Grapefruit Juice and Citrus Blossom Volatiles

Material Information

Grapefruit Juice and Citrus Blossom Volatiles
Jabalpurwala, Fatima
Place of Publication:
[Gainesville, Fla.]
University of Florida
Publication Date:
Physical Description:
1 online resource (127 p.)

Thesis/Dissertation Information

Doctorate ( Ph.D.)
Degree Grantor:
University of Florida
Degree Disciplines:
Food Science and Human Nutrition
Committee Chair:
Rouseff, Russell L.
Committee Members:
Sims, Charles A.
Marshall, Maurice R.
Ache, Barry W.
Graduation Date:


Subjects / Keywords:
Flavors ( jstor )
Grapefruit juice ( jstor )
Grapefruits ( jstor )
Juices ( jstor )
Mandarins ( jstor )
Predetermined motion time systems ( jstor )
Pummelos ( jstor )
Sulfides ( jstor )
Sulfur ( jstor )
Sulfur compounds ( jstor )
Food Science and Human Nutrition -- Dissertations, Academic -- UF
blossoms, citrus, grapefruit, pfpd, spme, sulfur
Electronic Thesis or Dissertation
born-digital ( sobekcm )
Food Science and Human Nutrition thesis, Ph.D.


GRAPEFRUIT JUICE AND CITRUS BLOSSOM VOLATILES A procedure to quantify the major volatile sulfur compounds (VSC's) in grapefruit juice (GFJ) has been developed for the first time. SPME headspace concentration conditions were optimized and coupled with a highly sensitive and selective sulfur specific detector. Over 30 VSC's were detected in GFJ's, of which 13 were positively identified. Three dissimilar GC column materials were employed to separate and characterize the early eluting VSC's. Quantification was achieved using three internal standards. Twenty VSC's were present in significantly higher concentrations (p < 0.05) in heated juices than in fresh juices. Principal component analysis (PCA) and discriminant analysis revealed unique volatile sulfur patterns which differentiated juices as fresh unpasteurized, pasteurized not from concentrate and reconstituted juice from concentrate. Hydrogen sulfide and dimethyl sulfide accounted for 78% of total variability. Five other VSC's helped distinguish between the three juice types. 1-p-menthene- 8- thiol, a grapefruit character impact compound, also increased dramatically with thermally processing and subsequent storage. Fifty one aroma active peaks were detected in fresh and canned reconstituted from concentrate juice using GC-Olfactometry. Six aroma peaks identified as VSC?s, including 1-p-menthene- 8-thiol, were perceived in higher intensities in canned juice. Four 'fresh grapefruit' or 'spicy pungent' smelling peaks were detected only in fresh juice and tentatively identified as nootkatone, 1,10- dihydronootkatone, beta- sinensal or isoeugenol and eugenol. Sulfur amino acids were determined to be precursors for specific GFJ VSC's. Methionine produced primarily methional and S-methyl methionine produced primarily dimethyl sulfide. Aromas of both were perceived more intensely in canned juice compared to fresh juice. Most monoterpenes, sesquiterpenes and 1-p-menthene-8-thiol are more highly retained in the pulp fraction of juice compared to the serum fraction. A study of intact blossom volatiles from 15 citrus cultivars using GC-MS, identified linalool, beta-myrcene, alpha-myrcene, limonene, (E)-ocimene, methyl anthranilate and indole as major volatiles. PCA revealed three widely separated clusters consisting of pummelo, mandarins and lemons- limes. Volatile composition of grapefruit blossom (hybrid of sweet orange and pummelo) was found more closely associated to sweet orange than to pummelo. ( en )
General Note:
In the series University of Florida Digital Collections.
General Note:
Includes vita.
Includes bibliographical references.
Source of Description:
Description based on online resource; title from PDF title page.
Source of Description:
This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Thesis (Ph.D.)--University of Florida, 2009.
Adviser: Rouseff, Russell L.
Electronic Access:
Statement of Responsibility:
by Fatima Jabalpurwala.

Record Information

Source Institution:
University of Florida
Holding Location:
University of Florida
Rights Management:
Copyright Jabalpurwala, Fatima. Permission granted to the University of Florida to digitize, archive and distribute this item for non-profit research and educational purposes. Any reuse of this item in excess of fair use or other copyright exemptions requires permission of the copyright holder.
Embargo Date:
Resource Identifier:
489203783 ( OCLC )
LD1780 2009 ( lcc )


This item has the following downloads:

Full Text




2 2009 Fatima A. Jabalpurwala


3 To my late father Mr. Abdulhussain Jabalpurwala


4 ACKNOWLEDGMENTS I would like to express my sincere gratitude to my advisor Dr. Rouseff, for his constant guidance, encouragement and support. Setting high standards by example, Dr. Rouseff has always motivated me to excel in my work. I have gain ed immensely from his knowledge and experiences in flavor chemistry. His multidisciplinary scientific appr oach to problem solving has driven me to think outside the box. He constantly challenged me to push the boundaries of flavor analytical techniques and concepts. Dr. Rouseff has been an excellent coach in transforming me from a fresh graduate student to a strong researcher. I am very thankful to him for the excellent opportunities in the past four years. My sincere gratitude also goes to University of Florida, Alumni Fellowshi p Foundation for providing financial support for the entire pe riod of my graduate studies. I would also like to extend my gratitude to my committee members, Dr. Charlie Sims, Dr. Barry Ache, Dr. Marty Marshall and initially Dr Murat Balaban for their guidance in this research. I have enjoyed learning advanced principles and applica tions in various specializations in Food Science from them. In particular, the se nsory evaluation and statis tics class of Dr. Sims, the instrumental analysis cla ss of Dr. Marshall and the food processing class taught by Dr. Balaban. I am also very thankful to Dr. Ac he for encouraging me to learn about the neurophysiological principles i nvolved in gustatory and olfact ory signal transductions and perceptions. It has given my unde rstanding of flavor science a w hole new dimension. I am very thankful to all my lab mates, John (Jack) Sm oot and June Rouseff for their friendship and kindness. I really appreciate Jack Â’s patience for training me on using various lab instruments. I have learned and applied his me ticulous methods in analytical chemistry, which enabled me to achieve more accuracy and reproducib ility of data. I admire June for being so positive and an enthusiastic human being. I have really enj oyed her friendship and am grateful for her


5 thoughtfulness and warmth. Thanks to her, Dr. R ouseff and Jack, I feel like I have a second family in the United States. I also wish to thank all my friends for ma king my stay in the US so enjoyable and memorable. Special thanks to my entire American and international friends, Juan Carlos, Rosalia and Juan Manuel for truly enhancing my diverse multicultural experiences. I will always cherish their friendship. Most importantl y, I would like to ac knowledge my best friend, Kamal. He has always been my inspiration, guidi ng me with his own experiences and challenging me to achieve the most difficult goals. He has helped me grow as a true professional. I owe a lot to him. My deepest gratitude goes to my family members for their unconditional love and for constantly standing by me. My biggest suppor t has been my mother, who has made many sacrifices to enable me, my brother and sister to achieve our dreams. My father taught us the value of humility and always emphasized that ther e was no substitute to hard work. Even with limited resources, my parents have provided us with the very best opportunities in life. I owe my achievements to them. To my grandfather, I w ould like to extend a special gratitude for his vision to see his granddaughters at tain highest honors in educati on, despite the traditional norms of my culture. I also want to acknowledge my grandparents and my Godparents for their unconditional love and ble ssings. Above all, I want to thank God for guiding my path always and for all the seemingly impossibl e miracles in my life!!


6 TABLE OF CONTENTS page ACKNOWLEDGMENTS ............................................................................................................... 4LIST OF TABLES ................................................................................................................ ...........9LIST OF FIGURES ............................................................................................................... ........10LIST OF ABBREVIATIONS ........................................................................................................1 1ABSTRACT ...................................................................................................................... .............12CHAPTER 1. INTRODUCTION ............................................................................................................ ......142. LITERATURE REVIEW ....................................................................................................... 17Grapefruit Juice Aroma ........................................................................................................ ..17Volatile Sulfur Compounds .............................................................................................18Analysis of Volatile Sulfur Compounds .................................................................................20Isolation and Concentration .............................................................................................20Distillation ................................................................................................................20Solvent extraction .....................................................................................................21Selective extraction of thiols ....................................................................................22Solid phase micro-extraction ....................................................................................23Detection ..................................................................................................................... .....24Flame photometric detector (FPD) ...........................................................................25Sulfur chemiluminescen ce detector (SCD) ..............................................................25Atomic emission detector (AED) .............................................................................26Pulsed flame photometric detector (PFPD) ..............................................................26Quantitation .................................................................................................................. ...27Thermal Processing Effects on Grapefruit Volatile Sulfur Compounds ................................28Precursors of Volatile Sulfur Compounds in Grapefruit Juice ...............................................29Sulfur Amino Acids .........................................................................................................29Cysteine-SConjugates ...................................................................................................30Aroma Distribution in Juice Matrix ........................................................................................31Citrus Blossom Volatiles ...................................................................................................... ..33Analytical Considerations ...............................................................................................34Chemotaxonomic Classification ......................................................................................34InsectPlant Interactions .................................................................................................353. SEPERATION AND IDENTIFICATI ON OF GRAPEFRUIT SULFUR VOLATILES USING SPME AND PULSED FLAME PHOTOMETRIXC DETECTION .........................36Introduction .................................................................................................................. ...........36


7 Materials and Methods ......................................................................................................... ..37Grapefruit Juice Samples .................................................................................................37Standard Chemicals .........................................................................................................37SPME Optimization .........................................................................................................38Fiber type ..................................................................................................................38Headspace atmosphere, extrac tion temperature and time ........................................39GC-PFPD Analysis ..........................................................................................................39Identification of VSC’s ....................................................................................................40Quantitation .................................................................................................................. ...40Statistical Analysis .......................................................................................................... 40Results and Discussion ........................................................................................................ ...41SPME Fiber Selection .....................................................................................................41Nitrogen Versus Air Versus Oxygen Headspace ............................................................42Time –Temperature SPME Optimization ........................................................................43Chromatographic Considerations ....................................................................................44Identification of Sulfur Peaks ..........................................................................................45Quantitation of Sulfur Volatiles ......................................................................................47Sulfur Volatile Differences Between Fresh and Canned GFJ .........................................48Conclusions ................................................................................................................... ..........494. PROCESSING PATTERNS, PRECUR SORS, AROMA ACTIVITY AND MATRIX DISTRIBUTION OF GRAPEFRUIT JUICE SULFUR VOLATILES .................................57Introduction .................................................................................................................. ...........57Materials and Methods ......................................................................................................... ..58Grapefruit Juice Samples .................................................................................................58Chemical Standards .........................................................................................................59Sulfur Characterization of Survey Samples ....................................................................60Thermally Accelerated Storage Study .............................................................................61Model Juice Precursor Study ...........................................................................................61GC-MS conditions ....................................................................................................62GasOlfactometry Study .................................................................................................62Volatile Matrix Distribution Study ..................................................................................63Statistical Analysis .......................................................................................................... 64Results and Discussion ........................................................................................................ ...64Sulfur Composition of Fresh and Processed Grapef ruit Juices .......................................64Thermally Accelerated Storage Study .............................................................................67Classification of Grapefruit Jui ces Based on Sulfur Composition ..................................68Nonvolatile Precursors of Volatile Sulfur Compounds ................................................69Comparison of Aroma Compounds in Fres h and Canned RFC Grapefruit Juice ...........72Matrix Distribution of Volatile Sulfur Compounds in Canned RFC Grapefruit Juice ....75Conclusions ................................................................................................................... ..........785. A COMPARISON OF CITRUS BLOSSOM VOLATILES ..................................................92Introduction .................................................................................................................. ...........92Materials and Methods ......................................................................................................... ..93


8 Sample Collection ...........................................................................................................93Headspace Volatile Analysis of Blossoms ......................................................................94GC/MS conditions ....................................................................................................94Peak identification ....................................................................................................94Statistical Analysis .......................................................................................................... 95Results and Discussion ...........................................................................................................95Blossom Volatile Composition ........................................................................................95MS identifications ....................................................................................................97Principal Component Analysis ......................................................................................100Possible Insect Plant Interactions ................................................................................102Conclusion .................................................................................................................... ........1036. SUMMARY ................................................................................................................. .........114REFERENCES .................................................................................................................... ........117


9 LIST OF TABLES Table page 3-1 Comparison of retention time beha vior of grapefruit su lfur volatiles on three different column materials .................................................................................................513-2 Quantitative analyses of sulfur volatiles in fresh and canned (reconstituted from concentrate) juices ........................................................................................................... ..524-1 Quantities in parts per billion (g /L) of volatile sulfur co mpounds in fresh, not from concentrate and reconstituted from concentrate grapefruit juices. ....................................794-2 Discriminant function analysis (f orward step) model for se gregation of fresh and heated juices. ................................................................................................................ ......854-3 Discriminant function analysis (f orward step) model for se gregation of fresh, not from concentrate and from concentrate juices. ..................................................................854-4 Aroma active compounds in Fr esh and Canned RFC grapefruit juice. .............................875-1 Citrus blossom cu ltivars included in this study. ..............................................................1055-2 Identifications of citrus blossom volatiles and their relative total ion current (TIC) peak area from eight citrus types. ....................................................................................1065-3 Average values with standard de viations for the 12 blo ssom volatiles found to be most differentiating using multiple analyses of variance. ...............................................112


10 LIST OF FIGURES Figure page 3-1 Comparison of average peak ar eas for volatile sulfur compounds in canned grapefruit juice using CAR/PDMS and DVB/CAR/PDMS SPME fibers at 40C. ...........533-2 Effects of different levels of headspace oxygen on canned grapefruit juice VSCÂ’s.. .......543-3 Effect of SPME exposure time (A) and temperature (B) on extraction efficiency of: DMS ( ), MeSH ( ), PMT ( ) and 3HMA ( ).. ..............................................................553-4 PFPD chromatogram of SPME headspace sulfur volatiles from canned grapefruit juice.. ....................................................................................................................... ...........564-1 Volatile sulfur comparison of unheat ed (0) and heated (95 C; 2, 6 and 15 hour) fresh hand squeezed Marsh grap efruit juices (n=3). ...................................................................814-2 Principal component analysis for VSC distribution in 10 fresh (), 11 NFC () and 9 RFC () grapefruit juices. ..............................................................................................824-3 Canonical roots from discriminant function analysis showing separation of unheated GFJÂ’s () from heated GFJÂ’s combined: NFC () and RFC (). ...................................834-4 Canonical roots from discriminant function analysis showing separation of unheated ( ), NFC () and RFC () GFJÂ’s. ....................................................................................844-5 Volatile sulfur breakdown products of sulfur containing amino acids heated (95C; 3 hr) in model GFJ system (pH 3.4).. ...................................................................................864-6 Aroma intensities of Fresh (grey bars) and Canned RFC (black bars) grapefruit juice volatiles ..................................................................................................................... .........894-7 Major volatile dist ribution in pulp (solid bars) and serum (bars with dots) fractions of canned RFC grapefruit ju ice analyzed using HSSPME coupled with GC-MS.. .........904-8 Volatile sulfur co mpound distribution in pulp (solid bars) and serum (bars with dots) fractions of canned RFC grapefruit jui ce analyzed using HSSPME coupled with GC-PFPD.. ..................................................................................................................... ....915-1 Chemical composition of citrus blossom headspace volatiles..........................................1095-2 Eigenvector values of PC 1vs PC 2 from PCA of 70 vo latile components. .....................1105-3 Load plot. Eigenvector valu es of PC_01vs PC_02 from PCA of 70 volatile components .................................................................................................................... ..111


11 LIST OF ABBREVIATIONS 2MTP 2methyl thiophene 3MTP 3methyl thiophene 4MMP 4mercapto-4-methyl-2-pentanone 4MMPol 4mercapto-4-methyl-2-pentanol 8MPM 8mercapto-p-menthan-3-one BMFDS Bis (2-methyl-3-furyl) disulfide CS2 Carbon disulfide Cys Cysteine DMS Dimethyl sulfide DMDS Dimethyl disulfide DMTS Dimethyl trisulfide EPM 2, 8-epithio-cis-p-menthen GC-O Gas chromatographyolfactometry GFJ Grapefruit juice Glu Glutathione H2S Hydrogen sulfide MeSH Methanethiol Met Methional Meth Methionine PFPD Pulsed flame photometric detector PMT 1-p-menthene-8-thiol SMM Smethyl methionine SO2 Sulfur dioxide VSC Volatile sulfur compound


12 Abstract of Dissertation Pres ented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy GRAPEFRUIT JUICE AND CITRUS BLOSSOM VOLATILES By Fatima Abdulhussain Jabalpurwala August 2009 Chair: Russell Rouseff Major: Food Science and Human Nutrition A procedure to quantify the major volatile su lfur compounds (VSC’s) in grapefruit juice (GFJ) has been developed for the first time. SPME headspace concentration conditions were optimized and coupled with a highly sensitive and selective sulfur sp ecific detector. Over 30 VSC’s were detected in GFJ’s, of which 13 we re positively identifie d. Three dissimilar GC column materials were employed to separate and characterize the early eluting VSC’s. Quantification was achieved using three internal standards. Twenty VSC’s were present in significantly higher concentrations (p < 0.05) in heated juices th an in fresh juices. Principal component analysis (PCA) and discriminant anal ysis revealed unique volatile sulfur patterns which differentiated juices as fresh unpasteu rized, pasteurized not from concentrate and reconstituted juice from concen trate. Hydrogen sulfide and dime thyl sulfide accounted for 78% of total variability. Five other VSC’s helped di stinguish between the three juice types. 1-pmenthene8thiol, a grapefruit character imp act compound, also increase d dramatically with thermally processing and subsequent storage. Fifty one aroma active peaks were detected in fresh and canned reconstituted from concentrate juice using GC-Olfactometry. Six ar oma peaks identified as VSC’s, including 1-pmenthene8-thiol, were perceived in higher intensities in canned ju ice. Four ‘fresh grapefruit or


13 spicy pungentÂ’ smelling peaks were detected only in fresh juice and tentatively identified as nootkatone, 1,10dihydronootkat one, betasinensal or isoeugenol and eugenol. Sulfur amino acids were determined to be prec ursors for specific GFJ VSCÂ’s. Methionine produced primarily methional and S-methyl meth ionine produced primarily dimethyl sulfide. Aromas of both were perceived more intensely in canned juice compared to fresh juice. Most monoterpenes, sesquiterpenes and 1-p-menthene-8 -thiol are more highly retained in the pulp fraction of juice compared to the serum fraction. A study of intact blossom volatiles from 15 citrus cultivars using GC-MS, identified linalool, beta-myrcene, alpha-myrcene, limonene, (E )-ocimene, methyl anthranilate and indole as major volatiles. PCA revealed three widely sepa rated clusters consisti ng of pummelo, mandarins and lemonslimes. Volatile composition of grap efruit blossom (hybrid of sweet orange and pummelo) was found more clos ely associated to sweet orange than to pummelo.


14 INTRODUCTION The pleasant aroma of citrus juices and blosso ms are highly desirable and find applications in the food, flavor and fragrance industries. Th e United States is the worldÂ’s second largest producer of oranges and grapefruits ( 1 ). Grapefruit ( Citrus paradisi Mcfayden) has a unique flavor due to a combination of sw eet-tart and slight bitter taste w ith a characteristic aroma. This aroma which differentiates grapefruit from ot her members of the citrus family has been attributed mainly to volatile sulfur compounds, VSCÂ’s. VSCÂ’s exist at ng/Kg levels in grapefruit juices. Therefore, their identification and quantification using conven tional aroma isolation and detection procedures is extremely difficult. The ideal technique for volatile su lfur analysis has to be selec tive and avoid artifacts, a common problem associated with these compounds. In or der to understand the sensory impact of these compounds, it is also important th at the method is sensitive an d enables quantification below threshold levels. We therefore propose to couple a solid phased microe xtraction (SPME) method with a sulfur specific detector such as a pulsed flame photometric detector (PFPD). SPME which is a rapid and solventless method that should pr ovide the necessary concentration levels, and PFPD should provide the sensitivity and selectivity needed for obt aining comprehensive analysis of a wide range of grapef ruit sulfur volatiles. Significant efforts have been ma de by previous researchers to investigate the effects of thermal processing on major volatile compounds in citrus juices ( 2-10 ). However, few studies have reported these effects on the sulfur composit ion of citrus juices, including grapefruit juice ( 8, 10, 11 ). This is of particular significance to grapefruit juice, where sulfur compounds contribute to its characteristic aroma. Some of the preliminary work in our laboratory has indicated that certain vo latile sulfur compounds reported in grapefruit juice may be generated during thermal processing. Nonvolatile precursors such as sulfurcontaining amino acids have


15 been proposed in the formation of volatile sulfur compounds in citr us juices. The exact contribution of these precursors to grapefruit ju iceÂ’s volatile sulfur composition remains to be established. In this study, we hypothesize that vo latile sulfur composition of fresh unheated and heat treated grapefruit juices may be very differe nt and would play a sign ificant role in defining the aroma composition of these juice types. Als o, different types of sulfur amino acids would differ in their thermal breakdown volatile sulfur products and thereby in their contribution to processed grapefruit juice aroma. The main objectives of this research were to (a) develop a routine procedure to quantify a wide range of VSC in GFJÂ’s (b) determine proce ssing related changes in gr apefruit juice sulfur aroma composition (c) identify sulf ur volatiles generated from sp ecific sulfur containing amino acid precursors and (d) determine VSC distribution within the juice matrix. The final portion of this re search was aimed at compari ng the volatile composition of grapefruit blossoms with other citrus cultiv ars. Although citrus blossoms are known for their pleasant, highly desirable aroma, few studies ha ve analyzed the volatil e composition of intact citrus blossoms. Historically, citrus blossom composition has b een studied using blossom oils obtained from steam distillates of the blossoms. It is well recognized that the heat applied during distillation process produces oils whose smell bear only limited resemblance to the natural scent of blossoms ( 12, 13 ). By employing solid phase microext raction, we hypothesized that the SPME headspace volatiles will not contain thermally ge nerated artifacts that are generated during the steam distillation volatile extraction process. Volatile secondary metabolites in citrus tree fruit peel oil ( 14, 15 ), as well as Petitgrain (leaf) and Neroli (sour orange blossom) oils ( 14 ) have been explored in devising taxonomic classifications of citrus cultivars. We hypothesi ze that citrus blossom volatile composition would


16 be characteristic of each cultivar. The primar y objective of this study was to examine headspace volatile composition of citrus cultivars blossoms using headspace SPME coupled with GCMS. A secondary goal was to explore the possibility that blossom vol atiles from individual cultivars were sufficiently unique that they could have chemotaxonomic potential or potential biological significance.


17 LITERATURE REVIEW Grapefruit Juice Aroma The unique flavor of GFJ is due to a combinati on of sweet-tart and slight bitter taste with a characteristic aroma which differe ntiates this fruit from other members in the citrus family. A low Brix/ acid ratio, particularly in early season grapefruit, is related to increased tartness and decreased consumer acceptability ( 16, 17 ). However, this ratio appears to play only a minor role in defining the overall flavor qua lity of grapefruit juice ( 17 ). Bitterness levels appear to be more important to consumer acceptance. Naringin, a flavanone neohesperidoside and limonoids such as limonin and nomilin are the major contributor s to bitterness in GFJ. In general, the combination of low sugar acid ratios and high bitterness can overpower the pleasant aroma of grapefruit juice. Considerable efforts have b een made to de-bitter GFJ to increase consumer acceptance ( 18 ). GFJÂ’s volatile aroma composition and subse quent changes related to processing and storage has been a subject of research since the early 1950Â’s. The work of Kirchner et al. ( 19 ) revealed that canning and prolonge d storage of canned juice increas ed volatile acids, methanol and furfural levels in the water soluble volatile s (essence) extracted from GFJ. In case of the essential oil extracted from these juices ( 20 ), they observed significan t decrease in limonene levels and increase in linalool monoxide, -terpineol and furfural le vels in the stored canned juice. These compounds when added to fresh juic e contributed an undesira ble flavor quality to juice. However, the technology at the time (1953) did not provide the ability to make certain that the added compounds did not contain impurities whic h may have contributed to the off flavor instead of the labeled chemical. In the 1960Â’s, nootkatone, a grapefruit peel oil constituent, was the first compound reported to possess a characteristic grapefruit arom a. This compound was initially detected as a


18 prominent GC peak in highly flavored grapefruit oil but was less apparent in weak flavored oils ( 21 ). Erdtman and Hirose ( 22 ) identified this peak as a sesq uiterpene ketone, nootkatone, based on the melting point and infrared spectrum. Macleod and Buiges ( 23 ) further established nootkatone as a native component of grapefruit juice, by anal yzing juice which was carefully separated from its peel-oil. The authors stated that the taste and odor of nootkatone was definitely recognizable as contributing to the di stinctive grapefruit flavor and its level in GFJ were typically at or above the taste and odor thresholds as determined by them. It was subsequently observed that the mother liquor from which nootkatone was crystallized retained much of the grapefruit flavor and at a higher intensity than that of pure nootkatone itself ( 24 ). Subsequent taste and odor panel studies ( 25 ) further revealed that the level of nootkatone in grapefruit o il was more important to the aroma of oil than to the flavor of grapefruit juice containing the same oil. All thes e studies led to the understanding that in addition to nootkatone, there were other related components pr esent in grapefruit that could contribute to its distinctive flavor. This hypothesis was later proved by the id entification of fifteen related sesquiterpene ketones from GFJ using MS and NMR spectra ( 26 ). Amongst these, 8, 9didehydronootkatone was found to possess a more intense grapefruit aroma than nootkatone at equimolar concentrations. Another compound 1, 10dihydronootkatone, was previously reported to be approximately 3.5 times more intense than pure nootkatone ( 24 ). However, their relative concentrations in GFJ with re spect to nootkatone were very small and hence their actual contribution to GFJ aroma remains to be established. Volatile Sulfur Compounds A new and probably more important class of ch emicals in GFJ aroma, to emerge in the early 1980Â’s, were the sulfur c ontaining volatile compounds. Of pa rticular significance was the identification of 1-p-menthene-8-t hiol (PMT) by Demole et al. ( 27 ). The authors isolated this


19 terpene thiol and its bicyclic epimer 2, 8-epithi o-cis-p-menthane (EPM) from a sulfurous heptane fraction from 100L of canned GFJ which had a char acteristic fresh grapefruit smell. The authors reported PMT as the most potent aroma compound in nature with a taste threshold of 0.1 ng/L in water (the lowest known taste threshold at the ti me). EPM, on the other hand, was found to be a less significant contributor to grapefruit aroma (tas te threshold as high as 9 g/L in water). Few commercial juices contain EPM at levels above its threshold. Recent studies employing HRGC Olfactometry and CI-MS ( 28, 29 ) have afforded identification of a new mercaptan in GFJ: 4mercapto-4-methylpentan-2-one (4MMP). This compound which was originally det ected in sauvignon blanc wine ( 30 ), is reminiscent of intense black current or catty odor but in combination with other odorous compounds in grapefruit, it is reported to produce an intense grapefruit-like aroma quality. With aroma extract dilution analysis, AEDA and flavor reconstitution model st udies, the authors suggest ed that the typical aroma of hand-squeezed grapefruit-juice is not only due to PMT, but is evoked much more by 4MMP. Other sulfur compounds which have been reported as important odorants in grapefruit juice are hydrogen sulfide (H2S), dimethyl sulfide (DMS), 4-mercapto-4-methyl-2-pentanol (4MMPol), 3-mercapto-1hexanol (3MHol), 3-me rcapto hexyl acetate (3MHA) and methional (Met) ( 10, 31-33 ). However, comprehensive information on the relative concentrations of these potent aroma compounds in grapefru it juice is still lacking. The presence of thes e volatiles at ng/L levels coupled with their highly reactive na ture (easily oxidized, dimerized or thermally/ enzymatically altered) has made their extracti on and analysis difficult and prone to artifact formation ( 34 ). By using a highly sensitive and specific sulfur analytical method, it should be possible to quantify levels of these potent compounds in grapef ruit juice and thereby establish their sensory significance to grapefruit juice aroma.


20 Analysis of Volatile Sulfur Compounds Analysis of sulfur volatile compounds in foods poses a unique challenge to flavor chemists. This is primarily because VSCÂ’s are extremely aroma potent and usually perceived at levels much below the detection limits of conve ntional analytical methods. In addition, these compounds present diverse structural chemistry and polarity in foods ranging from thiols, sulfides, polysulfides, thiophene s, thiazoles and other heteroat omic forms. This limits their simultaneous isolation, concentration and quantif ication using one single method. The two major approaches typically used for trace level analysis (g/ L) of these compounds are (a) extraction and substantial concentration of these compounds from the sample matrix and (b) sensitive and sulfur specific methods of detection. Isolation and Concentration The conventional method for sensitive analysis of sulfur volatiles has been isolation and concentration of these com pounds from large sample volum es through solvent extraction, vacuum distillation or resin adsorption. Various food matrices can pose unique challenges and sample specific extraction method ma y be required. Some of the widely used methods as well as some of the more recently developed sample ex traction techniques are presented below. Whereas most involve non-specific isolat ion methods, some have been a pplied for specific extraction of thiols. Distillation This is one of the easiest methods for isolat ing a wide boiling range of aroma compounds from complex food matrices. However, thermal ar tifacts are a common drawback of this method which involves heating sample for prolonged durations at high temperatures. This is mainly true for sulfur compounds, which are extremely reactive in nature and prone to thermal degradation. Vacuum assisted distillation, on the other hand, can largely overcom e this drawback and improve


21 sensitivity for sulfur volatiles. By using high vacuum distillation, Werkhoff et al. ( 35 ) isolated several character impacting sulfur volatiles from yellow passion fruit. However, the extensive equipment and time required for this method make it unsuitable for routine analysis. Solvent extraction This is, by far, the most widely used isolation procedures for sulfur volatiles in GFJ. Most procedures typically involve extr action of large sample volumes after which the organic extract is concentrated (by gentle dist illation) to increase the concentration of sulfur aroma compounds. Simultaneous distillationextract ion (SDE) and continuous liquidliquid extractions (LLE) are common variants of this procedure. In one such study ( 27 ), 1-p-menthen-8-thiol, PMT and 2,8epithio-cis-p-menthane, EPM were concentrated from 100L of canned GFJ using vacuum steam distillationheptane extracti on at 52C for 24 hours. The cr ude extract was subjected to evaporation at 50C followed by column chro matography to obtain a sulfurous fraction for subsequent detection and identification us ing GC-MS. More recently, 4-mercapto-4methylpentan-2-one (4MMP) and methional were extracted from five liters of GFJ after six hours of diethyl ether extraction at 34C followed by concentration on a Vigreux column at 38C ( 29 ). This concentration step enabled g/ Kg detection of sulfur compounds using two dimensional HRGC coupled with CI -MS in the selective ion mode. One major concern with solvent extraction proc edures is the uneven ex traction efficiencies. This is due to the structural diversity exhibited by sulfur compounds, ranging widely in their physical and chemical properties. Use of multiple internal standards or deuterated standards (as in isotope dilution analysis) or continuous liquid – liquid extractions for se veral hours, have been explored to compensate for une ven extraction efficiencies ( 34 ). Alternatively, some authors have employed a combination of solvent systems of different polarities to enhance extraction


22 efficiencies of wide chemical ra nge of grapefruit juice volatile s including some of the character imparting VSCÂ’s ( 10, 33 ). Another concern with using so lvents is the chromatographic limitation. The intense and broad solvent peak prevents chromatographic resolution of early eluting compounds from the solvent, making their separation, identification and analysis difficult. For example, in the above mentioned studies involving grapefruit juice VSCÂ’s ( 10, 27, 29, 33 ), the early eluting sulfur volatiles such as H2S, DMS, SO2 and COS could not be determined as these compounds typically elute at or near th e solvent front. Further, large sample volume concentration procedures are extremely time intensive. In addition, prolonged heating (even at moderate temperatures) during solvent extraction and concen tration can lead to severe alterations among highly reactive sulfur compounds. A typical out come could be loss of low boiling volatile compounds like H2S, methanethiol (MeSH) and DMS or oxidation of MeSH to dimethyl disulfide (DMDS). Due to th e unstable nature of VSCÂ’s, accurate quantification becomes extremely difficult using solv ent extraction procedures. Selective extraction of thiols Thiol specific analytical methods using a se lective chemical reaction has been employed for sensitive detection of aromatic thiols in wine ( 36, 37 ). The procedure involves a reversible reaction between a free thio l (-SH) group and organic mercury. Bouchilloux et al. ( 36 ) employed aqueous p-hydroxy mercury benzoate (pHMB) to concentrate 4MMP from purged headspace of Sauvignon wine (2L). Excess of cysteine or glutathi one was then used to re volatilize the thiol in solution. The method was prone to interferences and column chro matographic procedures were later employed to eliminate impurities ( 37, 38 ). Although powerful in obtaining purified extracts, these procedures are time consuming and could only trap and concentrate free thiol containing compounds. In addition to thiols, GFJ comprises several structurally di verse sulfur compounds


23 such as sulfides, polysulfides, and other hetero atomic compounds which cannot be concentrated using this method. More recently, derivatizing agents such as pe ntaflurobenzyl bromide (PFBBr) have been used for sensitive ng/ L analysis of poly-functional thiols including 4MMP in wine using negative chemical ionization mass spectrometry ( 39 ). Once again, this method is selective for thiols and would not provide a comprehensive detect ion of all sulfur compounds present in GFJ. Solid phase micro-extraction Solid phase microextraction (SPME) technique was develo ped by Arthur and Pawliszyn ( 40 ) as a rapid and solvent free extracting method fo r organic analytes. This technique combines isolation and concentration in a single step making it a relatively rapid method for sensitive analysis. The basic principle involves using small quantities of highly adsorbent extraction material (~ 1ul) coated on a fused silica fiber mo unted to a syringe like device. In the headspace SPME method, the fiber is exposed directly in to the sample headspace and volatiles from the food are adsorbed/ absorbed on the coated materi al. Two equilibria of volatiles exist: food/ headspace and headspace/ adso rbent and the amounts of volat iles extracted depend on their partition coefficients between these phases. Equi librium time for the less volatile compounds can be significantly shortened by agitation and by in creasing sampling temperature. Volatiles from the fiber coating are desorbed by in serting the fiber into the heated GC injector. A wide range of adsorbant materials (with different polarities) ar e available from Supelco (Bellefonte, PA, USA). The coatings are available individually or in co mbinations to provide selectivity for a broad range of analytes. SPME coupled with a highly sensitive and sulfur specific dete ctor, pulsed flame photometric detector (described la ter) has been used specifically for volatile sulfur analysis in various food systems. For e.g. beer ( 41 ), irradiated turkey ( 42 ), wine ( 43 ) and recently in


24 thermally processed milk ( 44 ). Most of these studies ( 42-44 ) were aimed at obtaining quantitative extraction of the low molecular weig ht (and highly volatile) sulfur compounds such as H2S, MeSH, CS2, and DMDS. By using pre-equilibri um conditions (15 min exposure at 30C), they were able to ra pidly and selectively extrac t these low boilers. Murray ( 45 ) demonstrated that the initial rate of adsorption by the SPME fiber is inve rsely proportional to the molecular weights, MW of the analytes. As equi librium sets in with increasing duration, the low MWs get gradually replaced by the high MWs. C onversely, equilibrium conditions with longer exposures (32 min exposure at 45C) were used to quantify semi-volatiles in beer with higher molecular weights ( 41 ). Similarly, SPME parameters may be optimized to enable a comprehensive profiling of a wide range of VSCÂ’s in GFJ. Although, SPME offers high sensitivity over a wide range of analytes, artifacts may still be a problem when using this sample concentra tion method. These include on fiber oxidation of MeSH to DMDS and DMS to DMSO either during extraction ( 46 ) or during the thermal desorption of SPME fiber in the GC injection port ( 47 ). More recent studies have minimized this problem with use of inert atmosphere during samp ling and/ or inactivation of GC injection liners ( 43, 44, 48 ). Detection Some of the widely used detection technique s such as flame ionization detector, FID and mass spectrometry, MS have largely faile d in detection of VSCÂ’s in GFJÂ’s ( 49, 50 ). Cadwallader and Xu ( 49 ) extracted 21 volatile components from small volumes (2 mL) of fresh GFJ using purge and trap coupled with GC-M S and GC-FID. However, no sulfur compounds were reported. Jella et al. ( 50 ) identified 52 volatile compounds in methylene chloride extracts of 8mL of GFJ using GC-MS. Once again, no su lfur volatiles were reported. Such nonspecific detection techniques, therefore, re quire large sample volumes to be extracted and concentrated in


25 order to provide concentrations above detection limits. Typically, sample concentrations from 3100L of juice have been employed for MS detection of some GFJ VSCÂ’s ( 27-29 ). On the other hand, use of sulfur specifi c and highly sensitive detectors: SCD ( 33 ), FPD ( 36 ), PFPD ( 44 ) and AED ( 51 ), have greatly reduced detecti on limits for VSCÂ’s in various food systems. Each of these detectors have been described in detail by Rouseff ( 34 ) and are briefly summarized below. Flame photometric detector (FPD) FPD is one of the first detectors to be deve loped for selective and sensitive analysis of sulfur compounds. It is based on the pr inciple of formation of excited S2 species by utilizing a H2/ air flame. The excited S2 species emit energy in the form of broadband emission spectra at a maximum wavelength of ~394 nm, which is measured using a photomultiplier tube. Trace levels (ng/L) of 4mercapto-4-methyl-2-pentanone were quantified in 500 mL of Sauvignon wine samples ( 36 ), by coupling dynamic purge and trap method with a thiol specific chemical reaction (using p-HMB) and flame photometric detection. The main limitation of this method is the redu ced sulfur response due to co-eluting, higher concentration hydrocarbons which quench the emitte d energy from excited sulfur. In addition, the response to the sulf ur concentration is exponential (i.e pr oportional to the square of sulfur concentration) due to emission from a bimolecular S2 species. This results in larger sulfur peaks being completely off-scale before the smaller pe aks can be seen. A square root mode for the sulfur emission output can be empl oyed to correct this problem. Sulfur chemiluminescence detector (SCD) This is based on the principle of thermal degr adation of all sulfur species into sulfur monoxide. This further reacts with ozone forming an excited sulfur dioxide (SO2) that emits light measured by a photomultiplier tube. This techni que has the advantage of linear response to


26 sulfur concentration. In additi on, quenching is largely avoide d by employing vacuum for the photochemical reaction. Lin et al. ( 33 ) demonstrated the presence of five sulfur containing volatile compounds in solvent extracted GFJ ( 25mL) using SCD. These were tentatively identified as 1-p-menthene-8-thiol, 4 –merpcat o-4-methyl -2-pentanone, methional, 3 mercapto hexyl acetate and 3-SH hexanol. One major difficu lty with the SCD is the lack of long term stability for the conversion of organic sulfur vol atiles into sulfur m onoxide which impacts the reproducibility of the detector response. Atomic emission detector (AED) This detects the emission from plasma exci ted sulfur atoms rather than hydrogen flame excited molecular species as seen for the other dete ctors. It has the advant age of lowest detection limits and a linear response to a wide range of su lfur concentrations. Howe ver, the high cost of this detector limit its use. Pulsed flame photometric detector (PFPD) PFPD overcomes the limitations of FPD and is mo re sensitive and selec tive. It has rapidly found application as a sulfur-specific detector Whereas flame photometric detection uses a continuous flame, PFPD uses a pulsed flame th at ignites 2 to 4 times per second to burn compounds eluting from the gas chromatography (G C) column. By setting a timed “gate delay” for the response, the luminescence emission of hydr ocarbons can be separated from that of later emitting sulfur compounds. The gated amplifiers r ecord light only for a specific time interval between 524 milliseconds, when emission from exc ited sulfur species takes place. Thus, the PFPD produces cleaner chromatograms than the flame photometric detector, making it possible to detect sulfur compounds at lo wer concentrations. This is more stable as compared to SCD and has the potential of offering detection limits almost as low as AED ( 34 ). The detector provides a square root option that makes the response linear. Although this is true for a limited


27 concentration range, PFPD has been accepted as the most practical and cost effective method for detection of trace quantit ies of sulfur containing volatiles. In the past decade, a combination of sensitive SPME extraction and sulfur specifi c GC-PFPD detection has been extensively employed for trace level (ng/L) VSC analysis in various food systems ( 39, 41-44 ). Quantitation Quantitation is typically achieve d by comparing the instrument al response (peak area) of an analyte with that of a st andard of known concentration. Th e simplest method involves using an external standard calibration method. It is well known that the analytefood matrix interactions as well as sampling method can c ontribute to uneven extrac tion efficiencies of analytes. This generally makes external standa rd calibration a poor choice for trace analyses of volatile sulfur compounds in foods which demand high accuracy and precision. As opposed to external standard method, in ternal standard method involves adding a standard of known concentration di rectly to the sample, which is then simultaneously extracted along with the analytes. This method can la rgely compensate for the uneven extraction efficiencies of analytes. Typical ly, a standard of similar physical and chemical properties as the analyte of interest and is not present in the sample is employed to closely match its extraction efficiency. For this purpose, deuterated standa rds (or stable isotopes) of the analyte compound make the best choice. Using stable isotope dilution assay, Buettner and Schieberle ( 29 ) were able to quantify trace levels of 4MMP (0.8 ug/L) a nd PMT (0.01 ug/L) in hand squeezed GFJ after extensive liquid-liquid extraction. The main lim itations however are th e cost, stability and availability of these isotopes. This procedure is also impractical for analys es of large number of analytes. A more practical approach for simultaneous an alysis of wide range of analytes is to employ multiple internal standards. In a recent study of wine sulfur volatiles using HS-SPME


28 ( 43 ), differences in SPME responses were reporte d for the different sulfur compounds. By using multiple internal standards, the authors were ab le to largely compensate for these differential responses and avoided errors th at would have occurred if only a single internal standard was employed. Thermal Processing Effects on Grap efruit Volatile Sulfur Compounds Impact of thermal processing on volatile com position and resulting aroma quality has been extensively studied in many food sy stems. Some of the characteri stic aroma compounds of fresh baked bread (thiazoles, thiazoline s), cooked meat (2-methyl-3-furna thiol and bis 2 methyl 3 furyl disulfide) and roasted co ffee (2 furfuryl thiol) are sulfur compounds. These are thermal products generated during a series of chemical reactions or Maillard type r eactions involving sulfur containing amino acids or vitamins Alternately, many of the offflavors associated with heat processed foods are also produced by sulfur-vol atiles. For example, the sulfurous cabbage offflavor defect of UHT processed milk was attribut ed to increased levels particularly of DMS, MeSH and DMTS ( 44 ). Significant efforts have been made to inve stigate the effects of thermal processing on major volatile compounds in citrus juices ( 2-10 ). However, few studies have reported these effects on the sulfur compos ition of citrus juices ( 8, 10, 11 ). This is of particular significance to grapefruit juice, where sulfur co mpounds are described as main c ontributors to its characteristic aroma. In a early 1980Â’s study ( 11 ), using packed column GC coupled with flame photometric detector, the authors compared headspace levels of sulfur compounds in fresh and commercially processed grapefruit juices. Th e authors reported significant amounts of one sulfur compound, tentatively identified as H2S, in freshly squeezed juices but none in the heat processed juices. Another sulfur compound, tentatively identified by these authors as DMS, was reported only in the heat processed juices.


29 More recently, GColfactometry technique was employed to determine the impact of thermal concentration on grap efruit juice aroma compounds ( 10 ). Substantial losses in ‘fresh/ citrusy’ odorants (octanal, nonanal, limonene) a nd ‘grapefruit/ sulfury’ smelling odorants (1,10dihydronootkatone, 3mercapto hexyl acetate, 3mercapto hexanol and 4mercapto4-methyl2pentanol) were reported in juice reconstituted from concentrate, with no restored flavors. It is interesting to note that aroma of the major ch aracter impacting 1-p-me nthene-8 –thiol was detected only in heated juice and not in fresh juice. The authors attributed its formation to thermal reaction between limonene and free H2S. Unfortunately, aroma activity of H2S and DMS, reported earlier by Shaw and Wilson ( 11 ), could not be established in the GC-O study ( 10 ) which employed solvent extraction for isolati on of aroma compounds. Bo th these compounds are extremely volatile and therefore minimally retain ed by GC columns and typically elute at the solvent front which limits their detection. Precursors of Volatile Sulfur Co mpounds in Grapefruit Juice Several precursors of volatile sulfur compounds have been proposed in citrus juices. These primarily comprise sulfur containing amino acids ( 8, 10, 11, 33 ) or vitamins ( 8, 52 ) and more recently cysteine-Sconjugates ( 53 ). In addition, certain volat ile sulfur compounds could undergo breakdown, oxidation or polymerization to form other volatile sulfur compounds in citrus juices ( 8, 32 ). Sulfur Amino Acids Sulfur containing amino acids are well known for their role as nonvolatile precursors in formation of many potent volatile su lfur compounds in foods. In ci trus juices, the formation of H2S, MeSH, DMS, DMDS, DMTS, Met and PMT has been largely attributed to sulfur amino acids ( 8, 10, 11, 33 ). Cysteine and glutathione were reported in citrus juices in the early 1950’s using paper chromatography ( 54 ) and subsequently in the la te 1970’s using HPLC with a


30 mercury based electrochemical detector ( 55 ). Trace levels of methioni ne were also reported in citrus juices ( 56 ), using ionexchange separation and ninhydrin detection. S-methyl methionine (SMM) a known precursor of dime thyl sulfide (DMS) has been reported mainly in mandarin juices ( 6, 57, 58 ). In these studies, SMM was shown to readily degrade to DMS during thermal processing or storage of manda rin juices. Belitz et al. ( 59 ) suggested that SMM could be formed from methylation of methionine during heat treatment in the pres ence of pectin. This in turn degraded to form DMS. Information on the amount s and/ or relative proportions of these amino acids is sparse, with only the 1950Â’s ( 54 ) study reporting 0.05mM cysteine and 0.19mM glutathione in GFJ. Also, there is no conclusive information on th e specific products of thermal breakdown of each of these amino acids under grap efruit juice conditions (pH < 4.0) and hence their unique contributions to grapefruit juice sulfur aroma remains to be established. Model food systems containing sulfur amino aci ds have been employed in the past to determine products from high temperature (180 C ) breakdown of these amino acids ( 60-62 ). In addition, these authors also investigated the eff ects of other factors su ch asreducing sugars, oxidative conditions and pH e ffects on amino acid thermal breakdown. By studying the thermal reactions between cysteine/ glutathione and gluc ose in model systems at pH 6 and above, these authors established the formation of some of th e typical Maillard sulfur flavors found in cooked meat ( 60 ). Cysteine-SConjugates In the past decade, significant research on wine sulfur aromahas led to the discovery of a new class of non-volatile precursors ( 63-69 ) called cysteinylated c onjugates or cysteine-Sconjugates. These nonvolatile compounds contai n an aromatic thiol which is bound to a cysteine molecule at the sulfur position. Cyst einylated precursors of mercaptans reported in Sauvignon blanc grape variety include cysteine Sconjugates of 4-mercapto-4-methyl-2-


31 pentanone, 4-mercapto-4-metyl-2-pentanol, 3me rcaptohexyl acetate and 3mercaptohexanol ( 63, 64, 67 ). Cysteinesconjugate precursors have also been identified in other fruit systems. These include the cysteine conjugate of 3-mercaptohexanol in passion fruit ( 70 ) and more recently S-(3-hydroxy-1,1-dimethylpropyl)-Lcyste ine, S-[3-(acetyloxy)-1,1-dimethylpropyl]-Lcysteine, and S-[1-(2-hydroxyethyl )butyl]-L-cysteine in Poncir us trifoliata (L.) Raf. ( 53 ), which is closely related to citrus. Thes e precursors have been found to be primarily locali zed in the fruit skin or fruit peel rather than the pulp or ju ice portion of the fruit. The aromatic thiols are typically released from their cysteine-sconjugate s by cleavage at the C-S linkage of cysteine. In case of grapes, this release has been shown to be assisted by the enzymatic activity of yeasts during fermentation. Aroma Distribution in Juice Matrix Citrus juices are a two phase system comprising an aqueous ‘Serum’ phase containing primarily soluble solids (sugars, or ganic acids, soluble pectins a nd salts) and a water insoluble phase of particulate matter. The insoluble matter ranges from very fine particles (< 2um) called ‘Cloud’ that remain largely suspended in the aq ueous phase to coarser particles (2 to several hundred um) called ‘Pulp’ that tend to settle down upon centrifugation. Pulp comprises mainly of membrane fragments, rag cell wall cellular orga nelles, polysaccharide, protein and lipid molecules. These larger particle s contribute mainly to visual opaqueness, mouthfeel, color and overall flavor to juice and are incorporated in commercial juices at levels that match consumer preference. Cloud forms only a minor fraction of w hole juice (< 1 % wt/wt) Volatile distribution within the juice matrix and its impact on aroma release have been shown to some extent in orange juices ( 3, 71-74 ). Aroma distribution within the ju ice matrix and its impact on flavor release has been studied to some extent in orange juices ( 3, 71-74 ).


32 Early research by Radford et al. ( 72 ) indicated unique matrix distribution patterns for citrus volatiles of differing chemical natu re. The authors examined headspace volatile concentrations of the pulp and serum fractions of citrus juices separate d by ultra centrifugation. Monoterpenes (a-pinene, myrcene, sabinene, li monene), sesquiterpenes (valencene) and long chain oxygenated compounds were reported as pr imarily associated with the pulp fraction accounting for ~ 80% of the tota l volatiles. On the other ha nd, smaller oxygenated compounds (esters, alcohols, aldehydes) a nd oxygenated terpenes (linalool) which typically contribute to citrus flavor, were reported as largely present in serum (cont aining ~20% of total volatiles). Similar volatile distributions were reported more recently for fresh squeezed ‘naveline’ orange juice ( 74 ). By employing model juice studies with washed pulp, Radford et al. ( 72 ) further demonstrated a suppressing effect of pulp on he adspace concentrations of certain aliphatic aldehydes, ketones and esters. Each of these co mpounds was observed to partition at a different extent in the pulp fraction, a tre nd seen dependent on its chemical nature and chain length. This could impact not only the flavor intensity from each compound but also overall aroma balance. The exact nature of interaction between aroma compounds and solid particles was not explained. Whether the aroma compounds found in pulp fraction were merely adsorbed oil droplets on rag particles or due to physical entrap ment inside the particle cell wa ll or fibrillar network remains to be established. More re cently, Brat et al. ( 74 ) reported orange juice pulp to comprise primarily polysaccharides, free sugars, proteins and lipids. These non-volatile components, particularly lipids, may play a role in re tention of mostly non-polar vola tiles such as monoterpenes and sesquiterpenes as described earlier.


33 The effect of pulp content on orange ju ice aroma quality was recently studied ( 71 ) by employing a trained sensory descriptive panel. Th e panel compared orange juice supernatant (no pulp) to juice enriched with increasing concentrations of pulp (6, 12 and 24%). This study reported significant increase in both desirable (natural flavor ) and undesirable flavors (cooked, fermented) resulting from addition of pulp to the supernatant. It is impor tant to note that these samples were subjected to heat treatment af ter the pulp addition step. Although pulp restored some of the desirable flavors back to juice, thermal treatment coul d modify pulp volatiles resulting in the formation/ releas e of ‘cooked’ flavor in juice. Matrix volatile distribution studies may bear relevance to developing sensitive aroma isolation procedures. In addition, it can provide useful insi ghts for improving the flavor of processed juices which are genera lly rated inferior to unpasteurized fresh squeezed juices. As far as we know, there is no comprehensive informati on for volatile distribution in grapefruit juices. The lack of a sensitive analytical procedure for the determination of aroma active sulfur compounds may be a limiting factor for such studies. Citrus Blossom Volatiles Citrus blossoms are known for their pleasan t and highly desirable aroma. Whereas extensive literature exists on the edible fruit volatiles of citrus ( 10, 27-29, 75-83 ), only few studies have analyzed the volatile composition of intact citrus blossoms. One of the more widely studied citrus blossoms is C. aurantium L. var amara, commonly named bitter or sour orange ( 13, 14, 84 ). This blossom variety is commercially cu ltivated for production of Neroli oil, which is a highly prized (and expens ive) perfume ingredient. However, information on the volatile composition of intact blossoms of th e major citrus species is limited.


34 Analytical Considerations One of the biggest challenges to perfumers is to recreate the natural scents of flowers and to use them in creating fragran ces. This requires knowledge of the volatile constituents that contribute to the aromatic scent of flowers. Hi storically, isolation of blossom volatiles was carried out using solvent extraction or steam distillation methods. These procedures yielded essential oils in substantial amounts required for perfumery purposes or for the analytical techniques available at that ti me. For example, in a 1966 study ( 85 ), the authors steam distilled essential oils from large quanti ties (>1 kg) of citrus blossoms, which were then separated using packed column gasliquid chromatography a nd identified using mass spectrometry, infra red spectroscopy and thin layer chromatography. It is well recognized th at the heat applied during a distil lation process produces oils whose smell bear only limited resemblance to the natural scent of blossoms ( 12, 13 ). As analytical detection limits have lowered, it is no longer necessa ry to prepare the blosso m oil. More recently, headspace analysis of blossom volatiles ha s gained popularity amongst diverse group of researchers studying differe nt flower species ( 13, 86-93 ). The main advantage of headspace method over distillation is the noninvasive sa mpling of volatiles emitted naturally from the blossoms. Therefore the aroma extracted closely resembles blossomÂ’s native volatile profile that is perceived by human nose. Most of these rese archers employed SPME, which offers substantial concentration over a wide range of volatile com pounds without requiring h eating of the sample. Chemotaxonomic Classification Conventional taxonomic classifications of the citrus subgenus are based on morphological and geographical characteristics. Numerous classification system s have been proposed, but those of Swingle ( 94 ) and Tanaka (95) are most widely accep ted. However, these two systems also proposed greatly differing number of citrus species: Swingle ( 94 ) proposed 16 species, whereas


35 Tanaka ( 95 ) described 162. Later taxonomi c studies concluded that th ere were only three ‘true’ citrus species and all other cultivars were hybrids of these true species ( 96 ). This theory has been further supported by recent phylogenic classification studies using DNA biochemical markers ( 97, 98 ). Volatile constituents from plants are also employed in devising taxonomic classifications. In the case of certain citrus cultivars such as sour ora nges and mandarins, chemotaxonomic classifications using volatile secondar y metabolites in fruit peel oil ( 14, 15 ), and Petitgrain and Neroli oils ( 14 ) have been explored. Other workers ( 13 ) have examined blossom volatiles of 13 citrus species cultivated in southern Japan, usi ng purge and trap methods. They employed cluster analysis (dendograms) and related resulting classifi cations to morphological characteristics, such as fruit size. InsectPlant Interactions Plant volatiles are well known for their biologica l role as chemical cues in communication with insects ( 99, 100 ). As reported in these studies, foragi ng insects were found to discriminate between plant species. This was based on their odor recognition of both complex mixtures of volatiles as well as specific i ndividual components. Blossom vol atile emission from two species of thistles, were shown to be specific to the pollination needs of these species ( 101 ). The authors found higher emission of blossom volatiles to coinci de with the thistle pollination stage. Overall fragrance decreased appreciably soon after pollination, whic h reduced the incidence of subsequent pollinator visits. However, only a few specific plant volatiles may be involved in any significant biological func tion. Thiery et al. ( 102 ) reported that only 24 of over 100 sunflower volatiles were able to stimulate olfactory respons es in worker honey bees as studied using GC coupled with EAG (electroantennography). Mo st of these stimulants were oxygenated compounds including short ch ain alcohols and aldehydes.


36 CHAPTER 3 SEPERATION AND IDENTIFICATION OF GR APEFRUIT SULFUR VOLATILES USING SPME AND PULSED FLAME PHOTOMETRIXC DETECTION Introduction Volatile sulfur compounds (VSCÂ’s), are key components in the aroma of grapefruit juice (GFJ). Unfortunately, they exist at ng/Kg levels. Thus, identifying and quantifying these volatiles are extremely difficult and partially e xplains the dearth of information on them. Hydrogen sulfide (H2S) was one of the first VSCÂ’s quantified after precipitation with lead acetate ( 31 ). H2S and dimethyl sulfide (DMS), were identif ied and quantified in GFJ using a sulfur specific flame photometric detector after concentration on gold fo il beads and derivatization with ethyl iodide ( 32 ). In the first use of GC-MS, Demole et al.( 27 ) identified 1-p-menthene-8-thiol (PMT), and its bicyclic epimer 2, 8-epithio-cis -p-menthane (EPM), in a concentrated heptane extract from 100L of canned GFJ. The authors reported PMT as the most potent aroma volatile in any food at that time, with a taste threshold (0.1 ng/L in water). More recently, 4-mercapto-4methylpentan-2-one (4MMP) and methional (Met ), were identified a nd quantified using two dimensional HRGC coupled with CI -MS in the selective ion mode ( 29 ). Volatiles were extracted from five liters of GFJ after six hours of continuous distillation w ith diethyl ether. Early eluting sulfur volatiles such as H2S, DMS, SO2, and carbonyl sulfide (COS) could not be determined as these compounds are obscured by the solvent peak. 4-Mercapto-4-methyl-2pentanol (4MMPol), 3-mercapto hexanol (3MHol), and 3-mercapto hexyl acetate (3MHA), were subsequently observed in GFJ from concentr ate in studies employing GCOlfactometry ( 10, 33, 103 ). These sulfur volatiles were only detect ed using human assessors as they were below instrumental detection limits. Additional stud ies employed sulfur chemiluminescence detection, SCD, to detect 4MMP, Met, PM T and 3MHA in pasteurized GFJ ( 10, 33, 103 )).


37 Currently, no comprehensive procedure has been reported which determines all the diverse sulfur volatiles which exist in GFJ. The first goal of this study was to develop a solventless, rapid method for concentrating sulfur headspa ce volatiles which could be used for routine analysis. Solventless techniques were required as solvent peaks masked early eluting sulfur volatiles. Another goal was to optimize extracti on conditions to increase analytical sensitivity without creating artifacts. Th e final objective was to devel op a set of chromatographic conditions which would unambiguously separate th ese sulfur volatiles so that they could be identified and quantified usi ng PFPD and MS where possible. Materials and Methods Grapefruit Juice Samples One commercial brand of canned GFJ (from c oncentrate) was used throughout the study. 5.5 oz cans of juice were purchased from a local supermarket. The same lot of canned juice was used for comparisons between treatments to avoid variation between samples. Fresh unpasteurized juice was extracted from four to six white Marsh grapefruit ( Citrus paradisi Macfed) using a reamer juicer from Sunkist Gr owers Inc (CA, USA). The fruits were picked fresh from groves at the University of FloridaÂ’s Citrus Research and Education Center in Lake Alfred, Florida and juiced the sa me day they were harvested. Standard Chemicals Gaseous standards of hydrogen sulfide (H2S) and sulfur dioxide (SO2) were purchased from Matheson Tri-gas (Houston, TX). Standard compounds such as dimethyl sulfide (DMS), carbon disulfide (CS2), 3-(methylthio) propionald ehyde or Methional (Met), 2-methyl thiophene (2MTP) and 3-methyl thiophene (3MTP) were procured from Si gma-Aldrich (Milwaukee, WI). Ethanethiol (EtSH), dimethyl di sulfide (DMDS), dimethyl trisul fide (DMTS) and bis (2-methyl3-furyl) disulfide (BMFDS) were obtained from Ac ros organics (Geel, Belgium). Methanethiol


38 (MeSH) and dimethyl sulfoxide (DMSO) were found in trace quantitie s in EtSH and DMS respectively. 3-mercaptohexyl acetate (3MHA) and 3-mercapto-1-hexanol (3MHol) were bought from Interchim (Montlugon, France). 1-p-menthene -8-thiol (PMT) was a gift from Ferminich (Lakeland, FL). 2, 8epithio-cis-p-menthane (EPM) and 4-Mercato-4-methyl-2-pentanone (4MMP) were purchased from Oxford Chemi cals (Hartlepool, UK). 4Mercato-4-methyl-2pentanol (4MMP-ol) was purchased from Endea vor Specialty Chemicals Ltd. (Northants, UK). The identities of these compone nts were confirmed by their rete ntion indices, odor qualities and mass spectra. Ethyl methyl sulfide (EMS), is opropyl disulfide (IPDS ) and 8-mercapto-pmenthan-3-one (8-MPM) from SigmaAldrich (Milwa ukee, WI) were used as internal standards. SPME Optimization Manual SPME assembly device comprisi ng a SPME holder 57330-U and fiber was purchased from SUPELCO (Bellefonte, PA, USA). The following parameters were examined to optimize SPME analysis of VSCÂ’s in GFJ. Fiber type Three widely used stationary phases of SPM E fibers were compared (n=3) (a) 75 m CAR/ PDMS, 1cm (b) 50/30m DVB/ CAR/ PDMS (triphase fiber), 1cm and (c) 65m PDMS/ DVB, 1cm. Each fiber was initially preconditi oned in a GC injection port as per company specification. In addition, fibers were conditioned at 200C for 5 minutes prior to each sample extraction. Canned juice samples (10 mL) were pl aced in 42 mL glass vials capped with a Teflon-faced silicone septa. Samples were e quilibrated for 20 minutes at 40C followed by SPME extraction for 30 minutes at the same temper ature. Extraction was assisted with magnetic stirring at 400 rpm.


39 Headspace atmosphere, extraction temperature and time The following headspace conditions during SPME analysis were compared (n=3): Inert nitrogen headspace (~0% oxygen); Air headspace (~21% oxygen); and ~100% oxygen headspace. Sample headspace was purged (10-15 s econds; flow rate of 300mL/min) with high purity compressed gas (nitrogen or oxygen) promptly after transferring of sample into vials. For the air headspace, samples were exposed to atmospheric air for 15 seconds. The SPME conditions were maintained as described in th e previous paragraph. A range of extraction temperatures (22, 30, 40, 50, 60 and 70C at consta nt extraction time of 30 minutes) and a range of extraction duration (5, 15, 30 to 60 minutes at c onstant temperature of 40C) were also tested (n=3). GC-PFPD Analysis HP-5890 Series II GC from Agilent (Santa Clara, CA) equipped with a 5380 PFPD detector from OI Analytical (C ollege station, TX) was used in the sulfur mode. VSCÂ’s from the SPME fiber were desorbed at 200C for 5 minutes in the GC injection port (splitless mode). The injection port was equipped with a Siltek deac tivated (silanized) SPME liner (1 mm x 6.3 x 78.5) from Restek (Bellefonte, PA). Separation of compounds was achieved on DB-Wax (30 m x 0.32 mm i.d. x 0.25m) purchased from Agilent J&W (Santa Clara, CA) and/ or Zebron ZB-5 (30 m x 0.32 mm i.d. x 0.5m) purchased from Phenomen ex (Torrance, CA). GC oven temperature was initially set at 40C and then ramped to 240C @7C/min (in case of DB-Wax) or to 265C @7C/min (in case of ZB-5). The final hold in each case was 5 minutes. GC was operated in a constant flow mode (2 mL/min) with helium as the carrier gas. PFPD detector was set at 250C and employed WinPulse32 Versi on 2.0 software. The PMT voltage was set at 525 V, and sulfur gate was opened between 524ms. PFPD output was recorded in the square root mode.


40 Chromatograms were recorded and integrat ed using ChromPerfect 5.5.5 (Justice Labs, Melbourne, Florida) data acquisition software. Identification of VSCÂ’s Linear retention index (LRI) values were determ ined for each column type from a series of alkane standards (C5-C25) using the carbon resp onse from the PFPD detector. Grapefruit VSCÂ’s were separated on: a nonpolar ZB-5, a polar DB-Wax and a hi ghly retentive GS-GasPro PLOT column (30 m x 0.32 mm i.d.) from Ag ilent J&W (Santa Clara, CA). In order to get accurate LRI values for early eluting sulfur compounds, PL OT column was calibrated using alkanes C4 through C14. The oven temperature for this colu mn was programmed from 175C and ramped to 250C @3C with final hold of 10min. Peak iden tifications were based on matching LRI values with those from pure sulfur sta ndards on all 3 column types. Quantitation Multiple internal standards were employed for quantitation of VSCÂ’s using HS-SPME similar to that used by Fang and Qian ( 43 ). A working standard solution was prepared by dissolving 1 g/mL ethyl dimet hyl sulfide, 0.5 g/mL isopropyl disulfide and 1 g/ mL 8mercapto-p-menthan-3-one in methanol. This solution was stored at -15C until used for analysis. Ten-L of solution was added to 10 mL of juice and thoroughly mixed. Each juice sample was equilibrated (40C for 30 min) and extracted usin g a 2cm DVD /CAR/ PDMS fiber (40C for 45 min). Analyses were carried out in quadruplicate. Statistical Analysis Data for SPME optimization study was aver aged over three replicates. A one-way ANOVA analysis was employed to determine significant differences between sulfur volatile levels under different headspace atmospheres. Me an separations were achieved using TukeyÂ’s HSD test at = 0.05 and box & whisker plots were used for representing these differences


41 between means. The above analysis was perfor med using Statistica 5 fro m StatSoft (Tulsa, OK). Quantitation data for sulfur volatiles in fresh and canned GFJ was averaged over four replicates. Results and Discussion SPME Fiber Selection Commercially available fibers range from the non polar, PDMS, to the highly polar, Carbowax/DVB. Three fiber types were compar ed for their sulfur volatile selectivity and extraction efficiencies in this study. DVB/CAR/P DMS or triphase fiber extracted the widest range of GFJ sulfur aroma compounds. A total of twenty one sulfur pe aks were detected in canned GFJ using the three SPME fiber types. Ei ghteen sulfur peaks were detected using triphase (DVB/CAR/PDMS) fiber. Fifteen GFJ sulfur peaks were observed using the CAR/PDMS fiber. In comparison, PDMS/ DVB coat ed fiber was able to extract only 7 sulfur compounds from the same GFJ. Al l three fibers collected the major sulfur volatiles to different degrees, the major differences between fibers were in the number of minor peaks observed. It is generally accepted that no single fi ber type can extract all analytes. Figure 3-1 summarizes peak area means and st andard deviations (n=3) for the major GFJ sulfur volatiles extracted us ing the two best SPME fibers. Since the PDMS/ DVB fiber was relatively ineffective in concentrating a broad ra nge of sulfur volatiles, this fiber type was excluded from further evaluation. As shown in Figure 3-1A, CAR/PDMS demonstrated a high affinity for the early eluting, low molecular weight (<100 amu), high vapor pressure sulfur volatiles. H2S, MeSH, DMS and DMDS responses were about 3, 2, 6 and 11 fold higher respectively using CAR/PDMS compared to DVB / CAR/ PDMS. Interestingly, CAR/PDMS did not collect SO2. The porous carboxen particles are pres ent in relatively higher amounts in CAR/PDMS fiber and thus have a greater capacity to retain smaller sulfur molecules. It is for


42 this reason that the CAR/PDMS fiber has been employed for the analysis of low MW sulfur volatiles in wine by a numb er of research groups ( 39, 43, 48 ). As shown in Figure 3-1A, the triphase DVB/CAR/PDMS concentrated low molecular weight sulfur volatiles adequately but not as well as CAR/PDMS. However, DVB/CAR/PDMS was superior in concentrating more of the la ter eluting higher mol ecular weight (>100 amu) volatiles (Figure 3-1B). Of particular importance to this study was its ability to concentrate 1-pmenthene-8-thiol which is the major grapefruit character impact compound. The inability of the CAR/PDMS fiber to concentrate one of the most sensorally important GFJ sulfur volatiles eliminated it from further consideration. Nitrogen Versus Air Versus Oxygen Headspace Many sulfur volatiles are highly reactive and prone to artifact formation, especially under oxidative, elevated temperature conditions. In a typical SPME an alysis, most solute compounds form a vapor phase composition that is proportional to its concentration in the liquid (juice). Vapor phase molecules are also concentrated on the fiber thus increasing the possibility of surface phase reactions in the pr esence of oxygen. Some authors have employed the use of inert gas for purging sample headspace prior to analysis to reduce oxidative reactions ( 39, 43 ). In a recent SPME wine headspace study ( 48 ), the use of nitrogen headsp ace achieved better peak area reproducibility versus air head space. More importantly, nitr ogen headspace decreased the amount of oxidative dimerization of DMS a nd diethyl sulfide observed with air. The results summarized as box and whisker pl ots in Figure 3-2 demonstrate how sulfur volatiles are affected by different levels of headspace oxygen. High levels of headspace oxygen (~100%) profoundly affected mean peak areas for al l but one (3MHA) of the GFJ sulfur volatiles (Figure 3-2). Levels of H2S, MeSH, DMS and PMT were significantly reduced by headspace oxygen. Not surprisingly, compounds such as SO2 were detected almost exclusively in the


43 presence of oxygen (data not shown). It seems likely that H2S was oxidized to SO2 in the presence of oxygen as it was diminished as SO2 increased. Similarly the d ecrease in the levels of MeSH may be associated with concomitant increases in DM DS and DMTS levels. This oxidative dimerization of MeSH to DMDS a nd subsequent conversion to DMTS has been described by Belitz et al. ( 59 ) and reported as thermally generated oxidation products in pasteurized milk ( 44 ). EPM was first reported in GFJ as a s pontaneous cyclization product of PMT ( 27 ). This reaction was found to occur at room temperatures in the presence of light. Thus the absence of PMT in ~100% oxygen headspace as seen in Figure 3-2 can be linked to the observed increases in amounts of EPM under similar conditions. Comparison of nitrogen and air headspace pr ovided additional understanding of the possible alterations that could result during SPME analysis under the conditions used in this study. Oxygen in air (~21%) was also found to alte r the amounts and kind s of sulfur volatiles extracted from the GFJ headspace. The primary effects of atmospheric oxygen ranged from poor reproducibility of peak areas (H2S) to significant alterations (as seen for DMTS and PMT). Thus, the VSC’s in GFJ are highly sensitive to oxidation and may undergo alterations during extraction from the sample headspace via SPME resu lting in artifact formation. It is for this reason that analyses in this study were car ried out under nitrogen headspace conditions. Time –Temperature SPME Optimization Exposure time and sample headspace temperat ure conditions are major factors affecting headspace SPME analysis. As shown in Figure 3-3A, longer exposure times increase the levels of volatiles observed. Rapid partitioning and fiber uptake of high vapor pressure compounds such as H2S, MeSH, DMS, SO2 and DMDS was achieved at 40 C. In the case of DMS and MeSH (Figure 3-3A), only marginal increases in peak intensities was observed beyond 5 minutes


44 of exposure time. Thus sensitive extraction of low MW compounds can be achieved even with short fiber exposure times (5-15 minutes). On th e other hand, slower fiber sorption was observed for the later eluting compounds. For polar co mpounds such as Met, EPM, PMT 3MHA, a minimum 15-30 minutes of SPME fiber exposure time was required to give a detectable PFPD response which increased with further exposure. However, very long exposure times may also allow for secondary reactions and displacement reactions to occur. Therefore, a 45 minute SPME exposure time was used for the quantitatio n study as this appeared to provide maximum sensitivity of the peaks of interest and mini mum artifact formation. To examine temperature effects, SPME fibers were only exposed for 30 minutes to maximize the number of analysis which could be carried out in one day. Increasing sample and headspace temperatur e increases solute vapor pressures and headspace concentrations as well as altering the distribution coe fficient between the fiber and headspace molecules. As seen at 22C in Figur e 3-3B, only the most volatile sulfur peaks are observed. As headspace temperature is increased, additional late eluting sulfur volatiles are observed. At 60C, some of the early eluting p eaks are diminished and many additional small peaks are observed suggesting that these small p eaks might be thermally generated artifacts. As seen in Figure 3-3B, DMS decreases rapidly and MeSH decreases slightly as temperature increases above 40C. Therefore, a headspace temp erature of 40C was chosen as it represented the best compromise of maximum headspace concen trations with minimal artifact formation for most sulfur volatiles. Chromatographic Considerations Eighteen GFJ sulfur volatiles were observ ed using the headspace SPME GC-PFPD method developed in this study. Figur e 3-4 represents a typical PFPD chromatogram for canned GFJ obtained using the optimized proced ure developed in this study. ZB -5 column type was selected


45 for further analysis as it offe red better reproducibility (with respect to LRI values, minor peaks and their areas) compared with a Wax column. As shown in this figure, a wide range of VSCÂ’s can be detected and quantified ranging from the highly volatile H2S (bp = -60C) to compounds with much higher boiling point such as bis (2-methyl-3-furyl) disulfide, BMFDS (bp = 270C). Identified peaks are labeled with their names and unidentified peaks are labeled as unknowns. Identification of Sulfur Peaks No prior study has examined the entire rang e of sulfur volatiles which exist in GFJ. Kirchner et al. ( 19 ) reported finding H2S in the headspace of fresh GFJ. Shaw and Wilson ( 11 ) reported finding H2S, DMS and trace levels of SO2. Demole et al. ( 27 ) identified PMT and EPM in their studies. Buettner and Schieberle ( 29 ) identified three GFJ sulfur volatiles, 4MMP, PMT and Met as having aroma activity. Lin et al. ( 10 ) tentatively identified ei ght sulfur components in solvent extracted GFJ using GC-O studies. Five of these, 3MHA, 3MHol, 4MMP, PMT and Met were confirmed using sulfur specific sulfur chemiluminescence detection ( 33 ). Thirteen sulfur volatiles listed in Table 31 have been identified by matching GFJ sulfur peak LRIÂ’s with those from standards on two or th ree dissimilar columns. Five additional VSCÂ’s were tentatively identified as: dimethyl sulfoxide, DMSO, 4MMP, 4MMpol, 3MHol and BMFDS based on LRI matching with standards on either a Wax or ZB-5 column only. Eleven sulfur compounds have been prev iously reported from scattere d GFJ reports and are noted in Table 3-1 and are denoted by the lett ers a-f. No prior study reported more than six volatile sulfur compounds in GFJ. Identifications in this study were based on (a ) sulfur specific PFPD re sponse indicating that the peaks detected contained sulf ur and (b) chromatographic behavi or of these sulfur peaks on 23 dissimilar column types. LRIÂ’s of standard su lfur compounds, on each column type, were then


46 employed for identification and matching with unknown sulfur peaks. Compounds such as H2S (12,600 Torr), SO2 (2530 Torr), MeSH (1900 Torr), CS2 (352 Torr), DMS (647 Torr), have very high vapor pressures and are extremely volatil e. Conventional column types (polar and non polar) do not retain these compounds to a sufficien t extent, often making resolution difficult and identifications inaccurate. A highly retentive porous layer open tubular, PLOT, column was therefore employed as an additional tool (a third set of LRI values ) to validate the identifications for these early eluting peaks. In addition, select ive ion monitoring of GC-MS analysis of these samples further confirmed identifications for the two major sulfur volatilesH2S (m/z 34) and DMS (m/z 62) in fresh and canned juice respectively. The presence of DMSO, 4MMP, 4MMPol and 3MHol and BMFDS in GFJ could not be cross checked with LRIÂ’s from a dissimilar column and their identifications were considered tentative. However, it should be me ntioned that four of the five ha d been also reported in prior studies. For example, Dreher et al. ( 52 ) identified BMFDS as a ther mal degradation product of thiamin in a model orange juice system. 4MMP has been described as a significant cont ributor to the characteristic GFJ aroma from AEDA and flavor reconstitution studies ( 28, 29 ). The limit of detection for the method developed in this study for 4MMP in GFJ was 0.1 g/ L. It appears that this compound, if present, may be at concentrations < 0. 1 g/L. This differs from earlier solvent extraction stable isotope dilution studies ( 29 ) where 0.8 g/Kg of 4MMP was reported. Multiple additional attempts to confirm the presence of 4MMP in this study included: (a ) solvent extraction and concentration of large volumes (1L) of juice, (b) use of 2c m CAR/ PDMS fiber (with higher sensitivity for low MW sulfur compounds), (c) y east fermentation or excessive thermal treatment of fresh GFJ in order to release 4MMP from its potential cysteine co njugate precursor. The


47 cysteine conjugate has been reported as the main precursor of 4MMP present in grapes ( 69 ) and shown to enzymatically release 4MMP during ferm entation and (d) GC-O analysis to determine 4MMP aroma activity even when compound not de tected by PFPD. No 4MMP was observed in any of these additional studies. Quantitation of Sulfur Volatiles A total of thirteen VSCÂ’s were quantified in various GFJÂ’s using the headspace SPME-GCPFPD procedure (Table 3-2). As seen in Figur e 3-4, three structurally dissimilar internal standards, ethyl dimethyl sulfide, EMS, isopropyl disulfide, IPDS and 8mercapto-p-menthan-3one, 8-MPM were employed for quantitation of VS CÂ’s in GFJ. They provided the necessary quantitative accuracy over the wide chemical range of sulfur anal ytes observed in GFJ. EMS peak areas were used to calculate concentrations of H2S, SO2, MeSH, DMS and CS2 by assuming they had similar headspace and PFPD responses With a similar assumption, the polysulfide, IPDS was used to calculate concentrations of DMDS, DMTS, Met, 2MTP and 3MTP. Finally, 8-MPM was used to quantify PMT, EPM, and 3MHA. Linear responses over the range of observed concentrations were seen for all three internal standards with a correlation coefficient R2 0.95. The significance of using multiple internal sta ndards was demonstrated in a recent analysis of sulfur volatiles in wine ( 43 ). The authors reported differe nces in SPME responses for the different sulfur compounds. By using multiple intern al standards, the authors were able to largely compensate for the differential responses and avoi ded errors that would have occurred if only one internal standard was employed. Past studies have been entirely qualitative or quantified less than five sulfur volatiles. In a 1953 study, the authors reported levels of 0.9mg/L of H2S in the headspace of fresh Marsh GFJ ( 19 ) after concentrating it by pr ecipitation in lead acetate. In contrast, Shaw et al. ( 32 ) reported


48 only 1.9 g/L of H2S and 0.5 g/L of DMS in fresh Marsh GFJ analyzed using sulfur specific flame photometric detector. As seen in Table 3-2, H2S (3.7 0.8 g/L), and five other sulfur volatiles ranging from 0.027-0.3 g/L concentrat ions were determined in fresh squeezed (unpasteurized/unheated) Marsh GFJ. Demole et al. ( 27 ) indicated that the amount of PMT in canned GFJ was at or below ppb (g/L) levels. The exact concentration of this potent GF character imparting compound could not be determined at that time. By employing stable isotope dilution analysis coupled with GC-CIMS, Buettner and Schieberle ( 29 ) reported levels of 0.01 g/L of PMT in fresh GFJ. In this study, PMT was not observed in fresh squeezed GFJ but found in canned GFJ sample (0.42 0.047 g/L). This is in agreement with an ea rlier solvent extraction GC-Olfactometry study ( 10 ) comparing the aroma active compounds in fresh ju ice with thermally processed juice. PMT was only observed in the thermally processed juice an d not the fresh squeezed juice. As expected, Met was only found in the thermally treated canned ju ice as it is generally accepted to be formed from the degradation of methionine. As discu ssed earlier 4MMP was only tentatively identified in this study as a trace component and its concentration in GFJ could not be established at this time. Sulfur Volatile Differences Between Fresh and Canned GFJ As seen from the concentration values in Ta ble 3-2, the profile of VSCÂ’s in fresh marsh (unheated) juice differed markedly from that of commercial canned juice (reconstituted from concentrate) both in kind and intensity. Commer cial canned juices are typically reconstituted from concentrate which means they have been heated twice (once during thermal concentration/evaporation and again when recons tituted with water) and hot filled. The greater heating history profoundly alters the volatile sulfur profile as many sulfur compounds are highly reactive.


49 Fresh Marsh grapefruit juice headspace containe d relatively few sulfur peaks compared to canned GFJ. H2S was the predominant sulfur volatile accoun ting for over 80% of its total volatile sulfur composition. The remaining five sulfur vo latiles quantified in Table 3-2 were comprised primarily of low molecular weight volatiles. The only sulfur volatile above amu 100 was 3MHA. Canned GFJ had higher amounts of total volatile s as well as a greater number of sulfur peaks compared to fresh squeezed (unheated) GFJ. As many as twelve sulfur compounds were observed in canned GFJ. Eleven of these have been quantified in Table 3-1. DMS dominated the headspace of this juice, accounting for about 58% of its total volatile su lfur composition. MeSH was the second most predominan t sulfur volatile in canned ju ice (~17%), followed by PMT (~ 7%). Although a major volatile in fresh GFJ, H2S was only a minor consti tuent in canned juice sulfur composition (5%). It appears that most of the H2S observed in fresh juice may have been lost during the thermal concentration/evaporation step due to its extrem ely low boiling point (bp = -60C) In addition to increased levels of the MeSH DMS and 3MHA, there were at least seven additional VSCÂ’s detected in canned juices that were not observed in fresh Marsh GFJ. These were DMDS, DMTS, 2MTP, 3MTP, Met, and most significantly EPM and PM T. It appears that thermal processing and typical prolonged am bient temperature storage of canned GFJÂ’s substantially alters the volatile sulfur com position including formation of the key character imparting compound, PMT. Conclusions A rapid and highly sensitive su lfur specific method was devel oped for quantitative analysis of GFJ sulfur volatiles us ing headspace SPME GC-PFPD. Cr itical SPME parameters were optimized to achieve increased sensitivity and minimize artifacts. Sulfur compounds of diverse


50 chemical nature were identified and quantif ied in GFJ ranging from highly volatile H2S to more polar compounds such as bis (2-methyl-3-furyl) disulfide. Fresh Marsh GFJ and commercial canned GFJ possessed profoundly different sulfur volat ile profiles. It appears that processing and subsequent storage of canned juice play a major ro le in the formation of several sulfur volatiles in GFJ, including the character im parting PMT. Future work is aimed at determining processing related changes in sulfur composition of wide va riety of fresh and processed GFJ juices and its impact on aroma quality of juice.


51 Table 3-1. Comparison of retention time behavior of grapefruit sulfur vol atiles on three different column materials. Data represents linear retention indices. Compounds ZB-5 ZB-5 DB-Wax DB-Wax PLOT PLOT obs std obs std obs std 01 hydrogen sulfide a, b, <400 <400 560 528 <400 <400 02 sulfur dioxide b <400 <400 835 831 414 411 03 methane thiol 430 428 669 675 414 411 04 dimethyl sulfide b, 524 519 740 736 718 718 05 carbon disulfide 545 540 736 722 443 440 06 dimethyl disulfide 743 744 1068 1064 860 858 07 2-methyl thiophene 770 771 1078 1083 08 3-methyl thiophene 781 780 1093 1098 09 dimethyl sulfoxide 868 1623 1610 10 methional d,e,f 915 914 1444 1450 11 4-mercapto-4-methyl-2pentanone d,e,f 949 1360 1366 12 dimethyl trisulfide 980 978 1344 1355 13 4-mercapto-4-methyl-2pentanol f 987 1544 1522 14 3-mercapto-1-hexanol e,f 1130 1844 1828 15 3-mercaptohexylacetate e,f 1256 1251 1721 1701 16 2,8-epithio-cis-p-menthan c 1268 1264 1507 1492 17 1-p-menthene-8-thiol c,d,e,f 1298 1295 1594 1580 18 Bis (2-methyl-3-furyl) disulfide f 1540 1537 2117 a – reported in ( 31 ) b – reported in ( 32 ), c – reported in ( 27 ), d – reported in ( 28 ), e reported in ( 10, 33 ), f reported in ( 10 ), tentative identifications, confirmed by MS


52 Table 3-2. Quantitative analyses of sulfur vol atiles in fresh and ca nned (reconstituted from concentrate) juices. Data represents Mean S.D values in g/L Compounds FHS Juice FC Juice MW 01 hydrogen sulfide 3.7 0.81 0.28 0.044 34 02 sulfur dioxide 0.30 0.068 64 03 methane thiol 0.23 0.041 0.94 0.11 48 04 dimethyl sulfide 0.24 0.073 3.3 0.074 62 05 carbon disulfide 0.065 0.025 76 06 dimethyl disulfide 0.071 0.015 94 07 2-methyl thiophene 0.053 0.019 98 08 3-methyl thiophene 0.024 0.016 98 09 dimethyl sulfoxide 78 10 methional 0.028 0.005 104 11 4-mercapto-4-methyl-2-pentanone 132 12 dimethyl trisulfide 0.075 0.030 126 13 4-mercapto-4-methyl-2-pentanol 134 14 3-mercapto-1-hexanol 134 15 3-mercaptohexylacetate 0.027 0.005 0.14 0.026 176 16 2,8-epithio-cis-p-menthan 0.25 0.019 170 17 1-p-menthene-8-thiol 0.42 0.047 170 18 bis (2-methyl-3-furyl)disulfide* 226 Tentative identifications


53 A B Figure 3-1. Comparison of aver age peak areas for volatile sulfur compounds in canned grapefruit juice using CAR/PDMS and DVB/CAR/PDMS SPME fibers at 40C. Black bars = CAR/PDMS and grey bars = DVB/CAR/PDMS


54 Figure 3-2. Effects of different levels of headspace oxygen on canned grapefruit juice VSCÂ’s. Nitrogen headspace ( 0 oxygen), air headspace ( 20% oxygen) and oxygen headspace ( 100%). Box and whisker plot values followed by different letters are significantly different (TukeyÂ’s HSD at = 0.05).


A B Figure 3 3 -3. Effect o f DMS ( ) vertical a x f SPME exp ) MeSH ( ) x is and ope n osure time ( A ) PMT ( ) a n symbols u s 55 A ) and tem p a nd 3HMA ( s e right han d p erature (B) ). Filled s y d side vertic a on extracti o y mbols use l e a l axis. o n efficienc y e ft hand sid e y of: e


56 Figure 3-4. PFPD chromatogram of SPME headsp ace sulfur volatiles from canned grapefruit juice. Volatiles separated on a DB-5 column. (IS) = internal standard.


57 CHAPTER 4 PROCESSING PATTERNS, PRECURSORS AROMA ACTIVITY AND MATRIX DISTRIBUTION OF GRAPEFRUIT JUICE SULFUR VOLATILES Introduction Volatile sulfur compounds (VSCÂ’s), are key aroma components in grapefruit juice (GFJ), flavor. Identification and quantitation of th ese trace level (ng/Kg) volatiles are extremely difficult using conventional aroma isolation and detection procedures. Using a recently developed sensitive HSSPME coupled with su lfur specific GCPFPD method (Chapter 3), a comprehensive set of thirteen VSCÂ’s were quantif ied for the first time in fresh and processed GFJÂ’s. Typically, VSCÂ’s are substa ntially higher both in number of volatiles and total peak area in canned reconstituted from con centrate juice compared to fresh hand squeezed juice. Thermal processing effects on major volatil es in orange juice have been extensively studied, including a few sulfur volatiles ( 2-5, 7-9, 11 ). Limited studies have been reported on the influence of thermal processing on potent VSCÂ’s impor tant to grapefruit juice aroma ( 10, 11 ). Decreased levels of hydrogen sulfide (H2S) ( 11 ), and aroma intensities of 3mercapto hexyl acetate (3MHA), 3mercapto hexanol (3MHol) and 4mercapto4-methyl2pentanol (4MMPol) ( 10 ) were attributed to thermal processing. On th e other hand, thermally processed juices were reported to contain higher levels of dimethyl sulfide (DMS) ( 11 ), and higher aroma intensities of 1-p-menthene-8-thiol (PMT) ( 10 ). Several precursors of volatile sulfur com pounds in citrus juices have been proposed. Proposed precursors include: sulfur containing am ino acids such as cysteine, glutathione, methionine and S-methyl methionine ( 8, 10, 11, 33, 58 ); vitamins ( 8, 52 ) and more recently cysteine-Sconjugates ( 53 ). In addition, some VSCÂ’s found in citrus juices could be generated due to thermal breakdown, oxidation or polymerization of other VSCÂ’s ( 8, 32 ). Model food systems containing sulfur amino acids have been employed in the past to determine thermal


58 degradation products at high temperature (180 C ) ( 60-62 ). There is no conclusive information on the specific products of th ermal breakdown of each of thes e amino acids under grapefruit juice conditions (pH < 4.0) and lo wer temperatures. Therefore, th e definite contributions of specific amino acids to grapefruit juice su lfur aroma remains to be established. The main objectives of this study were to : (a) quantify VSCÂ’s in fresh (unheated) and commercial NFC and RFC grapefruit juices to dete rmine if these VSCÂ’s were characteristic of their heating history; (b) examine the VSC cha nges in fresh GFJ during an accelerated storage study; (c) compare aroma active GFJ volatiles fres h squeezed and canned RF C GFJ with specific emphasis on determining which VSCÂ’s had aroma activity; and (d) determine thermal breakdown products of potential VSC precursors such as cy steine, glutathione, meth ionine and s-methyl methionine sulfonium ion in model GFJ systems. A final goal was to determine the sulfur volatile distribution be tween pulp and serum components of GFJ. Volatile dist ribution within the juice matrix and its significance to aroma release has been studied to some extent in orange juices ( 3, 71-74 ). These studies bear relevance to developing sensitive aroma isolation procedur es and also could provide useful insights for commercial processors to improve the flavor of juices containing different pulp levels. Materials and Methods Grapefruit Juice Samples Twenty nine grapefruit juice samples in cluding ten fresh handsqueezed and 19 commercially processed juices were included in a survey study completed during the period of March 7April 30, 2008. Fresh fruit samples compri sed four cultivars picked on multiple days: Marsh white seedless from two harvest dates, D uncan white (four harvest dates), Ruby red (one harvest date) and Flame (three harvest dates). Marsh and Ruby red were picked from the groves at University of FloridaÂ’s Citrus Research and E ducation Center in Lake Alfred, Florida. Duncan


59 and Flame were picked from dooryard trees within 50 miles from the research center. Fresh fruits were juiced the same day they were harvested. Ju ice was handextracted fr om four to six fruits using a model 8-R reamer jui cer (Sunkist Growers Inc CA, USA). The 19 commercial juice samples: ten refrigerated not from concentrate juices (NFC) and nine reconstituted from concentrate juices (RFC), were purchased from a local supermarket. Of the nine RFC juices, eight were refrigerated and one was a shelf stab le canned juice. Both white and pink grapefruit varieties were utilized for commercial juice sa mples as indicated on their labels. All samples were analyzed the same day as juiced or purchased. Chemical Standards Gaseous standards of hydrogen sulfide (H2S) and sulfur dioxide (SO2) were purchased from Matheson Tri-gas (Houston, TX). Standard compounds such as dimethyl sulfide (DMS), carbon disulfide (CS2), 3-(methylthio) propionald ehyde or Methional (Met), 2-methyl thiophene (2MTP) and 3-methyl thiophene (3MTP) were procured from Si gma-Aldrich (Milwaukee, WI). Ethanethiol (EtSH), dimethyl di sulfide (DMDS), dimethyl trisul fide (DMTS) and bis (2-methyl3-furyl) disulfide (BMFDS) were obtained from Ac ros organics (Geel, Belgium). Methanethiol (MeSH) and dimethyl sulfoxide (DMSO) were found in trace quantitie s in EtSH and DMS respectively. 3-mercaptohexyl acetate (3MHA) and 3-mercapto-1-hexanol (3MHol) were bought from Interchim (Montlugon, France). 1-p-menthene -8-thiol (PMT) was a gift from Ferminich (Lakeland, FL). 2, 8epithio-cis-p-menthane (EPM) and 4-Mercato-4-methyl-2-pentanone (4MMP) was purchased from Oxford Chemicals (Hartlepool, UK). 4-Mercato-4-methyl-2pentanol (4MMPol) was purchased from Endea vor Specialty Chemicals Ltd. (Northants, UK). The identities of these compone nts were confirmed by their rete ntion indices, odor qualities and mass spectra. Ethyl methyl sulfide (EMS), is opropyl disulfide (IPDS ) and 8-mercapto-pmenthan-3-one (8MPM) from SigmaAldrich (Milwa ukee, WI) were used as internal standards.


60 Sulfur Characterization of Survey Samples Static headspace SPME quantification using three internal standards was employed according to procedures developed in chapter 3. Ten milliliters of juice with added internal standard (10 L methanol solution contai ning 1g EMS, 0.5g IPDS and 1g 8MPM) was placed into a 42 mL glass vial with a micro stir ring bar and a Teflon-faced silicone septa screw cap. The vial headspace was purged (10-15 seconds ; flow rate of 300 mL/min) with high purity compressed gas (nitrogen) promptly after transf erring of sample into vials. The sample was equilibrated at 40C for 30 min in a water bath, and a 2 cm 50/30m divinylbenzene/Carboxen /polydimethylsiloxane (DVB/Car boxen/PDMS) Stable Flex fibe r (Supelco, Bellefonte, PA) was manually inserted into the vial and exposed fo r 45 min. Extraction was assisted with magnetic stirring at 400 rpm. Analyses we re carried out in quadr uplicate using four sample vials for each sample. The fiber was then transfe rred to the injector of the GC and desorbed for 5 min at 200C. A HP-5890 Series II GC from Agilent ( Santa Clara, CA) eq uipped with a 5380 PFPD detector from OI Analytical (C ollege station, TX) was used in the sulfur mode. Separation of compounds was achieved using a non-polar co lumn, 5% phenyl, 95% dimethylpolysiloxane (Zebron ZB-5; 30 m x 0.32 mm i .d. X 0.5m) purchased from Phenomenex (Torrance, CA). GC oven temperature was initially se t at 40C and then ramped to 265C @7C/min with 5 min final hold time. GC was operated in a constant flow mo de (2 mL/min) with helium as the carrier gas. PFPD detector was set at 250C and employed WinPulse32 Version 2.0 software. The PMT voltage was set at 525 V, and sulfur gate was opened between 524 ms. PFPD sulfur output was recorded in the square root mode. Chromat ograms were recorded and integrated using ChromPerfect version 5.5.5 (Justi ce Labs, Melbourne, Florida) so ftware. Peak identifications were achieved by matching linear re tention index (LRI) values on three column types using the method described in chapter 3.


61 Thermally Accelerated Storage Study Ten-mL samples of freshly handsqueezed Marsh white grapefruit (extracted by method described previously) were placed into 42 mL gla ss vials as described in the previous paragraph. Sample vials containing juice and air headspace were subjected to heating in water bath at 95C for 2, 6 and 15 hours. Sulfur volatiles in unheated fr esh juice and the heated juice samples (n =3 each) were analyzed using SPMEPFPD method de scribed previously without using internal standards. Change in sulfur composition with increasing heating durations was determined based on relative peak areas. Model Juice Precursor Study Model GFJ, pH 3.4 was prepared following the procedure described for model orange juice by Dreher et al. ( 52 ) with adjustments based on the compos ition of unsweetened grapefruit juice detailed in the Fineli database (National Health For Hea lth and Welfare, Helsinki, Finland). Model GFJ comprising % w/w of sucrose (2%), gl ucose (3.2%), fructose (3.8%), citric acid (0.3%), tripotassium citrate (0.5% ) was prepared using double disti lled water. pH was adjusted to 3.4 (using concentrated HCl) based on the aver age pH observed for various fresh grapefruit juices included in this research. Model juice solutions containing individual amino acids: 10mM Lmethionine from Acros organics (Geel, Belgium), 50mM Lcy steine hydrochloride anhydrous from Fluka (Sigma-Aldrich, India), 50 mM Lglutathione fr om SigmaAldrich (Milwaukee, WI) and 0.1 mM DL-methionine methyl sulfonium chloride (S-methyl methionine) from Fluka (Sigma-Aldrich, Japan) were prepar ed. 10mL solutions of each amino acid were placed in the 42 mL glass vials capped with Tefl on-faced silicone septa. Sample vials containing juice and air headspace were subjected to heatin g in water bath at 95C for 3 hours. Volatile sulfur compounds resulting from thermal breakdown of each amino acid system were analyzed


62 (n= 3) and identified using HSSPME method (DVB / Car/ PDMS 1 cm fiber, equilibration15 min; extraction30 min @ RT) coupl ed with GCMS and GC-PFPD. GC-MS conditions Analyses were performed with a Perkin-E lmer Clarus 500 quadrupole mass spectrometer equipped with Turbo Mass software version 5.4.0 (Perkin-Elmer, Shelton, CT) and a Stabilwax column (Restek, 60 m x 0.25 mm x 0.5m). Helium was used as car rier gas in constant flow mode of 2 mL/min. Source and in jection port temperatures were maintained at 200C and the transfer line temperature was 260C. The oven was programmed from 40 to 240C at 7C/min. Electron impact ionization (70 eV ) was operated in the total i on current mode by scanning m/z 25 to 300. Trace levels of various sulfur com pounds were detected using extracted ion monitoring option: carbon disulfid e (m/z 76); thiophene (m/z 84) and 2-methyl thiophene (m/z 98). Mass spectra matches were made by comparison of NIST 2005 version 2.0 standard spectra (NIST, Gaithersburg, MD). Only those compounds with spectral fit values 900 were considered positive identifications. Authentic sulfur standard LRI values were compared for identification. Further confirmation was based on presence of sulfur peaks on GCPFPD using DB-Wax column with matching LRI values. GasOlfactometry Study Marsh seedless freshly handsqueezed juices (extracted by method described in the grapefruit juice sample section) and commercia l canned RFC juices were compared. GC-O was performed using an HP-6890A GC (Hewlett-Pack ard Inc., Palo Alto, CA) equipped with a sniffing port. Samples were run on polar (DB-wax, 30 m x 0.32 mm i.d. X 0.5m; J&W Scientific, Folsom, CA) and a nonpolar co lumn, (ZB-5, 30 m x 0.32 mm i.d. X 0.5m; Phenomenex, Torrance, CA). The ZB-5 column oven temperature was programmed from 40 to 265C and from 40 to 240C for DB-wax at 7C/min, with a 5 min final hold. Helium was used


63 as carrier gas at flow rate of 2 mL/min. Splitless injector (200C) was us ed. The column effluent was split with one-third flow directed to FID and the other two-thirds to the humidified olfactory port for sniffing. Two trained assessors evaluate d each sample in triplicate on both ZB-5 and DB-wax columns. A timeintensity method using a linear potentiometer as previously described ( 104 ) was used to record their sensory impressions (retention times, aroma descriptors and intensities). Olfactory assessor and FID responses were separately recorded and integrated using two channels with ChromPerfect 5.5.5 (Justice Labs, Melbourne, Florida) so ftware. Intensities of each odor-active compound from all six replicates we re averaged and then normalized so that the peak with highest average inte nsity across all GC-O runs was given a score of 10. A peak was considered to be aroma-active if detected 3 times out of total six responses for each juice sample on each column type. Volatile Matrix Distribution Study Serum and pulp separation for a canned RFC juice sample was achieved by ultracentrifugation (average RCF of 103,700 x g for 1 hr) using a Beckman Coulter LE80K system (Fullerton, CA). A SW 28 roto r with 75.3 mm161 mm radius was operated at maximum RPM of 28,000. The clear serum supernatant was gently decanted from the centrifuge tubes and the tightly packed pulp pellets were removed with th e aid of glass rod. The pulp was reconstituted in double distill water (volume equali ng the serum fraction) using ma gnetic assisted stirring for 30 min. Samples (10mL) of each fraction were placed in the 42 mL glass vials capped with Teflon-faced silicone septa. The samples were purged with nitrogen and then extracted using SPME as described previously for the survey study. In case of volatile organic compounds, the analysis was completed using GC-MS procedur e described for the model GFJ precursor study, with total ion current monitoring from m/z 25300. Analysis of volatile sulfur compounds was achieved using GC-PFPD procedure also de tailed earlier for the survey study.


64 Statistical Analysis Quantitation data for sulfur volatiles in survey grapefruit juice samples were averaged over four replicates. A one-way ANOVA analysis us ing Statistica 5 from StatSoft (Tulsa, OK) was employed to determine significant differences be tween fresh, NFC and RFC juice types. Mean separations were achieved us ing TukeyÂ’s HSD test at = 0.05. Data for thermally accelerated storage study, model juice precu rsor study and matrix distribu tion study were peak areas averaged over three replicates for each sample type. Column charts with standard error bars ( = 0.05) were used to represent significant differe nces between means for the thermal accelerated storage study and matr ix distribution study Principal component analysis (PCA) in Unscrambler 9.7 from Camo (Woodbridge, NJ) and Discriminant function analysis in Statistica 5 from StatSoft (Tulsa, OK) were used to evaluate the survey data set. Results and Discussion Sulfur Composition of Fresh and Processed Grapefruit Juices A total of 33 sulfur volatiles were obser ved in fresh and processed GFJÂ’s using the headspace SPME GC-PFPD method developed in Chap ter 3. Of these, 13 VSCÂ’s were identified by matching GFJ sulfur peak LRIÂ’s with t hose from standards on two or three dissimilar columns, as described previously (Chapter 3). Table 4-1 lists the average concentrations and ranges of VSCÂ’s observed in 10 fresh, 11 NFC and 9 RFC juices quantified using multiple internal standards method. As seen from the c oncentration values in Ta ble 4-1, the profile of VSCÂ’s in unheated fresh juices differed markedly fr om that of heat processed juices in kind and intensity. Overall, fresh juice headspace contai ned relatively few sulfur peaks compared to commercially processed GFJÂ’s, FHS (21 VSCÂ’s, 3.1 g/L) << RFC (27 VSCÂ’s, 7.8 g/ L). This corroborates the findings of the earlier study (C hapter 3), wherein higher concentrations of


65 VSCÂ’s were observed in RFC canned GFJ sample as compared to a fresh hand-squeezed Marsh GFJ. Despite a wide range of values observed, se veral VSCÂ’s showed si gnificant differences between the three groups as determined by ANO VA. Of the 33 VSCÂ’s monitored, 29 showed significant differences (p 0.05) among unheated and commercial ly heated juices in general. Twenty of these were present in significantly higher concentra tions in the heated juices. NFC and RFC further differed in their composition with 11 VSCÂ’s present in significantly higher concentrations in RFC juices. Three sulfur peaks (ZB-5 LRIÂ’s: 872, 1547, 1768) de tected in fresh juic es were absent in the commercially processed juices. Only one of these peaks (LRI: 1547) has been tentatively identified as bis(2-methyl-3-furyly)disulfide. A lternatively, up to 14 sulf ur peaks were observed only in heat processed juices. These include d compounds such as DMDS, 2MTP, 3MTP, Met, DMTS, EPM, PMT and seven other unidentified compounds. Once again, the results are in confirmation of earlier findings wh ere at least seven additional peaks were detected in canned juice compared to fresh Marsh juice: DMDS, DM TS, 2MTP, 3MTP, Met, EPM and PMT. In this study, most of these VSCÂ’s (11 out of 14 found in pr ocessed juices) were present in significantly higher quantities in RFC juices compared to NFC ju ices, with at least five (unidentified) sulfur peaks observed only in the RFC juices and absent in NFC juices. Clearly, it appears that thermal processing of GFJÂ’s can substant ially alter the volati le sulfur composition. Interestingly, the grapefruit character impact PM T was observed in measurable qua ntities (0.87g/ L) only in the canned RFC juice samples (Table 4-1) and not the refrigerated RFC samples. Canned RFC which is a shelf-stable product would be typica lly stored without refrigeration for prolonged durations before retail or purchase. This lead s to the hypothesis that th e reaction that produced


66 PMT was largely induced by heating during ther mal processing and prolonged durations of storage at room temperatures. In a commercial process as detailed in ( 105 ), NFC juices are typi cally heated once during the initial pasteurizati on (95-98C, 10-30 sec). RFC juices ar e typically heated twice: initial pasteurization as NFC followed by vacuum assist ed evaporation (40C for 57 minutes) and second pasteurization post recons titution with water (80C95C, 1530 sec). The greater heating history of RFC juices can profoundly alte r the volatile sulfur profile as many sulfur compounds are highly reactiv e. Further, the aqueous essence (c ollected during the concentration/ evaporation step) that is restored to the rec onstituted juice has also experienced a thermal processing and may further contribute to the mark ed difference in composition compared to fresh juices. H2S was the predominant sulfur compound obser ved in the headspace of fresh grapefruit juices (59% of total VSC’s) ranging from 0.49 – 4.8g/L. An unidentif ied peak (ZB-5 LRI: 1775) was the second most abundant compound (13% of total VSC’s) seen in fresh juices. This peak did not possess aroma activity. The remaining 17 sulfur volatiles found in fresh juices were only present in minor concentrations ranging on an average from not detectable levels to 0.27g/L. In case of NFC juices, there were th ree major sulfur compounds DMS (0.9g/L), CS2 (0.65g/L) and MeSH (0.42g/L) contributing to an average 30%, 22% and 14% of total volatile sulfur composition re spectively. All three compounds were observed only as minor components of fresh juices. RFC juices contai ned dramatically higher levels of DMS (ranging from 2.110g/L; 62% total sulfur) compared to both fresh and NFC juices. The higher DMS content contributed almost exclusively to the increased total sulfur content observed in RFC juices (7.8g/L). MeSH and CS2 were present as two other major sulfur compounds of RFC


67 juices contributed to 13% and 5% of total sulfur composition re spectively. As observed in the previous study (Chapter 3), H2S which was a major volatile in fresh GFJÂ’s, but only a minor sulfur constituent in heat treate d NFC and RFC juices (2-3%). This confirms the speculation in chapter 3, that thermal proce ssing could result in loss of this extremely low boiling compound (bp = -60C). Similar observations were reported by Shaw and Wilson ( 11 ), when they examined headspace levels of two sulfur peaks in fresh a nd commercially processed grapefruit juices using packed column GC coupled with a flame photom etric detector. These au thors found significant amounts of one compound, tent atively identified as H2S in freshly squeezed juices but none in the heat processed juices. Thermally Accelerated Storage Study The objective of this study wa s to establish the role of thermal processing (in all commercially heated juices) and prolonged storage (canned RFC juice type) in the formation of various VSCÂ’s that were observed in these juic e types. Figure 4-1 repr esents volatile sulfur composition of freshly squeezed Marsh grapefruit juice as affected by increasing durations of heating (95C; 2, 6 and 15 hrs) in presence of atmospheric oxygen. This study simulates exaggerated conditions in order to explore the full extent of pro cessing as well as storage related abuse of sulfur composition in GFJ in a short time period. Fresh (unheated) juice contained few sulfur compounds. H2S was the major peak accounting for 59 % of total sulfur peak area. A significant increase both in kind and amount of su lfur compounds was observed as the juice was heated for increasing durations wi th a concomitant decrease in H2S. Increase in levels of MeSH, DMS, CS2, DMDS, DMTS, 2MTP, 3MTP, Met and EP M was observed within 26 hours heating. This study confirms the role of thermal tr eatment in formation of most of the VSCÂ’s that were observed in the commercially processed GF JÂ’s survey study (Table 4-1). Peak areas for MeSH, DMS and CS2 decreased with prolonged heating indicating the highly reactive or


68 thermally unstable nature of these compounds. PMT was only dete cted after 15 hours of heating. The sulfur volatile profile of this extensively heated (15hrs) GFJ sample closely resembled the pattern observed in RFC juices, particularly th e canned juice sample. This study confirmed the role of prolonged thermal conditions in the formation of PMT. Classification of Grapefruit Jui ces Based on Sulfur Composition Multivariate analysis of the survey data using principal component analysis (PCA) and discriminant function analysis (DFA) enabled separation of juices into fresh, not from concentrate and from concentrate categories. PCA is an unsupervised exploratory technique which reduces the multidimensional data set to lo wer dimensions or principal components(PCÂ’s), and helps identify inherent patterns in the data in an unbiased way. It highlights the overall similarities and differences amongst sa mples with respect to the overall sulfur composition (based on quantities of 33 VSCÂ’s). It also helps identif y those volatiles which contribute to maximum variability in the data. DFA, which involve s preassigning classes to all samples, identifies specific VSCÂ’s which are most discriminating between the sample classifications and can help predict samp le classifications for new samples. Figure 4-2 represents the eigenvector values for each juice sample mean using the first three PCÂ’s. The high degree of clustering and mi nimal overlap suggests unique sulfur volatile profiles for these clusters. The first two PCs, together account fo r 91% of the total variance in the data. PC1 can be thought of as a H2S (load value 0.75) and DMS (l oad value -0.61) axis. It accounts the complete separation of fresh juice group on the positive end of axis from RFC juice group on the negative end. PC2 axis was mainly defined by DMS (load value of 0.69) and CS2 (load value of -0.47), accounting for further separa tion of NFC juices from RFC juice. The third dimension, PC3, explained an additional 4% of th e total variabil ity attributed mainly to SO2 (loading of 0.9). Overall, RFC and FHS juices were most distinct clusters with no overlap


69 indicating the significant differences in their sulfur composition. NFC juices were positioned between these two widely distinct clusters, with some NFC juices overlapping onto the RFC juices indicating more closel y related sulfur composition. Figure 4-3 represents the canonical roots obt ained from performing forward stepwise multiple groups DFA. Using a minimum of 3 variables (steps) helped achieve significant separation (p< 0.0005) between the fresh juices and the heat treated juices (Figure 4-3). These compounds were Met, H2S and SO2 in decreasing order of sign ificance as determined by their WilkÂ’s Lambda listed in Table 4-2. Further se paration between NFC and RFC juice categories (p<0.005) could be achieved (as seen in Figur e 4-4) by including some additional variables (steps): MeSH, DMS, DMDS and DMTS (listed in Table 4-3). All four compounds were seen in significantly higher concentrati ons in RFC juices as obser ved from the Anova results summarized in Table 4-1. In an analogous discriminant functio n analysis and classification study ( 106 ) based on major (non sulfur) volatiles in fresh and comme rcial grapefruit juices, 16 compounds (steps) were required to achieve significant separati on between fresh, NFC and RFC juices. These included aliphatic alcohols (6), ketone (1), alde hyde (1), esters (2), terpenes and oxygenated terpenes (6). Since only 7 step s were required to achieve sim ilar separations in this study, the sulfur procedure is probably the more robust approach. Nonvolatile Precursors of Volatile Sulfur Compounds Sulfur containing amino acids su ch as cysteine, glut athione and trace leve ls of methionine have been previously repo rted in citrus juices ( 54-56 ). These nonvolatile compounds have been speculated as the likely sources of several volatile sulfur com pounds in citrus juices such as: H2S, MeSH, DMS, DMDS, DMTS, Met and PMT ( 8, 10, 11, 33, 58 ). S-methyl methionine, a known precursor of DMS, has also b een reported in mandarin juices ( 6, 57, 58 ).


70 The initial goal of this study was to identif y the sulfur thermal breakdown products from equimolar concentrations of i ndividual sulfur containing ami no acids (cysteine, glutathione, methionine) and s-methyl methionine in model GFJ systems (pH 3.4; air headspace). However, large variations in the amounts of breakdown pr oducts (measured as GC-MS TIC peak areas) were observed (data not shown). This made detec tion and identification fo r the trace level sulfur breakdown products for some sulfur amino acids di fficult. As a result, the molar concentrations of each amino acid were adjusted (as repres ented in Figure 4-5), in order to achieve approximately similar responses in term s of total GCMS TIC peak area. Figure 4-5 summarizes the major sulfur volatil es resulting from breakdown of these four amino acids. The ease or extent of thermal breakdow n could be inferred to some extent from the molar concentrations required to obtain measurable amounts of products. The four amino acids studied, varied substantially in relative amounts as well as kinds of thermal breakdown products in model GFJ. Overall, S-methyl methionine >> methionine> glutathione > cysteine in the total amounts of volatile sulfur compounds produced. S-methyl methionine was 100x more unstable than methionine. It is also worth noting th at these two related compounds produce completely different thermal degradation products. As show n in Figure 4-5, S-methyl methionine produces almost exclusively DMS whereas methionine produces primarily DMDS and methional. The formation of DMS from S-methyl methioni ne has been shown to follow a first order kinetic reaction ( 6, 107). Sawamura et al. ( 6 ) proposed that at low pH of 3.2, Smethyl methionine would typically undergo a hydrolysis reaction based on the observed formation of DMS and homoserine. As seen in Figure 4-5, DMS appeared fairly heat stable as less than 1% was oxidized to DMSO during the 3 hr heating @ 95C in an air headspace.


71 The major sulfur volatile products from me thionine thermal breakdown were methional (49%), DMDS (39%), DMTS (8%) and MeSH (4%). Various methionine thermal breakdown pathways have been proposed. It is well established that me thional (Met) is the Strecker degradation product of methionine ( 59 ). This aldehyde could further undergo -elimination to form MeSH which in turn can readil y oxidize to form DMDS and DMTS ( 59 ). Alternately, some authors ( 108 ) have proposed that methionine su lfoxide, a common oxi dation product of methionine, is a more likely precursor in gene ration of these polysulfides. These authors reported higher yields of DMDS and DMTS from methionine sulf oxide compared to methionine heated in model systems containing glucose. They further hypothesized that an additional Strecker degradation step was involved in the above reaction with methional sulfoxide as the intermediate. H2S was the predominant sulfur product obtained from heating cysteine and glutathione in model GFJ conditions. Some of the other minor breakdown products of cysteine comprised CS2 (another 15%), thiophene (5%) and trace levels of 2MTP. Majority of the thiophenes and thiazoles reported earlier ( 60 ) from high temperature (180C) r eactions of cystei ne/ glutathione with glucose were not observed in our study. Th e authors suggested that alkaline conditions (pH 7.5) were most conducive for formation of these sulfur compounds. Further, both cysteine and glutathione have been described as susceptible to oxidati on forming their respective dimers ( 61, 62 ), a phenomenon likely to have occurred in ou r model GFJ system heated in atmospheric headspace. In these studies ( 61, 62 ), a thermal reaction between th e reduced forms of cysteine/ glutathione and glucose resulted in formation of sulfur contai ning products such as thiophenes and thiazoles. Oxidized forms of cysteine and glutathione, on the othe r hand, did not produce any


72 sulfur containing compounds due to the possibl e impact of oxidation on the Maillard type reactions. As seen in Figure 4-5, up to five times as much H2S was released from glutathione than from equi-molar concentrations of cysteine. Th is observation was in agr eement with the findings of an earlier kinetics study ( 109 ). In this study, the release of H2S under acidic conditions (pH 3.0) was reported to require lo wer activation energy for glutathi one (18.8 Kcal/ mol) than that required for cysteine (31.3 Kcal/ mol). In conclusion, all four amino acids investigat ed in this study were found to generate some of the major VSC’s encountered in grapefruit juices (Table 41 and Figure 4-1). This finding substantiates their role as potenti al precursors of VSC’s in grapefru it juice. Their actual impact to grapefruit aroma quality must, however, be dete rmined based on the aroma activities of their breakdown products at levels in which they occu r in real juice systems. There are additional sulfur volatiles observed in thermally processe d GFJ for which the precursors have yet to be identified. Comparison of Aroma Compounds in F resh and Canned RFC Grapefruit Juice Fifty one aroma active peaks were detected in fresh and canned reconstituted from concentrate juice using GC-Olfactometry. These ar e listed in Table 4-4 in increasing order of LRI values on a polar column (DB-Wax). LRI’s on the nonpolar (ZB-5 column), odor descriptors, tentative chemical identificati ons and normalized aroma intensities are also summarized in this table. Inte nsities of all odor-ac tive compounds for each sample type were averaged and then normalized to a score of 10 given for the highest intensity peak amongst both juice types (peak #18/ octanal in canned RFC GF J). A comparison between fresh and RFC juice aroma intensities of all odoractiv e peaks is presented in Figure 4-6. The ‘citrusy’ octanal and ‘green’ hexanal were the most intense aroma peaks in FHS juice (normalized intensity of 8)


73 closely followed by ethyl butanoate and linal ool. Canned RFC juice co ntained fewer aroma peaks with the most intense aroma compounds co mprised octanal, Z-2-nonenal, b-ionone, PMT and linalool in decreasing order of intensity. Sulfur-containing aroma volatiles: Nine of the 51 odor-active GFJ volatiles belong to this category as their aromas were described as ‘sulfur/ cabbage/ cooked potato/ roasted coffee/ grapefruit’. Six of them we re positively identified as Me SH, DMS, Met, 3MHA DMTS and PMT. Identifications were base d on (a) sulfur specific PFPD res ponse indicating that the peaks detected contained sulfur and (b ) chromatographic behavior on 23 dissimilar column types and (c) aroma descriptor. This is apparently the first study to report aroma activity of MeSH, DMS and DMTS in GFJ. Earlier GC-O studies ( 10, 28, 33 ) employed solvent extraction for isolation of aroma compounds. The extremely high volati lity of some of these compounds (MeSH and DMS) makes their chromatogra phic resolution from the solvent peak difficult and health concerns would preclude huma ns from sniffing during the time solvents were eluting. As seen from the cross hatched bars in Figur e 4-6, the aroma intensities of these six VSC’s were substantially higher in canned RFC juice as compared to FHS juice. This was an expected outcome, as all six compounds were present in significantly higher conc entrations in canned RFC juice (chapter 3). Four of the six aroma active sulfur compounds were found to be thermal breakdown products resulting primarily from Smethyl methionine (producing DMS) and methionine (producing Met, MeSH and DMTS). This indicates th at Smethyl methionine and methionine are the major nonvolatile precurso rs in GFJ, significantly impacting its aroma quality. Interestingly, the aroma quality of PMT was de scribed as ‘fresh grap efruit’ when detected in FHS juice. The same peak in canned RFC jui ce was described to possess an ‘intense sulfur’


74 quality. This finding is in agreement with that re ported in the earlier GC -O work of Lin et al. ( 10 ) investigating the thermal concentration effects on GFJ aroma quality. The authors reported PMT aroma activity only in RFC juice and not in fresh juice thereby suggesting its formation during heat processing. Formation of this compound from limonene and H2S as suggested by these authors ( 10 ) remains to be established. Some prelim inary work in this study has indicated that such a reaction may occur over prolonged ther mal abuse or elevated temperature storage. Identifications for 4MMP, 4MMPol, and ne wly detected furfuryl mercaptan were considered only tentative as they did not meet all of the above criteria These three aroma active peaks were found in higher intensities in fres h juice as compared to canned juice. Flavor reconstitution studies in model solutions ( 29 ) suggested that 4MMP was a more significant contributor to fresh grap efruit aroma than PMT. The remaining VSC’s identified previously (T able 4-1) did not possess aroma activity at the levels in which they were extract ed. Of particular significance is H2S, which is the predominant sulfur volatile in fresh juice (T able 4-1). This compound has been previously speculated as important contributor to the aroma of fresh squeezed GFJ ( 11, 31, 32 ) but its aroma activity was not investigated in the earlier GC-O studies ( 10, 28, 33 ) because these studies employed solvent extraction. The lack of aroma activity of H2S observed in this study, suggests that both cysteine and glutathione, which are the primary precursors of H2S, are not immediate contributors to grapefru it aroma. However, H2S is relatively reactive and may undergo subsequent reactions to produce aroma activ e sulfur volatiles in heated juices. Nonsulfur aroma peaks : At least four nonsulfur p eaks contributed to the fresh grapefruit aroma quality of FHS juice. These we re described as ‘fresh GF/ GF peel/ spicy pungent’ and found present only in the FHS juice a nd not in canned RFC juice. These have been


75 tentatively identified as nootka tone, 1, 10dihydronootkatone, be tasinensal or isoeugenol, eugenol (peaks 51, 50, 49 and 48 respectively). Both nootkatone and 1, 10 dihydronootkatone have been reported as importa nt contributors to the characteristic grapefruit aroma ( 23, 24 ). The ‘sweet/ fruity/ minty’ odors were also f ound to be predominantly associated with FHS juice. As expected, these odor qualities were ma inly due to esters. The seven odoractive peaks in this category included: ethyl 2-methyl propanoate, ethyl buta noate, ethyl-2-methyl butanoate, ethyl hexanoate, 3-hexenol-acetate and two unknown peaks (peaks 15 and 40). Five of these peaks were perceived in detectab le levels only in fresh juic e and not the canned RFC juice. Similarly, 11 out of the 15 green/ fatty/ metallic/ mushroom notes were detected in higher aroma intensities in FHS juice as compared with canne d RFC juice. These odors were primarily due to aldehydes (9), a few ketones (2), terpenes (2), alcohol (1) and ot her unidentified peaks (3). Substantial quantitative and qualitative differe nces exist between the aroma volatiles of fresh unheated GFJ and canned RFC juice, with VSC’s being important contributors to these differences. Matrix Distribution of Volatile Sulfur Co mpounds in Canned RFC Grapefruit Juice Citrus juices are a multiphase system comprising an aqueous ‘Serum’ phase containing soluble solids and insoluble matte r called ‘Pulp’ ranging from 2 m to several hundred m. The main objective of this study was to determine th e overall volatile distri bution, including sulfur volatiles between the pulp and serum components of canned RFC juice. This juice type was selected mainly due to presence of measur eable amounts of numerous VSC’s including the grapefruit character imparting, PMT (as seen in Chapter 3). In addition, it represented a typical commercially homogenized pulp system. Figure 4-7 summarizes the overall volatile organic compound (VOC) distribution between the pulp and serum components of canned RFC grapef ruit juice separated by ultracentrifugation


76 and analyzed using GCMS. The 39 major VOC ’s found in canned RFC juice showed unique distributions between certain type s of juice volatiles. These type s were categorized broadly into four chemical groups consisting of monoterpene s, sesquiterpenes, oxygenated terpenes and oxygenated hydrocarbons. Pulp’s volatile composition was dominated by monoterpenes (96% of total VOC’s). As seen in Figure 4-7, monoterpenes were distribut ed in significantly hi gher amounts in pulp (about 1.5 times higher) than in serum. D-Limonene was the principal volatile accounting for up to 92% of the pulp’s monoterpene content. As seen in Table 4-4, the ‘citrusy’ smelling Dlimonene is only as a minor contributor to GFJ aroma. The other minor components in this category were pinene (green smell), myrcene (mushroom smell), myrcene, phellandrene, (E) ocimene, (Z) ocimene and terpinolene. Sesquiterpenes contri buted to another 3% of the VOC’s distributed in pulp. Of particular mention, is the ‘fresh GF’ smelling sesquiterpene ketone, nootkatone, which was about equa lly distributed between the pulp and serum components. Serum’s volatile composition differed from that of pulp to a great extent. Monoterpenes accounted for about 87% of the serum VOC’s. Oxygenated terpenes, which were only a minor component of pulp VOC’s, contri buted to as much as 8% the total VOC’s distributed in the serum. This was an expected outcome, considerin g the affinity of such polar volatiles for the aqueous serum phase. The major compounds in this category were alcohols such as: terpineol, terpineol, citronellol, the ‘floral’ smelli ng linalool and some terpene esters. Linalool content in serum was up to ten times higher th an in pulp. In addition, much of the oxygenated hydrocarbons found in juice, were also distri buted largely into th e serum fraction. These comprised two alcohols, four esters, four aldehydes and three ketones.


77 These observations confirm the early works ( 72 ) on VOC distribution between pulp and serum components of different citrus juices (i ncluding grapefruit). In their study, monoterpenes, sesquiterpenes and long chain oxygenated compounds were reported as primar ily associated with the pulp fraction ( ~ 80% of the total juice volatiles) and the citrusy smelling oxygenated compounds (small chain esters, alcohols, aldehy des) and oxygenated te rpene, linalool were reported as largely present in serum. Similar volatile distributions ha ve been reported more recently for solvent extracted serum and pulp fracti ons of fresh squeezed ‘naveline’ orange juice ( 74 ). However, none of these studies investigated VSC distribution within the juice matrix. Figure 4-8 represents the dist ribution of selected VSC’s identified in the canned RFC grapefruit juice matrix components. The pul p and serum fractions, separated by ultracentrifugation, were analyzed us ing GCPFPD. Most of the lo w MW (< 100) and high vapor pressure compounds: H2S, MeSH, DMDS, 2MTP and 3MTP were found distributed in the serum to a greater extent than in the pulp. An ex ception was the major VSC, DMS, which was found almost equally distributed betw een the pulp and serum fractions of canned RFC grapefruit juice Two middle MW VSC’s (100150) such as Met and DMTS presented a similar distribution trend as the other low MW sulf ur compounds. Both high polarities and vapor pressures of these VSC’s appear to enhance their distribution in to the aqueous serum phase. PMT (intense sulfur) and 3MHA (fresh GF), which represent the hi gh MW sulfur volatiles (> 150) were found be more strongly associated with the pulp portion of GFJ. Further the struct ural similarity between PMT and terpenes (for e.g. D-limonene) also e xplain the higher distri bution of this compound into the pulp fraction of juice. In summary, some of the important ‘g rapefruit’ smelling compounds were found distributed to a large extent in the pulp fr action. These findings are of significance to the


78 processing industry, where flavors lo st during concentration step ar e typically restored back to the RFC juice. This is often done without consid eration given to the propo rtions in which aroma compounds are present in the rec onstituted juice or amount of pul p added back to the juice. Conclusions Unique VSC patterns were observed in fresh an d commercially processed grapefruit juices. Several VSCÂ’s found in processed juices, were shown to be produc ts of thermal degradation of sulfur containing amino acids. Of particular si gnificance were methionine and its methylated form, smethyl methionine. At least six ar oma active sulfur compounds were identified in grapefruit juices. The altered aroma quality of RFC juice could be explained by the significant loss of some aroma compounds and the formation of some others particul arly sulfur containing compounds. The matrix distribution of VSCÂ’s in grapefruit juice has also been established.


79 Table 4-1. Quantities in parts per billion (g /L) of volatile sulfur compounds in fresh, not from concentrate and reconstituted from concentrate grapefruit juices. Fresh juice NOT FC Recons FC Peak Sulfur compound ZB-5 LRI mean range mean range mean range 01 hydrogen sulfide <500 1.8a 0.49 4.8 0.09b nd 0.25 0.14b 0.03 0.49 02 sulfur dioxide <500 0.27a nd 0.81 0.081b nd 0.35 0.072b nd 0.35 03 methanethiol <500 0.069b nd 0.64 0.42b nd 0.85 0.99a 0.28 2.64 04 dimethyl sulfide 526 0.24b 0.032 1.06 0.902b 0.24 0.08 4.8a 2.1 10 05 carbon disulfide 546 0.040b nd 0.23 0.65a 0.07 3.1 0.43a nd 0.99 06 unk 679 nd b nd nd b nd 0.017a nd 0.037 07 unk 704 0.002b nd 0.035 0.014a nd 0.067 nd b nd 08 dimethyl disulfide 750 nd c nd 0.065b nd 0.29 0.13a nd 0.36 09 2-methyl thiophene 777 nd c nd 0.012b nd 0.065 0.039a nd 0.088 10 3methyl thiophene 786 nd b nd tr b tr 0.024a nd 0.06 11 unk 872 0.017a nd 0.089 nd b nd nd b nd 12 methional 912 nd b nd 0.066a tr 0.19 0.041a tr 0.084 13 unk 921 nd b nd nd b nd 0.028a nd 0.19 14 dimethyl trisulfide 981 nd c nd 0.036b nd 0.14 0.101a tr 0.42 15 unk 995 tr b tr 0.007a nd 0.084 0.007a nd 0.06 16 unk 1007 nd b nd tr a tr 0.005a nd 0.075 17 unk 1037 tr tr tr tr tr tr 18 unk 1244 nd nd nd nd tr nd 0.089 19 3-mercaptohexyl acetate 1258 0.005a nd 0.048 0.009a nd 0.074 0.038a nd 0.34 20 2,8-epithio-cis-pmenthane 1274 nd b nd 0.13a nd 0.44 0.201a 0.092 0.58 21 1-p-menthene-8-thiol X 1298 nd nd nd nd nd 0.87


80 Table 4-1 Continued. Fresh juice NOT FC Recons FC Peak Sulfur compound ZB-5 LRI mean range mean range mean range 22 unk 1365 0.009b nd 0.055 0.048a nd 0.27 0.052a nd 0.32 23 unk 1370 0.009b nd 0.053 0.052a nd 0.27 0.060a nd 0.38 24 unk 1390 nd b nd nd b nd 0.043a nd 0.308 25 unk 1409 nd b nd nd b nd 0.081a nd 0.22 26 unk 1445 0.012b nd 0.085 0.023b nd 0.16 0.14a nd 0.34 27 bis(2-methyl3furyl)disulfide* 1547 0.039a nd 0.12 nd b nd nd b nd 28 unk 1577 0.038a nd 0.083 0.007b nd 0.086 0.008b nd 0.14 29 unk 1670 0.005b nd 0.037 0.074a nd 0.28 0.006b nd 0.096 30 unk 1768 0.022a nd 0.059 nd b nd nd b nd 31 unk 1775 0.42a 0.053 0.98 0.28b nd 0.91 0.23b nd 1.2 32 unk 1796 0.026a nd 0.095 0.004b nd 0.048 nd b nd 33 unk 1827 0.041a nd 0.13 0.017a nd 0.12 nd b nd total # sulfur compounds detected 21 25 27 average total sulfur content 3.10 3.00 7.80 Data are means of 10 fresh, 11 NFC and 9 RFC juices over four replications each. Means followed by a different letter in the sa me row are significantly different at 95% confidence level. Nd, not detected. tr, trace, detected but too small to quantify. Compounds tentatively identified. X 1-p-menthene-8-thiol detected only in canned RFC juice sample, hence mean and statistical differences could not be determined.


81 Figure 4-1. Volatile sulfur comparison of unheated (0) and heated (95C; 2, 6 and 15 hour) fresh hand squeezed Marsh grap efruit juices (n=3).


82 Figure 4-2. Principal component analysis for VS C distribution in 10 fresh (), 11 NFC () and 9 RFC () grapefruit juices.


83 Figure 4-3. Canonical roots from discriminant function analys is showing separation of unheated GFJÂ’s () from heated GFJÂ’s combined: NFC () and RFC ().


84 Figure 4-4. Canonical roots from discriminant function analys is showing separation of unheated (), NFC () and RFC () GFJÂ’s.


85 Table 4-2. Discriminant function analysis (forwa rd step) model for segregation of fresh and heated juices. # Volatile Sulfur Compounds Wilks' Lambda Partial Lambda p-level 01 Methional 0.32 0.55 0.0005 02 Hydrogen sulfide 0.29 0.59 0.0013 03 Sulfur dioxide 0.23 0.75 0.026 Table 4-3. Discriminant function analysis (forward step) model for segregation of fresh, not from concentrate and from concentrate juices. # Volatile Sulfur Compounds Wilks' Lambda Partial Lambda p-level 01 Methional 0.060 0.565 0.003 02 Dimethyl sulfide 0.063 0.533 0.001 03 Dimethyl trisulfide 0.067 0.503 0.001 04 Hydrogen sulfide 0.062 0.543 0.002 05 Sulfur dioxide 0.049 0.688 0.020 06 Methanethiol 0.045 0.744 0.045 07 Dimethyl disulfide 0.039 0.868 0.227


86 Figure 4-5. Volatile sulfur breakdown products of sulfur containing amino acids heated (95C; 3 hr) in model GFJ system (pH 3.4). Data average over three replicates. The Y axis represents a logarithmic scale.


87 Table 4-4. Aroma active compounds in Fresh and Canned RFC grapefruit ju ice. Identifications are tentative based on retention beha viors on two column materials. Aroma descriptor Compounds FHS RFC canned LRI Wax LRI ZB-5 01 sulfur methanethiol 5.6 6.2 661 <500 02 cabbage dimethyl sulfide 5.1 775 518 03 fruity ethyl 2-methyl propanoate 4.7 974 757 04 sour unknown 5.2 980 05 green a-pinene 4.2 6.0 1026 940 06 minty ethyl butanoate 7.9 7.5 1038 800 07 fruity ethyl-2-methyl butanoate 5.3 1059 850 08 green hexanal 8.4 1083 802 09 green, piney, metallic unknown 4.3 1092 10 mushroom unknown 7.1 1099 11 mushroom b-myrcene 7.7 6.9 1163 980 12 mushroom 2-octanol 3.9 1183 997 13 citrusy minty a-terpinene 6.1 1.5 1195 1010 14 citrusy minty limonene 5.8 5.3 1204 1034 15 minty unknown 7.8 6.4 1211 16 fruity ethyl hexanoate 5.8 1238 1002 17 floral unknown 5.4 1247 18 citrusy octanal 9.7 10 1294 1002 19 mushroom 1-octen-3-one 5.3 1.5 1308 980 20 sweet fruity 3-hexenol-acetate 5.5 1341 1004 21 jasmine rice, roasted 2 acetyl-1-pyrolline 6.0 5.4 1348 925 22 mushroom 1, 5 z-octadien-3-one 7.2 1.5 1376 982 23 sulfury, garlic dimethyl trisulfide 5.4 7.5 1388 975 24 sulfury 4-mercapto-4methyl-2-pentanone 4.1 1390 949 25 citrusy piney nonanal 3.5 5.7 1400 1110 26 roasted coffee furfuryl mercaptan 6.0 4.0 1438 910 27 cooked potato methional 4.3 6.8 1455 914 28 floral citrus unknown 5.2 1490 29 citrus detergent, geranium Z-2nonenal 6.4 9.9 1506 1147 30 fatty E-2-nonenal 4.0 1512 1155 31 floral unknown 7.2 7.9 1541 32 floral linalool 7.8 8.3 1551 1104 33 grapefruit, sulfury 4-mercapto-4-methyl2-pentanol 5.6 1556 1047 34 cucumber E, Z 2,6 nonadienal 4.3 3.9 1594 1162 35 grapefruit peel unknown 4.4 4.0 1605


88 Table 4-4. Continued. Aroma descriptor Compounds FHS RFC canned LRI Wax LRI ZB-5 36 fresh grapefruit Or intense sulfur 1-p-menthene-8-thiol4.6 8.7 1620 1295 37 green citrus fatty E-2-decenal 5.8 1650 1252 38 fatty dodecanal 5.1 5.1 1710 39 fresh grapefruit 3-mercapto hexyl acetate 5.2 1705 1257 40 fruity unknown 4.1 1730 41 lemon unknown 4.9 4.7 1750 42 cilantro E,Z-2,4-didecadienal 5.3 2.8 1763 43 fatty E,E-2,4-didecadienal 4.0 1823 1319 44 sweet medicine bdamascenone 5.0 4.6 1833 1390 45 spicy citrusy fatty unknown 4.3 5.7 1857 46 raspberry b-ionone 5.9 9.8 1955 1494 47 caramel unknown 4.6 4.2 2040 48 fresh grapefruit spice, pungent eugenol 5.6 2180 1365 49 fresh grapefruit spice isoeugenol or bsinensal 4.1 2290 1410 or 1681 50 fresh grapefruit 1, 10dihydronootkatone 5.5 2359 1681 51 grapefruit Nootkatone 4.8 2540 1820


89 Figure 4-6. Aroma intensities of Fresh (grey bars) and Canned RFC (black bars) grapefruit juice volatiles. Intensities are based on normalized peak heights. Sulfur containing aroma peaks are indicated by crosshatched bars.


90 Figure 4-7. Major volatile distributio n in pulp (solid bars) and serum (bars with dots) fractions of canned RFC grapefruit juice analyzed us ing HSSPME coupled with GC-MS. Data averaged over three replicates.


91 Figure 4-8. Volatile sulfur compoun d distribution in pulp (solid ba rs) and serum (bars with dots) fractions of canned RFC grapefruit jui ce analyzed using HSSPME coupled with GC-PFPD. Data averaged over three replicates.


92 CHAPTER 5 A COMPARISON OF CITRUS BLOSSOM VOLATILES Introduction Although citrus blossoms are known for their pl easant, highly desirable aroma, few studies have analyzed the volatile composition of intact citrus blossoms. Histor ically, citrus blossom composition has been studied using blossom oils obt ained from steam distillates of the blossoms. This approach produced samples with high volat ile concentrations which were necessary for separation and identification tec hnologies available at that time However, the heat applied during this process produced oils whose smell bore only limited resemblance to the original blossoms. Sour orange ( Citrus aurantium L. var amara) blossom oil from this process has been extensively examined, due to its comme rcial importance as oil of Neroli ( 14, 84 ). Attaway et al. ( 85 ) employed steam distillation to prepare blossom oils to study the compositional differences between orange, grapefruit and tang erine blossoms. However, as an alytical detection limits have lowered, it is no longer necessary to prepare th e blossom oil. Blossom volatiles can now be examined using headspace techniques which do not require heating and will more closely reflect the blossom’s native volatile profile. Conventional taxonomic classifications of th e citrus subgenus base d on morphological and geographical characteristics has produced two major systems ( 94, 95 ) with greatly differing number of citrus species. La ter taxonomic studies concluded th at there were only three ‘true’ citrus species and all other cultivars were crosses of these primary types ( 96 ). More recent phylogenic classification studies have employed DNA biochemical ma rkers to identify ‘true’ citrus species ( 97, 98 ). Chemical composition has also been employe d in devising taxonomi c classifications. Classifications of citrus cult ivars using volatile secondary me tabolites in fruit peel oil ( 14, 15 ), as


93 well as Petitgrain and Neroli oils ( 14 ) have been explored. Other workers ( 13 ) have examined citrus blossom volatiles using purge and trap methods. They examined their blossom volatile pattern data using cluster analysis (dendogram s) and related resulting classifications to morphological characteristics, such as fruit size. There is little information available on volat ile composition of intact blossoms of major citrus species. Therefore, th e primary objective was to examin e headspace volatile composition of 15 citrus cultivars blossoms to determin e if they possessed unique volatile profiles. A secondary goal was to explore the possibility that blossom volatiles from individual cultivars were sufficiently unique that they could have chemotaxonomic potential. Another secondary goal was to identify any potential biol ogical significance due to the observed differences in blossom volatile composition. Materials and Methods Sample Collection Samples were procured from the Florida Ci trus Arboretum (Winter Haven, Fl) and from the groves at the Citrus Research and Educati on Center (Lake Alfred, Fl). Blossoms were harvested during a nine day period from March 24 to April 1 2008 (with average temperatures: 66.9 + 6.2 F and average R.H. 68.1 + 8.8). Trees at their peak bloom stage were selected for sampling. Specifically, only those blossoms at an ear ly stage in their development (as determined by their partially opened petals) were picked. Ty pically, these young blossoms were picked from all around the tree. These parameters collectivel y ensured sampling of blossoms at a uniform development stage across cultivars, limited sun exposure and a more representative sampling. In a day, no more than three cultivars were harveste d and the blossoms were analyzed the same day as harvested. Whole blossoms including petal, pistil, stamen and a short pedicel were picked before 9 am and refrigerated in s ealed containers (4C, 3 h max.) unt il analysis to limit the rate of


94 respiration and deterioration if not immediately analyzed. Each blossom cultivar was analyzed in duplicate with the exception of ‘dancy’, ‘pummelo ’ and ‘Duncan’ being anal yzed in triplicate. A new sample vial containing fresh blossoms was used for each replicate analysis. Headspace Volatile Analysis of Blossoms Blossoms (3g) were placed into a 42 mL gl ass vial capped with a Teflonfaced silicone septa. Samples were equilibrated for 20 minutes at room temperature (RT) prior to analysis. A SPME device from Supelco (Bellefonte, PA) with 1 cm fiber coated with 75 m CAR/ PDMS was used to collect volatiles. In addition, to in itial conditioning, the fi ber was conditioned at 200C for 5 minutes prior to each blossom sampling. The SPME fiber was exposed in equilibrated blossom headspace for 30 min at RT to collect volatiles. GC/MS conditions A Clarus 500 quadruple GC-MS from Perkin El mer (Waltham, MA) was used for blossom volatile analyses. Blossom volatiles from the SPME fiber were desorbed in the GC injection port (splitless mode) at 200C. The fiber was removed after 5 minutes exposure in the injection port. Volatiles were separated using a Stabilwax (Restek, Bellefonte, PA) capillary column (60m x 0.25 mm i.d x 0.5 m). The oven was programmed from 40C (held for 2 min) to 240C at 7C/min with a final hold of 9.5 min. Helium was used as the carrier gas at 2.0 ml min-1. The MS was operated in the electron ioniza tion mode (EI+) at 70 eV with a transfer temperature of 240C and a source temperature of 180C. The mass spectrometer scanned masses from 25 m/z to 300 m/z at 0.2 sec scan time with 0.1 sec of inters can delay. Chromatogram s were recorded and integrated using Turbomass 5.4 data acquisition software from PerkinElmer (Suwannee, GA). Peak identification Peaks from the chromatogram were identif ied based on their mass spectra and linear retention index values. Mass sp ectra for unknowns were compared with standards from NIST


95 2005 (National Institute of Standards and Tec hnology, Gaithersburg, MD) mass spectra database. Only spectral match values of > 900 with an LRI value with 1% of standards were considered for identification. Linear retent ion index, LRI, values (forme rly, KovatÂ’s Index values) of chromatographic peaks on Stabilwax column were determined using a series of alkanes (C5C25) run under identical conditions for GC-MS. Statistical Analysis The dataset included analytical values from all volatile components measured. Principal component analysis (PCA) in Unscrambler 9.7 from Camo (Woodbridge, NJ) was used to evaluate the data set. Analysis of variance was employed to identify those volatiles that would be most differentiating between major citrus sp ecies. ANOVA was perfor med using Statistica 5 from StatSoft (Tulsa, OK). Means were separated usi ng TukeyÂ’s HSD test at = 0.05. Results and Discussion The 15 citrus cultivars shown in Table 5-1 were chosen as they represented a wide range of citrus blossoms. Some cultivars such as Bouquet de Fleurs' are highly prized for their fragrance and whose volatiles are extracted for high value pe rfumes. Other cultivars such as the Kaffir lime were included because of their unusual size and aroma. Blossom Volatile Composition Traditionally, blossom volatile composition ha s been examined after concentration using either steam distillation or solv ent extraction. Attaway et al. ( 85 ) used steam distillation to prepare oils from blossoms and different bl ossom parts. They found higher levels of sesquiterpenes such as -caryophylle ne and valencene in pistil oils compared to petal oils. Unfortunately, the aroma of extracted oils rare ly represents the delicate natural aroma of blossoms due to thermal artifacts produced during steam distillation ( 13 ). More recently, headspace analysis has been employed to sample bl ossom volatiles in a more representative way.


96 Toyoda et al.( 13 ) employed a tenax trap for capturing the headspace volatiles followed by solvent desorption of the extracts. This tec hnique concentrates many of the most volatile blossom components, but some of the most vol atile components are ma sked under the solvent peak and cannot be analyzed. Solid-phase micr oextraction (SPME) is a relatively new and solventless headspace technique which combines bot h isolation and concentr ation of volatiles in a single step. In this study static headspace SPME was employ ed to collect blossom volatiles. As seen in Figure 5-1, substantial differences in the to tal amount of volatiles emitted by blossoms were observed between the 15 cultivars as measured by GC-MS total ion current (TIC) peak areas. Pummelo blossoms produced highest levels of volatiles, whereas Kaffir lime blossoms emitted only ~23% as much. The height of the bars in Figure 5-1 represents the relative differences in total blossom volatile quantity in terms of peak area normalized to total peak area of pummelo. Individual stacked bars reflect the relative composition of blo ssom volatiles by chemical group. Volatiles were categorized into five chemical groups consisti ng of oxygenated terpenes, acyclic terpenes, cyclic terpen es, and nitrogencontaining com pounds. The fifth group comprised several miscellaneous compounds such as alipha tic hydrocarbons, alcohols esters, ketones and aldehydes. Aromatic compounds, se squiterpenes and furans were also included in this group. The 15 cultivars were grouped into six broad ci trus types consisting of pummelos, mandarins, grapefruits, sweet oranges, sour oranges and lemonslimes. Lemon and lime cultivars were grouped together due to similarities in thei r volatile composition, except Volkamer lemon and Kaffir lime which were distinctly different from the other lemons and limes as discussed later. Oxygenated terpenes, monoterpenes and nitrogen-containing compounds together constituted 90% or more of the total blo ssom volatile composition for all 15 cultivars.


97 Oxygenated terpenes were the predominant volatile s in pummelo and sour orange blossoms. In contrast, monoterpenes dominated the headspac e of lemonslimes, Volkamer lemons, Kaffir limes, mandarins, grapefruits and sweet oranges. Sour orange blossoms exhibited the largest relative amounts of esters and ketones perhaps ex plaining their value as perfume ingredients. Additional differences among the minor components in each chemical class further distinguished these blossoms from each other. MS identifications A total of 70 volatile compounds were detected in the headsp ace of the citrus blossoms analyzed in this study. Earlier studies of aromatic and medicinal plants ( 110 ) had found that a biphase fiber coating of polydimethylsiloxane (P DMS) and Carboxen was the most effective for detection of plant volatiles. Therefore, this fi ber type was employed for the current study. Of the 70 peaks reported, 66 were identified by matchi ng their retention ch aracteristics (DB-Wax column) and MS fragmentation patterns against sta ndards as well as library spectra. In addition, one peak could only be identified as an unknown terp ene and three others could not be identified at this time. Table 5-2 lists the 70 blossom volatiles de tected in this st udy. Relative amounts of individual blossom volatiles are presented as peak area percent of the total volatile peak area for each blossom type in Table 5-2. Twenty-nine vola tiles are reported in citrus blossoms for the first time. Thirty-seven blossom volatiles have been previously reporte d and are noted in the table along with the original c itation. Of the 66 volatiles identified, th ere were 16 oxygenated terpenes 17 monoterpenes (six acy clic, six cyclic and five bicyclic ), four alcohols, three esters, three ketones and five nitrogen volatiles. Attaway et al. ( 85 ) detected 59 volatiles in their citrus blossom study and identified 38, of which 25 were also identified in th e current study. Choi ( 93 ) employed SPME of only a singl e phase, PDMS, to analyze C. unshiu blossom volatiles. Of the


98 51 volatile compounds they identified, 12 monoterpe nes accounted for 84% of the total volatiles. This is not surprising as PDMS is known to be hi ghly selective for terpenes Twenty-four of the 66 volatiles identified in the current study were also reported by Flamini et al. ( 91 ) who also employed SPME to analyze blossoms and blossom parts from C. deliciosa. Oxygenated terpenes: This group comprised predominantly terpene alcohols, along with terpene esters, ethers and aldehydes, which were the minor constituents. Linalool was the most abundant terpene alcohol in citrus blossoms accounting for up to 45% of pummelo and sour orange total blossom volatiles. Linalool, a floral-smelling compound, is commonly found in many flowers including citrus blossoms ( 85, 88, 111-113 ). Other monoterpene alcohols such as citronellol were typically found as a trace component (<2 %). However, in Kaffir lime citronellol contributed to >25% of total volatiles, thus making this cultivar distinct from all other blossoms. Other terpene alcohols such as, nerol, geraniol and nerolidol were also identified. Acyclic monoterpenes: These non ringed terpenes tend to have lower aroma thresholds than similar ringed terpenes of comparable molecular weights. -myrcene, –myrcene and (E)ocimene were the major acyclic monoterpenes identified in this study. Mandarin blossoms, had the highest acyclic terpene content and were primarily comprised of -myrcene (~45% total), (E)-ocimene (~18% total) and -myrcene (~5% total) Cyclic monoterpenes: These terpenes have a single ring and are more stable with relatively high aroma thresholds compared to thei r acyclic equivalents. Limonene and p-cymene were the major cyclic monoterpenes found in this study. Limonene alone contributed to almost 60% of lemonlime total vol atile composition. Vo lkamer lemon blossom’s monoterpene composition resembled that of mandarins more th an other lemonslimes. However, appreciable levels of limonene (~12%) further differentiated this cultivar from ma ndarins. In contrast,


99 Pummelo blossoms also differed from all other cu ltivars due to high levels (13.6%) of p-cymene. No other cultivar had more than 4.1%, with most containing between 0.6 and 2% of total volatiles. Bicyclic monoterpenes: Sabinene, the major volatile in this category was found primarily in Kaffir lime blossoms as its second most abundant volatile (18% of total headspace composition). This further contributed to Kaffir li meÂ’s distinct volatile composition compared to all other blossoms (~2 % sabinene). Alpha and -pinene are chemically unstable bicyclic terpenes due to a strain ed 4 membered ring. As such they are found at low levels and evenly distributed in all the blo ssom types in this study. Nitrogen compounds: Nitrogen volatiles are potent floral volatiles that are difficult to detect and have been missed in some recent citrus blossom studies ( 93 ). As seen in Table 5-2, methyl anthranilate and indole we re nitrogencontaini ng compounds found in all citrus cultivars. Both these compounds have been previously report ed in oil extracted from bitter (sour) orange blossoms as summarized by Boelens and Gemert ( 84 ). Methyl anthranilate was initially identified by Attaway et al. ( 85 ) in citrus blossom oils and as seen from Table 5-2, found in relatively high levels (>5%) in pummelo, sweet orange and grapefruit headspace. It is a character impact volatile in Conc ord grapes and found in many othe r plant materials. In addition, it serves an important biological role in plant defense as a bird repellant. The chemosensory irritation caused by this compound has been employed to not only protect crops ( 114, 115 ) but also prevent bird/aircraft accidents ( 116 ). Indole, on the other hand, is relatively evenly distributed in all cultivars, but found in slightly higher levels in mandarin blossoms. It is a fecal aroma component which surprisingly has a fl oral character at trace concentrations.

PAGE 100

100 Principal Component Analysis The wide variability in citrus blossom volatile composition shown in Table 5-2 was unanticipated. In order to examin e this large data set without having to assign classifications ahead of time, Principal Component Analysis (PCA) was employed. It is an unsupervised learning technique used to determ ine hidden structure (a ssociations) in large data sets and to determine those volatiles whic h were most differentiating. PCA is an exploratory technique which reduces the multidimensional data set (70 blossom volatiles) to lower dimensions or principa l components which are comprised of linear combinations of variables (individual volatiles). This helps identify inherent patterns in the data in an unbiased way and highlights the similari ties and differences amongst samples (citrus cultivars). It also helps identif y those volatiles which are most di fferentiating within the entire data set. PCA Score plot: Eigenvector values for each sample in the first two dimensions are shown in Figure 5-2. The high degree of cluste ring and minimal overlap suggests that citrus blossoms have volatile profiles that are unique to each cultivar. The first two principal components, PCs, together account for 83% of th e total variance in the blossom peak area data. Citrus cultivars within the same subgenus were clustered more closely to one another than to those belonging to different subge nus. There were three widely se parated clusters in Figure 5-2 consisting of: pummelo (group A), lemonslimes (group B) and mandarins (group C). Sour oranges were also fairly tight ly clustered but are located be tween the lemon-lime and pummelo clusters. Overall, sweet ora nges and grapefruits were found to be more closely related to mandarins, whereas sour oranges were positioned closer to pummelos. This segregation of varieties using blossom volatile distribution is in agreement w ith phylogenic studies based on morphological and biochemical characteristic s (DNA markers) which identified citron (C.

PAGE 101

101 medica ), mandarin (C. reticulata ) and pummelo (C. grandis ) as the only ‘true’ citrus species with all others being cultivated biotypes ( 97, 98 ). Lemons and limes, which were found as a single distinct group in our study, are believed to be clos ely related to citron ( 117 ). This ancestral link of lemons and limes to citron has been also confirmed with DNA marker studies ( 98 ). Two cultivars, Volkamer lemon and Kaffir lime, whose common names suggest that they should belong in the lemon-lime group, were not found near the main lemon-lime cluster in Figure 5-2. This suggests that they do not belong in that group, at least in terms of their blossom volatile compositions. At least six DNA marker s which were common to both Volkamer lemon and mandarins were observed in a study by Nicolosi et al. ( 98 ). This genetic similarity could partly explain the observed proximity of Volkam er blossoms to the mandarin group in the PCA score plot. As seen in Figure 5-2, Kaffir lime ( C. hystrix ) occupied a distinct position on the PCs due to its unique volatile compos ition as discussed earlier. These re sults were also in accordance to the phylogenic findings of Nicolosi et al. ( 98 ), which placed C. hystrix in a cluster genetically distinct from the ‘citron’ cluster PCA Load plot: A 2D load plot (Figure 5-3) of th e data set compressed on first two PCs, reveals the influence of individua l volatiles in differentiating betw een citrus cultivars when the entire data set is considered. Those volatiles that account for maximum variance in the data set are given more weight or loading. Therefore, as seen in Figure 53, linalool (0.8 on PC1), limonene (-0.5 on PC1 and -0.6 on PC2) and myrcene (0.7 on PC2) were the most differentiating volatiles. If the PCA load plot is overlaid on the PCA score plot the relationship between heavily weighted volatiles to cultivar groups can be seen. For example, high linalool content differentiated pummelo varieties from le monslimes which had the lowest levels of linalool. However, high limonene content in lem onslimes differentiated them from mandarins

PAGE 102

102 which were also lower in linalool. Variation in myrcene content further segregated these cultivars. Load values for all other volatiles were located near origin (load values close to 0) suggesting that they do not play an important role in differentia ting between blossom cultivars. The third dimension, PC3, explained an additional 8% of the total variability (data not shown). In PC3, (E)-ocimene (loading of 0.4) appeared to be of some relevance to the observed variability amongst cultivars. This diverse volatile distribu tion between citrus blo ssoms demonstrated in Figures 5-1, 5-2 and 5-3 strongl y suggest that there are metabolic differences among citrus cultivar blossoms and substantiates the idea th at blossom volatile prof iling could serve as taxonomic markers. Possible Insect Plant Interactions The biological significance of plant volatiles as chemical cues for communication with insects has been widely accepted ( 99, 100 ). Discrimination between plant species by foraging insects is based on odor recogni tion of complex mixtures of volatiles as well as specific individual components ( 99, 100 ). Volatile emission by the flowers of some species appears to be specific to the pollination needs of individual plants. Theis and Raguso ( 101 ) observed two species of thistle, whose emissi on of blossom volatiles coincided with pollination stage. Overall blossom emissions were highest prior to pollination and decreas ed soon after pollination which reduced the incidence of subsequent pollinator visits. Thiery et al. ( 102 ) reported that only 24 of over 100 sunflower volatiles stim ulated olfactory responses in worker honeybees using GC coupled with EAG (electroantennography). Citrus blossom cultivars differed significantly in their headspace levels of certain volatile compounds Analysis of variance revealed a total of 12 out of 70 volatile compounds that were differen tiating between the eight cultivar groups. The remaining 58 volatiles showed no significant differe nces amongst the cultivar types. Table 5-3

PAGE 103

103 lists these 12 volatiles and the significant differences between group means determined using TukeyÂ’s HSD test ( = 0.05). Pummelo blossoms were found to emit significantly higher levels of -pinene, p-cymene, 1-hexanol and linalool as compared to all othe r major citrus types. Compounds such as 1hexanol and linalool have demonstrated olf actory stimulation of honeybee antennae in electroantennography ( 102 ) and calcium imaging studies ( 118 ). A possible explanation for pummelo to emit higher levels of honeybee attr acting volatiles could be the biological dependence of this species on polli nating insects. Whereas most ci trus species have both sexes on the same blossoms, most pummelo varieties a nd certain mandarin varieties (not used in our study) are self-incompatible ( 119 ). Since cross-pollination by honeybees is crucial for pummelo survival, it should not be surprising that pumme lo blossoms exhibited the observed levels of honeybee attracting volatiles. Conclusion HS-SPME analysis of citrus blossoms reve aled a diverse volatil e distribution between major citrus cultivars. Linalool, -myrcene, -myrcene, limonene, (E)-ocimene, methyl anthranilate and indole were the major volatiles produced by citrus blossoms. The distribution of citrus cultivar eigenvector scor es from PCA revealed three wi dely separated, tight clusters consisting of pummelo, mandarins and lemonslim es. Since PCA is a non supervised learning technique used to determine struct ure in data set, this widely separated clustering indicated that these cultivars were the most different amongst a ll the cultivars evaluate d. This clustering of citrus cultivars based on blossom volatile co mposition was found in good agreement with recent phylogenic studies ( 98 ). Therefore, it appears that it may be possible to use blossom volatile composition as a tool for classification in addi tion to other taxonomical/ phylogenetic methods.

PAGE 104

104 The biological significance of bl ossom volatile production could be explained specifically for pummelo blossoms. Survival of this citrus cultiv ar requires cross-pollination whereas the others in this study are self pollinati ng. Therefore, the pummeloÂ’s produc tion of the highest levels of total blossom volatiles as well as highest levels of honeybee attracting volatiles such as linalool and 1-hexanol serves to help perpetuate the species.

PAGE 105

105 Table 5-1. Citrus blossom cultivars included in this study. Citrus species Cultivar Common Name 01 C. sinensis (L.) Osbeck Navel sweet orange 02 C. sinensis (L.) Osbeck Valencia sweet orange 03 C. reticulata Blanco Dancy mandarin 04 C. reticulata Blanco Kimbrough Satsuma mandarin 05 C. reticulata Blanco Willowleaf mandarin 06 C. paradisi Macfad. Duncan grapefruit 07 C. aurantium L. Bittersweet sour orange 08 C. aurantium L. Bouquet de Fleurs sour orange 09 C. aurantium L. Common sour orange 10 C. grandis (L.) Osbeck Pummelo pummelo 11 C. limon (L.) Burm.f. Lisbon lemon 12 C. limon (L.) Burm.f. Ponderosa lemon 13 C. limon (L.) Burm.f. Volkamer lemon 14 C. aurantifolia (Christm.) Swingle Thornless Key Lime key lime 15 C. hystrix L. Kaffir lime

PAGE 106

106 Table 5-2. Identifications of citrus blossom volatiles and their re lative total ion current (TIC) pe ak area from eight citrus t ypes. Listed values are the percentage of the total peak area for each cultivar. Totals for each cultivar are listed at the bottom of the table. Major volatile s (>5%) are bolded. # Name CAS # Wax LRI Pummelo Sweet orange Grapefruit Sour orange Kaffir lime Volkamer lemon Lemonlime Mandarin 01 acetone 37-64-1 831 0.05 0.08 0. 04 0.46 0.25 0. 03 0.10 0.02 02 methyl acetate 79-20-9 841 0.26 0.03 03 2-methylfuran 534-22-5 910 0.01 0.15 0.05 0.12 04 isopropanol a 67-63-0 940 0.02 0.00 0.04 1.69 0.02 0.03 05 ethanol 64-17-5 949 0.05 0.05 0.03 0.45 0.65 0.00 0.03 0.05 06 -pinene a,b,c 80-56-8 1036 0.07 0.06 0.03 0.06 0.33 0.04 0.13 0.06 07 -thujene b 2867-05-2 1039 0.05 0.10 0.03 0.00 0.89 0.05 0.04 0.02 08 hexanal 66-25-1 1100 0.00 0.04 0.06 0.10 0.15 0.00 0.03 09 -pinene a,b,c 127-91-3 1125 1.64 0.20 0.07 0.51 0.78 0.14 1.00 0.09 10 sabinene a,b,c 3387-41-5 1137 0.03 2.10 1.05 0.23 18.11 1.12 0.42 0.05 11 -3carene a 13466-789 1166 0.00 0.34 0.13 12 -myrcene a,b,c 123-25-3 1176 0.73 24.00 21.36 8.23 2.57 43.41 2.06 45.02 13 -myrcene 1686-30-2 1183 5.02 12.72 17.84 15.88 1.42 10.47 0.67 5.34 14 -terpinene a,b,c 99-86-5 1199 0.79 0.28 1.67 0.44 0.30 15 limonene a,b,c 138-86-3 1219 2.09 4.47 2.87 2.11 0.59 11.41 56.57 0.87 16 unknown terpene 1222 9.39 2.80 3.20 4.94 1.00 0.78 17 1,8-cineole a,c 470-82-6 1232 1.15 1.85 0.00 18 (Z)-ocimene a,b 3338-55-4 1248 1.09 0.76 0.52 0.68 0.52 0.59 1.39 0.54 19 (E)-ocimene a,b,c 3779-61-1 1267 2.63 8.20 3.44 1.89 6.52 13.73 11.66 17.78 20 -terpinene a,b,c 99-85-4 1285 0.33 0.19 21 p-cymene a,b,c 99-87-6 1299 13.62 1.68 1.92 0.67 4.10 0.48 1.27 0.57 22 -terpinolene a,b,c 586-62-9 1307 0.15 1.86 0.36 0.69 0.28 23 unknown 1322 1.71 0.08 0.95 0.13 24 6-methyl-5hepten-2-one a,b 110-93-0 1361 0.20 0.27 0.14 1.81 0.06 0.04 0.30 0.05

PAGE 107

107 Table 5-2. Continued. # Name CAS # Wax LRI Pummelo Sweet orange Grapefruit Sour orange Kaffir lime Volkamer lemon Lemonlime Mandarin 25 1-hexanol 111-27-3 1364 1.07 0.01 0.09 0.45 26 neo-allo-ocimene 7216-56-0 1396 1.56 1.35 1.91 0.83 1.42 1.74 1.93 1.23 27 allo-ocimene b 673-84-7 1421 2.93 2.66 3.40 1.66 2.69 3.44 3.70 2.70 28 2,6-dimethyl-1,3,5,7octatetraene 90973-78-7 1459 0.13 0.10 0.06 0.10 0.13 0.10 0.15 29 -p-dimethylstyrene 1195-32-0 1471 1.01 0.62 0.87 0.31 0.55 0.38 0.39 0.43 30 carvyl acetate 1134-95-8 1475 0.90 0.96 0.90 1.00 1.07 0.83 0.79 0.83 31 sabinene hydrate a,b 546-79-2 1485 0.03 0.01 0.40 32 (Z)-limonene oxide 13837-75-7 1487 0.44 0.06 33 -elemene b 20307-84-0 1498 0.08 0.29 34 pentadecane 629-62-9 1501 0.07 0.29 35 citronellal a 106-23-0 1505 0.02 6.05 4 0.18 36 (Z)-epoxy-ocimene 1512 0.02 37 -copaene a,b 3856-25-5 1528 0.00 0.71 0.00 0.03 38 linalool a,b,c 78-70-6 1561 45.35 21.84 28.66 45.09 4.86 0.41 3.84 7.31 39 benzaldehyde 100-52-7 1573 0.04 0.04 0.08 0.01 0.10 40 linalyl acetate 115-95-7 1577 3.64 0.00 41 -bergamotene 17699-05-7 1615 0.21 0.08 0.64 42 thymol methyl etherc 1076-56-8 1621 0.20 0.00 0.10 43 -elemene a,b 515-13-9 1624 0.09 0.12 0.14 0.31 0.07 44 -caryophyllene a,b,c 87-44-5 1642 0.08 0.33 0.15 0.14 3.75 0.23 1.52 0.20 45 -farnesene 18794-84-8 1681 0.02 0.19 0.08 0.03 1.71 0.08 0.29 0.12 46 phenylacetaldehydec 122-78-1 1691 0.07 0.30 0.32 47 heptadecane 629-78-7 1699 0.06 0.13 48 -heptadecene 16369-12-3 1725 0.57 0.22 0.90 49 -bisabolene 495-61-4 1754 0.04 0.03 0.95 0.27 0.64 0.10 50 geranialc 141-27-5 1766 0.05 0.06 0.04 0.48 0.02 51 -citronellol 6812-78-8 1772 0.76 0.23 0.06

PAGE 108

108 Table 5-2. Continued. # Name CAS # Wax LRI Pummelo Sweet orange Grapefruit Sour orange Kaffir lime Volkamer lemon Lemonlime Mandarin 52 -citronellol a,c 106-22-9 1777 0.03 0.14 25.29 0.34 1.82 0.05 53 methylphenylacetate 140-39-6 1799 0.03 0.03 55 nerol a,c 106-25-2 1818 0.03 0.05 0.06 0.17 0.02 56 methyl salicylate 119-36-8 1832 0.02 0.03 57 geraniol a,c 106-24-1 1861 0.06 0.19 0.47 0.04 0.08 0.14 0.05 58 calamenene 483-77-2 1879 0.71 0.21 0.28 0.04 59 phenethyl alcohol b 60-12-8 1950 0.17 0.24 0.25 0.23 0.08 0.40 60 benzeneacetonitrile 140-29-4 1982 0.81 0.58 0.52 0.15 2.68 0.05 2.66 61 (Z)-jasmone a,b,c 488-10-8 1996 0.03 0.13 0.07 62 nerolidolc 142-50-7 2054 0.17 0.94 0.22 0.20 0.16 0.13 0.08 63 dimethyl anthranilate 85-91-6 2142 0.78 64 unknown 2181 0.18 0.11 0.18 0.16 65 unknown 2183 0.06 0.18 0.22 0.22 66 thymolc 89-83-8 2208 0.02 0.08 67 n-phenylformamide 103-70-8 2232 0.40 0.42 0.32 0.11 0.10 0.35 68 methyl anthranilateb,c 134-20-3 2299 6.17 5.03 5.18 4.81 2.81 1.53 1.62 3.50 69 farnesolc 4602-84-0 2375 0.05 0.07 0.04 70 Indole b,c 120-72-9 2507 1.41 3.98 2.42 2.61 1.19 2.00 1.19 4.50 Total Peak area (all volatiles) 1.62E+09 1.16E+09 1.23E+09 8.63E+08 3.70E+08 7.03E+08 9.71E+08 9.87E+08 a compounds reported in C unshiu blossoms ( 93 ) b compounds reported in C.deliciosa blossoms ( 91 ) c compounds reported in C. sinensis C. paradisi and C. reticulata blossoms ( 85 )

PAGE 109

109 Figure 5-1. Chemical composition of citrus bl ossom headspace volatiles. Values totaled over number of peaks (put in pare nthesis) in each chemical cla ss. All blossoms types were normalized to total peak area of pummelo blossoms.

PAGE 110

Figure 5 2. Eigenve c oranges ( ); Pumm e c tor values o ); Mandar i e lo (); Vol k o f PC 1vs P C i ns (); Le m k amer lemo n 110 C 2 from P C m ons and li m n (); Kaff i C A of 70 vol a m es (); Gr a i r lime (). a tile compo n a pefruit (); n ents. Swee t Sour orang e t e s (

PAGE 111

111 Figure 5-3. Load plot. Eigenvector values of PC_01vs PC_02 from PCA of 70 volatile components

PAGE 112

112 Table 5-3. Average values with standard deviations for the 12 blossom volatiles found to be most differentiating using multiple analyses of variance. Indivi dual values are group means + S.E. Values fo llowed by different lett ers are significantly different (TukeyÂ’s HSD at = 0.05) # Pummelo Sweet orange Grapefruit Sour orange L emon & L ime Mandarin Vol lem Kaff lim 01 -pinene 2.7E+07 9.3E+06 a 2.6E+06 1.2E+06 b 8.2E+05 5.2E+04 b 4.1E+06 1.1E+06 b 9.9E+06 3.6E+06 b 8.2E+05 6.6E+04 b 9.7E+05 1.4E+05 b 2.9E+06 3.9E+05 b 02 sabinene 5.1E+05 5.1E+04 d 2.4E+07 6.0E+06 b 1.3E+07 1.7E+06 b 1.9E+06 4.9E+05 d 4.0E+06 1.2E+06 c, d 4.4E+05 6.6E+04 d 7.7E+06 5.4E+05 c, d 6.6E+07 2.3E+06 a 03 -myrcene 1.2E+07 2.8E+05 c 2.8E+08 2.6E+07 b 2.6E+08 1.4E+07 b 7.3E+07 1.4E+07 c 2.0E+07 2.3E+06 c 4.3E+08 3.7E+07 a 3.1E+08 7.3E+07 a, b 9.4E+06 4.2E+05 c 04 myrcene 8.2E+07 2.5E+06 a, b, c 1.6E+08 7.3E+07 a 2.2E+08 5.5E+06 a 1.4E+08 1.8E+07 a, b 6.7E+06 1.3E+06 c 5.0E+07 5.7E+06 b, c 7.2E+07 3.1E+05 a, b, c 5.2E+06 1.1E+05 c 05 limonene 3.4E+07 4.5E+06 b 4.8E+07 1.3E+07 b 3.5E+07 6.5E+06 a 1.7E+07 7.4E+06 b 5.3E+08 2.8E+07 a 9.2E+06 4.4E+06 b 7.8E+07 6.2E+05 b 2.2E+06 2.1E+05 b 06 trans-ocimene 4.3E+07 9.7E+06 b 8.8E+07 2.4E+07 a, b, c 4.3E+07 5.9E+06 a 1.6E+07 2.7E+06 c 1.2E+08 2.7E+07 a, b 1.8E+08 2.9E+07 a 9.8E+07 2.2E+07 a, b, c 2.4E+07 5.2E+05 b 07 -terpinolene 1.5E+06 9.0E+05 b 6.7E+06 1.0E+06 a 2.7E+06 2.9E+05 b 2.4E+06 2.9E+05 b 6.8E+06 6.7E+05 a 08 p-cymene 2.8E+07 1.0E+07 a 1.1E+06 6.6E+05 b 1.2E+07 2.4E+06 b 1.1E+06 2.2E+05 b 09 1-hexanol 1.7E+07 7.4E+05 a 1.2E+05 7.0E+04 c 1.1E+06 1.9E+05 b 1.6E+06 1.8E+05 b 10 linalool 7.4E+08 2.1E+07 a 2.5E+08 9.1E+06 b 3.5E+08 1.9E+07 b 3.9E+08 3.5E+07 b 4.4E+07 2.7E+07 c 8.3E+07 4.0E+07 c 3.0E+06 8.5E+05 c 1.7E+07 4.4E+06 c

PAGE 113

113 Table 5-3. Continued. # Pummelo Sweet orange Grapefruit Sour orange L emon & L ime Mandarin Vol lem Kaff lim 11 benzeneacetonitrile 8.9E+06 2.4E+06 b, c 7.2E+06 6.6E+05 b, c 4.6E+06 7.3E+05 b, c 4.7E+05 8.6E+04 c 2.6E+07 3.9E+06 a 1.9E+07 7.4E+06 a, b 5.8E+05 1.3E+05 b, c 12 (Z)jasmone 2.9E+05 1.7E+05 b 1.1E+06 1.8E+05 a 6.7E+05 1.4E+05 a

PAGE 114

114 CHAPTER 6 SUMMARY A highly sensitive sulfur specific analysis employing three internal standards was developed to quantify volatile sulfur compounds (VSCÂ’s) f ound in fresh, pasteurized and concentrated grapefruit juices (GFJÂ’s). This is the first comprehensive method to determine GFJ sulfur volatiles and 13 VSCÂ’s were positively id entified. Solid phase microextraction (SPME) concentrated VSCÂ’s with no apparent artifact s and sulfur specific pulsed flame photometric detector (PFPD) was employed to selectively de tect VSCÂ’s at ng/Kg levels. SPME parameters (fiber type, headspace atmosphere, extraction time and temperature) were optimized to increase sensitivity. This study established the altera tions caused by headspace oxygen on the amounts of hydrogen sulfide, dimethyl sulfide and 1-p-ment hene-8-thiol (PMT) measured. The complete absence of PMT in 100% oxygen headspace was associ ated with a simultaneous increase in its bicyclic form: 2, 8-epithio-cis-p-menthane (E PM). An inert headspace SPME method was therefore employed. Using polar, non polar and porous layer open tubular columns coupled with PFPD the following VSCs were identified: hydroge n sulfide, sulfur dioxide, methanethiol, dimethyl sulfide, carbon disulfide, 2methyl thiophene, 3methyl thiophene, dimethyl disulfide, methional, dimethyl trisulfide, 3-mercaptohexy l acetate, PMT and EPM. These structurally diverse VSCÂ’s were subsequently quantified using three structurally different internal standards (ethyl methyl sulfide, isopropyl disulf ide and 8-mercapto-p-menthan-3-one). VSCÂ’s were quantified in 10 fresh hand-squeez ed (FHS), 11 not from concentrate (NFC) and 9 reconstituted from concentrate (RFC) ju ice samples using the above developed method. Multivariate statistics using principal component analysis and discriminant function analysis revealed unique volatile sulfur patt erns which segregated these juices into three distinct clusters. H2S (loading of 0.75 on PC1), DMS (loading value -0.61 on PC1) were the most differentiating

PAGE 115

115 volatiles accounting for 78% of tota l variability in data set. Ov erall, unheated GFJ comprised very few sulfur volatiles (total of 3 g/L), predominantly H2S (1.8 g/L) that maybe typically released during analysis at 40C.In contrast, RFC juice contained more than twice the quantity of sulfur volatiles (7.8g/ L), DMS being the prin cipal volatile (4.8 g/L). The aroma composition of these juices was characterized using GC-O analysis. Fresh GFJ aroma contained green, fresh grapefruity, fruity and floral notes. On the other hand, cooked, sulfury, garlic odors characterized aroma of heated juices. Thermally processed juices cont ained higher levels of DMDS, DMTS, 2MTP, 3MTP, methional, EPM a nd PMT (which was observed only in canned RFC juice). This suggests that several VSCs, including the character imparting PMT, were generated either during thermal processing or during subsequent prolonged storage. This hypothesis was further confirmed by the thermally accelerated storage studies on fresh GFJ. Model GFJ studies (pH 3.4) containing sulf ur amino acids revealed that thermal breakdown volatile sulfur products observed in grapefruit juices were a direct consequence of breakdown of S-methyl methionine and methi onine. Four of the six aroma active sulfur compounds were found to result from heating Smethyl methionine (producing DMS) and methionine (producing Met, MeSH and DMTS). This indicates th at Smethyl methionine and methionine are the major nonvolatile precurso rs in GFJ, significantly impacting its aroma quality. Heating of cysteine and gl utathione in model GFJ system ma inly resulted in formation of H2S, which lacked aroma activity in GFJ. This suggests that cysteine and glutathione may not immediate contributors to grapefruit aroma. Volatile distribution between serum and pul p fractions of ultracentrifuged canned RFC was examined. PulpÂ’s volatile composition was predominated by the presence of monoterpenes (primarily D-limonene). Sesquiterpenes (incl uding nootkatone) as well as PMT (structurally

PAGE 116

116 similar to Dlimonene) and 3MHA were found be more strongly associated with the pulp portion of GFJ. Nootkatone, PMT and 3MHA are im portant contributors to the characteristic aroma of grapefruit juice. On the other hand, oxygenated terpenes; oxygenated hydrocarbons as well as the high vapor pressure VSC’s such as H2S, MeSH, DMDS, 2MTP and 3MTP were found mainly distributed in the serum. The final study on citrus blossoms determined the volatile pattern of intact grapefruit blossom in comparison with blosso ms from other citrus species. Components reported to be honey bee attractants were found in highest conc entrations in those cultivars which require external pollination. Headspace-SPME coupled with GC-MS was employed to concentrate and identify the volatile composition of in tact blossoms. Linalool, -myrcene, -myrcene, limonene, (E)-ocimene, methyl anthranilate and indole were the major volatiles produced by citrus blossoms. Principal component analysis revealed a diverse volatile di stribution between major citrus cultivars. The distributi on of citrus cultivar eigenvector scores revealed three widely separated clusters consisting of pummelo, mandarins and lemonslimes. Interestingly, grapefruit blossom’s (hybrid of sweet orange and pummel o) volatile composition was found more closely associated to sweet orange than pummelo. This chemotaxonomic classification of citrus was found in good agreement with recent phyloge netic studies which designated pummelo, mandarins and citron as ‘true’ citr us species and all others as cu ltivated biotypes. This study has established that use of blossom volatiles as molecular markers can enable more robust identifications of citrus species than th e traditional geographical and morphological classifications.

PAGE 117

117 REFERENCES 1. USDA, Citrus Summary 2007-2008. In Department, F. A. S. S., Ed. USDA, NASS, Florida field office: 2009. 2. Shaw, P. E.; Buslig, B. S.; Moshonas, M. G. Classification of co mmercial orange juice types by pattern-recognition involving volatile constituents quantified by gaschromatography. J. Agric. Food Chem. 1993, 41, (5), 809-813. 3. Berlinet, C.; Guichard, E.; Fournier, N.; Ducruet, V. Effect of pulp reduction and pasteurization on the releas e of aroma compounds in industrial orange juice. J. Food Sci. 2007, 72, (8), S535-S543. 4. Kumazawa, K.; Wada, Y.; Masuda, H. Ch aracterization of potent odorants in handsqueezed and heat processed citrus juices. J. Jpn. Soc. Food Sci. Technol.-Nippon Shokuhin Kagaku Kogaku Kaishi 2007, 54, (6), 266-273. 5. Bylaite, E.; Meyer, A. S. Characterisation of volatile aroma compounds of orange juices by three dynamic and static headsp ace gas chromatography techniques. Eur. Food Res. Technol. 2006, 222, (1-2), 176-184. 6. Sawamura, M.; Shimoda, M.; Osajima, Y. Studies on off-flavor formed during heating process of Satsuma mandarin Juice .3. J. Agr. Chem. Soc. Jpn. 1978, 52, (7), 281-287. 7. Schreier, P.; Drawert, F.; Junker, A.; Mic k, W. Quantitative composition of natural and technologically changed aromas of plants .2. Aroma compounds in oranges and their changes during juice processing. Z. Lebensm. Unters. Forsch. 1977, 164, (3), 188-193. 8. Perez-Cacho, P. R.; Mahattanatawee, K.; Sm oot, J. M.; Rouseff, R. Identification of sulfur volatiles in canned orange juices lacking orange flavor. J. Agric. Food Chem. 2007, 55, (14), 5761-5767. 9. Nisperoscarriedo, M. O.; Shaw, P. E. Compar ison of volatile flavor components in fresh and processed orange juices. J. Agric. Food Chem. 1990, 38, (4), 1048-1052. 10. Lin, J. M.; Rouseff, R. L.; Barros, S.; Naim, M. Aroma composition changes in early season grapefruit juice produced from thermal concentration. J. Agric. Food Chem. 2002, 50, (4), 813-819. 11. Shaw, P. E.; Wilson, C. W. Volatile sulfides in headspace gases of fresh and processed citrus juices. J. Agric. Food Chem. 1982, 30, (4), 685-688. 12. Surburg, H.; Guentert, M.; Harder, H., Vola tile compounds from flowers: analytical and olfactory aspects In Bioactive volatile co mpounds from plants San Francisco, California, 1993; pp 168-186.

PAGE 118

118 13. Toyoda, T.; Nohara, I.; Sato, T. Headspace analysis of volatile compounds emitted from various citrus blossoms. ACS Symp. Ser. 1993, 525, (Bioactive Volatile Compounds from Plants), 205-19. 14. Boussaada, O.; Skoula, M.; Kokkalou, E.; Chem li, R. Chemical variability of flowers, leaves, and peels oils of f our sour orange provenances. J. Essent. Oil Bear. Pl 2007, 10, (6), 453-464. 15. Merle, H.; Moron, M.; Blazquez, M. A.; Boir a, H. Taxonomical cont ribution of essential oils in mandarins cultivars. Biochem. Syst. Ecol. 2004, 32, (5), 491-497. 16. Fellers, P. J.; Carter, R. D.; Dejager, G. Infl uence of the Ratio of Degrees Brix to Percent Acid on Consumer Acceptance of Pro cessed Modified Grapefruit Juice. J. Food Sci. 1988, 53, (2), 513-515. 17. Fellers, P. J. The relationship between the ratio of degrees brix to percent acid and sensory flavor in grapefruit juice. Food Technol. 1991, 45, (7), 68. 18. Fellers, P. J. A Review of limonin in grapefru it (Citrus-paradisi) juic e, its relationship to flavor, and efforts to reduce It. J. Sci. Food Agric... 1989, 49, (4), 389-404. 19. Kirchner, J. G.; Miller, J. M.; Rice, R. G.; Keller, G. J.; Fox, M. M. Volatile watersoluble constituents of grapefruit juice. J. Agric. Food Chem. 1953, 1, (7), 510-512. 20. Kirchner, J. G.; Miller, J. M. Volatile Oil Constituents of Grapefruit Juice. J. Agric. Food Chem. 1953, 1, (7), 512-518. 21. Beisel, C. G.; Johnson, V. E.; Kustel, P. A. J. Food Sci. 1964, 29, (565). 22. Erdtman, H.; Hirose, Y. Chemistry of na tural order cupressales .46. Structure of nootkatone. Acta Chem. Scand. 1962, 16, (6), 1311. 23. Macleod, W. D.; Buigues, N. M. Sesquiterp enes .1. Nootkatone new grapefruit flavor constituent. J. Food Sci. 1964, 29, (5), 565-568. 24. Stevens, K. L.; Guadagni, D. G.; Stern, D. J. Odour character and threshold values of nootkatone and related compounds. J. Sci. Food Agric. 1970, 21, (11), 590-593. 25. Shaw, P. E.; Wilson, C. W. Importance of noot katone to the aroma of grapefruit oil and the flavor of grapefruit juice. J. Agric. Food Chem. 1981, 29, (3), 677-679. 26. Demole, E.; Enggist, P. Further investiga tion of grapefruit ju ice flavor components (citrus paradisi Macfayden). Valencaneand Eudesmanetype sesquiterpene ketones. Helv. Chim. Acta 1983, 66, (5), 13811391. 27. Demole, E.; Enggist, P.; Ohloff, G. 1-p-Me nthene-8-thiol: a powerful flavor impact constituent of grapefruit jui ce (Citrus paradi si Macfayden). Helv. Chim. Acta 1982, 65, (6), 1785-94.

PAGE 119

119 28. Buettner, A.; Schieberle, P. Characterization of the most odor-active volatiles in fresh, hand-squeezed juice of grapefruit (Citrus paradisi Macfayden). J. Agric. Food Chem. 1999, 47, (12), 5189-5193. 29. Buettner, A.; Schieberle, P. Evaluation of key aroma compounds in hand-squeezed grapefruit juice (Citrus paradisi Macfayden) by quantitation and flavor reconstitution experiments. J. Agric. Food Chem. 2001, 49, (3), 1358-1363. 30. Darriet, P.; Tominaga, T.; Lavigne, V.; Boidr on, J.-N.; Dubourdieu, D. Identification of a powerful aromatic component of Vitis vinife ra L. var. Sauvignon wines: 4-mercapto-4methylpentan-2-one. Flav. Fragr. J. 1995, 10, (6), 385-92. 31. Kirchner, J. G.; Rice, R. G.; Miller, J. M.; Keller, G. J. The presence of hydrogen sulfide in citrus juices. Arch. Biochem. 1950, 25, (1), 231-232. 32. Shaw, P. E.; Ammons, J.; Braman, R. Volati le sulfur compounds in fresh orange and grapefruit juices: id entification, qu antitation, and possible impo rtance to juice flavor. J. Agric. Food Chem. 1980, 28, 778-781. 33. Lin, J.; Jella, P.; Rouseff, R. L., Gas chromatography-olfactometry and chemiluminescence characterization of grapef ruit juice volatile sulfur compounds. In Heteroatomic Aroma Compounds 2002; Vol. 826, pp 102-112. 34. Rouseff, R., Analytical methods to dete rmine volatile sulfur compounds in foods and beverages. In Heteroatomic Aroma Compounds Reineccius, G. A. R. T. A., Ed. ACS symposium series: 2002; Vol. 826, pp 2-24. 35. Werkhoff, P.; Guntert, M.; Krammer, G.; Sommer, H.; Kaulen, J. Vacuum headspace method in aroma research: Flavor ch emistry of yellow passion fruits. J. Agric. Food Chem. 1998, 46, (3), 1076-1093. 36. Bouchilloux, P.; Darriet, P.; Dubourdieu, D. Quantitative determination of 4-mercapto-4methylpentan-2-one in Sauvignon wines. J. Int. des Sci de la Vigne et du Vin 1996, 30, (1), 23-29. 37. Tominaga, T.; Murat, M. L.; Dubourdieu, D. Development of a method for analyzing the volatile thiols Involved in the charact eristic aroma of wines made from Vitis vinifera L. Cv. Sauvignon Blanc. J. Agric. Food Chem. 1998, 46, (3), 1044-1048. 38. Guth, H. Quantitation and sensory studies of character impact odorants of different white wine varieties. J. Agric. Food Chem. 1997, 45, (8), 3027-3032. 39. Mateo-Vivaracho, L.; Ferreira, V.; Cacho, J. Automated analysis of 2-methyl-3furanthiol and 3-mercaptohexyl acetate at ng L-1 level by headspace solid-phase microextracion with on-fibre derivatisation and gas chromatography-negative chemical ionization mass spectrometric determination. J. Chromatogr. A 2006, 1121, (1), 1-9.

PAGE 120

120 40. Arthur, C. L.; Pawliszyn, J. Solid-phase microextraction with th ermal-desorption using fused-silica optical fibers. Anal. Chem. 1990, 62, (19), 2145-2148. 41. Hill, P. G.; Smith, R. M. Determination of sulphur compounds in beer using headspace solid-phase microextraction and gas chro matographic analysis with pulsed flame photometric detection. J. Chromatogr. A 2000, 872, (1-2), 203-213. 42. Fan, X. T.; Sommers, C. H.; Thayer, D. W.; Lehotay, S. J. Volatile sulfur compounds in irradiated precooked turkey breast analyzed with pulsed flame photometric detection. J. Agric. Food Chem. 2002, 50, (15), 4257-4261. 43. Fang, Y.; Qian, M. C. Sensitive quantification of sulfur compounds in wine by headspace solid-phase microextraction technique. J. Chromatogr. A 2005, 1080, (2), 177-185. 44. Vazquez-Landaverde, P. A.; Torres, J. A.; Qian, M. C. Quantification of trace volatile sulfur compounds in milk by solid-phase micr oextraction and gas chromatography-pulsed flame photometric detection. J. Dairy Sci. 2006, 89, (8), 2919-2927. 45. Murray, R. A. Limitations to the use of so lid-phase microextraction for quantitation of mixtures of volatile or ganic sulfur compounds. Anal. Chem. 2001, 73, (7), 1646-1649. 46. Mestres, M.; Busto, O.; Gausch, J. Headsp ace solid-phase microextraction analysis of volatile sulphides and disu lphides in wine aroma. J. Chromatogr. A 1998, 808, (1-2), 211-218. 47. Haberhauer-Troyer, C.; Rosenberg, E.; Gr asserbauer, M. Evaluation of solid-phase microextraction for sampling of volatile organi c sulfur compounds in air for subsequent gas chromatographic analysis wi th atomic emission detection. J. Chromatogr. A 1999, 848, (1-2), 305-315. 48. Lopez, R.; Lapena, A. C.; Cacho, J.; Ferre ira, V. Quantitative determination of wine highly volatile sulfur compounds by using automated headspace solid-phase microextraction and gas chromatography-pulse d flame photometric detection Critical study and optimization of a new procedure. J. Chromatogr. A 2007, 1143, (1-2), 8-15. 49. Cadwallader, K. R.; Xu, Y. Analysis of vol atile components in fresh grapefruit juice by purge-and-trap gas-chromatography. J. Agric. Food Chem. 1994, 42, (3), 782-784. 50. Jella, P.; Rouseff, R.; Goodner, K.; Widmer, W. Determination of key flavor components in methylene chloride extracts from processed grapefruit juice. J. Agric. Food Chem. 1998, 46, (1), 242-247. 51. Schneider, R.; Kotseridis, Y.; Ray, J. L.; Augier, C.; Baumes, R. Quantitative determination of sulfur-containing wine odor ants at sub parts per billion levels. 2. Development and application of a stable isotope dilution assay. J. Agric. Food Chem. 2003, 51, (11), 3243-3248.

PAGE 121

121 52. Dreher, J. G.; Rouseff, R. L.; Naim, M. GC-olfactometric characterization of aroma volatiles from the thermal degradation of thiamin in model orange juice. J. Agric. Food Chem. 2003, 51, (10), 3097-3102. 53. Starkenmann, C.; Niclass, Y.; Escher, S. Vo latile organic sulfur-cont aining constituents in Poncirus trifoliata (L.) raf. (Rutaceae). J. Agric. Food Chem. 2007, 55, (11), 45114517. 54. Miller, J. M.; Rockland, L. B. Determination of cysteine and glutathi one in citrus juices by filter paper chromatography. Arch. Biochem. Biophys. 1952, 40, 416-23. 55. Saetre, R.; Rabenstein, D. L. Determination of cysteine and glutat hione in fruit by highperformance liquid chromatography. J. Agric. Food Chem. 1978, 26, (4), 982-983. 56. Clements, R. L.; Leland, H. V. Ion-Exchange Study of Free Amino Acids in Juices of 6 Varieties of Citrus. J. Food Sci. 1962, 27, (1), 20-25. 57. Ueda, Y.; Imahori, Y.; Yamanaka, H.; Kwak, S.; Chachin, K. Dimethyl sulfide production in stored satsuma mandarin: The pot ential role of ferrous ion and ascorbate. Acta. Hortic 1995 (398), 297-304. 58. Sakamoto, K.; Inoue, A.; Nakatani, M.; Kozuka, H.; Ohta, H.; Osajima, Y. Smethylmethionine sulfonium in fruits of citrus hybrids. Biosci. Biotech. Bioch. 1996, 60, (9), 1486-1487. 59. Belitz, H. D.; Grosch, W.; Schieberle, P. Food Chemistry 3rd ed.; Springer: 2004; p 365366. 60. Zhang, Y. G.; Ho, C. T. Comparison of the volatile compounds formed from the thermalreaction of glucose with cysteine and glutathione. J. Agric. Food Chem. 1991, 39, (4), 760-763. 61. Tai, C. Y.; Ho, C. T. Influence of cyst eine oxidation on thermal formation of maillard aromas. J. Agric. Food Chem. 1997, 45, (9), 3586-3589. 62. Tai, C. Y.; Ho, C. T. Influence of glutat hione oxidation and pH on thermal formation of Maillard-type vol atile compounds. J. Agric. Food Chem. 1998, 46, (6), 2260-2265. 63. Tominaga, T.; des Gachons, C. P.; Dubourdie u, D. A new type of flavor precursors in Vitis vinifera L cv Sauvignon bl anc: S-cysteine conjugates. J. Agric. Food Chem. 1998, 46, (12), 5215-5219. 64. des Gachons, C. P.; Tominaga, T.; Dubourdie u, D. Measuring the ar omatic potential of Vitis vinifera L. Cv. Sauvignon blanc grap es by assaying S-cysteine conjugates, precursors of the volatile thiols resp onsible for their varietal aroma. J. Agric. Food Chem. 2000, 48, (8), 3387-3391.

PAGE 122

122 65. Des Gachons, C. P.; Tominaga, T.; Dubourdieu, D. Localization of S-cysteine conjugates in the berry: Effect of skin contact on ar omatic potential of Vitis vinifera L. cv. Sauvignon blanc must. Am. J. Enol. Vitic. 2002, 53, (2), 144-146. 66. Chone, X.; Lavigne-Cruege, V.; Tominaga, T.; Van Leeuwen, C.; Castagnede, C.; Saucier, C.; Dubourdieu, D. Effect of vine nitrogen status on grape aromatic potential: Flavor precursors (S-cystein e conjugates), glutathione a nd phenolic content in Vitis vinifera L. cv. Sauvignon blanc grape juice. J. Int. des Sci de la Vigne et du Vin 2006, 40, (1), 1-6. 67. Dubourdieu, D.; Torninaga, T.; Masneuf, I.; Des Gachons, C. P.; Murat, M. L. The role of yeasts in grape flavor development during fermentation: The example of Sauvignon blanc. Am. J. Enol. Vitic. 2006, 57, (1), 81-88. 68. Starkenmann, C.; Troccaz, M.; Howell, K. The role of cysteine and cysteine-S conjugates as odour precursors in the fl avour and fragrance industry. Flav. Fragr. J. 2008, 23, (6), 369-381. 69. Darriet, P.; Tominaga, T.; Demole, E.; Dubour dieu, D. Evidence of the presence of a 4mercapto-4-methylpentan-2-one precursor in Vitis-Vinifera Sauvignon Blanc grape variety. Comptes Rendus Acad. Sci. Ser. III-Sci. Vie-Life Sci. 1993, 316, (11), 1332-1335. 70. Tominaga, T.; Dubourdieu, D. Identification of cysteinylated aroma precursors of certain volatile thiols in passion fruit juice. J. Agric. Food Chem. 2000, 48, (7), 2874-2876. 71. Rega, B.; Fournier, N.; Nicklaus, S.; Guicha rd, E. Role of pulp in flavor release and sensory perception in orange juice. J. Agric. Food Chem. 2004, 52, (13), 4204-4212. 72. Radford, T.; Kawashim.K; Friedel, P. K.; Pope, L. E.; Gianturc.Ma. Distribution of volatile compounds between pulp and serum of some fruit juices. J. Agric. Food Chem. 1974, 22, (6), 1066-1070. 73. Hernandez, E.; Chen, C. S.; Shaw, P. E.; Carter R. D.; Barros, S. Ultr afiltration of orange juice effect on soluble solids, suspended-solids, and aroma. J. Agric. Food Chem. 1992, 40, (6), 986-988. 74. Brat, P.; Rega, B.; Alter, P.; Reynes, M.; Brillouett, J. M. Distribution of volatile compounds in the pulp, cloud, and serum of freshly squeezed orange juice. J. Agric. Food Chem. 2003, 51, (11), 3442-3447. 75. Moshonas, M. G.; Shaw, P. E. Analysis of volatile flavor constituents from grapefruit essence. J. Agric. Food Chem. 1971, 19, (1), 119-120. 76. Moshonas, M. G.; Shaw, P. E. Quantities of Volatile Flavor Components in Aqueous Orange Essence and in Fresh Orange Juice. Food Technol. 1986, 40, (11), 100-103. 77. Moshonas, M. G.; Shaw, P. E. Quantitative-An alysis of Orange Juice Flavor Volatiles by Direct-Injection Gas-Chromatography. J. Agric. Food Chem. 1987, 35, (1), 161-165.

PAGE 123

123 78. Moshonas, M. G.; Shaw, P. E. Quantitation of volatile constituents in mandarin juices and its use for comparison with orange juices by multivariate analysis. J. Agric. Food Chem. 1997, 45, (10), 3968-3972. 79. Moshonas, M. G.; Shaw, P. E. Flavor and chemical comparison of pasteurized and fresh Valencia orange juices. J. Food Qual. 1997, 20, (1), 31-40. 80. Moshonas, M. G.; Shaw, P. E. Changes in volatile flavor constituents in pasteurized orange juice during storage. J. Food Qual. 2000, 23, (1), 61-71. 81. Nunez, A. J.; Maarse, H.; Bemelmans, J. M. H. Volatile flavor components of grapefruit juice (Citrus paradisi Macfadyen). J. Sci. Food Agric. 1985, 36, (8), 757-63. 82. Perez-Cacho, P. R.; Rouseff, R. Processi ng and Storage Effects on Orange Juice Aroma: A Review. J. Agric. Food Chem. 2008, 56, (21), 9785-9796. 83. Perez-Cacho, P. R.; Rouseff, R. L. Fr esh squeezed orange juice odor: A review. Critical Reviews in Food Science and Nutrition 2008, 48, (7), 681-695. 84. Boelens, M. H.; Gemert, L. J. v. Charact erization and sensory properties of volatile nitrogen compounds from natural isolates. Perfum. Flavor 1994, 19, 51-65. 85. Attaway, J. A.; Pieringer, A. P.; Barabas, L. J. Origin of citrus flavor components. II. Identification of volatile com ponents from citrus blossoms. Phytochemistry 1966, 5, (6), 1273-9. 86. Stashenko, E. E.; Martinez, J. R. Sampling flower scent for chromatographic analysis. Journal of Separation Science 2008, 31, (11), 2022-2031. 87. Takeoka, G. R.; Dao, L.; Rodriguez, D. M.; Patterson, R. Headspace volatiles of Scutellaria californica A. Gray flowers. Journal of Essential Oil Research 2008, 20, (2), 169-171. 88. Li, Z. G.; Lee, M. R.; Shen, D. L. Analysis of volatile compounds emitted from fresh Syringa oblata flowers in different flores cence by headspace solid-phase microextractiongas chromatography-mass spectrometry. Anal. Chim. Acta 2006, 576, (1), 43-49. 89. Custodio, L.; Serra, H.; Nogueira, J. M. F.; Goncalves, S.; Romano, A. Analysis of the volatiles emitted by whole flower s and isolated flower organs of the carob tree using HSSPME-GC/MS. J. Chem. Ecol. 2006, 32, (5), 929-942. 90. MacTavish, H. S.; Picone, J. M.; Clery, R. A. Identification of volatiles in headspace emitted from Mahonia japonica flowers. Journal of Essential Oil Research 2003, 15, (4), 231-233. 91. Flamini, G.; Cioni, P. L.; Morelli, I. Use of solid-phase micro-extraction as a sampling technique in the determination of volatiles emitted by flowers, isolated flower parts and pollen. J. Chromatogr. A 2003, 998, (1-2), 229-233.

PAGE 124

124 92. Fernando, L. N.; Grun, I. U. Headspace-SPME analysis of volatiles of the ridge gourd (Luffa acutangula) and bitter gour d (Momordica charantia) flowers. Flav. Fragr. J. 2001, 16, (4), 289-293. 93. Choi, H.-S. Characterization of Citrus uns hiu (C. unshiu Marcov. forma Miyagawa-wase) blossom aroma by solid-phase microextraction in conjunction with an electronic nose. J. Agric. Food Chem. 2003, 51, (2), 418-23. 94. Swingle, W.; Reece, P. The botany of Citrus and its w ild relatives of the orange subfamily University of California Press: 1943; Vol. I, p 129-474. 95. Tanaka, T. In Species Problem in Citrus Proc. Jpn. Soc. Prom. Sci, 1954; 1954. 96. Barrett, H. C.; Rhodes, A. M. A numerical ta xonomic study of affin ity relationships in cultivated Citrus and Its close relatives. Syst. Bot 1976, 1, (2), 105-136. 97. Moore, G. A. Oranges and lemons: clues to the taxonomy of C itrus from molecular markers. Trends Genet. 2001, 17, (9), 536-540. 98. Nicolosi, E.; Deng, Z. N.; Gentile, A.; La Ma lfa, S.; Continella, G.; Tribulato, E. Citrus phylogeny and genetic origin of important spec ies as investigated by molecular markers. Theor. Appl. Genet. 2000, 100, (8), 1155-1166. 99. Cunningham, J. P.; Moore, C. J.; Zalucki, M. P.; Cribb, B. W. Insect odour perception: recognition, of odour components by flower foraging moths. P. Roy. Soc. BBiol. Sci. 2006, 273, (1597), 2035-2040. 100. Kalinova, B.; Stransky, K.; Harmatha, J.; Ctvrtecka, R.; Zd'arek, J. Can chemical cues from blossom buds influence cultivar preference in the apple blossom weevil (Anthonomus pomorum)? Entomol. Exp. Appl 2000, 95, (1), 47-52. 101. Theis, N.; Raguso, R. A. The effect of pollination on floral fragrance in thistles. J. Chem. Ecol. 2005, 31, (11), 2581-2600. 102. Thiery, D.; Bluet, J. M.; Phamdelegue, M. H.; Etievant, P.; Masson, C. Sunflower aroma detection by the honeybee Study by coupling gas-chromatography and electroantennography. J. Chem. Ecol. 1990, 16, (3), 701-711. 103. Mestres, M.; Busto, O.; Guasch, J. Applica tion of headspace solid-phase microextraction to the determination of sulphur com pounds with low volatility in wines. J. Chromatogr. A 2002, 945, (1-2), 211-219. 104. Bazemore, R.; Goodner, K.; Rouseff, R. Vo latiles from unpasteurized and excessively heated orange juice analyzed with solid phase microextraction and GC-olfactometry. J. Food Sci. 1999, 64, (5), 800-803. 105. TetraPak. The orange book Tetra Pak Processing Systems AB: 1998; p 60-87.

PAGE 125

125 106. Shaw, P. E.; Moshonas, M. G.; Hearn, C. J.; Goodner, K. L. Volatile constituents in fresh and processed juices from grapef ruit and new grapefruit hybrids. J. Agric. Food Chem. 2000, 48, (6), 2425-2429. 107. Cremer, D. R.; Eichner, K. Formation of volatile compounds during heating of spice paprika (Capsicum annuum) powder. J. Agric. Food Chem. 2000, 48, (6), 2454-2460. 108. Yu, T. H.; Ho, C. T. Volatile compounds ge nerated from thermal-reaction of methionine and methionine sulfoxide with or without glucose. J. Agric. Food Chem. 1995, 43, (6), 1641-1646. 109. Zheng, Y.; Ho, C. T., Kinetics of the re lease of hydrogen sulfide from cysteine and glutathione during thermal treatment. In Sulfur compounds in foods Mussinan, C. J.; Keelan, M. E., Eds. ACS Symposium Se ries 564: Wahington, DC, 1994; pp 138-146. 110. Bicchi, C.; Drigo, S.; Rubiolo, P. Influen ce of fibre coating in headspace solid-phase microextraction-gas chromatographic analys is of aromatic and medicinal plants. J. Chromatogr. A 2000, 892, (1-2), 469-485. 111. Alissandrakis, E.; Daferera, D.; Tarantilis P. A.; Polissiou, M.; Harizanis, P. C. Ultrasound-assisted extraction of volatile compounds from citrus flowers and citrus honey. Food Chem. 2003, 82, (4), 575-582. 112. Jullien, F.; Gao, J.; Orel, G.; Legendre, L. Analysis of tissue-specific emission of volatiles by the flowers of six Camellia species. Flav. Fragr. J. 2008, 23, (2), 115-120. 113. Fernandes, F. M.; Figueiredo, A. C.; Barros o, J. G.; Pedro, L. G.; Wilcock, C. C.; de Carvalho, M. Volatiles of the inflorescen ces of the Madeiran orchids, Goodyera macrophylla Lowe and Gennaria diphylla (Lin k) Parl. and their role in pollination. Nat. Prod. Commun. 2007, 2, (4), 427-433. 114. Avery, M. L.; Tillman, E. A.; Laukert, C. C. Evaluation of chemical repellents for reducing crop damage by Di ckcissels in Venezuela. Int. J. Pest Manage 2001, 47, (4), 311-314. 115. Curtis, P. D.; Merwin, I. A.; Pritts, M. P.; Peterson, D. V. Chemical repellents and plastic netting for reducing bird damage to sw eet cherries, blueberries, and grapes. HortScience 1994, 29, (10), 1151-5. 116. Engeman, R. M.; Peterla, J.; Constantin, B. Methyl anthranilate aerosol for dispersing birds from the flight lines at Homestead Air Reserve Station. Int. Biodeter. Biodegr. 2002, 49, (2-3), 175-178. 117. Davies, F. S.; Albrigo, L. G. Citrus Wallingford, Oxon: CAB international: 1994; Vol. 2. 118. Sandoz, J. C. Odour-evoked responses to queen pheromone components and to plant odours using optical imaging in the antennal lobe of the honey bee drone Apis mellifera L. J. Exp. Biol. 2006, 209, (18), 3587-3598.

PAGE 126

126 119. Soost, R. K. In Self-incompatibility in Citr us Grandis (Linn.) Osbeck Proc. Am. Soc. Hort. Sci., 1964; 1964; pp 137-140.

PAGE 127

127 BIOGRAPHICAL SKETCH Fatima A. Jabalpurwala was born and raised in Mumbai, India. She received her Bachelor of Science in microbiology (1997) and Master of Science in foods, nutriti on and dietetics (1999) both from Mumbai University. Fatima was awar ded a research fellowship (19992000) from Bhabha Atomic Research Center, Mumbai for a project on wheat protein chemistry and bread quality. She later joined Consumer Science divisi on at Unilever Research, India and worked five years in the area of product deve lopment. Fatima enrolled in the PhD program in food science at University of Florida and was a recipient of a 4 year Alumni Fellowship. Her research years were spent at Citrus Research and Education Cent er, Lake Alfred, Florida. She received her PhD in 2009.