Abstract
Various commercial tomato juices with different flavors are available at markets worldwide. To clarify the marker compounds related to the flavor characteristics of tomato juice, we analyzed 15 pure commercial tomato juices by a combination of volatile profiling and sensory evaluation. The correlations among volatiles and the relationship between volatiles and sensory descriptors were elucidated by multivariate analyses. Consequently, the tomato juices made from fresh market tomatoes (including the popular Japanese tomato variety “Momotaro”) were clearly separated from other juices made from processing tomatoes, by both the volatile composition and sensory profiles. cis-3-Hexenol, hexanal, and apocarotenoids negatively contributed to the juices from fresh market tomatoes, whereas Strecker aldehydes and furfural showed positive contributions to the juices. Accordingly, the sensory characteristics of juices from fresh market tomatoes were related to cooked and fruity flavors but not to green or fresh notes.
Graphical abstract
Combined analysis by instrumental and sensory evaluation of various commercial pure tomato juices elucidated the difference of their flavor characteristics.
Tomato juice is a popular dietary vegetable drink with high nutritional value and a characteristic flavor. The flavor quality of commercial tomato juice is considered to be determined by the procedures used for the tomatoes’ processingCitation1,2) and the varieties of tomato materials.
The processing of commercial juice from tomato fruits generally includes a homogenizing process and thermal treatments such as blanching, pasteurization, and sterilization.Citation3) Enzymatic reactions during the process of homogenizing enhance the production of various aroma compounds. The C6 aldehydes and alcohols (hexanal, cis-3-hexenal, trans-2-hexenal, hexanol, cis-3-hexenol, and trans-2-hexenol) are the most important volatile compounds in fresh tomato fruit according to quantitative and qualitative analyses.Citation4–7) These compounds are produced by the breakdown of fatty acids via lipoxygenase and hydroperoxide lyase during homogenizing.Citation8–10) In a 2016 study, trans-2-hexenal was revealed to be produced from cis-3-hexenal by isomerases.Citation11) In addition, the enzymatic hydrolysis of glycosides as aroma precursors in tomato fruits is well known to release aglycon alcohol or phenol volatiles.Citation12,13) The composition of some aroma compounds can thus be under the control of homogenizing and the resting time of crushed tomatoes. Thermal treatment is generally known to accelerate the production of volatiles related to the Maillard reaction and the thermal decomposition of carotenoids.Citation14) Indeed, the thermal treatment of tomato products has been reported to increase the content of compounds such as furfural, dimethyl sulfide, and methional,Citation15–17) as well as decreasing C6 volatiles and compounds derived from carotenoids.Citation16,18) Thus, the volatile profile of tomato juice is likely to be variable and dependent on the processing procedure used.
The tomato variety may also influence the flavor characteristics of tomato juices. Tomato fruits have been classified as fresh market tomatoes and processing tomatoes. The use of fresh market tomatoes is preferred for fresh food dishes such as salad and salsa sauce, whereas processing tomatoes are popular for such cooked food dishes as tomato sauce and tomato soup. In Japan, various fresh market tomatoes are available and preferred, particularly the domestic tomato variety “Momotaro,” which is one of the most popular fresh market tomato varieties. The juices made from fresh market tomatoes are thus expected to have good fresh tomato flavor compared to the juices made from other materials. However, the flavor characteristics of juices made from fresh market tomatoes are unclear, and it has not been known how these characteristics differ from those of juices made from processing tomato varieties.
The combined analysis of the chemical composition and sensory properties is a promising approach for elucidating the factors that affect the flavor characteristics of food. Clarifying the direct relationships between the aroma/flavor of tomato juice and the responsible volatile compounds may be valuable for tomato juice evaluation, for producing the ideal tomato juice, and for predicting the quality of the product. The evaluation of tomato juice based on its chemical composition has been investigated previously,Citation19–23) but the combined analysis of chemical compounds and sensory properties has not been widely investigated. Vallverdu-Queralt et al. described the relationship between sensory properties (although not focused on aroma/flavor properties) and volatile compounds using commercial tomato juices obtained at Italian and Spanish markets.Citation24) An insufficient amount of other information is available about the specific direct correlations between volatile compounds and sensory descriptors focusing on the aroma/flavor properties for pure tomato juices. The identification of the responsible compounds in relation to flavor characteristics of tomato juices is significant for setting up desirable processing conditions and for the selection of suitable tomato materials.
In this study, we obtained various commercially available pure tomato juices including those from fresh market tomatoes, and subjected them to volatile profiling using gas chromatography-mass spectrometry (GC-MS). We also clarified the relationship between the sensory data related to aroma and flavor obtained by quantitative descriptive analysis (QDA) with the volatile profiles. For an efficient characterization of various tomato juices, we subjected the instrumental and sensory data to a multivariate data analysis, i.e. a principal components analysis (PCA), orthogonal partial least squares (OPLS), and two-way orthogonal partial least squares (O2PLS), and we determined the relationships among sensory attributes of tomato juices based on their chemical compositions.
Materials and methods
Materials
Fifteen pure tomato juices without salt or other additives were tested, comprising 11 commercially available juices from stores in Japan and four trial products processed in a laboratory. All of the tomato juices were directly packed after the processing of fresh tomato fruits, and not reduced from their concentrates. The basic features of the samples are listed in Table . Three of the juices (samples F, J and K) were made from fresh market tomatoes, and the other 12 juices were made from processing tomatoes. All four of the trial products (samples L–O) were processed under identical conditions, but the tomato varieties used were different. All of the samples were identical to those reported in a previous studyCitation25 and were stored unopened at 4 °C until the analyses. However, the detailed processing procedures including the homogenization and sterilization conditions for each commercial sample were unclear, because this information was not disclosed by the juices’ manufacturers.
Table 1. Characteristics of the tomato juice samples used in this study.
Headspace solid phase micro-extraction (HS-SPME)
The volatiles present in each juice were extracted by headspace solid phase micro-extraction (HS-SPME). A tomato juice sample (2 mL) was placed in a 20-mL vial, and CaCl2 (1.5 g) was added to saturation in the sample before the sample vial was sealed with PTFE/silicone septa and sonicated for 3 min. Before the volatiles were extracted, the sample was pre-incubated for 10 min at 50 °C with rotation at 250 rpm.
DVB/CAR/PDMS 50/30 fibers (2 cm, Supelco, Bellefonte, CA, USA) were exposed to the headspace of the sample for 20 min at 50 °C. In preliminary experiments, we confirmed that extraction at 50 °C was the most effective method for extracting without thermal influence. Each sample prepared by HS-SPME was injected automatically using an MPSII instrument for GC-MS analysis (Gerstel, Mülheim an der Ruhr, Germany). The volatile analysis was performed in triplicate for each tomato juice product. The consistency of the sensitivity for peak detection in successive data was confirmed by the fixed quality control sample prepared from the tomato juice samples.
GC-MS analysis
The GC-MS analysis was performed using an Agilent Technologies 6890N GC coupled to an Agilent Technologies 5973N mass spectrometer (Agilent Technologies, Santa Clara, CA). The volatiles absorbed on the SPME fiber were loaded into the GC system in the splitless mode and desorbed at the injection port (250 °C) for 5 min. The volatiles were separated using a DB-5 MS (30 m × 0.25 mm i.d., 250-μm thickness, Agilent J&W Scientific fused silica capillary column (Agilent Technologies).
The oven temperature was increased from 40 °C (5-min hold time) to 180 °C at a rate of 5 °C min−1 and to 250 °C at a rate of 15 °C min−1, before holding for 5 min. Helium was used as the carrier gas at a flow rate of 1 mL min−1. Mass spectra were obtained under the following conditions: ionization voltage, 70 eV (EI); ion source temperature, 230 °C; quadrupole temperature, 150 °C; and mass range, m/z 33–350. For peak detection, we set the scan speed as 3.15 scans/s.
Instrumental data processing
First, we attempted to normalize all of the peak data based on the area ratio of 3-nonanol (0.15 μg per sample), which was added to each sample as an internal standard. However, the release of 3-nonanol into the headspace was affected greatly by the viscosity of each tomato juice. We therefore analyzed a common standard 3-nonanol solution and a fixed quality control sample at intervals in the sequence analysis, and we confirmed that the response of them was stable throughout the experiment. Thus, we used the data without normalization in the following analytical steps. ChemStation software (Agilent Technologies) was used for the GC-MS data acquisition. The raw data were exported to the MetAlign program for peak detection and peak alignment.Citation26) First, ion peak-top heights of all m/z ions derived from all detected compounds were collected. The signal/noise level for peak selection was set to 2.0. Second, an alignment of m/z ion peaks among multiple sample data was performed. We set the limit of scan gap numbers for peak alignment as within 10 scans at the first scan number and 15 scans at the end scan number. This indicates that m/z ions detected almost within 10–15 scan gaps among multiple samples are aligned as the same peak.
The convergence of m/z ions for a single compound and identification were performed using the AIoutput program.Citation27) Here, all m/z ions collected by the Metalign program were aligned based on the retention time and the reconstructed mass spectra per detected compound. Subsequently, each compound was identified based on its retention index and mass spectral similarity compared with authentic standard compounds or databases (Aroma Office with NIST library, Nishikawa Keisoku, Tokyo). Tentative identification was performed based only on the NIST library search. The peak area of the base peak ion for each compound was used for the multivariate analysis.
Sensory data analysis by QDA
The sensory analysis data of tomato juices derived from the same sample lot with instrumental flavor analysis were obtained and reported in a recent study.Citation25) The detailed QDA procedure using trained panels and the data obtained are also described there.Citation25) The sensory data for aroma and flavor were differently obtained. Evaluations for aroma were performed by smelling the tomato juice samples. Each flavor was evaluated by swallowing a tomato juice sample (5 mL) in the mouth. The QDA attributes for the tomato juice aromas and flavors were determined using 18 and 16 descriptors, respectively (Table ). The QDA scores for each descriptor were subjected to a multivariate analysis.
Table 2. Compounds frequently detected as high VIP scores (>1.0) and their contribution to each sensory descriptor, lycopene content and Brix.
Multivariate analysis of instrumental data and QDA results
The PCA, OPLS, and O2PLS based on the GC-MS data were performed using SIMCA 13 software (Umetrics, Umeå, Sweden). The peak intensities of the detected compounds and the scores for sensory evaluation were converted by Pareto scaling or Unit Variance scaling and applied to the multivariate analysis. A hierarchical clustering analysis was performed using z-scores of the volatile compounds for each sample.
Results and discussion
Volatile profiles of tomato juices
The GC-MS analysis of the 15 tomato juice samples detected 160 compounds, among which 89 compounds were identified or tentatively identified. The other 71 compounds were detected as peaks, but they did not match any compounds in the library search. In all samples, the most common compounds were hexanal, cis-3-hexenol, and 6-methyl-5-hepten-2-one, which are well known to be abundant in tomato volatiles.Citation15)
To characterize the volatile profiles of the samples, we performed a PCA using the GC-MS data-set. First, the PCA results from 160 compounds including unknown compounds were compared with those from 89 tentatively identified compounds. The score plots and loading plots showed very similar patterns, indicating that the amounts of unknown compounds did not significantly influence the PCA. We therefore focused on the results from 89 tentatively identified compounds (Fig. (A) and (B)).
Fig. 1. Principal components analysis of volatile compounds. Score plot (A) and loading plot (B) for the detected volatile compounds are shown.
The score plot by Pareto scaling showed that most of the samples could be separated based on PC1 (38.6% of the variance, Fig. (A)). The samples separated into negative fields for PC1 (F, G, H, I, J, and K) contained the juices of fresh market tomatoes as their base material (F, J, and K) (Table ). PC2 explained 21.0% of the variance, and it separated samples L, M, N, and O into the fourth quadrant. All of these samples were produced from different materials but using the same processing conditions. Nevertheless, they had similar GC-MS profiles, indicating that the volatile profile of tomato juice is appreciably affected by the processing conditions as well as by the tomato materials employed.
The loading plot shown in Fig. (B) indicates the compounds that contributed to the characterization of each sample. For the samples in the first quadrant (A and E), ethyl acetate, cis-3-hexenol, hexanol, ethanol, 3-methyl-1-nitrobutane, and 2-phenylethanol made positive contributions. In particular, it should be noted that sample E was separated from the others based on its high ethyl acetate content, which was more than three times that found in the other samples (Fig. (A)). Ethyl acetate is a small volatile molecule that can be detected in fresh tomato fruit and tomato products.Citation28) It was reported that the ethyl acetate content is not affected by tomato juice treatments, i.e. cold breaking or hot breaking,Citation17) which suggests that it differed among the samples due to the properties of the materials used.
Fig. 2. Relative peak intensities of ethyl acetate (A), dimethyl sulfide (B), cis-3-hexenol (C) and furfural (D) among 15 tomato juices.
cis-3-Hexenol was one of the most variable compounds among the samples (Fig. (B)). For example, samples F–K, which were plotted as negative scores for PC1, contained little cis-3-hexenol as a volatile. In tomato fruit, cis-3-hexenol is produced from cis-3-hexenal by alcohol dehydrogenase,Citation29) and cis-3-hexenal is well known to be one of the main compounds responsible for the “green note,” together with hexanal, hexanol, and cis-3-hexenol. Thus, cis-3-hexenal has been reported as a potent aroma compound in tomato fruit.Citation29) However, cis-3-hexenal was not detected at significant levels in any of the tomato juice samples in the present study, despite their high cis-3-hexenol contents. This indicates that cis-3-hexenal is either converted into cis-3-hexenol by alcohol dehydrogenaseCitation29) or trans-2-hexenal by isomeraseCitation11) during tomato processing, or that it is released or degraded during the heating process. Indeed, the cis-3-hexenal content was previously reported as zero or at a low level in tomato juices and tomato products.Citation15,17,24)
In the second quadrant in Fig. (B), dimethyl sulfide made the highest contribution, although its content was considerable in all samples (Fig. (C)). Dimethyl sulfide has been detected in processed tomato products such as tomato juice, canned tomato, and tomato paste, which are derived from thermally treated tomatoes.Citation15,30) Furfural was also plotted in the second quadrant in the present study. The content of furfural was higher in samples F, H, and J (Fig. (D)) and low in samples L, M, N, and O. The samples in the fourth quadrant (L, M, N, and O) had major contributions from 6-methyl-5-hepten-2-one, 1-nitro-2-phenylethane, hexanal, phenylacetaldehyde, and geranylacetone.
To investigate the correlation among compounds and samples, we performed a hierarchical cluster analysis using the z-scores of each volatile compound among the samples (Fig. ). The clusters grouped by samples showed patterns similar to the classifications in the PCA. The samples were mainly clustered in four groups (I–IV in Fig. ), although three samples from B were placed in different groups. Group I, composed of F, H, J, and K, was basically divergent from the other samples. Most of these samples were shown to be made from fresh market tomato varieties, indicating that the juices of fresh market tomato varieties have characteristic flavor profiles. This group was positively related to the clusters composed of thermally produced volatiles such as Strecker aldehydes, furfural, and 2-acetylfuran.
Fig. 3. Hierarchical clustering of tomato juices.
Group IV derived from samples L, M, N, and O was positively correlated with the clusters composed of apocarotenoid volatiles such as 6-methyl-5-hepten-2-one, geranylacetone, and β-ionone. The peak areas of 6-methyl-5-hepten-2-one and geranylacetone were strongly correlated with each other in all of the tomato juices (R2 = 0.87, Fig. (A)). In addition, we performed correlation analyses of the lycopene contents and the contents of 6-methyl-5-hepten-2-one and geranylacetone for all of the juice samples. It is known that 6-methyl-5-hepten-2-one and geranylacetone are generated from lycopene via degradation by carotenoid cleavage dioxygenase (CCD, Fig. (D)).Citation31,32) The coefficient of determination for each relative to the lycopene contents was 0.52 and 0.33, respectively, indicating their contents could be affected by the lycopene content (Fig. (B) and (C)).
Fig. 4. Correlation between apocarotenoid volatiles and lycopene content, and the pathway of synthesis of apocarotenoids from lycopene by carotenoid cleavage dioxygenase in tomato.
The relationships between the volatile and sensory profiles
We investigated the relationships between the volatile and sensory profiles to identify the significant volatile compounds responsible for specific descriptors. First, a correlation analysis was performed using the PC1 values from the PCA analyses of both the volatile compositions and the QDA profiles. Spearman’s correlation coefficient was –0.880, which indicated that the values obtained from the volatile and QDA profiles were negatively correlated. The PC1 scores obtained from the PCAs of the volatile composition (x-axis) and the QDA analysis (y-axis) for all samples are plotted in Fig. . Some samples were exceptional, but most of the samples were plotted close to the approximate line (R2 = 0.774). These results suggest that the volatile profiles were related to the sensory descriptors.
Fig. 5. Correlation analysis of the PC1 values for the volatile compounds (x-axis) and sensory descriptors (y-axis) determined for the tomato juice samples.
We next performed an OPLS regression with each of the 34 descriptors, resulting in good linearity (R2 > 0.9) between the measured and predicted values (Table ). In particular, “fresh tomato aroma,” “whole processed tomato aroma,” “grated carrot aroma,” and “herb aroma” had the highest predictive value (R2 = 0.992). We selected the compounds that contributed to each sensory descriptor using the VIP (variable importance in the projection) values and loading scores. Generally, explanatory variables indicating VIP values as >1.0 are significantly recognized as more important for the prediction of modeling.Citation33)
We filtered volatiles expressing VIP > 1.0 for each descriptor, and evaluated those using loading scores whether they contribute positively or negatively (Suppl. Table 2). Tens of multiple volatiles were filtered for all descriptors, suggesting that each aroma/flavor descriptor is not explainable by the composition of specific compounds. Among them, the volatiles frequently selected for several descriptors are listed in Table . cis-3-Hexenol and 6-methyl-5-hepten-2-one were detected frequently with higher scores as positive contributors compared to the other compounds. They made positive contributions to descriptors that express freshness, such as “fresh tomato aroma/flavor,” “fresh grass aroma/flavor,” “fresh cut grass aroma/flavor,” and “fresh tomato aftertaste flavor.” In contrast, “whole processed tomato aroma/flavor,” which is defined as cooked tomato, was detected as a negative contributor. Therefore, the contents of these compounds should be strong indicators of fresh aroma production in tomato juice.
The characteristics of tomato juice made from fresh tomato varieties
Both the PCA and the hierarchical clustering separated the juices of fresh tomato varieties from those of processing varieties. To investigate their flavor characteristics in both the instrumental and sensory profiles, we performed an O2PLS analysis using both the GC-MS and QDA data.Citation34) O2PLS can bidirectionally predict variations between explanatory variables (X-variables) and objective variables (Y-variables).Citation35) We set 89 volatile compounds as X-variables, and we set 34 QDA descriptors, lycopene content and Brix as Y-variables. Unit variance scaling was applied to normalize differences in the intensity of the volatiles and QDA scores. The result of the O2PLS analysis demonstrated that 66.7% of the variation in the volatile profiles could explain 90.0% of the variation in sensory profiles.
Fig. shows the score plot values for the samples (A), the loading QDA descriptors (B), and the plot for volatile compounds (C) from the O2PLS analysis. The values of the samples, QDA descriptors, and volatiles were expressed in the same direction, indicating their positive correlations. In the score plot (Fig. (A)), the juices from fresh market tomatoes (F, J and K) were clearly separated from the other samples. Samples F and K are made from var. Momotaro, and sample J is from another fresh market tomato variety. The aroma/flavor characteristics of the juices from fresh market tomatoes were more cooked (“whole processed tomato aroma/flavor” and “grated carrot aroma/flavor”) and more fruity (“peach-like aroma/flavor” and “apple-like aroma/flavor”) than the others, whereas they showed less green fresh notes (“fresh tomato flavor aroma/flavor,” “fresh grass aroma/flavor,” and “fresh cut grass aroma/flavor”) (Fig. (B)).
Fig. 6. The first predictive component values under score and loading plots by O2PLS analysis using volatile profiles (X-variables) and sensory scores by QDA analysis (Y-variables).
In previous studies, C6-compounds such as hexanal and cis-3-hexenal were detected as the most common aroma compounds expressing “green note” in fresh tomato fruit based on aroma extract dilution analyses.Citation4,6,7,36) Even in the fruits of cv. Momotaro, cis-3-hexenal was reported as the most odor-active compound.Citation37) As shown in Fig. (C), however, all of these compounds were negatively correlated with the juices from fresh market tomatoes. The generation of C6-compounds is known to increase by ripening, because ripening-dependent lipoxygenase is induced.Citation8,37) Therefore, the ripening stages of fruit materials could affect the concentration of C6-compounds in the juices from fresh market tomatoes.
All apocarotenoids, including 6-methyl-5-hepten-2-one, geranylacetone, β-damascenone, and β-ionone were also negatively correlated with the juices of fresh market tomatoes. This is explainable by the negative contribution of lycopene content to the juices of fresh market tomatoes, because carotenoids are precursors of these volatiles (Fig. ). β-Damascenone and β-ionone derived from the degradation of β-carotene, are known to show low odor thresholds, i.e. 0.002 and 0.007 ppb/water, respectively. β-Damascenone was reported as one of the most significant contributing aroma compounds in tomato fruits. The negative correlation of juices from fresh market tomatoes with aroma-contributing compounds such as C6-compounds and apocarotenoids thus supports the flavor mildness of fresh market tomato juices.
In contrast, we detected aldehydes such as 2-methylbutanal, furfural, and methional as positive contributors to the juices from fresh market tomatoes. Furfural and methional were reported as typical compounds generated by Maillard reaction during the thermal processing of tomato products such as tomato paste.Citation15,30) Fresh market tomatoes showed a higher contribution of Brix as well, suggesting that this reaction occurs easily in their processing.
In the present study, we investigated the differences of flavor characteristics among commercial tomato juices by a combined analysis of instrumental and sensory evaluations. Multivariable analyses identified significant marker volatiles that can be used for characterizing commercial tomato juices, and we observed correlations among volatiles and a relationship between volatiles and sensory descriptors. Consequently, the juices from fresh market tomatoes were found to be differentiated from those from processing tomatoes based on flavor characteristics. However, more systematic experiments are needed to elucidate the detailed effects of processing and tomato varieties on the flavor characteristics of tomato juices. A recent preference mapping for tomato fruit indicated a high contribution of aroma properties, but the preferences of consumers were not homogeneous.Citation38,39) Modifying the volatile composition using a predictive model based on tomato varieties and processing procedures could help to improve the production of desirable tomato juices.
Author contributions
Y. Iijima, Y. Otagiri and A. Obata conceived and designed the experiments. Y. Iijima and Y. Iwasaki performed the instrumental experiments. Y. Otagiri, H. Otomo, and T. Sato performed the sensory evaluation. H. Tsugawa assisted with the data analysis. Y. Sekine and Y. Otagiri prepared the juice samples. Y. Iijima, Y. Otagiri and A. Obata wrote the manuscript.
Disclosure statement
The authors declare that there are no potential conflicts of interest.
Funding
This study was financially supported by the Japanese Ministry of Agriculture, Forestry, and Fisheries under the “Development of Non-destructive Technology Evaluating Various Qualities of Agricultural Products” Project.
Supplemental material
The supplemental material for this paper is available at http://dx.doi.org/10.1080/09168451.2016.1222264.
TBBB_1222264_Supplemental_Table_2.xlsx
Download MS Excel (79 KB)TBBB_1222264_Supplemental_Table_1.xlsx
Download MS Excel (47.1 KB)Notes
Abbreviations: GC-MS, gas chromatography-mass spectrometry; OPLS, orthogonal partial least squares; O2PLS, two-way orthogonal partial least squares; PCA, principal components analysis; QDA, quantitative descriptive analysis; SPME, solid-phase microextraction.
Related Research Data
References
- Jacob K, Garcia-Alonso FJ, Ros G, et al. Stability of carotenoids, phenolic compounds, ascorbic acid and antioxidant capacity of tomatoes during thermal processing. Arch. Latinoam. Nutr. 2010;60:192–198.
- Goodman CL, Fawcett S, Barringer SA. Flavor, viscosity, and color analyses of hot and cold break tomato juices. J. Food Sci. 2002;67:404–408.10.1111/jfds.2002.67.issue-1
- Gould WA. Tomato production, processing and technology. 3rd ed. Cambridge: Woodhead Publishing; 1992.10.1533/9781845696146
- Selli S, Kelebek H, Ayseli MT, et al. Characterization of the most aroma-active compounds in cherry tomato by application of the aroma extract dilution analysis. Food Chem. 2014;165:540–546.10.1016/j.foodchem.2014.05.147
- Buttery RG, Teranishi R, Ling LC. Fresh tomato aroma volatiles: a quantitative study. J. Agric. Food Chem. 1987;35:540–544.10.1021/jf00076a025
- Tandon KS, Jordan M, Goodner KL, et al. Characterization of fresh tomato aroma volatiles using GC-olfactometry. Proc. Florida State Hort. Soc. 2001;114:142–144.
- Mayer F, Takeoka GR, Buttery RG, et al. Studies on the aroma of five fresh tomato cultivars and the precursors of cis-and trans-4, 5-epoxy-(E)-2-decenals and methional. J. Agric. Food Chem. 2008;56:3749–3757.10.1021/jf0732915
- Riley JC, Willemot C, Thompson JE. Lipoxygenase and hydroperoxide lyase activities in ripening tomato fruit. Postharvest Biol. Technol. 1996;7:97–107.10.1016/0925-5214(95)00032-1
- Aguiló Aguayo I, Soliva Fortuny R, Martín Belloso O. Volatile compounds and changes in flavour-related enzymes during cold storage of high-intensity pulsed electric field- and heat-processed tomato juices. J. Sci. Food Agric. 2010;90:1597–1604.10.1002/jsfa.v90:10
- Chen G, Hackett R, Walker D, et al. Identification of a specific isoform of tomato lipoxygenase (TomloxC) involved in the generation of fatty acid-derived flavor compounds. Plant Physiol. 2004;136:2641–2651.10.1104/pp.104.041608
- Kunishima M, Yamauchi Y, Mizutani M, et al. Identification of (Z)-3:(E)-2-hexenal isomerases essential to the production of the leaf aldehyde in plants. J. Biol. Chem. 2016. doi:10.1074/jbc.M116.726687.
- Birtić S, Ginies C, Causse M, et al. Changes in volatiles and glycosides during fruit maturation of two contrasted tomato (Solanum lycopersicum) lines. J. Agric. Food Chem. 2009;57:591–598.10.1021/jf8023062
- Ortiz-Serrano P, Gil JV. Quantitation of free and glycosidically bound volatiles in and effect of glycosidase addition on three tomato varieties (Solanum lycopersicum L.). J. Agric. Food Chem. 2007;55:9170–9176.10.1021/jf0715673
- Kanasawud P, Crouzet JC. Mechanism of formation of volatile compounds by thermal degradation of carotenoids in aqueous medium. 1. beta.-Carotene degradation. J. Agric. Food Chem. 1990;38:237–243.10.1021/jf00091a052
- Buttery RG, Teranishi R, Ling LC, et al. Quantitative and sensory studies on tomato paste volatiles. J. Agric. Food Chem. 1990;38:336–340.10.1021/jf00091a074
- Servili M, Selvaggini R, Begliomini A, et al. Effect of thermal treatment in the headspace volatile compounds of tomato juice. In: Contis ET, Ho CT, Mussinan CJ, et al. editors. Food flavors: formation, analysis and packaging influences. Proceedings of the 9th International Flavor Conference The George Charalambous Memorial Symposium; Amsterdam: Elsevier; 1998. p. 315–330.
- Mirondo R, Barringer S. Improvement of flavor and viscosity in hot and cold break tomato juice and sauce by peel removal. J. Food Sci. 2015;80:S171–S179.10.1111/jfds.2015.80.issue-1
- Marković K, Vahčić N, Ganić KK, et al. Aroma volatiles of tomatoes and tomato products evaluated by solid-phase microextraction. Flavour Frag. J. 2007;22:395–400.10.1002/(ISSN)1099-1026
- Di Lecce G, Martínez-Huélamo M, Tulipani S, et al. Setup of a UHPLC-QqQ-MS method for the analysis of phenolic compounds in cherry tomatoes, tomato sauce, and tomato juice. J. Agric. Food Chem. 2013;61:8373–8380.10.1021/jf401953y
- Sánchez-Moreno C, Plaza L, de Ancos B, et al. Nutritional characterisation of commercial traditional pasteurised tomato juices: carotenoids, vitamin C and radical-scavenging capacity. Food Chem. 2006;98:749–756.10.1016/j.foodchem.2005.07.015
- Lin CH, Chen BH. Stability of carotenoids in tomato juice during storage. Food Chem. 2005;90:837–846.10.1016/j.foodchem.2004.05.031
- Seybold C, Fröhlich K, Bitsch R, et al. Changes in contents of carotenoids and vitamin E during tomato processing. J. Agric. Food Chem. 2004;52:7005–7010.10.1021/jf049169c
- Krumbein A, Peters P, Brückner B. Flavour compounds and a quantitative descriptive analysis of tomatoes (Lycopersicon esculentum Mill.) of different cultivars in short-term storage. Postharvest Biol. Technol. 2004;32:15–28.10.1016/j.postharvbio.2003.10.004
- Vallverdú-Queralt A, Bendini A, Tesini F, et al. Chemical and sensory analysis of commercial tomato juices present on the italian and spanish markets. J. Agric. Food Chem. 2013;61:1044–1050.10.1021/jf304631c
- Trivittayasil V, Tsuta M, Imamura Y, et al. Fluorescence fingerprint as an instrumental assessment of the sensory quality of tomato juices. J. Sci. Food Agric. 2016;96:1167–1174.10.1002/jsfa.2016.96.issue-4
- Lommen A. MetAlign: interface-driven, versatile metabolomics tool for hyphenated full-scan mass spectrometry data preprocessing. Anal. Chem. 2009;81:3079–3086.10.1021/ac900036d
- Tsugawa H, Tsujimoto Y, Arita M, et al. GC/MS based metabolomics: development of a data mining system for metabolite identification by using soft independent modeling of class analogy (SIMCA). BMC Bioinformatics. 2011;12:131.10.1186/1471-2105-12-131
- Chung T-Y, Hayase F, Kato H. Volatile components of ripe tomatoes and their juices, purees and pastes. Agric. Biol. Chem. 1983;47:343–351.
- Baldwin EA, Scott JW, Shewmaker CK, et al. Flavor trivia and tomato aroma: biochemistry and possible mechanisms for control of important aroma components. Hortscience. 2000;35:1013–1022.
- Miers JC. Formation of volatile sulfur compounds in processed tomato products. J. Agric. Food Chem. 1966;14:419–423.10.1021/jf60146a025
- Simkin AJ, Schwartz SH, Auldridge M, et al. The tomato carotenoid cleavage dioxygenase 1 genes contribute to the formation of the flavor volatiles β-ionone, pseudoionone, and geranylacetone. Plant J. 2004;40:882–892.10.1111/tpj.2004.40.issue-6
- Vogel JT, Tieman DM, Sims CA, et al. Carotenoid content impacts flavor acceptability in tomato (Solanum lycopersicum). J. Sci. Food Agric. 2010;90:2233–2240.10.1002/jsfa.4076
- Chong I-G, Jun C-H. Performance of some variable selection methods when multicollinearity is present. Chemometr. Intell. Lab. 2005;78:103–112.10.1016/j.chemolab.2004.12.011
- Trygg J, Wold S. O2-PLS, a two-block (X-Y) latent variable regression (LVR) method with an integral OSC filter. J. Chemometrics. 2003;17:53–64.10.1002/(ISSN)1099-128X
- Bylesjö M, Eriksson D, Kusano M, et al. Data integration in plant biology: the O2PLS method for combined modeling of transcript and metabolite data. Plant J. 2007;52:1181–1191.10.1111/tpj.2007.52.issue-6
- Krumbein A, Auerswald H. Characterization of aroma volatiles in tomatoes by sensory analyses. Nahrung. 1998;42:395–399.10.1002/(SICI)1521-3803(199812)42:06<395::AID-FOOD395>3.3.CO;2-Z
- Hayata Y, Maneerat C, Kozuka H, et al. Flavor volatile analysis of ‘house momotaro’ tomato fruit extracts at different ripening stages by Porapak Q column. J. Japan Soc. Hort. Sci. 2002;71:473–479.10.2503/jjshs.71.473
- Berna AZ, Lammertyn J, Buysens S, et al. Mapping consumer liking of tomatoes with fast aroma profiling techniques. Postharvest Biol. Technol. 2005;38:115–127.10.1016/j.postharvbio.2005.05.013
- Sinesio F, Cammareri M, Moneta E, et al. Sensory quality of fresh French and Dutch market tomatoes: a preference mapping study with Italian consumers. J. Food Sci. 2010;75:S55–S67.10.1111/jfds.2010.75.issue-1