particular style of wine that a growing region produces (Easingwood et al., 2011). Determining the characteristics of specific regions can help growers predict the quality of grapes or wines before they are harvested or vinified, as well as identify environmental, viticultural, and sensory factors that can improve marketing strategies to wine consumers.

1.2.1 Regionality Studies

Numerous studies have compared wine regions in a number of countries and found differences in aroma, flavor, and mouthfeel aspects in both commercial wines and wines that were made with a standard protocol to eliminate stylistic differences (Green et al., 2011;

Heymann et al., 2015; King et al., 2014; Marais et al., 2017). These studies also found differences in the aromatic profiles of investigated wines. For example, Green et al. found that New Zealand

9 Sauvignon blanc had higher concentrations of green aromatic compounds including IBMP, 3- mercaptohexanol (3MH), and hexanol, while Austrian Sauvignon blanc samples had higher concentrations of esters contributing to fruity aroma, including ethyl butanoate and isoamyl acetate (Green et al., 2011). Some of these studies have been able to relate aroma and flavor differences to environmental conditions. For example, King and colleagues found that Malbec wines from hotter climates had higher red fruit aroma and pH, and lower earthy aroma, sour taste, and TA, even when winemaking was controlled. Samples from Mendoza, Argentina, were also grown at higher elevation, which could affect the aroma and flavor differences found (King et al,

2014).

While it may be expected for grapes grown thousands of miles from each other to produce wines with noticeably different characteristics, studies have also shown that differences can be found within one country of origin. In Australia, analysis of commercial Cabernet

Sauvignon found that presence of specific volatile and non-volatile compounds was region- specific (A. L. Robinson et al., 2012). In Chile, regional proximity to the Andes Mountains negatively impacted ester and acid concentration in commercial Carignan (Úbeda et al., 2017).

Studies in Canada have examined differences within a viticultural area, namely the Niagara

Peninsula. Differences in aroma and flavor exist between sub-appellations of the region, both in commercial and research samples (Douglas et al., 2001; Willwerth et al., 2015). In Spain,

Albariño wines from different sub-regions of the Rias Baixas region were found to differ in aroma intensities, with regional aroma profiles emerging and ranging from fruity and floral

(Condado do Tea sub-region) to lactic in aroma (Val do Salnés sub-region) (Vilanova & Vilariño,

2006). This study examined commercial wines, and stylistic differences were likely the cause of some regional trends found, especially with wines from Val do Salnés, which were the only samples to undergo malolactic fermentation.

10 Vintage can have a varying impact on final wine quality (Asproudi et al., 2016; Fischer et al., 1999), therefore a majority of regionality studies are carried out over a number of growing seasons to ensure that standard weather conditions are captured. When comparing the impacts of vintage, winery, and vineyard location within the Rheingau region in Germany, Fischer and colleagues found that all three factors affected sensory properties of Riesling wines. The researchers also found that some vineyard sites had less climatic variation over two vintages than other sites in the region (Fischer et al., 1999). While many studies have found aroma and flavors to be characteristic of specific wine regions regardless of winemaking practice, controlling vinification by means of employing a standard winemaking protocol is the only way to guarantee differences in final aroma and flavor are directly due to geographic origin and not winemaking technique.

1.3 Grüner Veltliner

Grüner Veltliner (GV) (Vitis vinifera) is a white fleshed wine grape variety that is the signature variety of Austria, comprising 31% of the country’s total wine growing area (Austrian

Wine Marketing Board, 2018). The variety is also the most widely planted white variety in

Slovakia, accounting for 20% of total vineyard area for the country (J. Robinson et al., 2012). The variety is also grown in other European countries, including the Czech Republic, Hungary,

Bulgaria, and Italy.

GV is a relatively new variety to the United States and has not been the subject of many studies in the country, however, the variety has been studied more frequently in Austria. Regner et al. (2009) examined viticultural and enological aspects of a number of GV clones from winegrowing regions in Austria. A trained sensory panel rated apricot, white pepper, and citrus aromas highest among the research wines regardless of growing region or clone. The researchers

11 also found that there were small differences in volatile composition of research wines, and identified terpenes as minor aromatic components of GV wines (Regner et al., 2009). Compounds identified and quantified included (E)-2-hexenol, as well as the monoterpenes linalool, nerol, and geraniol, which have been associated with apricot aroma in Chardonnay and Viognier wines

(Siebert et al., 2018).

Monoterpenes in their free form can also be associated with flora and citrus aroma in wine, and their concentration in berries is heavily dependent on both sun exposure and berry temperature (Belancic et al., 1997; Black et al., 2015; Darriet, Thibon, & Dubourdieu, 2012; A. L.

Robinson et al., 2014). Monoterpenes can be found in berries in both free and bound forms, but are in higher concentration in the bound form, and are found in both the skin and pulp of berries

(Waterhouse et al., 2016). The sesquiterpenoid (-)rotundone has also been identified in Austrian

GV wines (Caputi et al., 2011; Nauer et al., 2018), which can provide a characteristic peppery aroma. Rotundone is primarily found in grape skins, and cooler climates typically favor development of rotundone in berries (Black et al., 2015; Waterhouse et al., 2016).

While GV is typically made into a dry wine and not aged, a recent study examined sur lie aging with GV, a vinification method wherein wines are aged on lees. GV wines aged for 6 weeks and 5 months were evaluated by a panel of wine experts, and aged wines were rated significantly higher in “autolytic” and “body” attributes, which are characteristics associated with sur lie aging (Pavelescu et al., 2012).

GV has also been the subject of regionality studies in Austria. Commercial GV wines from the Weinviertel region were examined to determine if aroma and flavor differences existed within the region. It was found that while wines did not have significant differences in aroma composition via gas chromatography-mass spectrometry (GC-MS), a sensory panel did find wines from the western area of the Weinviertel region to be higher in pear/quince aroma than those from the eastern side of the region (Flak et al., 2007). These wines were then compared to

12 commercial wines from Wagram, a smaller GV producing region in Austria. Wagram GV was found to be characterized by citrus and apple notes, and when compared to Weinviertel GV, was higher in citrus notes and lower in pepper and pear taste (Flak et al., 2009). Although regional trends were found in these studies, it is important to note samples were commercial wines, and so were subject to winemaker influence.

While studied in Austria and popular in European countries, GV is a lesser-known variety in New World wine producing countries, planted in New Zealand and Australia in 2008 and

2010, respectively (J. Robinson et al., 2012). GV has recently been planted in the United States in

Oregon, California, and Washington, and was first planted in Pennsylvania (PA) in 2003 in the

Lehigh Valley (J. Robinson et al., 2012; Pompilio, 2015). In PA, GV is most widely planted in the southeastern growing region but is found in vineyards in the northwest and northeast regions as well.

While GV has gained popularity among more PA wine growers in recent years, wine consumers in the state are still relatively unfamiliar with GV wine. PA winery owners have reported that while consumers may refrain from trying it due to their unfamiliarity with the name and the style, consumption of GV has steadily grown since first introduced on PA tasting room menus (Lenton, 2017). A variety of factors, including price, grape variety, and country of origin, influence consumer willingness to purchase wine (Lockshin & Corsi, 2012). Willingness to try new wines has been shown to increase with consumer wine knowledge, while the same study found that willingness to try new wines decreased with age (Ristic et al., 2016). However, older millennials are more likely to try wines they are unfamiliar with or have never tried before compared to Generation X and Baby Boomers (Caparoso, 2016).

13 1.4 Comparison of Austrian and Pennsylvanian Growing Conditions

The effect of the Danube River on temperature ranges in Austrian GV growing regions allows for favorable growth in vineyards located on sloped or terraced hillsides along the river

(Brook, 2019). Regions in Austria are typically characterized by cooler evening temperatures during the growing season. Elevation levels of Austrian regions vary, with some vineyards at 117 m above sea level, and others located at more than 1,000 m above sea level (Austrian Wine

Marketing Board, 2018).

Some Pennsylvania areas are also impacted by bodies of water, namely the Lake Erie

American Viticultural Area (AVA), which generally has a milder climate compared to other northern growing areas and favors the growth of Vitis vinifera, native Vitis labrusca, and hybrid varieties. Vineyards are located at lower elevations in PA compared to Austria, as the highest elevation in the state is approximately 950 m (Duiker, 2020); however, PA vineyards are commonly located at elevations between 100 m and 600 m above sea level. Annual precipitation is moderate in both PA and Austria, with average annual precipitation in PA being slightly higher, ranging from 812.8 to 1244.6 mm of rainfall each year compared to Austria, which receives 488.6 to 1017.1 mm on average based on 13 years of data (Austrian Wine Marketing

Board, 2018; Duiker, 2020).

The wine regions of Austria that predominately grow GV, namely Niederösterreich

(Lower Austria), Wien (Vienna), and Burgenland, are characterized by a variety of soil types, and

GV thrives in the loess and loam soils of these regions (Brook, 2019). PA regions also contain a variety of soil types, including limestone and silt loams, most of which are well drained and suited for agricultural use (Duiker, 2003). In agricultural production, heat accumulation is typically reported in growing degree days (GDD). GDD is on average higher in southern regions of PA compared to the central and North West regions. Historically, the North West region will

14 accumulate 270-386 GDD (10 °C) in each of the summer months, while the South East will accumulate 335-345 GDD (10 °C) in these months (M. Centinari, personal communcation, July 3,

2020). Austrian growers typically use average sunlight hours instead of GDD to examine growing season patterns, with the country’s wine regions receiving between 1,815.7 and 2,185.2 sunlight hours annually in recent years (Austrian Wine Marketing Board, 2018).

Winter temperatures can become very low in PA growing regions, which can impart bud damage to vinifera that are not as cold hardy as inter-specific Vitis hybrids or Vitis Labrusca. GV damage has been reported in PA with temperatures below 0°F, and particularly cold periods have resulted in temperatures as low as -11°F in PA regions where GV is currently grown (Pompilio,

2015). However, in general, PA and Austria’s similar climate of warmer, wet summers and relatedly short growing season which provides favorable growing conditions to produce high quality GV wines in PA.

1.5 Purpose and Significance

The wine industry in PA has consistently grown in both wine production and number of wineries in recent years, ranking as the 5th largest wine producing state in 2018 (Economic Impact of the Pennsylvania Wine and Grape Industries, 2018). In the same year, the industry generated

$4.8 billion in economic activity for the state, including $476.5 million in tourism revenue

(Economic Impact of the Pennsylvania Wine and Grape Industries, 2018). The success of the PA industry does not come without challenges, however. These challenges include excessive precipitation, winter and late spring cold damage, and pest pressures that the industry must overcome while producing wines from high-quality grapes in order to remain competitive in the national wine market (Duke, 2017).

15 While GV is growing in popularity among PA wine producers, there are few studies on the variety itself. Currently, GV wine produced in the US is understudied and has not been studied from a regional perspective. Examining both the chemical and sensory differences in PA

GV wines will give new insight into the variety and potentially unveil chemical factors that correlate to quality aspects in finished wines. Understanding key sensory attributes of GV made in PA may be useful for wineries and tasting room managers to describe GV wines to consumers and increase recognition of the variety.

Additionally, researchers have only begun to study the wine regions of PA in recent years, and regional profiles for PA wines have yet to be uncovered. Defining regional characteristics for the PA wine industry can be useful in marketing specific characteristics of wines from a designated region and building a reputation for high quality wine produced in that area. Researchers have examined commercial PA white wines for regional trends and found wines from the northwest region to be sweeter; however, winemaker style likely influenced final wine composition and masked regional characteristics (Donohue, 2020). Examining regionality through a controlled winemaking study will mitigate this effect and allow for a more direct examination of regional wine characteristics of PA.

1.6 Hypothesis and Objectives

I hypothesize that GV wines made under controlled conditions with grapes grown in different geographic regions of PA will have greater variation in perception of appearance, aroma, and flavor, as well as greater variation in concentration of grape- and fermentation- derived volatile and non-volatile compounds. I also hypothesize that wines made from grapes grown in the same region will have fewer differences in these sensory attributes and these volatile and non-volatile compounds.

16 To test these hypotheses, I propose to:

1. Evaluate the perceived intensities of appearance, aroma, taste, mouthfeel, and flavor

of these wines using a trained descriptive analysis panel

2. Use instrumental analysis, including solid phase microextraction – gas

chromatography – mass spectrometry (HS-SPME-GC-MS) and high-performance

liquid chromatography (HPLC) with diode array detection (DAD), to corroborate

findings revealed by sensory analysis

17 Chapter 2

Materials and Methods

2.1 Site Selection

Experimental sites were chosen in order to capture climate variability of the growing regions and to be representative of the various PA regions that currently grow Grüner Veltliner

(GV). A total of 9 sites in four PA growing regions were chosen for the experiment: two in the

Northwest (NW), one in the North Central region (NC), one in the Northeast (NE), and five in the

Southeast region (SC). A map of the sites within the growing regions is provided in Figure 2.1.

Canopy management practices including shoot thinning, leaf pulling, and cluster thinning were standardized to minimize confounding effects from vine management practices during the growing season. A HOBOÒ weather sensor and datalogger (Onset Computer Corporation,

Bourne, MA) recorded vineyard air temperature, solar radiation, and rainfall measurements in 15- minute intervals during both growing seasons (May 1 to October 31) at all sites except NW2 and

NC1. At sites NW2 and NC1, weather data was collected from on-site stations that were associated with the Network for Environment and Weather Applications (NEWA; http://newa.cornell.edu) during both growing seasons.

Severe winter freeze damage occurred at multiple sites between growing seasons. At least three sites recorded temperatures below -5°F during a polar vortex event in the winter months of

January and February 2019 that caused significant primary bud damage at these sites. This resulted in the removal of NE1, SE4, and SE5 in Year 2 of the study. In Year 2, there were two harvests at site NW1, and the second harvest was treated as a separate site (NW3). It should also be noted that sites SE2 and SE3 are located at the same commercial vineyard; however, grapes for these sites come from different areas of the vineyard property. The SE2 plot is on a slope

18 located at the edge of a wooded area with northeast-southwest row orientation, while the SE2 plot is located on a hill with east-west row orientation.

Figure 2.1 Location of experimental sites within PA wine regions recognized by the PA Winery Association (PWA; pennsylvaniawine.com).

2.2 Vinification

Fruit was hand harvested at each experimental site between September 14 and September

24, 2018 for Year 1, and September 11 and October 2, 2019 for Year 2. In Year 1, yield ranged from 6.9 to 120.4 kg, and in Year 2 yield ranged from 61.5 to 140.1 kg. The harvested fruit was transported to the Department of Food Science at The Pennsylvania State University (University

Park, PA) where it was stored for less than 48 hours at 4°C before processing. Fruit weights were measured to estimate yield. Fruit was crushed and destemmed, and the must was pressed using a bladder press to a yield of 15 gallons of juice per 113.4 kg of grapes. After pressing, one 50 mL sample was collected for total soluble solids (TSS, pH, titratable acidity (TA), and yeast assimilable nitrogen (YAN) measurements. An additional 50 mL sample was frozen at

-80 °C for aroma analysis. Juice was treated with Cinn Free (Scott Laboratories, Petaluma, CA), a pectinolytic enzyme, at a rate of 15 µL/L. Juice was also treated with SO2 prior to cold settling;

19 adjustment rates were dependent on cluster rot assessment and ranged from 30-50 ppm in Year 1 and 30-40 ppm in Year 2.

After cold settling overnight, juice was racked and chaptalized with sucrose to achieve a final TSS level of 22 °Brix. If a given juice was above pH 3.4, its pH was adjusted to 3.4 using tartaric acid (Presque Isle Wine Cellars, Norther East, PA). The adjusted juice was racked into two 5-gallon glass carboys prior to inoculation. Due to low juice yields for some sites in Year 1, two experimental sites were fermented in 1-gallon glass jars and one site was fermented in 3- gallon carboys. In Year 2, one of the 7 sites was fermented in 3-gallon carboys, and another one was fermented in triplicate due to large juice volume. Each fermentation replicate was inoculated with Saccharomyces cerevisiae EC-1118 yeast (Lallemand, Petaluma, CA) at a rate of 0.25 g/L with an addition of 0.30 g/L Go-Ferm nutrient (Lallemand). Fermentations were temperature controlled in Year 1 using glycol jackets (Gotta Brew, Sacramento, CA) set at 15°C and in Year 2 using a chilled water bath system with a glycol chiller set to 12.8°C. Alcoholic fermentation was monitored daily by TSS readings via hydrometry and temperature readings via thermocouple

(Fluke Corporation, Everett, WA). YAN adjustments were performed 24 hours post-inoculation and at one-third sugar depletion using Fermaid K nutrient (Lallemand) to achieve a final concentration of 0.25 g/L.

When TSS values measured below 0 °Brix, as assessed by hydrometry, AimTab

Reducing Substance tablets (Germaine Laboratories, San Antonio, TX) were used to confirm dryness of wines, defined as £1% residual sugar. Fermentations were held at 4 °C for cold settling, and were subsequently racked off lees, treated with the enzyme blend Scottzyme KS

(Scott Laboratories) at a rate of 79.25 µL/L, and SO2 was adjusted to achieve 0.85 mg/L molecular SO2. SO2 levels were monitored weekly via aeration oxidation and Ripper methods in

Year 1, and by way of a Vinmetrica kit (Vinmetrica, Carlsbad, CA) in Year 2. A sample of each

20 replicate (250 mL) was collected for basic wine chemistry analysis. Final SO2 adjustments were made based on these results one day before bottling. Replicates were manually bottled in 375 mL green glass bottles and sealed with Saranex-lined Stelvin screw cap closures. Bottled wines were stored at 3°C until sensory evaluation and chemical analysis (within three months).

2.3 Descriptive Analysis

Descriptive analysis protocol was evaluated by The Pennsylvania State University

Institutional Review Board (protocol #STUDY00008551) and found to be exempt under category

6 (Taste and Food Quality Evaluation). A panel of 8 participants was trained to evaluate Year 1 wines three months after bottling. The participants (4 females; ages 24-60) were selected from a pool of previous alcoholic beverage panels conducted at the Sensory Evaluation Center at The

Pennsylvania State University. Panelists were trained for 14 hours over a 4-week period on aroma, flavor (in-mouth aroma), taste, and mouthfeel attributes found in the wines using generic descriptive analysis (DA) (Lawless & Heymann, 2010a). Panelists generated terms and agreed upon corresponding physical and verbal references (see Table 2.1A and B) for each attribute in the training sessions. Attribute terms were generated by blindly exposing the panelists to the different wines, with each wine replicate presented at least once during training.

The panel developed a list of two appearance, 15 aroma, 5 taste, three mouthfeel, and 15 flavor attributes. Two of the attributes (citrus aroma/flavor and other fruit aroma/flavor) used two reference standards to describe the term. After the attribute list was generated, panelists were trained on attribute rating and scaling. Intensities of each attribute were rated on line scales that had anchor terms on both ends. Appearance attributes (yellow color and haziness) were anchored with the terms “No (yellow/haze)” at the left end of the scale to “Very (yellow/hazy)” at the right end. Aroma, taste, mouthfeel, and flavor attributes were anchored with “None” at the end of the

21 scale to “A lot” at the right end. For the complete questionnaire and example of scales, see

Appendix A1. For initial training on scale usage, the line scales were split into four equal sections by tick marks, but these were removed for sample evaluation in the sensory booths.

Blind duplicate samples were used to evaluate panel performance and rating agreement.

Panelists were considered trained when they could rate blind duplicate samples and find no significant differences (p < 0.05) between them using analysis of variance (ANOVA). Fourteen of the 18 wine samples (= 9 sites x 2 fermentation replicates) were evaluated in sensory triplicate, while four wines were evaluated in sensory duplicate due to low sample volume. Panelists evaluated samples in 7 sessions over a period of three weeks. Evaluation was conducted in individual booths in the Penn State Sensory Evaluation Center under white light and positive pressure and results were collected using Compusense Cloud software (Academic Consortium,

Guelph, ONT). Panelists were presented with 30 mL of each sample at room temperature in clear

ISO certified tasting glasses covered with clear plastic Petri dishes. Each panelist evaluated 6 samples per session that were labeled with randomized three-digit blinding codes and presented in a modified Williams Latin square block design. Panelists were prompted to cleanse their palate with water and unsalted crackers (Mondelēz Global LLC, East Hanover, NJ) that were provided during a forced 60 second break between samples. One panelist missed two evaluation sessions, and the missing values were imputed using the panelist’s mean replicate values.

Six of the panelists from Year 1 were able to participate in evaluation of Year 2 wines.

Three additional individuals were screened for previous descriptive analysis experience for a total of 9 panelists (6 females; ages 32-60). Panel training started one month after wines were bottled, and training sessions were held over a 5-week period in which panelists received approximately

19 hours of training.

The panelists were trained on the same reference standards that were used in the Year 1

DA. The panel was given the option to add references to this list, but after being presented with

22 all replicates of Year 2 wines, the panel determined additional references were not needed.

Panelists were asked to identify the aroma references blindly, and results from this exercise were used to determine the references for which the panel needed additional training. The following sessions focused on a few attributes at a time, and attributes were paired together in training if there was confusion among the panelists on those attributes. For example, pear and green apple references were paired together and discussed after evaluating a “pure” sample, the aroma reference standard, and a sample of control GV wine spiked with that reference. Panelists were prompted to discuss how they determined the differences between the references.

The final training sessions were used to introduce panelists to scaling and scale usage, as well as blindly identify spiked wine samples. The panel was deemed trained when they evaluated blind duplicate samples and found no significant differences between them.

Sample evaluation took place over 5 sessions in which panelists evaluated samples in sensory triplicate using Compusense Cloud software, with fermentation replicates treated as separate samples. A modified Williams Latin square block design was used for evaluation.

Panelists were presented with 30 mL of each sample at room temperature in clear ISO certified tasting glasses covered with clear plastic Petri dishes. Panelists evaluated 10 samples per session for the first four sessions and evaluated 5 samples during the last session. Panelists evaluated samples in individual booths in the Sensory Evaluation Center at Penn State under white light.

Panelists received 5 samples at a time with a forced 60 second break between samples, and a forced break of 10 minutes between two sets of 5 samples. Panelists were encouraged to cleanse their palate with water and unsalted crackers during the forced 60 second breaks, as well as leave the booth area during the 10-minute break. Analysis of sensory data was completed using R

Studio software (version 1.1.463, Boston, MA) and the SensoMineR package (version 1.23) (Le

& Husson, 2008) for principal component analysis (PCA).

23 Table 2.1A Appearance, aroma, and flavor reference standards used for training of sensory attributes for Year 1 and Year 2 DA. All standards were created in Bota Box Pinot Grigio (Bota Box Vineyards, Manteca, CA) base wine unless otherwise noted. Attribute Reference Name Reference Standard Appearance Yellow Color Rated on line scale with anchors “No Yellow” to “Very Yellow” Haziness Rated on line scale with anchors “No Haze” to “Very Hazy” Aroma & Flavor Green Apple Green Apple 10.0 g fresh granny smith apple (Wegmans, State College, PA) in 20 mL wine Pear Pear 25.0 g fresh Bartlett pear (Wegmans, State College, PA) in 25 mL wine Other Fruit Stone Fruit 10.0 g fresh nectarine and 10.0 g fresh peach (Wegmans, State College, PA) in 40 mL wine Mixed Fruit 10 mL canned Wegman's Fruit Cocktail juice (Wegmans, State College, PA) in 20 mL wine Grape Grape 3 halved fresh green grapes (Wegmans, State College, PA) in 20 mL wine Citrus Lemon 3x2 cm fresh lemon peel (Wegmans, State College, PA) in 20 mL wine Orange/Grapefruit 3x2 cm fresh orange peel and 3x2 cm fresh grapefruit peel (Wegmans, State College, PA) in 40 mL wine Floral Floral 4 drops of Floral stock in 20 mL wine (Floral stock: 1 drop lavender essential oil (Aura Cacia, Norway, IA) in 25 mL wine) Earthy Earthy 1.0 g soil (Indoor Potting Mix, Miracle Gro Lawn Products Inc., Marysville, OH) in 20 mL wine Thiol Thiol 0.075 mL 30 µM 4-methyl-4-mercaptopentan-2-one in 50 mL RO water Canned Vegetable Canned Vegetable 2.5 mL canned pea juice and 2.5 mL canned green bean juice (Wegmans, State College, PA) in 20 mL wine Rotten Egg Rotten Egg 0.4 g hardboiled egg yolk in 20 mL base wine Sulfur Sulfur 2 pinches potassium metabisulfite (Presque Isle Wine Cellars, North East, PA) in 25 mL RO water Yeasty Yeasty 2 pinches baker's yeast (Fleischman's ActiveDry Yeast, ACH Food Companies, Memphis, TN) in 5 mL RO water Oxidized Oxidized 5 mL dry sherry (Taylor Wine Company, Canandaigua, NY) in 20 mL wine Chemical/Solvent Chemical/Solvent 1 drop ethyl acetate (VWR International, Radnor, PA) in 50 mL wine Ethanol Ethanol 10% (v/v) ethanol (Decon Labs, Inc., King of Prussia, PA) solution

24

Table 2.1B Taste and mouthfeel reference standards used for training of sensory attributes for Year 1 and Year 2 DA. All standards were created in Bota Box Pinot Grigio (Bota Box Vineyards, Manteca, CA) base wine unless otherwise noted. Attribute Reference Name Reference Standard Taste & Mouthfeel Sour Sour 1 1.5 g/L tartaric acid (≥99.7%, Sigma-Aldrich, St. Louis, MO) in RO water Sour 2 2.0 g/L malic acid (≥99%, Sigma-Aldrich) in RO water Sweet Sweet 20.0 g/L sucrose (Domino Foods, Inc., Yonkers, NY) in RO water Salty Salty 2.0 g/L salt (Morton Salt, Chicago, IL) in RO water Bitter Bitter 0.8 g/L caffeine (Sigma-Aldrich) in RO water Umami Umami 5.0 g/L monosodium glutamate (B&G Foods, Inc., Parsippany-Troy Hills, NJ) in RO water Viscous/Thick Viscous/Thick 1.0 g/L carboxymethyl cellulose (Tic Gums, Belcamp, MD) in RO water Astringent Astringent 0.8 g/L alum (McCormick, Hunt Valley, MD) in RO water Warm/Hot Warm/Hot 6% (v/v) ethanol (Decon Labs, Inc.) solution

25

2.4 Chemical Analysis

2.4.1 Juice and Wine Chemical Analysis

Juice samples were analyzed for TSS, pH, TA, and YAN before chaptalization and acid adjustment. TSS was measured using a handheld refractometer (Master, Atago, Nellevue, WA). pH was measured using an Orion Star A111 pH meter (Thermo Fisher Scientific, Waltham, MA).

TA was measured using an autotitrator (G20, Mettler Toledo, Columbus, OH). A 10 mL sample of juice was diluted to 40 mL with deionized water and titrated to a pH of 8.2 using 0.1 N sodium hydroxide, and results were recorded in g/L tartaric acid equivalents. YAN was measured by an enzymatic assay (Vintessential Laboratories, Victoria, AUS).

Wines were sampled prior to bottling for basic wine chemical analysis, including residual sugar (RS), ethanol content, volatile acidity (VA), free and total sulfur content, TA, pH, lactic acid, and malate content by the Cornell Craft Beverage Analytical Laboratory (Geneva, NY) according to validated methods.

2.4.2 Aroma Analysis of Juice via HS-SPME-GC-MS

Headspace – solid phase microextraction – gas chromatography – mass spectrometry

(HS-SPME-GC-MS) was used to determine aroma composition of wine and juice samples. An

Agilent Technologies 7890B GC 5977B MS (Santa Clara, CA) equipped with an MPS autosampler (Gerstel, Inc., Linthicum, MD) and Rtx-Wax 30 m x 0.25 mm x 0.25 µm GC column

(Restek, Bellefonte, PA) was used for sample analysis. Agilent MassHunter GC/MS software and

26 Gerstel MAESTRO software was used to control the GC and autosampler, respectively. Sample preparation methods were based off work from Pedneault et al. with some modifications

(Pedneault et al., 2013). For sample preparation, 3 g sodium chloride (>99.9%, Dot Scientific,

Burton, MI) and 0.5 g d-gluconic acid lactone (Sigma-Aldrich) were added to 20 mL crimp top headspace vials (Restek). Next, 2 mL of juice and 10 µL of internal standards d8-napthalene and

2-octanol (both ³90%, Sigma-Aldrich) (9.9 mg/L d8-napthalene and 13.7 mg/L 2-octanol in

100% HPLC grade methanol, Fisher Scientific) were added to the vials before being capped and vortexed for 20 seconds. Samples were prepared in triplicate for each experimental site.

Samples were incubated at 30°C for 5 minutes and then extracted for 30 minutes at the same temperature using a 2 cm DVB/CAR/PDMS SPME fiber (Supelco, Bellefonte, PA). The

SPME fiber was desorbed at 250°C for 10 minutes in splitless mode. Data was analyzed using

PARADISe software (version 3.1) (Johnsen et al., 2017) and concentrations were reported in internal standard (IS) equivalents.

2.4.3 Aroma Analysis of Wine via HS-SPME-GC-MS

For sample preparation, 2 mL of wine and 50 µL of internal standard (as described in

2.4.2) were added to 20 mL crimp cap headspace vials (Restek) containing 2 g of sodium chloride

(Dot Scientific). Analysis was completed using the method and instrumentation described in

2.4.2. Sampling occurred during the DA evaluation sessions so analysis could be completed on sensory samples. In Year 1, samples for aroma analysis were prepared in 6 replicates (2 replicates from each sample evaluation day), and in Year 2 samples were prepared in 5 replicates (1 replicate from each evaluation day).

27 2.4.4 Phenolic Analysis of Wine via HPLC

Individual phenolics were quantified using high performance liquid chromatography

(HPLC). A Shimadzu HPLC (Shimadzu, Torrance, CA) equipped with LC-10ADvp pumps, a reverse phase Eclipse Plus C18 column (4.6 x 250 mm, 5 µm; Agilent Technologies, Santa Clara,

CA), SPD-M10Avp controller, DGU-14A degasser, SCL-10Avp UV-DAD detector, SIL-20AC

HT autosampler, and Brinkman CH-30 column heater with an Eppendorf TC-45 temperature controller was used for analysis. The binary mobile phase system consisted of 1% phosphoric acid in water (A) and methanol (B).

Samples were centrifuged at 13,500 x g for 5 minutes (Microfuge 16, Beckman Coulter,

Indianapolis, IN) and 200 µL of sample was transferred into 11 mm glass HPLC vials (VWR,

Radnor, PA) with 250 µL reduced volume vial inserts (6 x 31 mm; MicroSolv Technology

Corporation, Leland, NC) and 11 mm snap caps with PTFE/red rubber septum (VWR, Radnor,

PA). Injection volume was 20 µL, flow rate was 0.7 mL per minute, and column temperature was

30°C. Phenolics were eluted using the following sequence: 0-5 min, 5% B; 7 min, 15% B; 10 min, 17.5% B; 22 min, 19.75% B; 26-33 min, 50% B; 34-35 min, 85% B, and re-equilibration at

5% B for 7 min before the next injection. This method was modified and adapted from Sheridan and Elias (Sheridan & Elias, 2016).

Samples were prepared in single vials with each sample injected twice. Gallic acid, (+)- catechin, and (-)-epicatechin were chosen as phenolic standards based on previous work in identifying phenolic compounds in white wine and juice (Betés-Saura et al., 1996). Detection was completed at 210 and 280 nm; however, more consistent data was collected at 210 nm and thus, used for quantitation, while data collected at 280 nm served as the qualifier wavelength.

28 2.4.5 Color Analysis of Wines via CIE-Lab Color Measurements

Chromatic characteristics of wine samples were obtained following the protocol of the

Compendium of International Analysis of Methods of Wine and Must Analysis (OIV, 2006).

Samples were centrifuged at 2500 x g for 5 minutes and transferred into 10 mm glass cuvettes.

Transmittance of the sample was measured every 5 nm from 380 to 780 nm, with Illuminant D65 and the 10° Observer used as standard conditions. Ultrapure water was used as a blank and samples were measured in triplicate. Transmittance values were converted to the colorimetric coordinates L*, a*, and b* (Lindbloom, 2011).

2.5 Statistical Analysis

Statistical analyses were performed using RStudio software (version 1.1.463) (RStudio

Team, 2015). For analysis of sensory data, ANOVA was carried out with wine, judge, fermentation replicate and all two-way interactions to determine significant attribute differences among the samples with wine, judge, and fermentation replicate treated as main, fixed effects. A pseudo-mixed model was used to calculate F-ratio for samples that had significant wine-by-judge interaction in addition to a significant wine effect (Gay, 1998). ANOVA with region as the main effect was then used in combination with Least Square Means (LS Means) post-hoc comparison to examine potential regional differences in the attributes, using the emmeans package (version

1.4.6) (Lenth et al., 2020). The SensoMineR package (version 1.23) (Le & Husson, 2008) was used for principal component analysis (PCA). The panellipse.session function in the SensoMineR

R package was used for PCA, with fermentation replicate being used as the session factor. These statistical methods were also used for juice and wine aroma analyses.

29 Statistical analysis for basic wine chemical analysis was completed using one-way

ANOVA with wine as a main factor. ANOVA with region as the main factor, and the agricolae package (version 1.2-8) was used for Tukey’s post-hoc comparison to examined potential regional differences (de Mendiburu, 2017). For phenolic analysis, gallic acid, catechin, and epicatechin were quantified in wine samples at 210 nm. One-way ANOVA was used to determine if significant differences in relative abundance for the compounds existed between wine samples and between regions.

L*, a*, and b* coordinates for wine samples were obtained from transmittance spectra.

While ANOVA could be used to determine differences in each color value, a significant ANOVA result does not necessarily reflect a significant visual difference in color between samples.

Instead, calculation of Euclidean distance (DE) between two samples using L*, a*, and b* values provides a numerical value that can be translated to detection of a visual color difference

(Mokrzycki & Tatol, 2011).

Correlation of sensory and instrumental data was measured using the Pearson linear correlation coefficient. The ggcorplot (version 0.1.3) and ggplot2 (version 3.1.0) packages were used to complete correlation analysis (Kassambara, 2019; Wickham et al., 2020). Significance for all analyses was defined as p < 0.05.

30

Chapter 3

Results and Discussion

3.1 Descriptive Analysis

3.1.1 Year 1

The trained panel found the wines to differ significantly (p < 0.05) in both appearance attributes, three of the 15 aroma attributes, two of the 8 taste and mouthfeel attributes, and two of the 15 flavor attributes. There was a significant (p < 0.05) fermentation replicate effect for haziness and sulfur aroma. A one-way ANOVA with region as a factor was then used, in combination with Least Square Means (LS Means) post-hoc comparison, to examine potential regional differences in the attributes.

Among the significantly different attributes, only yellow color and haziness were different by region. A summary of the significant region effects for each wine is provided in

Table 3.1. Wines from the NW region were rated significantly higher in yellow color than the other regions (5.52 vs. 2.51, 2.87, and 4.20 on a 10-point scale). Additionally, wines from the NE region were significantly higher in yellow color than those from the SE and NC regions (4.20 vs.

2.51 and 2.87). Grape maturity at harvest may have affected the resulting wine color, with more mature grapes producing wines more yellow in color. Previous work on white wines has found grapes harvested at an earlier date to be lighter in color and less intense, as indicated by higher L* and lower C* and b* values (Gómez-Míguez et al., 2007). NW2 and NE1 were the last sites to be harvested and had two of the highest TSS levels (19.6 and 16.4 °Brix, respectively), indicating these sites provided more mature grapes for vinification compared to sites harvested earlier at

31 lower TSS levels. Potential weather effects may have also contributed to the regional color differences found. Elevated temperatures in these regions could contribute to accumulation of phenolic compounds that influence wine color. Research has shown that increased daytime temperature in the vineyard can increase berry color ratings in a white vinifera variety (Sadras et al., 2013), which affect final wine color.

Wines from the NW region were significantly less hazy than the wines from the other 3 regions, while NC, SE, and NE wines did not differ in haziness ratings. Haze formation in wines is often attributed to the presence of proteins after fermentation, exacerbated by elevated storage temperature (Van Sluyter et al., 2015). Previous work has suggested that protein accumulation and composition in wine grapes can be dependent on viticultural factors such as disease pressure or environmental conditions (Girbau et al., 2004; Monteiro et al., 2003). When visually inspected, grapes from the NC, SE, and NE regions appeared to have higher rot levels at harvest. It is possible that unfavorable conditions such as higher rainfall and humidity increased disease pressure in these regions, resulting in higher protein accumulation in grapes and more haze in these wines.

Principal Component Analysis (PCA) allows us to examine potential regional differences visually while displaying the greatest percentage of variance in the data. All 7 significant attributes were used in the PCA, which captured 80.99% of sample variation within the first two dimensions. Some regional grouping among the wines were noted in the PCA biplot (Figure

3.1A). NW wines, located in the top left quadrant of the plot, separated from other wines but were not found to be different from each other, seen in the overlapping 95% confidence ellipses between the NW1 and NW2 wines. NE wine, located in the top right quadrant, separated from other samples as well. Three wines from the SE region (SE3, 4, and 5) were not found to be different from each other, as indicated by overlapping 95% CI ellipses, and the NE wine was not different from the SE3 and SE2 wines. While not all wines from the SE region have overlapping

32 confidence ellipses, four of the five SE wines were located in the bottom quadrants, negatively loaded on PC 2.

Wines were separated by haziness along PC 1, with wines high in haziness located on the right side of the plot. NW wines were rated higher in yellow color and lower in haziness than the other wines, and low haziness ratings drive their negative loading along PC 1. Wines from the other regions were higher in haziness, and so show a positive correlation to PC 1. SE1 had the highest haziness ratings among all samples and is positioned to the farthest right-hand side of PC

1; this was likely in part due to an excessive addition of sulfur to those wines prior to bottling.

Although sulfur aroma ratings were not significantly different by wine, SE1 wines were rated higher in sulfur than all sites with the exception of SE5, which had the same average sulfur aroma rating as SE1 (2.79).

While appearance differences drive sample separation along the first dimension, sour taste is driving sample separation along PC 2. NE wine, located in the first quadrant, was rated highest in sour taste, while wines lower in sour taste ratings, i.e., SE3 and SE4, are loaded negatively on PC 2.

When examining variables along the dimensions, we can see yellow color and haziness ratings are negatively correlated to each other, indicated in their opposing direction on the biplot.

Sour taste and warm/hot mouthfeel are also negatively correlated to each other. Attributes that form 90° angles are statistically independent of each other, such as yellow color and ethanol flavor.

It is likely that haziness is obscuring regional differences in this analysis due to the excess sulfur addition to SE1 wines. Therefore, the data was re-analyzed by PCA excluding the haziness attribute. This PCA (Figure 3.1B) captured 74.86% of the total variance in two dimensions, and so it is still a suitable visualization of the data, even though less variance is captured due to haziness left out of the analysis. Examining this biplot (Figure 3.1B), we see

33 more consistent regional trends among the wines without the haziness attribute obscuring sample differences. The wines from the NC region, located in the bottom left quadrant of the biplot, separated from all other wine samples. While the two NW wines are significantly different from each other, as indicated by the not overlapping 95% CI ellipses, the wines were both positively loaded on PC 1. Wines NE1 and NW2 were not significantly different from each other and showed overlapping CI ellipses, and similarly, wines SE5 and NE1 showed not significantly different sensory profiles as indicated by their overlapping CI ellipses. All 5 SE wines, with the exception of SE5, were positively loaded on PC 2, and three of the 5 samples (SE1, 2, and 3) were located on the negative PC 1 axis.

Yellow color is a clear driver of separation between the different wines from the different regions along the first dimension, PC 1. Wines that were rated highest in yellow color (NW1 and

NW2) are positively loaded along the first dimension, while wines from the SE, located along the negative PC 1 axis, were lowest in yellow color. Conversely, thiol aroma and flavor, ethanol flavor, and warm/hot mouthfeel as opposed to sour taste are driving separation along the second dimension (PC 2); the three samples which were rated highest in these attributes (SE1, 4, and 3), are positively loaded on PC 2, while the NC wine was rated highest in sour taste, and thus, was located on the negative PC 2 axis. Sour taste showed a negative correlation to warm/hot mouthfeel, which was significantly loaded onto PC2, and ethanol flavor.

Table 3.1 Significant differences between PA regions for Year 1 GV wines. Values that share the same letter within column are not significantly different according to Tukey’s post-hoc comparison (p < 0.05). Yellow Color Haziness NC (n=2) 2.51a 2.73b NE (n=2) 4.20b 4.10b NW (n=4) 5.52c 1.29a SE (n=10) 2.87a 3.43b

34

A B

Figure 3.1 PCA biplots for Year 1 GV wines A) with the haziness and B) without the haziness attribute. Wines are shown with their 95% confidence ellipses and colored by region. Attributes ending in A indicate aroma, F indicate flavor, MF indicate mouthfeel, and T indicate taste attributes.

3.1.2 Year 2

In Year 2, the trained panel found the wines to differ significantly in both appearance attributes, three aroma attributes, four taste and mouthfeel attributes, and three flavor attributes.

There was a significant fermentation replicate effect for green apple flavor. Yellow color, haziness, thiol aroma and flavor, and sour taste were significantly different in both years (see

Appendix Table 2).

Regional differences were examined again using ANOVA and LS Means post-hoc comparison. A summary of the significantly different attributes and LS Means results for Year 2 are provided in Table 3.2. All of the significantly different wine attributes, with the exception of floral and canned vegetable aromas, also showed significant differences by region (p <0.05).

Wines from the NW region (n=6) were again found to be highest in yellow color ratings and were rated significantly higher than wines from the NC and SE regions (Table 3.2).

35 While grape maturity likely contributed to higher yellow color ratings for NW wines in

Year 1, fruit harvested from two of the three NW sites were least mature in Year 2 (see Appendix

Table 4). However, since these wines had higher acidity, less sulfur dioxide was added to these wines after fermentation since less sulfur dioxide was needed to achieve 0.8 g/L molecular SO2 at the lower pH values. Since there was less sulfur in these wines to act as an antioxidant, it is possible oxidation reactions caused a darker, more yellow color to form in these wines, resulting in panelists rating NW wines highest in yellow color for Year 2 despite the fruit being less mature. NW wines were also rated as significantly less hazy than SE wines, which is consistent with Year 1 results. In Year 1, haziness and yellow color were negatively correlated to each other, however, these attributes were independent of each other in Year 2. NW wines were also rated highest in sour taste, astringent mouthfeel, citrus flavor, and green apple flavor, and differed significantly in sour taste, sweet taste, and citrus flavor attributes from the NC and SE regions.

The NC sample was highest in sweet taste and differed significantly from wines from the

NW and SE regions. These ratings were expected, as the NW1 wine, followed by NW2 and

NW3, had the highest TA values among the samples (see Appendix Table 4), while both fermentation replicates of the NC region wine had the highest RS values. It is possible that panelists rated NW samples highest in citrus and green apple flavors due to the association of sourness with these attributes. Pearson’s correlation coefficient was 0.82 for green apple flavor and sour taste, and 0.89 for citrus flavor and sour taste. Last, the 3 SE wines were rated significantly higher in thiol aroma and flavor than wines from other regions (Table 3.2). This was again similar to the results from Year 1.

When examining the PCA biplot for Year 2 results, we can see similar sample separation trends compared to Year 1 (Figure 3.2). All significant attributes were used in the PCA, and

71.67% of variation was captured within the first two dimensions. The three NW wines again grouped together in the bottom right quadrant. The NC sample was again located in the bottom

36 left quadrant; however, this sample was not significantly different from the SE1 and SE2 wines as it was in Year 1. The last wine from the SE region, SE3, differed significantly from the other SE wines and was the only wine to be positively correlated to PC 2.

PC 1 captured 48.36% of the total sample variation. Both appearance attributes (yellow color, haziness), sour and salty taste, astringent mouthfeel, canned vegetable aroma, and citrus and green apple flavors were all positively loaded along PC 1. Sour and sweet tastes were negatively correlated with each other along PC 1, which is expected, as mixture suppression can cause sweet taste to be reduced in the presence of acids in food or beverages (Lawless &

Heymann, 2010b). Another explanation for this could be the ripeness levels of the grapes used for making these wines – grapes used in vinifying NW1 and NW3 wines were the least ripe of all sites (harvested at 14.0 and 16.6 °Brix, respectively, Appendix Table 6), which could explain some of the higher ratings for attributes associated with unripeness, such as canned vegetable aroma or sour taste. Research examining influence of fruit maturity in red wine varieties has found that fruit harvested at lower TSS results in wines higher in vegetal aroma and sourness ratings (Casassa et al., 2013; Sherman et al., 2017). Similar vegetal aroma differences were found in Fiano wines harvested at different maturities, however, this study compared normal harvest and late harvest maturity (Genovese et al., 2007).

Separation along PC 2, which captured 23.31% of sample variation, was driven by thiol aroma and flavor ratings, which showed a positive correlation to PC 2. When examining the significant attributes along with the samples on the biplot, we can see that the SE3 wine separated from the other wines, including those from the SE region, due to significantly higher ratings for thiol aroma. Volatile thiol compounds, specifically 3-mercaptohexanol (3MH) and 3- mercaptohexyl acetate (3MHA), can contribute to the tropical aroma of Marlborough Sauvignon blanc that is unique to the region (Benkwitz et al., 2012). Other work has shown that 4-mercapto-

4-methylpentan-2-one (4MMP) to be an important contributor to thiol aroma in Sauvignon blanc,

37 and concentrations of 4-MMP increased with berry maturity and mild water deficit stress (Peyrot des Gachons et al., 2005). It is possible that water stress due to weather conditions from veraison to harvest enhanced or diminished accumulation of precursors that provide thiol aroma in finished wines.

It should be noted that SE3 separated from SE2 in the PCA. Grapes for SE2 and SE3 wines originate from different vineyards at the same geographic site. This suggests that mesoclimatic differences can lead to aroma and flavor differences in GV wines. The two vineyards experience the same weather patterns, however, are located at different elevations (187 vs. 219 m above sea level). While this is a relatively small difference in elevation, air displacement can still vary among sites with relative elevation differences (Kurtural, 2003). The vineyards also differ in row orientation (northeast-southwest vs. east-west), which could impact fruit sun exposure and photosynthetically active radiation (PAR) levels in the canopy (D. Grifoni et al., 2008). These factors could be the reason for the observed sensory differences in the wines, as winemaking was the same and controlled in this experiment.

Table 3.2 Significant differences between PA regions for Year 2 GV wines. Values that share the same letter within column are not significantly different according to Tukey’s post-hoc comparison (p < 0.05). Attributes that end in F indicate flavor, A indicate aroma, T indicate taste, and MF indicate mouthfeel attributes. Yellow Color Haziness Green Apple_F Citrus_F Thiol_F NC (n=2) 2.44a 2.90ab 3.39a 2.30a 1.15a NW (n=6) 4.28c 3.13b 4.01b 3.51b 1.30a SE (n=7) 3.50b 2.10a 3.73ab 2.86a 1.76b Thiol_A Sour_T Sweet_T Salty_T Astringent_MF NC (n=2) 1.36a 3.69a 3.67c 1.29a 3.83a NW (n=6) 1.73a 5.09c 2.36a 1.70b 4.76b SE (n=7) 2.37b 4.32b 2.98b 1.50ab 4.39ab

38

Figure 3.2 PCA biplot for Year 2 GV wines. Wines are shown with their 95% confidence ellipses and colored by region. Attributes ending in A indicate aroma, F indicate flavor, MF indicate mouthfeel, and T indicate taste attributes.

When examined together, sensory results from both years of this study suggest differences between PA regions that produce GV wines. This study included a controlled winemaking protocol, which eliminated winemaker influence and allowed for a more direct examination of PA wine regional characteristics. While GV is most commonly grown in the SE

39 region, our study was able to include sites in other PA growing regions in which GV is becoming more popular to grow and produce.

Some consistent regional trends were found across both years of study, including NW wines rated as higher in yellow color and less hazy compared to the other growing regions. Thiol aroma and flavor were also more highly rated in SE wines in both years. Differences in thiol aroma ratings were found within the SE region, specifically in two wines produced with grapes from the same experimental site, suggesting mesoclimatic differences in the vineyard can influence GV aroma. While a number of volatile thiols have been identified as important varietal aroma compounds in white vinifera varieties such as Sauvignon blanc, Chardonnay,

Gewürztraminer, and Riesling (Green et al., 2011; Schüttler et al., 2015; Tominaga et al., 2000), the contribution of volatile thiols to GV aroma is understudied. Further examination of thiols in

GV could help gain understanding into which specific compounds may be present and how they may influence perceived wine aroma.

This study included wines from two vintages, which allows further examination of consistent weather patterns in wine regions. However, weather conditions in Year 1 were abnormally wet with high amounts of rainfall in multiple regions shortly before and during the harvest season, which may have obscured regional trends. For example, many Year 1 wines were harvested at lower TSS levels than usual, due to rainfall and wet conditions that prevented sugar accumulation and increased rot levels in the vineyard. A second year of study was included to provide a more accurate depiction of typical weather conditions across the growing regions.

However, a third year of study could give a more definitive characterization of weather patterns that influence final wine aroma and flavor.

40 3.2 Aroma Analysis of GV Juices by HS-SPME-GC-MS

3.2.1 Year 1

Of the 44 compounds detected by HS-SPME-GC-MS for juice samples, 32 were found to differ significantly in terms of concentration (measured in internal standard equivalents) across the juice samples from each experimental site, and 23 of the 32 compounds were validated using the NIST library GC database based on chromatographic retention indices (see Appendix Table

7). PCA was completed using the validated aroma compounds as variables (Figure 3.3), and

70.06% of variance was captured within the first two dimensions, with dimension 1 accounting for 50.13% and dimension 2 accounting for 19.93%.

The dimdesc function in the FactoMineR R package was used to determine which variables were driving separation along the principal components. Isobutanol and phenylethyl alcohol were the compounds driving separation along the first dimension. Other compounds that were correlated with the first dimension (r(7) > 0.80, p < 0.05) include hexyl acetate, isoamyl acetate, ethanol, hexanol, and methyl hexanoate. Separation along the second dimension was driven by hexanal, octanal, and acetic acid.

Juice samples did not have clear regional grouping in the PCA biplot. SE1, SE5, NW1, and NE1 juices were positive along PC 1 and so were more likely to contain higher amount of isobutanol, phenylethyl alcohol, and other compounds driving PC 1 separation. NW1, NC1, and

SE4 were more likely to be higher in hexanal and octanal, as these samples were positive along

PC 2. NE1 and NW2 were positively correlated with acetic acid.

ANOVA with region as a factor used in combination with LS Means found significant region effects (p < 0.05) for 5 of the aromatic compounds (Table 3.3). Juice from the NC region had significantly higher concentration of hexanal and 2-hexenal than other regions, which caused

41 the NC samples to be negative along the first dimension and positive along the second dimension.

The NC region had lower concentrations of acetoin than the other regions, while NE and NW juices were highest in acetoin concentration. SE and NE juices had higher isobutanol concentration than NC juice. Juices from the SE region were also higher in 2-hexen-1-ol acetate concentration than that from the NC region.

Hexanal Octanal

p-Cymene SE1 1-Hexanol SE4 Hexyl acetate NC1 Isoamyl acetate 2-Hexenal Ethyl vinyl carbinol 2-Heptenal 3-Hexen-1-ol Methyl hexanoate Linalool oxide Ethanol Nerol oxide Phenylethyl alcohol NW1 Methyl acetate Methyl butyrate SE5 Dim 2 (19.93%) Isobutanol SE3 NW2 1-Hexen-3-ol NE1

Methyl 2-methylbutyrate SE2 Acetoin

2-Hexen-1-ol acetate

Acetic acid

Dim 1 (50.13%)

Figure 3.3 PCA biplot with volatile compounds that were significantly different by juice for Year 1 GV juice samples.

42 Table 3.3 Significant differences between regions for Year 1 GV juice samples. Values that share a letter in column are not significantly different according to Tukey’s post-hoc comparison (p < 0.05). Hexanal Isobutanol 2-Hexenal Acetoin 2-hexen-1-ol acetate NC 0.4665b 0.0354a 4.44b 0.00989a 0.00593a NE 0.0731a 0.1519b 1.44a 0.05237c 0.02882ab NW 0.2143a 0.1149ab 2.68a 0.03904bc 0.01980ab SE 0.1851a 0.1220b 1.66a 0.03146b 0.03581b

3.2.2 Year 2

In Year 2, 27 volatile compounds were found to differ significantly (p < 0.05) in concentration among juice samples from each experimental site and validated using the NIST library GC database. PCA captured 76.68% of variance in the data, with dimension 1 accounting for 50.56% and dimension 2 capturing 26.12% (Figure 3.4). The compounds driving separation along PC 1 included 2-hexen-1-ol acetate, methyl hexanoate, acetoin, and acetic acid. Juice samples from NW2, SE1, and SE3 were positive along the first dimension and likely contained these compounds in higher concentrations. Samples that were negative along PC 1 were more likely to be higher in 2-hexenal, p-cymene, octanal, and hexanal as these compounds were negatively correlated with PC 1. Hotrienol was the primary driver of separation for PC 2, and SE2 and SE3 juices were likely highest in hotrienol and linalool oxide due to their location along the second dimension.

All SE samples were positive along the second dimension, and so were likely to be higher in hotrienol, linalool oxide, and ethyl vinyl carbinol. All NW samples were negative along this dimension, and so were likely higher in 2-ethylhexanol. NC1 was also negative along the second dimension. SE2 and SE3 samples appear to be more similar to each other than to SE1, which was expected as SE2 and SE3 sites are located on the same vineyard property, while SE1 is located at a different property in the region. NW1 and NW3 samples are from the same experimental site

43 but were harvested at different points. Because these samples were from the same site, it was expected that these samples would be more similar to each other than to the NW2 sample.

The concentrations of 9 volatile compounds were also significantly different by region (p

< 0.05) (Table 3.4). Juices from the SE region were significantly higher in 1-hexanol, 3-hexen-1- ol, and 2-hexen-1-ol than the other regions. These compounds give green, grassy aroma and are formed enzymatically when grapes undergo physical damage due to mechanical crushing

(Waterhouse et al., 2016). NC juice was significantly higher in 2-heptenal than other regions.

Octanal and p-cymene were significantly higher in NC juice compared to SE samples. SE samples were higher in ethyl vinyl carbinol and hotrienol compared to NW juice. These differences can explain why SE wines are positive and NC wines are negative along the first dimension.

Hotrienol SE2

Nerol oxide 1-Hexanol 3-Hexen-1-ol 2-Hexen-1-ol SE3 Ethyl vinyl carbinol Linalool oxide

1-Hexen-3-ol

2-Heptenal Methyl hexanoate SE1 Methyl 2-methylbutyrate Methyl butyrate

Dim 2 (26.12%) Hexyl acetate 2-Hexenal Isoamyl acetate Octanal Amyl acetate 2-Hexen-1-ol acetate p-Cymene NC1 NW1 Acetic acid Hexanal Acetoin Methyl acetate Ethanol Phenylethyl alcohol NW2 NW3 Isobutanol

2-ethylhexanol Dim 1 (50.56%)

Figure 3.4 PCA biplot for Year 2 samples with volatile compounds that were significantly different by juice.

44 Table 3.4 Significant differences between regions for Year 2 GV juice samples. Values that share a letter in column are not significantly different according to Tukey’s post-hoc comparison (p < 0.05). Ethyl vinyl carbinol p-Cymene Octanal 2-Heptenal 1-Hexanol NC 0.0212ab 0.0262b 0.00930b 0.0326b 4.58a NW 0.0235a 0.0223ab 0.00769ab 0.0134a 3.44a SE 0.0313b 0.0196a 0.00694a 0.0161a 6.53b

3-hexen-1-ol 2-hexen-1-ol 2-ethylhexanol Hotrienol NC 0.125a 2.22a 1.77ab 0.939ab NW 0.191a 2.40a 1.80b 0.334a SE 0.365b 5.32b 1.36a 1.612b

3.3 Basic Wine Chemistry Results

One-way ANOVA was used to evaluate regional differences in basic wine chemical parameters. Statistical analysis could not be conducted to examine differences between wines due to one analytical replicate value provided by the external analytical laboratory for each parameter.

In Year 1, regional differences were found in pH, malate, lactic, and total SO2 content

(Table 3.5). SO2 was added to wines based on pH to obtain 0.8 mg/L molecular SO2. The variation in total SO2 is likely due to a combination of pH differences between the wine samples as well as results from aeration oxidation. The aeration oxidation apparatus used in measuring free SO2 malfunctioned during use and gave false low readings and resulted in excess sulfur added to wines before bottling.

Malate and lactic acid concentrations also differed by region. Research has shown that elevated temperatures during veraison and ripening can reduce malate content in V. vinifera

(Sweetman et al., 2014). Temperature differences in the vineyard during veraison and ripening may have caused differences in malate content in GV juice. Malate is an organic acid that is metabolized during the grape ripening process and so can be an indicator of grape maturity throughout the growing season (Sweetman et al., 2009). Therefore, it is possible that GV grapes

45 from the NW and NE regions were more mature at harvest compared to other regions due to higher temperatures through veraison and ripening.

Table 3.5 Significant differences between regions for Year 1 GV wine chemistry parameters.

pH Malate (g/L) Lactic acid (g/L) Total SO2 (ppm) NC 3.280ab 2.40a 0.00b 138.50ab NE 3.170 ab 1.35bc 0.55a 122.50ab NW 3.098b 1.15c 0.45a 106.25b SE 3.282a 1.74b 0.04b 157.60a

In Year 2, alcohol content, RS, pH, TA, malate, and free and total SO2 content significantly differed by region (Table 3.6). NC wines were significantly higher in TA and lower in pH. This is likely due to gapes from the NW1 and NW3 site being less ripe at harvest than those from other sites. NC wines were higher in RS than those from other regions, however since juices were chaptalized this difference is likely due to an incomplete fermentation rather than a regional difference. Since juices were chaptalized, the significant difference in alcohol content is also likely due to slight variation in fermentation dynamics. NW wines had lower free and total

SO2 concentrations due to achieving 0.8 mg/L molecular SO2 at a low pH.

Table 3.6 Significant difference between regions for Year 2 GV wine chemistry parameters.

Alcohol (%) RS (g/L) pH TA (g/L) Free SO2 (ppm) Total SO2 (ppm) NC 12.70ab 8.95a 3.22a 8.33a 15.50a 92.00a NW 12.15b 2.13b 2.98b 6.65b 9.00b 57.50ab SE 12.74a 2.10b 3.17a 7.34b 10.43ab 89.14a

3.4 Aroma Analysis of GV Wines by HS-SPME-GC-MS

3.4.1 Year 1

Samples for aroma analysis were collected from the same bottles used in sensory evaluation in order to most accurately relate the chemical and sensory data. A total of 36

46 compounds were detected, and 12 were found to differ significantly (p < 0.05) among wine samples from each region. In the PCA, dimension 1 separation was caused by fermentation- derived esters, including ethyl decanoate, ethyl hexanoate, ethyl octanoate, as well as hexyl acetate and 2-hexen-1-ol acetate (Figure 3.5). These compounds contribute to a general fruity aroma in wine and include odors of apple and peach (Waterhouse et al., 2016). Dimension 2 separation was driven by octanoic acid and phenylethyl alcohol. Dimension 1 captured 52.75% of variance and dimension 2 captured 31.57%, resulting in 84.32% total variance accounted for in the PCA.

There were some regional trends displayed by the PCA, however, these trends did not necessarily reflect what was found in descriptive analysis. Both NW wines were negative along the second dimension, and NW2 was associated with higher phenylethyl alcohol content. All SE wines except SE2 were positive along PC 2. NC wine was more similar to some SE samples than to those from the NE region. NW wines were found to be different from each other, and NW1 was found to be more similar to SE2 than to the other NW sample. SE1 and SE5 were the only wines from the SE region to not significantly differ from another regional sample. This suggests that when examining aroma alone, regional characteristics are less clear than when examined along with appearance, taste, and flavor attributes.

Significant regional effects were found in 6 of the volatile compounds, the majority of which were driving sample separation in the PCA (Table 3.7). Phenylethyl alcohol and ethyl octanoate were found to be higher in concentration in NW wines compared to SE wines. While these compounds are associated with rose and peach aromas respectively, NW wines were not rated higher than all SE wines in floral aroma or other fruit aroma. The NC region was highest in hexyl acetate, explaining why the region is positive along PC 1. NC juice was found to have the highest concentrations of 2-hexenal and hexanal, both precursors for hexyl acetate, compared to other regions, which may explain why NC wine was highest in hexyl acetate. Ethyl hexanoate

47 was found to be higher in SE wines than NC and NW wines, and 1-hexanol was found in higher concentrations in the NE region compared to the NC region.

Figure 3.5 PCA biplot for Year 1 GV wines with volatile compounds that were significantly different by wine sample. Wines are shown with 95% confidence ellipses.

Table 3.7 Significant differences between regions for Year 1 GV wines. Internal standard equivalent values that share a letter in column are not significantly different according to Tukey’s post-hoc comparison (p < 0.05). Ethyl Hexyl 2-Hexen-1-ol Ethyl Phenylethyl hexanoate acetate acetate 1-Hexanol octanoate alcohol NC 3.02b 1.323b 0.0205b 0.442a 13.2ab 10.1ab NE 2.18ab 0.896a 0.0120ab 0.606b 10.0ab 10.7ab NW 2.80b 0.895a 0.0114a 0.535ab 13.1b 13.5b SE 2.15a 1.033a 0.0149ab 0.483ab 9.4a 10.6a

48 3.4.2 Year 2

Year 2 wines were found to differ significantly (p < 0.05) in concentration of 20 volatile compounds. These compounds were used as variables in the PCA, and it was determined that separation in dimension 1, which captured 43.86% of variance, was driven by acetic acid and acetate esters (hexyl, amyl, and butyl acetates). Dimension 2 captured 25.15% of variance and was highly correlated with ethyl butyrate and d-limonene. The PCA biplot for Year 2 wines is shown in Figure 3.6.

SE and NW wines separated along the first dimension, with SE wines characterized by higher concentrations of hexyl acetate, amyl acetate, and acetic acid and NW wines higher in methyl hexanoate and methyl octanoate. Wines from the NC region were similar to SE2. While wines from the same region were found to separate similarly along the first dimension, there were not clear regional trends along the second dimension. One of the 3 NW samples was negative along PC 2, while the others were positive. Conversely, 2 SE samples were negative along PC 2 while one was positive.

Fifteen of 20 compounds that differed by wine also differed significantly (p < 0.05) by region. Compounds that had significant region effect in combination with LS Means, along with

Tukey’s post-hoc comparison, are provided in Table 3.8. SE wines had significantly higher concentration of the following compounds compared to the NW and NC regions: ethyl acetate, 1- propanol, butyl acetate, isobutanol, amyl acetate, 1-hexanol, acetic acid, and 2-hexen-1-ol. While some of these compounds are associated with apple aroma, namely butyl acetate and amyl acetate, SE1 and SE3 were the two wines with the lowest green apple aroma ratings, suggesting these compounds are not contributing to perceived apple aroma. Although chemical/solvent aroma was not significantly different between the wines, the three SE samples had the highest chemical/solvent aroma ratings among the samples, which could be caused by their higher

49 concentrations of ethyl acetate. Separation along PC 1 is consistent with these findings, as SE wines were positive along PC 1 and many of these compounds were driving separation for this

PC. In contrast, SE wines were significantly lower in ethyl butyrate than NW wines. Wines from the NW region were lowest in 3-hexen-1-ol, which gives a grassy aroma in wines. NC wines were highest in methyl hexanoate and methyl octanoate, while NC wines were lowest in these compounds. Methyl octanoate can provide a range of odors including orange, herbal, and vegetable. Wines from the NW region were higher in canned vegetable aroma and citrus flavor, and it is possible methyl octanoate could be contributing to panelists rating these attributes higher in NW wines.

Figure 3.6 PCA biplot for Year 2 GV wines with volatile compounds that were significantly different by wine sample. Wines are shown with 95% confidence ellipses.

50 Table 3.8 Significant differences between regions for Year 2 GV wines. Internal standard equivalent values that share a letter in column are not significantly different according to Tukey’s post-hoc comparison (p < 0.05). Ethyl Ethyl acetate Isobutyl acetate butyrate 1-Propanol Buytl acetate NC 2.03a 0.1004ab 0.476ab 0.0816a 0.01116a NW 2.02a 0.0846a 0.518b 0.0853a 0.00989a SE 2.37b 0.1301b 0.469a 0.1304b 0.01986b Amyl Methyl Isobutanol Isoamyl acetate acetate hexanoate 1-Hexanol NC 0.247a 10.08ab 0.0234a 0.0219a 0.537a NW 0.248a 9.24a 0.0225a 0.0456c 0.486a SE 0.276b 12.06b 0.0280b 0.0335b 0.644b 3-Hexen-1-ol Methyl octanoate Acetic acid Octyl acetate 2-Hexen-1-ol NC 0.0241b 0.164a 0.253a 0.185b 0.00703a NW 0.0117a 0.301c 0.234a 0.150a 0.00682a SE 0.0286b 0.239b 0.314b 0.166ab 0.00930b

3.5 Phenolic Analysis of GV Wines by HPLC

3.5.1 Year 1

Relative abundance significantly differed (p < 0.05) for all compounds by wine (Figure

3.7A, B, and C). Relative abundance of epicatechin was also significantly different by region

(Figure 3.8B), however, catechin and gallic acid abundances did not differ between regions

(Figure 3.8A and C).

Regional differences were likely found due to epicatechin being undetectable in NC wines. NW, SE, and NE regions were not significantly different in epicatechin content, and NE and NC regions were not different even though epicatechin was not detected in the latter samples.

When examined with wine as the main factor, NC wine was significantly lower in epicatechin than all samples, while NW2 was significantly higher than all samples.

51 When examining catechin abundance by wine, SE1, NW2, and SE3 wines had significantly higher catechin content than other samples. Catechin was not detected in NC1 and

SE5 samples. NE wines were significantly lower in catechin than NW2 and SE samples in which catechin was detected.

Gallic acid was detected in all wine samples. SE3 was higher in gallic acid compared to

SE1, 2, and 5, as well as NW1. NW2 was also higher in gallic acid compared to NW1.

3.5.2 Year 2

In Year 2 samples, relative abundance significantly differed (p < 0.05) for all compounds by wine (Figure 3.7D, E, and F); however, no compounds were significantly different by region

(Figure 3.8D, E, and F). NW2 was significantly higher in both gallic acid and epicatechin than all other wines. NW1 and NW3 wines were lowest in all compounds (epicatechin and catechin were not detected in these samples), which may be due to limited cluster thinning at these sites.

Clusters per vine counts at these sites were high (54 clusters per vine in Year 2). Cluster thinning is used to improve grape quality and control crop load in over cropping vines, and phenolic levels have been found to increase in grapes when cluster thinning is employed (Guidoni et al., 2002;

Kliewer & Weaver, 1971; Rutan et al., 2018). It is likely that lack of adequate crop thinning at the

NW1 and NW3 sites altered phenolic composition and resulted in lower phenolic concentrations in the wines.

Although not significantly different by wine, the NW1 and NW3 samples still had the highest ratings of astringent mouthfeel in Year 2. Astringency can be caused by a number of compounds present in wines, including phenolics. However, high acidity can also lead to high astringency ratings in white wines regardless of total phenolic concentration (Gawel et al., 2014).

52 NW1 and NW3 wines had the lowest pH and highest TA values of the samples, suggesting that astringency ratings were due to acidity and not concentration of phenolics.

53

Figure 3.7 ANOVA by wine and Tukey’s post-hoc comparison results for relative abundance of catechin (A,D), epicatechin (B,E), and gallic acid (C,F) in GV wines. Results for Year 1 wines are shown in A, B, and C, while Year 2 wine results are shown in D, E, and F.

54

Figure 3.8 Results for ANOVA by region with Tukey’s post-hoc comparison for relative abundance of catechin (A,D), epicatechin (B,E), and gallic acid (C,F). Year 1 results are shown in A, B, and C, while Year 2 results are in D, E, and F.

55 3.6 Color Analysis of GV Wines

3.6.1 Year 1

Colorimetric coordinates give clarity, color, and lightness values to a sample. The L* coordinate defines lightness of a sample on a scale from 0-100, with 0 being black and 100 being white. The a* coordinate defines the red-green color of a sample, with positive values being more red and negative values more green. The b* coordinate measures the yellow-blue component, with a more yellow sample having a positive b* value and a more blue sample having a negative b* value (Wrolstad & Smilth, 2017).L*, a*, and b* coordinates for each sample were plotted to determine which samples differed the most in these aspects (Figure 3.9A, B, and C). In general, there was variation in L* values for samples regardless of region. All but four replicates of SE wines and all but one replicate of NW wines had L* values higher than 99.00. NE wines and one replicate of NC wines had L* values less than 99.00. a* values also varied within region for SE wines, ranging from -0.37 to -0.81. NW wines generally had lower a* values than wines from other regions, ranging from -0.78 to -0.88. A clearer regional trend was seen in b* values, with

SE and NC wines having b* values less than 3.00 and NW and NE wines having b* values greater than 3.00. These results are consistent with visual sensory evaluation, in which the trained panel found NW and NE wines to be significantly higher in yellow color than wines from the SE and NC regions.

Differences in L*, a*, and b* coordinates yielded some visual differences between samples. A DE ³ 2.0 indicates both experienced and unexperienced observers can notice a color difference. The first replicate of NW2 was visually different from both replicates of SE1 and NC1 wines, likely due to NW2 measuring highest in b* value. Samples with the highest and lowest L* had a DE of 0.83, and samples that had highest and lowest a* values had a DE of 1.5.

56 3.6.2 Year 2

L* values were generally similar across regions for Year 2, consisted with Year 1 results

(Figure 3.9D, E, and F). All but 3 of 15 fermentation replicates had L* values measuring between

98.00 and 99.00. There was variation in a* measurements across wines, ranging from -0.40 to

0.03 for SE wines and -0.36 to -0.16 for NW wines. NW wines had the highest b* values ranging from 4.36 to 5.07. These results are consistent with visual sensory evaluation, in which the trained panel rated NW wines significantly higher in yellow color than other regions. The trained panel rated NC wines higher in yellow color than SE wines. However, b* values for NC wines were lower than those of SE wines.

While DE values indicated a noticeable color difference for both experienced and unexperienced observers between some Year 1 samples, samples that had the greatest difference in b* value (replicates of NC1 and NW1) had a DE of 1.96, suggesting that only an experienced observer, such as a trained panelist, can detect a visual difference between samples. Similar results were found for samples that were most different in L* value (NW1 and NW3 replicates), which had a DE of 1.35. Samples that were most different in a* value had a DE of 0.77, indicating no visual difference.

57

Figure 3.9 Comparison of L*, a*, and b* color coordinates for Year 1 (A,B,C) and Year 2 (D,E,F) wines. Measurements shown for each fermentation replicate are the average of analytical replicate values.

58 3.7 Correlation of Sensory and Instrumental Data

Significant sensory and wine instrumental variables were correlated using Pearson’s correlation coefficient. Statistical correlations may indicate potential relationships between instrumental and sensory data. Examining these correlations does not prove causality, however, it can give insight into which relationships should be further studied and which ones may be causal.

Significant linear correlations between variables for both years are shown in Figure 3.11. Year 1 resulted in fewer significant linear correlations between sensory attributes and instrumental variables than Year 2 (n=122 vs. n = 244), possibly because there were fewer significant differences between sensory attributes and aroma compounds in Year 1 data analysis.

Some correlations were found to be consistent between Year 1 and Year 2, and these will be reviewed first (Fig. 3.11). In Figure 3.11, squares in red shades show positive correlations (p <

0.05), while those in blue show negative correlations at the same significance level. Yellow color was positively correlated with b* values (Year 1: r(16) = 0.88, p = 1.26e-06; Year 2: r(13) = 0.64, p = 1.378e-02) and negatively with a* values (Year 1: r(16) = -0.89, p = 7.23e-07; Year 2: r(13)

= -0.56, p = 0.04) across both years. Higher b* values indicate a sample is more yellow in color, so b* and yellow color correlation was expected. a* values were negative for the samples, which indicate a sample is greener and less red, and for white wines a negative a* value is not unexpected since there is limited skin contact to give a more orange or red color to wines. Thiol aroma and flavor correlation was consistent in both years as well (Year 1: r(16) = 0.75, p = 3.1e-

04); Year 2: r(13) = 0.89, p = 1.36e-05). Sour taste was positively correlated with TA values and negatively with pH values in both years. This was also expected as lower pH values indicate higher acidity, while higher TA values also indicate higher acidity.

There were a number of significant correlations that were only found in the Year 1 data

(Figure 3.11A). In Year 1, thiol aroma, i.e., the aroma associated with 4MMP (Table 2.1A), was

59 positively correlated with catechin (r(16) = 0.68, p = 0.002) and free SO2 (r(16) = 0.59, p = 0.01).

It is possible that sulfur acted as an antioxidant and prevented oxidation of thiols and subsequent loss of thiol aroma. Research has shown that additions of sulfur dioxide can prevent loss of thiols in the presence of catechin in model wine (Blanchard et al., 2004). Work by Saucier and

Waterhouse has also found catechin to have a synergistic effect in combination with SO2 in preventing oxidation in model wine (Saucier & Waterhouse, 1999). Sulfur aroma, i.e., the aroma associated with the smell of potassium metabisulfite (Table 2.1A), was also positively correlated with free SO2 (r(16) = 0.49, p = 0.037). This was expected, as aroma of SO2 is detected when a high proportion of SO2 is in the free form (Blesic et al., 2014).

There were several significant negative correlations between sensory and instrumental variables in Year 1. Sulfur aroma was negatively correlated with multiple esters, including ethyl hexanoate, hexyl acetate, 2-hexen-1-ol acetate, ethyl octanoate, and ethyl decanoate. Sulfur dioxide can bind with carbonyl compounds and reduce concentrations of aroma compounds such as b-damascenone or acetaldehyde (Daniel et al., 2004), however, in the case of volatile esters, additions of sulfur dioxide can retain ester concentrations after three months of aging (Garde-

Cerdán & Ancín-Azpilicueta, 2007), so it is likely this negative correlation is not causal.

Gallic acid was negatively correlated with sour taste (r(16) = -0.54, p = 0.021) and positively correlated with pH (r(16) – 0.62, p =0.006). Gallic acid can provide both bitter taste and astringent mouthfeel in wines (Robichaud & Noble, 1990), and it is possible higher concentrations of gallic acid in samples reduced perception of sour taste due to mixture suppression (Lawless & Heymann, 2010b). Phenylethyl alcohol was negatively correlated with octanoic acid (r(16) = -0.79, p = 1.0e-04) and phenethyl acetate (r(16) = -0.64, p = 0.004), and gallic acid was negatively correlated with methyl octanoate. Phenethyl acetate is produced enzymatically from phenylethyl alcohol, so the negative correlation between these compounds is expected (Waterhouse et al., 2016). Phenylethyl alcohol was identified in juice samples as well as

60 in wines in Year 1, while octanoic acid was only present in wine samples. The wine with the highest concentration of phenylethyl alcohol, NW2, was also lowest in octanoic acid, which may be driving the negative correlation. Higher alcohols such as phenylethyl alcohol are a result of amino acid metabolism, while octanoic acid is a result of fatty acid metabolism during the fermentation process, so it is unlikely that the correlation between these compounds is causal.

Correlations for Year 2 sensory and instrumental data is shown in Figure 3.11B. Thiol aroma was positively correlated with 1-propanol and 3-hexen-1-ol, which provide an alcohol and grassy aroma, respectively. It is unlikely that these compounds are contributing to the thiol aroma perceived in wines, as panelists were trained using a 4-MMP solution to define thiol aroma (Table

2.1A). Gallic acid and isoamyl alcohol, which provides a fruity aroma, were also positively correlated. Floral aroma was positively correlated with all three phenolic compounds, VA, and total SO2. Gallic acid, epicatechin, and catechin are not believed to impart aroma or flavor characteristics to wines, so the correlation with floral aroma is likely coincidental. As previously mentioned, additions of sulfur dioxide can decrease floral aromas by binding carbonyl compounds (Daniel et al., 2004), so it is unlikely that SO2 bound to floral compounds in the wine samples since these two variables were positively correlated. Sweet taste was positively correlated with RS, indicating panelist perceived wines with higher sugar content as sweeter than other samples. Green apple flavor was positively correlated with TA. Higher TA values indicate a sample is more acidic, and it is possible panelists associated more sour samples with green apple flavor. This flavor attribute was also correlated with methyl octanoate, which can give a waxy and green aroma.

Haziness was negatively correlated with a number of volatile compounds in Year 2, including ethanol, isobutanol, isoamyl alcohol, and 1-hexanol. Haziness was also negatively correlated with L*, indicating that the trained panel rated darker samples as hazier. Floral aroma

61 was negatively correlated with ethyl hexanoate, which can give a green apple aroma. Ethyl hexanoate was also negatively correlated with the three phenolic compounds quantified.

The most prominent difference in correlation results across years is that sulfur aroma was negatively correlated with many ethyl and acetate esters in Year 1 while esters were negatively correlated with haziness in Year 2. It is possible that higher SO2 levels in Year 1 wines provided a sulfur aroma that overpowered fruity notes provided by esters.

Figure 3.10 Significant (p < 0.05) correlation of sensory and chemistry parameters for Year 1 (A) and Year 2 (B) GV wines. Positive correlations are shown in red shades while negative correlations are shown in blue.

62 Chapter 4

Conclusions and Future Work

4.1 Overall Conclusions

This study examined regional differences in PA-grown GV using a controlled winemaking method, along with both sensory and instrumental analysis. The study was completed across two vintages in order to capture any year-to-year climate variation. This was the first study to examine regional differences in PA using a controlled winemaking method, and also the first time PA GV was examined sensorially.

Differences in aroma, taste, mouthfeel, and flavor were found between the wines when compared to each other; however, fewer differences were found when examining the wines based on region. In both years, wines from the NW region were highest in yellow color, and in Year 1 they were also the least hazy. These wines were also highest in citrus flavor in Year 2 and associated with earthy aroma in Year 1. In Year 2, the SE region wines were highest in both thiol aroma and flavor. NC wines were most sweet and least sour in Year 2, however, this was likely due to higher RS present at the end of fermentation for these samples since chaptalization and acidity were controlled in vinification. While NE wines were not examined in Year 2 due to severe winter damage, these wines were associated with sour taste in Year 1.

Regional differences were less clear when aspects were examined individually via analytical methods. Differences in aroma profiles of GV juice found variation in compounds that provide mainly grassy and green aromas. When wines were examined for volatile aroma composition, NC wine from Year 1 was highest in hexyl acetate, possibly due to presence of the precursors 2-hexenal and hexanal. In Year 2, differences were found by region, namely SE wines being highest in some acetate esters and NW wines highest in the esters methyl hexanoate and

63 methyl octanoate. When phenolics were examined, Year 1 saw regional differences in epicatechin, with NW and SE regions higher in this compound than NE and NC. No differences in phenolic compounds were found when examined by region in Year 2. Instrumental color results were similar to those found with the trained panels, in that NW wines were higher in b* and therefore more yellow in color than samples from other regions. In both years, there was variation in L* across all regions.

Overall, regional differences were found in both years of study, and in both instrumental and sensory analysis. Even though abnormally high amounts of rainfall were experienced in Year

1, some regional trends were seen across both years, such as the NW region consistently producing wines higher in yellow color. Since winemaking was controlled throughout the study, differences in aroma and flavor are likely due to the presence or absence of specific precursor compounds, and the accumulation of these compounds is dependent on environmental conditions in the vineyard.

4.2 Future Work

Weather data was collected across both growing seasons at each experimental site, and future work will include examining the weather data to further explain regional differences found in this study. Statistical analysis such as correlation or partial least squares regression (PLSR) will give insight into what weather parameters may be driving differences found in the wines.

Including a third year of data collection for the study would also be useful in confirming environmental characteristics for PA growing regions. The end of the Year 1 growing season was unusually wet with high amounts of rainfall before harvest, and so a third year of winemaking could further validate findings from both years of study.

64 Further examination of weather patterns across the experimental sites could also validate the use of current growing regions. The regions recognized by the Pennsylvania Winery

Association, which were used in this study, were created based on county jurisdictions. However, it is unlikely that county jurisdictions capture a winegrowing region accurately, and other parameters may be more useful in determining winegrowing regions in PA. Studying the weather patterns at each of the experimental sites may reveal that some sites may experience similar climates even though they are not in the same current growing region, and suggest that regional classifications should be made using other means such as environmental characteristics.

Regional differences were found in both sensory and instrumental analysis in this study, as well as differences between wines within the same growing region. Specifically, SE2 and SE3 showed differences in aroma perception, even though these sites are located at the same commercial vineyard. This suggests that mesoclimate variation will contribute to differences in final wine aroma. Further investigation into how these sites differ (i.e. differences in soil, elevation, microclimate, etc.) may provide explanation for differences in these wines.

Additionally, these sites produced wines that were high in thiol aroma and flavor. While expression of thiol aroma in wines is typical when using damaged or mechanically harvested fruit, winemaking was controlled in this study, and so expression of these compounds is likely due to variations in the vineyard. While volatile thiols have been studied extensively in other vinifera varieties such as Sauvignon blanc or Riesling, characterizing varietal thiols in GV has not been completed. With variation in thiol aroma in wines from this study, there is potential that volatile thiols may play a part in characteristic PA GV aroma and flavor.

65

References

Adams, D. O. (2006). Phenolics and ripening in grape berries. American Journal of Enology and Viticulture, 57(3), 249–256.

Allegro, G., Pastore, C., Valentini, G., & Filippetti, I. (2019). Effects of sunlight exposure on flavonol content and wine sensory of the white winegrape grechetto gentile. American Journal of Enology and Viticulture, 70(3), 277–285. https://doi.org/10.5344/ajev.2019.17108

Asproudi, A., Petrozziello, M., Cavalletto, S., & Guidoni, S. (2016). Grape aroma precursors in cv. Nebbiolo as affected by vine microclimate. Food Chemistry, 211, 947–956. https://doi.org/10.1016/j.foodchem.2016.05.070

Aubert, C., Baumann, S., & Arguel, H. (2005). Optimization of the analysis of flavor volatile compounds by liquid-liquid microextraction (LLME). Application to the aroma analysis of melons, peaches, grapes, strawberries, and tomatoes. Journal of Agricultural and Food Chemistry, 53(23), 8881–8895. https://doi.org/10.1021/jf0510541

Austrian Wine Marketing Board. (2018). Austrian Wine Statistics Report 2017. (September). Retrieved from http://www.ris.bka.gv.at/GeltendeFassung.wxe?Abfrage=Bundesnormen&Gesetzesnummer =20006757

Belancic, A., Agosin, E., Ibacache, A., Bordeu, E., Baumes, R., Razungles, A., & Bayonove, C. (1997). Influence of Sun Exposure on the Aromatic Composition of Chilean Muscat Grape Cultivars Moscatel de Alejandría and Moscatel rosada. 48(2), 181–186.

Benkwitz, F., Tominaga, T., Kilmartin, P. A., Lund, C., Wohlers, M., & Nicolau, L. (2012). Identifying the chemical composition related to the distinct aroma characteristics of New Zealand Sauvignon blanc wines. American Journal of Enology and Viticulture, 63(1), 62– 72. https://doi.org/10.5344/ajev.2011.10074

Bergqvist, J., Dokoozlian, N., & Ebisuda, N. (2001). Sunlight exposure and temperature effects on berry growth and composition of Cabernet Sauvignon and Grenache in the central San Joaquin Valley of California. American Journal of Enology and Viticulture, 52(1), 1–7.

Bernhard, R. A., Shibamoto, T., Yamaguchi, K., & White, E. (1983). The Volatile Constituents of Schinus molle L. Journal of Agricultural and Food Chemistry, 31(2), 463–466. https://doi.org/10.1021/jf00116a075

66 Betés-Saura, C., Andrés-Lacueva, C., & Lamuela-Raventós, R. M. (1996). Phenolics in White Free Run Juices and Wines from Penedès by High-Performance Liquid Chromatography: Changes during Vinification. Journal of Agricultural and Food Chemistry, 44(10), 3040– 3046. https://doi.org/10.1021/jf9601628

Black, C. A., Parker, M., Siebert, T. E., Capone, D. L., & Francis, I. L. (2015). Terpenoids and their role in wine flavour: Recent advances. Australian Journal of Grape and Wine Research, 21, 582–600. https://doi.org/10.1111/ajgw.12186 Blanchard, L., Darriet, P., & Dubourdieu, D. (2004). Reactivity of 3-mercaptohexanol in red wine: Impact of oxygen, phenolic fractions, and sulfur dioxide. American Journal of Enology and Viticulture, 55(2), 115–120.

Blesic, M., Spaho, N., & Murtic, M. S. (2014). Threshold of Sensory Perception of Sulphur Dioxide in Herzegovinian White Wines. (August 2015), 3–7.

Brook, S. (2019). The Wines of Austria (Second). Infinite Ideas. Buttery, R. G., Ling, L. C., & Kamm, J. A. (1982). Volatile Components of Alfalfa Flowers and Pods. Journal of Agricultural and Food Chemistry, 30(4), 739–742. https://doi.org/10.1021/jf00112a028 Cantergiani, E., Brevard, H., Krebs, Y., Feria-Morales, A., Amadò, R., & Yeretzian, C. (2001). Characterisation of the aroma of green Mexican coffee and identification of mouldy/earthy defect. European Food Research and Technology, 212(6), 648–657. https://doi.org/10.1007/s002170100305 Caparoso, R. (2016). Older Millennials are now America’s most important wine consumers. Retrieved from Lodi Winegrape Commission website: https://www.lodiwine.com/?method=blog.blogDrilldown&blogEntryID=1523D253-988C- 9421-9220-01FA7B5E59ED&originalMarketingURL=blog/Older-Millennials-are-now-our- most-important-wine-consumers Caputi, L., Carlin, S., Ghiglieno, I., Stefanini, M., Valenti, L., Vrhovsek, U., & Mattivi, F. (2011). Relationship of changes in rotundone content during grape ripening and winemaking to manipulation of the “peppery” character of wine. Journal of Agricultural and Food Chemistry, 59(10), 5565–5571. https://doi.org/10.1021/jf200786u

Casassa, L. F., Beaver, C. W., Mireles, M., Larsen, R. C., Hopfer, H., Heymann, H., & Harbertson, J. F. (2013). Influence of fruit maturity, maceration length, and ethanol amount on chemical and sensory properties of Merlot wines. American Journal of Enology and Viticulture, 64(4), 437–449. https://doi.org/10.5344/ajev.2013.13059

Castellarin, S. D., Matthews, M. A., Di Gaspero, G., & Gambetta, G. A. (2007a). Water deficits accelerate ripening and induce changes in gene expression regulating flavonoid biosynthesis in grape berries. Planta, 227(1), 101–112. https://doi.org/10.1007/s00425-007-0598-8

67 Castellarin, S. D., Pfeiffer, A., Sivilotti, P., Degan, M., Peterlunger, E., & Di Gaspero, G. (2007b). Transcriptional regulation of anthocyanin biosynthesis in ripening fruits of grapevine under seasonal water deficit. Plant, Cell and Environment, 30(11), 1381–1399. https://doi.org/10.1111/j.1365-3040.2007.01716.x

Castello, G., Timossi, A., & Gerbino, T. C. (1988). Gas chromatographic pounds on non-polar columns. 454, 129–143.

Chandravadana, M. V., Vekateshwarlu, G., Bujji Babu, C. S., Roy, T. K., Shivashankara, K. S., Pandey, M., … Selvaraj, Y. (2005). Volatile flavour components of dry milky mushrooms (Calocybe indica). Flavour and Fragrance Journal, 20(6), 715–717. https://doi.org/10.1002/ffj.1653

Chaves, M. M., Zarrouk, O., Francisco, R., Costa, J. M., Santos, T., Regalado, A. P., … Lopes, C. M. (2010). Grapevine under deficit irrigation: hints from physiological and molecular data. Annals of Botany, 105(5), 661–676. https://doi.org/10.1093/aob/mcq030

Chen, C., Kuo, M., Liu, S., & Wu, C. (1986). Volatile Components of Salted and Pickled Prunes ( Prunus mume et Zucc .). 08903, 140–144.

Coombe, B. G. (1989). The Grape Berry as a Sink. Acta Horticulturae, (239), 149–158. Crespo, J., Rigou, P., Romero, V., García, M., Arroyo, T., & Cabellos, J. M. (2018). Effect of seasonal climate fluctuations on the evolution of glycoconjugates during the ripening period of grapevine cv. Muscat à petits grains blancs berries. Journal of the Science of Food and Agriculture, 98(5), 1803–1812. https://doi.org/10.1002/jsfa.8656 Daniel, M. A., Elsey, G. M., Capone, D. L., Perkins, M. V., & Sefton, M. A. (2004). Fate of damascenone in wine: The role of SO2. Journal of Agricultural and Food Chemistry, 52(26), 8127–8131. https://doi.org/10.1021/jf048582h Darriet, P., Thibon, C., & Dubourdieu, D. (2012). Aroma and aroma precursors in grape berry. The Biochemistry of the Grape Berry, (September 2015), 111–136. https://doi.org/10.2174/978160805360511201010111 de Mendiburu, F. (2017). Agricolae: Statistical Procedures for Agricultural Research. Deluc, L. G., Quilici, D. R., Decendit, A., Grimplet, J., Wheatley, M. D., Schlauch, K. A., … Cramer, G. R. (2009). Water deficit alters differentially metabolic pathways affecting important flavor and quality traits in grape berries of Cabernet Sauvignon and Chardonnay. BMC Genomics, 10, 1–33. https://doi.org/10.1186/1471-2164-10-212 Donohue, M. (2020). Understanding Regional Sensory Profiles of White Wines in Pennsylvania. The Pennsylvania State University. Douglas, D., Cliff, M. A., & Reynolds, A. G. (2001). Canadian terroir: Characterization of Riesling wines from the Niagara Peninsula. Food Research International, 34(7), 559–563.

68 https://doi.org/10.1016/S0963-9969(01)00071-0

Downey, M. O., Dokoozlian, N. K., & Krstic, M. P. (2006). Cultural practice and environmental impacts on the flavonoid composition of grapes and wine: A review of recent research. American Journal of Enology and Viticulture, 57(3), 257–268. Downey, M. O., Harvey, J. S., & Robinson, S. P. (2008). The effect of bunch shading on berry development and flavonoid accumulation in Shiraz grapes. Australian Journal of Grape and Wine Research, 10(1), 55–73. https://doi.org/10.1111/j.1755-0238.2004.tb00008.x

Duiker, S. W. (2003). The Agronomy Guide 2003.

Duiker, S. W. (2020). The Agronomy Guide 2019–2020.

Duke, A. (2017). Penn State Grape and Wine Team Boosts Industry Through Research, Education.

Easingwood, C., Lockshin, L., & Spawton, A. (2011). The drivers of wine regionality. Journal of Wine Research, 22(1), 19–33. https://doi.org/10.1080/09571264.2011.550759 Economic impact of the Pennsylvania Wine and Grape Industries. (2018).

Fischer, U., Roth, D., & Christmann, M. (1999). The impact of geographic origin, vintage and wine estate on sensory properties of Vitis vinifera cv. Riesling wines. Food Quality and Preference, 10(4–5), 281–288. https://doi.org/10.1016/s0950-3293(99)00008-7

Flak, W., Krizan, R., Kutscher, W., Tscheik, G., & Wallner, E. (2007). Charakterisierung von Weinen der Sorte Grüner Veltliner aus verschiedenen Herkünften im Weinbaugebiet Weinviertel. Mitteilungen Klosterneuburg, 57, 131–139.

Flak, W., Krizan, R., Passmann, G., Tscheik, G., & Wallner, E. (2009). Charakterisierung von Leitweinen des Weinbau- gebietes Wagram . Teil 1 : Grüner Veltliner. Mitteilungen Klosterneuburg, 59, 227–236. Garde-Cerdán, T., & Ancín-Azpilicueta, C. (2007). Effect of SO2 on the formation and evolution of volatile compounds in wines. Food Control, 18(12), 1501–1506. https://doi.org/10.1016/j.foodcont.2006.11.001 Gas chromatography analysis of organic solvents using capillary columns (No. 2). (2003). Retrieved from Shimadzu website: http://www.shimadzu.com/apps/form.cfm Gawel, R., Day, M., Van Sluyter, S. C., Holt, H., Waters, E. J., & Smith, P. A. (2014). White wine taste and mouthfeel as affected by juice extraction and processing. Journal of Agricultural and Food Chemistry, 62(41), 10008–10014. https://doi.org/10.1021/jf503082v Gay, C. (1998). Invitation to comment on the “Medical Record Linkage Study”. Food Quality and Preference, 9(3), 166. https://doi.org/10.5694/j.1326-5377.1995.tb140033.x

Genovese, A., Gambuti, A., Piombino, P., & Moio, L. (2007). Sensory properties and aroma

69 compounds of sweet Fiano wine. Food Chemistry, 103(4), 1228–1236. https://doi.org/10.1016/j.foodchem.2006.10.027

Girbau, T., Stummer, B. E., Pocock, K. F., Baldock, G. A., Scott, E. S., & Waters, E. J. (2004). The effect of Uncinula necator (powdery mildew) and Botrytis cinerea infection of grapes on the levels of haze-forming pathogenesis-related proteins in grape juice and wine. Australian Journal of Grape and Wine Research, 10(2), 125–133. https://doi.org/10.1111/j.1755-0238.2004.tb00015.x Gómez-Míguez, M. J., Gómez-Míguez, M., Vicario, I. M., & Heredia, F. J. (2007). Assessment of colour and aroma in white wines vinifications: Effects of grape maturity and soil type. Journal of Food Engineering, 79(3), 758–764. https://doi.org/10.1016/j.jfoodeng.2006.02.038 Gouot, J. C., Smith, J. P., Holzapfel, B. P., & Barril, C. (2019). Grape berry flavonoid responses to high bunch temperatures post véraison: Effect of intensity and duration of exposure. Molecules, 24(23). https://doi.org/10.3390/molecules24234341 Green, J. A., Parr, W. V., Breitmeyer, J., Valentin, D., & Sherlock, R. (2011). Sensory and chemical characterisation of Sauvignon blanc wine: Influence of source of origin. Food Research International, 44(9), 2788–2797. https://doi.org/10.1016/j.foodres.2011.06.005 Grifoni, D., Carreras, G., Zipoli, G., Sabatini, F., Dalla Marta, A., & Orlandini, S. (2008). Row orientation effect on UV-B, UV-A and PAR solar irradiation components in vineyards at Tuscany, Italy. International Journal of Biometeorology, 52(8), 755–763. https://doi.org/10.1007/s00484-008-0168-1 Grifoni, D., Mancini, M., Maracchi, G., Orlandini, S., & Zipoli, G. (2006). Analysis of Italian wine quality using freely available meteorological information. American Journal of Enology and Viticulture, 57(3), 339–346. Guidoni, S., Allara, P., & Schubert, A. (2002). Effect of cluster thinning on berry skin anthocyanin composition of Vitis vinifera cv. Nebbiolo. American Journal of Enology and Viticulture, 53(3), 224–226.

Hed, B., & Centinari, M. (2018). 2018 Growing Season Recap. Heymann, H., Robinson, A. L., Buscema, F., Stoumen, M. E., King, E. S., Hopfer, H., … Ebeler, S. E. (2015). Effect of region on the volatile composition and sensory profiles of malbec and cabernet sauvignon wines. ACS Symposium Series, 1203, 109–122. https://doi.org/10.1021/bk-2015-1203.ch007 Hickey, C. C., Kwasniewski, M. T., & Wolf, T. K. (2018). Leaf removal effects on cabernet franc and petit verdot: Ii. grape carotenoids, phenolics, and wine sensory analysis. American Journal of Enology and Viticulture, 69(3), 231–246. https://doi.org/10.5344/ajev.2018.17107

70 Humpf, H., & Schreier, P. (1991). Bound Aroma Compounds from the Fruit and the Leaves of Blackberry ( Rubus laciniata L .). 1830–1832.

Jensen, K., Christensen, L. P., Hansen, M., Jrgensen, U., & Kaack, K. (2001). Olfactory and quantitative analysis of volatiles in elderberry (Sambucus nigra L) juice processed from seven cultivars. Journal of the Science of Food and Agriculture, 81(2), 237–244. https://doi.org/10.1002/1097-0010(20010115)81:2<237::AID-JSFA809>3.0.CO;2-H

Johnsen, L. G., Skou, P. B., Khakimov, B., & Bro, R. (2017). Gas chromatography – mass spectrometry data processing made easy. Journal of Chromatography A, 1503, 57–64. https://doi.org/10.1016/j.chroma.2017.04.052

Kabinett, L. Q. (1993). Environmental and Management Practices Affecting Grape Composition and Wine Quality - A Review. 44(4), 409–430. Kassambara, A. (2019). Package ggcorrplot: Visualization of a Correlation Matrix using “ggplot2.” Keller, M. (2010). Environmental Constraints and Stress Physiology. The Science of Grapevines: Anatomy and Physiology, 227–310. https://doi.org/10.1016/b978-0-12-374881-2.00007-6

Khan, M., Verma, S. C., Srivastava, S. K., Shawl, A. S., Syamsundar, K. V., Khanuja, S. P. S., & Kumar, T. (2006). Essential oil composition of Taxus wallichiana Zucc. from the Northern Himalayan region of India. Flavour and Fragrance Journal, 21(5), 772–775. https://doi.org/10.1002/ffj.1682

King, E. S., Stoumen, M., Buscema, F., Hjelmeland, A. K., Ebeler, S. E., Heymann, H., & Boulton, R. B. (2014). Regional sensory and chemical characteristics of malbec wines from mendoza and california. Food Chemistry, 143, 256–267. https://doi.org/10.1016/j.foodchem.2013.07.085 Kliewer, W. M., & Torres, R. E. (1972). Effect of Controlled Day and Night Temperatures on Grape Coloration. 23(2), 71–77.

Kliewer, W. M., & Weaver, R. J. (1971). Effect of Crop Level and Leaf Area on Growth, Composition, and Coloration of Tokay Grapes. König, H., Unden, G., & Fröhlich, J. (2017). Biology of microorganisms on grapes, in must and in wine. In Biology of Microorganisms on Grapes, in Must and in Wine. https://doi.org/10.1007/978-3-319-60021-5 Kotsaki, E., Reynolds, A. G., Brown, R., Jollineau, M., Lee, H. S., & Aubie, E. (2020). Proximal sensing and relationships to soil and vine water status, yield, and berry composition in ontario vineyards. American Journal of Enology and Viticulture, 71(2), 114–131. https://doi.org/10.5344/ajev.2019.19018 Kuhn, N., Guan, L., Dai, Z. W., Wu, B. H., Lauvergeat, V., Gomès, E., … Delrot, S. (2014).

71 Berry ripening: Recently heard through the grapevine. Journal of Experimental Botany, 65(16), 4543–4559. https://doi.org/10.1093/jxb/ert395

Kurtural, S. K. (2003). Vineyard Site Selection. In University of Kentucky Cooperative Extension Service. Retrieved from http://www.ext.vt.edu/pubs/viticulture/463-020/463-020.html Kwasniewski, M. T., Vanden Heuvel, J. E., Pan, B. S., & Sacks, G. L. (2010). Timing of cluster light environment manipulation during grape development affects C13 norisoprenoid and carotenoid concentrations in riesling. Journal of Agricultural and Food Chemistry, 58(11), 6841–6849. https://doi.org/10.1021/jf904555p

Lawless, H. T., & Heymann, H. (2010a). Descriptive Analysis. In Sensory Evaluation of Food (Second, pp. 227–257). https://doi.org/10.1007/978-1-4419-6488-5 Lawless, H. T., & Heymann, H. (2010b). Physiological and Psychological Foundations of Sensory Function. In Sensory Evaluation of Food (Second, pp. 19–56). https://doi.org/10.1007/978-1-4419-6488-5_2 Le, S., & Husson, F. (2008). Sensominer: A package for sensory data analysis. Journal of Sensory Studies, 23(1), 14–25. https://doi.org/10.1111/j.1745-459X.2007.00137.x

Leeuwen, C. Van, Tregoat, O., Choné, X., Bois, B., Pernet, D., & Gaudillére, J. P. (2009). Vine water status is a key factor in grape ripening and vintage quality for red bordeaux wine. How can it be assessed for vineyard management purposes? Journal International Des Sciences de La Vigne et Du Vin, 43(3), 121–134. https://doi.org/10.20870/oeno- one.2009.43.3.798 Lenth, R., Singmann, H., Love, J., Buerkner, P., & Herve, M. (2020). Package ‘emmeans’: Estimated Marginal Means, aka Least-Squares Means. In R package version 1.4.6. https://doi.org/10.1080/00031305.1980.10483031>.License Lenton, G. (2017, May 31). Could popular grape from Austria become one of Pennsylvania’s signature wines? Reading Eagle. Retrieved from https://www.readingeagle.com/berks- country/could-popular-grape-from-austria-become-one-of-pennsylvanias-signature- wines/article_7237bdf8-f58b-539d-89ea-d5ba0195ee95.html Li, J. H., Guan, L., Fan, P. G., Li, S. H., & Wu, B. H. (2013). Effect of sunlight exclusion at different phenological stages on anthocyanin accumulation in red grape clusters. American Journal of Enology and Viticulture, 64(3), 349–356. https://doi.org/10.5344/ajev.2013.12130 Lindbloom, B. (2011). Spectral Calculator Spreadsheets. Retrieved from http://www.brucelindbloom.com/index.html?SpectCalcSpreadsheets.html

Lockshin, L., & Corsi, A. M. (2012). Consumer behaviour for wine 2.0: A review since 2003 and future directions. Wine Economics and Policy, 1(1), 2–23. https://doi.org/10.1016/j.wep.2012.11.003

72 MacLeod, A. J., & Pieris, N. M. (1983). Volatile Components of Papaya (Carica papaya L.) with Particular Reference to Glucosinolate Products. Journal of Agricultural and Food Chemistry, 31(5), 1005–1008. https://doi.org/10.1021/jf00119a021

Marais, J., Hunter, J. J., & Haasbroek, P. D. (1999). Effect of Canopy Microclimate, Season and Region on Sauvignon Blanc Grape Composition and Wine Quality. South African Journal of Enology & Viticulture, 20(1), 19–30.

Marais, J., Versini, G., van Wyk, C. J., & Rapp, A. (1992). Effect of Region on Free and Bound Monoterpene and C13-N orisoprenoid Concentrations in Weisser Riesling Wines. South African Journal of Enology & Viticulture, 13(2), 71–77. https://doi.org/10.21548/13-2-2177

Marciniak, M., Reynolds, A. G., & Brown, R. (2013). Influence of water status on sensory profiles of Ontario Riesling wines. Food Research International, 54(1), 881–891. https://doi.org/10.1016/j.foodres.2013.08.030

Martin, D., Grose, C., Fedrizzi, B., Stuart, L., Albright, A., & McLachlan, A. (2016). Grape cluster microclimate influences the aroma composition of Sauvignon blanc wine. Food Chemistry, 210, 640–647. https://doi.org/10.1016/j.foodchem.2016.05.010

Mendez-Costabel, M. P., Wilkinson, K. L., Bastian, S. E. P., Jordans, C., Mccarthy, M., Ford, C. M., & Dokoozlian, N. (2014). Effect of winter rainfall on yield components and fruit green aromas of Vitis viniferaL. cv. Merlot in California. Australian Journal of Grape and Wine Research, 20(1), 100–110. https://doi.org/10.1111/ajgw.12060

Mendez-Costabel, Martin P., Wilkinson, K. L., Bastian, S. E. P., McCarthy, M., Ford, C. M., & Dokoozlian, N. (2013). Seasonal and regional variation of green aroma compounds in commercial vineyards of vitis vinifera L. Merlot in California. American Journal of Enology and Viticulture, 64(4), 430–436. https://doi.org/10.5344/ajev.2013.12109

Mokrzycki, W., & Tatol, M. (2011). Color difference Delta E - A survey Colour difference ∆ E - A survey Faculty of Mathematics and Informatics. Machine Graphics and Vision, 20(4), 383–411. https://doi.org/10.1007/s10817-009-9143-8

Monteiro, S., Piçarra-Pereira, M. A., Teixeira, A. R., Loureiro, V. B., & Ferreira, R. B. (2003). Environmental conditions during vegetative growth determine the major proteins that accumulate in mature grapes. Journal of Agricultural and Food Chemistry, 51(14), 4046– 4053. https://doi.org/10.1021/jf020456v

Morales, A. L., Albarracín, D., Rodríguez, J., Duque, C., Riaño, L. E., & Espitia, J. (1996). Volatile Constituents from Andes Berry (Rubus glaucus Benth). HRC Journal of High Resolution Chromatography, 19(10), 585–587. https://doi.org/10.1002/jhrc.1240191011

Mori, K., Goto-Yamamoto, N., Kitayama, M., & Hashizume, K. (2007). Loss of anthocyanins in red-wine grape under high temperature. Journal of Experimental Botany, 58(8), 1935–1945. https://doi.org/10.1093/jxb/erm055

73 Nauer, S., Brandes, W., Patzl-Fischerleitner, E., Hann, S., & Eder, R. (2018). Analyse von (- )rotundon mittels SPE-SPME-GC-MS in Österreichischen qualitätswinen der rebsorte “Grüner Veltliner.” Mitteilungen Klosterneuburg, 68, 107–119.

Nishimura, O., Yamaguchi, K., Mihara, S., & Shibamoto, T. (1989). Volatile constituents of Guava Fruits (Psidium guajava L.) and Canned Puree. Flavour and Fragrance Journal, 37(1), 139–142. https://doi.org/10.1002/ffj.2730060314

OIV. (2006). COMPENDIUM OF INTERNATIONAL ANALYSIS OF METHODS -OIV Chromatic Characteristics Method OIV-MA-AS2-11 Determination of chromatic characteristics according to CIELab. Retrieved from http://www.oiv.int/oiv/files/6 - Domaines scientifiques/6 - 4 Methodes d analyses/6-4-1/EN/OIV-MA-AS2-11.pdf

Ojeda, H., Andary, C., Kraeva, E., Carbonneau, A., & Deloire, A. (2002). Influence of pre- and postveraison water deficit on synthesis and concentration of skin phenolic compounds during berry growth of Vitis vinifera cv. Shiraz. American Journal of Enology and Viticulture, 53(4), 261–267. Osorio, C., Alarcon, M., Moreno, C., Bonilla, A., Barrios, J., Garzon, C., & Duque, C. (2006). Characterization of odor-active volatiles in champa (Campomanesia lineatifolia R. & P.). Journal of Agricultural and Food Chemistry, 54(2), 509–516. https://doi.org/10.1021/jf052098c Pavelescu, D., Fiechter, G., Steidl, R., Eder, R., & Kneifel, W. (2012). Sensory and chemical characteristics of wines from the variety “Grüner Veltliner” in connection with the sur lie method. Mitteilungen Klosterneuburg, 62, 21–30. Pedneault, K., Dorais, M., & Angers, P. (2013). Flavor of Cold-Hardy Grapes: Impact of Berry Maturity and Environmental Conditions. Journal of Agricultural and Food Chemistry, 61, 10418–10438. Peyrot des Gachons, C., Van Leeuwen, C., Tominaga, T., Soyer, J. P., Gaudillère, J. P., & Dubourdieu, D. (2005). Influence of water and nitrogen deficit on fruit ripening and aroma potential of Vitis vinifera L cv Sauvignon blanc in field conditions. Journal of the Science of Food and Agriculture, 85(1), 73–85. https://doi.org/10.1002/jsfa.1919

Politeo, O., Jukic, M., & Milos, M. (2007). Chemical composition and antioxidant capacity of free volatile aglycones from basil (Ocimum basilicum L.) compared with its essential oil. Food Chemistry, 101(1), 379–385. https://doi.org/10.1016/j.foodchem.2006.01.045

Pompilio, R. (2015, September). Grüner Veltliner Finds a Home. Wines & Vines. Retrieved from https://winesvinesanalytics.com/features/article/156978/Gruner-Veltliner-Finds-a-Home

Regner, F., Hack, R., Hanak, K., & Santiago, J. L. (2009). Variability within the cultivar “gruener veltliner.” Acta Horticulturae, 827, 245–252. https://doi.org/10.17660/actahortic.2009.827.42

74 Reynolds, A. G., & Wardle, D. (1989). Influence of fruit microclimate on monoterpene levels of Gewurztraminer. Am. J. Enol. Vitic., 40(3), 149–154. https://doi.org/10.1017/CBO9781107415324.004

Rezende, C. M., & Fraga, S. R. G. (2003). Chemical and aroma determination of the pulp and seeds of murici (Byrsonima crassifolia L.). Journal of the Brazilian Chemical Society, 14(3), 425–428. https://doi.org/10.1590/S0103-50532003000300014

Ristic, R., Johnson, T. E., Meiselman, H. L., Hoek, A. C., & Bastian, S. E. P. (2016). Towards development of a Wine Neophobia Scale (WNS): Measuring consumer wine neophobia using an adaptation of The Food Neophobia Scale (FNS). Food Quality and Preference, 49, 161–167. https://doi.org/10.1016/j.foodqual.2015.12.005

Riu-Aumatell, M., Castellari, M., López-Tamames, E., Galassi, S., & Buxaderas, S. (2004). Characterisation of volatile compounds of fruit juices and nectars by HS/SPME and GC/MS. Food Chemistry, 87(4), 627–637. https://doi.org/10.1016/j.foodchem.2003.12.033

Robichaud, J. L., & Noble, A. C. (1990). Astringency and Bitterness of Selected Phenolics in Wine. Journal of the Science of Food and Agriculture, 53(3), 343–353.

Robinson, A. L., Adams, D. O., Boss, P. K., Heymann, H., Solomon, P. S., & Trengove, R. D. (2012). Influence of geographic origin on the sensory characteristics and wine composition of Vitis vinifera cv. Cabernet Sauvignon wines from Australia. American Journal of Enology and Viticulture, 63(4), 467–476. https://doi.org/10.5344/ajev.2012.12023

Robinson, A. L., Boss, P. K., Solomon, P. S., Trengove, R. D., Heymann, H., & Ebeler, S. E. (2014). Origins of grape and wine aroma. Part 1. Chemical components and viticultural impacts. American Journal of Enology and Viticulture, 65(1), 1–24. https://doi.org/10.5344/ajev.2013.12070

Robinson, J., Harding, J., & Vouillamoz, J. (2012). Wine Grapes. Allen Lane. Roubelakis-Angelakis, K. A., & Kliewer, W. M. (1986). Effects of exogenous factors on phenylalanine ammonia-lyase activity and accumulation of anthocyanins and total phenolics in grape berries. Am. J. Enol. Vitic., 37(4), 275–280. Rowan, D. D., Lane, H. P., Allen, J. M., Fielder, S., & Hunt, M. B. (1996). Biosynthesis of 2- Methylbutyl, 2-Methyl-2-butenyl, and 2-Methylbutanoate Esters in Red Delicious and Granny Smith Apples Using Deuterium-Labeled Substrates. Journal of Agricultural and Food Chemistry, 44(10), 3276–3285. https://doi.org/10.1021/jf9508209 RStudio Team. (2015). RStudio: Integrated Development for R. Retrieved from http://www.rstudio.com/

Rutan, T. E., Herbst-Johnstone, M., & Kilmartin, P. A. (2018). Effect of Cluster Thinning Vitis vinifera cv. Pinot Noir on Wine Volatile and Phenolic Composition. Journal of Agricultural and Food Chemistry, 66(38), 10053–10066. https://doi.org/10.1021/acs.jafc.8b04062

75 Sadras, V. O., Petrie, P. R., & Moran, M. A. (2013a). Effects of elevated temperature in grapevine. II juice pH, titratable acidity and wine sensory attributes. Australian Journal of Grape and Wine Research, 19(1), 107–115. https://doi.org/10.1111/ajgw.12001

Sadras, V. O., Moran, M. A., & Bonada, M. (2013b). Effects of elevated temperature in grapevine. I Berry sensory traits. Australian Journal of Grape and Wine Research, 19(1), 95–106. https://doi.org/10.1111/ajgw.12007

Saucier, C. T., & Waterhouse, A. L. (1999). Synergetic activity of catechin and other antioxidants. Journal of Agricultural and Food Chemistry, 47(11), 4491–4494. https://doi.org/10.1021/jf990352t

Schüttler, A., Friedel, M., Jung, R., Rauhut, D., & Darriet, P. (2015). Characterizing aromatic typicality of riesling wines: Merging volatile compositional and sensory aspects. Food Research International, 69, 26–37. https://doi.org/10.1016/j.foodres.2014.12.010

Selli, S., Rannou, C., Prost, C., Robin, J., & Serot, T. (2006). Characterization of aroma-active compounds in rainbow trout (Oncorhynchus mykiss) eliciting an off-odor. Journal of Agricultural and Food Chemistry, 54(25), 9496–9502. https://doi.org/10.1021/jf0619582

Sheridan, M. K., & Elias, R. J. (2016). Reaction of Acetaldehyde with Wine Flavonoids in the Presence of Sulfur Dioxide. Journal of Agricultural and Food Chemistry, 64(45), 8615– 8624. https://doi.org/10.1021/acs.jafc.6b03565 Sherman, E., Greenwood, D. R., Villas-Boâs, S. G., Heymann, H., & Harbertson, J. F. (2017). Impact of grape maturity and ethanol concentration on sensory properties of Washington State merlot wines. American Journal of Enology and Viticulture, 68(3), 344–356. https://doi.org/10.5344/ajev.2017.16076

Shibamoto, T., Kamiya, Y., & Mihara, S. (1981). Isolation and Identification of Volatile Compounds in Cooked Meat: Sukiyaki. Journal of Agricultural and Food Chemistry, 29(1), 57–63. https://doi.org/10.1021/jf00103a015 Siebert, T. E., Barker, A., Pearson, W., Barter, S. R., De Barros Lopes, M. A., Darriet, P., … Francis, I. L. (2018). Volatile Compounds Related to “Stone Fruit” Aroma Attributes in Viognier and Chardonnay Wines. Journal of Agricultural and Food Chemistry, 66(11), 2838–2850. https://doi.org/10.1021/acs.jafc.7b05343

Slizhov, Y., & Gavrilenko, M. (2001). Effect of thermal treatment of poly(ethylene glycol) modified with europium acetylacetonate on its chromatographic properties. Russian Journal of Physical Chemistry, 75(6), 1012–1013. Spayd, S. E., Tarara, J. M., Mee, D. L., & Ferguson, J. C. (2002). Separation of sunlight and temperature effects on the composition of Vitis vinifera cv. Merlot berries. American Journal of Enology and Viticulture, 53(3), 171–182. Spencer, M. D., Pangborn, R. M., & Jennings, W. G. (1978). Gas Chromatographic and Sensory

76 Analysis of Volatiles from Cling Peaches. Journal of Agricultural and Food Chemistry, 26(3), 725–732. https://doi.org/10.1021/jf60217a052

Sweetman, C., Sadras, V. O., Hancock, R. D., Soole, K. L., & Ford, C. M. (2014). Metabolic effects of elevated temperature on organic acid degradation in ripening Vitis vinifera fruit. Journal of Experimental Botany, 65(20), 5975–5988. https://doi.org/10.1093/jxb/eru343

Sweetman, Crystal, Deluc, L. G., Cramer, G. R., Ford, C. M., & Soole, K. L. (2009). Regulation of malate metabolism in grape berry and other developing fruits. Phytochemistry, 70(11– 12), 1329–1344. https://doi.org/10.1016/j.phytochem.2009.08.006

Tarara, J. M., Lee, J., Spayd, S. E., & Scagel, C. F. (2008). Berry temperature and solar radiation alter acylation, proportion, and concentration of anthocyanin in Merlot grapes. American Journal of Enology and Viticulture, 59(3), 235–247. Tatsuka, K., Suekane, S., Sakai, Y., & Sumitani, H. (1990). Volatile Constituents of Kiwi Fruit Flowers: Simultaneous Distillation and Extraction versus Headspace Sampling. Journal of Agricultural and Food Chemistry, 38(12), 2176–2180. https://doi.org/10.1021/jf00102a015 Teixeira, A., Eiras-Dias, J., Castellarin, S. D., & Gerós, H. (2013). Berry phenolics of grapevine under challenging environments. International Journal of Molecular Sciences, 14(9), 18711–18739. https://doi.org/10.3390/ijms140918711 The Geography of Wine. (2012). In P. H. Dougherty (Ed.), Journal of Chemical Information and Modeling (Vol. 53). https://doi.org/10.1007/978-94-007-0464-0

Toda, H., Yamaguchi, K., & Shibamoto, T. (1982). Isolation and Identification of Banana-like Aroma from Banana Shrub (Michellia figo Spreng). Journal of Agricultural and Food Chemistry, 30(1), 81–84. https://doi.org/10.1021/jf00109a017

Tominaga, T., Balktenweck-Guyot, R., Peyrot des Gachons, C., & Dubourdieu, D. (2000). Contribution of Volatile Thiols to the Aromas of White Wines Made From Several Vitis viniferia Grape Varieties. 51(2), 178–181.

Tressl, R., Friese, L., Fendesack, F., & Koppler, H. (1978). Studies of the Volatile Composition of Hops During Storage. Journal of Agricultural and Food Chemistry, 26(6), 1426–1430. https://doi.org/10.1021/jf60220a036

Úbeda, C., del Barrio-Galán, R., Peña-Neira, Á., Medel-Marabolí, M., & Durán-Guerrero, E. (2017). Location effects on the aromatic composition of monovarietal cv. Carignan wines. American Journal of Enology and Viticulture, 68(3), 390–399. https://doi.org/10.5344/ajev.2017.16086

Umano, K., Shibamoto, T., Shoji, A., & Hagi, Y. (1986). Volatile Constituents of Peel of Quince Fruit, Cydonia oblonga Miller. Journal of Agricultural and Food Chemistry, 34(4), 593– 596. https://doi.org/10.1021/jf00070a003

77 Van Leeuwen, Cornelis, Friant, P., Choné, X., Tregoat, O., Koundouras, S., & Dubourdieu, D. (2004). Influence of climate, soil, and cultivar on terroir. American Journal of Enology and Viticulture, 55(3), 207–217.

Van Sluyter, S. C., McRae, J. M., Falconer, R. J., Smith, P. A., Bacic, A., Waters, E. J., & Marangon, M. (2015). Wine protein haze: Mechanisms of formation and advances in prevention. Journal of Agricultural and Food Chemistry, 63(16), 4020–4030. https://doi.org/10.1021/acs.jafc.5b00047 Vilanova, M., & Vilariño, F. (2006). Influence of geographic origin on aromatic descriptors of Spanish Albariño wine. Flavour and Fragrance Journal, 21(2), 373–378. https://doi.org/10.1002/ffj.1677

Viña, A., & Murillo, E. (2003). Essential Oil Composition from Twelve Varieties of Basil (Ocimum spp) Grown in Colombia. Journal of the Brazilian Chemical Society, 14(5), 744– 749. https://doi.org/10.1590/S0103-50532003000500008

Vinogradov, B. A. (2004). Production, composition, properties and application of essential oils. Waterhouse, A. L., Sacks, G. L., & Jeffery, D. W. (2016). Understanding Wine Chemistry. In Understanding Wine Chemistry. https://doi.org/10.1002/9781118730720 Wickham, H., Chang, W., Henry, L., Pedersen, T. L., Takahashi, K., Wilke, C., … Dunnington, D. (2020). Package ‘ggplot2’: Create Elegant Data Visualisations Using the Grammar of Graphics.

Willwerth, J. J., Reynolds, A. G., & Lesschaeve, I. (2015). Sensory analysis of riesling wines from different sub-appellations in the Niagara Peninsula in Ontario. American Journal of Enology and Viticulture, 66(3), 279–293. https://doi.org/10.5344/ajev.2015.14081

Wrolstad, R. E., & Smilth, D. E. (2017). Color Analysis. In S. Nielsen (Ed.), Food Analysis (pp. 545–555). https://doi.org/10.1007/978-3-319-45776-5 Yamaguchi, K., & Shibamoto, T. (1981). Volatile Constituents of Green Tea, Gyokuro (Camellia sinensis L. var Yabukita). Journal of Agricultural and Food Chemistry, 29(2), 366–370. https://doi.org/10.1021/jf00104a035 Zsófi, Z. S., Tóth, E., Rusjan, D., & Bálo, B. (2011). Terroir aspects of grape quality in a cool climate wine region: Relationship between water deficit, vegetative growth and berry sugar concentration. Scientia Horticulturae, 127(4), 494–499. https://doi.org/10.1016/j.scienta.2010.11.014

78

Appendix

Chapter 2 Supplementary Data

A1 – Descriptive Analysis Evaluation Ballot

(continued on page 78)

79

80

81

82

Chapter 3 Supplementary Data

Appendix Table 1. ANOVA for Year 1 sensory attributes showing the F-values for the factors Wine (W), Fermentation Replicate (FR), Judge (J), and their interactions. Bolded F-values are significant (p < 0.05). Factor W FR J W:J W:FR J:FR DF 8 1 7 56 8 7 Yellow Color 76.333 1.828 28.1084 2.4537 6.3427 0.9678 Haziness 76.9784 89.6667 9.2575 1.2694 21.8328 1.4682 Green Apple A 1.2619 1.3766 5.7968 1.443 2.1405 0.9372 Pear A 0.4796 0.0208 13.0532 1.3514 1.2616 0.6488 Other Fruit A 1.8178 0.2147 30.0189 0.9899 0.8332 0.1366 Grape A 1.1381 1.4633 18.2456 0.8199 1.4307 0.4723 Citrus A 5.813 0.0203 52.4984 3.1789 1.124 0.6919 Floral A 0.8409 1.593 13.9592 1.2506 1.1039 0.6875 Earthy A 4.4809 1.006 37.6537 2.0365 1.3713 0.8386 Thiol A 2.9519 0.0286 16.9742 1.0681 0.4442 0.4573 CannedVeg A 3.1145 0.3936 30.6272 1.6694 1.6952 1.3842 RottenEgg A 3.9512 0.0611 31.551 2.5372 1.3908 1.8079 Sulfur A 2.1203 5.1131 38.9351 0.8727 0.8806 1.7315 Yeasty A 1.5529 0.0133 54.8933 1.0731 0.5049 0.3407 Oxidized A 0.8086 0.0206 19.291 1.2466 1.5322 0.8333 ChemicalSolvent A 0.7631 0.0356 43.4047 0.9106 1.0687 0.4222 Ethanol A 1.0804 2.8731 32.4814 1.2261 0.8929 2.0856 Sour T 2.3825 0.0018 21.2335 0.6709 0.6826 0.1717 Sweet T 0.7055 0.0014 25.199 0.54 1.2749 0.3599 Salty T 1.71023 2.9871 8.2585 1.6311 1.6019 3.7142 Bitter T 0.3341 0.4167 24.579 0.6507 1.115 0.3658 Umami T 1.9491 0.0948 21.2355 1.2563 1.0582 0.1799 ViscousThick MF 0.9412 0.305 139.2328 1.4324 1.1921 1.3115 Astringent MF 1.5199 0.8669 54.1689 1.3935 0.7956 0.9003 WarmHot MF 2.4404 0.0068 37.3172 1.2985 0.4014 0.2196 Green Apple F 0.9296 0.8328 5.1555 0.926 0.8763 0.404 Pear F 0.4807 1.4591 23.9595 0.5949 0.6073 1.7458 Other Fruit F 1.2672 0.9201 41.7183 0.7707 0.4027 2.2549 Grape F 0.4902 0.1362 23.7382 0.6943 0.4746 1.2012 Citrus F 3.2364 1.6032 43.8834 1.7453 1.098 0.3823 Floral F 1.0258 0.1907 21.443 0.8224 0.8633 0.6515 Earthy F 2.7308 1.432 27.2294 2.2268 1.7859 0.9889 Thiol F 2.6128 0.0002 36.3526 1.0415 0.5039 1.0077 CannedVeg F 1.1252 0.0005 17.1047 0.9628 0.5506 0.7334

83 Appendix Table 1 continued. ANOVA for Year 1 sensory attributes showing the F-values for the factors Wine (W), Fermentation Replicate (FR), Judge (J), and their interactions. Bolded F-values are significant (p < 0.05). Factor W FR J W:J W:FR J:FR DF 8 1 7 56 8 7 RottenEgg F 0.8456 1.4039 14.0358 1.8118 0.7294 2.7527 Sulfur F 1.4026 0.5049 43.2664 1.149 1.1522 0.3467 Yeasty F 1.5049 7.1638 125.8509 3.0934 0.887 5.1608 Oxidized F 0.8907 2.7423 19.6613 0.9856 1.9371 0.4941 ChemicalSolvent F 0.3083 3.0232 40.204 0.7731 0.3199 0.0637 Ethanol F 5.2121 1.4956 106.3201 1.2422 1.1409 1.1963

84 Appendix Table 2. ANOVA for Year 2 sensory attributes showing the F-values for the factors Wine (W), Fermentation Replicate (FR), Judge (J), and their interactions. Bolded F-values are significant (p < 0.05). Factor W FR J W:J W:FR J:FR DF 6 1 8 48 6 8 Yellow Color 71.5429 1.4075 18.5469 5.1505 2.468 1.4342 Haziness 457.2942 2.5551 36.2051 5.3688 11.6356 1.226 Green Apple A 0.9842 0.3143 17.5022 1.2524 0.7881 0.7784 Pear A 3.4361 4.6044 14.1936 1.5055 0.8262 0.9044 Other Fruit A 1.3232 0.0256 26.6302 1.3168 0.5998 0.8937 Grape A 0.8078 0.379 16.7828 0.7771 1.0046 0.5259 Citrus A 4.6851 0.2612 36.3845 2.1269 0.2257 1.0963 Floral A 2.3137 1.1701 17.1813 1.4164 0.7683 1.4034 Earthy A 1.5877 0.5087 26.9782 1.4491 0.5021 0.6841 Thiol A 13.8627 0.1151 17.3366 0.9554 1.472 0.3301 CannedVeg A 2.9468 2.6759 50.9468 1.3882 0.5577 0.3013 RottenEgg A 0.7955 0.232 18.1385 1.0419 1.3697 0.179 Sulfur A 2.1208 0.1264 55.7862 1.3646 1.4597 1.621 Yeasty A 2.018 3.0704 75.7223 1.2475 0.4882 1.7676 Oxidized A 2.0349 2.0729 46.2555 1.5096 0.4553 1.0905 ChemicalSolvent A 2.173 2.2366 70.0558 1.724 1.2862 1.9582 Ethanol A 1.2701 0.1046 41.8155 1.7883 0.9341 0.5344 Sour T 13.6751 0.2475 40.2103 2.0936 0.2377 1.0862 Sweet T 14.3618 0.0049 65.5766 1.7736 1.6289 0.7862 Salty T 2.5907 1.1292 26.7285 0.937 1.6423 0.1191 Bitter T 2.974 0.4633 27.8264 1.6723 0.8818 0.2096 Umami T 2.1608 1.287 65.5446 1.1634 1.3348 1.0948 ViscousThick MF 2.4833 0.031 76.8036 1.5634 0.4837 0.2175 Astringent MF 5.7097 1.4053 49.2053 1.0906 0.2433 0.3737 WarmHot MF 0.9309 2.128 23.7111 0.9801 0.7974 0.4532 Green Apple F 2.938 5.8585 22.6736 1.3776 2.1127 1.3618 Pear F 0.9038 3.8534 34.4786 1.5438 1.7477 1.175 Other Fruit F 4.0516 0.1282 81.5271 2.3359 1.3999 1.6899 Grape F 0.9653 0.0401 62.7602 0.7977 1.2683 0.3332 Citrus F 15.8517 1.1568 34.4787 3.2374 0.358 0.6443 Floral F 1.0764 0.7222 80.363 1.0772 0.768 0.5254 Earthy F 1.4019 0.1002 56.4623 1.8599 0.5909 0.8135 Thiol F 14.5672 0.017 96.2999 2.9494 0.4156 1.1568 CannedVeg F 0.4605 0.0727 165.4384 1.2343 1.4963 0.6458 RottenEgg F 0.7588 0.4646 63.7974 1.2712 0.6534 1.5029 Sulfur F 2.3352 0.0306 85.799 1.4537 0.473 0.6424 Yeasty F 1.3828 0.0202 194.7571 1.2158 0.9004 0.8915 Oxidized F 2.1783 6.3646 72.9348 2.0949 0.6308 1.3212 ChemicalSolvent F 2.2333 0 106.4048 1.0963 0.1764 0.6664 Ethanol F 0.8931 1.9043 32.3386 1.5583 1.2128 1.7381

85

Appendix Table 3. Wine chemistry data for Year 1 wines.

Titratable Residual Malate Lactic Volatile Free SO Total SO Fermentation % pH Acidity 2 2 Sugar (g/L) (g/L) (g/L) Acidity (g/L) (ppm) (ppm) Wine Replicate Alcohol (g/L) NC1 1 12.5 1.9 3.29 6.5 2.4 0 0.26 39 136 NC1 2 12.4 2.3 3.27 6.4 2.4 0 0.27 45 141 SE4 1 13.2 4.1 3.34 5.8 1.7 0 0.27 51 154 SE4 2 13.2 4.5 3.31 5.9 1.7 0 0.3 42 144 SE3 1 12.6 0.7 3.38 6.1 2.3 0 0.29 54 165 SE3 2 12.7 0.8 3.39 6.1 2.3 0 0.3 49 163 SE2 1 12.6 1 3.31 5.7 1.6 0 0.26 46 135 SE2 2 12.6 0.8 3.29 5.7 1.4 0 0.29 42 132 SE1 1 11.3 0.1 3.15 6.1 1.6 0 0.17 78 201 SE1 2 11.0 0.1 3.18 6 1.6 0 0.24 87 196 NW2 1 13.1 0.7 3.2 6 1 0.4 0.32 26 91 NW2 2 13.1 0.7 3.2 6 1.1 0.4 0.31 48 126 NE1 1 12.5 0.8 3.17 6.3 1.4 0.3 0.27 28 121 NE1 2 11.9 0.9 3.17 6.6 1.3 0.8 0.27 35 124 NW1 1 13.2 1 2.99 6.6 1.3 0.5 0.23 39 102 NW1 2 13.2 1 3 6.6 1.2 0.5 0.25 42 106 SE5 1 11.7 0.5 3.26 6 1.6 0 0.33 48 138 SE5 2 11.6 0.5 3.21 6.4 1.6 0.4 0.3 49 148

86

Appendix Table 4. Wine chemistry data for Year 2 wines.

Titratable Residual Malate Lactic Volatile Free SO Total SO Fermentation % pH Acidity 2 2 Sugar (g/L) (g/L) (g/L) Acidity (g/L) (ppm) (ppm) Wine Replicate Alcohol (g/L) NC1 1 12.6 9.6 3.21 6.9 2.1 0.2 0.24 15 92 NC1 2 12.8 8.3 3.24 6.4 1.9 0.0 0.29 16 92 SE3 1 12.9 2.3 3.12 7.8 2.0 0.3 0.30 11 103 SE3 2 13.2 2.3 3.17 7.7 2.6 0.2 0.27 13 107 SE2 1 12.5 4.2 3.10 7.0 2.0 0.2 0.18 13 104 SE2 2 13.0 3.2 3.29 7.2 2.2 0.0 0.09 10 101 SE1 1 12.4 1.0 3.18 7.2 1.9 0.4 0.26 10 70 SE1 2 12.6 0.9 3.20 7.2 2.0 0.3 0.31 9 71 SE1 3 12.6 0.8 3.16 7.3 1.9 0.5 0.28 7 68 NW2 1 12.3 3.4 3.01 7.6 2.5 0.0 0.31 11 79 NW2 2 12.4 2.6 3.05 8.2 2.6 0.3 0.27 10 80 NW1 1 11.8 0.8 2.81 9.2 2.6 0.5 0.23 7 40 NW1 2 11.9 0.6 2.89 9.2 2.8 0.4 0.11 4 36 NW3 1 12.8 3.1 3.05 7.8 2.1 0.0 0.10 11 56 NW3 2 11.7 2.3 3.05 8.0 2.1 0.1 0.15 11 54

87 Appendix Table 5. Harvest and juice chemistry results for Year 1 experimental sites. Titratable Site Yield (gal) TSS (°Brix) pH Acidity (g/L) YAN (mg N/L) NC1 13.5 14.2 3.22 7.409 248.18 NE1 13 16.4 3.25 7.708 221.17 NW1 16.2 18 3.06 7.331 57.55 NW2 13 19.6 3.41 5.683 128.2 SE1 0.75 14.2 3.56 6.446 98.93 SE2 11 16.4 3.36 6.152 177.26 SE3 13 15.6 3.52 6.428 241.06 SE4 9 16.4 3.28 6.256 215.96 SE5 3 16 3.48 5.542 261.24

Appendix Table 6. Harvest and juice chemistry results for Year 2 experimental sites. Titratable Site Yield (gal) TSS (°Brix) pH Acidity (g/L) YAN (mg N/L) NC1 12.25 20 3.29 6.731 139.303 NW1 18 14 3.04 9.514 105.512 NW2 15 19.2 3.11 8.545 113.484 NW3 8.25 16.6 3.07 8.143 107.6 SE1 19 18 3.69 5.362 201 SE2 11.25 19.6 3.51 6.809 137.5 SE3 13 19.6 3.75 7.118 157.8

88 Appendix Table 7. Volatile compounds that significantly differed by juice or wine samples for both years of study, retention time, and calculated and reported retention indices (RI). Retention Calculated Reported Volatile Compound time (min) RI RI Reference 1-Hexanol 8.34558296 1315.50938 1316 (Riu-Aumatell et al., 2004) 1-Hexanol 8.42775059 1331.29168 1332 (Yamaguchi & Shibamoto, 1981) 1-Hexanol 8.56 1337.82193 1337 (Tressl et al., 1978) 1-Hexen-3-ol 6.90506649 1222.64884 1225 (Vinogradov, 2004) 1-Propanol 4.13665009 1027.07918 1014 (Slizhov & Gavrilenko, 2001) 2-ethylhexanol 10.1248665 1460.1617 1470 (Selli et al., 2006) 2-Heptenal 7.8798337 1291.19787 1287 (Chen et al., 1986) 2-Hexen-1-ol 9.11923313 1364.56773 1363 (Khan et al., 2006) 2-Hexen-1-ol 9.0392828 1376.55683 1380 (Buttery et al., 1982) 2-Hexen-1-ol acetate 8.05836678 1303.95017 1305 (Morales et al., 1996) 2-Hexenal 6.5069499 1194.70403 1205 (Yamaguchi & Shibamoto, 1981) 3-Hexen-1-ol 8.44006634 1332.20328 1334 (Nishimura et al., 1989) Acetic acid 9.35941696 1400.26737 1400 (Shibamoto et al., 1981) Acetoin 7.24366665 1246.46038 1255 (Nishimura et al., 1989) Amyl acetate 5.8564167 1149.40228 1160 (Spencer et al., 1978) Butyl acetate 4.49166679 1052.40134 1049 (Osorio et al., 2006) d-Limonene 6.13953352 1163.30107 1180 (Viña & Murillo, 2003) Ethanol 2.96496654 925.494458 925 (Umano et al., 1986) Ethanol 2.99976659 949.898544 940.7 (Castello et al., 1988) Ethyl acetate 2.54419994 919.756513 917 (Tressl et al., 1978) Ethyl butyrate 4.07918358 1022.98028 1026 (Rezende & Fraga, 2003) Ethyl decanoate 12.48 1621.06703 1624 (Riu-Aumatell et al., 2004) Ethyl hexanoate 7.07 1230.58321 1223 (Riu-Aumatell et al., 2004) Ethyl octanoate 9.89 1435.01805 1428 (Riu-Aumatell et al., 2004) Ethyl vinyl carbinol 5.62453318 1133.2544 1145 (Cantergiani et al., 2001) Hexanal 4.5984664 1060.70676 1066 (Chandravadana et al., 2005) Hexyl acetate 7.57 1266.14509 1258 (Aubert et al., 2005) Hotrienol 11.5531502 1575.65022 1570 (Politeo et al., 2007) Isoamyl acetate 5.14843321 1100.09981 1102 (Toda et al., 1982) Isoamyl alcohol 6.48430014 1185.55843 1187 (Tressl et al., 1978) Isobutanol 4.758533 1072.17285 1085 (Humpf & Schreier, 1991) Isobutanol 4.93493366 1084.01809 1092 (Tatsuka et al., 1990) (“Gas chromatography analysis of organic solvents using capillary Isobutyl acetate 3.80215001 1003.2204 1013 columns (No. 2),” 2003) Isubutanol 4.758533 1072.17285 1085 (Humpf & Schreier, 1991) Linalool oxide 9.53180027 1413.75589 1414 (Yamaguchi & Shibamoto, 1981) Methyl 2- methylbutyrate 3.70783329 995.908369 1010 (Rowan et al., 1996) Methyl acetate 2.14821672 863.349172 828 (Yamaguchi & Shibamoto, 1981) Methyl butyrate 3.42670012 969.260674 975 (MacLeod & Pieris, 1983) Methyl hexanoate 6.03701687 1161.97889 1177 (MacLeod & Pieris, 1983)

89 Appendix Table 7 continued. Volatile compounds that significantly differed by juice or wine samples for both years of study, retention time, and calculated and reported retention indices (RI). Retention Calculated Reported Volatile Compound time (min) RI RI Reference Methyl octanoate 9.12 1379.0287 1378 (Riu-Aumatell et al., 2004) Methyl octanoate 8.88528347 1349.73262 1352 (Bernhard et al.,1983) Nerol oxide 9.87244987 1440.41079 1450 (Jensen et al., 2001) (Yamaguchi & Shibamoto, Octanal 7.45708323 1261.46858 1270 1981) Octanoic acid 16.7 1931.52348 2070 (Tatsuka et al., 1990) Octyl acetate 10.0467997 1425.64671 1424 (Riu-Aumatell et al., 2004) p-Cymene 7.18005037 1241.98666 1245 (Bernhard et al.,1983) Phenylethyl alcohol 14.795517 1847.51261 1858 (Nishimura et al., 1989)