University of Florida Thesis Or Dissertation Formatting

DEVELOPING HIGH FLAVOR QUALITY PROCESSED TOMATO PRODUCTS

YAOZHOU ZHU

A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY

UNIVERSITY OF FLORIDA

2016

To my family and faculty advisors for their constant support and encouragement, especially my late grandfathers, who taught me how good fresh garden fruits and vegetables can taste

ACKNOWLEDGMENTS

I would like to express my utmost gratitude to my committee chair, Dr. Paul J.

Sarnoski for the tremendous amount of time he has given me throughout this research and his invaluable assistance with all the difficulties I have encountered along the way. I would like to thank Dr. Charles Sims, Dr. Lisa House, Dr. Maurice Marshall and Dr.

Harry Klee for spending precious time to help me build my career and providing useful suggestions about my research whenever I needed it. Your guidance and advice will stay with me for many years to come. I also appreciate the support I received from Dr.

Renee Goodrich-Schneider, Dr. Keith Schneider and Dr. George Baker, whose office doors were always open for me. The faculty members in the Food Science and Human

Nutrition Department at UF have also played a crucial role in my academic career. I am particularly thankful to Dr. Asli Odabasi Kirli, Dr. Harry Sitren, Dr. Susan Percival, Dr.

Jesse Gregory, Dr. Anita Wright, Dr. Liwei Gu and Dr. Michelle Danyluk.

I would like to acknowledge Dr. Denise Tieman, Ms. Sara Marshall, Ms. Bridget

Stokes, Dr. Yavuz Yagiz, and Ms. Isabella Goddoy for patiently assisting me with tomato harvest and laboratory techniques, Mr. Andres Hincapie for the survey and Mr.

Nathan Mechulan, Mr. Kaipeng Xu, Mr. Changjie Xu, Ms. Natalia Bedoya, and Mr.

Stephen Koltun for their help on tomato processing.

Special thanks go to Ms. Marrianne Mongone, Ms. Mary Spitzer, Ms. Shelia

Parker-Hall, Mr. Herschel Johnson, Ms. Jenna Grogan, Ms. Mindy Edwards and Ms.

Julie Barber for their help with the seemingly endless forms and paperwork associated with graduate school. My colleagues Taylor Dole, Kelly Brown, Michael Torti, La’Oshiaa

Reed and Jing Bai made my lab life joyful. Thanks are also due to Dr. Changqi Xiu, Dr.

Meng Shen, Dr. Kayla Ou, Dr. Changmou Xu and Dr. Lu Zhao for their helpful insights related to this field of research.

I am especially thankful and indebted to my family, who have given me so much support and love over the years. In particular, I would like to thank my parents, my husband Haomiao Fan and my in-laws for the solid foundation they have provided throughout; it is only as a result of their invaluable advice and constant encouragement that I have been able to successfully achieve my goals.

Lastly, I would like to thank my friends, who have played a crucial role in my happiness throughout this research and my life here in Gainesville, Tallahassee, and

Shanghai.

TABLE OF CONTENTS

page

ACKNOWLEDGMENTS ...... 4

LIST OF TABLES ...... 9

LIST OF FIGURES ...... 11

ABSTRACT ...... 12

CHAPTER

1 INTRODUCTION ...... 14

2 LITERATURE REVIEW ...... 16

Tomato Varieties and Common Uses ...... 16 What Is Garden Gem?...... 18 Tomato Products ...... 19 Factors Affecting Tomato Flavor ...... 22 Environmental and Seasonal Effects on Tomato Flavor ...... 22 Effect of Processing on Tomato Flavor ...... 23 Flavor, Volatiles and Instrumental Analysis ...... 28 Flavor Isolation ...... 30 Flavor Identification ...... 34 Bound Volatiles ...... 38 Consumer Acceptance of Food Products ...... 41 Consumer Perceptions of Tomatoes ...... 41 Processed Product Acceptability ...... 43 Exploring Palatability and Willingness to Pay for Fruit Products ...... 44

3 SENSORY AND CHEMICAL CHARACTERISTICS OF TOMATO JUICE ...... 53

Abstract ...... 53 Background Information and Objectives ...... 54 Materials and Methods...... 56 Tomatoes ...... 56 Juice Processing ...... 56 Fresh Juice Production ...... 56 Flavor Isolation ...... 57 Identification and Quantification ...... 57 Sensory Analysis ...... 58 Chemical Analysis ...... 59 Statistics ...... 59 Results and Discussion...... 60 Chemical Analysis of the Products ...... 60

Sensory Analysis Results ...... 62 Correlations between Overall Liking and Sensory, Chemical and Flavor Compounds for Processed Juice ...... 63 Volatile Metabolites Identified using Different Columns ...... 65 Volatile Metabolites of Fresh Tomato Cultivars ...... 66 Volatile Metabolites of Garden Gem, Roma and Commercial Juice ...... 67 Seasonal Effect for Garden Gem, Roma and Commercial Juice ...... 69 Summary ...... 70

4 CONSUMER PERCEPTIONS OF TOMATO JUICE ...... 83

Background Information and Objectives ...... 83 Materials and Methods...... 85 Tomatoes ...... 85 Juice Processing ...... 85 Sensory Analysis Survey ...... 85 Online Survey ...... 87 Choice Experiment ...... 88 Models and Statistics ...... 89 Group Segmentation ...... 91 Results and Discussion...... 91 Sensory Analysis Demographic and Opinion ...... 91 Panelists’ Willingness to Pay ...... 94 Online Survey Demographic and Opinions...... 95 Tomato Juice Consumption and Purchasing Opinions ...... 96 Survey Willingness to Pay ...... 98 Summary ...... 99

5 TOMATO ESSENCE CREATION AND CHARACTERIZATION ...... 113

Abstract ...... 113 Background Information and Objectives ...... 114 Materials and Methods...... 116 Tomatoes ...... 116 Juice Processing ...... 116 Fresh Juice Production ...... 117 Essence Production ...... 117 Flavor Isolation ...... 117 Identification and Quantification ...... 118 Statistics ...... 118 Results and Discussion...... 118 Flavor Profile Change after Processing ...... 118 Flavor Profile of Tomato Essence ...... 123 Summary ...... 128

LIST OF REFERENCES ...... 136

BIOGRAPHICAL SKETCH ...... 151

LIST OF TABLES

Table page

2-1 Different stages of tomato ripening...... 47

2-2 Summary of various microorganisms, cases of tomato-related outbreaks for fresh tomatoes...... 48

2-3 Comparison of different isolation methods ...... 49

2-4 Common molecules occurring as glycosidically bound volatiles in tomatoes. .... 50

3-1 Chemical analysis of thermally processed tomato juice using Duncan’s multiple range test mean separation. Means (± standard deviation) with different letters within the same column are significantly different at p< 0.05. .... 72

3-2 Sensory ratings using a nine-point hedonic scale (1-dislike extremely, 9-like extremely) for thermally processed tomato juice (2015 Summer Season) with Duncan’s multiple range test mean separation...... 72

3-3 Sensory means of thermally processed tomato juice (2015 Fall Season). Means were separated using Duncan’s multiple range test for mean separation...... 72

3-4 Significant correlations (r) for sensory analysis of Summer 2015 season processed tomato juice (significant at p≤0.05) ...... 73

3-5 Significant correlations (r) for sensory analysis of Fall 2015 season processed juice (significant at p≤0.05) ...... 73

3-6 Correlation coefficient (r) for sensory, chemical and flavor analysis (significant at p≤0.05 *) of processed tomato juice for both seasons...... 74

3-7 Comparison of quantitative analyses of major volatiles (in ppb) for processed tomato juices...... 76

3-8 Major Volatiles in fresh Garden Gem (GG) and Roma (RA)...... 79

4-1 Profile analysis of the segments (Season 1) of sensory evaluation participants according to descriptive variables(N=119) ...... 101

4-2 Profile analysis of the segments (Season 2) of sensory evaluation participants according to descriptive variables(N=119) ...... 102

4-3 Purchase intent difference of sensory testing panelists before and after sensory analysis (Season 1)...... 103

4-4 Purchase intent difference of sensory testing panelists before and after sensory analysis (Season 2)...... 103

4-5 Name your own price of sensory testing panelists before and after sensory analysis (Season 1)...... 103

4-6 Name your own price of sensory testing panelists before and after sensory analysis (Season 2)...... 103

4-7 Characteristics preventing consumer purchase of tomato juice (N=238)...... 104

4-8 Occasions and frequency for tomato juice consumption (N=238)...... 104

4-9 Factor scores for sensory analysis (Season 1 and Season 2) segmentation according to tomato juice consumption opinions...... 105

4-10 Descriptive statistics of the online survey participants (N=983)...... 106

4-11 Health and lifestyle information related to tomato juice consumption of online survey participants (N=983)...... 107

4-12 Occasions and frequency for tomato juice consumption (N=983)...... 108

4-13 Label and tomato origin opinion of tomato juice purchasers (N=983)...... 109

4-14 Characteristics preventing consumer purchase of tomato juice (N=983)...... 110

4-15 Multinomial logit model estimates for respondents...... 111

4-16 Coefficient estimates in different willingness to pay models (N=909)...... 111

4-17 Premiums for tomato juice using Krinsky-Robb bootstrapping simulation...... 111

5-1 Changes in volatile compounds during thermal processing (Garden Gem)...... 129

5-2 Changes in volatile compounds during thermal processing (Roma)...... 131

5-3 Important volatile compounds in fresh tomato juice, tomato essence and fractions by GC-MS...... 133

5-4 Odor active values of important volatile components in tomato essence fractions by GC-MS...... 134

LIST OF FIGURES

Figure page

2-1 Relationship between flavor and instrumental analysis...... 51

2-2 General structure of bound volatiles ...... 52

3-1 Biplot of tomato juice with compounds on PC1 VS PC2 (Pearson model) explaining 81.14% of the summer season data variation...... 81

3-2 Biplot of tomato juice with compounds on PC1 VS PC2 (Pearson model) explaining 84.46% of the fall season data variation...... 82

4-1 Example of choice set ...... 112

5-1 Chromatograms of four tomato essence fractions...... 135

Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy

DEVELOPING HIGH FLAVOR QUALITY PROCESSED TOMATO PRODUCTS

Yaozhou Zhu

December 2016

Chair: Paul J. Sarnoski Major: Food Science

Despite a high nutrient content, processed tomatoes are among the least liked fruit products. This research therefore sought to characterize the flavor of processed products made from a superior Florida tomato variety and explore consumer perceptions of such products. The two pilot tomato juices (Garden Gem and Roma) and a popular commercially available tomato juice produced in two different seasons were compared using sensory evaluation and flavor analysis. The results of the sensory and flavor evaluation were further abstracted to nationwide sample of representative tomato juice consumers, who examined and evaluated important attributes related to tomato juice consumption and purchase behavior using a discrete choice experiment. Overall, the tomato juice made from Garden Gem tomatoes was significantly (p<0.05) liked by the sensory testing panelists for overall liking, tomato flavor, and sweetness, regardless of the seasonal variation. Garden Gem juice was found to contain significantly (p<0.05) higher sweet fruity related aroma compounds: 6-methyl-5-hepten-2-one, linalool and β- ionone. These results suggest that a premium quality tomato juice with fresh notes and better taste made from a premium tomato such as Garden Gem will encourage consumer purchase intent. Consumers stated that they would be willing to pay up to

30% more if trusted sensory review information was provided. Commercially produced tomato juice revealed that processed tomato juice had a markedly different volatile composition from fresh juice. The possibility of using a tomato essence to boost the fruity and green flavor note compounds that enhance the tomato flavor of thermally treated juice was therefore explored. The compounds 6-methyl-5-hepten-one (1793 ppb), α-citral (766 ppb), 3-hexen-1-ol (710 ppb), 2-isobutylthaziole (354.2 ppb) and geranylacetone (317.9 ppb) were found to be particularly dominant in tomato essence.

The findings of this research represent the detailed flavor profile of processed tomato products made from Garden Gem. They also demonstrate how sensory preference can confer a potential market advantage over existing commercial products. The creation of a tomato essence that has potential to add desirable flavor attributes to processed tomato products.

CHAPTER 1 INTRODUCTION

The tomato is one of the most widely used agricultural commodities in the world, and the United States is a leading producer of tomatoes. Florida produces 40% of the tomatoes in the United States, and supplies most of the fresh market (USDA, 2010).

Processed tomato consumption has generally increased over the past 50 years.

Tomatoes are a good-tasting source of vitamins and minerals such as Vitamin C, folic acid, potassium, and thiamin. Most noticeably, lycopene and other carotenoids found in tomatoes are antioxidants that have been shown to lower the risk of cardiovascular disease (Canene-Adams, Campbell, Zaripheh, Jeffery, & Erdman, 2005), prostate, lung, stomach, pancreas, colon, rectal, esophageal, breast and cervical cancer (Giovannucci,

1999). However, the tomatoes used to make processed tomato products are not usually high in flavor quality.

Flavor, which has a major impact on consumer purchases and consumption, is comprised of aroma, taste, and somatic sensation. Taste is perceived in the mouth by the sensations of sweet, sour, salty, bitter, and umami, a savory taste, while aroma is sensed in the nasal cavity and encompasses most of what is considered flavor. Most of these aroma compounds consist of volatile or semi-volatile compounds that are perceived by the orthonasal and retronasal olfactory systems. Orthonasally perceived aromas are taken directly through the nose and retronasally perceived aroma volatiles are produced when food is chewed and air is exhaled through the nasal cavity. The retronasal impact is most likely to leave a lasting impression on the individual.

Florida produces many fresh (traditional varieties, generally better tasting) tomatoes and these have tended to dominate the fresh tomato market in the state.

However, fresh tomatoes can be clearly linked to food safety outbreaks. It can take more than two weeks for regulatory authorities to identify the real source of an outbreak, which is sometimes not the tomato itself, but improper handling or cross contamination, and after two weeks of holding, the highly perishable tomatoes are no longer fit for sale.

Irregular sized fresh tomatoes are often donated to charity or dumped, even though they may have good flavor quality. Good quality raw materials are essential for producing good quality processed tomato products, and misshapen or otherwise undesirable tomatoes could thus be used in processed products to reduce waste, add value and enhance food safety. However, there is limited research on consumer perceptions of high-quality processed tomato products.

It is clearly important for a variety of sensory attributes and proper models to be applied when assessing a consumer’s willingness to pay for particular food products.

However, research in this area has tended to focus on product development solutions, food safety regulations or to simply accept the perceptions of consumers that there is a lack of good quality tomatoes available. This is the first research ever to explore the willingness of consumers to pay more for a good flavored-processed tomato product.

The overall objective of this research was therefore to develop a processed tomato product with high flavor quality using Florida-grown tomatoes. The quality was determined using an instrumental analysis of the aroma compounds and sensory evaluation sessions were conducted. The potential market for these processed tomatoes and consumer willingness to pay for this type of product was also investigated.

CHAPTER 2 LITERATURE REVIEW

Tomato Varieties and Common Uses

The word tomato (Solanum lycopersicum) usually refers to a round berry of a perennial plant that originated in the Andean regions of South America (Smith, 1994). It was considered a non-edible ornamental berry by the sixteenth century Spanish explorers who discovered it because they thought it resembled berries in the nightshade family, which are poisonous. The tomato was initially named “Lycopersicon”, meaning

“wolf peach”, a term used by the Greek naturalist Galen to describe a poison fruit. Later, botanists classified tomatoes in the genus “Solanum” based on their morphological and molecular characteristics (Peralta, Spooner, Razdan, & Mattoo, 2007). Interestingly, this classification places tomatoes in a sister group with potatoes.

Whether considered a fruit from the botanical perspective or a vegetable according to customs regulations (Nix v. Hedden,1893), tomatoes now appear frequently in different cuisines around the world, depending on the type of tomato (TGA,

2014). Classic round tomatoes (70-100g, 4.7-6.7 cm. in diameter) are among the most popular varieties and are widely used for salads, grilling, baking, frying, soup, and sauces. Cherry and cocktail tomatoes are smaller and much sweeter and may have a yellowish color depending on the variety. These types of tomatoes are commonly sold attached to the stem and are often eaten whole or sliced in raw foods (salads, snacks), rather than cooked (barbecue, sauces, soups, etc.). Plum tomatoes, which have a firm texture and a less juicy taste, are mainly consumed as an ingredient in prepared foods.

Beefsteak tomatoes are larger than other varieties, with twice the weight and a larger locular portion compared to the classic round tomatoes; the locules are surrounded by

the fleshy tissue of the tomato. A larger tomato is more likely to be sliced and used in sandwiches, burgers, and salads.

Vine tomatoes are a special category of tomatoes that are harvested by keeping the stems on the fruit to enhance flavor development after harvesting. Most high quality, heirloom tomato (cultivated for a long time and usually with good taste) varieties are sold at a higher price. Tomatoes are quite perishable and suffer physiological injury when exposed to cold temperatures below 12 °C. In order to prevent injury and lengthen shelf life, most fresh-market tomatoes are handpicked and harvested around stages 1 to

3 (where 0=immature, 1=mature-green, 2=breaker, 3=turning, 4=pink, 5=light-red,

6=red-ripe) well before the red-ripe stage (Table 2-1). Exogenous ethylene treatment is then applied to accelerate the ripening process. Tomatoes with thicker skins, picked at the red stage by mechanical harvesters, are mainly used for processed food products.

Tomatoes are available and relatively inexpensive all year, either as fresh or processed fruit. In 2012, 5130 kg of tomatoes were consumed every second worldwide

(FDA, 2014). Thanks to international free trade agreements, fresh tomatoes can be consumed year round in the US; in Florida, tomatoes are in season from mid-October to mid-June and imported into Florida from other sources such as California, North

Carolina, Canada, Mexico from July to September (FTC, 2014). In 2012, the average price for fresh tomatoes was $ 1.28/pound and for canned tomatoes it was $0.77/pound respectively in 2011 (Stewart, Hyman, Buzby, Frazao, & Carlson, 2011). Their attractive appearance has been favored ever since they were first discovered by the conquistadors and they continue to be cultivated as a common garden plant. The sour

and sweet taste of tomato is used as a supporting ingredient in many dishes by chefs in many different world cuisines.

A typical 100 g tomatoes consists of 95% moisture (Ensminger, Ensminger,

Konlande, & Robson, 1995) and a substantial amount of dietary fiber, Vitamin A,

Vitamin C, Vitamin B, beta-carotene, calcium, phosphorus, iron, sodium, potassium and zinc. As a good source of micronutrients and phytonutrients, tomatoes are a recommended way to address micronutrient malnutrition according to the FAO (Chadha

& Oluoch, 2003). Tomatoes are popular among weight-conscious populations as a low- calorie food. As noted earlier, the lycopene and other carotenoids in tomatoes are thought to lower the risk of cardiovascular disease (Canene-Adams, Campbell,

Zaripheh, Jeffery, & Erdman, 2005), prostate, lung, stomach, pancreas, colon, rectal, esophagus, breast and cervix cancers (Giovannucci, 1999).

What Is Garden Gem?

Garden Gem is a new hybrid premium tomato product of two tomato varieties

(Maglia Rosa and FLA-8059) that was developed by Dr. Harry Klee’s research group, in

2011 (Schatzker, 2015). The Malia Rosa is a productive cherry tomato that is one of the most popular heirloom varieties. As a typical heirloom tomato, the Maglia Rosa has a superior taste but is very susceptible to disease and insect pests. The other parent variety “FLA-8059” is a fresh-market tomato line released in 2006 that has high disease resistance and a relatively long shelf life. The FLA 8059 is also a parent of Tasti-Lee, a popular grocery store tomato (Scott, Baldwin, Klee, Brecht, Olson, Bartz, et al., 2008).

The Garden Gem was reported to be statistically identical to the Maglia Rosa in taste tests and one of the top 2 among thirty-seven popular tomato varieties (Klee,

2015). This high yield hybrid crops early, needing only about two months from

transplanting to harvest. The 2-2.5 ounce tomatoes produced are good for salads, marinara, and high-end sauces. The possibility of making premium tomato juice using

Garden Gem is thus worth exploring.

Tomato Products

Thanks to the development of modern food technology and logistics, perishable fruit like the tomato can now be preserved and processed promptly, thus reducing the amount of deterioration that occurs and the possibility of foodborne illness (Table 2-2).

The thicker skin tomatoes picked by mechanical harvesters are mainly used as ingredients in processed food products. The state of California leads the US in processed tomato production, with 12.9 million tons of canned, frozen and dried tomatoes produced in the year 2012 (Thornsbury & Jerardo, 2012). The acidic content of the tomato makes it a prime candidate for canning because the low pH environment prevents pathogen growth, which is one of the main reasons the tomato was being preserved by canning more than any other fruit or vegetable by the end of the nineteenth century.

Tomatoes can be processed in a number of different ways, including as juices, purees, ketchup, salsas, canned tomato products, and sauces. The thicker skin tomatoes used for processing are mechanically harvested at the red ripe stage and then sold under contract between farms and food manufacturers at a relatively lower price.

California produces most tomatoes for processing. In contrast, Florida produces mostly fresh tomatoes, supplying more than 50% of all the fresh market tomatoes produced in the US. These tomatoes are sold in supermarkets and grocery stores and are supplied to food service industry such as fast-food chain restaurants McDonalds and Taco Bell

(McDonald's to Pay Tomato Farmers More, 2007). The price of fresh tomatoes can

fluctuate depending on demand (Pleven & Cui, 2010). However, when foodborne illness outbreaks are reported to be possibly associates with tomatoes from a particular source, even when this has not been officially confirmed, the demand for all tomatoes, including those from sources not linked to the suspect source, declines sharply and tons of fresh tomatoes may have to be discounted, donated to a food bank or left to rot in the field (Zhang & Etter, 2008). There is no report of Florida’s fresh tomatoes being processed to extend their shelf life or used in processed food products. Interestingly, most fresh tomatoes purchased from supermarkets are cooked by consumers anyway – few are eaten raw.

Americans consume three-fourths of their tomatoes in processed form. The

United States consumption of processed tomatoes began a steady climb that accelerated in the late 1980s with the rising popularity of pizza, pasta, and salsa. The largest processed use of tomatoes is in sauces (35 %), followed by paste (18 %), canned whole tomato products (17 %), and catsup and juice (each about 15 %)

(Heuvelink, 2005). About one-third of all processed tomato products are purchased away from home at foodservice outlets such as pizza parlors and restaurants. Food manufacturers usually concentrate tomatoes to reduce transportation costs and extend shelf life.

Tomato juice contains unfermented liquid extracted from mature tomatoes or reconstituted from concentrated juice or paste for direct consumption. Peel, seeds, and other coarse objects are removed from the juice, which is generally concentrated and later reconstituted with water and/or tomato juice to a tomato soluble solids content of no less than 5% percent by weight, according to the standards of identity of federal

regulation (FDA, 2015). Homogenized tomato juice may contain added ingredients such as salt, organic acids, and other flavoring ingredients. Tomato juice is preserved by heat treatment, sterilization (canning), refrigeration, or freezing. However, tomatoes have been reported as the cause of foodborne illness in numerous cases. Table 2-2 lists the microorganisms known to be associated with cases of foodborne illness deriving from fresh tomato consumption since 1990.

Although no illness has been reported related to processed tomato consumption, several voluntary recalls have been posted due to the potential presence of botulism in canned tomatoes/canned tomato juice (Peng, Mah, Somavat, Mohamed, Sastry, &

Tang, 2012; Pritzker, 2014; Sandoval, Barreiro, & Mendoza, 1992). Clostridium botulinum has not been reported as associated with fresh tomato because of its strictly anaerobic characteristics.

Most processed tomato products such as sauces and paste are heated in an effort to concentrate the product, and although this method does lower the microorganism count, it can also reduce the palatability of the resulting product.

Minimum processing has gained popularity in recent years because in addition to being sufficient to inactivate foodborne pathogens, it also preserves the flavor. A temperature of 165°F (74°C) or above for 15s at the center of the product generally reduces the bacteria to an acceptable level that poses no health risk (FDA, 2011; Schneider,

Schneider, Hubbard, & Richardson, 2014). Theoretically, certain strains of thermal resistant molds can survive at 85°C for 20 minutes if initial numbers are above 105

CFU/ml (Silva & Gibbs, 2004), but no cases of illnesses caused by mold in these products have been reported.

Factors Affecting Tomato Flavor

Environmental and Seasonal Effects on Tomato Flavor

As a crop, the tomato plant requires an optimum temperature for maximum yields and quality, though the plant can adapt to different climates. Seasonal differences refer to the same fruit cultivar grown in different dates and/or locations. The environmental factors affecting tomato flavor have been investigated, but the main focus has been on seasonal effects on fresh fruit and there have been few reports of research being conducted on the quality of processed tomato products using raw material grown at different times of the year.

Temperature and light have been reported to be important factors for both the growth of tomato plants and flavor development. Color, shape, size, soluble solids (SS), titratable acidity (TA) and aroma compounds may all be affected by the growth temperature (climate) and post-harvest storage. Light duration and intensity may also influence the content of the various flavor components. This effect may be associated with the fruit’s metabolism and color development. For example, inferior taste has been linked with autumn season and greenhouse strawberries compared to those grown in the summer and in fields (Anza, Riga, & Garbisu, 2006). Higher soluble solids and reduced sugars and sugar/acid ratios have been recorded in fruits grown on the plains compared to those grown in the mountains, while fertilization and irrigation can also affect tomato flavor. Tomatoes with higher levels of TA, SS, and the volatiles hexenal,

2-hexanone, benzaldehyde, phenylacetaldehyde, α-ionone, and 6-methyl-5-hepten-2- one have been grown by increasing the supply of N–K or N levels within certain ranges by changing the fertilizer used (Wang, Huang, Liu, & Jin, 2007; Wang, Liu, Huang, &

Jin, 2009; Wright & Harris, 1985). Restricting the water supply, though not sufficiently to

cause water stress, produces tomatoes that are higher in sugars, titratable acids, vitamin C, and aroma volatiles (Veit‐Köhler, Krumbein, & Kosegarten, 1999) and differences in the concentrations of volatile compounds between field-grown, greenhouse-grown, and artificially ripened tomatoes have been reported (Dalal,

Salunkhe, Olson, Do, & Yu, 1968; Salunkhe, Jadhav, & Yu, 1974). In general, high concentrations of the aroma volatiles are found in field-grown tomatoes and greenhouse-grown tomatoes contain the lowest concentrations.

Effect of Processing on Tomato Flavor

Approximately 400 compounds have been identified on tomato and only a few of have an effect on flavor. Tomato flavor is largely affected by the variety, environment, and processing conditions. Differences between the flavors of different varieties arise due to differences in the quantitative proportions of the volatile substances and the sugar/acid ratio. The environmental conditions can affect the tomato fruit development, which in turn affects the volatile substances developed during ripening and post- harvest. Fresh tomato flavor can be affected by the conditions discussed previously and the aspects of flavor released during consumption. Unlike fresh tomatoes, processed tomatoes undergo a standard procedure during post-harvest and processing; value added processing also plays a vital role in a product’s unique flavor.

Processed tomato flavor comes partly from the hydrolysis of the glycosides present in tomato fruits. Please see the Bound Volatiles section later in this chapter for more details. Buttery, Teranishi, Ling and Turnbaugh (1990) isolated a glycoside fraction from fresh tomatoes and hydrolyzed it at various pH levels to give the volatile aglycone. The major volatiles identified that were present at all pH levels included 3- methyl-butyric acid, β-damascenone, phenylacetaldehyde, 2-phenylethanol, linalool,

linalool oxides, α-terpineol, 4-vinylguaiacol and 4-vinylphenol. Buttery, Teranishi, Ling and Turnbaugh (1991) also identified a number of other compounds present in smaller proportions.

Tomatoes can suffer significant volatile loss during processing, with many compounds being lost during heat processing due to their high volatility. For example,

C-6 products, which are mostly responsible for the green note in tomatoes, are highly vulnerable; although slightly increased during cell breakage, the C-6 compounds are reduced substantially during processing and the amount of this reduction depends on the processing conditions used. These C-6 products are oxidation products of linolenic acid in the presence of oxygen and enzyme (lipoxygenase, alcohol reductase), therefore conditions that influence lipoxygenase activity during processing determine the loss or retention of C-6 products. The temperature used not only affects the volatility of these C-6 compounds, but also determines the interactions between them. When heat is applied at 42 °C, cis-3-hexenal is the predominate form during vacuum distillation, but when distilled at 100 °C, trans-2-hexenal becomes the predominate form

(Stone, Hall, & Kazeniac, 1975). Buttery et al. (1990) identified 21 additional volatile compounds in tomato paste that had not been previously reported, with dimethyl sulfide and 1-octen-3-one being the most potent odorants. They attributed the creation of these compounds to the hydrolysis of glycosides.

Hydrogen sulfide and methyl sulfide have also been reported to affect the overall aroma of processed tomato products. Buttery et al. (1990) reported that methyl sulfide with an odor threshold value of 0.3 ppb is consistently found in the aroma of heated tomato products and linalool, with an odor threshold value of 6ppb, is another

compound that contributes to the cooked or fruity aroma of processed tomato products.

The precursor for the methyl sulfide was determined to be an S-methyl methionine sulfonium salt that when heated yields homoserine and methyl sulfide. Processed tomato products contain several sulfur-containing amino acids, all of which may react during processing to produce sulfides. During tomato processing, the compounds responsible for flavor development undergo changes that may or may not be desirable; some of the volatile compounds evaporate while others are formed due to the breakdown of sugars and carotenoids. Depending on the time and temperature used during the processing, sugars and amino acids decrease and the acid content rises due to the formation of pyrrolidone carboxylic acid (Gancedo & Luh, 1986).

The undesirable flavors that occasionally appear in heat-treated vegetable products have been attributed to the formation of pyrrolidone carboxylic acid.

Shallenberger and Moyer (1961) reported pyrrolidone carboxylic acid to have an unpleasant phenolic or bitter flavor, particularly in the pH range of 5-6. Nelson and Hoff

(1969) reported a marked difference in the volatile components of raw and heat- processed tomato samples, commenting that methyl sulfide, which was absent in the fresh fruit, was likely formed during heat processing. They also reported significant changes in the concentrations of acetaldehyde, acetone, methanol and hexanal due to processing. Sieso and Crouzet (1977) also reported the effect of processing on the volatiles present in tomato products, focusing mainly on canned tomato juice and paste.

They identified 16 components and indicated that the important heat-induced volatiles produced included furfural, linalyl acetate, and 2-methyl-2-hepten-6-one. Chung,

Hayase and Kato (1983) reported a decrease in low boiling volatiles and an increase in

middle and high-boiling compounds during processing of juice into paste. Furfural, which is formed by the dehydration of sugars, is considered a negative indicator of fruit juice quality as it. Furfural can affect the flavor in various ways. At concentrations above

3000 ppb in water, it produces a sweet or slightly burnt aroma. At concentrations below

2mM, furfural has been reported to exert an inhibitory effect on saccharomyces and alcohol dehydrogenase (Modig, Liden, & Taherzadeh, 2002). Alcohol dehydrogenase is reported to play a vital role in the formation of C-6 compounds could thus easily affect the overall flavor (Feussner, 2002).

In addition to amino acids and fatty acids, the pigments and sugar components in tomatoes play a major role in determining the flavor of processed products. The carotenoids in tomato products undergo oxidative degradation during processing leading to the formation of terpenes and terpene-like compounds. Cole and Kapur

(1957) reported the formation of 6-methyl-hepten-2-one and acetone when lycopene becomes oxidized, while Ilg, Bruno, Beyer and Al-Babili (2014) showed that farnesyl acetone and generylacetone are derived from lycopene. The formation of α and β- ionones, toluene, and para-xylene has also been reported in heat-processed tomato products. β-carotene is destroyed in an oxidation coupled reaction with linoleate which is catalyzed by lipoxidase. Blain, Patterson and Pearce (1968) identified a water-soluble fraction in tomato that bleached carotenoids in the presence of linoleate and Stevens

(1970) found a relationship between the polyene-carotene content and the volatile composition of tomatoes. Many Maillard reaction products, volatile carbonyl and sulfur compounds, are formed during processing and these may also affect the aroma of processed tomato products, as do other decomposition products reported in tomato

products such as furan, pyrrole and pyrazine, which also influence the aroma of processed tomato products.

In tomato products, a critical reaction during thermal processing is the degradation of the red pigment lycopene. Originally in the trans form, lycopene isomerizes to the cis form when heated resulting in a significant color change due to heat. Moreover, the changes in color and flavor can also be affected by non-enzymatic browning. The consistency of tomato products, which refers to their viscosity and the ability of their solid portion to remain in suspension throughout the shelf life of the product, is strongly affected by the pectin composition. Controlling the breakdown or retention of the pectin and the enzymes that affect the pectin is thus of great importance during processing as the degradation of the pectin chains reduces the viscosity of the juice. Two enzymes, pectin methylesterase (PME) and polygalacturonase (PG) are involved in the breakdown of pectins. The action of PME makes the pectin susceptible to further degradation by PG because PG acts only on segments of the pectin chain that have already been demethylated by PME.

Two different processes are commonly used in the production of tomato products. In the hot-break process, tomatoes are rapidly heated to 90–95°C immediately after homogenization. This process inactivates enzymes rapidly, particularly those involved in pectin degradation, producing a product with high viscosity. Temperatures higher than 78°C for 40 s and 90°C for 5 min, completely inactivate PME and PG, respectively, in tomato juices. Carotenoid oxidative products such as ß-ionone, ß-damascenone, geranyl acetone and 6-methyl-5-heptene-2-one that are responsible for the fruity, floral flavor of tomato are relatively stable during the

heating process and are considered desirable flavor compounds in both fresh and processed tomatoes. In the cold-break process, the homogenized tomatoes are heated only to around 60°C. This lower temperature process reduces the amount of thermal abuse the product undergoes, supporting a greater retention of color and flavor components and reducing the production of undesirable compounds. However lower temperatures may not entirely inactivate the enzymes PME and PG, allowing them to break down more of the pectin and thus reducing the viscosity of the juice produced.

The major advantage of cold break over hot break is that the final product has a more natural color. However, Fonseca and Luh (1976) found that aroma and flavor were rated better for hot break tomato juice. The hot break method is generally used to produce those tomato products where a fresher aroma is desired.

Pasteurization is critical to maintaining product safety. High temperature short time (HTST) is a standard method used to pasteurize fruit juices, although the retort can also be employed as an alternative method for pasteurization. Low temperature long time (LTLT) is the least preferred pasteurization method for maintaining the quality of juices in terms of nutrient and flavor retention, but it is less expensive to implement and can be more affordable for small-scale processors.

Flavor, Volatiles and Instrumental Analysis

Flavor is one of the most important reasons for the pleasure associated with food consumption. We perceive the minor components of flavor in a delicate way. Our brain works very efficiently to collect and organize the information it receives from the nose, mouth, tongue and somatosenses (images, sounds, tactile elements) (Auvray &

Spence, 2008). Olfaction (smell) and gustation (taste) are the major components of the human perception system related to flavor. Taste is linked with perceptions of

sweetness, bitterness, sourness, saltiness and umami from the taste buds; while smell refers to the aromas (volatile molecules) perceived. However, recent evidence has indicated that volatiles may contribute more towards flavor perception than we previously thought because they may not only be aroma-active but also linked to basic tastes such as sweetness (Schwieterman et al., 2014; Tieman, et al., 2012; Vogel,

Tieman, Sims, Odabasi, Clark, & Klee, 2010). Volatile compounds can be introduced through inhalation (orthonasal olfaction), or bought through the nasal epithelium by inhalation when chewing (retronasal olfaction). The 853 research studies devoted to

“tomato volatiles” and the 706 to “tomato flavor” listed on the Web of Science Database

(Tompson Reuters, New York, USA) since 1950 utilized, a variety of approaches for analyzing tomato flavor as part of the effort to understand the characteristics of fresh fruit and its processed products, extend the shelf-life, improve the quality and develop new and innovative products. These studies all involved either sensory evaluation, instrumental analysis, or both.

The underlying rationale for the sensory evaluation method is to use human participants, as the ultimate consumer, as an instrument to evaluate and validate the differences, preferences or attributes that characterize a food (Lawless & Heymann,

1988; Meilgaard, Carr, & Civille, 2006). Sensory evaluation provides a general picture of flavor perception from the consumers’ perspective, but the components that stimulate this perception may not be easily communicated or named. Outputs usually include difference (yes/no or intensity descriptors) or preference measures for a food; information regarding which component, and why and how this affects preference/attributes is not normally included. Where attempts are made to include this

information (descriptive analysis), the necessary sensory evaluation methods are generally expensive, time consuming and suffer from a lack of reproducibility.

Instrumental analysis is often used to provide supplementary information that addresses many of the shortcomings of sensory evaluation (Figure 2-1) as it provides standardized and repeatable values, enabling individual compounds to be isolated and qualitatively/quantitatively identified in a consistent and repeatable manner. In principle, the compounds identified can then be combined to reproduce the flavor analyzed and thus guide quality improvement or flavor innovation (Buttery, Teranishi, & Ling, 1987).

However, in practice, although these approaches may provide useful information, they also include information and compounds that do not affect the flavor. It is easy to underestimate or overestimate the contributions of individual volatile compounds to flavor (Reineccius, 2004) even given ideal sensitivity. Though challenging, incorporating the results obtained from instrumental analysis into sensory evaluations should eventually become possible as we improve our understanding of human physiology and statistics (Chambers & Koppel, 2013). Moreover, the concept of flavoromics

(Reineccius, 2008) suggests that we should focus not only on compounds already known to influence the flavor, but also the unknown low molecular weight compounds that are also present. Adopting an unbiased chemical stimuli approach opens up new avenues for researchers seeking to identify new flavor compounds and better flavor prediction for food products (Charve, Chen, Hegeman, & Reineccius, 2011).

Flavor Isolation

Since volatile compounds are usually present at very low levels (down to ppm and even ppb levels), the extraction process utilized is critical for the successful isolation of representative profiles of the flavor compounds from among the many other

highly complex food components present such as carbohydrates, lipids, proteins and vitamins. The food matrix makes the isolation of flavor compounds even more challenging as multiple ingredients such as thermally labile constituents (e.g. sugars, proteins, lipids and vitamins), emulsifiers (e.g. lecithins and proteins), and water are all involved. The matrix has a number of different characteristics that can interact with the flavor compounds, for example by altering the solubility and volatility of the flavor compounds. It seems no single method is capable of presenting a comprehensive picture of the volatile composition in food products. Different methods applied in similar studies (Ishida, Baldwin, Buttery, Chui, & Ling, 1993; Tandon, Baldwin, Scott, &

Shewfelt, 2003; Tieman, et al., 2012) have failed to agree on the volatile composition and key compounds of tomato flavor. Differences in extraction methods, cultivar, and handling could be responsible for this variation, but even when using the same batch of tomato samples, any differences in the isolation technique/materials/condition used will affect the tomato volatile spectrum produced and influence the results. Since the potential aroma compounds do not have unique physical or chemical characteristics, multiple approaches may be needed to assemble the whole picture for a representative sample.

Solvent extraction is one of the classical methods used to identify differences in the solubility of volatile compounds and separate important and unimportant constituents (mainly water and water-soluble macromolecules). Concerns related to this technique include:

1. the methodology is usually time consuming and the solvent added to the food may affect the results;

2. foods with a high lipid content may have an adverse effect on the final extract if not properly removed; and

3. the level of impurities may adversely affect further separation steps, and identification using GC.

A recent study used dichloromethane extracted aroma compounds in cherry tomatoes (Selli, Kelebek, Ayseli, & Tokbas, 2014) and the resulting aromatic extracts were further evaluated for similarity and intensity compared to a reference sample.

Solvent extraction is generally a good way to extract non-volatiles and semi-volatiles, but often excludes the most highly volatile components. In an attempt to address this shortcoming, high efficiency supercritical CO2 was applied to extract the non-volatile lycopene in tomato juice in another study (Egydio, Moraes, & Rosa, 2010), although the high cost and pilot size scalability of this supercritical CO2 extraction prevents its utilization at the industrial level.

Headspace (either static or dynamic), distillation and direct injection methods were most commonly used to extract volatile compounds before GC separation techniques became available. The distillation approach uses boiling point differences to separate components. However, the heat applied to the volatile compounds may cause detrimental effects such as decomposition and degradation. High vacuum distillation equipment can overcome these issues but at a relatively high cost; it also suffers from an inability to eliminate water vapor as water and volatiles are co-distilled. Although the simultaneous distillation extraction method has been used recently on tomatoes (cultivar

Momotaro), it appears to be less efficient when used on low and high boiling point volatiles (Maneerat, Hayata, Kozuka, Sakamoto, & Osajima, 2002). A combination of distillation, simultaneous distillation extraction (SDE) and solid phase extraction was used for a study of changes in the volatile composition of tomatoes during storage. In the common tomato, hexanal, (E)-2-hexanel, 6-Methyl-5-hepten-2-one, (E)-2-octanel

and (E,E)-2,4-hexadienal were proposed as biomarkers for quality control (Mathieu, Cin,

Fei, Li, Bliss, Taylor, et al., 2009; Zhang, Zeng, & Li, 2008).

Headspace analysis, which is a GC method, generally collects vapor from a point approximately at the same distance and content at which the human nose could or would smell it. It is non-destructive (Rouseff & Cadwallader, 2001) and provides a relatively clean method compared to solvent extraction and distillation because little or no residue remains in the GC. The static headspace method can be very simple, usually drawing the gas phase into a sampler at concentrations close to what the human nose can smell. However, only the most abundant volatiles are usually collected by simple static headspace methods. Oxygen and water vapor are also introduced into the GC by headspace methods, which slowly degrades the GC column.

Solid phase micro extraction (SPME) uses a coated polymer fiber as the stationary phase suspended over the headspace of the sample to extract volatiles and then desorbs those volatiles with heat into a GC (Gorecki & Pawliszyn, 1997). It is simple, portable, and requires minimal training for field use and has been widely used for flavor analysis since 1997. Compared to other headspace techniques, SPME greatly reduces the amount of collected water vapor and increases the collection of the less volatile compounds. However, the ability of the volatiles to bind to the coating materials is not consistent and this method may therefore neglect compounds that are important to human perception. Also, the SPME fibers are expensive and degrade quickly, requiring frequent replacement. Thousands of studies involving static SPME have been reported, but these drawbacks mean that dynamic SPME is now gaining popularity

(Alonso, Vazquez-Araujo, Garcia-Martinez, Ruiz, & Carbonell-Barrachina, 2009).

Compared to the static approach, dynamic SPME collects more volatile and semi- volatile compounds compared with static SPME approaches (Munoz-Gonzalez et al.,

2014).

Dynamic headspace approaches utilize a “purge and trap” process for extracting compounds where the purged gas passes through the sample and boosts the volatility of the odor compounds compared to static headspace methods. A trapping unit is used to collect the volatiles and then desorbs them at a later time for analysis. Cold, solvent or solid adsorbent traps have been used for flavor molecule collection. A cold trap is usually less selective, but has the ability to trap the water vapor that would otherwise interfere with the GC analysis. The Tenax® trap is used extensively but has a relatively low adsorption capacity because of its low surface area, while an active carbon trap can absorb most aroma constituents but also produces artefacts upon desorption. Although a Super Q hydrocarbon trap (Tieman et al., 2006) and 2,6-diphenyl-p-phenylene oxide based traps such as the Tenax trap (Buttery, Teranishi, & Ling, 1987) have been used for tomato analyses, these methods involve solvent desorption and are thus subject to the same artefact and bias problems as solvent extraction. It has been argued that a

Tenax trap with thermal desorption is a relatively clean method, with fast sampling and high sensitivity (Sucan & Russell, 2001). Operation conditions can have an influence but a good quality isolate can be achieved by optimization of conditions (time, temperature, flow rate etc). Table 2-3 describes the advantages and disadvantages of the aroma isolation methods commonly used for flavor analysis.

Flavor Identification

Chromatography is used to separate and identify molecules using either gas or liquid as the mobile phase. The separation process also involves various types of

stationary phase columns based on a number of different adsorption mechanisms, including partition, ion-exchange, size exclusion or affinity, to ensure compounds are separated and eluted at different retention times (Taylor & Linforth, 2009). Liquid chromatography (LC), which is used for non-volatile analytes, has a large sample capacity but LC columns have a lower separation efficiency than gas chromatography

(GC), which has become one of the major methods for flavor separation and identification due to its high resolution, separation efficiency, and availability. Although high in resolution, the primary disadvantage of the capillary columns that are currently available is their limited sample capacity, which may cause problems for fraction collection or gas chromatography-olfactory analyses (human GC detection). GC columns with different polarities have been used for previous studies on tomatoes

(Alonso, Vazquez-Araujo, Garcia-Martinez, Ruiz, & Carbonell-Barrachina, 2009;

Cosmai, Summo, Caponio, Paradiso, & Gomes, 2013; Ishida, 1993; Petro‐Turza, 1986;

Tandon, Baldwin, 2003; Tieman, et al., 2006).

The mass spectrometer (MS) is the detector that is most often used for tomato flavor research (Beltran, Serrano, Lopez, Peruga, Valcarcel, & Rosello, 2006; Dittmann,

Zimmermann, Engelen, Jany, & Nitz, 2000).The mass spectrometer ionizes molecules to enable them to be controlled by a magnetic or electromagnetic field to filter out the ions of interest. Structural information for these complex molecules is gathered via fragmentation, which provides information on the structural orientation of the atoms within the molecule. Nowadays, a compound that has been separated can often be matched to a sample in the MS library databank, which gives a score for the probability of the match and makes the process of identification viable without the need for a great

deal of training, making it possible to perform qualitative and quantitative identifications simultaneously. Time-of-flight mass spectrometry (TOFMS) provides a relatively high efficiency and short time detection method compared to conventional quadrupole MS

(Song, Fan, & Beaudry, 1998).

In contrast, flame ionization detectors (FID) detect hydrocarbons such as volatiles by monitoring the current produced by charged species as they are destroyed, although it does not provide structural information directly. FID detectors are less sensitive to inorganic molecules such as sulfur compounds, which does limit their utility for detecting some of the volatiles in tomatoes (Buttery, Teranishi, Flath, & Ling, 1989).

Pulsed flame photometric detector (PFPD) is more sensitive to sulfur-containing compounds, however, and two of the sulfur volatiles that contribute to tomato flavor were identified using PFPD (Du, Song, Baldwin, & Rouseff, 2015).

More than 400 of the volatiles in tomatoes have been detected and identified using the instruments described above. However, only a few of these are well perceived by humans, which have very different odor threshold. For example, we can perceive the unpleasant rotten egg flavor of sulfur-containing compounds at concentrations below the ppm level, far lower than most instruments can detect. At the same time, volatiles at relatively high levels may still fall below the threshold of human capacity and will not stimulate any perception. Most research using the instruments described above can assist humans directly or indirectly at some point to determine the odor potential of a volatile. GC-olfactometry (GC-O) uses humans as detectors to assess the volatile compounds eluted from GC, providing a “sniff port” for human analysts to detect odor- active compounds (Delahunty, Eyres, & Dufour, 2006). The specific volatile compounds

that contribute to human perception can be identified when these odorants are eluted from the GC with the aid of a humidified carrier gas. An FID or MS detector is also often used simultaneously for comparison. The chromatogram and aromagram (GC-O output) often overlap to some extent but are seldom identical due to the odor threshold issue.

This information does, however, draw more attention to the odor active compounds and a great deal of information can be provided by human assessors, depending on the methodology used. The frequency method needs the least training time and only requires 6-12 panelists to record the presence and duration of an odor that is perceived.

An aromagram can collect panelists’ perception information and identify those compounds perceived by most panelists. It should be noted that this method may not reflect the effect of concentration if a compound is present at levels above the odor hreshold of all participants; the aromagram will simply record the same concentration above this range (Linssen, Janssens, Roozen, & Posthumus, 1993).

Dilution analyses are sometimes used in studies of tomato flavor to pinpoint the aroma active compounds (Mayer, Takeoka, Buttery, Nam, Naim, Bezman, et al., 2003;

Mayer, Takeoka, Buttery, Whitehand, Naim, & Rabinowitch, 2008; Selli, Kelebek, Ayseli,

& Tokbas, 2014). Aroma Extract Dilution Analysis (AEDA) (Ullrich & Grosch, 1987) or

Charm analysis (Acree, Barnard, & Cunningham, 1984) are used for the quantification of the potent aroma active compounds. These methods all use the ratio of volatile concentration to odor threshold concentration to estimate the intensity of an odor.

However, it is important to bear in mind that the odorant threshold may be different in the food matrix compared to the solution/air. This is because of solubility differences in different solvents (food components) and the various synergy or suppression effects

between odor component perceptions. Thus, the intensity of the odor might not necessarily be reflected by the odor threshold. The intensity of sensation described by panelists using the GC-O does, however, show a strong correlation to that of sensory analyses (van Ruth & O'Connor, 2001). The dynamic time-intensity GC-O methods that have been developed use small groups of panelists to record odor intensity and duration. This intensity measurement can be expressed as a 16 point scale via a pressure-resistant button or cross modality matching based on the assessor's finger span (Etievant, Callement, Langlois, Issanchou, & Coquibus, 1999). and require intensive training for the panelists. Similar to descriptive analysis, the time-intensity method requires sensitivity and reproducibility of the panelists. The odorants that elute from the GC may be very similar but not necessarily the same if they are for a food sample. A further comparison of descriptors generated from the GC-O and a sensory panel may also be needed.

Bound Volatiles

Volatiles in tomatoes exist in different forms. Free volatiles can be directly extracted and identified, while aroma precursors are often bound volatiles released into free form at a later stage (Table 2-4). Bound volatiles are present in plant tissues and can potentially release free volatiles by enzymatic or chemical cleavage. These conditions include, but are not limited to, plant maturation, industrial pretreatments, or processing at low pH, all of which involve tissue disruption (Cai, Liu, Ling, & Su, 2002).

A number of researchers have explored the chemical structures and possible mechanisms of bound volatiles. Glycosidic derivatives of O-β-D-glucosides, O- diglycosides or diglycosides are the bound volatile forms most often reported. The glucopyranosyl unit is bound to an aglycone via a β-glycosidic linkage. Aglycones are

volatile-related molecules such as: terpenes (monoterpenes, sesquiterpenes, diterpenes, triterpenes), C-13-Norisoprenoids (megastigmane and non-megastigmane), benzene derivatives (Figure 2-2) and aliphatic alcohols. In an acid environment, the glycosidic bond may not exist since some glycones are cleaved under these conditions.

Carotenes may spontaneously convert to β-ionone and bind to O-β-D-glucosides (Stahl-

Biskup, Intert, Holthuijzen, Stengele, & Schulz, 1993). Another C-13 norisoprenoid, β-

Damascenone, is an important aroma active component in tomato flavor. The biochemical pathway explaining the presence of β-Damascenone has previously been elucidated in grapes (Williams, Sefton, & Wilson, 1989). Since most plants are high in carbohydrates, not only the fruits, but also the leaves, seeds and roots can be sources of glycosidic volatiles. The monoterpenoid aglycones most commonly found in fruits are geraniol, nerol, linalool, and α-terpineol.

The release of bound volatiles involves hydrolysis, usually with acids and/or enzymes. The β-glycosidase enzyme has been used to hydrolyze both monoglycosidic and diglycosidic (exoglycosidase) precursors, while Buttery et al. (1990) used different acid pH conditions to free bound tomato volatiles. 3-methylbutyric acid, 2-methybutryric acid, β-Damascenone, phenylacetaldehyde, 2-phenylethanol, linalool, α-terpineol, linalool oxides, 4-vinylguaiacol, 4-vinylphenol, hexanal and (E)-2-hexenal were the most common volatiles at pH 4.0, which is close to the tomato’s pH. Traceable amounts of eugenol, benzoic acid, geranial and (E)-6-methyl-3,5-heptadien-2-one have also been reported and (E)-4-(2,3’,6’-trimethylphenyl)-3-buten-2-one was identified at the extreme pH of 3.0. However, volatiles found in abundance in tomatoes such as hexanal, (E)-2- hexenal, 6-methyl-5-hepten-2-one are actually likely to be “fake” bound volatiles under

acidic conditions because they form acetals with sugar in blended tomato during hydrolysis at extremely acidic pH. Also, although cyclic norisoprenoids are predominately found in hydrolyzed conditions, few open-chain norisoprenoids (6-methyl-

5-hepten-2-one, geranylacetone and pseudo-ionone) are found under non-hydrolysis conditions. This suggests that compared to enzyme systems, acidic conditions may help generate more compounds and at higher intensities.

Enzymes for releasing bound volatiles are more complicated due to the specificity of the enzyme sources involved. β-glucosidase and pectinase have been reported as a way of freeing bound volatiles, yielding similar amounts of benzaldehyde, phenylacetaldehyde, 2-phenylethanol, 3-methylbutanol and (E)-2-hexenal, while floral/fruity note related terpenols such as geraniol, β-citronellol, α-terpineol and trans-

/cis-linalool oxidates were identified only in bound glycosides by a yeast-generated β- glucosidase (Ortiz-Serrano & Gil, 2007). Overall, 40% of the aglycons detected are not found in their free form. Compared to more abundant volatiles, linalool, benzyl alcohol, trans-2-hexenal, methyl salicylate, nerol, 2-phenylethanol and eugenol are found at a ratio 1:28 higher in the bound form. Cultivars that are low in free volatiles have been observed to have high amounts after hydrolysis. Winery and fruit juice producers have successfully used glycosidases for aroma improvement (Francis, Sefton, & Williams,

1992). However, it is not clear whether processing effects (mainly heat) actually release free volatiles or not. Thus, it is worth exploring how heat may affect the bound volatiles in future studies involving thermally processed tomato products.

Consumer Acceptance of Food Products

Consumer Perceptions of Tomatoes

The preferences of 179 Italian consumers (residents of the metro areas of Milan and Naples) regarding 16 cultivars of fresh French and Dutch tomatoes were studied

(Sinesio, et al., 2010). Eighteen sensory attributes, including flavor, appearance, odor, aftertaste, overall liking and texture characteristics, were rated using a 9-point scale (1- dislike extremely, 9-like extremely). Three-way ANOVA and 2-way interactions, generalized Procrustes analysis, partial linear regression, principal component analysis

(PCA), internal preference mapping, and hierarchical cluster analysis were then used to analyze the data collected. Texture and flavor were identified as the most important drivers of consumer preferences. Four heterogeneous consumer groups were identified by the between groups analysis, which revealed little agreement on the preferences of these cultivars. These differences of opinion may depend on the usage habits, demographic/behavioral characteristics and purchase patterns of the consumers.

However, these factors were not explored in this research.

Knowledge, consumption, and willingness to pay for organic tomatoes were examined in a study conducted among Spanish consumers (Mesías Díaz, Martínez-

Carrasco Pleite, Miguel Martínez Paz, & Gaspar García, 2012). Personal interviews with

361 regular tomato consumers, all of whom were the primary shoppers in their household, in the Murcia and Extremadura regions were conducted using random stratified sampling weighted in proportion to the population of each town. A contingent valuation was used for willingness to pay for organic and non-organic tomatoes. Cluster analysis, multivariate logit model, and the Hanemann model were used to estimate the average and maximum willingness to pay among different clusters of consumers. The

results showed a mean maximum willingness to pay a premium of 0.81 euro/kg for the organic product, which is 45.26% higher than the price of regular tomatoes. However, there was no significant link for informed consumers with regular access to organic food.

The willingness of consumers to pay for U.S. grown, non-label, or non-U.S. grown (Mexican, Dutch, and Canadian) fresh tomatoes and apples were studied among

335 regular consumers in Gainesville (FL), Atlanta (GA), and Lansing (MI). On a Vickrey auction (fifth-bid scaled price) method and double hurdle probit model, the consumers were willing to pay a premium of $0.49/lb for apples and $0.48/lb for tomatoes labeled

“U.S.A. grown.” However, the Lansing consumers were the least willing to pay for both apples and tomatoes grown in the U.S., the participants from Gainesville were willing to pay the highest price for fresh tomatoes produced in the U.S., and the consumers from

Atlanta were willing to pay the highest average price premium for fresh apples (Mabiso,

Sterns, House, & Wysocki, 2005).

Though a few previous studies have investigated the flavor and textural aspects of fresh products, little attention has been devoted to consumer perceptions of minimally processed tomato products. Goodman, Fawcett and Barringer (2002) studied consumer perceptions of the flavor quality of tomato juice, comparing hot break and cold break treatments. The break stage is an essential processing step for yielding good flavor. A trained panel of 153 consumers on a Columbus, OH campus evaluated the color, sweetness, sourness, ripeness, paint, tomato character, and tomato flavor intensity in a descriptive analysis using the flavor profile scale on a Columbus, Ohio campus, after tasting the samples. The freshness and overall enjoyment of each of the juice samples

were rated using a six-inch lines scale. Simple Statistical analysis: t-test mean comparisons and post hoc tests (Fisher LSD) were used for the data analysis. The 15 min cold break juice was judged to yield superior quality and was significantly preferred by the consumers. However, this study was designed from a product development perspective and there was no indication of consumer willingness to pay for a processed product.

Processed Product Acceptability

Novel commercial non-thermal processing improves the safety and quality of food, although this may not be well communicated to the public. A study of 1204 adults participating in an online nation-wide survey in the U.S. (Hicks, Pivarnik, McDermott,

Richard, Hoover, & Kniel, 2009) assessed their knowledge, attitude, and willingness to pay for high pressure processed (HPP) ready-to-eat salads containing tomatoes as one of the ingredients. A contingent valuation approach with product reference price range was used to estimate the maximum willingness to pay. The analysis of variance

(ANOVA), chi-square statistics, and Cronbach’s alpha measure of internal consistency were used for data analysis. Only 8% of the consumers had previously heard of the

HPP technology and 15% remained unwilling to pay a premium price for products processed using HPP even after the benefits of the technology had been explained.

Processed blueberry base product (pure blueberry jam, blueberry-lime jam, blueberry yogurt, blueberry fruit rollups, blueberry dry muffin mix and blueberry raisinettes) were used to explore the preferences of Kentucky consumers regarding organic features, Kentucky-grown claims, sugar free claims and willingness to pay (Hu,

Woods, & Bastin, 2009). These regular consumers were interrupted during grocery shopping. A D-optimal fractional random design involving pairs of processed products in

4 choice sets were provided for the choice experiment. According to the mixed logit specification model analysis, the Kentucky-grown label received the most support. Low to moderate-income younger and middle-aged consumers valued the Kentucky-grown products much higher than the organic equivalent for pure blueberry jam. Sugar-free claims may be positively or natively linked with food. Some consumers may be attracted by the organic claim, but this had no impact on their preference for blueberry fruit rollups and dry muffin mix.

A payment card contingent valuation method with product reference price ranges and choice options was used to estimate the maximum willingness to pay for value- added processed blueberry products (blueberry herbal tea, blueberry basil vinegar and blueberry syrup) (Hu, Woods, Bastin, Cox, & You, 2011). A log likelihood function was used for the data analysis. Similar to the previous study on blueberry products, 604 representative regular Kentucky consumers were interrupted while grocery shopping.

The study found consumers would be willing to pay (WTP) positive amounts for these commodities. However, heterogeneous preferences on different products were found.

Consumers’ socio-economic characteristics and information on health benefit claims were important determinants for WTP on some products.

These research studies explored the willingness of consumers to pay for processed products extensively. However, there has been little or no research on sensory attributes, which are important factors for the consumption of food.

Exploring Palatability and Willingness to Pay for Fruit Products

In order to examine the role of sensory attributes in consumer decisions to purchase fruit, a recent research program on citrus (mandarin oranges) used a probit model (House, Gao, Spreen, Gmitter, Valim, Plotto, et al., 2011) to examine the

relationship between appearance (color, taste, flavor, sweetness, acidity, juiciness, ease of peeling, size, shape, and amount of seeds) and purchase patterns. Equal numbers of consumers from three cities (Tampa, FL, Chicago, IL and Baltimore, MD) were interrupted at the shopping mall for a taste test and a 5-point likelihood test to purchase/eat. The results from this study successfully identified sweetness, shape, acidity, and flavor as the most important factors leading adults and children to increase their willingness to purchase or to eat a mandarin orange, while such factors as number of seeds, size of fruit, color, and overall appearance were less important.

The relationship between sensory attributes and the consumer’s willingness to pay for the preservation method (ethylene treatments) were examined for Anjou pears

(Zhang, Gallardo, McCluskey, & Kupferman, 2010). In this research, a combination of sensory tests and online surveys were used. Five thousand consumers representative of the U.S. census population participated in an online survey for Anjou pear consumption, demographics, and received an invitation to participate in further sensory tests in Portland, OR. The sensory test with 120 participants subsequently conducted in the Portland metropolitan area accessed the overall desirability, flavor, sweetness, juiciness, firmness and texture of the pears on a 9-point hedonic scale. A double blind, dichotomous choice contingent valuation model was employed to estimate the willingness to pay for Anjou pears treated with ethylene for different lengths of time.

Consumers were willing to pay a premium of $0.25/lb for pears that received six-day treatments compared to non-treated market products. Three sensory variables

(firmness, sweetness and juiciness) were significant factors related to willingness to

pay. Among these participants, consumers with at least one child under the age of 18 expressed a higher willingness to pay for this product.

Table 2-1. Different stages of tomato ripening.

Stages Description Picture

Green The surface of the tomato is completely green in color. The shade of green may vary from light to dark.

Breakers There is a definite “break” in color from green to tannish yellow, pink or red on no more than 10% of the tomato surface.

Turning 10%-30% of the tomato surface, in the aggregate, shows a definite change in color from green to tannish-yellow, pink, red or a combination of thereof.

Pink 30%-60% of the tomato surface, in aggregate, shows pink or red in color.

Light red 60%-90% of the tomato surface shows pinkish-red or red color.

Red More than 90% of the surface of tomato is red.

(California Tomato Commission, 2014)

Table 2-2. Summary of various microorganisms, cases of tomato-related outbreaks for fresh tomatoes. Microorganism Cases Year Norovirus NR 1992, 1996, 2010 Hepatitis A 92 1994 Giardia.lamblia 21 1989 Campylobacter NR 1996 Listeria monocytogenes 20 1979 Escherichia coli O157 NR 1995 Bacillus cereus NR 2003 Salmonella Javiana 176, 429 1990, 2004 Montevideo 100 1993 Newport 7, 52, 65, 55 2002, 2005, 2007, 2010

Braenderup 137, 84 2004, 2005 Typhimurium 429, 8 2004, 2006 Anatum 429 2004 Muenchen 429 2004 Thompson 429 2004 Berta 16 2006 Saintpaul 21 2009 NR: not reported, but case≥1 for spore botulism, case≥2 for other microorganism in the short period of time can be considered an outbreak (FDA, 2015). (Bennett, Littrell, Hill, Mahovic, & Behravesh, 2015; CDC, 2014; Sandoval, Barreiro, & Mendoza, 1992; Schneider, Schneider, Hubbard, & Richardson, 2014; Valadez, Schneider, & Danyluk, 2013)

Table 2-3. Comparison of different isolation methods. Methods Sample Destructive to Sensitivity Quick convenience Training Cost collecting sample Solvent Indirect Yes Depends on No No A lot Low extraction solubility

Super- Indirect Yes High, but high No No A lot High critical CO2 solvent costs

Distillation Indirect Yes Medium to high No No A lot Low/High

Static Direct No Abundant Yes Yes Little Low/High headspace volatiles

Dynamic Direct No Abundant Yes Yes/No Little Medium headspace volatiles

SPME Direct No Depends on Yes Yes No High coating materials

Table 2-4. Common molecules occurring as glycosidically bound volatiles in tomatoes. Compound Compound Compound Aliphatic alcohols Monoterpenes Monoterpenes Butan-1-ol Borneol Neo-isomenthol 2-methylbutan-1-ol Carvacrol Myrtenol Pentan-1-ol cis-Carveol Neral Pentan-2-ol (-)-cis-Chrysanthenol Nerol Pent-1-en-3-ol 1,8-Cineole,2-hydroxy Nerol oxide Cis-Pent-2-en-1-ol Citronellol Z(1S,5R)-β-Pinen-10- ol Hexan-1-ol Dihydrocarveol Rose oxide cis-Hex-3-en-1-ol Isodihydrocarveol cis-Sabinene hydrate Heptan-2-ol Neo-isodihydrocarveol trans-Sabinene hydrate 6-Methylhept-5-en-2- 2,6-Dimethyllocta-3,7-diene-1,6- Sabinol ol diol Octan-2-ol 2,6-Dimethyllocta-2,6-diene-1,8- α-Terpineol diol Octan-3-ol Geranial Terpinen-4-ol Oct-1-en-3-ol cis-Geranic acid Thujol Oct-2-en-1-ol trans-Geranic acid Isothujol Nonanol Geranial Neo-isothujol Geraniol Thymol Geranylacetone Thymol, 2-hydroxy- Lavandulol Verbenol Sesquiterpenes Linalool Aromadendrol Linalool,1-hydroxy- α-Cadinol Linalool,9-hydroxy- Phenylpropane derivatives and related compounds T-Cadinol Linalool oxide(fur.) Benzaldehyde Elemol Linalool oxide(pyr.) Benzyl alcohol γ-Eudesmol p-Mentha-1,5-dien-7-ol trans-Coniferyl alcohol cis-Nerolidol p-Mentha-1,5-dien-8-ol Eugenol Spathulenol p-Mentha-2,8-dien-1-ol Isoeugenol α-Bisabolol p-Mentha-1,8-dien-7-ol Hydroquinone 10-epi-Cubenol Menthol Methyl salicylate Hydroxygermacrene Isomenthol 2-Methyoxy-4- methylphenol Viridiflorol Neomenthol 2-Methoxy-4- vinylphenol (Krammer, Buttery, &Takeoka, 1995)

Figure 2-1. Relationship between flavor and instrumental analysis.

Figure 2-2. General structure of bound volatiles.

(Krammer, Buttery, & Takeoka, 1995)

CHAPTER 3 SENSORY AND CHEMICAL CHARACTERISTICS OF TOMATO JUICE

Abstract

Despite its high nutrient content, tomato juice is among the least liked fruit/vegetable juices due to its generally poor flavor. To date, Florida tomatoes are primarily produced for the fresh market and have not been widely used in processed tomato products such as sauces, pastes and juices. The objective of this study was therefore to characterize the flavor of a premium Florida tomato variety that has significant potential for producing a high quality processed juice product.

A high-quality Florida plum tomato variety (Garden Gem), and a typical grocery store plum tomato variety (Roma) were thermally processed into tomato juices without adding salt, sugar or flavors. The two pilot products (Garden Gem and Roma) and a popular commercially available tomato juice (low sodium with sugar and flavor added) were compared using sensory evaluation and flavor analysis. Flavor compounds in these products were identified using dynamic headspace purge and trap-gas chromatography-mass spectrometry (PT-GC-MS) by MS library match and retention index and were semi-quantitated using internal standards. Analysis of variance, mean comparison and principal component analysis were used to analyze both sensory and flavor data.

Among the three products, Garden Gem juice was rated significantly (p<0.05) higher for overall liking, tomato flavor, and sweetness by the 119 panelists in both seasons. Uniformity and color were most linked with the commercial product. Garden

Gem juice was found to contain significantly (p<0.05) higher levels of 3 sweet/fruity related aroma compounds: 6-methyl-5-hepten-2-one, linalool and β-ionone. The

commercial tomato juice contained a high level of the Maillard reaction-related notes furfural, dimethyl sulfide and the least amount of green-related notes (hexanal, E-2- hexenal and Z-2-heptenal). Despite lower levels of fruity and green-related notes, the commercial tomato juice contained ethyl acetate, which was not found in the tomato juices without additives. The flavor profile of the Roma tomato juice was close to that of

Garden Gem juice except it contained substantially lower amounts of hexanal and 2- isobutylthiazole. No fruity or sweet-related β-ionone were detected in either the commercial or Roma juice. This research reports the detailed flavor profile of a processed Garden Gem juice; and how its flavor profile demonstrates a potential market advantage over current commercial products.

Background Information and Objectives

Tomatoes are one of the largest agriculture commodity in the world. The United

States is one of the world-leading producers of tomatoes. Florida produces 43% of the tomatoes grown in the United States, all of which are sold as fresh market (USDA,

2010).

Processed tomato consumption has increased markedly over the past 50 years and tomatoes are considered good tasting and a source of vitamins and minerals, including Vitamin C, folic acid, potassium, and thiamin; a single 8 oz. glass of tomato juice at breakfast meets the daily requirement of Vitamin C. In particular, lycopene and the other carotenoids found primarily in tomatoes are antioxidants and have been shown to lower the risk of both cardiovascular disease (Canene-Adams et al.,

Campbell, Zaripheh, Jeffery, & Erdman, 2005) and prostate, lung, stomach, pancreas, colon, rectum, esophagus, breast and cervical cancers (Giovannucci, 1999). However,

tomatoes commonly used for processed tomato products are generally not high in flavor quality.

Flavor is one of the major influences on consumer purchase and consumption.

Flavor is comprised of aroma, taste, and somatic sensation. Taste is perceived in the mouth by the sensations of sweet, sour, salty, bitter, and umami. Aroma is sensed in the nasal cavity and encompasses most of what is considered flavor. Most of these aroma compounds are volatile or semi-volatile compounds perceived by the orthonasal and retronasal olfactory systems, where orthonasal refers to the aromas taken directly through the nose and retronasal refers to the aroma volatiles produced when food is chewed and air is exhaled through the nasal cavity. The aroma volatile partition is affected by saliva interaction and mastication. The retronasal impact is most likely to leave a lasting impression on the consumer.

Good quality raw materials are essential for producing good quality tomato products. The overall objective of the study reported in this chapter was to develop a processed tomato product with high flavor quality using Florida tomatoes. First the most common methods for measuring tomato flavor, including both formal and informal sensory studies, soluble solids (SS) measurements, pH, titratable acidity (TA), and the

SS/TA ratio were used. Various instrumental measurement of sugars, acids ,and flavor volatiles were then compared to data obtained from both experienced and trained sensory panels for 3 thermally processed tomato juice samples (Garden Gem, Roma and commercial juice) made from tomatoes harvested in two different seasons, a total of 6 different treatments. The purpose of this comparison was to identify which

instrumental measurements best reflect sensory observations for different aspects of tomato flavor.

Materials and Methods

Tomatoes

Fully ripe Garden Gem tomatoes were harvested in Summer 2015 from a field

(Live Oak, FL) and in Fall 2015 from a greenhouse (Gainesville, FL). Ripe Roma tomatoes were purchased from Publix (Gainesville, FL) and Winn-Dixie (Gainesville, FL) grocery stores in both the Summer and Fall of 2015.

Juice Processing

Twenty pounds of tomatoes from each cultivar were each cut into three pieces by toss and chop kitchen scissors (Comfify, Fernadale, WA, USA). The small tomato pieces were then fed into a tomato electric milling machine (Model 2400, O.M.R.A.

MTD, Brooklyn, NY, USA) to remove the skin and seeds in 15 seconds. The resulting puree was then continuously fed into a steam kettle and heated (a process called “hot break”) at 95ºC for 3 min, after which the hot break juice was poured into a sanitized carboy and stored at 4ºC. Later, the juice was pasteurized (Model 5024-F 25 HV,

Microthermics, Raleigh, NC, USA) at 73 ºC for 30 s and then stored in another newly sanitized carboy at 4ºC. The whole process was repeated three times for each variety.

The commercial source, Campbell’s Low Sodium Tomato Juice (Campbell Soup

Company, Camden, NJ, USA), was purchased from a local Publix supermarket for comparison with the Garden Gem and Roma juices.

Fresh Juice Production

As in the juice processing procedure, twenty pounds of tomatoes from each cultivar were each cut into three pieces by toss and chop kitchen scissors (Comfify,

Fernadale, WA, USA) and the resulting small pieces were fed into the tomato electric milling machine (Model 2400, O.M.R.A. MTD, Brooklyn, NY, USA) to remove the skin and seeds. The puree produced was mixed with food grade saturated calcium chloride

(Modernist Pantry, Portsmouth, NH, USA) at a 3:1 ratio and stored at -20ºC in 1-gallon sample bags.

Flavor Isolation

Tomato juice treated with or without heat was filtered by glass microfiber binder free paper GF/C™ (Whatman, Pittsburgh, PA, USA) using a Buchner vacuum filtration set to remove solids. The liquid was collected and placed in a 40 mL glass vessel, after which 3 µL of antifoam reagent (Trans-400, Bristol, WI, USA) was added and mixed well. Carvone (150 ppb) was added into each sample vial as the internal standard.

Flavor extraction was performed using a Stratum dynamic headspace unit (Teledyne

Tekmar, Mason, OH, USA). A total of 5 mL sample was pulled by autosampler and purged with a nitrogen flow at 200 mL/min for 30 min. The volatile compounds were adsorbed by a trap containing Tenax® and desorbed at 180°C, then sent by transfer line to the GC injecton port.

Identification and Quantification

Identification was conducted on an Agilent (Santa Clara, CA, USA) gas chromatograph equipped with quadrupole mass spectrometer detector (5975C MSD). A

ZB-WAXplus column (30 m x 250 µm x 0.25 µm) and a ZB-5MSplus column (30 m x

250 µm x 0.25 µm) (Phenomenex, Torrance, CA, USA) were used for separation. The

GC oven temperature was held at 45°C for 7 min, and then ramped up to 150°C at

3°C/min. Continued ramping was applied at a rate of 10°C /min until 210°C while a

30°C/min rate was further applied to 240°C. The column was then held at 240°C for 4

min. Helium gas flow rate was 1mL/min. The MSD used scan mode to detect compounds in the range of 45-300 m/z. The peaks were identified by conducting a mass spectral comparison with the 2011 NIST mass spectral library and a retention index comparison. The retention index was calculated using a C5-C20 mixture of linear organic chemical standards.

Sensory Analysis

Recruiting for both the summer and fall sessions for the sensory testing was done through e-mails and announcements posted around the Food Science and Human

Nutrition Department at the University of Florida (UF) Gainesville campus. Panelists were screened based on their familiarity with tomato juice, their frequency of consumption and their availability. Only those panelists who normally consumed tomato juice were selected to participate.

The sensory laboratory in the Food Science and Human Nutrition Department at

UF has ten individual booths with computer workstations. Panelists were seated, assigned a panelist number and asked to sign in to the sensory program

(CompuSense® Five Sensory Analysis Software for Windows, CompuSense, Guelph,

Canada) for that session. Each booth was furnished with a tray containing water, crackers, a napkin, and a refuse cup.

The tomato juice consumers (119 for the Summer season, 119 for the Fall season) were invited to complete a sensory analysis using a 9 point hedonic scale rating the color, texture (before taste), flavor, sweetness uniformity (with chunks – Roma and Garden Gem; without chunks – commercial; after taste) and overall acceptance

(Schwieterman, et al., 2014; Tieman, et al., 2012). Three thermally processed juices

(Garden Gem, Roma and commercial juice) were randomly coded with a three-digit

number and presented in a balanced experimental design. All samples used for tasting were served at room temperature. For each session (Summer and Fall seasons), panelists were asked to complete the same questionnaire.

Chemical Analysis

Chemical analysis for pH, soluble solids content, and titratable acidity (TA) were completed in triplicate according to the methods of Baldwin (1998). Soluble solids (Brix) and pH were measured using a refractometer and pH meter, respectively. To measure titratable acidity, 15g of juice was titrated to pH 8.0 using 0.1 N NaOH (Fisher Scientific,

Waltham, MA, USA). To calculate TA in g/100 mL citric acid, the following formula was used (Baldwin, 1998):

TA =[(mL base) (N base) *100]/sample weight (3-1) where N is the normality of the base.

Statistics

Sensory data collected from CompuSense® Five (Guelph, Ontario, Canada) was transferred into a Microsoft Excel® worksheet for use with Statistical Analysis

Systems™ 9.1 (SAS). The chemical data collected (Table 3-1) was subjected to a 1- way analysis of variance (ANOVA), and the sensory data was subjected to a 2-way

ANOVA. To identify any differences among means between the cultivars, Duncan’s multiple-range test for mean separation was conducted. Differences were considered significant at an alpha level of less than or equal to 0.05.

The mean scores of each cultivar for sensory attributes, chemical attributes, and flavor attributes were then compiled, and a Principal Component Analysis and

Pearson’s Linear Correlation Analysis was completed using SAS to observe the relationships between all variables.

Results and Discussion

Chemical Analysis of the Products

Tomato juice made from Garden Gem (GG), Roma (RA) and commercial tomatoes grown in two different seasons were assayed for titratable acidity (TA), pH, soluble solids (SS) and SS/TA. Significant differences were found in all chemical parameters between the seasons except for the SS/TA ratio (Table 3-1).

The titratable acidity for tomato juice ranged from 5.38 to 8.58g citric acid/100mL and a significant seasonal difference for each variety was observed. The % titratable acidity was found to range from 0.24-0.54, which is quite close to the typical amounts of

0.20-0.50 previously reported for tomato puree (Baldwin, 2004). Juice made from the summer season was significantly higher in titratable acidity except for the RA variety.

The RA summer season juice had the lowest TA value, while the commercial summer season and GG summer season juice had the highest value. Overall, the commercial variety showed significantly higher TA than the sample varieties (RA, GG). The GG variety was higher than the RA variety in both TA and SS, regardless of the season.

The commercial fall juice had the lowest pH (high acidity), while the RA fall juice had the highest pH (low acidity). One would expect TA and pH values to be related, but there were several instances when they did not match up for the treatments. This lack of agreement may be due to the slight differences in acid measurement by titratable acidity and pH: the pH focuses on the concentration of free hydrogen ions regardless of whether the weak acids contribute to taste, while the TA measures the amount of acid anions present, including citric acid, glutamic acid, malic acid and phosphate acid, etc.

(Anthon & Barrett, 2010). Thus, the TA was considered to be a more important indicator since it directly relates to sourness and ripeness. The lower TA values for Roma may be

explained by the fact that the Roma tomatoes purchased in the supermarket were likely not at optimum ripeness or had been picked green and ethylene-treated for ripeness. A change in the amount of citric acid in the tomatoes is common if the tomato has been ethylene treated (Boe & Salunkhe, 1967).

A soluble solids measurement is a common indicator for assessing the glucose and fructose concentrations. The soluble solids in tomato juice made from different varieties ranged from 5.53 to 6.63 in the Summer season and 5.27 to 5.90 in the Fall season. This seasonal difference may be explained by the better nutrients and sunlight conditions available in the Summer than in the Fall. Similar trends were observed by

Anthon and Barrett (2010), who found that later season vine tomatoes suffer a 4-7% decrease in soluble solids. The GG juice consistently showed the highest (p<0.05) soluble solids across seasons, noticeably higher than that of the commercial juice. RA also had a significantly higher SS than commercial juice in the Summer, but there was no significant difference for the Fall season. The year-round supply of RA and commercial variety tomatoes, which may be obtained from different regions, could lead to some inconsistency in the product.

The ratio between soluble solids and percent titratable acidity is used widely as it is considered a better sensory indicator than either soluble solids or acidity alone.

These findings have also been demonstrated in research with other fruits such as table grapes (Jayasena & Cameron, 2008). The values obtained ranged from 10.08 to 16.67 for all varieties across the seasons. However, there was no clear pattern among varieties or seasonal effects for the SS/%TA ratio.

Sensory Analysis Results

After tasting the samples, GG was rated significantly (p<0.05) higher in overall likability, sweetness, and tomato flavor than RA and commercial tomatoes across both seasons (Tables 3-2 and 3-3). The same trend in variety preference of the attributes was seen independent of season, but there were different rating orders before tasting and after tasting. The visual parameters (color and uniformity) were rated before tasting; the commercial variety was rated highest (p<0.05) in this aspect. The GG has a comparable color to the commercial juice, but was rated significantly lower in uniformity.

The RA rated significantly lower in both color and uniformity. This suggests that the GG has superior flavor quality but the uniformity of the sample varieties (RA and GG) needs to be improved, possibly by homogenizing the juice before pasteurization.

Overall likability values ranged from 4.93-6.08 and 4.60-5.91 for the summer and fall seasons, respectively. The rating scores were relatively low on a 9-point scale, indicating that tomato juice as a product has room to improve if it is to gain greater consumer acceptance. However, when questioned about tomato flavor, the participants rated the preferred GG with means of 6.29 (summer season) and 6.10 (fall season), which is actually quite high on the 9-point scale. Therefore, tomato flavor may play a vital role in tomato/tomato juice overall likability. When comparing the RA and commercial varieties, which had similar ratings for tomato flavor, the RA had a relatively higher sweetness. However, the inferior mouthfeel of RA may have contributed to its low overall likability. The low flavor rating was the major problem with the commercial product.

Correlations between Overall Liking and Sensory, Chemical and Flavor Compounds for Processed Juice

To identify which measurements best reflect consumer preferences for tomato juice, Pearson correlations and Principal Component (PCA) were performed. The

Pearson correlation helps to elucidate the relationship between two measurements, and

PCA assists in gathering relationships of all parameters into a single picture.

Looking at the sensory evaluation data for the two seasons separately (Tables 3-

4 and 3-5), all the sensory variables showed a significant positive correlation (the r values were all p < 0.05). For both seasons, tomato flavor and overall likability were observed to be very strongly positively correlated, with r values of 0.845 and 0.832.

These results are in agreement with a previous study that found flavor to be the most important factor in determining consumer acceptance (Moskowitz & Krieger, 1995).

Sweetness and mouthfeel were correlated moderately with overall likability. Appearance

(color and uniformity) were weakly correlated with most parameters. The similarity between the Pearson correlations for the independent seasons (Summer and Fall) demonstrate that variability between seasons and planting location (field vs. greenhouse) had only a minimal impact on the sensory analysis.

The important sensory, and chemical and flavor compounds for both seasons were combined (Table 3-6). Tomato flavor was strongly correlated (p<0.05) with hexanal, heptanal, 2-octenal, benzaldehyde, β-cyclocitral, and β-ionone. Interestingly, sweetness showed a strong correlation with hexanal, heptanal, 2-octenal, benzaldehyde and β-ionone. This similarity demonstrates the likelihood of a relationship between volatiles and sweetness perceptions. The odor-induced taste enhancement theory could support the hypothesis that flavor compounds in a mixture often give a higher

sweetness rating than a sweet solution alone (Prescott, Johnstone, & Francis, 2004).

Hexanal, heptanal, and 2-octenal, which are derived from tomato lipids, are thought to play a major role in giving tomato it’s green “top-note” (Sucan, 2002). β-cyclocitral and

β-ionone are degradation products from the carotenoid pathway, imparting a fruity/floral note in tomatoes (Auldridge, McCarty, & Klee, 2006). The top three compounds that correlated with overall liking were benzaldehyde (0.909), methyl salicylate (0.849), and

β-cyclocitral (0.827). Methyl salicylate is an important derivative of salicylic acid that contributes to tomato product flavor (Chung, 1983). However, as the data in Table 3-7 shows, not all these compounds were present in high enough concentrations to exceed the odor threshold in water (Buttery, 1990), although this odor threshold was passed for compounds that affect green and fruity notes (Baldwin, Goodner, Plotto, Pritchett, &

Einstein, 2004).

Chemical measurements (TA, Brix, pH) saw no significant correlation with overall likability, although the Brix measurement was correlated with tomato flavor and sweetness. Soluble solids (Brix) were positively correlated with sweetness (0.822) and tomato flavor (0.853). This suggests that quality control measurements that focus only on sugar and acid level may not be sufficient to identify products with better flavor quality.

The data were set up in a matrix containing values for the two sets of variables

(sensory, and flavor) for each cultivar tested. Principal component analysis (PCA) was run on the data matrix and the output is presented in Figures 3-1 and 3-2. Principal

Component 1 and Principal Component 2 (PC2) explained 81.14% of the variation in the collected observed measures for the Summer season juice, while PC1 and PC2

explained the 84.46% of the data variation in the biplot. For both seasons, the Garden

Gem juice was linked with green and fruity note compounds more often, whereas the commercial juice was linked closely with cooked note and few green note compounds.

However, the RA lacked fruity note and was thus less favored.

In general, the PCA load plot confirmed the results of the correlation analysis.

The variables found on one side of PC1 of the load plot correlated positively with those around them and negatively with those on the opposite side of the axis. It was observed that in this biplot, the angles between green noted compounds were the smallest. This indicates that these variables had the highest positive correlation.

The PCA biplot showed groupings of cultivars that were related. The replicates of each cultivar (GG, RA and commercial) in both seasons have very small angles between their biplot points as well as a distinct separation of spacing from the other cultivars, which means these replicates groups are highly similar. The PCA biplot confirms that the variables expected to be related to flavor actually were. The variables of green note and fruity note flavor are clearly related and clustered very closely in GG and commercial quadrant. PCA can display these attributes that are closely related as well as where the sample falls relative to the variables most responsible for their differences.

Volatile Metabolites Identified using Different Columns

The reason for using both the ZB5 and ZB-WAX columns was to determine which degree of column polarity best separated the volatile compounds in tomato juice.

The ZB-5, consisting of 5% phenol/ 95% dimethylpolysiloxane, is designed to separate non-polar volatiles and semi-volatile (acid/base, neutral) analytes. The ZB-WAX is the lowest bleed polyethylene glycol phase available and is widely used for profiling

samples composed of mixtures of polar compounds. Here, 59 and 85 volatile compounds were identified using the ZB5 and ZB WAX columns, respectively, for fresh

Garden Gem tomatoes harvested in Fall 2015. As more compounds were separated using the ZB-WAX column, this column was deemed superior for separating tomato juice volatiles. Most flavor compounds reported in tomato products are non-polar alcohols, aldehydes, and aromatics such as essential oils and glycols. As shown in

Table-3-7, acetone, glutaraldehyde, 1-penten-3-ol, methyl propyl sulfide, dimethyl sulfide, 2-heptenal, methional, benzeneacetaldehyde, 2,5-dimethyl-benzaldehyde, 6- methyl-5-hepten-2-ol, eugenol and furfural were only identified using the ZB-WAX column. Significantly higher levels of methyl salicylate were quantified using the ZB-5 than the ZB-WAX column, but overall, the ZB-WAX column was found to better separate a greater variety of volatile compounds from tomato juice samples.

Volatile Metabolites of Fresh Tomato Cultivars

A total of 71 and 90 volatile compounds were identified in fresh Roma and

Garden Gem samples (Table 3-8), respectively. More alcohols were identified in Garden

Gem. The most abundant compounds quantified in Roma and Garden Gem were generally similar: hexanal, 2-hexenal, 3-hexen-1-ol, 1-hexanol, 1-penten-3-one, 1- penten-3-ol, 6-methyl-5-hepen-2-one and 2-methyl-butanol. However, the flavor profiles for Roma and Garden Gem were quite different. Quantities of the following compounds: hexanal (2 fold), 2-hexenal (4 fold), 2-heptenal (1 fold), 1-penten-3-one (3 fold), 1- penten-3-ol (2 fold) and geranylacetone (2 fold) were significantly higher in Roma than in Garden Gem. However, Garden Gem did contain larger amounts of 6-methyl-5- hepten-2-one (4 fold), benzaldehyde (4 fold), 3-methyl-butanal (2 fold), β-citral (4 fold),

β-cyclocitral (2 fold), linalool (2 fold), acetone (2 fold), and 2-isobutylthiazole (15 fold

higher). β -damascenone was observed in the Garden Gem samples only. This compound is derived from cyclic carotenoid metabolism (Vogel et al., 2010). This compound is characterized by fruity aroma and sweet taste and has been linked with a rose like odor from GC-O descriptors (Du et al., 2015). β -damascenone and 6-methyl-

5-heptene-2-one are produced by oxidative cleavage of carotenoids, and mostly contribute to the tomato’s fruity/floral green note (Berna, Lammertyn, Saevels, Di

Natale, & Nicolaı̈, 2004). Several fruity note compounds have been reported to enhance sweet taste (Baldwin et al., 2004).

The abundant volatiles found in Roma were mainly due to lipid oxidation products of linoleic acid and linolenic acid, which are again green-note related compounds.

Interestingly, 1-penten-3-one has also been reported to enhance the green aroma perception and sweet taste and linalool is thought to enhance the overall, fruity and aftertaste in combination with sugars and acids in fresh tomatoes (Baldwin et al., 2004) .

Volatile Metabolites of Garden Gem, Roma and Commercial Juice

More flavor impact compounds were identified in Garden Gem juice than in either

Roma or commercial juice. Among the 36 reported flavor impact compounds

(concentrations found above the odor threshold), 91.6%, 44%, and 52.7% were identified in GG juice, RA, and commercial juice, respectively (Table 3-7).

Garden Gem juice was found to contain significantly (p<0.05) higher sweet fruity related aroma compounds: 6-methyl-5-hepten-2-one, linalool and β-ionone. The compound β-ionone has been reported to be derived from 9, 10 and/or 9′, 10’ bond cleavage of cyclic carotenoids (Lewinsohn, et al., 2005), while 6-methyl-5-hepten-2-one is known to be a product of lycopene oxidative cleavage (Vogel, Tan, McCarty, & Klee,

2008). Linalool is likely formed as a monoterpene product during maturity from geranyl

diphosphate, an indicator of ripening fruit (Schwab, Davidovich‐Rikanati, & Lewinsohn,

2008). It is known to contribute a sweet and floral note (Buttery et al., 1990; Buttery,

Seifert, Guadagni, & Ling, 1971). The presence of this compound may be an indication of the Garden Gem’s ripeness, while the Roma tomatoes used here were likely treated with ethylene gas for faster ripening. The commercial tomato juice lacked all these compounds, probably because of the extended heating process it had undergone. Most commercial tomato juice is produced by diluting a paste, and the process of making paste requires significant heating to remove excess water (Thakur, Singh, & Nelson,

1996).

The commercial tomato juice contained a high amount of the Maillard reaction- related note furfural (283 ppb), dimethyl sulfide and the least amount of green-related notes (hexanal, E-2-hexenal and Z-2-heptenal). Most of these volatiles are derived from linoleic and linolenic acids degradation through 13-lioxygenase (Feussner &

Wasternack, 2002). These C-6 and C-7 products are characterized by the green, grass, stem-like, and fresh fruity notes found in fresh tomato products. Despite less fruity and green-related notes, the commercial tomato juice contained ethyl acetate, which was not found in either of the tomato juices without additives. The ethyl acetate found in the commercial tomato juice may be from natural flavors generated from other fruit and vegetables added by the manufacturer to enhance the flavor (Guiné et al., 2010).

The flavor profile of the Roma tomato juice was close to that of the Garden Gem juice except for its substantially lower levels of hexanal and 2-isobutylthiazole. The 2- isobutylthiazole is known to be a degradation product of leucine/isoleucine (Wang, Chin,

Ho, Hwang, Polashock, & Martin, 1996). Like hexanal, this compound gives a pungent,

tomato vine aroma note, but here 2-hexenal was found to be abundant in Roma juice.

Although these compounds can be characterized as green notes, minor differences in odor threshold and descriptors may be observed for specific compounds. No fruity and sweet-related β-ionone was detected in either the commercial or Roma juices. The concentration of furfural was below its reported odor threshold (Buttery, 1990).

However, furfural’s potential synergistic effect with other compounds could not be ruled out.

The PCA plots for both seasons (Figures 3-1 and 3-2) further illustrate the fruity note compounds (6-methyl-5-hepten-one, β-citral, β-ionone, gerenylacetone, linalool, β- cyclocitral and β-damascenone) clustered together for the Garden Gem juice in both seasons. The Roma processed tomato juice was linked to 2-hexenal and a few other compounds. However, other C6 volatiles were more related to Garden Gem after thermal processing. Garden Gem contains relatively high levels of green and fruity note compounds. It is not surprising that the cooked flavor-related compounds were clustered with commercial tomato juice; few of the fruity related compounds were close to the commercial product, as expected, since the commercial product was found to contain minimal levels of fruity related compounds. Compounds such as β- damascenone are known (Buttery, 1990) to be free volatiles released from their bound form after acid or enzymatic cleavage (such as β-glucosidase and pectinase). It is therefore possible that either the enzyme or the heat, or some combination of both, helps the bound volatile release such compounds.

Seasonal Effect for Garden Gem, Roma and Commercial Juice

As demonstrated in Table 3-7, higher concentrations of volatile compounds were identified in the Summer than in the Fall for all three varieties. The compound

concentrations were not statistically significant for Garden Gem except for acetone, 6- methyl-5-hepten-2-one, linalool, β-cyclocitral, benzeneacetaldehyde, β-ionone and 3- hexen-1-ol. However, most of the concentration differences between the compounds in the RA and commercial samples were significant. For the Fall season, RA suffered a

91% loss of citral, a 91% loss of 6-methyl-5-hepten-2-one, and a 6% loss of benzaldehyde. However, β-damascenone was observed to rise by 164% in the Fall. For the commercial tomato juice, the Summer season showed a 128% increase in 2-methyl butanal, a rise of 244% in cis-geranylacetone, a 69% loss in benzaldehyde and a 94% loss of furfural.

Summary

Overall, tomato juice made from Garden Gem tomatoes was rated significantly

(p<0.05) higher for overall likability, tomato flavor, and sweetness regardless of seasonal variation. However, uniformity and color aspects may need some improvement to achieve an acceptable level for a commercial product. Garden Gem juice was found to contain significantly (p<0.05) higher amounts of several sweet fruity related aroma compounds: 6-methyl-5-hepten-2-one, linalool and β-ionone. Seasonal variations in SS/TA were not significant. This may indicate that the flavor compounds are better indicators for tomato juice acceptance than traditional non-specific chemical analysis. The commercial tomato juice tested contained relatively high amounts of the

Maillard reaction-related note furfural and dimethyl sulfide and the least amount of green-related notes (hexanal, E-2-hexenal and Z-2-heptenal). Despite its less fruity and green-related notes, the commercial tomato juice contained ethyl acetate, which was not found in either of the tomato juices without additives. The flavor profile of the Roma tomato juice contained many similar volatile compounds to those in the Garden Gem

juice, but the Roma juice contained substantially lower amounts of hexanal and 2- isobutylthiazole. No fruity and sweet-related β-ionone was detected in either the commercial or the Roma juice. Most of the flavor compounds identified in the Roma and

Garden Gem thermally processed juice corresponded to those found in the fresh state, but at lower levels. This chapter reports the detailed flavor profile of a processed

Garden Gem tomato juice and demonstrates its potential market advantage over current commercial products, despite some seasonal variation.

Table 3-1. Chemical analysis of thermally processed tomato juice using Duncan’s multiple range test mean separation. Means (± standard deviation) with different letters within the same column are significantly different at p< 0.05. Variety Titratable pH Soluble SS/TA Acidity(g/100mL) Solids(°brix) Commercial fall 7.15 B ±0.08 3.64E±0.01 5.30D±0.00 11.58D±0.13

Commercial summer 8.58A±0.05 3.94D±0.02 5.53C±0.06 10.08E±0.16

GG fall 6.22C±0.02 3.94D±0.01 5.90B±0.00 16.08B ±0.05

GG summer 8.59A±0.06 4.23B±0.03 6.63A±0.05 12.08D±0.13

RA fall 5.88D±0.02 4.35A±0.04 5.27D±0.11 14.00C±0.26

RA summer 5.38E±0.04 4.07C±0.04 5.73B±0.23 16.67A±0.79

N=3 for each sample

Table 3-2. Sensory ratings using a nine-point hedonic scale (1-dislike extremely, 9-like extremely) for thermally processed tomato juice (2015 Summer Season) with Duncan’s multiple range test mean separation. Means (± standard deviation) with different letters within the same column are significantly different at p< 0.05. Variety After tasting Before tasting Overall Sweetness Tomato Mouthfeel Color Uniformity Likability Flavor Commercial 5.41B±2.07 4.79B±1.83 5.51B±2.04 5.83A±2.09 6.86A±1.40 6.89A±1.45 summer GG summer 6.08A±1.76 5.90A±1.85 6.29A±1.82 5.86A±1.90 6.76A±1.40 6.30B±1.40 RA summer 4.93B±1.86 4.98B±1.82 5.33B±1.91 4.58B±2.02 5.66B±1.54 5.31C±1.45

N=119 for each sample

Table 3-3. Sensory means of thermally processed tomato juice (2015 Fall Season). Means were separated using Duncan’s multiple range test for mean separation. Means (±standard deviation) with different letters within the same column are significantly different at p< 0.05. Variety After tasting Before tasting Overall Sweetness Tomato Mouthfeel Color Uniformity Likability Flavor Commercial 4.91B±1.99 4.63B±1.96 5.18B±1.91 5.49A±1.98 6.77A±1.52 7.02A±1.22 fall GG fall 5.81A±1.80 5.98A±1.79 6.10A±1.70 5.66A±2.00 6.57A±1.35 5.86B±1.56 RA fall 4.60B±1.79 4.76B±1.77 5.18B±1.85 4.58A±2.06 5.16B±1.75 5.18C±1.80 N=119 for each sample

Table 3-4. Significant correlations (r) for sensory analysis of Summer 2015 season processed tomato juice (significant at p≤0.05) Overall Tomato Variables Liking Flavor Sweetness Mouthfeel Color Uniformity Overall Likability 1 0.845 0.782 0.746 0.356 0.336 Tomato Flavor 0.845 1 0.757 0.671 0.335 0.285 Sweetness 0.782 0.757 1 0.664 0.286 0.252 Mouthfeel 0.746 0.671 0.664 1 0.355 0.431 Color 0.356 0.335 0.286 0.355 1 0.675 Uniformity 0.336 0.285 0.252 0.431 0.675 1 N=119 for each sample

Table 3-5. Significant correlations (r) for sensory analysis of Fall 2015 season processed juice (significant at p≤0.05) Overall Tomato Variables Liking Flavor Sweetness Mouthfeel Color Uniformity Overall Likability 1 0.832 0.809 0.724 0.463 0.366 Tomato Flavor 0.832 1 0.716 0.636 0.396 0.290 Sweetness 0.809 0.716 1 0.611 0.361 0.209 Mouthfeel 0.724 0.636 0.611 1 0.522 0.505 Color 0.463 0.396 0.361 0.522 1 0.724 Uniformity 0.366 0.290 0.209 0.505 0.724 1

Table 3-6. Correlation coefficient (r) for sensory, chemical and flavor analysis (significant at p≤0.05 *) of processed tomato juice for both seasons. Variables Overall liking Sweetness Tomato flavor Overall likability 1* 0.866* 0.967* Sweetness 0.866* 1* 0.957* Tomato flavor 0.967* 0.957* 1* Dimethyl sulfide -0.336 -0.693 -0.559 Acetone 0.649 0.263 0.520 Ethyl Acetate -0.326 -0.377 -0.396 Glutaraldehyde 0.560 0.271 0.439 Butanal, 2-methyl- -0.391 -0.381 -0.434 Butanal, 3-methyl- 0.357 -0.007 0.211 Disulfide, dimethyl -0.362 -0.731 -0.551 Hexanal 0.659 0.920* 0.828* 1-Penten-3-ol 0.508 0.689 0.656 Heptanal 0.722 0.911* 0.861* 2-Hexenal, (E)- -0.371 0.018 -0.142 1-Pentanol 0.364 0.549 0.443 Sulfide, methyl propyl 0.127 -0.368 -0.107 2-Heptenal, (Z)- 0.494 0.681 0.633 5-Hepten-2-one, 6-methyl- 0.587 0.540 0.600 1-Hexanol 0.466 0.758 0.630 3-Hexen-1-ol, (Z)- 0.537 0.705 0.663 2-Isobutylthiazole 0.361 -0.125 0.118 2-Octenal, (E)- 0.748 0.822* 0.853* Linalool oxide -0.182 -0.595 -0.413 Methional 0.224 -0.232 0.016 Furfural -0.155 -0.588 -0.392 5-Hepten-2-ol, 6-methyl- 0.155 -0.297 -0.055 2,4-Heptadienal, (E,E)- -0.090 0.352 0.157 Decanal 0.668 0.602 0.659 1-Hexanol, 2-ethyl- 0.163 0.559 0.401 Benzaldehyde 0.909* 0.817* 0.856* Linalool 0.419 0.484 0.500 2-Furancarboxaldehyde, 5-methyl- -0.210 -0.634 -0.450 3,5-Heptadien-2-one, 6-methyl-, (E)- 0.671 0.455 0.646 β-Cyclocitral 0.827* 0.709 0.844* Benzeneacetaldehyde -0.101 -0.458 -0.308 α-Terpineol 0.258 0.438 0.313 β-Citral 0.611 0.587 0.630 2,6-Dimethylbenzaldehyde 0.621 0.679 0.729

Table 3-6. Continued Variables Overall liking Sweetness Tomato flavor Methyl salicylate 0.849* 0.631 0.741 β-Damascenone 0.377 0.475 0.456 Geranylacetone 0.117 0.488 0.285 Phenylethyl Alcohol 0.408 0.157 0.331 β-Ionone 0.807 0.883* 0.904* Eugenol -0.102 0.263 0.078 TA 0.099 -0.221 -0.065 Brix 0.806 0.822* 0.853* pH -0.305 -0.032 -0.127 SS/TA 0.255 0.553 0.416 *Significant correlation p<0.05

Table 3-7. Comparison of quantitative analyses of major volatiles (in ppb) for processed tomato juices. DB-5 WAX GGP- GGP- RAP- commercia commerci WAX RT Compound RAP-fall TD Note RI RI summer fall summer l-summer al-fall

741 835 2.26 Dimethyl sulfide n.d n.d n.d 1.68 6.73 9.50 0.3 Cooked

744 871 2.48 Acetone 182.49 51.87 n.d 36.83 133.9 92.86 40 Fruity

753 918 2.96 Ethyl Acetate n.d n.d n.d n.d 8.27 4.57 50 -

759 929 3.07 Glutaraldehyde 17.34 11.34 n.d 3.19 40.24 92.07 - - Butanal, 2- 761 938 3.22 n.d n.d 6.08 15.54 n.d n.d - - methyl- Butanal, 3- 776 939 3.23 18.17 11.78 n.d 25.82 n.d n.d 0.2 Fruity methyl- Disulfide, 781 952 6.63 n.d 1.0 n.d 1.00 3.02 1.80 9.8 Cooked dimethyl 852 1084 6.84 Hexanal 91.19 85.48 36.69 56.22 24.47 18.75 4.5 Green Fruity/ 759 1168 10.60 1-Penten-3-ol 9.82 5.05 6.08 5.43 2.54 n.d 1 green 1183 11.28 Heptanal 3.15 2.38 1.29 1.30 n.d n.d - Green

855 1213 12.76 2-Hexenal, (E)- 19.45 22.17 23.22 63.8 2.69 2.06 17 Green

789 1215 12.87 1-Pentanol 11.42 8.47 n.d 5.38 n.d n.d 1600 Green

Sulfide, methyl - 1253 14.76 2.91 1.32 2.04 1.22 8.41 3.80 - Cooked propyl

5-Hepten-2- 988 1334 18.55 702.3 195.5 421.2 46.22 212.5 112.6 50 Fruity one, 6-methyl-

874 1359 19.71 1-Hexanol 15.75 12.06 n.d 6.04 n.d n.d 200 Green

Table 3-7. Continued DB-5 WAX GGP- GGP- RAP- commercia commerci WAX RT Compound RAP-fall TD Note RI RI summer fall summer l-summer al-fall 3-Hexen-1-ol, 858 1387 20.99 84.98 35.34 n.d 25.47 n.d n.d - Green (Z)- 2- 1140 1398 21.52 7.18 4.55 1.95 n.d n.d n.d 3.5 Green Isobutylthiazole

1005 1420 22.44 2-Octenal, (E)- 2.61 1.92 n.d 1.44 n.d - Green

- 1440 23.31 Linalool oxide 1.90 2.04 - -

- 1445 23.53 Methional 1.91 1.46 1.83 0.96 4.11 1.76 0.2 Cooked

874 1455 24.17 Furfural 11.69 5.92 n.d n.d 0.14 0.01 3000 Cooked

5-Hepten-2-ol, - 1475 24.54 2.11 0.59 n.d n.d n.d n.d 2000 Fruity 6-methyl- 2,4- - 1482 25.14 Heptadienal, 0.71 0.71 n.d 1.11 n.d n.d - - (E,E)- - 1492 25.55 Decanal 1.56 2.22 n.d 1.20 n.d n.d - - 1-Hexanol, 2- 1031 1494 25.66 2.36 2.65 n.d 3.58 n.d n.d - Fruity ethyl- 1100 1551 27.98 Linalool 38.35 16.63 25.15 18.21 13.54 13.96 6 Fruity

1216 1605 30.14 β-Cyclocitral 5.55 3.24 6.46 n.d n.d n.d 5 Fruity

Benzeneacetal - 1628 31.03 9.04 3.27 4.33 3.94 2.45 n.d 4 - dehyde

Table 3-7. Continued DB-5 WAX GGP- GGP- RAP- commercia commerci WAX RT Compound RAP-fall TD Note RI RI summer fall summer l-summer al-fall Citrus, 1238 1670 32.64 β-Citral 6.80 2.43 n.d 0.30 n.d n.d 30 flora 2,6- - 1711 33.74 Dimethylbenzal 2.59 0.67 0.91 0.77 n.d n.d - - dehyde Methyl 1188 1758 35.85 3.73 3.39 1.88 0.81 2.69 3.04 40 Green salicylate β- 1376 1809 37.69 3.45 2.31 n.d 2.20 0.47 1.61 0.002 Fruity Damascenone Fruity, - 1931 41.70 β-Ionone 1.09 0.59 n.d n.d n.d n.d 0.007 sweet *RI of DB-5 and DB-WAX Plus column was reported based on C8-C20 alkaline standard Aroma note was summarized from Kazeniac & Hall (1970), Rowe (1998), Galliard (1977), Buttery (1993), Yilmaz (2001), Klee (2010) and Burdock (2016)

Table 3-8. Major Volatiles in fresh Garden Gem (GG) and Roma (RA). RT Identification Mol GG- RA- (min) Weight Concentration Concentration (amu) (ppb) (ppb) 2.49 Acetone 58 47.57±1.44 23.25±0.16 2.88 Glutaraldehyde 100 4.93±0.53 12.05±4.56 2.96 Ethyl Acetate 88 8.95±0.29 n.d 3.08 2-Butanone 72 21.13±1.96 n.d 3.26 Butanal, 3-methyl- 86 28.0±3.35 17.30±0.73 4.13 3-Pentanone 86 34.23±2.00 n.d 4.27 Butanoic acid, methyl ester 102 1.14±0.13 n.d 4.45 3-Pentanone, 2-methyl 100 1.70±0.30 n.d 5.00 1-Penten-3-one 86 30.6±0.46 114.9±3.93 5.28 2-Pentanol, 2,4-dimethyl- 116 8.46±5.09 n.d 5.71 3-Buten-2-ol, 2-methyl- 86 20.82±0.15 n.d 6.51 5-Hexen-2-ol, 5-methyl- 114 1.66±0.23 n.d 6.61 Disulfide, dimethyl 94 0.84±0.06 3.31±0.16 6.85 Hexanal 110 320.0±5.71 669.3±13.90 7.24 2-Butenal, 2-methyl- 84 12.32±0.01 25.11±0.40 8.24 3-Pentanol 88 0.48±0.19 n.d 8.58 Butanenitrile, 3-methyl- 83 10.36±0.23 n.d 8.70 2-Pentenal, (E)- 84 3.43±0.47 22.48±1.23 9.22 3-Hexenal 98 188.6±11.35 305.9±6.40 9.90 1-Hexene, 2-methyl- 98 4.72±0.79 n.d 10.59 1-Penten-3-ol 86 46.18±0.87 64.36±12.10 11.32 Heptanal 114 1.79±0.23 7.84±2.27 11.86 2-Butenal, 3-methyl- 84 4.02±0.15 n.d 12.04 2-Hexenal, (E)- 98 15.80±3.51 42.53±1.06 12.75 2-Hexenal 98 122.3±10.97 698.2±3.43 12.89 Butanol, 2-methyl-,(S) 88 65.17±41.10 n.d 14.01 4-Heptenal, (Z)- 112 1.00±0.03 5.14±0.09 14.60 Heptane, 2,3-epoxy- 114 0.65±0.04 n.d 14.92 1-Pentanol 88 9.74±0.03 15.41±1.58 16.08 6-Methyl-2-Heptanol, acetate 172 0.94±0.01 n.d 16.22 Octanal 128 0.98±0.02 4.01±0.07 16.37 Butane, 1-nitro- 103 0.47±0.04 n.d 16.84 3-Heptanol, 3-methyl- 130 1.08±0.03 n.d 17.69 2-Heptenal, (Z)- 112 48.43±0.44 10.99±0.26 17.78 2-Penten-1-ol, (Z)- 86 6.13±0.10 23.66±0.31 17.93 1-Pentanol, 4-methyl- 102 1.60±0.07 n.d 18.12 Pentane, 1-nitro- 117 34.26±0.92 n.d 18.59 5-Hepten-2-one, 6-methyl- 126 642.2±33.19 112.9±0.76 19.15 5-Heptenal, 2,6-dimethyl- 140 0.78±0.15 n.d 19.71 1-Hexanol 102 59.57±5.12 61.11±0.45 20.42 2-Propylthiazole 127 0.50±0.03 n.d 20.99 3-Hexen-1-ol, (Z)- 100 171.4±15.14 110.3±0.12 21.11 Cyclohexanone, 2-ethyl-, oxime 141 2.53±0.04 n.d 21.19 2,4-Hexadienal, (E,E)- 96 2.43±0.06 3.68±0.08 21.34 Citronellyl formate 184 4.03±0.44 n.d 21.51 2-Isobutylthiazole 141 63.31±2.48 3.91±0.11 22.01 2-Hexen-1-ol, (E)- 100 1.83±0.06 4.36±0.10 22.43 Furan, 3-(4-methyl-3-pentenyl)- 150 4.60±0.18 n.d 3-Methyl-2-butenoic acid, undec- 22.65 252 0.90±0.24 n.d 10-enyl ester

Table 3-8. Continued RT Identification Mol GG- RA- (min) Weight Concentration Concentration (amu) (ppb) (ppb) 23.10 1,9-Nonanediol 160 1.46±0.01 n.d 23.28 Linalool oxide 170 0.30±0.11 n.d 23.97 1-Nonen-3-ol 128 8.47±0.27 n.d 24.54 5-Hepten-2-ol, 6-methyl- 128 15.95±0.00 n.d 25.12 2,4-Heptadienal, (E,E)- 110 3.18±0.11 17.70±0.11 25.52 Decanal 156 0.69±0.09 n.d 25.63 1-Hexanol, 2-ethyl- 130 1.91±0.03 n.d 25.93 S-(3-Hydroxypropyl) thioacetate 134 3.76±0.08 n.d 26.17 Benzaldehyde 106 26.41±0.05 6.26±0.01 3,5-Octadiene, 2,7-dimethyl-, 26.70 138 6.33±0.12 n.d (Z,Z)- 26.90 2-Nonenal, (E)- 140 0.44±0.09 1.72±0.00 27.24 Isopulegol 154 0.84±0.29 n.d 27.94 Linalool 154 6.15±1.43 2.90±0.29 28.40 1-Octanol 130 1.00±0.12 2.53±0.20 28.54 cis-Verbenol 152 0.65±0.12 n.d 2(3H)-Furanone, dihydro-4,4- 28.83 114 0.46±0.08 n.d dimethyl- 3,5-Heptadien-2-one, 6-methyl-, 29.22 124 4.11±0.33 n.d (E)- 29.28 2-Methyl-oct-2-enedial 154 0.68±0.16 n.d 29.63 p-Menth-8-en-2-one 152 0.46±0.03 0.61±0.04 Cyclopentane, (3- 29.96 138 1.01±0.07 n.d methylbutylidene)- 30.12 β-Cyclocitral 152 43.69±0.06 1.81±0.02 30.62 2-Octen-1-ol, (E)- 128 0.46±0.01 0.81±0.00 30.99 Benzeneacetaldehyde 120 1.86±0.07 n.d 32.61 β-Citral 152 31.16±1.06 6.55±0.29 33.14 1-Nonen-4-ol 142 0.61±0.01 n.d 33.32 2,4-Nonadienal 138 2.16±0.04 n.d 33.50 L-α-Terpineol 154 1.42±0.24 n.d 34.54 Citral 152 55.00±0.54 8.90±0.24 34.95 Dibutyl carbitol 218 0.92±0.11 n.d 35.81 Methyl salicylate 152 3.56±0.31 n.d 36.40 Methyl 2-hydroxydecanoate 202 0.46±0.05 n.d 37.65 β-Damascenone 190 0.95±0.17 n.d 39.02 Geranylacetone 194 3.80±0.97 6.38±0.15 39.10 1,3-Heptadiene, 2,3-dimethyl- 124 0.66±0.18 n.d 39.81 Benzyl alcohol 108 4.57±0.33 n.d 41.13 Benzyl nitrile 117 1.19±0.01 n.d 41.66 β-Ionone 192 0.84±0.09 0.90±0.06 Formic acid, (2- 45.69 150 1.44±0.14 n.d methylphenyl)methyl ester 46.54 Eugenol 164 0.34±0.18 0.75±0.03

A 2

3,5-Heptadien-2-one, 6-methyl-… Acetone Methyl salicylate 1.5 β-Cyclocitral Decanal BenzaldehydeGGP-summer 2-Isobutylthiazole 2-Octenal, (E)- Butanal, 3-methyl- 1 Phenylethyl Alcohol trans-β-Ionone commericial-summer 2,6-Dimethylbenzaldehyde Sulfide, methyl propyl Heptanal 5-Hepten-2-ol, 6-methyl- Hexanal GGP-summer β-Damascenone Linalool oxideFurfural Benzeneacetaldehyde 1-Hexanol, 2-ethyl- 0.5 Methional cis-Geranylacetone 1-Penten-3-ol Disulfide, dimethyl GGP-summer 3-Hexen-1-ol, (Z)- commericial-summer linalool 5-Hepten-2-one, 6-methyl- 2-Furancarboxaldehyde, 5- β-Citral methyl- commericial-summer 2-Heptenal, (Z)- 0 Glutaraldehyde

F2 (25.65 %) (25.65 F2 1-Hexanol α-Terpineol -0.5 2,4-Heptadienal, (E,E)- Dimethyl sulfide RAP-summer 1-Pentanol 2-Hexenal, (E)- -1

Butanal, 2-methyl-

-1.5 Ethyl Acetate RAP-summer

RAP-summer -2 -2.5 -2 -1.5 -1 -0.5 0 0.5 1 1.5 2 2.5 F1 (55.49 %)

Figure 3-1. Biplot of tomato juice with compounds on PC1 VS PC2 (Pearson model) explaining 81.14% of the summer season data variation.

Benzaldehyde 1.5 5-Hepten-2-one, 6-methyl- GGP-fall GGP-fall Methyl salicylate GGP-fall β-Citral Glutaraldehyde 1 trans-β-Ionone 1-Pentanol Methi… Heptanal 2-Isobutylthiazole Decanal α-Terpineol 1-Hexanol commericial-fall Hexanal Butanal, 3-methyl- 5-Hepten-2-ol, 6-methylAcetone- 0.5 3-Hexen-1-ol, (Z)- cis-Geranylacetone β-Damascenone β-Cyclocitral Benzeneacetaldehyde 2-Furancarboxaldehyde, 5- 2-Octenal, (E)- Ethyl Acetatemethyl- Linalool oxide Furfural 2-Heptenal, (Z)-Phenylethyl Alcohol Sulfide, methyl propyl Dimethyl sulfide 0 3,5-Heptadien-2-one, 6- Butanal, 2-methyl- methyl-, (E)-

.m-Eugenol F2 (19.59 %) (19.59 F2 1-Penten-3-ol commericial-fall 1-Hexanol, 2-ethyl- commericial-fall -0.5 2,6-Dimethylbenzaldehyde linalool Disulfide, dimethyl 2,4-Heptadienal, (E,E)-

RAP-fall -1 2-Hexenal, (E)-

RAP-fall -1.5 RAP-fall

-2 -2.5 -2 -1.5 -1 -0.5 0 0.5 1 1.5 2 2.5 F1 (64.87 %)

Figure 3-2. Biplot of tomato juice with compounds on PC1 VS PC2 (Pearson model) explaining 84.46% of the fall season data variation.

CHAPTER 4 CONSUMER PERCEPTIONS OF TOMATO JUICE

Background Information and Objectives

Sales of soft drinks in the United States have declined in recent years and are expected to continue to lose market share due to the increasing health awareness of consumers. Conversely, most fruit drinks and juices have seen a double-digit growth in sales over the same period (Nielsen, 2015). Juices may be the driving force behind rising beverage sales in the U.S., which are estimated to continue increasing from $131 billion in 2013 to $164 billion by 2018 (BMC, 2015). Tomato-related processed products have witnessed a 14% increase in sales in 2015 alone. Second only to oranges, tomatoes are Florida’s leading commodity (Putnam, 2013). Recently, citrus production has declined worldwide due to citrus greening disease (Halbert & Manjunath, 2004).

Therefore, tomatoes are becoming a more important source of juice. A niche market for value-added, high-quality tomato juice may present a new opportunity for both tomato growers and consumers.

Researchers have found that nutrition knowledge, and demographic factors such as age, income and education level, significantly affect willingness to pay for value- added products (Loureiro & Hine, 2002). Research on tomato juice preferences has mostly used a central location test approach (Goodman, 2002) where the test is administered in a designated place such as a sensory panel school or company. This means that the participants tend to come from the neighboring area or consist of employees working at the same campus. A different approach to consumer research focuses instead on using online surveys, which collect data from multiple regions and can thus examine the attitude, preference, and willingness to pay of consumers,

irrespective of any regional bias (Auger, Devinney, Louviere, & Burke, 2010; Okechuku,

1994).

However, a major limitation of online surveys is that they typically focus on the comparison of non-taste or non-health factors. Many online food studies therefore investigate the impact of factors such as genetically modified or organic labels on consumers’ willingness to pay a premium price for a regular product (Wier & Calverley,

2002). Most consumers in these studies respond that they received a more flavorful, nutritious and safer product through the information provided by these labels (Aarset et al., 2004; Hughner & Kleine, 2004). However, these impressions may or may not be supported by current scientific research comparing organic food to non-organic food

(Lairon, 2011; Magkos, Arvaniti, & Zampelas, 2006; Yeung & Morris, 2001).

The objective of this research was to elicit consumer’s preferences and willingness to pay for tomato juice. The intent was to find which characteristics motivate consumers to drink tomato juice and whether particular groups of consumers are willing to pay a premium price for a healthy and better-tasting product. It would not be surprising if consumers with different backgrounds showed different opinions. In this regard, we first examined the preference pattern of panelists participating in a tomato juice sensory panel at the University of Florida. These participants evaluated a superior flavor product, Garden Gem juice, and were asked about their willingness to pay more for it compared with other products tasted. A follow-up online survey was then conducted among a US-census representative consumer group to determine their awareness of the health and nutrition benefits and taste quality of tomato juice. Also, a

new “label” system, including taste review ratings, was tested to verify whether this information successfully assists consumers in their purchase decisions.

Materials and Methods

Tomatoes

Fully ripe Garden Gem tomatoes were harvested in Summer 2015 from a field

(Live Oak, FL) and in Fall 2015 from a greenhouse (Gainesville, FL). Ripe Roma tomatoes were purchased from local Publix (Gainesville, FL) and Winn-Dixie

(Gainesville, FL) grocery stores in both Summer and Fall 2015.

Juice Processing

Twenty pounds of tomatoes from each cultivar were cut into three pieces by toss and chop kitchen scissors (Comfify, Fernadale, WA, USA). The tomato pieces were then fed into a tomato electric milling machine (Model 2400, O.M.R.A. MTD, Brooklyn,

NY, USA) to remove the skin and seeds. The puree produced was continuously fed into a steam kettle and heated (a process called “hot break”) at 95ºC for 3 min. Then the hot break juice was poured into a sanitized carboy and stored at 4ºC. Later, the juice was pasteurized (Model 5024-F 25 HV, Microthermics, Raleigh, NC, USA) at 73ºC for 30 s and stored in another newly sanitized carboy at 4ºC. The whole process was repeated three times for each variety. A commercial Campbell’s Low Sodium Tomato Juice

(Campbell Soup Company, Camden, NJ) was purchased from Publix for comparison with the sample juices.

Sensory Analysis Survey

Recruiting for the sensory analysis was done through e-mails and announcements posted around the Food Science and Human Nutrition Department at the University of Florida (UF) Gainesville campus. Panelists were screened based on

their familiarity with tomato juice as well as their frequency of consumption and availability. Only those panelists who consume tomato juice were selected to participate.

The sensory laboratory in the Food Science and Human Nutrition Department at

UF has ten individual booths with computer workstations. Panelists were seated, assigned a corresponding panelist number and asked to sign in to the sensory program

(CompuSense® Five Sensory Analysis Software for Windows, CompuSense, Guelph,

Canada) for the session. Each booth was furnished with a tray containing water, crackers, a napkin, and a refuse cup.

For the first part of the experiment, tomato juice consumers (119 in October,

2015, 120 in February 2016) were invited to rate their purchase intent (1-very unlikely to purchase, 5-very likely to purchase) and willingness to pay (name their own price for a

64 o.z juice, with a reference price of $2.98 provided) for each product before and after the three types of tomato juice samples were tasted. The three thermally processed juices (Garden Gem, Roma, and commercial juice) were randomly coded with a three- digit number and presented in a balanced experimental design. All the samples were served at room temperature; the pilot juices (Garden Gem and Roma) contained no additives. The panelists received the same questionnaire for the sessions in both the

Summer and Fall.

In the second part of the experiment, panelists were asked a series of questions concerning their demographics, consumer behavior towards tomato juice, and perceptions about Florida tomatoes. In the third part of the experiment, participants were asked what they considered to be the most important attributes for drinking tomato

juice and for not purchasing tomato juice. The set of twenty-nine attributes offered to them were generated and summarized from a previous focus group consisting of regular tomato juice consumers in Fall 2014. These attributes were rated by sensory analysis participants using a 5-point scale (1-strongly disagree, 5-strongly agree).

Online Survey

A nationwide online survey was conducted from September 7 to September 12,

2016. Participants age eighteen or above were recruited according to the U.S demographic criteria provided by the United States Census Bureau (USCB, 2012). A total of 1299 participants qualified from among the 1864 participants after excluding those who did not complete the survey, declined the privacy policy, or would not want to consume tomato juice under any circumstances. Among these 1299 qualifiers, 75.7%

(983 respondents) completed the screening questions satisfactorily. Three hundred and sixteen consumers were excluded from the survey because they either failed the validation questions used to control for response quality (Jones, House, & Gao, 2015) or were below the age of eighteen.

The survey consisted of four sections. In the first section, participants were asked nutrition and health related questions, then questioned about their consumer behavior towards tomato juice, labels and perceptions about Florida tomatoes specifically. In the second section, individuals were asked their opinions on the most important attributes for drinking tomato juice and not purchasing tomato juice. The twenty-nine attributes

(See Tables 4-7, 4-11) were the same as those used in the sensory analysis survey conducted previously. These attributes were rated by survey participants on a 5-point scale (1-strongly disagree, 5-strongly agree). The third section was a set of choice experiment questions (see next section) designed to elicit their willingness to pay when

given a new labeling system that included taste review scores (reflect sensory testing score in previous chapter). The important factors listed in the choice experiment options were ranked by individuals, who then indicated the characteristics of tomato juice that would cause them to pay a higher price. The final section was a set of demographic questions regarding their socioeconomic status, age, gender, income, and education.

Choice Experiment

In each question, individuals were required to choose from three options (Figure

4-1). Two of those options described the tomato juice attributes and the other consisted of the statement “I wouldn’t purchase either product”.

To identify important tomato juice attributes, the sensory analysis results reflecting tomato juice purchasing intent were employed. Four attributes were selected for inclusion in the choice experiment: price, taste review, origin, and concentrate/ingredients added. The prices offered were $1.98, $2.98, $3.98, reflecting the range of tomato juice prices in grocery stores during the period of the study. The taste review consisted of three levels: 2.5 stars, 3.5 stars and 4.5 stars in a 5-star consumer review system. “From Florida” as the origin or none was the third attribute.

For concentrate/ingredients combination, the following three options were included: from concentrate, ingredients added; not from concentrate, no ingredients added; not from concentrate, ingredients added. The unrealistic combination “from concentrate, no ingredients added” was excluded.

A full factorial experimental design that included all possible combinations would require 2916 choice sets. Since it is not reasonable to ask consumers to select from such a large number of choice sets, 15 purchasing scenarios were generated via a

fractional experimental design with minimized D-error. To avoid possible ordering effects, the choice sets were randomized except for the alternative placed last.

Models and Statistics

Choice experiments are based on the assumption that consumers make rational decisions and that the decision will be based on the maximization of the benefit (or the utility) they perceive from a given set of alternatives. In this study, the alternative could be the price, taste review score, origin, concentrate/ingredients combination and unknown factors. For an individual consumer i, the utility of choosing alternative k can be expressed as:

Uki = Vki + ki (4-1) where Vki is the deterministic part of utility that can be explained by the model and ki captures the unobserved factors that may affect utility. The observed part of the utility function is assumed to be linear in parameters, and the probability that individual i would choose alternative k over alternative j is:

Pki= P(Vki + ki > Vji + ji ) = P( ji < Vki - Vji+ ki) ∀ j ≠ k; kCi, (4-2) where Ci denotes the choice set that respondent i faces. Therefore, the probability of choosing alternative k is the cdf of the random error at point (Vki - Vji+ ki). If the unobserved factors are iid of type I extreme value distributed, the choice probability can be expressed as:

푢푉푘푖 (4-3) Pki = 푒 ⁄ 퐽 푢푉푗푖 ∑푗=1 푒 which is the multinomial logit (MNL) model (Hensher, Greene, & Chorus, 2013; Train,

2002). The scale u is inversely related to the variance of the error terms and is not

identifiable in a single model, hence the estimated parameters are compounds made up of the preference and scale parameters.

Compared to the MNL model, the mixed logit model allows for variation among consumers regarding taste. Therefore, there is no specific behavioral assumption to be restricted (Train, 2002). The probability that an individual consumer i will choose alternative k is conditional on the individual specific parameter βi. Assuming that the coefficients vary among individuals but are stable for each consumer, the probability that an individual chooses a particular sequence of choices is the product of the conditional probability function. The integral of this product over all β values gives the unconditional probability function:

(4-4) 푁 푒푥푝⁡(훽푖푥푘푖푛) Pkin = ∫ (∏ [ ]) 푓(훽)푑훽 푛=1 ∑퐽 푗 푒푥푝⁡(훽푖푥푗푖푛) where N represents the number of choice sets that each individual faces (Petrin & Train,

2003). The density function f(β) is assumed to be normal, and takes the form βi=b+ei, where b is the population mean,  is the coefficient standard deviation and the iid random error term ei ~N(0,1). If  is zero, the mixed logit model collapses to the MNL model. We assume that the quality attributes (taste, origin, etc) and the opt-out alternative follows a normal distribution, while the price parameter is fixed. Because the choice probability for a mixed logit model has no closed form solution, parameters were estimated by maximizing the simulated maximum likelihood using 1,000 bootstrapping model draws.

The willingness to pay (WTP) estimates can be calculated according to the logit model results and compared across the models. The average WTP are calculated as the ratio of the attribute coefficient divided by the price coefficient. For the mixed logit

model, the means and standard deviations are derived by the Krinsky and Robb method with 1,000 simulations (Krinsky & Robb, 1986).

Group Segmentation

A factor analysis was used to segment participants into groups according to their opinions on tomato juice consumption. All the variables were examined with both a chi- square test and a 1-way analysis of variance (ANOVA) using SAS 9.4 software package

(SAS Institue Inc, NC, USA).

Results and Discussion

Sensory Analysis Demographic and Opinion

Descriptive statistics for the sensory panelists’ demographic information are summarized in the right hand column of Tables 4-1 and 4-2. A typical participant for the sensory testing was female, aged 18-29, with a college degree, an income of $23,000-

$34,999, and a resident of central Florida. This sample may not represent the population of Florida or the United States well, but this will generally be the case in a central-location-test sensory testing at the University of Florida. Central location tests are known to result in high response rates and to give relatively reliable results (Lawless

& Heymann, 2010). The measure of consumer approval can be efficiently achieved by using a well-designed experiment and well-scripted staff-consumer interactions.

These tomato consumers were recruited from panelists on a listserv of the

University of Florida –Sensory Panel. Most of the participants were students or staff associated with the university who had an interest in supporting research and were willing to devote some of their time to tasting samples. These panelists may represent generally well educated, Millennial consumers who are the target consumers of tomato juice (Prado, Parada, Pandey, & Soccol, 2008). Though the income of these panelists

was at the low end of the scale, the status of their income was likely to change dramatically after graduation. However, tomato juice is not an expensive product and is one that a college student can easily afford.

Consumers tend to alter their purchase intent for tomato juice according to the characteristics they expect and experience. Since all the samples were randomized and coded with a 3-digit number, the consumer relied solely on the appearance characteristics and taste experience. The summary of the responses for purchase intent and price are presented in Tables 4-3, 4-4, 4-5, and 4-6. After tasting, Garden Gem was most likely to be purchased by consumers at a significantly (p<0.05) higher level.

However, before tasting, the commercial and Garden Gem juice showed no significant difference relating to purchase intent and willingness to pay. This change in preference is a result of the taste test itself. Consumers liked the appearance of the commercial tomato juice when the taste experience was not available (i.e., before tasting), but when given the chance to experience the taste, participants favored Garden Gem significantly with respect to both overall likability and tomato flavor. It is likely that the sensory experience affected the purchase intent and may further influence purchase behavior.

Interestingly, the commercial juice significantly discouraged consumer’s willingness to pay after tasting the juice. The panelists dropped the price they would be willing to pay for the commercial juice to such an extent that they would pay only a 30% lower price compared with the Roma (which consistently rated low in the sensory analysis both before and after taste). The vivid color and uniformity of the commercial juice may provide consumers with high expectations about tomato juice when other information is not available and they are then disappointed when the taste of the

commercial juice fails to live up to their expectations. This meant consumer’s purchase intent dropped dramatically when tested about future purchases.

On a scale from 1 (not at all important) to 5 (very important), more than 50% of the sensory testing panelists rated bad taste, lack of freshness and concerns about microbial contamination as very important attributes that prevented them from purchasing tomato juice (Table 4-7). This demonstrates a need for a tomato juice that is enhanced with tomato flavor and freshness.

More than 50% of consumers purchased tomato juice from the grocery stores, restaurants or at a bar with an alcoholic drink (Table 4-8). Approximately 40% of the panelists had never consumed tomato juice in an airplane or made fresh juice at home.

Interestingly, a study by Acree (2015) revealed an increased preference for tomato juice on airplanes compared with consumption at normal atmospheric pressure. Their habits of tomato juice consumption confirmed that most college students drank tomato juice outside the home, with or without blending with alcoholic drink.

A factor analysis was used to segment tomato juice consumers into different groups according to their opinion on characteristics for purchase of tomato juice. The segmentation was determined by eigenvalues (>1) from fifteen factors that influence tomato juice consumption. Three segments (Table 4-9) of consumers were formed in both Summer and Fall season, with a slight seasonal variation among consumers.

Segment 1 on the sensory testing represents consumers (36.13%-Season 1; 47.06%

Season 2) who desired a safe, nutritious and healthy product, with a good appearance, from a trusted brand tomato juice. About 39.05% and 26.05% of consumers were represented in Segment 2 in Season 1 and Season 2, respectively. Consumers from

Segment 2 most favored a combination of sourness, sweetness, saltiness and texture.

About 24.37% (Season 1) and 36.89% (Season 2) of consumers comprised Segment 3, who consumed tomato juice for its freshness and tomato flavor. When questioned about opinions on “Florida Tomatoes,” these tomatoes were perceived to be better tasting and were rated significantly higher. This was agreed upon among all segments. More than

20% of the Segment 2 consumers believed in the taste and premium quality characteristics of Florida tomatoes. These numbers show sensory analysis participants have a generally good impression of Florida tomatoes.

Panelists’ Willingness to Pay

According to panelists, a price premium of $0.37 and $0.41 for Garden Gem in the Summer and Fall seasons, respectively, was acceptible compared with commercial juice (64 oz). A linear regression price and demographic information showed consumers with a higher income or education tended to be willing to pay more for Garden Gem.

Across all groups, consumers’ purchase intent and stated willingness to pay was significantly affected by their taste experience. As the most direct and easy way to convey the consumer’s willingess to pay, contingent valuation (name your own price) is used extensively to estimate how much people value a product that is not yet on the market. Despite these advantages, the “name your own price” method suffers a serious drawback – that the respondents may not accurately state their price sensitivity

(Chhabra, 2015) as the respondents may not consider the willingness to pay question they answer to reflect a real-life purchase scenario. In daily life, the consumer sees the product in the grocery store with its labels. The labels may contain information they care about, such as origin, type, and nutrition, which taken together initiate the purchase behavior. Therefore, there is a need to develop an improved model capable of

uncovering people’s actual willingness to pay for tomato juice made from Florida tomatoes, especially those consumers living elsewhere in the nation.

Online Survey Demographic and Opinions

The descriptive statistics of the national online survey participants are shown in

Table 4-10. The sample generally corresponded well with the U.S. census population

(USCB, 2010) by gender, region and age over 18 except for the variables of education, income, race, and ethnicity. Since the respondents were screened to be tomato juice consumers, the sample is not expected to be representative of the entire U.S. in all aspects. Among the tomato juice consumers in the survey, the African-American population and Hispanic/Latino ethnicity were slightly under-represented at 4.23% and

4.49%, respectively. This is not uncommon in online panels (Couper, 2000; Daugherty,

Lee, Gangadharbatla, Kim, & Outhavong, 2005). The majority (53%) of participants held a college degree or above and had an annual household income of $50,000 or above.

The vast majority (72%) of tomato juice consumers lived in a house with 1-3 children under the age of18.

Survey participants and their families generally reported being in good health

(Table 4-11). Of the participants, 73.43% and 52.42% claimed no diabetes or high cholesterol in their household, respectively. However, the rate of those who were overweight was 34.76% for the participants themselves and 20.47% for other household members. Many reported trying diet or exercise within one year, at 48.12% and 59.72%, respectively. Approximately three-quarters (76.09%) of the participants searched for health information related to food and beverages, and 58.61% of these searched for that type of material at least once a week. Therefore, it is not surprising that 70.30% of the survey-takers claimed moderate or above average knowledge of the health benefits

of tomato juice on a 7-point scale (1-not at all knowledgeable, 4-moderate knowledge,

7-expert).

Tomato Juice Consumption and Purchasing Opinions

Participants consumed tomato juice at a moderate frequency (Table 4-9) and during different occasions (Table 4-12). They indicated that they purchased tomato juice from the grocery store (75.35%), made it at home (47.24%), or drank it in a bar

(35.14%). A total of 32.25% and 24.42% of the participants indicated they had consumed tomato juice during the previous week and previous month, respectively, with more than 80% consuming tomato juice within the past year. With respect to reading labels, more than half of the survey-takers have noticed that the tomato juice could be labeled either “from concentrate” or “not from concentrate.”

Views on the origin of tomato juice diverged from person to person, with generally positive remarks about Florida tomatoes (Table 4-13). A quarter of the participants claimed the origin of tomato juice does not matter to them, while a second quarter of respondents liked tomatoes from other states or countries such as Ohio,

South Carolina, Italy, Canada, and Mexico. In particular, 28.46% and 21.84% of respondents preferred to purchase tomatoes from California or Florida, respectively.

When asked their impression of Florida tomatoes, the top answers given by participants were “premium quality,” “taste better,” and “more nutrients.”.

Similar to the opinion of the sensory testing panelists, a vast majority of the online survey takers agreed that the problem with tomato juice is its “lack of freshness,”

“bad taste,” and “microbial contamination concern”, all rated at 4 or 5 on a 5 point scale

(1-strongly disagree, 5-strongly agree) (Table 4-14). There has not been a foodborne pathogen outbreak associated with tomato juice in the US in the past few decades.

Perhaps participants may have associated “microbial contamination concern” with recent US recalls related to foodbourne illness outbreaks linked to fresh tomato consumption. More than 60% of the participants agreed that “sugar added” and “salt added” as well as “price” were the major reasons that prevented them from purchasing tomato juice.

When it comes to their reason for tomato juice consumption, survey participants were divided into two segments using factor analysis. Segment 1 mostly summarized the reason as taste and additive free driven and comprised 45.17% of the respondents, while the remaining 54.83% (Segment 2) considered natural tomato flavor, freshness, nutrition and label information (origin, organic, concentrate) as their motivation to purchase.

The multinomial logit model (MNL) tested the effects of price, taste review, origin, type parameter and “not variable” (not choose either one) on consumers. When examining responses for the 983 online participants, 74 chose “not to buy either one” for all fifteen choice experiment questions. These participants may simply not be willing to purchase tomato juice, or may not be willing to participate in the choice experiment.

Two MNL models using 983 and 909 participants, respectively, were established as outlined in Table 4-15. Both models showed a similar trend, but the likelihood ratio value for the latter model was higher, indicating a better fit.

The price coefficients were negative and significant for all model specifications.

The negative coefficient is an indication that tomato juice would be less likely to be chosen for a higher price. This is the expected relationship according to economic theory and fits with the claim that 60% of consumers think the price of tomato juice

prevents them from purchasing. The coefficients for taste review were significantly positive (0.68), suggesting the probability that tomato juice was favored by consumers as taste reviews increased. The results for tomato type coefficient were significant, which demonstrated that consumers in general prefer tomato juice without ingredients added and not from concentrate. Besides our pilot juice made from Garden Gem and

Roma, none of the tomato juice products currently on the market fulfills “without added ingredients and not from concentrate”. However, among all the current tomato juice products, two relatively new products have the claim “not from concentrate” and are sold at a premium price. Recently, Hirzel Canning Company and Farms launched a new tomato juice labeled Dei Fratelli’s Truly™ Tomato! Juice (2015). Made from north west

Ohio and southeast Michigan vine tomatoes, the juice has a “not from concentrate” claim but with sea salt added. Similarly, Lakewood’s pure organic tomato juice added organic lemon juice into the product (2016) and is sold as a high end juice.

A relatively low but significant coefficient was found in the origin parameter.

Consumers may choose tomato juice with a Florida origin more often than a tomato juice with no origin information. Interestingly, a highly positive coefficient of “none” was observed significantly in this model. Since the “none” refers to the “not buy either one” option in choice experiment questions, the significant “none” coefficient indicated consumers might choose not to buy tomato juice in some circumstances. This is reasonable as the previous questions conveyed that tomato juice is not a product that consumers tend to purchase and consume frequently.

Survey Willingness to Pay

The willingness to pay (WTP) estimates were derived from the MNL model coefficients shown in Table 4-16. Consumers were willing to pay $1.33 more for

premium tomato juice as the taste review score increased by one star. Participants would pay about $1.04 more for tomato juice with no ingredients added or for not from concentrate. However, only a $0.24 premium was observed when a tomato juice was labeled “From Florida” compared to no information. Hence, consumers were likely to pay for a tomato juice product with a high indication of taste or a more natural product.

As Table 4-17 shows, the estimated 95% confidence interval of taste and origin showed negative numbers in the lower bounds. This indicates heterogeneity in preferences, especially in taste review and origin information. The mean premium for the taste review and type were $1.28 and $1.02, respectively. However, the origin mean premium ($0.64) in the mixed logit model simulation was much higher than the MNL results. This may suggest that consumers value tomato attributes differently.

After the choice experiment, 75% of the participants indicated they found the taste review helpful in making the decision about which product they chose during the experiment. At the same time, the higher taste score was rated only second to fresh tomatoes as a way to encourage a consumer to pay more for tomato juice. Therefore, consumers have a high demand for tomato juice with a good taste and would be willing to pay a premium for such a product. Specifically, good taste may be linked to fresh tomato notes, which most processed tomato juice lacks.

Summary

Overall, this research suggests that consumers would be willing to pay a premium price for tomato juice with high sensory quality. However, some effort is needed to improve the sensory quality of the tomato juice currently on the market. Most tomato juice consumers are concerned with health and there is a link with higher education and/or income. Perhaps not surprisingly, a bad taste discouraged consumers’

purchase intent and willingness to pay; consumers demand a premium quality tomato juice with a fresh note and better taste such as that is provided by the Garden Gem tomato. Consumers would be willing to pay more if trusted taste review information was provided on the label. This information is a signal of better taste and freshness, and it could encourage consumer purchase with less fear of a sensory quality penalty after tasting. Consumers may be aware of the type of tomato juice and have a generally positive impression of Florida tomatoes. In order to promote the consumption and purchase of tomato juice, improving the quality and then communicating information about improved quality is key to increasing consumer willingness to pay.

100

Table 4-1. Profile analysis of the segments (Season 1) of sensory evaluation participants according to descriptive variables(N=119) Variables S1(%) S2(%) S3(%) Total(%) 36.13% 39.5% 24.37% Gender Male 46.51% 40.43% 37.93% 42% Female 53.49% 59.57% 62.07% 58% Age <18 0.00% 2.13% 0.00% 0.8% 18-29 69.77% 70.21% 65.52% 68.9% 30-44 13.95% 14.89% 13.79% 14.3% 45-65 16.28% 12.77% 20.69% 16.0% >65 0.00% 0.00% 0.00% 0.0% Income <14,999 20.93% 19.15% 27.59% 21.8% $15,000-$24,999 25.58% 36.17% 27.59% 30.3% $25,000-$34,999 6.98% 12.77% 17.24% 11.8% $35,000-$49,999 16.28% 8.51% 3.45% 10.1% $50,000-$74,999 18.60% 10.64% 10.34% 13.4% $75,000-$99,999 6.98% 4.26% 3.45% 5.0% >$100,000 4.65% 8.51% 10.34% 7.6% Education Less than high school 0.00% 8.51% 0.00% 3.4% High school/GED 13.95% 8.51% 10.34% 10.9% Some college 11.63% 10.64% 6.90% 10.1% 2-year college 4.65% 12.77% 0.00% 6.7% 4-year college 27.91% 42.55% 51.72% 39.5% Postgraduate 34.88% 10.64% 20.69% 21.8% Doctoral degree 6.98% 6.38% 10.34% 7.6% Florida tomato opinion*** Less pesticides 6.98% 8.51% 20.69% 10.9% Better tasting 13.95% 21.28% 20.69% 18.5% Nutritious 11.63% 8.51% 3.45% 8.4% Envrionmentally 12.6% friendly 13.95% 12.77% 10.34% Safer 9.30% 8.51% 6.90% 8.4% Premium quality 11.63% 14.89% 6.90% 11.8% None of the above 32.56% 25.53% 31.03% 29.4% **Significant differences among segments at 95% ***Which of the following statements about “Florida tomatoes” do you agree with? (select all that apply)?

101

Table 4-2. Profile analysis of the segments (Season 2) of sensory evaluation participants according to descriptive variables(N=119) Variables S1(%) S2(%) S3(%) Total(%) 47.06% 26.05% 26.89% Gender Male 48.21% 38.71% 34.38% 42.0% Female 51.79% 61.29% 65.63% 58.0% Age <18 0.00% 3.23% 0.00% 0.8% 18-29 78.57% 70.97% 71.88% 74.8% 30-44 10.71% 6.45% 12.50% 10.1% 45-65 10.71% 19.35% 12.50% 13.4% >65 0.00% 0.00% 3.13% 0.8% Income <14,999 33.93% 32.26% 21.88% 30.3% $15,000-$24,999 21.43% 19.35% 43.75% 26.9% $25,000-$34,999 8.93% 19.35% 12.50% 12.6% $35,000-$49,999 12.50% 9.68% 9.38% 10.9% $50,000-$74,999 8.93% 9.68% 3.13% 7.6% $75,000-$99,999 8.93% 6.45% 6.25% 7.6% >$100,000 5.36% 3.23% 3.13% 4.2% Education** Less than high school 0.00% 6.45% 0.00% 1.7% High school/GED 7.14% 9.68% 18.75% 10.9% Some college 10.71% 9.68% 21.88% 13.4% 2-year college 1.79% 12.90% 6.25% 5.9% 4-year college 35.71% 32.26% 21.88% 31.1% Postgraduate 33.93% 22.58% 18.75% 26.9% Doctoral degree 10.71% 6.45% 12.50% 10.1% Florida tomato opinion*** Less pesticides 12.50% 12.90% 9.38% 11.8% Better tasting 21.43% 16.13% 28.13% 21.8% Nutritious 3.57% 6.45% 3.13% 4.2% Environmentally 14.29% 3.23% 9.38% 10.1% friendly Safer 3.57% 6.45% 3.13% 4.2% Premium quality 17.86% 22.58% 3.13% 15.1% None of the above 26.79% 32.26% 43.75% 32.8% **Significant differences among segments at 95% ***Which of the following statements about “Florida tomatoes” do you agree with? (select all that apply)?

102

Table 4-3. Purchase intent difference of sensory testing panelists before and after sensory analysis (Season 1). Before Taste After Taste Difference Garden Gem 3.34a 3.1a -0.24b Roma 2.83b 2.3c -0.29ab Commercial 3.48a 2.7b -0.50a *1-Definitely would not buy, 5-Definitely would buy

Table 4-4. Purchase intent difference of sensory testing panelists before and after sensory analysis (Season 2). Before Taste After Taste Difference Garden Gem 3.13a 2.98a -0.15c Roma 2.66b 2.23b -0.43b Commercial 3.32a 2.42b -0.90a *1-Definitely would not buy, 5-Definitely would buy

Table 4-5. Name your own price of sensory testing panelists before and after sensory analysis (Season 1). Before Taste After Taste Difference Garden Gem 2.89a 2.64a -0.24b Roma 2.39b 2.10b -0.29ab Commercial 2.77a 2.27a -0.50a

Table 4-6. Name your own price of sensory testing panelists before and after sensory analysis (Season 2). Before Taste After Taste Difference Garden Gem 2.62a 2.46a -0.16c Roma 2.34b 1.86b -0.48b Commercial 2.75a 2.05b -0.70a

103

Table 4-7. Characteristics preventing consumer purchase of tomato juice (N=238). Strongly Somewhat Somewhat Strongly Variable disagree disagree Neutral agree agree Price 0.86% 10.34% 18.10% 41.38% 29.31% Sugar added 7.26% 14.96% 26.92% 33.33% 17.52% Salt added 7.30% 16.31% 29.18% 28.76% 18.45% Lack of freshness 2.59% 6.90% 12.50% 31.90% 46.12% Flavor enhancer added 5.60% 13.36% 28.02% 27.59% 25.43% Microbial 4.27% 7.69% 17.95% 24.36% 45.73% contamination concern Do not like the texture 2.56% 7.69% 17.95% 44.87% 26.92% No habit of consumption 3.43% 9.01% 36.91% 25.32% 25.32% Lack of nutrients 4.70% 15.38% 32.05% 30.34% 17.52% Bad taste 4.27% 4.70% 9.40% 20.09% 61.54%

*What do you consider to be the most important attributes that may prevent your purchase of tomato juice?

Table 4-8. Occasions and frequency for tomato juice consumption (N=238). Variable Never Rarely Sometimes Often Always Airplane 48.28% 20.26% 18.53% 8.19% 4.74% Restaurant 35.22% 29.57% 22.61% 8.26% 4.35% Bar 31.76% 27.47% 30.04% 6.87% 3.86% Homemade blend with other 20.09% 21.37% 34.19% 15.81% 8.55% Grocery store 12.82% 19.66% 34.19% 18.80% 14.53% Homemade blend tomato 43.59% 16.67% 18.80% 11.54% 9.40% only

104

Table 4-9. Factor scores for sensory analysis (Season 1 and Season 2) segmentation according to tomato juice consumption opinions. Variables F1 F2 F3 Freshness 0.305 0.012 0.388* Nutrition and health benefit 0.444* 0.069 0.353 Sourness -0.088 -0.612* 0.172 Saltiness -0.133 -0.682* 0.077 Sweetness -0.063 -0.527* 0.245 Safety 0.286* -0.045 0.153 Tomato flavor 0.114 -0.097 0.328* Color 0.318* -0.217 -0.173 Type 0.396* -0.112 -0.173 Origin 0.608* -0.062 -0.057 Organic 0.707* 0.126 0.031 Additive free 0.657* 0.109 0.159 Texture 0.105 -0.341* -0.264 Brand 0.479* -0.283 -0.394 Calories 0.376* -0.071 -0.177 *Denotes the highest factor score for the variable

105

Table 4-10. Descriptive statistics of the online survey participants (N=983). Variable Group Percentage Gender Female 53.61% Male 46.39% Family size 1-3 65.62% 4 or more 34.38% Children under 18 0-2 74.67% 2 or more 27.36% Income level <$15,000 7.83% $15,000-$24,999 8.24% $25,000-$34,999 11.60% $35,000-$49,999 11.39% $50,000-$74,999 20.24% $75,000-$99,999 19.74% $100,000-$149,999 13.73% $150,000 or over 7.02% Age (years) 18-34 24.82% 35-64 54.22% 65 and above 20.96% Race White/Caucasian 83.47% African American 4.23% Hispanic 4.49% Asian 3.00% Other 4.81% Education Less than high school 1.53% High school/GED 14.95% Some college 21.26% 2-year college 8.55% 4-year college 32.96% Postgraduate 20.65% Region Northeast 22.99% Midwest 19.02% South 33.57% West 24.11% Other 0.20%

106

Table 4-11. Health and lifestyle information related to tomato juice consumption of online survey participants (N=983). Variable Group Percentage

Diet Yes 48.12% No 51.88%

Exercise Yes 59.72% No 40.28% Diabetes Yourself or family 26.57% No 73.43% High cholesterol Yourself or family 51.42% No 52.42% Overweight Yourself or family 55.23% No 44.77% Health information look up* Do not look for information 23.91% Less than once a week 31.43% once a week 23.60% More than once a week 21.06% Knowledge on tomato 1 24.82% health benefit ** 2 54.22% (1-not as all, 3 20.96% 7-expert) 4 83.47% 5 4.23% 6 4.49% 7 3.00% Last time consumed Last week 32.25% tomato juice Last month 24.42% Last year 18.01% More than a year 19.63% Never drink 5.70%

*Generally, how often do you spend searching for health information related to food and beverages? ** Please indicate how knowledgeable you are about the health benefits of tomato juice.

107

Table 4-12. Occasions and frequency for tomato juice consumption (N=983). Variable Never Rarely Sometimes Often Always Airplane 61.56% 15.01% 11.77% 6.05% 5.62% Restaurant 44.70% 22.73% 16.13% 9.74% 6.71% Bar 48.37% 16.49% 21.48% 9.76% 3.90% Homemade blend with other 40.76% 12.00% 20.54% 17.19% 9.51% Grocery store 7.78% 16.86% 30.05% 24.86% 20.43% Homemade blend tomato only 53.29% 10.79% 13.92% 11.76% 10.25%

108

Table 4-13. Label and tomato origin opinion of tomato juice purchasers (N=983). Variable Group Percentage Aware concentrate label * Yes 59.61% No or do not know 40.39% Origin** Canada 5.48% California, US 28.46% Florida, US 21.84% Italy 8.48% Mexico 3.07% Spain 4.68% Other 2.67% Does not matter 25.32% Impression of Florida Grown using less pesticides 9.31% tomatoes*** Taste better than other 12.11% tomatoes More nutritious 12.41% Environmentally friendly 11.44% Safer 9.66% Premium quality 19.63% Willing to pay more 7.43% Other 1.42% None of the above 16.58% *Are you aware that tomato juice can be labeled as “from concentrate” or “not from concentrate”? **Given the option, would you prefer to purchase tomato juice from a specific origin, if so, please indicate which origin(s) you prefer (select all that apply)? ***Which of the following statements about “Florida tomatoes” do you agree with? (select all that apply)?

109

Table 4-14. Characteristics preventing consumer purchase of tomato juice (N=983). Strongly Somewhat Somewhat Strongly Variable* disagree disagree Neutral agree agree Price 4.80% 8.98% 24.08% 36.43% 25.71% Sugar added 4.28% 8.46% 26.91% 31.50% 28.85% Salt added 5.39% 10.99% 30.42% 30.32% 22.89% Lack of freshness 3.06% 4.08% 20.49% 32.52% 39.86% Flavor enhancer added 4.07% 9.77% 28.69% 32.76% 24.72% Microbial contamination concern 4.59% 5.91% 22.94% 24.36% 42.20% Do not like the texture 6.31% 10.59% 25.15% 32.48% 25.46% No habit of consumption 9.28% 9.79% 42.20% 21.81% 16.92% Lack of nutrients 11.46% 11.26% 29.99% 25.90% 21.39% Bad taste 3.78% 4.59% 13.67% 22.04% 55.92%

*What do you consider to be the most important attributes that may prevent your purchase of tomato juice?

110

Table 4-15. Multinomial logit model estimates for respondents. Original Adjusted* Number of respondents 983 909 Number of observations 14745 13635 Log likelihood -14597 -12999 Likelihood ratio 3204.8 3962.2 Price -0.48 -0.52 Taste score 0.64 0.68 Origin 0.33 0.35 Type 0.49 0.54 None 2.01 1.83 Adjusted* denotes remove the seventy-four participants choose “would not buy either one” for all fifteen choice experiment questions

Table 4-16. Coefficient estimates in different willingness to pay models (N=909). Multinomial Logit Model Mixed Logit Model Fixed mean Mean Standard Deviation Price -0.52*** -0.58*** - Taste score 0.68*** 0.74*** 0.41*** Origin 0.35*** 0.37*** 0.21 Type 0.54*** 0.59*** -0.15 None 1.83*** 1.51*** - **Denotes coefficient parameters that are significant at p<0.05 level ***Denotes coefficient parameters that are significant at p<0.001 level

Table 4-17. Premiums for tomato juice using Krinsky-Robb bootstrapping simulation. Premiums Mean Standard 95 % confidence interval 90% confidence interval deviation Taste score 1.28 0.71 -0.11 to 2.67 0.12 to 2.44 Origin 0.64 0.36 -0.07 to 1.35 0.05 to 1.23 Type 1.02 0.26 0.13 to 1.15 0.21 to 1.07

111

Figure 4-1. Example of choice set

112

CHAPTER 5 TOMATO ESSENCE CREATION AND CHARACTERIZATION

Abstract

Despite a high nutrient content, processed tomatoes are among the least liked processed fruit products. The undesirable flavor quality incurred by the change from fresh to processed products has long been a major issue for consumers and there is thus a need to develop solutions for enhancing the flavor profile of processed tomato products. The objective of this research was therefore to: 1) observe how flavor compounds change during thermal processing; 2) create and characterize tomato essences; 3) determine how an essence differs from a fresh tomato’s flavor.

The volatile compounds in juices from two tomato varieties, Garden Gem and

Roma, and the essences created for this study were characterized using purge and trap gas chromatography-mass spectrometry (PT-GC-MS). Identification was conducted using an MS-NIST library match and retention indices; semi-quantification was carried out using internal standards. Four essence fractions were created from a high-quality plum tomato variety (Garden Gem).

Upon thermal processing, tomato juice made from both varieties suffered significant losses in their green note and fruity note related compounds. Among the four essence fractions created, Fraction 1 and Fraction 2 were most complementary to the undesirable flavor changes typically encountered during thermal processing (loss of freshness and presence of cooked flavors). Fraction 1 was characterized as “green tomato note” with substantially high amounts of 2-isobutylthiazole, hexanal, 3-hexanal,

3-hexen-1-ol, 1-penten-3-ol and 6-methyl-5-hepten-2-one. Fraction 2 was described as

“fruity tomato note” with the balanced compounds mentioned above plus linalool, α-

113

citral, β-citral, β-cyclocitral, β-damascenone, cis-geranylacetone, phenylethyl alcohol, and benzaldehyde. Fraction 3 and Fraction 4 were described as “cooked tomato note” and “faint tomato note”, respectively. Fraction 3 consisted of excessive amounts of geranylacetone, benzaldehyde, and relatively low amounts of hexanal. Most of the compounds in Fraction 4 were reduced in concentration by more than 50% when compared to those in Fraction 3, and by approximately 10% compared to Fraction 1.

This research is the first known report of the creation of a tomato essence. The results demonstrate the feasibility of adding tomato essence to impart desirable flavor attributes to processed tomato products.

Background Information and Objectives

The undesirable change in flavor quality from fresh to processed tomato juice

(pasteurized / concentrate) has long been a major issue. These undesirable changes have been reported in tomato products (juice, puree, paste) due to the flavor profile change (Buttery, 1990; Chung, 1983). Efforts have been made to resolve this problem, for example by adding important fruity and green related flavors to tomato products to achieve a balanced profile. In 1991, Buttery patented a fresh tomato flavor mixture that could be added to processed tomato products that consisted of 2-methylhept-2-en-6- one, cis-3-hexenal, eugenol and β-ionone. However, the source of these flavor chemicals must also be considered. If these compounds are isolated in relatively pure form from natural ingredients, a relative higher cost may result even for a large scale operation. Also, if diluted food grade synthetic chemicals are added to a product, then the products may no longer carry a “clean label”.

An alternative way to maintain the “clean label” designation in a cost-efficient way is to add natural flavor enhancement ingredients or enzymes to help produce desirable

114

flavors. To enhance the fruity flavor of tomato products, soybean lipoxygenase, linoleic acids, and β-carotene were added to tomato pulp, resulting in an increase in both β- cyclocitral (10 ppb) and β-ionone (10 ppb) (Deutz, Dunphy, & Van, 2003). However, linoleic acid can undergo lipid oxidation and other enzymatic reactions, producing an artifact that negatively affects the overall flavor of the product in an unexpected way.

Therefore, one needs to consider carefully not only the flavor compounds or precursors added but also the potential issues that can arise in a given food system. This needs to be dealt with on a case by case basis since different food systems contain different desirable or undesirable flavors.

There is clearly a need for solutions that enhance the flavor profile of juice products with no artifacts and without significantly changing the labeling of the product.

Essences are water-soluble, clear liquids that are produced at the pre-heater stage of the evaporator during the juice concentration process.

Essence production has served as an excellent means of characterizing the flavor of fresh citrus, apples, berries (strawberry, raspberry, cherry, and blackberry), peaches, pears, plums, pumpkins, mangoes and Concord grapes. Orange essence has been one of the most widely used essences in the food industry since the early 1960s

(Attaway, Wolford, & Edwards, 1962). Orange essence was first collected from saturated vapor during juice processing. This steam was then condensed and used to form a concentrated mixture that is both aqueous and oil-soluble, rich in alcohols, aldehydes, ketones, and esters (Wolford, Alberding, & Attaway, 1962). The water insoluble portion, the so-called “essence oil”, is further separated from the water-soluble portion by centrifugation. The water-soluble portion of the orange that remains is called

115

the “essence” and has been found to contribute a fresh, fruity, orange aroma (Ringblom,

2004). The major components in essence are ethanol, acetaldehyde, and D-limonene

(Moshonas & Shaw, 1983). However, as yet there has been no similar attempt to fraction an essence to enhance tomato flavor.

The research presented in this chapter proposes a “tomato essence” for tomato juice flavor enhancement. The flavor profile change was first examined during tomato processing using instrumental analysis; fractions of tomato essence were characterized, and differences between the fresh samples and the fractions were examined.

Materials and Methods

Tomatoes

Fully ripe Garden Gem tomatoes were harvested in Fall 2015 from a greenhouse

(Gainesville, FL). Ripe Roma tomatoes were purchased from a local Publix

(Gainesville, FL) and Winn-Dixie (Gainesville, FL) grocery stores in Fall 2015.

Juice Processing

Twenty pounds of Garden Gem and Roma tomatoes were each cut into three pieces by toss and chop kitchen scissors (Comfify, Fernadale, WA, USA). Small tomato pieces were fed into a tomato electric milling machine (Model 2400, O.M.R.A. MTD,

Brooklyn, NY) to remove the skin and seeds in 15 s. The puree produced was continuously fed into a steam kettle and heated (a process called “hot break”) at 95ºC for 3 min. Then the hot break juice was poured into a sanitized carboy. Later the juice was pasteurized (Model 5024-F 25HV, Microthermics, Raleigh, NC, USA) at 73ºC for 30 s and stored in another newly sanitized carboy at 4ºC. The whole process was repeated three times for each variety.

116

Fresh Juice Production

As in the juice processing procedure, twenty pounds of tomatoes for each cultivar were cut into three pieces by toss and chop kitchen scissors. Small tomato pieces were fed into the tomato electric milling machine to remove skin and seeds in 15 s. The puree produced was mixed with food grade saturated calcium chloride (Modernist Pantry,

Portsmouth, NH, USA) at a 3:1 ratio and stored at -20ºC in a 1-gallon sample bag.

Essence Production

Fresh tomato juice was distilled promptly under vacuum using a rotary evaporator (Buchi R-100, Flawil, Switzerland) to give four chilled condensate (4ºC) fractions. A 200 mL sample of fresh tomato juice was distilled at 35ºC in a water bath under negative pressure to give 160 mL of essence in total (40 mL for each fraction).

Each fraction was collected in a timely manner. The evaporator rotation speed was 200 rpm. Four individuals categorized the fractions into general note classifications via olfactory analysis.

Flavor Isolation

Tomato juice treated with or without heat was filtered by glass microfiber binder free paper GF/CTM (Whatman, Pittsburgh, PA, USA) using a Buchner vacuum filtration set. The liquid was collected and placed in a 40 mL glass vial. Then 3 µL of an antifoaming reagent (Trans-400, Bristol, WI, USA) was added and mixed well. Carvone

(150 ppb) was added into each sample vial as the internal standard. The tomato essence skipped these pre-treatment process steps, and instead was placed directly in a vial. Flavor extraction was performed using a Stratum purge and trap unit (Teledyne

Tekmar, Mason, OH, USA). A total of 5 mL sample was pulled by autosampler and purged with a nitrogen flow at 200 mL/min for 30 min. The volatile compounds were

117

adsorbed by a trap containing Tenax® and desorbed at 180ºC, then sent to the GC injection port via transfer line.

Identification and Quantification

Identification was conducted on an Agilent (Santa Clara, CA, USA) gas chromatograph equipped with quadrupole mass spectrometer detector (5975 MSD). A

ZB-WAXplus (Phenomenex, Torrance, CA, USA) column (30 m x 250 µm x 0.25 µm) was used for separation. The GC oven temperature was held at 45ºC for 7 min, then the temperature was ramped to 150ºC at 3ºC/min. Continued ramping was applied at a rate of 10ºC/min until 210ºC. A 30ºC/min rate increase in oven temperature was further applied to 240ºC. The column was then held at 240ºC for 4 min. Helium gas flow rate was 1 mL/min. The MSD used a scan mode to detect compounds in the range of 45-

300 m/z. The peaks were identified by a mass spectrometer comparison with the 2011

NIST library. The retention indices were calculated using a C5-C20 mixture of linear alkane chemical standards (Sigma-Aldrich, St. Louis, MO, USA).

Statistics

All collected flavor chemical data was subjected to a 1-way analysis of variance

(ANOVA) using SAS 9.4 software package (SAS Institue Inc, NC, USA); differences were considered significant at an alpha level of less than or equal to 0.05.

Results and Discussion

Flavor Profile Change after Processing

Overall, the flavor compounds in tomato juice changed greatly after thermal processing. The post-thermal processing flavor profile changes for Garden Gem (GG) and Roma are given in Tables 5-1 and 5-2. Both the GG and Roma juice had 38 compounds that were observed to be statistically different by 1-way ANOVA analysis

118

(p<0.05) after thermal processing. This result demonstrates that thermal processing significantly influences the volatile profile of tomato juice, and hence also likely the sensory perception of the final product. Both the GG and Roma fractions lost flavor compounds during thermal processing. However, in general, the GG variety maintained more vital flavor chemicals through the course of thermal processing.

A group of volatiles associated with “green” odor are considered “green note compounds” that contribute to tomato flavor. The C6 alcohols and aldehydes are the most studied green note compounds (Goodman, Fawcett, & Barringer, 2002; Klee,

2010; Stone, 1975). The C6 volatiles are likely formed from the disruption of cell walls during homogenization or juicing because enzymes and substrates are released from the cells through the action of grinding and become reactive during this step (Goodman,

2002). As the cell wall breaks up mechanically or thermally, the amino acids and unsaturated fatty acids that make up the cellular membranes are susceptible to oxidation catalyzed by enzymes in the cell. Hexanal and 3-hexenal are known to be formed through the action of 13-lipoxygenase and 13-hydroperoxide lyase acting upon linoleic acid and linolenic acid, respectively (Klee, 2010), and then hexanol and 3- hexenol are formed from alcohol oxidoreductase of hexanal and 3-hexenal (Stone,

1975). In this study, most of the green related notes (hexanal, 3-hexenal, 1-hexanol, 3- hexen-1-ol, 2-octenal, 2-isobutylthiazole) were observed in significantly lower amounts after thermal processing. Hexanal, 3-hexenal, 1-hexanol, 3-hexen-1-ol, 2- isobutylthiazole were reduced by roughly 80% after thermal processing (Tables 5-1, 5-

2). The reduction was expected and is in accordance with previous studies. Goodman et al. (2002) reported significantly lower total peak areas of green note compounds

119

(hexanal, 3-hexenal, 2-hexenal, 3-hexenol and hexanol) in 93ºC treated juice compared to those in 70ºC treated juice. Heat plays a key role in the inactivation of lipoxygenase and hydroperoxytrienoic lyase (Anthon & Barrett, 2003). Therefore, heat treatment leads to limited generation of C-6 volatiles. Buttery et al. (1990) also reported a remarkable loss of green note compounds in heat-processed tomato paste compared to fresh juice.

In another study, as the processing temperature increased, the low-boiling volatiles such as hexanal, heptanal were found to decrease in thermally processed tomato juice

(Chung, 1983). When heat is applied to the juice, relatively lower boiling point C-6 alcohols and aldehyde evaporate quickly. The partition coefficient of volatiles also changes in the headspace with the addition of heat. The compound 2-isobutylthiazole has been described as a non-C6 origin green note attributor and patented for use as a tomato flavor enhancer (Christiansen et al., 2011).

Despite these undesirable losses of important flavor compounds, it is both common and necessary to use heat treatment to hot break (for inactivating enzymes) and pasteurize tomato juice. The heat applied during tomato juice processing helps to maintain a desirable texture by inactivating pectinmethylesterase (PME) and endopolygalactuonase (EPG), both of which degrade pectin and adversely affect juice texture (Hayes, Smith, & Morris, 1998).

Many volatiles are described as having “fruity/floral” attributes. Compared to

“green note” volatiles, the “fruity note” compounds are commonly found only in low abundance in both watermelons and tomatoes (Lewinsohn et al., 2005). However, much lower odor thresholds have been observed for many fruity related compounds

(Teranishi, Buttery, Stern, & Takeoka, 1991). Fruity flavor notes have lower human

120

perception odor thresholds than green notes, so even when the concentration of fruity note compounds is low, the perception of these compounds is strong due to the ability of humans to detect these compounds at low levels. More importantly, the fruity related volatiles enhance sweetness perceptions in fresh tomatoes (Baldwin, Goodner, &

Plotto, 2008). These fruity note compounds are mainly carotenoid-derived apocarotenoid volatiles such as geranylacetone, 6-methyl-5-hepten-2-one (MHO), β- ionone and β-damascenone. These apocarotenoid volatiles are formed either by enzymatic (carotenoid cleavage dioxygenases) or non-enzymatic oxidative cleavage of carotenoids (linear/cyclic) (Vogel, Tieman, Sims, Odabasi, Clark, & Klee, 2010).

In this research, most carotenoid related compounds (6-methyl-5-hepten-one, β- citral, geranylacetone and β-ionone) suffered significant loss after processing. The 6- methyl-5-hepten-one and β-citral were found to be reduced by approximately 60% and

90%, respectively (Tables 5-1 and 5-2). The compound 6-methyl-5-hepten-one has been reported as “fruit-like” and “sweet” (Kazeniac & Hall, 1970), while the compound β- ionone has been described as “fruity floral” and “sweet” (Tandon, Jordan, Goodner, &

Baldwin, 2001). β-citral is known to be largely derived from the carotenoid pathway and linked to “citrus” and “fruity” (Alonso et al., 2009; Tandon, Jordan, Goodner, & Baldwin,

2001).

Surprisingly, several compounds that contribute positively to tomato aroma were found to increase under thermal processing conditions. In particular, the compounds linalool and β-damascenone were observed to increase significantly. These two compounds have been described as “fruity” (Baldwin, Goodner, Plotto, Pritchett, &

Einstein, 2004; Pineau, Barbe, Van Leeuwen, & Dubourdieu, 2007). Both linalool and β-

121

damascenone are likely to be carotenoid pathway degradation products and have been found in fresh and processed tomato products (Landy, 2002; Vallverdu-Queralt, 2013).

In fresh treatments, linalool and β-damascenone are likely generated from the enzymatic action on carotenoids. In processed tomato products (GG and Roma), heat may have catalyzed the action of the carotenoid degrading enzymes, leading to increased yields of linalool and β-damascenone (Tables 5-1 and 5-2). Similarly, phenylalanine degradation products (benzaldehyde, benzenacetaldehyde, 2,6- dimethylbenzaldehyde) were found to increase in concentration; however, the increase in the concentration of benzaldehyde, benzenacetaldehyde would not be expected to be nearly as important because of the high odor threshold of benzeneacetaldehyde (Table

5-1).

In contrast to the increase in desirable “fruity” notes, a number of “cooked flavors” were detected or were present to a greater extent as heat was applied to tomato juice. Most of the “cooked” compounds were sulfur-containing (methyl propyl- sulfide, methional) or a Maillard reaction product (furfural). Methyl propyl-sulfide was reported as “cooked” in fruit and vegetable products such as onion and garlic (Rowe,

1998; Villière , Le Roy, Fillonneau, Guillet, Falquerho, Boussely, et al., 2015).

Methional was characterized as cooked or baked potato or potato pungent aroma, and it was considered an essential constituent of fresh tomato aroma (Mayer, Takeoka,

Buttery, Whitehand, Naim, & Rabinowitch, 2008). Furfural has also been found in tomatoes subjected to heat treatment (Servili, Selvaggini, Taticchi, Begliomini, &

Montedoro, 2000). It has been suggested that the “cooked” aroma of furfural is generated by the Maillard reaction in the absence of sulfur-containing amino acids

122

(Mottram, 2007). However, the amount of furfural in the processed tomato juice was below odor threshold so it is unlikely that furfural would affect sensory perception in a processed tomato product.

Flavor Profile of Tomato Essence

As the composition results of tomato essence in Table 5-3 demonstrate, most of the green note compounds (hexanal, 3-hexenal, 1-hexanol, 3-hexen-1-ol, 2-octenal and

2-isobutylthiazole) and fruity note compounds discussed previously were observed in similar or higher concentration to those in fresh tomato juice. The compounds found in the essence demonstrate a complementary effect on the desirable note loss during processing (Tables 5-1 and 5-2). The addition of an essence may thus offer a way to rebuild the fresh/fruity aroma of processed tomato juice. More importantly, the tomato essence yielded substantial amounts of fruity and green note that may further intensify the desired sweet/floral/green aroma characteristics.

The volatile compounds observed in the essence at relatively high concentrations were 6-methyl-5-hepten-one (1793 ppb), α-citral (766.8 ppb), 3-hexen-1-ol (710 ppb), 2- isobutylthaziole (354.2 ppb) and geranylacetone (317.9 ppb). Compared to fresh tomato juice made from GG, a 100-fold, 15-fold, 10-fold increase of geranylactone, α-citral and

β-citral, respectively, were observed. The most potent compounds detected in the essence were β-ionone (60.14 ppb), β-damascenone (9.31 ppb), and acetone (70.95 ppb), which were 8590, 4654 and 236–fold higher than their respective odor thresholds.

Most of the compounds mentioned above contributed to “sweet”, “fruity”, and “floral” notes except for green-note related 3-hexen-1-ol and 2-isobutylthaziole.

Geranylacetone and 6-methyl-5- hepten-2-one were likely derived from acyclic carotenoids (lycopene or phytofluene), while α-citral and β-citral were derived from δ-

123

carotene and β-carotene, respectively (Lewinsohn et al., 2005). Unlike C-6 compounds

(green-note related 3-hexen-1-ol and 2-isobutylthaziole), the carotenoid pathway products are known to be affected by both the oxidation and enzymatic release of glycosides (Parker, Elmore, & Methven, 2014).

Compared to fresh tomatoes, some important green note related compounds were observed in higher concentrations in the tomato essence. These compounds included hexanal, 2-hexenal, 3-hexenal. 3-hexen-1-ol, and 1-hexanol. The 3-hexenal has a relatively lower odor threshold compared with its isomers (hexanal, 2-hexenal); and 3-hexenal was detected at levels two-fold higher in the essence compared with fresh tomato. When heat is applied, 3-hexen-1-ol might form from Z-3-hexenal via NAD oxidoreductase activity; Stone et al. (1975) reported the formation of 3-hexanal (as the major C-6 compound) via NAD oxidoreductase at 42°C during vacuum distillation of fresh tomatoes. Similarly, 1-hexanol is a reduction product of hexenal. The green related note 1-hexanol, was found at levels approximately six times higher in the essence compared to fresh, unprocessed tomato juice. Although a significant concentration of 1-hexanol was observed at 279 ppb, this may only contribute in a minor way to the green aroma due to its concentration being barely above the odor threshold.

Compounds at concentrations above the human odor threshold observed in the tomato essence were β-ionone, β-damascenone, 1-pentanol, acetone, 1-penten-3-ol, 3- hexenal, 2-isobutylthiazole, hexanal, 6-methyl-5-hepten-2-one, 1-penten-3-one, 2- hexenal, β-citral, linalool, geranylacetone, 2-octenal, 1-octen-3-ol and β-cyclocitral.

Seven of these compounds were used by Buttery (1993) to replicate fresh tomato aroma. Since this essence was created at a relatively low temperature (35ºC), the

124

similarity of the flavor profile between the essence and fresh tomatoes was expected.

The most potent compound observed in essence is β-ionone, which is present at about

8500 times higher than its odor threshold value (0.007 ppb).

Two compounds, ψ-ionone and ethyl sorbate (2,4-hexadienoic acid, ethyl ester), were observed only in the tomato essence and were not detected in either the fresh or processed tomato juice samples made from GG. However, a previous study (Buttery,

Seifert, Guadagni, & Ling, 1971) identified ψ-ionone as a component of California tomatoes (VF-145) after hexane extraction. Ψ-Ionone is cataloged as a “sweet” and

“juicy” flavor in the handbook of flavor ingredients (Burdock, 2016). Ψ-Ionone may be derived from δ-carotene and ξ-carotene (Lewinsohn et al., 2005) or aldol condensation of citral with acetone (Selli, Kelebek, Ayseli, & Tokbas, 2014). Since the tomato juicer removed the skin and seeds, breaking down the tomato tissue in the process, carotenoid cleavage dioxygenases (LeCCD1 A and LeCCD 1b) (Simkin, Schwartz,

Auldridge, Taylor, & Klee, 2004) and substrates in the tomato were likely mixed thoroughly. The combination of these factors would be expected to favor the cleavage of linear carotenoids (Vogel, 2010) or aldol condensation. Ψ-ionone was detected at a concentration (both in cis and trans form) of 118 ppb; this concentration is thousands of times higher than the odor threshold of β-ionone (0.007 ppb); The odor threshold comparison information between β-ionone and Ψ-ionone is currently unavailable so it is unclear how this concentration may affect human perceptions of Ψ-ionone. Ethyl sorbate has been described as a “warm/fruity odor” and has been observed in orange essence oil (Coleman, Lund, & Moshonas, 1969). Ethyl sorbate is quite stable and has

125

been widely used as a synthetic flavoring in beverages and ice cream (Hui & Evranuz,

2012).

The only compound in the essence that showed a slight decrease in concentration was 3-methyl-butanal, which has been characterized as a “malty” flavor

(Yilmaz, 2001). However, the 2-methyl-butanal (green, pungent note) was observed in relatively higher abundance. Besides its natural appearance in fresh tomatoes, the isomerization of 3-methyl-butanal to 2-methyl-butanal might have occurred under the experimental conditions of this study, which may explain its presence.

Among the four essence fractions (Figure 5-1), Fraction 1 and Fraction 2 are the most complementary to the thermally treated juice. Fraction 1 was characterized as a

“green tomato note” by smell with a substantial amount of 2-isobutylthiazole, hexanal, 3- hexanal, 3-hexen-1-ol, 1-penten-3-ol and “fruity note” related 6-methyl-5-hepten-2-one.

Fraction 2 was characterized as a “fruity tomato note” (linalool, α-citral, β-citral, β- cyclocitral, β-damascenone, geranylacetone, phenylethyl alcohol and benzaldehyde), with slightly less green note character. It is feasible that Fractions 1 and 2 could be added to a thermally treated tomato product to rebalance the product by supplying the green and fruity aroma notes that these products typically lack. Fraction 3 and Fraction

4 were described as “cooked tomato note” and “faint tomato note”, respectively.

Fraction 3 consist of excessive geranylacetone, benzaldehyde and relatively low amounts of hexanal; while Fraction 4 was reduced more than 50% in most compounds compared to Fraction 3.

A total of seventeen, twelve, eleven and eight compounds with odor active values greater than one were observed in Fractions 1 through 4, respectively (Table 5-4). All

126

four fractions contained β-damascenone, β-ionone and 1-penten-3-ol as the top three components with the highest odor-active values (OAV). Although the threshold of one compound may vary from person to person, OAV has been used extensively to access the aroma potency of volatile compounds. The OAV of several compounds in Fractions

2, 3 and 4 were below 1. This indicates that a human might not be able to perceive the flavor note of these compounds, thus altering the overall profile of the fractions. Other than in Fraction 1, the OAV of 2-octenal, 1-octen-3-ol, β-cyclocitral and β- geranylacetone in Fraction 2 was below 1, indicating that these compounds may be less likely perceived by humans. Meanwhile, the major fruity related potent compounds (the top three odors plus β-citral) retained relatively high OAV values across all fractions.

Some green related compounds: 1-penten-3-ol, 2-isobutylthiaole, hexanal, 2-hexenal decreased considerably in Fraction 2, and fell even further in the later fractions. This could partially explain Fraction 2’s major green-flavor reduction and relatively stable fruity notes. In Fraction 3, both the fruity and green related aroma active notes were further diminished, leaving the “cooked notes” to take center stage. Interestingly, slight increases in 2-isobutylthiazole and geranyl acetone were observed in Fraction 3 compared to Fraction 2. These two compounds may be formed due to thermal degradation of tomato glycosides. The composition of the “cooked note” in Fraction 3 is surprisingly different from the sulfur-containing/Millard reaction compounds described in

Tables 5-1 and 5-2. This may suggest that the fruity and green note compounds are essential for a fresh tomato aroma.

Essence Fractions 1 and 2 may be used to recover the flavor lost during processing and enhance the green and fruity aroma of tomato juice since the

127

condensed essence may be able to enrich or balance a more “green” or “fruity” aroma profile. Fractions 1 and 2 both contain a considerable amount of potent green related and fruity related compounds, and hence these essence fractions can be utilized in a diluted form to achieve the desired result.

Summary

Overall, this study demonstrated a clear difference in volatile composition from fresh to thermally processed tomato juice. Most green note related compounds

(hexanal, 3-hexenal, 1-hexanol, 3-hexen-1-ol, 2-octenal, 2-isobutylthiazole) and fruity related compounds (6-methyl-5-hepten-one, β-citral, β-ionone, geranylacetone) were lost due to thermal processing. Sulfur containing (methyl propyl-sulfide, methional) or a

Mailard reaction product (furfural) were detected in the thermally processed juice.

Phenylalanine degradation products (benzaldehyde, benzenacetaldehyde, 2,6- dimethylbenzaldehyde), linalool and β-damascenone were all observed to increase significantly due to thermal processing.

A tomato essence could provide a substantial amount of fruity and green note compounds to enhance the tomato flavor of thermally treated juice. 6-methyl-5-hepten- one (1793 ppb), α-citral (766 ppb), 3-hexen-1-ol (710 ppb), 2-isobutylthaziole (354.2 ppb) and cis-geranylacetone (317.9 ppb) were the most dominant compounds in tomato essence. By further segmenting the essence into four fractions and characterizing the results, Fraction 1 and Fraction 2 were most complementary to the undesirable flavor changes produced during thermal processing. Fraction 3 and Fraction 4 were described as “cooked tomato note” and “faint tomato note”, with less potent odor. The essence of fractions 1 and 2 may be better suited for the recovery of the flavor lost during thermal processing.

128

Table 5-1. Changes in volatile compounds during thermal processing (Garden Gem). RI RT Identification TD Processed Fresh % Change Aroma note (min) (ppb) (ppb) (ppb) 871 2.48 Acetone 40 51.87 50.00 3.7 Fruity 918 2.96 Ethyl Acetate 50 n.d 8.95 - - 929 3.07 Glutaraldehyde - 11.34 n.d - - 939 3.23 Butanal, 3-methyl- 0.2 11.78 28.05 -58.0 Fruity 952 6.63 Disulfide, dimethyl 9.8 1.00 0.46 119 Cooked 1084 6.84 Hexanal 4.5 85.48 320.6 -73.3 Green 1139 9.22 3-Hexenal 2 n.d 188.6 - Green 1168 10.60 1-Penten-3-ol 1 5.05 46.18 -89.1 Fruity/green 1183 11.28 Heptanal - 2.38 1.77 34.3 Green 1213 12.76 2-Hexenal, (E)- 17 22.17 122.3 -81.9 Green 1215 12.87 1-Pentanol 1600 8.47 48.43 -82.5 Green 1253 14.76 Sulfide, methyl propyl - 1.32 n.d - Cooked 1315 17.71 2-Heptenal, (Z)- 13 1.77 6.13 -71.1 Green 1334 18.55 5-Hepten-2-one, 6-methyl- 50 195.5 642.2 -69.6 Fruity 1359 19.71 1-Hexanol 200 12.06 59.57 -79.8 Green 1387 20.99 3-Hexen-1-ol, (Z)- - 35.34 174.1 -79.7 Green 1398 21.52 2-Isobutylthiazole 3.5 4.55 61.31 -92.6 Vine green 1420 22.44 2-Octenal, (E)- - 1.92 4.51 -57.3 Green 1440 23.31 Linalool oxide - n.d 0.30 - - 1445 23.53 Methional 0.2 1.46 n.d - Cooked 1455 24.17 Furfural 3000 5.92 n.d - Cooked 1475 24.54 5-Hepten-2-ol, 6-methyl- - 0.59 15.95 -96.3 Fruity 1482 25.14 2,4-Heptadienal, (E,E)- - 0.71 3.18 -77.6 - 1492 25.55 Decanal - 2.22 0.49 348 - 1494 25.66 1-Hexanol, 2-ethyl- - 2.65 1.91 38.7 Fruity 1507 26.21 Benzaldehyde 3500 71.91 26.41 172 Fruity 1551 27.98 Linalool 6 16.63 6.15 170 Fruity 3,5-Heptadien-2-one, 6- 1563 29.24 380 3.19 4.11 -22.4 - methyl-, (E)- 1583 30.14 β-Cyclocitral 5 3.24 2.26 43.7 Fruity 1605 31.03 Benzeneacetaldehyde 4 3.27 1.84 77.5 Floral 1628 32.53 α-Terpineol 330 1.52 1.52 -0.1 Fruity

129

Table 5-1. Continued RI RT Identification TD Processed Fresh % Change Aroma note (min) (ppb) (ppb) (ppb) 1694 32.64 β-Citral 30 2.43 31.16 -92.2 Citrus, floral 1670 33.74 2,6-Dimethylbenzaldehyde - 0.67 n.d - - 1711 35.85 Methyl salicylate 40 3.39 3.56 -4.9 Green 1758 37.69 β-Damascenone 0.002 2.31 0.95 143 Fruity 1809 39.05 Geranylacetone 60 3.44 3.80 -9.5 Fruity 1849 40.97 Phenylethyl Alcohol 1100 1.92 4.57 -58.0 Fruity 1906 41.70 β-Ionone 0.007 0.59 0.84 -29.6 Floral, sweet % change as calculated as (Processed-Fresh)/Fresh *100% n.d represents not detected; TD represents threshold value in water solution reported by Buttery et al. (1971), Burdock (2016) Aroma note was summarized from Kazeniac & Hall (1970), Rowe (1998), Galliard (1977), Buttery (1993), Yilmaz (2001), Klee (2010) and Burdock (2016) RI of DB-Wax plus column was reported based on C8-C20 alkane standard. Compounds quantified were for fall season fresh and processed tomato juice

130

Table 5-2. Changes in volatile compounds during thermal processing (Roma). RI RT Identification TD Processed Fresh % Change Aroma note (min) (ppb) (ppb) (ppb) 835 2.26 Dimethyl sulfide 0.3 1.68 n.d - Cooked 871 2.48 Acetone 40 36.83 23.25 58.4 Fruity 918 2.96 Furan,2-ethyl - 6.18 1.77 249 - 929 3.07 Glutaraldehyde - 3.19 29.13 -89.0 - 938 3.22 Butanal, 2-methyl- - 15.54 17.30 -10.2 - 939 3.23 Butanal, 3-methyl- 0.2 25.82 45.46 -43.2 Fruity 952 6.63 Disulfide, dimethyl 1.00 3.31 -69.8 Cooked 1084 6.84 Hexanal 4.5 56.22 669.3 -91.6 Green 1139 9.22 3-Hexenal 2 n.d 305.9 - Green 1168 10.60 1-Penten-3-ol 1 5.43 64.36 -91.6 Fruity/Green 1183 11.28 Heptanal - 1.30 7.84 -83.4 Green 1213 12.76 2-Hexenal, (E)- 17 63.8 42.53 50.0 Green 1215 12.87 1-Pentanol 1600 5.38 15.41 -65.1 Green 1253 14.76 Sulfide, methyl propyl - 1.22 n.d - Cooked 1315 17.71 2-Heptenal, (Z)- 13 1.67 10.99 -84.8 Green 1334 18.55 5-Hepten-2-one, 6-methyl- 50 46.22 112.85 -59.0 Fruity 1359 19.71 1-Hexanol 200 6.04 61.11 -90.1 Green 1387 20.99 3-Hexen-1-ol, (Z)- - 25.47 110.3 -76.9 Green 1398 21.52 2-Isobutylthiazole 3.5 n.d 3.91 - Green 1420 22.44 2-Octenal, (E)- 4 1.44 13.88 -89.6 Green 1445 23.53 Methional 0.2 0.96 n.d - Cooked 1455 24.17 Furfural 3000 7.24 n.d - Cooked 1475 24.54 5-Hepten-2-ol, 6-methyl- - n.d 6.70 - Fruity 1482 25.14 2,4-Heptadienal, (E,E)- - 1.11 17.7 -93.7 - 1492 25.55 Decanal - 1.20 n.d - - 1494 25.66 1-Hexanol, 2-ethyl- - 3.58 n.d - - 1507 26.21 Benzaldehyde 3500 9.86 6.26 57.5 Fruity 1551 27.98 Linalool 6 18.21 2.9 528 Fruity 3,5-Heptadien-2-one, 6-methyl-, 1563 29.24 380 3.33 n.d - - (E)- 1583 30.14 β-Cyclocitral 5 n.d 1.81 - Fruity 1605 31.03 Benzeneacetaldehyde 4 3.94 n.d - Fruity 1628 32.53 α-Terpineol 330 1.08 0.60 80.0 Fruity

131

Table 5-2. Continued RI RT Identification TD Processed Fresh % Change Aroma note (min) (ppb) (ppb) (ppb) 1694 32.64 β-Citral 30 0.30 6.55 -95.4 Citrus, floral 1670 33.74 2,6-Dimethylbenzaldehyde - 0.77 n.d - - 1758 37.69 β-Damascenone 0.002 2.20 n.d - Fruity 1809 39.05 Geranylacetone 60 2.83 6.38 -55.6 Fruity 1849 40.97 Phenylethyl Alcohol 1100 1.12 n.d - Fruity 1906 41.70 β-Ionone 0.007 n.d 0.90 - Fruity % change was calculated as (Processed-Fresh)/Fresh *100% n.d represents not detected; TD represents threshold value in water solution reported by Buttery et al. (1971), Burdock (2016) Aroma note was summarized from Kazeniac & Hall (1970), Rowe (1998), Galliard (1977), Buttery (1993), Yilmaz (2001), Klee (2010) and Burdock (2016) RI of DB-Wax plus column was reported based on C8-C20 alkane standard. Compounds quantified were for fall season fresh and processed tomato juice

132

Table 5-3. Important volatile compounds in fresh tomato juice, tomato essence and fractions by GC-MS. RI RT Compounds TD Fresh Essence F1 F2 F3 F4 (min) (ppb) (ppb) (ppb) (ppb) (ppb) (ppb) (ppb) 871 2.47 Acetone 40 47.57 78.35 68.90 4.80 2.60 2.05 918 2.96 Ethyl Acetate 50 8.95 10.78 7.24 1.90 1.14 0.50 939 3.24 Butanal, 3-methyl- 0.2 28.0 12.95 8.24 3.90 0.81 n.d 938 4.11 Butanal, 2-methyl- - 30.0 37.69 24.5 7.82 5.37 n.d 1168 4.98 1-Penten-3-one 1 30.60 31.24 25.02 5.60 0.62 n.d 1084 6.84 Hexanal 4.5 320.0 373.6 213.8 101.9 40.38 17.53 1139 9.22 3-Hexenal 2 188.6 215.8 89.71 81.96 33.74 10.36 1168 10.60 1-Penten-3-ol 1 46.18 150.1 85.71 26.51 26.35 11.49 1213 12.76 2-Hexenal 17 122.3 169.5 88.52 49.86 19.36 11.75 1-Butanol, 2- 1214 12.86 - 65.17 297.4 131.8 96.12 44.91 24.59 methyl- 1215 12.90 1-Pentanol 1600 48.43 279.0 130.4 95.58 50.32 2.70 1315 17.71 2-Heptenal 13 6.13 17.26 10.91 3.02 2.2 1.13 1325 18.14 2-Penten-1-ol - 1.07 58.22 38.84 9.58 7.19 2.61 5-Hepten-2-one, 1334 18.55 50 642.2 1793 1244 240.9 200.4 108.5 6-methyl- 1359 19.71 1-Hexanol 200 59.57 223.9 127.5 45.98 35.48 14.93 1387 20.99 3-Hexen-1-ol - 174.1 710.0 387.9 156.1 123.8 42.09 1398 21.52 2-Isobutylthiazole 3.5 61.31 354.2 228.6 47.36 51.52 26.73 1420 22.44 2-Octenal, (E)- 4 4.51 17.12 11.07 2.97 2.35 0.73 1440 23.31 Linalool oxide 4 0.30 0.75 0.57 0.18 n.d n.d 1455 23.97 1-Octen-3-ol 14 8.66 54.98 31.67 11.22 8.20 3.89 5-Hepten-2-ol, 6- 1475 24.54 - 15.95 78.52 45.61 19.74 9.82 3.35 methyl- 1482 25.14 2,4-Heptadienal - 3.18 11.24 7.41 2.10 1.73 n.d 1-Hexanol, 2- 1494 25.66 - 1.91 60.49 33.35 12.25 10.13 4.76 ethyl- 2,4-Hexadienoic 1500 25.93 - n.d 846.2 587.0 114.0 96.66 48.54 acid, ethyl ester 1507 26.21 Benzaldehyde 3500 26.41 74.98 51.28 10.18 8.96 4.56 1551 27.98 linalool 6 6.15 35.63 18.26 8.63 5.90 2.84 3,5-Heptadien-2- 1563 29.24 one, 6-methyl-, 380 4.11 13.43 7.29 3.64 2.10 0.40 (E)- 1583 30.14 β-cyclocitral 5 43.69 17.32 10.37 3.78 2.02 1.15 1694 32.64 β-citral 30 31.16 286.8 136.0 91.27 40.25 19.28 1698 33.33 2,4-Nonadienal - 2.16 17.55 8.34 4.94 2.69 1.58 1722 34.57 α- citral - 55.00 766.8 344.0 240.9 119.9 62.01 1711 35.85 Methyl salicylate 40 3.56 15.26 9.87 2.57 1.86 0.96 1758 37.66 β-Damascenone 0.002 0.95 9.31 4.15 2.46 1.74 0.96 1809 39.02 Geranylacetone 60 3.80 317.9 76.97 58.52 96.34 86.11 Phenylethyl 1849 40.93 1100 4.57 17.04 5.14 5.12 3.29 3.49 Alcohol 1906 41.70 β-Ionone 0.007 0.84 60.14 19.22 14.34 15.72 10.86 2023 44.43 cis-.psi.-Ionone - n.d 19.13 8.09 5.81 3.10 2.13 2079 46.06 psi.-Ionone - n.d 89.60 36.83 33.92 10.35 8.50 n.d represent not detected RI of DB-Wax plus column was reported based on C8-C20 alkane standard. TD represents threshold value reported by Buttery et al. (1971) or Burdock (2016) F1-F4 refer to the different fractions of tomato essence The essence is the combined total of Fractions 1-4

133

Table 5-4. Odor active values of important volatile components in tomato essence fractions by GC-MS. Compounds F1 F2 F3 F4 β-Ionone 2746 2049 2246 1551 β-Damascenone 2075 1230 870.0 480.0 1-Penten-3-ol 85.71 26.51 26.35 11.49 2-Isobutylthiazole 65.32 13.53 14.72 7.64 Hexanal 47.51 22.64 8.97 3.90 3-Hexenal 44.86 40.98 16.87 5.18 Butanal, 3-methyl- 41.2 19.5 4.05 - 1-Penten-3-one 25.02 5.60 0.62 - 5-Hepten-2-one, 6- 24.87 4.82 4.01 2.17 methyl- 2-Hexenal 5.21 2.93 1.14 0.69 β-citral 4.53 3.04 1.34 0.64 Linalool 3.04 1.44 0.98 0.47 2-Octenal, (E)- 2.77 0.74 0.59 0.18 1-Octen-3-ol 2.26 0.80 0.59 0.28 β-cyclocitral 2.07 0.76 0.40 0.23 Acetone 1.72 0.12 0.07 0.05 Geranylacetone 1.28 0.98 1.61 1.44 Odor active value (OAV)=concentration of compound/ threshold of compound Theoretically, compounds with OAV>1 should be detectable by humans F1-F4 refer to different fractions of the tomato essence

134

Figure 5-1. Chromatograms of four tomato essence fractions.

135

LIST OF REFERENCES

Aarset, B., Beckmann, S., Bigne, E., Beveridge, M., Bjorndal, T., Bunting, J., McDonagh, P., Mariojouls, C., Muir, J., & Prothero, A. (2004). The European consumers' understanding and perceptions of the “organic” food regime: The case of aquaculture. British Food Journal, 106(2), 93-105.

Acree (2015). Why you crave tomato juice in-flight. Accessed on July 19th, 2015. http://www.beveragemarketing.com/shop/us-fruit-beverages-through-2019- p665.aspx.

Acree, T. E., Barnard, J., & Cunningham, D. G. (1984). A procedure for the sensory analysis of gas-chromatographic effluents. Food Chemistry, 14(4), 273-286.

Alonso, A., Vázquez-Araújo, L., García-Martínez, S., Ruiz, J., & Carbonell-Barrachina, Á. A. (2009). Volatile compounds of traditional and virus-resistant breeding lines of Muchamiel tomatoes. European Food Research and Technology, 230(2), 315- 323.

Anthon, G. E., & Barrett, D. M. (2003). Thermal inactivation of lipoxygenase and hydroperoxytrienoic acid lyase in tomatoes. Food Chemistry, 81(2), 275-279.

Anthon, G. E., & Barrett, D. M. (2010). Changes in tomato paste during storage and the effects of heating on consistency of reconstitued tomato paste Journal of Texture Studies, 41(3), 262-278.

Anza, M., Riga, P., & Garbisu, C. (2006). Effects of variety and growth season on the organoleptic and nutritional quality of hydroponically grown tomato. Journal of Food Quality, 29(1), 16-37.

Attaway, J. A., Wolford, R. W., & Edwards, G. J. (1962). Fruit flavors and odors, the isolation and identification of some volatile carbonyl components from orange essence. Journal of Agricultural and Food Chemistry, 10(2), 102-104.

Auger, P., Devinney, T. M., Louviere, J. J., & Burke, P. F. (2010). The importance of social product attributes in consumer purchasing decisions: A multi-country comparative study. International Business Review, 19(2), 140-159.

Auldridge, M. E., McCarty, D. R., & Klee, H. J. (2006). Plant carotenoid cleavage oxygenases and their apocarotenoid products. Current Opinion in Plant Biology, 9(3), 315-321.

Auvray, M., & Spence, C. (2008). The multisensory perception of flavor. Consciousness and Cognition, 17(3), 1016-1031.

136

Baldwin, E., Scott, J. W., Einstein, M. A., Malundo, T. M. M., Carr, B. T., Shewfelt, R. L., & Tandon, K. S. (1998). Relationship between sensory and instrumental analysis for tomato flavor. Journal of the American Society for Horticultural Science, 123(5), 906-915.

Baldwin, E., Goodner, K., Plotto, A., Pritchett, K., & Einstein, M. (2004). Effect of volatiles and their concentration on perception of tomato descriptors. Journal of Food Science, 69(8), S310-S318.

Baldwin, E., Goodner, K., & Plotto, A. (2008). Interaction of volatiles, sugars, and acids on perception of tomato aroma and flavor descriptors. Journal of Food Science, 73(6), S294-S307.

Baldwin, E., Goodner, K., Plotto, A., Pritchett, K., & Einstein, M. (2004). Effect of volatiles and their concentration on perception of tomato descriptors. Journal of Food Science, 69(8), S310-S318.

Beltran, J., Serrano, E., Lopez, F. J., Peruga, A., Valcarcel, M., & Rosello, S. (2006). Comparison of two quantitative GC-MS methods for analysis of tomato aroma based on purge-and-trap and on solid-phase microextraction. Analytical and Bioanalytical Chemistry, 385(7), 1255-1264.

Bennett, S., Littrell, K., Hill, T., Mahovic, M., & Behravesh, C. B. (2015). Multistate foodborne disease outbreaks associated with raw tomatoes, United States, 1990–2010: a recurring public health problem. Epidemiology and Infection, 143(07), 1352-1359.

Berna, A. Z., Lammertyn, J., Saevels, S., Di Natale, C., & Nicolaı̈, B. M. (2004). Electronic nose systems to study shelf life and cultivar effect on tomato aroma profile. Sensors and Actuators B: Chemical, 97(2), 324-333.

Beverage Marketing Corporation (2015). U.S fruit beverages through 2019. Accessed on July 19th, 2015. http://www.beveragemarketing.com/shop/us-fruit-beverages- through-2019-p665.aspx.

Blain, J., Patterson, J., & Pearce, M. (1968). Carotene bleaching activity of tomato extracts. Jounal of the Science of Food and Agriculture, 19(12), 713-717.

Boe, A., & Salunkhe, D. (1967). Ripening tomatoes: ethylene, oxygen, and light treatments. Economic Botany, 21(4), 312-319.

Burdock, G. A. (2016). Fenaroli's handbook of flavor ingredients: CRC Press.

Buttery, Teranishi, R., Ling, L. C., & Turnbaugh, J. G. (1990). Quantitative and sensory studies on tomato paste volatiles. Journal of Agricultural and Food Chemistry, 38(1), 336-340.

137

Buttery, R. (1993). Quantitative and sensory aspects of flavor of tomato and other vegetables and fruits. Flavor science: Sensible principles and techniques, 259- 286.

Buttery, R., Seifert, R., Guadagni, D., & Ling, L. (1971). Characterization of additional volatile components of tomato. Journal of Agricultural and Food Chemistry, 19(3), 524-529.

Buttery, R. G., Teranishi, R., Flath, R. A., & Ling, L. C. (1989). Fresh tomato volatiles- composition and sensory studies. Acs Symposium Series, 388, 213-222.

Buttery, R. G., Teranishi, R., & Ling, L. C. (1987). Fresh tomato aroma volatiles-a quantitative study. Journal of Agricultural and Food Chemistry, 35(4), 540-544.

Buttery, R. G., Teranishi, R., Ling, L. C., & Turnbaugh, J. G. (1990). Quantitative and sensory studies on tomato paste volatiles. Journal of Agricultural and Food Chemistry, 38(1), 336-340.

Buttery, R. G., Teranishi, R., Ling, L. C., & Turnbaugh, J. G. (1991). Cooked tomato flavor composition. Google Patents.

Cai, J., Liu, B., Ling, P., & Su, Q. (2002). Analysis of free and bound volatiles by gas chromatography and gas chromatography–mass spectrometry in uncased and cased tobaccos. Journal of Chromatography A, 947(2), 267-275.

California Tomato Commission (2014). Stages of tomato ripening. Accessed on July 19th, 2015. http://www.lagorio.com/assets/pdf/lagorio-tomato-guide.pdf

Canene-Adams, K., Campbell, J. K., Zaripheh, S., Jeffery, E. H., & Erdman, J. W. (2005). The tomato as a functional food. The Journal of Nutrition, 135(5), 1226- 1230.

CDC. (2014). How Restaurants Handle Tomatoes. Accessed on July 19th, 2015. http://www.cdc.gov/nceh/ehs/ehsnet/plain_language/how-restaurants-handle- tomatoes.htm#fio.

Chadha, M., & Oluoch, M. (2003). Home-based vegetable gardens and other strategies to overcome micronutrient malnutrition in developing countries. Food Nutrition and Agriculture(32), 17-23.

Chambers, E. T., & Koppel, K. (2013). Associations of volatile compounds with sensory aroma and flavor: the complex nature of flavor. Molecules, 18(5), 4887-4905.

Charve, J., Chen, C., Hegeman, A. D., & Reineccius, G. A. (2011). Evaluation of instrumental methods for the untargeted analysis of chemical stimuli of orange juice flavour. Flavour and Fragrance Journal, 26(6), 429-440.

138

Chhabra, S. (2015). Determining the optimal price point: Using Van Westendorp’s price sensitivity meter. In Managing in Recovering Markets, (pp. 257-270): Springer.

Christiansen, K., Olsen, E., Vegarud, G., Langsrud, T., Lea, P., Haugen, J., & Egelandsdal, B. (2011). Flavor release of the tomato flavor enhancer, 2- isobutylthiazole, from whey protein stabilized model dressings. Food Science and Technology International, 1082013210381935.

Chung, T.-Y., Hayase, F., & Kato, H. (1983). Volatile components of ripe tomatoes and their juices, purees and pastes. Agricultural and Biological Chemistry, 47(2), 343- 351.

Cole, E., & Kapur, N. (1957). The stability of lycopene. I.‐Degradation by oxygen. Journal of the Science of Food and Agricture, 8(6), 360-365.

Coleman, R. L., Lund, E. D., & Moshonas, M. G. (1969). Composition of orange essence oil. Journal of Food Science, 34(6), 610-611.

Cosmai, L., Summo, C., Caponio, F., Paradiso, V. M., & Gomes, T. (2013). Influence of the thermal stabilization process on the volatile profile of canned tomato-based food. Journal of Food Science, 78(12), C1865-1870.

Couper, M. P. (2000). Review: Web surveys: A review of issues and approaches. The Public Opinion Quarterly, 64(4), 464-494.

Dalal, K., Salunkhe, D., Olson, L., Do, J., & Yu, M. (1968). Volatile components of developing tomato fruit grown under field and greenhouse conditions. Plant and Cell Physiology, 9(2), 389-400.

Daugherty, T., Lee, W. N., Gangadharbatla, H., Kim, K., & Outhavong, S. (2005). Organizational virtual communities: Exploring motivations behind online panel participation. Journal of Computer‐Mediated Communication, 10(4), 00-00.

Delahunty, C. M., Eyres, G., & Dufour, J. P. (2006). Gas chromatography-olfactometry. Journal of Separation Science, 29(14), 2107-2125.

Dei Frantelli. (2015). Trully tomato juices. Accessed on Dec 19th, 2015. http://www.deifratelli.com/truly/

Deutz, I. E. U. R. V., Dunphy, P. J., & Van, D. H. H. T. (2003). Tomato products enriched in beta-cyclocitral. Google Patents.

Dittmann, B., Zimmermann, B., Engelen, C., Jany, G., & Nitz, S. (2000). Use of the MS- sensor to discriminate between different dosages of garlic flavoring in tomato sauce. Journal of Agricultural and Food Chemistry, 48(7), 2887-2892.

139

Du, X., Song, M., Baldwin, E., & Rouseff, R. (2015). Identification of sulphur volatiles and GC-olfactometry aroma profiling in two fresh tomato cultivars. Food Chemistry, 171(0), 306-314.

Egydio, J. A., Moraes, Â. M., & Rosa, P. T. V. (2010). Supercritical fluid extraction of lycopene from tomato juice and characterization of its antioxidation activity. The Journal of Supercritical Fluids, 54(2), 159-164.

Ensminger, A. H., Ensminger, M., Konlande, J. E., & Robson, J. R. (1995). The concise encyclopedia of foods & nutrition: CRC Press Inc.

Etievant, P. X., Callement, G., Langlois, D., Issanchou, S., & Coquibus, N. (1999). Odor intensity evaluation in gas chromatography olfactometry by finger span method. Journal of Agricultural and Food Chemistry, 47(4), 1673-1680.

FDA. (2011). Food code 2009. Accessed on July 19th, 2015. http://www.fda.gov/Food/GuidanceRegulation/RetailFoodProtection/FoodCode/u cm186451.htm.

FDA. (2015). Accessed on July 19th, 2015. http://www.fda.gov/Food/ResourcesForYou/Consumers/ucm180323.htm.

Feussner, I., & Wasternack, C. (2002). The lipoxygenase pathway. Annual review of plant biology, 53(1), 275-297.

Florida Tomato Committee (2015). Tomato 101 fact sheet. Accessed on July 19th, 2015. http://www.floridatomatoes.org/tomato-101/

Fonseca, H., & Luh, B. (1976). Effect of break temperature on quality of tomato juice reconstituted from frozen tomato concentrates. Journal of Food Science, 41(6), 1308-1311.

Francis, I. L., Sefton, M. A., & Williams, P. J. (1992). Sensory descriptive analysis of the aroma of hydrolyzed precursor fractions from Semillon, Chardonnay and Sauvignon blanc grape juice. Journal of Agricultural and Food Chemistry, 59(4), 511-520.

Gancedo, M., & Luh, B. (1986). HPLC analysis of organic acids and sugars in tomato juice. Journal of Food Science, 51(3), 571-573.

Giovannucci, E. (1999). Tomatoes, tomato-based products, lycopene, and cancer: review of the epidemiologic literature. Journal of the National Cancer Institute, 91(4), 317-331.

Goodman, C., Fawcett, S., & Barringer, S. (2002). Flavor, viscosity, and color analyses of hot and cold break tomato juices. Journal of Food Science, 67(1), 404-408.

140

Gorecki, T., & Pawliszyn, J. (1997). Field-portable solid-phase microextraction fast GC system for trace analysis. Field Analytical Chemistry and Technology, 1(5), 277- 284.

Guiné, R. P., Martín-Belloso, O., Mínguez-Mosquera, M. I., Paliyath, G., Pessoa, F. L., Le Quéré, J.-L., Sidhu, J. S., Sinha, N., Peggy Stanfield, R., & Hui, Y. (2010). Handbook of fruit and vegetable flavors: John Wiley and Sons.

Halbert, S. E., & Manjunath, K. L. (2004). Asian citrus psyllids (Sternorrhyncha: Psyllidae) and greening disease of citrus: a literature review and assessment of risk in Florida. Florida Entomologist, 87(3), 330-353.

Hayes, W., Smith, P., & Morris, A. (1998). The production and quality of tomato concentrates. Critical Reviews in Food Science and Nutrition, 38(7), 537-564.

Hensher, D. A., Greene, W. H., & Chorus, C. G. (2013). Random regret minimization or random utility maximization: an exploratory analysis in the context of automobile fuel choice. Journal of Advanced Transportation, 47(7), 667-678.

Heuvelink, E. (2005). Tomatoes (Vol. 13): CABI.

Hicks, D. T., Pivarnik, L. F., McDermott, R., Richard, N., Hoover, D. G., & Kniel, K. E. (2009). Consumer awareness and willingness to pay for high‐pressure processing of ready‐to‐eat food. Journal of Food Science Education, 8(2), 32-38.

House, L. A., Gao, Z., Spreen, T. H., Gmitter, F. G., Valim, M. F., Plotto, A., & Baldwin, E. A. (2011). Consumer preference for mandarins: implications of a sensory analysis. Agribusiness, 27(4), 450-464.

Hu, W., Woods, T., & Bastin, S. (2009). Consumer acceptance and willingness to pay for blueberry products with nonconventional attributes. Journal of Agricultural and Applied Economics, 41(01), 47-60.

Hu, W., Woods, T., Bastin, S., Cox, L., & You, W. (2011). Assessing consumer willingness to pay for value-added blueberry products using a payment card survey. Journal of Agricultural and Applied Economics, 43(2), 243.

Hughner, R. S., & Kleine, S. S. (2004). Views of health in the lay sector: a compilation and review of how individuals think about health. Health:, 8(4), 395-422.

Hui, Y. H., & Evranuz, E. Ö. (2012). Handbook of animal-based fermented food and beverage technology: CRC press.

Ilg, A., Bruno, M., Beyer, P., & Al-Babili, S. (2014). Tomato carotenoid cleavage dioxygenases 1A and 1B: Relaxed double bond specificity leads to a plenitude of dialdehydes, mono‐apocarotenoids and isoprenoid volatiles. FEBS open bio, 4(1), 584-593.

141

Ishida, B. K., Baldwin, E. A., Buttery, R. G., Chui, S. H., & Ling, L. C. (1993). Flavor volatiles, sugars and color development in ripening in-vitro-cultured tomato fruit and calyx. Physiologia Plantarum, 89(4), 861-867.

Jayasena, V., & Cameron, I. (2008). ° Brix/acid ratio as a predictor of consumer acceptability of Crimson Seedless table grapes. Journal of Food Quality, 31(6), 736-750.

Jones, M. S., House, L. A., & Gao, Z. (2015). Respondent screening and revealed preference axioms: testing quarantining methods for enhanced data quality in Web panel surveys. Public Opinion Quarterly, 79(3), 687-709.

Kazeniac, S., & Hall, R. (1970). Flavor chemistry of tomato volatiles. Journal of Food Science, 35(5), 519-530.

Klee, H. J. (2010). Improving the flavor of fresh fruits: genomics, biochemistry, and biotechnology. New Phytologist, 187(1), 44-56.

Klee, H. J. (2015). Heirloom variety trials. Accessed on July 19th, 2015. http://hos.ufl.edu/kleeweb/newcultivars.html

Krammer, G. E., Buttery, R. G., & Takeoka, G. R. (1995). Studies on tomato glycosides. In R. L. Rouseff & M. M. Leahy (Eds.), Fruit Flavors: Biogenesis, Characterization, and Authentication, vol. 596 (pp. 164-181).

Krinsky, I., & Robb, A. L. (1986). On approximating the statistical properties of elasticities. The Review of Economics and Statistics, 715-719.

Lairon, D. (2011). Nutritional quality and safety of organic food. In Sustainable Agriculture Volume 2, (pp. 99-110): Springer.

Lakewood. (2016). Organic tomato juice pure. Accessed on July 19th, 2016. http://www.lakewoodjuices.com/product_detail/id-145/

Landy, P., Boucon, C., Kooyman, G. M., Musters, P. A., Rosing, E. A., de Joode, T., & Haring, P. G. (2002). Sensory and chemical changes in tomato sauces during storage. Journal of Agricultural and Food Chemistry,50(11), 3262-3271.

Lawless, H., & Heymann, H. (1988). Sensory evaluation of food: Practices and principals. Chapman and Hall: New York, NY, USA.

Lawless, H. T., & Heymann, H. (2010). Sensory evaluation of food: principles and practices: Springer Science & Business Media.

Lewinsohn, E., Sitrit, Y., Bar, E., Azulay, Y., Ibdah, M., Meir, A., Yosef, E., Zamir, D., & Tadmor, Y. (2005). Not just colors—carotenoid degradation as a link between pigmentation and aroma in tomato and watermelon fruit. Trends in Food Science & Technology, 16(9), 407-415.

142

Linssen, J. P. H., Janssens, J., Roozen, J. P., & Posthumus, M. A. (1993). Combined gas-chromatography and sniffing port analysis of volatile compounds of minearl water packed in polyethylene laminated packages. Food Chemistry, 46(4), 367- 371.

Loureiro, M. L., & Hine, S. (2002). Discovering niche markets: A comparison of consumer willingness to pay for local (Colorado grown), organic, and GMO-free products. Journal of Agricultural and Applied Economics, 34(03), 477-487.

Mabiso, A., Sterns, J., House, L., & Wysocki, A. (2005). Estimating consumers’ willingness-to-pay for country-of-origin labels in fresh apples and tomatoes: A double-hurdle probit analysis of American data using factor scores. In American Agricultural Economics Association Annual Meeting, Providence, Rhode Island, (pp. 24-27).

Magkos, F., Arvaniti, F., & Zampelas, A. (2006). Organic food: buying more safety or just peace of mind? A critical review of the literature. Critical Reviews in Food Science and Nutrition, 46(1), 23-56.

Maneerat, C., Hayata, Y., Kozuka, H., Sakamoto, K., & Osajima, Y. (2002). Application of the Porapak Q column extraction method for tomato flavor volatile analysis. Journal of Agricultural and Food Chemistry, 50(12), 3401-3404.

Mathieu, S., Cin, V. D., Fei, Z., Li, H., Bliss, P., Taylor, M. G., Klee, H. J., & Tieman, D. M. (2009). Flavour compounds in tomato fruits: identification of loci and potential pathways affecting volatile composition. Journal of Experimental Botany, 60(1), 325-337.

Mayer, F., Takeoka, G., Buttery, R., Nam, Y., Naim, M., Bezman, Y., & Rabinowitch, H. (2003). Aroma of fresh field tomatoes. In K. R. Cadwallader & H. Weenen (Eds.), Freshness and Shelf Life of Foods, vol. 836 (pp. 144-161). Washington: Amer Chemical Soc.

Mayer, F., Takeoka, G. R., Buttery, R. G., Whitehand, L. C., Naim, M., & Rabinowitch, H. D. (2008). Studies on the aroma of five fresh tomato cultivars and the precursors of cis- and trans-4,5-epoxy-(E)-2-decenals and methional. Journal of Agricultural and Food Chemistry, 56(10), 3749-3757.

Mayer, F., Takeoka, G. R., Buttery, R. G., Whitehand, L. C., Naim, M., & Rabinowitch, H. D. (2008). Studies on the aroma of five fresh tomato cultivars and the precursors of cis-and trans-4, 5-epoxy-(E)-2-decenals and methional. Journal of Agricultural and Food Chemistry, 56(10), 3749-3757.

McDonald's to Pay Tomato Farmers More. (2007) Wall Street Journal Eastern edition New York, N.Y.

Meilgaard, M. C., Carr, B. T., & Civille, G. V. (2006). Sensory evaluation techniques: CRC Press.

143

Mesías Díaz, F. J., Martínez-Carrasco Pleite, F., Miguel Martínez Paz, J., & Gaspar García, P. (2012). Consumer knowledge, consumption, and willingness to pay for organic tomatoes. British Food Journal, 114(3), 318-334.

Modig, T., Liden, G., & Taherzadeh, M. J. (2002). Inhibition effects of furfural on alcohol dehydrogenase, aldehyde dehydrogenase and pyruvate dehydrogenase. Biochemical Journal, 363(3), 769-776.

Moshonas, M. G., & Shaw, P. E. (1983). Composition of aqueous essence and essence oil from Citrus temple. Journal of Agricultural and Food Chemistry, 31(2), 334- 336.

Moskowitz, H. R., & Krieger, B. (1995). The contribution of sensory liking to overall liking: An analysis of six food categories. Food Quality and Preference, 6(2), 83- 90.

Mottram, D. S. (2007). The Maillard reaction: source of flavour in thermally processed foods. In Flavours and Fragrances, (pp. 269-283): Springer.

Munoz-Gonzalez, C., Feron, G., Guichard, E., Rodriguez-Bencomo, J. J., Martin- Alvarez, P. J., Moreno-Arribas, M. V., & Pozo-Bayon, M. A. (2014). Understanding the role of saliva in aroma release from wine by using static and dynamic headspace conditions. Journal of Agricultural and Food Chemistry, 62(33), 8274-8288.

Nelson, P., & Hoff, J. (1969). Tomato volatiles: Effect of variety, processing and storage time. Journal of Food Science, 34(1), 53-57.

Nielsen. (2015). Despite growing health sales, consumers are caught with their hands in the cookie jar. Accessed on July 19th, 2015. http://www.nielsen.com/us/en/insights/news/2015/despite-growing-health-sales- consumers-are-caught-with-their-hands-in-the-cookie-jar.html

Nix v. Hedden. In. (1893). U. S. Court.

Okechuku, C. (1994). The importance of product country of origin: a conjoint analysis of the United States, Canada, Germany and The Netherlands. European Journal of Marketing, 28(4), 5-19.

Ortiz-Serrano, P., & Gil, J. V. (2007). Quantitation of free and glycosidically bound volatiles in and effect of glycosidase addition on three tomato varieties (Solanum lycopersicum L.). Journal of Agricultural and Food Chemistry, 55(22), 9170-9176.

Parker, J. K., Elmore, S., & Methven, L. (2014). Flavour development, analysis and perception in food and beverages: Elsevier.

144

Peng, J., Mah, J.-H., Somavat, R., Mohamed, H., Sastry, S., & Tang, J. (2012). Thermal inactivation kinetics of Bacillus coagulans spores in tomato juice. Journal of Food Protection, 75(7), 1236-1242.

Peralta, I. E., Spooner, D. M., Razdan, M., & Mattoo, A. (2007). History, origin and early cultivation of tomato (Solanaceae). Genetic Improvement of Solanaceous Crops: Tomato, 2.

Petrin, A., & Train, K. (2003). Omitted product attributes in discrete choice models. In): National Bureau of Economic Research.

Petro‐Turza, M. (1986). Flavor of tomato and tomato products. Food Reviews International, 2(3), 309-351.

Pineau, B., Barbe, J.-C., Van Leeuwen, C., & Dubourdieu, D. (2007). Which impact for β-damascenone on red wines aroma? Journal of Agricultural and Food Chemistry, 55(10), 4103-4108.

Pleven, L., & Cui, C. (2010). Dying on the vine: Tomato Prices. In Wall Street Journal Eastern edition. New York, N.Y.

Prado, F. C., Parada, J. L., Pandey, A., & Soccol, C. R. (2008). Trends in non-dairy probiotic beverages. Food Research International, 41(2), 111-123.

Prescott, J., Johnstone, V., & Francis, J. (2004). Odor–taste interactions: effects of attentional strategies during exposure. Chemical Senses, 29(4), 331-340.

Pritzker, F. (2014). Washington State Deli Recalls Canned Tomato Juice and Adjika. Food Poisoning Bulletin. Accessed on July 19th, 2015. http://foodpoisoningbulletin.com/2014/washington-state-deli-recalls-canned- tomato-juice-and-adjika/.

Putnam, A. (2013). Florida agriculture by the numbers. In, (pp. 180). Accessed on July 19th, 2015. http://www.freshfromflorida.com/Divisions-Offices/Marketing-and- Development/Education/For-Researchers/Florida-Agriculture-Overview-and- Statistics: Florida Department of Agriculutre and Consumer Services.

Reineccius, G. (2004). Flavor chemistry and technology: CRC press.

Reineccius, G. A. (2008). Flavoromics - the next frontier? Abstracts of Papers of the American Chemical Society, 235, 1.

Ringblom, U. (2004). The orange book: Tetra Pak.

Rouseff, R., & Cadwallader, K. (2001). Headspace techniques in foods, fragrances and flavors - An overview. In R. L. Rouseff & K. R. Cadwallader (Eds.), Headspace Analysis of Foods and Flavors: Theory and Practice, vol. 488 (pp. 1-8).

145

Rowe, D. (1998). Aroma chemicals for savory flavors. Perfumer and Flavorist, 23, 9-18.

Salunkhe, D., Jadhav, S., & Yu, M. (1974). Quality and nutritional composition of tomato fruit as influenced by certain biochemical and physiological changes. Qualitas Plantarum, 24(1-2), 85-113.

Sandoval, A., Barreiro, J., & Mendoza, S. (1992). Thermal resistance of Bacillus coagulans in double concentrated tomato paste. Journal of Food Science, 57(6), 1369-1370.

Schatzker, M. (2015). This is the perfect tomato. Accessed on Oct, 18th, 2015. http://www.slate.com/articles/life/food/2015/07/garden_gem_tomato_why_harry_ klee_s_perfect_cultivar_isn_t_sold_in_supermarkets.html

Schneider, K. R., Schneider, R. G., Hubbard, M. A., & Richardson, S. (2014). Preventing Foodborne Illness: Salmonellosis. Accessed on July 19th, 2015. http://edis.ifas.ufl.edu/fs096

Schwab, W., Davidovich‐Rikanati, R., & Lewinsohn, E. (2008). Biosynthesis of plant‐ derived flavor compounds. The Plant Journal, 54(4), 712-732.

Schwieterman, M. L., Colquhoun, T. A., Jaworski, E. A., Bartoshuk, L. M., Gilbert, J. L., Tieman, D. M., Odabasi, A. Z., Moskowitz, H. R., Folta, K. M., Klee, H. J., Sims, C. A., Whitaker, V. M., & Clark, D. G. (2014). Strawberry flavor: diverse chemical compositions, a seasonal influence, and effects on sensory perception. PLoS One, 9(2), e88446.

Scott, J. W., Baldwin, E. A., Klee, H. J., Brecht, J. K., Olson, S. M., Bartz, J. A., & Sims, C. A. (2008). Fla. 8153 hybrid tomato; Fla. 8059 and Fla. 7907 breeding lines. Hort Science, 43(7), 2228-2230.

Selli, S., Kelebek, H., Ayseli, M. T., & Tokbas, H. (2014). Characterization of the most aroma-active compounds in cherry tomato by application of the aroma extract dilution analysis. Food Chemistry, 165, 540-546.

Servili, M., Selvaggini, R., Taticchi, A., Begliomini, A. L., & Montedoro, G. (2000). Relationships between the volatile compounds evaluated by solid phase microextraction and the thermal treatment of tomato juice: optimization of the blanching parameters. Food Chemistry, 71(3), 407-415.

Shallenberger, R., & Moyer, J. (1961). Sugar-starch transformation in peas, relation between changes in glucose, fructose, galactose, sucrose, and stachyose, and the formation of starch in peas. Journal of Agricultural and Food Chemistry, 9(2), 137-140.

Sieso, V., & Crouzet, J. (1977). Tomato volatile components: Effect of processing. Food Chemistry, 2(4), 241-252.

146

Silva, F. V., & Gibbs, P. (2004). Target selection in designing pasteurization processes for shelf-stable high-acid fruit products. Critical Reviews in Food Science and Nutrition, 44(5), 353-360.

Simkin, A. J., Schwartz, S. H., Auldridge, M., Taylor, M. G., & Klee, H. J. (2004). The tomato carotenoid cleavage dioxygenase 1 genes contribute to the formation of the flavor volatiles β‐ionone, pseudoionone, and geranylacetone. The Plant Journal, 40(6), 882-892.

Sinesio, F., Cammareri, M., Moneta, E., Navez, B., Peparaio, M., Causse, M., & Grandillo, S. (2010). Sensory quality of fresh French and Dutch market tomatoes: a preference mapping study with Italian consumers. Journal of Food Science, 75(1), S55-S67.

Smith, A. F. (1994). The tomato in America: early history, culture, and cookery: University of Illinois Press.

Song, J., Fan, L. H., & Beaudry, R. M. (1998). Application of solid phase microextraction and gas chromatography time-of-flight mass spectrometry for rapid analysis of flavor volatiles in tomato and strawberry fruits. Journal of Agricultural and Food Chemistry, 46(9), 3721-3726.

Stahl-Biskup, E., Intert, F., Holthuijzen, J., Stengele, M., & Schulz, G. (1993). Glycosidically bound volatiles: A review 1986-1991. Flavour and Fragrance Journal, 8(2), 61-80.

Stevens, M. A. (1970). Relationship between polyene-carotene content and volatile compound composition of tomatoes. Journal of the American Society of Horticultural Science, 95, 461-464.

Stewart, H., Hyman, J., Buzby, J. C., Frazao, E., & Carlson, A. (2011). How much do fruits and vegetables cost? : DIANE Publishing.

Stone, E., Hall, R., & Kazeniac, S. (1975). Formation of aldehydes and alcohols in tomato fruit from U‐14 C‐labeled linolenic and linoleic acids. Journal of Food Science, 40(6), 1138-1141.

Sucan, M. K., & Russell, G. F. (2002). Effects of processing on tomato bioactive volatile compounds. Bioactive Compounds in Foods, 816, 155-172.

Sucan, M. K., & Russell, G. F. (2001). Dynamic headspace analysis of fresh tomato juices. In R. L. Rouseff & K. R. Cadwallader (Eds.), Headspace Analysis of Foods and Flavors: Theory and Practice, vol. 488 (pp. 165-173).

Tandon, K. S., Baldwin, E. A., Scott, J. W., & Shewfelt, R. L. (2003). Linking sensory descriptors to volatile and nonvolatile components of fresh tomato flavor. Journal of Food Science, 68(7), 2366-2371.

147

Tandon, K. S., Jordan, M., Goodner, K. L., & Baldwin, E. A. (2001). Characterization of fresh tomato aroma volatiles using GC-olfactometry. Proceedings of the Florida State Horticultural Society, vol. 114, 142-144.

Taylor, A. J., & Linforth, R. (2009). Food flavour technology: John Wiley & Sons.

Teranishi, R., Buttery, R. G., Stern, D. J., & Takeoka, G. (1991). Use of odor thresholds in aroma research Food Science and Technology-Lebensmittel-Wissenschaft & Technologie, 24(1), 1-5.

Thakur, B., Singh, R., & Nelson, P. (1996). Quality attributes of processed tomato products: a review. Food Reviews International, 12(3), 375-401.

Thornsbury, S., & Jerardo, A. (2012). Vegetables and pulses outlook. Economic Research Service VGS-350, USDA.

Tieman, D., Bliss, P., McIntyre, L. M., Blandon-Ubeda, A., Bies, D., Odabasi, A. Z., Rodriguez, G. R., van der Knaap, E., Taylor, M. G., Goulet, C., Mageroy, M. H., Snyder, D. J., Colquhoun, T., Moskowitz, H., Clark, D. G., Sims, C., Bartoshuk, L., & Klee, H. J. (2012). The chemical interactions underlying tomato flavor preferences. Current Biology, 22(11), 1035-1039.

Tieman, D. M., Zeigler, M., Schmelz, E. A., Taylor, M. G., Bliss, P., Kirst, M., & Klee, H. J. (2006). Identification of loci affecting flavour volatile emissions in tomato fruits. Journal of Experimental Botany, 57(4), 887-896.

Train, K. E. (2002). Discrete choice methods with simulation (Vol. 8): Cambridge University Press Cambridge.

Ullrich, F., & Grosch, W. (1987). Identification of the most intense volatile flavor compoudns formed during autoxidation of linoleic acid. Zeitschrift Fur Lebensmittel-Untersuchung Und-Forschung, 184(4), 277-282.

USCB. (2012). Age and sex composition: 2010. Accessed on Sep 20th, 2016. http://www.census.gov/prod/cen2010/briefs/c2010br-03.pdf

USDA. (2010). U.S. tomato statistics. Accessed on Sep 20th, 2016. http://usda.mannlib.cornell.edu/MannUsda/viewDocumentInfo.do?documentID=1 210

Valadez, A. M., Schneider, K. R., & Danyluk, M. D. (2013). Outbreaks of foodborne diseases associated with tomatoes. IFAS Extension.

Vallverdú-Queralt, A., Regueiro, J., de Alvarenga, J., Torrado, X., & Lamuela-Raventos, R. (2015). Carotenoid profile of tomato sauces: effect of cooking time and content of extra virgin olive oil. International Journal of Molecular Sciences, 16(5), 9588.

148

van Ruth, S. M., & O'Connor, C. H. (2001). Evaluation of three gas chromatography- olfactometry methods: comparison of odour intensity-concentration relationships of eight volatile compounds with sensory headspace data. Food Chemistry, 74(3), 341-347.

Veit‐Köhler, U., Krumbein, A., & Kosegarten, H. (1999). Effect of different water supply on plant growth and fruit quality of Lycopersicon esculentum. Journal of Plant Nutrition and Soil Science, 162(6), 583-588.

Villière, A., Le Roy, S., Fillonneau, C., Guillet, F., Falquerho, H., Boussely, S., & Prost, C. (2015). Evaluation of aroma profile differences between sué, sautéed, and pan-fried onions using an innovative olfactometric approach. Flavour, 4(1), 1-19.

Vogel, J. T., Tan, B.-C., McCarty, D. R., & Klee, H. J. (2008). The carotenoid cleavage dioxygenase 1 enzyme has broad substrate specificity, cleaving multiple carotenoids at two different bond positions. Journal of Biological Chemistry, 283(17), 11364-11373.

Vogel, J. T., Tieman, D. M., Sims, C. A., Odabasi, A. Z., Clark, D. G., & Klee, H. J. (2010). Carotenoid content impacts flavor acceptability in tomato (Solanum lycopersicum). Journal of the Science of Food and Agriculture, 90(13), 2233- 2240.

Wang, C., Chin, C.-K., Ho, C.-T., Hwang, C.-F., Polashock, J. J., & Martin, C. E. (1996). Changes of fatty acids and fatty acid-derived flavor compounds by expressing the yeast Δ-9 desaturase gene in tomato. Journal of Agricultural and Food Chemistry, 44(10), 3399-3402.

Wang, Y. T., Huang, S. W., Liu, R. L., & Jin, J. Y. (2007). Effects of nitrogen application on flavor compounds of cherry tomato fruits. Journal of Plant Nutrition and Soil Science-Zeitschrift Fur Pflanzenernahrung Und Bodenkunde, 170(4), 461-468.

Wang, Y. T., Liu, R. L., Huang, S. W., & Jin, J. Y. (2009). Effects of Potassium Application on Flavor Compounds of Cherry Tomato Fruits. Journal of Plant Nutrition, 32(9), 1451-1468.

Wier, M., & Calverley, C. (2002). Market potential for organic foods in Europe. British Food Journal, 104(1), 45-62.

Williams, P. J., Sefton, M. A., & Wilson, B. (1989). Nonvolatile conjugates of secondary metabolites as precusors of varietal grape flavor components. Acs Symposium Series, 388, 35-48.

Wolford, R. W., Alberding, G. E., & Attaway, J. A. (1962). Analysis of recovered natural orange essence by gas chromatography Journal of Agricultural and Food Chemistry, 10(4), 297-&.

149

Wright, D. H., & Harris, N. D. (1985). Effect of nitrogen and potassium fertilization on tomato flavor. Journal of Agricultural and Food Chemistry, 33(3), 355-358.

Yeung, R. M., & Morris, J. (2001). Food safety risk: Consumer perception and purchase behaviour. British Food Journal, 103(3), 170-187.

Yilmaz, E. (2001). The chemistry of fresh tomato flavor. Turkish Journal of Agriculture and Forestry, 25(3), 149-155.

Zhang, H., Gallardo, R. K., McCluskey, J. J., & Kupferman, E. M. (2010). Consumers' willingness to pay for treatment-induced quality attributes in Anjou pears. Journal of Agricultural and Resource Economics, 105-117.

Zhang, J., & Etter, L. (2008). Tomato Industry Seeks Compensation. In Wall Street Journal Eastern edition. New York, N.Y.

Zhang, Z. M., Zeng, D. D., & Li, G. K. (2008). Study of the volatile composition of tomato during storage by a combination sampling method coupled with gas chromatography/mass spectrometry. Journal of Agricultural and Food Chemistry, 88(1), 116-124.

150

BIOGRAPHICAL SKETCH

Yaozhou Zhu has lived in several places including Shanghai, Tallahassee, and

Gainesville, FL. She graduated with honors with a Bachelor of Food Science and

Technology and a minor in finance from Shanghai Ocean University and Fudan

University in 2010. She then earned her Master of Science in food science and human nutrition from Florida State University in 2013 and began the pursuit of her doctoral degree in food science, with a minor in food resource economics, at the University of

Florida in the same year. She has a strong passion for flavor and consumer preference and intends to pursue a career in this discipline.

151