Clone candidates differentiation of grapevine Vitis vinifera ' Škrlet bijeli ' using aroma compounds detected by gas chromatography-mass spectrometry

The aim of this work was to investigate existence presence and stability of must specific aroma compounds (monoterpenes C13-norisoprenoids, C6-alcohols, alcohols, esters and carbonyl compounds) and which can be used to establish differences among clone candidates of 'Škrlet bijeli' (Vitis vinifera L.) grapevine variety. The compounds responsible for the varietal aroma profile were determined by gas chromatographymass spectrometry (GC-MS), in must samples of ten clone candidates grown on two vineyard sites for three consecutive years. Significant variation among clone candidates is shown in 22 out of the total 35 identified aroma compounds. Significant impact of the vineyard site on the clone candidate’s aroma profile was identified. Differences in primary aroma compounds responsible for flavour of 'Škrlet bijeli' variety, linalool, terpinolen, nerol and α-terpineol, were not significant among clone candidates, while remarkable differences were established for β-damascenone. Contrary to expectation, monoterpene geraniol was not detected. Other identified aroma compounds (trans-ocimene, 2-methyl-1butanol, myrcene, α-phelandrene, cis-ocimene and 3-methyl1-butanol) noticeably less participate in total flavour description, but they still enable notable clone candidates discrimination.


INTRODUCTION
Individual clonal selection is the method most commonly used for genetic improvement of autochthonous grapevine varieties, which results by divergent clones.Usually the selection is performed on grape quality parameters such as sugars and acids content, while aroma profile is checked in late selection stages.How the wine aroma profile is important for wine identity and quality and as aroma precursors originated from must, it is very important to include monitoring of must aroma compounds in early stages of clonal selection.Specific varietal wine aroma originates from volatile compounds such as monoterpenes, norisoprenoids, aliphatic compounds, phenylpropanoids, methoxypyrazine and volatile sulfur compounds synthetized in grapes, which in numerous combinations make a unique, distinctive, typical varietal aroma (Coombe and McCarthy, 1997;Ebeler and Thorngate, 2009).Listed compounds can be used for varietal identification (Marais, 1983;Rapp and Mandery, 1986) because characteristic wine aroma of specific variety is attributed to the aroma compounds of grape.
The most important group of grape compounds with the greatest contribution to distinctiveness of aroma variety are terpenes (Marais, 1983;Câmara et al., 2007) to which belongs groups of monoterpene and C 13 norisoprenoids (Mateo and Jiménez, 2000).The biggest impact on wine primary aroma provide monoterpenes that are present in wine in free form, while they are present in grape in both free and bound glycoside forms.Monoterpens have very important contribution on white wines aroma of Muscat varieties, but also on other aromatic varieties (Mateo and Jiménez, 2000;del Caro et al., 2012).Free monoterpenes are present in less aromatic and non-aromatic varieties in significantly lower concentration (Iyer et al., 2010;Genovese et al., 2013).Aroma precursor's analysis has been used as a strategy to determine the aroma potential of grape both from aromatic and non-aromatic varieties (Loscos et al., 2009).Monoterpenes have strong impact on wine flavor character that is verified by strong correlation of linalool and α-terpineol content and floral description of wine (Komes et al., 2006;Skinkis et al., 2008;Sánchez-Palomo et al., 2012).
Aroma profile is extremely important in clonal selection procedure despite that main subject is always wine aroma profile but not grape aroma.Koch et al. (2010) highlighted a significance of aroma compound 2-methoxy-3-isobutylpyrazine in clonal selection of 'Cabernet Sauvignon' variety as characteristic compound of primary aroma, while Boidron (1995) quoteed that wine of two clones of 'Chardonnay' variety have pronounced Muscat aroma tint in comparison to other studied clones and recognizes this as positive and desirable clone characteristic.Versini et al. (1990) stated that it is for quality of clone grapes comparison necessary to study specific aroma compounds on which content in grapes the effect of environmental is at minimum, they are monoterpenes and by them were defined differences between clones of 'Traminer Red' and 'Chardonnay' variety. Marais and Rapp (1991) concluded that it is possible to distinguish clones based on terpene content.They proved that clones 457/48, 14Gm D35, 925/643 and FR46/106 of 'Gewiirztraminer' variety appeared to have a greater potential to produce aroma-rich and variety-typical wines than N20 Kieselberg.With respect to 'Weisser Riesling', two clones, namely 37 and 327, could possibly be selected as more flavorful than the others.McCarthy (1992) studied grapes of 10 clones 'Muscat à petite grains blanc' variety and found that there was no difference in the free volatile terpene concentration between clones, but there were significant differences in bounded form of monoterpene concentration.In order to assess the suitability of some genotypes for functional genomics studies on terpenol synthesis in grapevine, Duchene et al. (2009) studied two varieties differing in their aromatic pattern: 'Gewurztraminer' and 'Sauvignon Rose' and two clones of 'Chardonnay' (76 and 809).There are evidences that clonal variation, through somatic mutations, can modify the aromatic profile of fruits.Genovese et al. (2013) reported results that showed different aroma profile (free and bound volatile compounds) in 'Aglianica' and 'Uva di Troia' grapes.
Various authors are using different methods for determination of terpenes in grape and wines.Setkova et al. (2007) developed a rapid headspace solid-phase micro extraction-gas chromatographic-time-of-flight mass spectrometric method for qualitative profiling volatile fraction of wines.Volatile compounds of grapes are responsible of varietal aroma.In order to obtain an appropriate technique to study grape volatile compounds in pulp and skins of 'Muscat' grapes, Sánchez-Palomo et al., (2005) have developed HS-SPME method coupled with GC-MC.Sixteen volatile compounds have been quantified.Prosen et al. (2007) using synthetic solution developed an extraction procedure for the aroma compounds from musts and wines, using solid-phase micro extraction.The method was suitable for analyzing free aroma compounds in must of different varieties and for monitoring of their release after enzymatic or acidic hydrolysis.Coelho et al. (2007) propose headspace-solid phase micro extraction (HS-SPME) for the variety-and pre-fermentationrelated volatile compounds of 'Fernão-Pires' (FP) white grape berries.Two C 13 norisoprenoids, two aromatic alcohols, two C6 aldehydes, and three C6 alcohols were identified by gas chromatographyquadruple mass spectrometry (GC-qMS).Bordiga et al. (2013) suggest using combination of HS-SPME technique with GCxGC/TOF-MS system for the analysis of wine volatile compounds.
In a last few years, not only on Croatian but also on European wine market, there is a growing interest for wines made of local grapevine varieties with distinguish quality.These wines contribute in raising a regional wine identity and tourism potential.

Samples
Ten clone candidates of 'Škrlet bijeli' variety produced by propagation of elite vines, selected in process of mass positive clonal selection, and planted in two sites, Popovača and Repušnica, during three vintages were investigated.Both locations are in viticulture region of continental Croatia, sub-region Moslavina.Production viticulture zone is B (Winkler et al., 1974).Both field trails were planted in period from year 2001 to 2004.
Clone candidates coded as ŠK-07, ŠK-11, ŠK-29, ŠK-32, ŠK-33, ŠK-57, ŠK-60, ŠK-69, ŠK-74, ŠK-77 represent progeny of individual elite vine selected from old vineyards.All clone candidates originated from mass clonal selection of 'Škrlet bijeli' which was performed on agricultural traits (yield and sugar accumulation) as well as on good vigour of mother plants.In the field trails, each clone candidate was represented with 3-5 vines planted in same row, that were grafted, at the place, with gem originated from elite vine by method "green on green" on virus-free rootstock: Leaf Roll Virusis.(LR1 and LR3) and Raspberry Ringspot Virus (RRSV).Harvest date was determined based on phenotype evaluation and refractometry tracking of sugar accumulation dynamics and was harvested at the same date for all clone candidates at about 85 ºOe as it is common sugar content at harvest of 'Škrlet bijeli'.The basic agricultural traits (yield and sugar accumulation) are listed in Annexes 1-2.

Head-space SPME extraction
The SPME extraction conditions were 10 ml of sample that was spiked with internal standard of 3octanol in concentration of 0,0844 µg l -1 (Sigma-Aldrich, St. Louis, MO, USA) in a 20 ml glass headspace vials, with addition of 1,5 g NaCl, extraction time of 45 min and extraction temperature of 50 °C under stirring at 350 r min -1 .The headspace was sampled using a 50/30 μm divinylbenzene-carboxen-poly(dimethylsiloxane) (DVB-CAR-PDMS) coated fiber in a Supelco fiber holder (Bellefonte, PA, USA).After equilibration, the fiber was removed from the sample and the analytes were thermally desorbed in the injector port of the GC.

GC-MS
A multiPurpose autosampler MPS2 (Gerstel GmbH, Germany) with an agitator and SPME fiber conditioning station was used to extract the volatiles from sample vial headspace.All chromatographic analysis was performed using an Agilent 7890A Series GC system with an Agilent 5975C Mass Selective Detector (Agilent, Palo Alto, USA).The apparatus used was equipped with split/split less injector, J&W DB-Wax column (60 m length x 0.32 i.d.x 0.25 μm film thickness (J&W Scientific, Folsom, CA, USA).The temperature program used was 40 °C for 5 min; 4 °C min -1 to 230 °C; 20 min at maximum temperature.Carrier gas (He) flow was 1.2 ml min - 1 .Injections of 1 μl were performed in split less mode while the injector port and the ion source were maintained at 230 °C and 250 °C, respectively.Positive electron impact spectra were recorded at 70 eV in a range m/z 30 -250.Mass spectrometric information of each chromatographic peak was compared to NIST (National Institute for Standards and Technology, USA) mass spectra library.Data is given as relative peak area (RPA) ± standard deviation (SD) presented ratio of area peak of identified compound and peak area of internal standard.

Statistical analysis
Significant differences between clones of 'Šklet bijeli' were determined on the basis of the most abundant aroma compounds: linalool, βdamascenone, terpinolen, nerol and α-terpineol by one-way analysis of variance using software package Statistics (version 8.0, Statsoft Inc., Tulsa, USA), while differences between averages of RPA by Student-Newman-Keul test.Principle component analysis (PCA) and linear discriminant analysis (LDA) were performed to classify the grapevine clone candidates regarding to vineyard site and vintage.A total of 35 major aroma compounds were included in analysis.PCA was performed to provide a data structure study over a reduced dimension, covering the maximum amount of the information present in the basic data.
In order to determine if there are differences among clones on the level of free terpene compounds linalool, β-damascenone, terpinolen, nerol and α-terpineol expressed as mean RPA values for all ten clones, the belonging rank of significant differences was determined by analysis of variance.The results are presented in Table 1.

Clone
The impact of vintage and vineyard site on the RPA values of linalool, β-damascenone, terpinolen, nerol and α-terpineol in grape must of clone candidates of 'Škrlet bijeli' with the belonging rank of significant differences was determined by analysis of variance.The results were presented in Table 2 and 3.
Year in column marked by the same letter are not significantly different (p ≤ 0.05); p-value is the significance; Pr > F -This is the p-value associated with the F-statistic.It is used in testing the null hypothesis that all of the model coefficients are 0; F -This is the F-statistic is the mean square model divided by the mean square error; F exp -this is the variance between treatments devided by the variance within treatments Table 3: The influence of vineyard site on the content of linalool, β-damascenone, terpinolen, nerol and α-terpineol (mean value of RPA ± SD) with belonging rank of significant differences for 10 clone candidates of 'Škrlet bijeli', for both vineyard sites through three years.Preglednica 3: Vpliv lokacije vinograda na vsebnosti linaloola, β-damascenona, terpinolena, nerola in α-terpineola (povprečna vrednost relativne ploščine vrha ± SD) s pripadajočim rangom značilnih razlik za 10 klonskih kandidatov 'Škrlet bijeli' za obe lokaciji.It is not possible to make a real judgment on must aroma based on processing data of individual aroma compounds because different combination of compound concentration at the end bring different wine olfactory experience (Robinson, 2011;Botelho, 2008).To determine if there is stable correlation between clones, vintage years and location and all 35 detected aroma compounds, PCA method was used.
The other five parameters isoamyl acetate, acetaldehyde, Z-linalool oxide, ethyl acetate and limonene also contribute significantly to better separation among the samples.When the LDA was applied to the data (48 samples, 22 variables), three discriminate functions explained 80 % of the total variance.Function 1 explains 43.8 % of the total variance, function 2 explains 23.4 %.The scores of the samples and parameters for these first two functions are plotted on Figure 1.As it can be seen, the samples are well separated depending on vineyard site and clone candidates.The accuracy of the placement of each sample into 20 groups was 100 %.From Figure 1 it can be seen, that only two clone candidates (ŠK-69 and ŠK-33) on two vineyard sites (Popovača and Repušnica) do not show differences in analyzed parameters.

Influence of vintage year on aroma compounds in vineyard site Popovača
Using LDA method, four parameters were selected for vintage year of grape production and clone candidate as the most discriminating variables: myrcene and α-phelandrene, as well as transocimene and cis-ocimene.The other ten parameters: (2-hexene-1-ol (E), α-terpinene, acetaldehyde, 2-hexenal, 3-methyl-1-butanol, 2methyl-1-butanol, isoamyl acetate, hexanal, ethyl acetate and geranic oxide also contribute significantly to better separation among the samples.When the LDA was applied to the data (24 samples, 22 variables), three discriminate functions explained 84.6 % of the total variance.Function 1 explains 48.0 % of the total variance, function 2 explains 26.1 % and function 3 10.5 %.The scores of the samples and parameters for these first two functions are plotted on Figure 2. As it can be seen, the samples are well separated depending on grape production year.The accuracy of the placement of each sample into 10 groups was 1007%.Clone candidates' ŠK-60, ŠK-69, ŠK-77, ŠK-29, ŠK-11 and ŠK-07 during three or two vintage years did not show differences in analyzed parameters.

Influence of vintage year on aroma compounds in vineyard site Repušnica
Using LDA method, six parameters were selected for vintage year of grape production and clone candidate as the most discriminating variables: 3methyl-1-butanol, ethyl acetate and trans-ocimene, as well as 2-methyl-1-butanol, α-phelandrene and α-terpinene.The other eight parameters: 2-methyl-1-butanol, acetaldehyde, hexanal, geranic oxid, isomyl acetate, myrcene, 2-hexanal and hexyl acetate also contribute significantly to better separation among the samples.When the LDA was applied to the data (24 samples, 22 variables), two discriminant functions explained 82.7 % of the total variance.Function 1 explains 60.8 % of the total variance, function 2 explains 21.9 %.The scores of the samples and parameters for these first two functions are plotted on Figure 3.As it can be seen, the samples from site Repušnica are well separated depending on vintage year.The accuracy of the placement of each sample into 10 groups was 100 %.Clones ŠK-32, ŠK-57, ŠK-77 and ŠK-29 during three or two production years did not show differences in analyzed parameters.

CONCLUSION
Analyzing free volatile terpene compounds responsible for primary aroma compounds responsible for flavour of 'Škrlet bijeli' such as linalool, terpinolen, nerol, α-terpineol and βdamascenone were detected, while contrary to expectation, monoterpene geraniol was not detected.Differences of RPA values for the first four compounds were not significant among clone candidates.However, remarkable differences of RPA values among clone candidates were established for some less represented compounds, as eg.norisoprenoid compound β-damascenone.It is noteworthy that significantly higher RPA values for all five dominant aroma compounds were established in the 2008 vintage, which indicates favourable climatological conditions for their synthesis and they can be used as quantitate indicator for prediction of wine aroma intensity.At the same time, RPA values for all five the most abundant aroma compounds were not significantly influenced by vineyard site and therefore could not effectively discriminated among them.
Meanwhile, other aroma compounds were identified (trans-ocimene, 2-methyl-1-butanol, myrcene, α-phelandrene, cis-ocimene and 3methyl-1-butanol) that noticeably less participate in total flavour description, but they still enable notable clone candidates discrimination using LDA method according to the individual compounds (only within individual vineyard site).Results of these aroma compounds showed that influence of vineyard site (soil, climate, fertilization and other) was dominant over clone genetic potential when it is placed in other environment (or ecological) conditions, that their RPA values for individual clones and their order (rank) were not consistent on different location.However, within individual site, clones were through three very different vintages retained their mutual relations regarding to aroma compounds synthesis and it was possible to differentiate them.
Analysis of must aroma compounds enabled positive discrimination of two clones ŠK-32 and ŠK-57 in comparison to others.It is necessary to initiate additional comparative study on must and further the wines of these clones in order to find answer what is must aroma profile analysis can help in clonal selection procedure.

Figure 3 :
Figure 3: Projection of the scores of the samples (right) and parameters (left) for 24 must samples from vineyard site Repušnica depending on vintage and clone candidates used in the plane defined by the two standardized canonical discriminant function coefficients.Slika 3: Projekcija rezultatov vzorcev (desno) in parametrov (levo) za 24 vzorcev mošta iz lokacije vinograda Repušnica v odvisnosti od letnika trgatve in klonskih kandidatov v ravnini, ki ju določata standardizirani funkciji diskriminantnih koeficientov.If compared both vineyard sites it can be concluded that only two clones ŠK-77 and ŠK-29 during investigated vintage years did not show differences in analyzed parameters.
Yield (kg of grape per stock) of clone candidates in investigated vintages and vineyard sites Priloga 1: Pridelek (kg grozdja po trsu) klonskih kandidatov v proučevanih letnikih in lokacijah Newman-Keul test; values in column marked by the same letter are not significantly different (p ≤ 0.05) Neither linalool, with the highest RPA values which ranged from 32705 to 98760, nor terpinolen (RPA ranged from 1522 to 9573), nerol (RPP ranged from 869 to 3168) nor α-terpineol (RPP ranged from 754 to 2983) content were not statisticaly significantly different among clone candidates.RPA values of norisoprenoid compound β-damascenone were statistically different and clone candidats ŠK-29 and ŠK-69 had lower RPA values then clones ŠK-57 and ŠK- Keul test; values in column marked by the same letter are not significantly different (p ≤ 0.05); p-value -this is the significance; Pr > F -This is the pvalue associated with the F-statistic.It is used in testing the null hypothesis that all of the model coefficients are 0; F -This is the F-statistic is the Mean Square Model divided by the Mean Square Error; F exp -this is the variance between treatments/variance within treatments