Electronic Nose Evaluation of the Effects of Canopy Side on Cabernet franc (Vitis vinifera L.) Grape and Wine Volatiles

Transcription

1 Electronic Nose Evlution of the Effects of Cnopy Side on Cbernet frnc (Vitis vinifer L.) Grpe nd Wine Voltiles Ymun S. Devrjn, 1 Bruce W. Zoecklein, 2 * Kumr Mllikrjunn, 3 nd Denise M. Grdner 1 Abstrct: The effect of grpevine cnopy side (north versus south nd est versus west) on grpe nd wine voltiles of Cbernet frnc ws evluted during two growing sesons using two electronic nose systems bsed on conducting polymers nd surfce coustic wves. Dt from three smpling dtes per seson from both electronic noses were compred with physicochemistry nd wine rom sensory evlutions. Univrite nd multivrite sttisticl nlyses generlly indicted grpe physicochemistry indices could not differentite consistently (p > 0.05) between cnopy sides cross growing sesons nd smpling dtes. Both electronic nose (ENose) systems provided complete discrimintion of cnopy sides for grpes nd wine using cnonicl discriminnt nlysis. On verge, the surfce coustic wve-bsed ENose explined <50% of vrition for grpes nd <60% for wine using the first principl component, compred to >80% for the conducting polymer-bsed ENose. Wine rom sensory evlution differentited cnopy sides in three of four evlutions. Key words: cnopy side, grpe voltiles, wine voltiles, electronic nose, Cbernet frnc Fruit from different grpevine cnopy loctions my hve different mturity becuse of vritions in het nd light exposure (Downey et l. 2006), cusing ltered grpe composition (Jckson nd Lombrd 1993). Severl studies hve evluted the effects of mngement prctices on grpe composition, using both nturl nd rtificil methods for mnipulting light intensity: plstic sheet (Kliewer et l. 1967), wxed bgs (Wever nd McCune 1960), shde cloth (Smrt et l. 1988), shded- nd light-exposed berries from the sme cluster (Price et l. 1995), nd cnopy sides (Bergqvist et l. 2001). Some studies hve shown tht incresed exposure to sunlight delys ripening by inhibiting sugr ccumultion (Bergqvist et l. 2001, Kliewer 1977) nd my ripen berries unevenly compred to clusters growing in shde (Kliewer nd Lider 1968). In contrst, other studies hve shown tht the percentge of soluble solids is not ffected by light exposure (Crippen nd Morrison 1986, Spyd et l. 2002). Berry weight ws higher in shded berries in some studies (Crippen nd Morrison 1986), but not others (Crippen nd Morrison 1986b). Exposed clusters hve been shown to hve lower titrtble cidity (Bergqvist et l. 2001, Crippen nd Morrison 1986, Kliewer nd Lider 1968) nd ph (Bergqvist et l. 2001) thn shded clusters. The pprently contrdictory results on grpe composition due to light exposure suggest the need for exploring lterntive evlution tools. 1 Grdute Student, 2 Professor, Enology-Grpe Chemistry Group, nd 3 Associte Professor, Biologicl Systems Engineering, Virgini Tech, Blcksburg, VA *Corresponding uthor (emil: bzoeckle@vt.edu; fx: ) Mnuscript submitted Mr 2010, revised Sep 2010, ccepted Sep 2010 Copyright 2011 by the Americn Society for Enology nd Viticulture. All rights reserved. doi: /jev Electronic nose (ENose) technology hs been used in vriety of food industry pplictions, including qulity ssurnce, process monitoring, study of storge conditions, mturtion nd ging of wine, nd product-pckge interction (Mielle 1996), nd lso used to evlute the effect of vineyrd mngement technique on grpe nd wine voltiles (Mrtin et l. 2008). ENose is chemosensor rry-bsed technology usully consisting of series of sensors. Upon exposure to chemicl vpor, sensors undergo physicochemicl chnges (Mllikrjunn 2005). Bsed on the sensor, electronic noses cn be clssified into the following systems: conducting polymer (CP), qurtz microblnce (QMB), metl oxide sensors (MOS), metl oxide semiconductor field effect trnsistors (MOSFET), nd surfce coustic wve (SAW). All systems consist of three mjor prts: sensors, system controls, nd dt processing units (Mllikrjunn 2005). Precision nd ccurcy of n ENose depend on severl chrcteristics, including sensor selectivity, operting temperture, humidity, sensor drift, nd sensitivity to prticulr compound. Hence, the use of prticulr sensor could be limited to certin pplictions. CP, SAW, nd MOS electronic noses re the most commonly used due to sensor drift over time (Mielle 1996). Conducting polymer-bsed systems consist of severl sensors. A chnge in the resistnce of ech sensor is stored s smell print during introduction of stndrd smples, nd n unknown smple is compred with the vilble smell prints for identifiction (Cyrno Sciences 2000). CP sensors re composed of different polymers: polyniline, polypyrrole, polythiophene, polycetylene, nd polyindole t different oxidtion-reduction sttes to provide selectivity to different compounds (Mllikrjunn 2005, Pinheiro et l. 2002). The bility of CP ENose to mesure grpe voltiles nondestructively hs been demonstrted (Athmneh et l. 2008). SAWbsed systems consist of single sensor, which simultes virtul sensor rry s if consisting of 100 orthogonl sensors 73

2 74 Devrjn et l. (Mllikrjunn 2005). The commercil znose uses SAWbsed sensors nd opertes s miniture fst-reding gs chromtogrphy (GC) unit (Mllikrjunn 2005). Smples re drwn into the smpling port nd sent through column. Identifiction of voltile compounds is bsed on pek retention time in the column nd quntifiction is determined by frequency shifts of the qurtz crystl detector, which correspond to the mount of mteril deposited (Electronic Sensor Technology 2001). The dt obtined from this system cn be nlyzed using chromtogrphic or spectroscopic pproches (Lmmertyn et l. 2004, Mllikrjunn 2005). Primry problems reported with the use of CP sensors for wines re the influence of ethnol (Rgzzo-Snchez et l. 2006) nd wter vpor. However, n ethnol bseline with conducting polymer ENose system hs been used to minimize the ethnol interference with sensor redings, bsing discrimintion on wine rom voltiles (Sntos et l. 2004). Sensitivity generlly vries ccording to sensory type nd polymer type for CP ENose systems. Reported sensitivity is in the µg/l to ng/l rnge (Mllikrjunn 2005). In this study, we sought to determine the bility of CP nd SAW electronic nose systems to discriminte grpevine cnopy sides, compred to trditionl nlyses. Mterils nd Methods Field design. This study ws performed during 2007 nd 2008 on Cbernet frnc grpes grown on Bllerin trining system in Chrlottesville, Virgini. Het summtion for 2007 nd 2008 ws C nd C, respectively. Men monthly precipittion from April through October ws 81 mm nd men reltive humidity ws 75% in September. Both vineyrds were dry frmed. Grpevine rows with cnopy sides fcing true north/south (±5 ) nd true est/west (±5 ) were plnted in 2003 on m 2 nd m 2 plots, respectively. Vines were clone 4 grfted onto rootstock. Vines on both plots were spced 2.13 x 3.05 m prt. In 2007, 10 grpevines were selected using rndomized block design for both the est/west- nd north/south-oriented plots. Fruit ws smpled bsed upon cnopy side on three smpling dtes postbloom (once per week) in both sesons. Smpling ws conducted on weeks 12, 14, nd 15 postbloom in 2007, nd weeks 14, 15, nd 17 postbloom in The lst smpling dte for both sesons ws t commercil hrvest. Degrees Brix by refrctometer (model 10430; AO Scientific Instruments, Keene, NH), %RH by digitl hygrometer (model 4187; Trceble, Control Compny, Friendswood, TX), nd temperture by infrred thermometer (model 42529; Extech Instruments, Wlthm, MA) were mesured on both sides within the vine cnopy on ll smpling dtes, between 0800 nd 1100 hr. At the end of the growing seson, ~80.0 kg of fruit ws hrvested from ech cnopy side. Severl components of yield were determined: shoots/meter, clusters/vine, clusters/shoot, cluster weight, berry weight, fruit weight/vine, nd fruit weight/cnopy side. Lbortory nlysis. Twenty-five berries per vine were rndomly collected from ech side of ech vine t ech smpling dte, processed, nd nlyzed immeditely. Berry smples were crushed in commercil blender (Wring model 13BL91; New Hrtford, CT) for one second nd plced into filter bg (model 400; Stewrd Stomcher Lb System, London, UK). The expelled juice ws filtered through 0.45-µm syringe filters (Whtmn, Clifton, NJ). Berry weight, Brix, ph, nd titrtble cidity (TA) were determined s described previously (Zoecklein et l. 1999). Color intensity (A A 520 ), hue (A 420 /A 520 ), nd totl phenols (A 280 ) were estimted using Genesys 5 spectrophotometer (Spectronic, Leeds, UK). Yest ssimilble nitrogen ws determined enzymticlly (Megzyme, Bry, Irelnd). The totl glycosyl-glucose (TGG) nd phenol-free glycosyl glucose (PFGG) nlyses were performed s described (Willims et l. 1995) nd modified (Zoecklein et l. 2000). For wines, nlyses of mlic cid, yest ssimilble nitrogen, lcohol concentrtion, residul sugr, nd voltile cidity were lso conducted. l-mlic cid ws determined enzymticlly (R-Biophrm AG, Drmstdt, Germny). Alcohol ws determined by FTIR (model FT 120; Foss WineScn, Eden Pririe, MN) nd residul sugr concentrtion by Clinitest (Byer, Elkhrt, IN). Wine smples for GC-MS nlysis were prepred using 4-mL smple with 1.0 g NCl in 10-mL cler glss vils seled with sept (MicroLiter Anlyticl Supplies, Suwnee, GA), s described (Whiton nd Zoecklein 2000). Vils were pre-incubted 30 sec t 30 C with gittion t 250 rpm. A CAR/DVB/PDMS grey SPME fiber (Supelco Sigm-Aldrich, St. Louis, MO) ws used to penetrte vils to 32-mm depth. A GC-MS (model 6890N, Network GC System, 5975B inert MSD; Agilent Technologies, Snt Clr, CA) with injector temperture t 250 C, DB-wx column (30 m x 0.25 mm), nd helium crrier gs with flow rte of 1 ml/min ws used. Oven temperture ws 40 C with rmp rte of 6 C/min to 230 C. Thirty-two stndrd compounds from ech wine smple were mnully integrted nd quntified. Processing nd fermenttion. Grpes hrvested from ech cnopy side were crushed nd destemmed using Wottle (Anton, Poysdorf, Austri) destemmer-crusher to bout 50% berry brekge, estimted visully. Fruit ws distributed into three open-top 60-L Nlgene fermenting bins (Thermo Fisher Scientific, Wlthm, MA) of equl height nd volume for ech cnopy side. Ech bin ws treted with 250 mg/l Velcorin (Scott Lbortories, Petlum, CA) dimethyl-dicrbonte. Bins were held for 24 hr t 7 C in cooler, followed by 25 mg/l potssium metbisulfite ddition. Grpes were cold soked for 6 dys t 7 C, with dily punching of the must. Must juice nlysis ws then performed, following ddition of 0.24 g/l FermAid K (Lllemnd, Blgnc, Frnce). Scchromyces cerevisie ICV-D254 yest (Lllemnd) (0.24 g/l) ws inoculted following the cold-sok tretment. Go-Ferm (Lllemnd) yest nutrient ws prepred ccording to mnufcturer instructions nd dded during yest rehydrtion. After inocultion, cps were punched three times dily. Fermenttion ws monitored by hydrometer nd crried out t 25 ± 2 C until dryness (<1% residul sugr), s determined by Clinitest (Byer). Following fermenttion, wines were dejuiced using Willmes press (model 100; Bensheim, Germny) to 0.5 br. Free run nd press run frctions were

3 Effects of Cnopy Side on Cbernet frnc Voltiles 75 combined nd plced into snitized, crbon dioxide-filled glss crboys. Wine ws kept t 7 C for 24 hr, rcked into 3.80-L glss bottles, nd stored t 12 C. Electronic nose nlysis. Two electronic nose (ENose) systems, conducting polymer (CP) (Cyrnose 320, Smiths Detection, Psden, CA; used in the field nd in the lbortory) nd surfce coustic wve (SAW) (ZNose 730, Electronic Sensor Technology, Newbury Prk, CA; lbortory only) were used to determine voltile differences between cnopy sides in grpes nd in wine produced from those grpes. The CP ENose optimiztion method ws s described in n erlier study on the impct of field vribles in our climtic region, such s het nd humidity, on the CP system (Athmneh et l. 2006). Optimiztion of the wine evlution method for the CP ENose ws s described previously (Grdner 2009). Optimiztions involved evlutions of smple temperture, incubtion time, pump speed, purge, nd individul sensor response. Cnonicl projection plots showed no seprtions between replictions, indicting minimum sensor drift. The optimum smple temperture ws 30 C. Ethnol stndrd solutions (three stndrds per tretment) were used to crete n ethnol bseline to evlute the impct on polymerbsed sensor response nd to minimize differences in lcohol concentrtions mong wine smples while using the ENose. Concentrtions for ethnol stndrds were bsed on the lcohol concentrtion recorded for ech wine tretment. For field nlysis, two clusters (one from ech cnopy side) were chosen t rndom from ech of 10 selected vines. Clusters were bgged with n HDPE bg (Inteplst, Livingston, NJ) nd equilibrted for 45 min. In 2007, dditionl clusters from neighboring vines were collected for lbortory nlysis. Clusters were plced into individul 1.50-L glss jrs in wter bth t 30 C for 20 min nd then nlyzed by CP ENose. Berry nlysis ws lso performed, using 50 g of berries picked from the cluster nd plced in 200-mL glss Mson jr. Wines were nlyzed twice: immeditely postfermenttion nd gin 6 months postfermenttion. Five replictes of 20- ml wine smples were plced in 40-mL GC cler glss vils seled with teflon/silicone 3-mm sept (MicroLiter Anlyticl Supplies). The smples were plced in wter bth t 30 C for 20 min nd subsequently nlyzed by CP ENose. For the SAW ENose, the defult settings of the DB-5 column system were used, except for the sensor temperture of 45 C. The system ws tuned with C6 to C14 lkne stndrds ech dy. The sme smpling technique for the CP ENose wine nlysis ws used for this system. Sensory nlysis. Tringle difference tests on wine rom were conducted under stndrd conditions 6 months postfermenttion s described (Meilgrd et l. 2007), compring est versus west nd north versus south. Wines were prescreened for sulfur-like off odors nd ssigned rndomized 3-digit code. Pnelists were given three smples nd sked to identify the odd smple. Stndrd ISO glsses were filled with 10 ml wine, covered with plstic petri dish, nd presented to untrined consumer pnelists t 19 C under red light. Ech yer, 32 pnelists with ges between 21 nd 27 yers were used (α = 0.05, β = 0.30, rd = 40%, 16 correct responses required for significnt difference). Approximtely equl numbers of mle nd femle pnelists were selected, with prerequisite of wine consumption t lest once week. Written instructions were provided nd pnelists received two 20-min sessions outlining the procedures. Pnelists smelled two sets of smples t different times. Sttisticl nlysis. The physicochemistry, CP ENose, nd SAW ENose dt were nlyzed nd compred using univrite (one-wy nlysis of vrince [ANOVA] nd lest significnt difference [LSD]) nd multivrite (cnonicl discriminnt nlysis [CDA] nd principl component nlysis [PCA]) sttisticl methods using SAS JMP version 7 softwre (SAS Institute, Cry, NC). The SAW ENose dt ws nlyzed using the GC chromtogrphic pproch s described (Lmmertyn et l. 2004). For the sensory dt, the significnce of the number of correct responses ws determined s described (Meilgrd et l. 2007). Results nd Discussion Components of yield t hrvest did not differ mong grpevine cnopy sides for either seson, with the exception of berry weight in 2008 (Tble 1). Cnonicl discriminnt nlysis (CDA) of seven grpe physicochemistry indices (Brix, berry weight, ph, TA, color intensity, hue, nd totl phenols) demonstrted tht the bility of these mesures to identify grpevine cnopy side incresed with weeks postbloom (Tble 2). However, these indices did not consistently predict cnopy side on ny smpling dte, unlike both ENoses. CDA plots of the physicochemistry dt for 2007 nd 2008 generlly illustrted tht these indices could not differentite both Est from West nd North from South (Figure 1). Such plots represent the multivrite men of the dt points s circles whose size indictes the 95% confidence limit for the men. Nonintersecting circles indicte significnt differences. The p vlues for the nlyses of nine physicochemistry indices (PFGG, TGG, nd the seven indices listed bove) were Tble 1 Components of yield of different cnopy sides for Cbernet frnc grpes t hrvest in Cnopy side Shoots/meter Clusters/side Fruit wt (kg) Cluster wt (kg) Berry wt (g) Est ± ± ± ± 0.04 b 1.77 ± 0.07 b West ± ± ± ± 0.04 b 1.63 ± 0.07 b North ± ± ± ± ± 0.07 South ± ± ± ± 0.04 b 1.89 ± 0.07 p vlue Columns with different letters indicte 95% significnt difference between tretments.

4 76 Devrjn et l. determined (Tble 3). Pirwise comprison (t-test) results showed tht Brix (2007), TA (2008), nd ph (both sesons) vlues were not significntly correlted with cnopy side on most smpling dtes, similr to previous results (Athmneh et l. 2008). However, differences between cnopy sides were evident for Brix (2008) nd TA (2007). Berry weight ws differentited between cnopy sides in both sesons for ll smpling dtes. Fruit color intensity showed cnopy differences in 2007, but not in In generl, the physicochemistry indices did not show consistent differences between cnopy sides cross both sesons. Cnonicl plots demonstrted the bility of the CP ENose to distinguish cnopy sides t most mturity stges evluted in the field (Figure 2). The bility to minimize the environmentl vribles of het nd humidity ws confirmed by compring the CP field results with CP lbortory nlyses, which demonstrted exctly the sme trends (dt not shown). Bsed on the ANOVA of in-field CP ENose dt, most sensors were sensitive (p < 0.05) to Cbernet frnc grpe voltiles, with the exceptions of S19, S24, nd S32 (dt not shown). The cnonicl plots of the SAW ENose dt on cnopy sides of fruit from three smpling dtes in 2008 re shown (Figure 3). This dt is representtive of both sesons nd demonstrted tht the SAW system showed similr bility to discriminte cnopy side. Principl component nlysis (PCA) of physicochemistry nd CP nd SAW ENose dt showed tht, for ech dt source, the first three components together explined 100% of the vrition (Tble 4). CP ENose dt explined most of the vrition (>90%) in single (PC1) xis, while physicochemistry nd the SAW ENose dt explined similr vrition using PC1 nd PC2. PCA is multivrite sttisticl method Tble 2 Cnonicl discrimintion of Cbernet frnc grpes t three times postbloom; physicochemistry nd conducting polymer-bsed (CP) in 2007 nd 2008 nd surfce coustic wve-bsed (SAW) ENose in Smpling dte Chemistry CP ENose SAW ENose 2007 Week 12 80% 100% n b Week 14 90% 100% n Week 15 95% 100% n 2008 Week 14 80% 100% 100% Week 15 85% 100% 100% Week 17 95% 100% 100% Vlues indicte the percentge of correct predictions of est vs west nd north vs south. b n indictes dt not vilble. Figure 1 Cnonicl distribution of differences detected by physicochemistry nlyses of Cbernet frnc juice on three smpling dtes during 2007 postbloom weeks 12 (A), 14 (B) nd 15 (C), nd 2008 postbloom weeks 14 (D), 15 (E), nd 17 (F). Significnt differences t α = 0.05 re indicted by nonintersecting circles. Tble 3 p Vlues (ANOVA) of cnopy side differences in Cbernet frnc juice detected by physicochemicl nlyses during 2007 nd 2008 on three smpling dtes (postbloom weeks 12, 14, nd 15 in 2007, nd weeks 14, 15, nd 17 in 2008) Physicochemistry indices Week 12 Week 14 Week 15 Week 14 Week 15 Week 17 Brix Berry weight (g) ph Titrtble cidity (g/l) Color intensity (A 420 +A 520 ) Hue (A 420 /A 520 ) Totl phenols (A 280 ) PFGG (μm) n n 0.11 TGG (μm) n n 0.00 p vlues 0.05 indicte significnce; n indictes dt not vilble.

5 Effects of Cnopy Side on Cbernet frnc Voltiles 77 in which the vrition of dt is summrized in the form of principl components. This method explins the vrition in dt by replcing the lrger set of vribles correlted with cnopy side with smller set of uncorrelted vribles. The conducting polymer-bsed system discriminted better between cnopy sides thn the surfce coustic wvebsed system (Tble 4). This could be ttributed to the types of dt sets in the two systems. Better discrimintion cn be observed with lrger dt set (Vndeventer nd Mllikrjunn 2003). Bsed on biplot ry lengths for physicochemistry dt in 2007 nd 2008, Brix ws the lest-effective prmeter in detecting cnopy side difference, followed by TA nd hue (dt not shown). Wine chemistry prmeters (PFGG, TGG, color intensity, hue, nd totl phenols) of fruit from Figure 2 Cnonicl plots of cnopy side differences for Cbernet frnc grpe berries, detected by conducting polymer-bsed ENose in the field, in 2007 postbloom weeks 12 (A), 14 (B), nd 15 (C), nd 2008 postbloom weeks 14 (D), 15 (E), nd 17 (F). Significnt differences t α = 0.05 re indicted by nonintersecting circles. Figure 3 Cnonicl plots of cnopy side differences for Cbernet frnc grpe juice, detected by surfce coustic wve-bsed ENose in the lbortory, in 2008 postbloom weeks 14 (A), 15 (B), nd 17 (C). Significnt differences t α = 0.05 re indicted by nonintersecting dt groups. Tble 4 Principl component nlysis of Cbernet frnc juice, showing the difference between cnopy side detected by physicochemicl nlyses, conducting polymer-bsed (CP) ENose in 2007 nd 2008 nd surfce coustic wve-bsed (SAW) ENose in Smpling dte Physicochemistry CP SAW Smpling Physico- CP SAW ENose ENose dte chemistry ENose ENose Principl component Week n Week PC n PC n PC3 Week n Week PC n PC n PC3 Week n Week PC n PC n PC3 Vlues indicte the percentge of vrition explined by principl components 1, 2, nd 3; n indictes dt not vilble.

6 78 Devrjn et l. Tretment Tble 5 Pirwise comprison dt of 2008 Cbernet frnc wine chemistry indices. Color intensity (A 420 +A 520 ) Hue (A 420 /A 520 ) Totl phenols (A 280 ) Est ± 0.03 b ± 0.01 c ± 0.03 c 62.7 ± ± b West ± 0.03 c ± 0.01 b ± 0.03 d 28.2 ± 5.30 bc 1323 ± b North ± ± 0.01 b ± ± 5.30 b 1608 ± South ± ± ± 0.04 b 15.2 ± 5.30 c 1598 ± Different letters within columns indicte 95% significnt difference between tretments. PFGG (µm) TGG (µm) Figure 4 Cnonicl plots of cnopy side differences for Cbernet frnc wine, detected by conducting polymer-bsed ENose in the lbortory, in 2008 immeditely postfermenttion (A), nd 6 months postfermenttion (B). Significnt differences t α = 0.05 re indicted by nonintersecting circles. Figure 5 Cnonicl distribution of 2008 Cbernet frnc wine 6 months postfermenttion, using (A) surfce coustic wve-bsed ENose nd (B) GC-MS. Significnt differences t α = 0.05 re indicted by nonintersecting dt groups. Tble 6 Pirwise comprisons nd ANOVA on 2008 est, west, north, nd south Cbernet frnc wine voltiles 6 months postfermenttion. Concentrtion (µg/l) Compound Est West North South p vlue Ethyl cette ± ± ± ± Methyl propnol ± ± ± ± 0.81 b Isomyl cette ± c ± c ± ± b < n-butnol 9.12 ± 0.02 b 9.13 ± ± 0.02 b 9.08 ± 0.02 b Methyl butnol ± 2.13 b ± 2.13 b ± ± Ethyl hexnote ± 5.98 b ± 5.98 c ± ± Hexyl cette 4.17 ± ± 0.08 d 3.54 ± 0.08 b 2.61 ± 0.08 c < Ethyl heptnote 5.01 ± 0.11 b 5.05 ± 0.11 b 5.14 ± 0.11 b 6.27 ± n-hexnol 1.54 ± 0.03 b 1.69 ± ± 0.03 c 0.99 ± 0.03 c < Ethyl octnote ± 2.54 b ± 2.54 b ± ± 2.54 < Ethyl-1-hexnol 6.60 ± 0.24 b 8.14 ± ± 0.24 b 5.36 ± 0.24 c Ethyl nonnote ± 0.03 c ± 0.03 b ± ± 0.03 b Octnol ± 2.51 b ± ± 2.51 c ± 2.51 d < Terpinene-4-ol ± 0.35 c ± ± 0.35 b 9.13 ± 0.35 d < Ethyl decnote ± 1.53 d ± 1.53 c ± ± 1.53 b < Isomyl octnote ± 0.05 d ± 0.05 c ± ± 0.05 b < Nonnol 7.92 ± 0.14 b ± ± 0.14 c 5.58 ± 0.14 d < Isovleric cid 2.23 ± 0.06b c 2.05 ± 0.06 c 2.40 ± 0.06 b 2.57 ± Diethyl succinte ± ± 9.19 b ± ± Methionol 1.56 ± ± ± ± Citronellol ± 1.20 c ± ± 1.20 b ± 1.20 b Phenethyl cette ± 2.44 b ± 2.44 b ± ± β-dmscenone ± 0.41 b ± ± 0.41 c ± 0.41 c < Hexnoic cid 1.04 ± ± ± ± Ethyl dodecnote ± 0.41 b ± 0.41 b ± ± 0.41 < Benzyl lcohol ± 4.60 b ± 4.60 b ± ± Phenethyl lcohol ± 0.85 b ± 0.85 b ± 0.85 b ± γ-nonlctone 0.08 ± 0.00 b 0.09 ± ± 0.00 c 0.07 ± 0.00 c < Ethyl myristte ± ± 0.16 b ± 0.16 b ± 0.16 b Octnoic cid 1.63 ± ± ± ± Ethyl plmitte ± 0.66b c ± 0.66 c ± 0.66 b ± 0.66 < Different letters within row indicte 95% significnt difference between tretments.

7 Effects of Cnopy Side on Cbernet frnc Voltiles 79 ech cnopy side were nlyzed using pirwise comprisons nd ANOVA. Generlly, results indicted minor differences in cnopy side ech seson. For exmple, in 2008, cnopy side differences were seen except for TGG (Tble 5). Percentge lcohol (v/v), ph, TA, nd mlic cid generlly illustrted nonsignificnt differences between wines produced from the different cnopy sides ech seson (dt not shown). Wine voltiles were nlyzed both immeditely postfermenttion nd 6 months postfermenttion using the CP ENose. These dt explined most of the vrition long the PCA PC1 xis (>80%), both immeditely postfermenttion nd 6 months postfermenttion. Six months postfermenttion gve better discrimintion of cnopy sides (PC1: 97.6%, PC2: 1.7%, nd PC3: 0.7%) thn did immeditely postfermenttion (PC1: 81.1%, PC2: 17.9%, nd PC3: 1.0%). The SAW ENose nlysis performed immeditely postfermenttion ws ble to explin <50% vrition using PC1 (PC1: 48.4%, PC2: 35.6%, nd PC3: 16.0%). However, both CP ENose nd SAW ENose nlyses of wine voltiles were ble to explin 100% of the vrition (bsed on CDA) immeditely postfermenttion nd 6 months postfermenttion (Figure 4, Figure 5). ANOVA on CP ENose sensor responses shows tht most of the sensors were sensitive to cnopy side differences in the wine. Concentrtion differences in voltiles were detected in wines produced from different cnopy sides using trditionl GC-MS 6 months postfermenttion (Tble 6). CDA nd PCA (PC1: 67.5%, PC2: 18.6%, nd PC3: 13.9%) explined 100% of the vrition in the dt cross cnopy sides. PCA biplot rys indicted tht ethyl myristte, citronellol, ethyl nonnote, nd hexyl cette were the compounds most ssocited with cnopy side differences (dt not shown). Pnelists were ble to significntly (α = 0.05) differentite wine rom between est nd west (18 nd 17, of 32 correct responses, for 2007 nd 2008, respectively; Tble 7), but between north nd south in 2007 only (16 correct responses, compred to 13 of 32 in 2008). These results might hve been different if trined pnel hd been used, lthough trined pnel is not necessrily indictive of consumer response. Lck of differences detected by sensory tests does not lwys illustrte tretment similrities, since it my involve bis nd vribility over time (Meilgrd et l. 2007). Although the ENose my produce results similr to sensory nlysis (Mllikrjunn 2005), it evlutes both rom nd nonrom voltiles (Hugen nd Kvl 1998). Additionlly, electronic noses hve the bility to objectively evlute voltiles in complex mtrix such s wine. Tble 7 Cbernet frnc wine rom tringle difference sensory results (n = 32) of est versus west nd north versus south tretments for 2007 nd Tretment Correct responses Signf Est vs west, yes Est vs west, yes North vs south, yes North vs south, no α = 0.05, β = 0.10, ρ mx = 40%, 16 or more correct responses corresponds to significnt difference. Conclusion A mjor chllenge for the grpe nd wine industry is to replce time-consuming lbortory nlyses with new techniques tht re fst, precise, nd ccurte. A further chllenge is to understnd the reltionships mong vineyrd mngement, fruit chemistry, wine chemistry, nd sensory response. While most industry prctitioners understnd the potentil differences in grpe nd, therefore, wine composition tht my result from differences in grpevine cnopy side, it cn be difficult to quntify those differences. Such quntifiction is required to exmine the economics of differentil hrvest. This study ws performed to determine if conducting polymer-bsed nd surfce coustic wve-bsed electronic nose systems could distinguish nd discriminte between grpe nd wine voltiles cross vine cnopy sides. Results were compred with trditionl physicochemistry indices. The physicochemistry nlyses did not differentite between cnopy sides on most smpling dtes nd did not show trends cross growing sesons. However, this study demonstrted the bility of both ENose systems to distinguish between cnopy sides t ll mturity stges evluted. Sensory evlutions of wine rom lso showed cnopy side differences. The differences detected by GC-MS for individul wine voltile compounds (25 out of 32) support the vrition detected by the ENose systems. Considering the time efficiency nd nondestructive procedure, the CP ENose my be more useful thn the SAW ENose to differentite between cnopy sides. Literture Cited Athmneh, A.I Electronic nose evlution of grpe mturity. Thesis, Virgini Polytechnic Institute nd Stte University, Blcksburg. Athmneh, A.I., B.W. Zoecklein, nd K. Mllikrjunn Electronic nose evlution of Cbernet Suvignon fruit mturity. J. Wine Res. 19: Bergqvist, J., N. Dokoozlin, nd N. Ebisud Sunlight exposure nd temperture effects on berry growth nd composition of Cbernet Suvignon nd Grenche in the Centrl Sn Joquin Vlley of Cliforni. Am. J. Enol. Vitic. 52:1-7. Crippen, Jr., D.D., nd J.C. Morrison The effects of sun exposure on the compositionl development of Cbernet Suvignon berries. Am. J. Enol. Vitic. 37: Crippen, Jr., D.D., nd J.C. Morrison. 1986b. The effects of sun exposure on the phenolic content of Cbernet Suvignon berries during development. Am. J. Enol. Vitic. 37: Cyrno Sciences Inc The Prcticl Guide to the Cyrnose 320 TM. Cyrno, Psden, CA. Downey, M.O., N.K. Dokoozlin, nd M.P. Krstic Culturl prctice nd environmentl impcts on the flvonoid composition of grpes nd wine: A review of recent reserch. Am. J. Enol. Vitic. 57: Electronic Sensor Technology Fst GC-Anlyzer Opertion Mnul. EST, Newbury, CA. Grdner, D.M Chrcteriztion of cold sok on Vitis vinifer L. cv. Cbernet Suvignon grpe nd wine voltiles using n electronic nose system. Thesis, Virgini Polytechnic Institute nd Stte University, Blcksburg. Hugen, J.E., nd K. Kvl Electronic nose nd rtificil neurl network. Met Sci. 49:S273-S286.

8 80 Devrjn et l. Jckson, D.I., nd P.B. Lombrd Environmentl nd mngement prctices ffecting grpe composition nd wine qulity A review. Am. J. Enol. Vitic. 44: Kliewer, W.M., L.A. Lider, nd H.B. Schultz Influence of rtificil shding of vineyrds on the concentrtion of sugr nd orgnic cid in grpes. Am. J. Enol. Vitic. 18: Kliewer, W.M., nd L.A. Lider Influence of cluster exposure to the sun on the composition of Thompson seedless fruit. Am. J. Enol. Vitic. 19: Kliewer, W.M Influence of temperture, solr rdition nd nitrogen on colortion nd composition of Emperor grpes. Am. J. Enol. Vitic. 28: Lmmertyn, J., E.A. Ververbeke, nd J. Irudyrj znose TM technology for the clssifiction of honey bsed on rpid rom profiling. Sens. Actutors, B Chem. 98: Mllikrjunn, P Electronic nose pplictions in the food industry. In Nondestructive Testing of Food Qulity. J. Irudyrj et l. (eds.), pp Wiley-Blckwell, Ames, IA. Mrtin, A., K. Mllikrjunn, nd B. Zoecklein Discrimintion of wines produced from Cbernet Suvignon grpes treted with queous ethnol post-bloom using n electronic nose. Int. J. Food Eng. 4(2). doi: / Meilgrd, M.C., G.V. Civille, nd B.T. Crr Sensory Evlution Techniques. 4th ed. CRC Press, Boc Rton, FL. Mielle, P Electronic noses: Towrds the objective instrumentl chrcteriztion of food rom. Trends Food Sci. Technol. 7: Pinheiro, C., C. Rodrigues, T. Schäfer, nd J. Crespo Monitoring the rom production during wine-must fermenttion with n electronic nose. Biotechnol. Bioeng. 77: Price, S.F., P.J. Breen, M. Vlldo, nd B.T. Wtson Cluster sun exposure nd quercetin in Pinot noir grpes nd wine. Am. J. Enol. Vitic. 46: Rgzzo-Snchez, J.A., P. Chlier, D. Chevlier, nd C. Ghommidh Electonic nose discrimintion of rom compounds in lchoholised solutions. Sens. Actutors, B 114: Sntos, J.P., J. Lozno, M. Aleixndre, I. Sygo, M.J. Fernndez, L. Ares, J. Gutierrez, nd M.C. Horrillo Discrimintion of different romtic compounds in wter, ethnol nd wine with thin film sensor rry. Sens. Actutors, B 103: Smrt, R.E., S.M. Smith, nd R.V. Winchester Light qulity nd quntity effects on fruit ripening for Cbernet Suvignon. Am. J. Enol. Vitic. 39: Spyd, S.E., J.M. Trr, D.L. Mee, nd J.C. Ferguson Seprtion of sunlight nd temperture effects on the composition of Vitis vinifer cv. Merlot berries. Am. J. Enol. Vitic. 53: Vndeventer, D., nd P. Mllikrjunn Comprtive performnce nlysis of three electronic nose systems using different sensor technologies in odor nlysis of retined solvents on printed pckging. J. Food Sci. 67: Wever, R.J., nd S.B. McCune Influence of light on color development in Vitis vinifer grpes. Am. J. Enol. Vitic. 11: Whiton, R.S., nd B.W. Zoecklein Optimiztion of hedspce solid-phse microextrction for nlysis of wine rom compounds. Am. J. Enol. Vitic. 51: Willims, P.J., W. Cynkr, I.L. Frncis, J.D. Gry, P.G. Ilnd, nd B.G. Coombe Quntifiction of glycosides in grpes, juices, nd wines through determintion of glycosyl glucose. J. Agric. Food Chem. 43: Zoecklein, B.W., L.S. Dougls, nd Y.W. Jsinski Evlution of the phenol-free glycosyl-glucose determintion. Am. J. Enol. Vitic. 51: Zoecklein, B.W., K.C. Fugelsng, B.H. Gump, nd F.S. Nury Wine nlysis nd production. Kluwer Acdemic Publishers, New York.