Search

Hybrid American-European hazelnuts (Corylus americana x C. avellana) are being developed in the Upper Midwest of the United States to enhance ecological and economic sustainability, while providing consumers with healthy, local food. Researchers are currently working on pre-commercialization aspects of this perennial crop, including the investigation of flavor quality, a key driver of food choice. This dissertation's overall aim was to evaluate the flavor of developing hybrid varieties, both through sensory and chemical analysis, and to identify key aroma compounds that impact perception and consumer liking. In the first phase of this project, consumer acceptance and perception of twelve different varieties of roasted hybrid hazelnuts were evaluated alongside two European hazelnut varieties, revealing unique aroma and flavor profiles that impacted liking. Key aroma and flavor attributes driving liking were identified as roasty, hazelnut-like, caramel-like, chocolate-like, sweet, and salty, while earthy, burnt, and bitter were identified as primary drivers of disliking. In phase two, gas chromatography/mass spectrometry/olfactometry (GC/MS/O) was utilized to identify odor-active compounds in a hybrid hazelnut variety, which was differentiated by consumers for its roasty and nutty aroma profile. Application of aroma extract dilution analysis (AEDA) revealed 33 odor-active compounds with flavor dilution values ≥ 16, including 2-acetylpyrazine and 2-aminoacetophenone as first reported odorants in hazelnut. Quantitative analysis of the identified compounds showed differences compared to the prior literature on European hazelnuts and highlighted the importance of roasting on the development of hazelnut aroma. Sensory descriptive analysis confirmed the contribution of 27 compounds quantified above literature threshold values to the roasted hazelnut aroma, which was broken into six main attributes: roasty, nutty, fruity, sweet/caramel, earthy, and fatty. The third phase of this project applied targeted and non-targeted GC/MS analyses with multivariate orthogonal partial least squares regression (OPLS) modeling to identify volatile compounds related to consumer aroma liking of 10 roasted hybrid hazelnut varieties. Highly predictive and positively correlated compounds from the targeted (n = 12) and non-targeted (n = 4) OPLS models were quantified and validated for their sensory relevance. Four new odor threshold values were determined for the predictive compounds furfuryl alcohol, 2-methyl-6,7-dihydro-5H-cyclopentapyrazine, 3-phenylfuran and methyl picolinate, the latter two of which were first identified in hazelnuts in this work. Consumer recombination testing confirmed that the sample with the addition of twelve compounds at above-threshold concentrations found in the most-liked hazelnut was preferred (p < 0.001, Δ aroma liking = 2.2) and perceived as more roasty, nutty, and sweet (p < 0.05), as compared to a control sample. Sensory testing also revealed that the addition of the four threshold-determined compounds at subthreshold levels was preferred (p = 0.02) and perceived as less earthy and mushroom-like than the control (p < 0.05).In summary, this work advanced the understanding of the flavor properties of roasted hybrid hazelnuts and identified molecular and sensory targets for improved consumer acceptance. These results provide better guidance for hybrid hazelnut breeding and post-processing strategies in the United States to enhance sensory quality.

Coffee is the second most-consumed beverage around the world. The popularity of coffee is favored by its stimulating effect and characteristic flavor. The bitterness of coffee is generally considered an important aspect of coffee flavor, however, at elevated levels is largely considered a negative attribute for consumer acceptance. Understanding the chemical components that impact the bitter perception of coffee provides a basis to optimize coffee quality. Historically, compounds that directly contribute to bitter perception in coffee have been identified. Currently, there is a limited understanding of flavor interactions, such as modulation, on the bitter perception of coffee. The overall goal of this project was to utilize untargeted LC/MS flavoromics analysis to identify chemical compounds that impact the bitter perception in coffee. This research consisted of two main phases focused on the investigation of comprehensive chemical profiles of coffee that were highly predictive and (1) negatively correlated or (2) positively correlated to the bitter perception of coffee.In phase I, untargeted LC/MS flavoromics profiling was applied to identify chemical compounds that suppressed bitter perception of coffee brew. Orthogonal partial least squares (OPLS) analysis with good model quality was established to correlate the chemical profiles and bitter intensity ratings of fourteen coffee brews. Ten chemical markers that were highly predictive and negatively correlated to the bitter perception of coffee were isolated and identified using multi-dimensional preparative LC/MS and NMR. Sensory recombination analysis validated three compounds, namely 4-caffeoylquinic acid, 5-caffeoylquinic acid, and 2-O-β-D-glucopyranosyl-atractyligenin, as bitter modulators which significantly decreased the bitter perception in coffee. Subsequently, in phase II, untargeted LC/MS flavoromics profiling was applied to identify chemical compounds that increased or enhanced the bitter perception of coffee brew. Five compounds that were highly predictive and positively correlated to bitter intensity were selected from the OPLS model and further isolated with more than 90% purity by multi-dimensional preparative LC/MS. Sensory recombination analysis determined the five compounds at subthreshold levels significantly enhanced the bitter perception of coffee when presented as a mixture. The compounds had no perceivable impact on bitterness when evaluated individually both in coffee and in water, indicating flavor modulation by the mixture of the five compounds. A set of roasting experiments demonstrated all five compounds were generated during the coffee roasting process. Overall, this study demonstrates the successful application of untargeted flavoromics to identify bitter modulators that significantly impact the bitter perception in the coffee brew. Novel compounds that suppressed and enhanced bitterness perception were identified. This study improves the understanding of bitter perception in coffee and provides a novel basis to develop strategies to optimize the bitter taste of coffee to increase consumer acceptance.

Reducing added dietary sugar intake is needed for the better health, leading to longer lifespan, less prevalence of preventable diseases, and better overall wellness. A common practice for reducing added calories, while maintaining sweetness, is to replace caloric sweetener ingredients with non-nutritive, high intensity sweeteners; however, consumer opinions of high intensity sweeteners remain a challenge, due to lower flavor quality, and often artificial compounds utilized. There has been an increased use of plant-based high intensity sweeteners, such as those found within the leaves of Stevia rebaudiana, which contain a variety of highly sweet di-terpene glycosides called rebaudiosides and steviosides. The most used stevia-based sweetener is Rebaudioside A (Reb A) driven primarily by its high relative sweetness intensity, large abundance in the plant leaves, and relatively clean taste. However, the use of stevia sweeteners is limited by lower flavor performance compared to common nutritive sweeteners. An improved understanding of the chemical drivers of stevia flavor is needed. The overall aim of this dissertation was to define the impact of storage on the flavor stability of Reb A in beverage systems. This research consisted of two phases: (1) to define the impact of Reb A degradation on the flavor profile and (2) to investigate mechanisms of Reb degradation. In the first phase of the project, untargeted LC-MS flavoromic chemical profiling was used to identify compounds related to sensory change in Reb A sweetened beverages during storage. Beverage samples consisting of 80%, 90% and 98% purity Reb A were stored under accelerated conditions for six weeks at 35 ⁰C. Good quality orthogonal partial least squares analysis (OPLS) models were developed from the chemical profiles and the d’ values from tetrad sensory testing with reasonable predictive ability. Four highly predictive compounds were selected based on VIP scores consisting of three positively correlating compounds and one negatively correlating compound. Based on accurate mass and NMR analysis, the compounds were identified as Reb A (negatively correlated) and three Reb A degradation compounds (positively correlated), that included a rearrangement, a hydration, and an epoxidation/rearrangement of Reb A, termed compounds 1, 2 and 3, respectively. The concentrations of compounds 1-3 in the stored beverages were determined to be below the sensory best estimate recognition threshold values. However further sensory recombination testing of the three compounds as a mixture revealed a perceivably significant sensory change. Thus, Reb A degradation products have an additive-type sub-threshold activity which causes a shift in perception in both unsweetened and sweetened beveragesThe second phase investigated the pathways of rebaudioside A (Reb A) degradation in acidic beverages during storage. Three degradation products of known sensory importance were monitored that included a rearrangement, a hydration, and an epoxidation/rearrangement of Reb A, termed compounds 1, 2 and 3, respectively. Using deuterium labeled water (D2O) experiments, compounds 1-2 were reported to be generated by acid-catalyzed mechanisms involving the carbocation on carbon position 16 (C16), followed by either proton elimination on C15 to form the more thermodynamically stable alkene (compound 1) or by the Markovnikov addition of water to form a tertiary alcohol (compound 2). Compound 3 was generated by the epoxidation of the alkene at C16-17 position, followed by ring opening and rearrangement to form a new alkene bond between C15-C16 and a primary alcohol on C17. Further analysis of beverage ingredients on the formation of compounds 1-3 indicated the addition of caramel color significantly increased the concentration of compounds 1-2, whereas the flavoring material (coffee) and caramel color were reported to significantly reduce the formation of compound 3 during storage. In summary, three Reb A degradation compounds were identified as key chemical changes during storage that impacted the flavor stability of beverage systems. Two mechanisms of degradation were reported, an acid-catalyzed carbocation intermediate pathway and an epoxidation/rearrangement pathway both involving the exomethylene functional group of Reb A. Further analysis also demonstrated the addition of caramel color and added flavor materials significantly altered the formation of these Reb A degradation compounds. Overall, this work provides a basis to improve the performance of Reb A by providing chemical markers to monitor Reb A flavor stability and an improved understanding of the mechanisms of degradation, ultimately to increase market utilization.

Increasing global demand for premium coffee beverages has challenged coffee producers to provide higher quality products. The overall aim of this project was to characterize the chemical composition of coffee brews that contribute to different flavor quality and ultimately to provide a chemical basis for the optimization of coffee production. During the initial phase of this study, current industrial methods for green bean selection were evaluated in regard to their ability to predict coffee brew sensory properties. Physico-chemical changes observed in the green coffee beans stored under different conditions were not indicative of coffee brew flavor changes in this study. Overall, these results supported the need to develop new methods for coffee quality prediction. In the second phase of this project, an untargeted Liquid Chromatography/Mass Spectrometry (LC/MS) Flavoromics approach was used to define chemical markers predictive of coffee brew quality. Eighteen coffees ranging in qualities from below specialty to excellent specialty quality, were chemically profiled and modeled against the official Specialty Coffee Association cup score (obtained by Q-graders) using Orthogonal Partial Least Square (OPLS). A successful model was generated and showed high prediction ability (R2Y=0.99, Q2=0.97). A total of 10 highly predictive compounds were selected, of which six were positively and four were negatively correlated to cup score. Their sensory impact was evaluated using a recombination study performed by five Q-graders. Addition of the compounds individually and in combination to a coffee brew base caused significant changes in cup scores. Three compounds predictive of high cup score: 3-O-caffeoyl-1-O-3-methylbutanoylquinic acid (m/z 437), 3-O-caffeoyl-1-O-3-methylbutanoyl-1, 5-quinide (m/z 419) and a chlorogenic acid derivative (m/z 671) significantly increased the quality of a below specialty grade coffee brew to excellent specialty grade. Four compounds predictive of low cup score; ent-16α,17-kauran-19-diglucoside (m/z 659), its aglycone form ent-16α,17-dihydroxy-kauran-19-oic acid (m/z 335), unknown compounds m/z 351 and m/z 9.48_333 significantly decreased the coffee quality score of a high cup score coffee. All the identified compounds had no or slight flavor activity when evaluated in water at the levels present in coffee. Yet they were able to significantly modulate the flavor of coffee, likely through the existence of perceptive interactions. The presence of these seven markers of coffee brew quality in the green bean was further investigated. Markers 3-O-caffeoyl-1-O-3-methylbutanoylquinic acid and ent-16α,17-dihydroxy-kauran-19-diglucoside were both detected in green coffee beans at levels correlated to the coffee brew cup scores. Overall, this research showed a successful application of a newer flavor analysis method, Flavoromics, that allowed for the discovery of seven flavor compounds predictive of the coffee cup quality. These compounds can be used as markers of quality for raw material selection and process optimization to achieve desirable coffee brew flavor.

Consumer acceptability, or “liking,” is a complex consumer reaction to food which impacts food-derived enjoyment, purchasing behavior, and satiety, and is thus of interest to the food industry as well as consumers and researchers. While numerous factors impact acceptability, food flavor plays an outsize role in determining if a food is liked or disliked. Traditional methods of analyzing flavor identify individual compounds in isolation and out of context, primarily due to instrumental limitations. These methods have limited capability to understand acceptability, itself more a consumer reaction to the holistic experience provided by food than to individual sensory attributes or impact compounds. Chemical profiling and multivariate modeling of metabolites to find compounds predictive of effects, termed “metabolomics,” has recently been applied to the study of flavor. The resulting technique, termed flavoromics, offers the unprecedented ability to directly tie the underlying chemistry of a food to acceptability and other complex sensory percepts through statistical modeling. Compounds predicted to be important to acceptability can be recombined with the original food system and tasted to determine if they drive liking or are merely correlated, providing crucial confirmation for further application and use of findings. While showing great potential for compound discovery, flavoromics remains an emerging technique. Within this dissertation, flavoromics was used across two studies to understand chemical drivers of liking in fruit spreads. Sugar-free fruit spreads have been observed to universally present atypical flavor defects not seen in traditional products and are less-liked than full-sugar spreads. To understand chemical drivers of positively perceived flavors in traditional spreads, flavoromics was used to find universal differences between sugar-free and traditional jams across eight fruit varieties through class-based modeling to remove the impact of fruit variety. Four non-volatile compounds predictive of traditional spreads as compared to sugar-free spreads, regardless of fruit, were identified and tasted. Two compounds, including one which had not previously been described for flavor activity, were revealed to have sensory impact which made sugar-free spreads taste more-similar to traditional spreads, although further sensory validation is needed. This study showed the ability of flavoromics to relate chemistry to flavor-relevant class-based differences regardless of orthogonal variations such as fruit type, and to uncover flavor-active compounds linked to an effect. This study will provide molecular targets to the jam industry for use in producing better-liked sugar-free products, helping dieting consumers achieve sugar-reduction goals without sacrificing flavor enjoyment. The second study focused on linking chemistry directly to consumer ratings of acceptability in strawberry preserves. Fifteen unique strawberry preserves were manufactured using an identical process. Preserves were profiled using untargeted GC/MS and LC/MS to capture flavor-relevant chemical information and were rated by a consumer acceptability panel. Collected chemical data was correlated to acceptability to uncover compounds directly related to liking. Four non-volatile and nine volatile compounds were identified as highly-correlated to acceptability, including two novel non-volatiles and one novel flavor active volatile. Compounds were recombined with spreads and rated by a consumer preference panel, revealing a bimodal distribution of preference which partially broke down by gender. It was found that women significantly preferred jams spiked with acceptability-linked non-volatile compounds over control jams, while men did not display a preference. Although more work is required to investigate causes of preference distributions, this work successfully showed the capability of flavoromics to identify compounds which drive liking in foods. This second study serves to provide crop breeders and jam producers with molecular targets for marker-assisted crop breeding, raw ingredient selection, and process optimization. Together, the two studies discussed in this dissertation present a proof-of-concept for the use of flavoromics to understand acceptability and open the door to greater use of the technique towards understanding drivers of complex sensory percepts in foods.

Flavor research has long utilized targeted methods to understand and identify flavor active compounds in food systems. While this has advanced the field, there are limitations to only investigating compounds that are directly flavor active. Flavor is a multi-model sensation consisting of a complex set of chemical stimuli that are perceived as a mixture, and more inclusive investigation can supplement targeted methods. Increasing the number of analytes allows for more comprehensive understanding of food systems which can lead to new discoveries. This dissertation illustrates the development of untargeted analytical methods to identify flavor active materials in citrus extracts that relate to aging. This dissertation uses an untargeted workflow to identify age related compounds, and establish their sensory significance through recombination modeling and structural elucidation. Preprocessing methods were optimized using an uneven multi-level two factor design of experiment and further untargeted modeling. Signal to noise ratio and thresholding were varied during pre-processing which produced data sets varying in size from >50,000 features to 500. Part per billion differentiation was achieved using both unsupervised (principle component analysis) and supervised (projection to latent structures) multivariate modeling, indicating a high quality analytical and statistical framework. To understand how a food system ages ethanol extracts from different citrus varietals (Navel, Mineola, and Valencia) were aged and chemically profiled using Ultra Performance Liquid Chromatography Mass Spectrometry. Machine learning (Random forest) and multivariate (projection to latent structures) models were implemented to model the age of the extracts. Varietal difference was leveraged and models were adapted to understand the age of the samples, rather than model the varietal differences. Statistically important compounds were isolated using food grade mass spectrometry directed fractionation. These isolated compounds underwent sensory evaluation using descriptive analysis in both a Solvent Assisted Flavor Evaporation (SAFE) extract of orange juice and a volatile orange flavor (VOF) model beverage. The isolated compounds showed significant impact in both tasting mediums. All compounds identified increased with sample age, and when evaluated in the SAFE extract showed significant decrease in orange character. Compound 413 showed a significant increase in cooked character and green bean character and a suppression of floral character. While compound 383 showed a significant increase in green bean character. In the volatile orange flavor (VOF) compound 413E2 and compound 457 showed an increase in sweetness over the control, which was the only noted change to the ‘taste’ attributes noted among the compounds isolated. Compound 383 showed a decrease in cooked character over the control. Compound 661 showed suppression of floral aroma over the sample blank. Compounds that positively correlated with age were reported to increase the cooked and green bean character and suppression of floral and orange character notes, which indicate degradation in flavor quality, or a deviation from fresh character. Structural elucidation using Nuclear Magnetic Resonance (1H, HMBC, TOCSY, HSQC) and accurate mass (TOF) revealed compound 693 was Nomilin 17-O-beta-D-glucopyranoside, while compounds 383 and 661 were shown to be novel compounds. The systematic name of compound 661 was (5-(((2R,3S,4R,5R)-4,5-dihydroxy-3-((3-hydroxy-3-methyl-5-oxohexanoyl)oxy)-6-(4-(3-hydroxypropyl)-2,6-dimethoxyphenoxy)tetrahydro-2H-pyran-2-yl)methoxy)-3,3-dimethyl-5-oxopentanoic acid. There are two 3-Hydroxy-3-methylglutaric acid units and a dihydrosinapyl alcohol moiety bonded to a sugar backbone. Compound 383 was identified as (3,5,5-trimethyl-4-((E)-3-oxo-4-(((2S,3S,4S,5S,6R)-3,4,5-trihydroxy-6-(hydroxymethyl)tetrahydro-2H-pyran-2-yl)oxy)but-1-en-1-yl)cyclohex-2-en-1-one) and suggested reaction product of a sugar and terpene moiety. The final piece of this research was to characterize aging in lemon extracts and identify contextual variable interactions associated statistically significant compounds. Six types of lemon extracts which were aged and profiled using Ultra Performance Liquid Chromatography Mass Spectrometry. The data was modeled using 9 different machine learning algorithms.. Random Forest produced the highest quality initial model, which went through further development and tuning. Within the random forest model number of trees, features tried at each node, max number of terminal samples were all optimized, as was the final model using both gini and entropy for decision criteria. The final model had a training fit of 0.951 and test score of 0.928(+/- 0.0049). Statistically important variables from this analysis were investigated for contextual data interactions that would illuminate data trends for further study. This approach aims to help provide additional value from untargeted flavoromics and better understand contextual interactions in data sets.

This thesis introduces an alternative approach to predict flavor and is referred to as flavoromics; this research adapts concepts and tools from metabolomics investigations. It is a non-targeted strategy combined with chemometrics which considers for study all (ideally) low molecular weight compounds in foods as candidate chemical stimuli in human flavor perception. The feasibility of flavoromics to predict the intensities of various flavor attributes was tested on mandarin juice. Forty-six juices were characterized by both instrumental and descriptive sensory analyses. Volatiles and non-volatiles were analyzed by headspace solid-phase micro extraction gas chromatography and solid-phase extraction ultra high performance liquid chromatography- time of flight mass spectrometry, respectively. The developed methods were a compromise between the number of compounds extracted and detected, throughput, and repeatability. The capability of distinguishing samples based on mass spectral information collected from the different instruments using chemometrics was confirmed. The descriptive sensory analysis of the mandarin juice samples revealed very different flavor profiles between and within hybrids and cultivars, and juices made from fruits with common genetic background tended to share some sensory characteristics. Compositional variations across mandarin juice samples and their sensory profile were correlated using partial least squares regression, from which different predictive models of sensory quality were developed. The explanatory and predictive performances of the models were improved when combining all instrumental data into one single data set as opposed to individual ones, indicating that each individual subset conveyed complementary information and that merging them improved the overall description of the sensory profile. The best PLS model was obtained with mid-level data fusion, for which a preliminary variable selection was done. The predictive power of the selected model was tested using a calibration and prediction sample sets. A fairly robust model was obtained and a strong relationship between instrumental and sensory measurements was observed. The resulting model showed that prediction of sensory scores was possible to a certain extent for a majority of the sensory descriptors, demonstrating the applicability of using a data-driven approach to predict flavor irrespective of whether the chemical identity of the instrumental signals was known or not.