Determination of major volatile compounds during the production of fruit vinegars by static headspace gas chromatography–mass spectrometry method

Depósito de investigación de la Universidad de Sevilla https://idus.us.es/ “This is an Accepted Manuscript of an article published by Elsevier in FOOD RESEARCH INTERNATIONAL on 2011, available at: https://doi.org/10.1016/j.foodres.2010.10.025.” 1 TITLE: DETERMINATION OF MAJOR VOLATILE COMPOUNDS DURING THE 1 PRODUCTION OF FRUIT VINEGARS BY STATIC HEADSPACE GAS 2 CHROMATOGRAPHY-MASS SPECTROMETRY METHOD 3 AUTHORS: C. Ubeda1, R.M. Callejón1, C. Hidalgo2, M.J. Torija2, A. Mas2, A.M. 4 Troncoso1, M.L. Morales1* 5 ADDRESSES: 6 1Área de Nutrición y Bromatología. Facultad de Farmacia. Universidad de Sevilla. 7 C/ P. García González nº2, E- 41012. Sevilla, Spain 8 2Departamento de Bioquímica y Biotecnología. Facultad de Enología. Universitat 9 Rovira i Virgili. C/ Marcel·lí Domingo s/n. E- 43007. Tarragona, Spain. 10 *Corresponding author: e-mail: [email protected]; Tel.: 34-954-556760; Fax: 34-954- 11 233765 12 ABSTRACT 13 A static headspace gas chromatography coupled to mass spectrometry (SHS-GC-MS) 14 method was validated to determine several major volatile components during the 15 production process of fruit vinegars. The method is simple, fast, linear in the working 16 range, suitably sensitive, repeatable and reproducible, and has a good degree of 17 accuracy for most of the compounds studied. Different conditions were tested in the 18 production process of vinegars by means of double fermentation. The addition of SO2 19 and pectolytic enzymes produced a considerable increase in methanol and acetaldehyde, 20 especially in strawberry purees, whereas pressing led to a loss of these volatile 21 compounds. In the alcoholic fermentation of persimmon and strawberry purees, the 22 Saccharomyces cerevisiae strain used had a great influence on the production of 23 acetaldehyde and higher alcohols in wines. Considering the influence of these studied 24 compounds in the final profile of the vinegars, our results showed that the 25 2 Saccharomyces cerevisiae strain isolated in this study produced the most suitable wine 26 substrates for the production of vinegars. Moreover, semisolid fruit substrate provides 27 better results than liquid substrate. Inoculated acetification in wood recipients yielded 28 vinegars with a better volatile profile, as these contained higher levels of most 29 compounds except acetaldehyde. 30 KEYWORDS: Volatile compounds; persimmon; strawberry; vinegar; wine; SHS-GC- 31 MS 32 1. Introduction 33 Vinegar is one of the most widespread and common products in the world because it is 34 available in every country in several different varieties (Mazza & Murooka, 2009). The 35 traditional use and integration of vinegars in numerous cultures can be traced back to 36 ancient times. Today, the most widely marketed vinegar is wine vinegar, although 37 vinegar can be produced from a variety of very different raw materials. 38 In today’s market, there is a growing demand for fruit vinegar sold as a health food 39 product (Ou & Chang, 2009). This consumer trend has led to the development of new 40 products with the aim of expanding the range of vinegars available on the market. 41 Furthermore, the production of these vinegars provides a use for surpluses of second 42 quality fruit. 43 Different quality parameters should be studied in selecting the best production 44 procedure for new fruit vinegars. Such parameters should include volatile compounds 45 responsible for aroma and close attention should be paid to which of these compounds 46 might be influenced by the production process. 47 Aroma is certainly one of the most important determinants of food quality and 48 acceptance. The particular aroma of vinegar is the result of high quantities of volatile 49 compounds. These compounds may come from the raw material or may be formed 50 3 during the production process. Different authors have pointed out the importance of the 51 production process in the final aroma of vinegars and therefore in their organoleptic 52 qualities (Morales et al., 2001; Callejón et al., 2009). Moreover, the content of several 53 major volatile compounds found in vinegar such as methanol is restricted by Spanish 54 legislation (<1g/L) (Presidencia del Gobierno, 1993). 55 Gas chromatography coupled to a mass spectrometry detector is widely used in the 56 study of volatile compounds. To analyse these constituents in a liquid sample, the 57 sample is introduced into a gas chromatograph, the volatile components are evaporated, 58 and their vapour is carried through the column by the mobile phase (Ettre, 2002). 59 However, the non-volatile matrix remains in the injector, thereby contaminating it. 60 Researching volatile components present in a solid sample is even more complicated. 61 This type of sample obviously cannot be introduced into an instrument; it requires an 62 elaborate sample preparation procedure that includes extracting the volatile components, 63 among other steps (Ettre, 2002). 64 Headspace is a fast, simple, efficient and environmentally friendly sampling method 65 used with capillary GC for the analysis of volatile fractions in many food samples. 66 Headspace (HS) is essentially a sampling method that permits analysts to take an aliquot 67 of the gas phase in equilibrium with a liquid or solid phase (Ettre, 2002). During static 68 HS analysis, equilibrium between the sample and the headspace above is achieved, and 69 a fraction of this headspace gas phase is withdrawn for GC analysis (Bylaite & Meyer, 70 2006). In equilibrium, the distribution of the analytes between the two phases depends 71 on their partition coefficients. The composition of the original sample can therefore be 72 established from the analytical results of this aliquot (Ettre, 2002). 73 Static HS-GC works well with high precision and accuracy for liquid samples since 74 calibration can be performed easily by either external or standard addition without any 75 4 serious problems (Li et al., 2009). With static headspace sampling, sample headspace 76 volatiles are automatically brought directly to the GC, thus offering good validation as 77 well as the possibility for a high number of samples to be processed (Sriseadka et al., 78 2006). The main disadvantage of static HS-GC compared to dynamic HS-GC is its 79 relatively low sensitivity (Snow & Slack, 2002). However, sensitivity can be increased 80 by salting-out, pH control or increasing the equilibration temperature during sample 81 heating (B’Hymer, 2003). Static headspace GC is mostly useful for applications in the 82 high-ppb to percent concentration ranges (Wang at al., 2008). In the headspace analysis, 83 parameters such as temperature and equilibrium time, headspace volume and 84 instrumental conditions must be carefully standardized (Arisseto and Toledo, 2008). 85 The overall goal of this work was to develop and to optimize a simple and fast method 86 based on GC-MS to monitor the evolution of major volatile compounds in the 87 production process of fruit vinegars. Firstly, to monitor changes in these compounds a 88 sampling method had to be selected that was suitable for all three products studied: raw 89 material (fruit puree), fruit wine and fruit vinegar1, which all have very different 90 consistencies. We decided to test headspace sampling. Next, we optimized the static 91 headspace sampling and injection conditions. Finally, the method was successfully 92 applied to determine the major volatile compounds in these kinds of matrices. 93 2. Materials and methods 94 2.1. Chemicals and reagents 95 All the chemicals used were analytical-reagent grade and provided from the following 96 sources: acetaldehyde, methyl acetate, methanol, ethyl acetate, 1-propanol, isobutanol, 97 isoamyl acetate, 2-methyl-1-butanol, 3-methyl-1-butanol, ethanol, acetic acid and 4- 98 methyl-2-pentanol (IS) from Merck (Darmstadt, Germany); sodium chloride from 99 5 Sigma–Aldrich (Madrid, Spain); and water from a Milli-Q purification system 100 (Millipore, Bedford, USA). 101 2.2. Standards and sample preparation 102 6 g of sample saturated in sodium chloride (2 g) and 10 L of internal standard (391 ug 103 kg-1) were placed into a 20 mL HS vial and sealed immediately with a white 104 silica/PETF lined septum and aluminium crimp cap (VWR International Eurolab S.L., 105 Barcelona, Spain) and then placed in the autosampler tray for HS sampling. 106 A standard mix was used to establish the best injection volume. A dearomatised fruit 107 puree spiked with standards was used to select sample incubation temperature and time. 108 Fruit was dearomatised as follows: 5 mL of dichloromethane were added to 20 g of fruit 109 puree. This mixture was stirred with a stir bar over night, and then was centrifuged at 110 4,000 rpm for 10 min and the dichloromethane was withdrawn. This procedure was 111 repeated. To eliminate remains of dichloromethane, the puree was submitted to a 112 nitrogen stream for 20 min. After this, 5 mL of acetone were added and the mixture was 113 stirred for three hours, followed by centrifugation (4,000 rpm for 10 min), the solvent 114 was withdrawn and a nitrogen stream was subsequently applied for 20 min. We spiked a 115 commercial fruit puree and vinegar with the analytes for repeatability, intermediate 116 precision and recovery assays. 117 2.3. Vinegars production and samples studied 118 Fruit processing and pre-treatment was performed as follows: fruit was crushed with a 119 beater; 60 mg L-1 of sulphur dioxide were added to prevent the growth of undesirable 120 micro-organisms; 15 mg L-1 of each of two kinds of pectolytic enzymes (Depectil extra- 121 garde FCE® and Depectil clarification® from Martin Vialatte Oenologie, Epernay, 122 France), were then added to the puree. 50 g L-1 and 75 g L-1 of sucrose were also added 123 to 2008 and 2009 strawberry puree respectively to ensure an appropriate final acidity in 124 6 the resulting vinegar. Samples of fruit puree were taken before and after the addition. 125 One portion of 2008 strawberry fruit puree was pressed to study the effect of two types 126 of starting substrates (semisolid and liquid) (Table 1). 127 The alcoholic fermentation of the fruit substrate was similar in persimmons and 2008 128 strawberries and slight modifications were made in the case of 2009 strawberries. 6 L of 129 fruit puree was distributed into various glass recipients: six for persimmons, eight for 130 2008 strawberries (four of purees and two of liquid substrate) and eight for 2009 131 strawberries. These recipients were then divided into two groups: half of them were 132 inoculated with the oenological yeast Saccharomyces cerevisiae QA23 at a 133 concentration of 2x106cells mL-1, and spontaneous alcoholic fermentation was allowed 134 to take place in the other half. The inoculated fermentation in the 2009 strawberries was 135 performed with the yeast strain Saccharomyces cerevisiae RP1, isolated during the 136 spontaneous alcoholic fermentation of the 2008 strawberry puree. 137 Acetification was carried out in glass vessels by spontaneous processes except for 138 strawberry wines from the 2009 harvest. These wines were acetified in three different 139 containers: a glass vessel, and oak and cherry wood barrels. Each of them was filled 140 with 5.5 L of wine. All the wine obtained from inoculated alcoholic fermentation was 141 mixed and dispensed in the abovementioned recipients and inoculated with acetic acid 142 bacteria. The wines from spontaneous alcoholic fermentation were processed in the 143 same way and acetified spontaneously. 144 All vinegars obtained in 2007 and 2008 were pressed. Additionally two different final 145 treatments were applied to strawberry vinegars from the 2008 harvest: some were 146 centrifuged and others pasteurized. Strawberry vinegars from 2009 were only 147 pasteurized. The 2007 persimmon vinegars presented an average acetic degree between 148 4.4 (from inoculated wines) and 4.5 (from spontaneous wines). The acetic acid contents 149 7 average in 2008 strawberry vinegars were 4.8 (from spontaneous wines) and 4.9 (from 150 inoculated wines). Finally, inoculated vinegars from 2009 harvest reached an acetic 151 degree of 5.5 (glass vessel), 6.6 (oak barrel) and 6.3 (cherry barrel). 152 Furthermore, part of the puree from the 2009 strawberries was concentrated by heating 153 to test another form of increasing the sugar content and prevent having to add it in; the 154 resulting product was a cooked must (Table 1). One litter of this substrate was 155 fermented by a spontaneous process and 1 L by inoculating it with RP1 strain yeast. 156 Finally, the inoculated wines were acetified by adding the selected acetic acid bacteria 157 and the spontaneous wines were left to acetify spontaneously. 158 Different samples were taken throughout these production processes and a total of 53 159 samples were analysed: 6 fruit purees and 1 liquid substrate, 22 wines and 24 vinegars. 160 All the samples were stored in 30 mL amber glass flasks at -20°C until the analysis. The 161 codes and characteristics of the samples are shown in Table 1. 162 2.4. Optimization of static headspace conditions and method validation 163 Several headspace conditions were optimized: spit ratio, injection volume, time and 164 temperature of incubation. Different split ratios (2, 5, 10, 15, 20 and 40) and injection 165 volumes (250 and 350 uL) were tested. 166 We studied different incubation times (10, 20, 30 and 40 min) and temperatures (55, 65, 167 75 and 85ºC). A sample of commercial fruit puree was spiked with all the compounds 168 studied for these trials. The quantities added were roughly 25 mg kg-1 except for ethyl 169 acetate, which was 150 mg kg-1. 170 The method was validated with respect to linearity, sensitivity (LOQ), precision 171 (repeatability and intermediate precision) and accuracy. 172 8 The quantification limits were obtained injecting successive dilutions of standards and 173 were calculated as the concentration which would result in a signal-to-noise ratio higher 174 than or equal to 10. These values were determined for liquid and semisolid matrices. 175 Repeatability and intermediate precision were checked using a dearomatized 176 commercial fruit puree and vinegar spiked with the analytes. These spiked samples were 177 injected six times in a single day for the repeatability assay and three times a day on six 178 different days for the intermediate precision assay. The results, expressed as relative 179 standard deviation (%RSD). 180 The accuracy of the method was evaluated only in the case of vinegar since the 181 calibration lines were built using hydroacetic solutions instead of a real matrix. A 182 commercial vinegar was spiked with standards at three levels of concentration. 183 184 2.5. Static headspace GC-MS instrumentation and conditions 185 Analyses were conducted using an Agilent 6890 GC system coupled to an Agilent 186 5975inert quadrupole mass spectrometer and equipped with a Gerstel MP2 headspace 187 autosampler (Müllheim an der Ruhr, Germany). 188 Static headspace equilibration was performed at 65ºC for 20 min, while a low shaking at 189 250 rpm was applied during sample heating. 350 L of headspace gas were injected 190 using a heated (85ºC) gastight syringe (1 mL) in split mode 10:1. The split/splitless inlet 191 temperature was 200ºC. Syringe injection speed was 50 L s-1. 192 Separation was performed on a CPWax-57CB column (50m×0.25mm, 0.20m film 193 thickness, Varian, Middelburg, The Netherlands). The carrier gas was He at a constant 194 flow rate of 1mL/min. The column oven temperature was initially set at 35ºC for 5 min, 195 and then was increased to135ºC at 4ºC min-1 and then at 10ºC min-1 to 200ºC and held 196 for 5 min. 197 15 was isolated from 2008 spontaneous wines. Therefore, the strain involved in the 347 fermentation process has a strong influence on the end levels of these compounds in 348 wines (Torrea et al., 2003; Ribereau-Gayon et al., 2006). 349 Methanol levels increased in persimmon and 2008 strawberry during alcoholic 350 fermentation, although these differences were only statistically significant in the case of 351 persimmon. Methanol is a non-fermentative alcohol; therefore, the only source of this 352 compound during alcohol fermentation is the hydrolysis of pectins. In these reactions, 353 ester bonds between galacturonic acid and methanol are cleaved, releasing this alcohol 354 into the medium, which is carried out by pectin esterases (Fernandez-Gonzalez et al., 355 2005). Several authors have shown that some Saccharomyces cerevisiae strains have 356 pectin-esterase activity (Pretrorius and Van der Westhuisen, 1991; Gainvors et al., 357 1994; Fernandez-Gonzalez et al., 2005). Thus, the increase in methanol in this 358 fermentative stage may have come from two possible hydrolytic pathways: due to the 359 pectin esterase activity of the yeast and/or to the pectolytic enzymes added to the 360 substrate that continued to act. 361 Acetaldehyde is a secondary product of yeast alcoholic fermentation; it is produced 362 during the first days of fermentation (Bosso and Guaita, 2008). This aldehyde increased 363 in persimmon case, being slightly higher in inoculated fermentations than in 364 spontaneous fermentations, although the changes were not statistically significant. 365 Meanwhile, in strawberry alcoholic fermentation acetaldehyde values decreased, 366 especially in spontaneous fermentation. Strawberries are rich in anthocyanins, which are 367 responsible for the berry’s red colour. In the production of red wines, these compounds 368 undergo condensation reactions in which different molecules are linked by acetaldehyde 369 bridges (Bosso and Guaita, 2008). These reactions involve a loss of this aldehyde. 370 These types of reactions could explain the diminution of acetaldehyde in strawberry 371 16 wine production. Opposing trends were found in terms of the final amount of this 372 compound in strawberry wines depending on the year of harvest. In 2008, strawberry 373 wines from inoculated fermentation, “inoculated wines”, were found to have higher 374 values than “spontaneous wines”. However in 2009 strawberry wines, the highest 375 results for acetaldehyde were found in spontaneous wines. As mentioned above, the 376 yeast strain employed for the production of 2009 inoculated wines was the same as that 377 used for 2008 spontaneous wines. Furthermore, these 2008 spontaneous wines and 2009 378 inoculated wines presented similar values for this compound. The influence of the 379 Saccharomyces cerevisiae strain on the differing production of acetaldehyde has been 380 reported by several authors (Antonelli et al., 1999; Regodon et al. 2006). 381 Among the esters studied, the most abundant in our fruit wines was ethyl acetate 382 followed by methyl acetate and isoamyl acetate, this last related to a fruity aroma. 383 Ethyl acetate is the most prevalent ester in grape wines (Ribereau-Gayon et al., 2006). 384 In persimmon puree, the concentration of this ester was below the quantification limit; 385 however the wines presented extremely high levels compared to the normal values in 386 grape wines (30-110 mg kg-1, Regodon et al., 2006). In 2008 strawberry, ethyl acetate 387 underwent a slight increase only during alcoholic fermentation in the inoculated wines. 388 Although the starting concentrations in 2009 wines were very low (below the 389 quantification limit), the wines obtained through spontaneous fermentation presented 390 high concentrations (633-761 mg kg-1) while in those obtained through inoculated 391 fermentation this compound was not detected. Several authors have shown that the 392 formation of esters during alcoholic fermentation is closely related to the enzymatic 393 activity of the yeast strain (Barre et al., 2000). In keeping with this, we observed that 394 this compound was not produced in the 2009 inoculated process and it was only 395 17 produced in one case in 2008 spontaneous wines (Table 5-6). The ester isoamyl acetate 396 increased in all cases studied. 397 Methyl acetate is formed by the condensation of methanol and acetic acid. We found 398 that during the alcoholic fermentation of persimmon the amount of this ester doubled. 399 This is consistent with the high levels of methanol found in persimmon substrate. 400 This compound remained practically unchanged in strawberry wine production except 401 in the case of the 2009 spontaneous process, in which the levels of methyl acetate 402 concentration increased. Finally, all compounds were found to have increased in the 403 alcoholic fermentation of strawberry liquid substrate. Figuring among the most 404 outstanding changes, we might mention a considerable increase (up to 70%) in 405 acetaldehyde, higher alcohols and isoamyl acetate. The liquid substrate was fermented 406 in the absence of solid colorants so the binding reaction between acetaldehyde and 407 monomeric anthocyanins did not frequently occur. This is a likely explanation for why 408 levels of this aldehyde were found to increase in wines from this substrate. Furthermore, 409 the largest increase in acetaldehyde occurred in inoculated alcoholic fermentation. We 410 observed the same behaviour for higher alcohols as in the fermentation of semisolid 411 substrate, showing the highest contents of 1-propanol in inoculated wines and the other 412 three higher alcohols in spontaneous wines. These results again indicate the relevance of 413 the yeast strain in the production of higher alcohols. 414 Comparing the final content of the volatile compounds analysed in wines from different 415 substrates (liquid and semisolid), it is clear that methanol and 1-propanol reached higher 416 values in liquid wines than in wines from semisolid substrate. Wines from liquid 417 resulted in lower values of methyl and ethyl acetate than wines from the other type of 418 substrate. 419 Acetic fermentation 420 18 In the acetic fermentation of persimmon wine, levels of acetaldehyde increased in most 421 cases. In 2008 strawberry vinegar, concentrations of this compound increased in all 422 cases. The transformation of ethanol to acetic acid takes place in two steps, with 423 acetaldehyde being the intermediary product. These reactions can be performed by 424 acetic acid bacteria as well as by chemical oxidation. When performed by a micro- 425 organism, each step is catalyzed by different enzymes (alcohol deshidrogenase and 426 aldehyde deshydrogenase, respectively). In chemical oxidation, the step from 427 acetaldehyde to acetic acid depends on the presence of oxygen (Ribereau-Gayon et al., 428 2006). 429 The acetification process in samples from the 2009 harvest was carried out in different 430 containers (glass vessels, cherry and oak wood barrels). In the vinegar from glass 431 vessels, we noticed a remarkable amount of acetaldehyde together with lower levels of 432 ethanol and acetic acid than in vinegar produced in wood barrels. The main difference 433 between these kinds of recipients is the better oxygen transference that occurs through 434 wood pores. This might suggest that ethanol is being transformed into acetaldehyde 435 while the second reaction is not taking place at a similar rate, probably due to the lower 436 proportion of oxygen in the glass vessel. This result coincides with that reported by 437 other authors on the accumulation of this aldehyde during acetification due to oxygen 438 impoverishment (Polo and Sanchez-Luengo, 1991). Acetaldehyde tends to accumulate 439 under low oxygen conditions instead of being oxidized to acetic acid (Zoecklein et al., 440 1995). Furthermore, we have observed increases in acetaldehyde in previous studies 441 during glass bottle aging of red vinegars in which acetification and aging processes took 442 place simultaneously (Callejon et al., 2010). And during accelerated aging in glass 443 vessels with wood chips we observed an increase in acetaldehyde due to the chemical 444 oxidation of ethanol (Tesfaye et al., 2004). Although these studies prove that the 445 19 accumulation of acetaldehyde in vinegars can take place by means of the two 446 abovementioned pathways (microbiological or chemical oxidation), in our case, 447 microbiological transformation is the most likely cause of the accumulation of this 448 compound. 449 The samples from cherry wood barrels had higher concentrations of acetaldehyde than 450 those from oak barrels, regardless of the type of acetification. This compound may be 451 released into the liquid medium from this type of wood, as this phenomenon has been 452 observed in white wine vinegars aged in different kinds of wood (oak, cherry, chestnut 453 and acacia) (Callejon et al., 2010). 454 A loss of higher alcohols occurred during the acetification stage. Callejon et al. (2009) 455 showed that acetic acid bacteria consume other alcohols apart from ethanol, with 3- 456 methyl-1-butanol being the most frequently consumed followed by isobutanol and 2- 457 methyl-1-butanol, in keeping with the abundance order in the substrate. In our case, a 458 similar behaviour was observed, and in agreement with these authors, the pattern of 459 higher alcohols consumption varied depending on the abundance of these alcohols in 460 the starting wines. In other words, the higher the concentration of the alcohol, the more 461 it was consumed. 462 The 2009 strawberry wines were divided into two groups: one underwent spontaneous 463 fermentation and the other was inoculated with acetic acid bacteria. In the inoculated 464 processes the vinegars reached 6ºAc while spontaneous processes they only reached 465 4ºAc as a consequence of the unexpected halt of the acetification process. Therefore, in 466 terms of the changes in higher alcohols, the consumption of these compounds was more 467 pronounced in vinegars produced using selected acetic acid bacteria. 468 Although the consumption of methanol by acetic acid bacteria has not been previously 469 reported, the acetification process implied a decrease in this alcohol. Generally, these 470 20 micro-organisms have a defence mechanism that transforms alcohols into less toxic 471 products such esters. Persimmon vinegars showed a reduction in the concentration, with 472 about 150 mg kg-1, and a similar diminution was observed for 2008 strawberry samples. 473 In the 2009 acetification processes, spontaneous fermentation produced a larger 474 decrease in methanol than did inoculated fermentation and this difference was more 475 pronounced in samples produced in glass vessels. The concentration of methanol in all 476 final products was below the legal level allowed for vinegars (Presidencia del Gobierno, 477 1993). 478 On the other hand, methanol is involved in the synthesis of methyl esters, in this case, 479 especially of methyl acetate. We observed higher levels of methyl acetate in persimmon 480 vinegars, and as in alcoholic fermentation, during the production process the content of 481 this ester doubled. In 2008 strawberry samples, acetic fermentation produced significant 482 increases in this compound. However, these condensation reactions alone are not 483 sufficient to explain the diminution of methanol mentioned above. 484 In samples from the 2009 harvest, both strawberry vinegars produced in glass vessels 485 experienced a similar decrease in methyl acetate. However, an increase in methyl 486 acetate was found in the vinegar produced in wood barrels, with slightly higher levels 487 recorded in the case of oak barrels, which may be due to concentration phenomena. 488 Furthermore, we might point out a considerable increase in inoculated processes in 489 barrels. In general, despite the different evolutions observed, the final concentrations of 490 methyl acetate in vinegars were correlated with initial concentrations of methanol 491 (r=0.7). 492 Different trends were found in levels of ethyl acetate, a characteristic compound of 493 vinegar, which were especially conditioned by the fruit substrate used. In persimmon, 494 the concentrations of this ester in the resulting vinegars were similar to those in wines 495 21 and no clear tendency was observed (Table 4). In 2008 strawberry vinegars, ethyl 496 acetate reached more than twice the concentration of that in wines. From the 2009 497 harvest, the vinegars obtained through inoculated acetification showed values between 498 83 for glass vessels and 663-682 for the others. This indicates a considerable formation 499 along with a slight concentration of this compound in wood recipients. The results of 500 the spontaneous acetifications in the 2009 samples were the opposite because a 501 hydrolysis of ethyl acetate was taking place. This behaviour has been observed by 502 several authors who have shown that the active consumption of ethanol by acetic acid 503 bacteria induces the hydrolysis of most ethyl esters (Callejon et al., 2009). 504 Isoamyl acetate usually increases during surface acetification processes, however, in our 505 vinegars in most cases it was found to diminish. This might be explained again by a 506 hydrolysis reaction due to the consumption of alcohol 3-methyl-1-butanol by acetic acid 507 bacteria. 508 Comparing the two final treatments applied to the 2008 strawberry vinegars, 509 pasteurization and centrifugation, no statistically significant differences in the volatile 510 compounds studied between them were found (Table 5). 511 Special vinegars were also produced for this study which used cooked strawberry must 512 (Table 6). Only inoculated acetifications we obtained final products. The main 513 difference in these heated strawberry vinegars was the high levels of acetaldehyde 514 compared to vinegars obtained from uncooked strawberry fruit puree. These high levels 515 would adversely affect the organoleptic properties of the end product. 516 3.4. Principal component analysis 517 The compounds studied underwent a series of changes during the production of the 518 vinegars. Several principal component analyses were performed to evaluate whether 519 these changes were great enough to distinguish the different samples obtained 520 22 throughout the production process based on substrate, production stage or production 521 method. In the case of persimmons, the PCA allowed us to separate the samples into 522 three groups: the substrate, wines and vinegars, with the first three components 523 accounting for 93.9% of the variance. Similar results were obtained when the PCA was 524 applied to the 2008 strawberry sample data. However, in the products obtained from the 525 2009 harvest the separation was not so clear. 526 Moreover, this analysis was applied to the data of the strawberry puree substrates to 527 study the influence of the addition of enzymes and SO2. Each sample appears in a 528 different quadrant in the plan of the two principal components. The PC1 is able to 529 separate the substrates depending on the harvest and the PC2 separates the samples with 530 and without treatment. 531 PCA of strawberry wines from 2008 harvest reveals that substrate pressing affects more 532 than the inoculation. This is deducted from the samples separation into the plan of two 533 first PC. The liquid wines inoculated and spontaneous appear in the same quadrant 534 whist the group of wines from inoculated semisolid substrate are separated in different 535 quadrant from the spontaneous group. 536 On the other hand, the result of this analysis on the data obtained from all the wine 537 samples showed that the principal three components explained 92.6% of the variance. 538 Data scores and variable loadings are plotted simultaneously into the plan made up of 539 the first two principal components in Figure 3. This figure shows that the samples are 540 distributed into three groups. The figure shows that PC2 successfully separates the 2008 541 strawberry spontaneous and 2009 inoculated wines from the other strawberry wines. 542 Thus, the wines obtained through the use of the same yeast strain appear together in the 543 same quadrant. This reinforces the theory that the yeast strain has a strong influence on 544 these compounds of the aromatic profile. We confirmed a high degree of association 545 23 between strawberry wines inoculated with the RP1 strain and the production of higher 546 alcohols such as 2-methyl and 3-methyl-1-butanol and isobutanol. Moreover, if we 547 consider only the persimmon and 2008 strawberry wines, the PCA revealed that PC1 548 allows us differentiates between persimmon wines and strawberry wines and PC2 549 distinguishes between inoculated and spontaneous wines. PC1 was positively correlated 550 with acetaldehyde, the three acetates and methanol, and PC2 was positively correlated 551 with acetaldehyde, isoamyl acetate and propanol. In the analysis of the final vinegars, 552 the score plot obtained by selecting the first two PCs as axes showed that the samples 553 were distributed in three groups, one formed by persimmon vinegars, another which 554 included 2008 strawberry vinegars and 2009 strawberry vinegars produced in a glass 555 vessel, and a third group, very far from the previous ones, comprised of the 2009 556 strawberry vinegars produced in barrels. This shows the importance of the type of 557 recipient in which the acetification is carried out on the final content of these 558 compounds. 559 Conclusions 560 The headspace sampling method proposed has proved to be a valuable methodology for 561 the determination of major volatile compounds during the production process of fruit 562 vinegars. From a practical point of view, this method does not require any complicated 563 sample preparation. The validation of the method was satisfactory, recovery values and 564 limits detection are acceptable for most of the compounds studied, and the method was 565 successfully applied to real samples. 566 The addition of SO2 and pectolytic enzymes produced a considerable increase in 567 methanol and acetaldehyde, especially in the strawberry samples. However, pressing led 568 to a loss of these volatile compounds. In alcoholic fermentation, the Saccharomyces 569 cerevisiae strain used had a great influence on the production of acetaldehyde and 570 24 higher alcohols in wines. Taking into account the influence of these compounds studied 571 in the final profile of vinegar, the results show that the Saccharomyces cerevisiae strain 572 isolated in this study produces the most suitable wine substrates for the production of 573 vinegars. Moreover, the use of semisolid fruit substrate provides better results than the 574 use of a liquid substrate. 575 In terms of acetic fermentation, inoculated acetifications in wood recipients resulted in 576 vinegars with better volatile profiles as these presented higher levels of most 577 compounds except acetaldehyde. 578 Acknowledgments 579 This research was made possible thanks to financial support from the Spanish 580 Government by means of a predoctoral grant and the research project AGL2007-66417- 581 C02-01 funded by the Ministry of Science and Innovation. Moreover, the researchers 582 are grateful to the enterprises Hudisa S.A., Agromedina and Grupo Alconeras for 583 providing the fruit used in vinegar production. 584 References 585 Antonelli, A., Castellari, L., Zambonelli, C., & Carnacini, A. (1999). Yeast Influence on 586 Volatile Composition of Wines. Journal of Agricultural and Food Chemistry, 47(3), 587 1139–1144. 588 AOAC (1993) Peer verified methods program. Manual on policies and procedures. 589 Ariseto, A. P., & Toledo, M. C. F. (2008). Furano: um contaminante fromado durante o 590 processamento de alimentos. Brazilian Journal of Food Technology, 11(1), 1-11. 591 Barre, P., Blondin, B., Dequin, S., Feuillat, M., Sablayrolles, J. M., & Salmon, J. M. 592 (2000). La levadura de fermentación alcoholica. In C. Flanzy Ed., Enología: 593 Fundamentos científicos y tecnológicos (pp. 274-322). Madrid: AMV Editions. 594 31 Table 1. Treatment and codex of samples. 727 728 Fruit and harvest Treatment Puree Sample Treatment Substrate sample Alcoholic fermentation Wine sample Acetification Treatment o Recipient Vinegar sample Persimmon 2007 Crushed K7Z1 SO2 Pectolytic enzymes K7Z2 Inoculated K7WI1-K7WI3 Spontaneous Pressing Centrifugation K7VE1- K7VE3 Spontaneous K7WE1-K7WE3 Spontaneous Pressing Centrifugation K7VI1- K7VI3 Strawberry 2008 Crushed F8P1 SO2 Pectolytic enzymes sucrose F8P2 Inoculated F8WI1- F8WI3 Spontaneous Centrifugation F8SVI1C-F8SVI2C Pasteurization F8SVI1P-F8SVI2P Spontaneous F8WE1- F8WE3 Centrifugation F8SVE1C-F8SVE2C Pasteurization F8SVE1P-F8SVE2P - F8P2 Pressing F8L Inoculated F8LWI - - - Spontaneous F8LWE Strawberry 2009 Crushed F9P1 SO2 Pectolytic enzymes sucrose F9P2 Inoculated F9WI1- F9WI4 Inoculated glass vessel F9SVIG oak barrel F9SVIO cherry barrel F9SVIX Spontaneous F9WE1- F9WE4 Spontaneous glass vessel F9SVEG oak barrel F9SVEO cherry barrel F9SVEX - Heating Concentrated F9MC Inoculated - Inoculated glass vessel F9MCVI1-F9MCVI2 32 Spontaneous - Spontaneous glass vessel - 729 Table 2. Analytical characteristics of the method for vinegar. 730 731 Compound Retention Time (min) m/z Linear range (mg kg-1) r2 LOQ (ug kg-1) Added (mg kg-1) Recovery (%) Mean Recovery (%) Repeatability (%RSD) Intermediate Precision (%RSD) Acetaldehyde 3.95 44 1-200 0.998 0.30 37 50 62 65.1 67.0 72.0 68.0 ± 3.6 1.88 3.80 Methyl acetate 5.01 74 2-500 0.998 0.15 15 20 25 103.5 100.0 103.5 102.3 ± 2.0 2.64 4.10 Ethyl acetate 6.03 61 74-2002 0.9995 0.095 450 600 750 82.0 70.8 68.3 73.7 ± 7.3 3.90 1.60 Methanol 6.54 31 10-700 0.9992 4.0 150 200 250 90.0 90.0 85.6 88.5 ± 2.5 1.65 2.22 Propanol 10.8 31 1-75 0.9999 0.24 3.37 4.50 5.62 90.3 91.0 85.0 88.8 ± 3.3 2.09 3.00 Isobutanol 12.7 43 1-124 0.9998 0.21 9 12 15 96.0 109.7 102.0 102.6 ± 6.9 1.54 1.97 Isoamyl acetate 13.3 55 0.57-20.5 0.9999 0.015 0.375 0.500 0.625 84.1 89.4 77.0 83.5 ± 6.2 4.92 5.20 2-Methyl-1-butanol 16.9 57 1-75 1.000 0.11 2.62 3.50 4.37 102.7 95.0 96.5 98.1 ± 4.1 0.87 2.52 33 3-Methyl-1-butanol 17.0 55 1-76 0.9993 0.13 10 14 17 99.0 108.7 117.0 108.2 ± 9.0 2.54 3.97 732 733 Table 3. Analytical characteristics of the method for wine and puree of fruit. 734 735 Compound Linear range (mg kg-1) r2 LOQ (mg kg-1) Repeatability (%RSD) Intermediate Precision (%RSD) Acetaldehyde 1-200 0.9986 4.63 4.85 5.75 Methyl acetate 0.9-170 0.9982 2.77 3.12 4.19 Ethyl acetate 61-4500 0.9960 3.1 4.24 4.30 Methanol 51-3000 0.9991 38.1 4.26 5.19 Propanol 1-200 0.9989 2.40 4.96 6.00 Isobutanol 1-200 0.9991 1.54 4.70 6.88 Isoamyl acetate 0.05-10.4 0.9989 0.17 2.86 7.08 34 2-Methyl-1-butanol 1-200 0.9989 0.27 6.93 8.15 3-Methyl-1-butanol 1-202 0.9967 0.30 0.83 5.73 736 737 738 Table 4. Changes in volatile compounds during the elaboration of persimmon vinegars. 739 740 Samples Mean concentration of compounds (mg kg-1) ± SD Acetaldehyde Methyl acetate Ethyl acetate Methanol 1-Propanol Isobutanol Isoamyl acetate 2-Methyl-1-butanol 3-Methyl-1-butanol K7Z1 10.4  0.3 9.5  0.9 n.q. 343  9 2.27  0.01 1.340  0.003 0.1177  0.0004 0.140  0.004 n.q. K7Z2 28.2  1.9a 18.1  1.7a n.q. 376  39 2.97  0.08a 1.99  0.05a 0.140  0.004 0.31  0.02a n.q. K7WE1 32.1  1.9 36.1  1.3b 1221  45b 551  17b 8.9  0.6b 15.8  1.3b 0.94  0.03b 7.69  0.25b 27.98  1.04b K7WE2 25.1  3.2 34  3b 1046  107b 554  41b 8.5  0.5b 15.6  1.3b 0.82  0.06b 8.5  0.4b 33  3b K7WE3 30.8  1.9 42.2  0.7b 1459  17b 758  18b 11.10  0.05b 20.5  0.5b 1.33  0.08b 8  1b 38  4b K7WI1 39.2  0.7b,c 38.8  0.9 1094  58b 581  40b 14.8  1.2b,c 15.3  0.9b 1.31  0.15b 10.466  0.024b,c 40.6  0.5b,c K7WI2 40.47  0.14b,c 67  5 1942  90b 695  6b 15.46  0.15b,c 16.67  0.03b 2.87  0.19b 10.93  0.03b,c 42.1  0.3b,c K7WI3 36.8  1.6b,c 47.8  0.9 1354  140b 539  74b 16  2b,c 16.3  1.7b 1.86  0.07b 9.3  0.9b,c 41  3b,c K7VE1 37  3 103  7b 1447  152b 471  42 3.07  0.07b 7.01  0.03b 1.25  0.16 5.19  0.11b 16.0  0.6b K7VE2 32.81  0.19 79.89  0.17b 1203  24b 444  31 3.42  0.14b 7.59  0.15b 0.89  0.04 5.3  0.4b 17.9  0.3b 35 K7VE3 47  3 86  6b 1278  100b 464  12 3.37  0.07b 8.17  0.13b 0.90  0.07 5.91  0.16b 17.5  0.3b K7VI1 61  4 86.10  5.03b 1094  59d 374  19b,d 4.8  0.1b,d 5.83  0.04b,d 0.9980  0.0001 4.9  0.4b 17.55  0.24b K7VI2 33.4  2.1 67  4b 921  70d 326  22b,d 4.47  0.15b,d 5.38  0.13b,d 0.89  0.03 5.14  0.07b 17.1  0.3b K7VI3 38.1  2.4 87  6b 1024  84d 385  8b,d 4.169  0.002b,d 5.11  0.12b,d 0.95  0.12 4.6  0.1b 15.3  0.3b n.q.: concentration under quantification limit. a: significant differences (p<0.05) with respect to the initial fruit puree (ANOVA) b: significant differences (p<0.05) with respect to its substrate (ANOVA) c: significant differences (p<0.05) with respect to spontaneous process (ANOVA) d: significant differences (p<0.05) with respect to the vinegars obtained from spontaneous wines (ANOVA) 741 742 Table 5. Changes in volatile compounds during the elaboration of strawberry vinegars in harvest 2008. 743 744 Samples Mean concentration of compounds (mg kg-1) ± SD Acetaldehyde Methyl acetate Ethyl acetate Methanol 1-Propanol Isobutanol Isoamyl acetate 2-Methyl-1-butanol 3-Methyl-1-butanol F8P1 9.9  0.5 6.9  0.6 96  10 190  14 4.46  0.11 11.6  0.5 0.249  0.004 3.4  0.1 15.5  0.9 F8P2 52.9  2.2a 9.30  0.11a 140  7a 292.  9a 5.97  0.14a 16.9  0.7a 0.360  0.007a 5.3  0.4a 22.5  1.2a F8L 7.2  0.3e 1.93  0.08e 53  2e 244  7e 3.51  0.09e 10.3  0.2e 0.107  0.015e 3.7  0.4e 12  9e F8LWE1 9.2  0.6 2.1  0.4f 59  9f 489  33 49.9  2.1 65.0  2.4f 0.96  0.07f 16  2f 80.9  1.8f F8LWI1 27.4  2.4c, f 2.9  0.4f 86  6 c, f 530  14f 90.3  0.4c 40.44  0.08c, f 0.678  0.024c, f 11.0  0.3c 72.0  1.1c, f F8WE1 21.8  0.9b 9.2  0.7 170  11 462  9 37  2b 83  3b 0.95  0.06b 34.0  0.3b 108.3  0.6b 36 F8WE2 19.0  0.6b 7.62  0.22 127  5 290.1  2.3 24.9  0.9b 61.2  2.3b 1.14  0.08b 24.3  1.3b 80.1  1.9b F8WE3 19.8  1.4b 8.1  0.4 129  11 303  22 26.0  1.7b 69  7b 0.969  0.021b 26.2  1.9b 92  7b F8WI1 51  5c 8.47  0.09 173  6b 305  3b 43.9  0.8b,c 33.9  0.4b,c 1.382  0.007b,c 14.4  0.3b,c 62  5b,c F8WI2 46  4c 8.9  0.5 184  11b 317  16b 44.8  1.7b,c 35  1b,c 1.50  0.12b,c 12.1  0.7b,c 64  3b,c F8WI3 52.5  1.4c 9.07  0.25 207  17b 327  28b 45  4b,c 34.3  2.4b,c 1.52  0.15b,c 12.3  1.6b,c 58.0  1.9b,c F8SVE1C 34.3  0.4b 17.7  0.5b 439  31b 259  13 4.40  0.09b 11.93  0.04b 0.610  0.023b 5.74  0.21b 13.7  0.4b F8SVE1P 40  4b 19.4  1.9b 483  47b 246  33 4.3  0.3b 11.7  0.5b 0.61  0.08b 5.1  0.3b 13.7  0.3b F8SVE2C 75.9  3.0b 13.00  0.11b 368  15b 195  12 4.16  0.23b 11.6  0.6b 0.46  0.07b 5.47  0.16b 14.2  0.6b F8SVE2P 79  4b 14.2  0.9b 374  34b 181  14 4.05  0.13b 11.4  0.3b 0.31  0.03b 5.4  0.3b 14.4  0.3b F8SVI1C 67.9  2.3b 14.0  0.5b 374  11b 174  10b 6.78  0.24b,c 5.32  0.22b,c 0.271  0.003b,c 3.0  0.4b,c 9.7  0.4b,c F8SVI1P 77.4  2.3b 16.5  0.9b 446  24b 180  5b 7.3  0.4b,c 5.8  0.3b,c 0.251  0.013b,c 2.98  0.19b,c 10.5  0.8b,c F8SVI2C 89.2  1.6b 15.50  0.19b 424.11  2.23b 179  5b 7.72  0.25b,c 5.87  0.19b,c 0.260  0.016b,c 2.93  0.03b,c 10.85  0.15b,c F8SVI2P 98  4b 18.2  0.8b 498  34b 181  15b 8.24  0.13b,c 6.3  0.3b,c 0.248  0.014b,c 2.9  0.3b,c 11.6  0.3b,c a: significant differences (p<0.05) with respect to the initial fruit puree (ANOVA) b: significant differences (p<0.05) with respect to its substrate (ANOVA) c: significant differences (p<0.05) with respect to spontaneous process (ANOVA) d: significant differences (p<0.05) with respect to the vinegars obtained from spontaneous wines (ANOVA) e: significant differences (p<0.05) with respect to F8P2 sample (ANOVA) f: significant differences (p<0.05) with respect to semisolid wines obtained with similar alcoholic process (spontaneous or inoculated) (ANOVA) 745 746 37 747 748 749 750 751 752 753 754 755 756 757 758 759 760 Table 6. Changes in volatile compounds during the elaboration of strawberry vinegars in harvest 2009. 761 762 Samples Mean concentration of compounds (mg kg-1) ± SD Acetaldehyde Methyl acetate Ethyl acetate Methanol 1-Propanol Isobutanol Isoamyl acetate 2-Methyl-1-butanol 3-Methyl-1-butanol F9P1 5.4  0.1 3.21  0.05 n.q. 159  12 2.34  0.02 1.371  0.011 0.118  0.001 0.211  0.002 n.q. F9P2 95  8a 4.6  0.4a n.q. 293  31a 3.47  0.12a 2.60  0.08a 0.133  0.001 1.37  0.07a 3.9  0.3a F9WE1 65.1  0.7b 11.5  0.3b 639  11b 237  10b 14.4  0.5b 25.0  0.7b 2.94  0.04b 16.5  0.5b 48.8  0.9b F9WE2 55.1  0.6b 12.4  0.1b 761  8b 254  11b 15.0  0.5b 26.0  0.8b 2.85  0.04 b 13.4  0.4b 48.6  1.3b F9WE3 44  1b 11.3  0.4b 667  31b 239  8b 14.4  0.4b 25.0  0.4b 2.66  0.18b 15.19  0.11b 47.6  0.7b 38 F9WE4 49  4b 11.1  0.7b 633  47b 222  3b 13.6  0.3b 23.9  0.3b 2.55  0.11b 12.69  0.15b 46.0  0.5b F9WI1 23.6  1.3b,c 4.72  0.07c n.q. 303  4 12.81  0.22b,c 69.7  0.5b,c 2.64  0.06b 52.7  1.3b,c 171  7b,c F9WI2 25.1  1.9b,c 4.45  0.15c n.q. 279  16 12.05  0.22b,c 67.6  1.1b,c 2.60  0.17b 42.4  0.8b,c 167  10b,c F9WI3 23.2  1.3b,c 4.02  0.12c n.q. 235  5 11.1  0.1b,c 59.4  0.7b,c 1.98  0.06b 39  3b,c 152  5b,c F9WI4 20.0  0.6b,c 4.52  0.03c n.q. 277  12 11.9  0.5b,c 67.2  2.4b,c 2.72  0.08b 44  3b,c 173  11b,c F9SVEG 1.43  0.07 7.0  0.5 45  5 120  1 0.71  0.01 1.569  0.022 n.q. 2.111  0.003 2.739  0.004 F9SVEO 23.6  0.6 16.2  0.5 148  5 165.6  0.4 1.16  0.01 3.036  0.012 0.065  0.007 2.914  0.008 5.64  0.07 F9SVEX 63.15  0.11 14.22  0.02 439  17 198.2  1.1 2.001  0.003 5.176  0.014 0.158  0.014 4.67  0.07 9.5  0.3 F9SVIG 129  5 3.4  0.3 83  5 146.7  0.9 1.493  0.024 11.5  0.3 0.27  0.04 8.81  0.22 27.0  0.7 F9SVIO 42  3 20.4  1.4 682  41 276  5 2.364  0.012 24.7  0.9 1.4  0.1 21.2  0.6 47.5  0.06 F9SVIX 64.4  1.0 17.3  1.1 663  5 278  16 2.82  0.09 26.2  0.8 1.282  0.023 23.3  1.0 52.1  0.4 F9MCVI1 719  58 22.8  2.1 341  17 318  15 11.3  0.4 9.9  0.6 0.57  0.03 9.1  0.6 43  3 F9MCVI2 410  17 25.4  2.1 452  39 370  27 15.1  0.8 11.3  0.4 0.65  0.05 11.3  1.0 48.9  1.9 n.q.: concentration under quantification limits a: significant differences (p<0.05) with respect to the initial fruit puree (ANOVA) b: significant differences (p<0.05) with respect to its substrate (ANOVA) c: significant differences (p<0.05) with respect to spontaneous process (ANOVA) d: significant differences (p<0.05) with respect to the vinegars obtained from spontaneous wines (ANOVA) 763 764 765 39 766 767 768 769 770 771