Comparing the Flavor Characteristics of 71 Tomato (Solanum lycopersicum) Accessions in Central Shaanxi

Flavor is an important quality of mature tomato fruits. Compared with heirloom tomatoes, modern commercial tomato cultivars are considerably less flavorful. This study aimed to compare the flavor of 71 tomato accessions (8 pink cherry, PC; 11 red cherry, RC; 15 pink large-fruited, PL; and 37 red large-fruited, RL) using hedonism scores and odor activity values. Taste compounds were detected using high-performance liquid chromatography. Volatiles were detected using gas chromatography–olfactometry–mass spectrometry. The flavor of tomato accessions can be evaluated using the DTOPSIS analysis method. According to the results of DTOPSIS analysis, 71 tomato accessions can be divided into 4 classes. Tomato accessions PL11, PC4, PC2, PC8, RL35, RC6, and RC10 had better flavor; accessions PC4, PC8, RC10, RL2, and RL35 had better tomato taste; and accessions PL11, PC2, and RC6 had better tomato odor. The concentrations of total soluble solids, fructose, glucose, and citric acid were shown to positively contribute to tomato taste. Tomato odor was mainly derived from 15 volatiles, namely, 1-hexanol, (Z)-3-hexen-1-ol, hexanal, (E)-2-hexenal, (E)-2-heptenal, (E)-2-octenal, (E,E)-2,4-decadienal, (Z)-3,7-dimethyl-2,6-octadieal, 2,6,6-timethyl-1-cyclohexene-1-carboxaldehyde, (2E)-3-(3-pentyl-2-oxiranyl)acrylaldehyde, 6-methyl-5-hepten-2-one, (E)-6,10-dimetyl-5,9-undecadien-2-one, methyl salicylate, 4-allyl-2-methoxyphenol, and 2-isobutylthiazole. Significant positive correlations (P < 0.05) were detected between the compound concentrations and flavor scores. The above-mentioned compounds can be used as parameters for the evaluation of flavor characteristics and as potential targets to improve the flavor quality of tomato varieties.


INTRODUCTION
Tomato fruits are important dual-use (vegetable and fruit) products (Razifard et al., 2020). Because of their high nutritional value and various volatiles with delicious tastes and odors, tomato fruits are widely consumed worldwide (Zhu Y. et al., 2018). In 2018, tomato production reached 182.26 million tons all over the world (Food and Agriculture Organization of the United Nations [FAO], 2020). However, compared with heirloom tomatoes, modern commercial tomatoes have poor flavor, which can cause consumer dissatisfaction (Klee, 2010;Kanski et al., 2020). The flavor of tomato fruit mainly comes from soluble sugars, organic acids, amino acids, and volatile compounds. Compared with the wild or heirloom tomato varieties, modern tomato varieties have decreased many flavor compounds (fructose, glucose, citric acid, and at least 13 volatiles) throughout the process of domestication and improvement (Tomato Genome Consortium, 2012;Lin et al., 2014) because breeders serve growers, not consumers. Growers require tomato varieties with high yield, strong disease resistance, and a long shelf-life to ultimately ensure high returns (Giovannoni, 2018;Zhu G. et al., 2018;Razifard et al., 2020). A negative correlation has been observed between fruit weight and sugar concentration (Folta and Klee, 2016;Tieman et al., 2017). In order to obtain higher yield, breeders ignored the improvement of flavor quality. However, as living standards rise, consumers need not only sufficient food supplies but also nutritious, healthy, and delicious tomato fruits and are willing to pay more for them. Therefore, there is an increasing demand for the restoration of heirloom tomato flavors.
It is not supported by growers to sacrifice yield to increase sugar concentration of tomatoes (Goff and Klee, 2006). Volatiles may affect the flavor at very low concentrations, so they would be a candidate for flavor improvement without sacrificing yield. Volatile compounds are mainly derived from essential nutrients, such as fatty acids, carotenoids, and amino acids (Tieman et al., 2012), and can be responsible for tomato fruits having very different odor profiles. Volatiles can be affected by genotype, cultivation conditions, harvest-stage maturity, and postharvest treatment Zhang et al., 2016;Zhao et al., 2019), which could change the level of precursor supply, gene expression, enzyme activity, and frequency of enzyme contact with substrates (Klee and Tieman, 2013). Although previous studies have conducted more comprehensive and deeper evaluations of tomato flavor compounds, the key tomato flavor compounds that are screened differ greatly in the published literature. To further clarify the flavor compounds of tomato and their effects on tomato flavor, we analyzed the taste compounds and volatiles, selected the key flavor factors through hedonism scores and odor activity values, and compared the flavor of 71 tomato accessions to supply reference data for the cultivation of tomato varieties with excellent flavor.

Tomato Materials
The 71 tomato accessions were inbred lines (Table 1), which were screened, segregated, and fixed by our lab-the Tomato Genetic Breeding and Quality Improvement Lab of Northwest A&F University. We used 8 pink cherry (PC1-PC8), 11 red cherry (RC1-RC11), 15 pink large-fruited (PL1-PL15), and 37 red large-fruited (RL1-RL8) tomato accessions. All of them are fresh market tomatoes rather than processing tomatoes. Seedling cultivation was conducted in a specialized seedling factory in January 2019. Tomato seedlings were then planted in a standardized research greenhouse in Yangling Zone (34 • N, 108 • E, 500 m altitude) of Shaanxi Province in March 2019. Tomatoes on the third inflorescence were picked when they reached the red ripe stage (i.e., mature tomato fruits with 90% surface coloring) (Shinozaki et al., 2018) from mid-June to early July 2019. Tomato samples were required to be in consistent size, have uniform coloring, and have no deformities, cleft fruit, or rot (Cheng et al., 2020).

Soluble Solid
A small hole was made on the tomato, and a drop of tomato juice was squeezed out by hand and dropped on a PAL −1 digital refractometer (Atago Co., Ltd., Japan) to measure the concentration of soluble solid (Xu et al., 2018).  vacuum. Then, the sample was derivatized with methoxyamine hydrochloride and N-methyl-N-trimethylsilyltrifluoroacetamide (MSTFA) sequentially. After derivatization, the sample was filtrated twice by 0.22-µm microfiltration membrane for HPLC analysis. The citric acid and malic acid were also detected using an ultraviolet detector (VWD, Agilent Inc., United States). The mobile phase used to detect fructose and glucose was acetonitrile/water = 7:3 (v/v), and the mobile phase used to detect citric acid and malic acid was 0.2% metaphosphate. The flowrate was 1.67 × 10 −5 L s −1 . Column temperature was 35 • C, and injection volume was 10 −5 L.

Volatiles
Qualitative and quantitative analyses of volatile compounds were conducted using the headspace solid-phase microextraction (HS-SPME) gas chromatography-olfactometry-mass spectrometry (GC-O-MS) method (Farneti et al., 2017;Liu et al., 2019). Briefly, the samples were crushed with a homogenizer, and a 0.005-kg sample was added to 0.005 kg of anhydrous NaCl. The mix was vortexed to deactivate the tomato enzymes and filtered through a glass wool (Birtic et al., 2009). The supernatant was transferred to a 40-ml headspace bottle and added a magnetic rotor and 10 −5 L of chromatographically pure 3.284 × 10 −5 kg L −1 3nonanone (Meryer Chemical Technology Co., Ltd., Shanghai, China) were added. The volatile 3-nonanone was used as the internal standard because it is not found in tomato and is stable under normal temperature and pressure. The retention time (RT) of 3-nonanone appeared at 19.26 min. The RTs of volatiles in tomato appeared between 6.60 and 35.07 min. In the chromatogram, there are many volatiles peaks near to the peak of 3-nonanone, but the different peaks can be clearly distinguished. The recovery rate of 3-nonanone was as high as 98.423%. The headspace bottle containing the sample was placed on a magnetic stirrer (Troemner Inc., United States) for 2,400 s at 50 • C. At the same time, the volatiles were extracted using an solidphase microextraction (SPME) carboxen/polydimethylsiloxane (CAR/PDMS) fiber assembly (Product ID: 57318, 7.5 × 10 −5 m; particle size, 0.01 m length) (Supelco Inc., United States). The fiber assembly was used in conjunction with the manual injection handle 57330U. As the temperatures increases, the fiber coating begins to lose its ability to adsorb analytes. Although the SPME fiber can be used at least 50 times, we replaced it after 45 uses. The volatiles were detected using the GC-MS instrument (ISQ & TRACE ISQ) (Thermo Fisher Scientific Inc., United States), which had a polar elastic quartz capillary chromatographic column HP-INNOWAX (0.25 mm i.d., 60 m length, 0.25 µm film thickness). In order to remove the residual solvents and volatiles from the filler and promote the uniform distribution and fixation of the liquid film on the filler surface, the SPME fiber and HP-INNOWAX column must be deactivated before extraction and detection of volatiles. The deactivation methods were as follows: the SPME fiber and HP-INNOWAX column were connected to the GC-MS instrument; the helium flowrate of carrier gas was 1.67 × 10 −4 L s −1 ; the initial column temperature was 40 • C, then increased to 230 • C at a rate of 0.08 • C s −1 , and maintained for 7200 s. The GC conditions used were as follows: inlet temperature of 230 • C, over 99.999% helium carrier gas, no split injection, column flowrate of 1.67 × 10 −5 L s −1 , a split ratio of 20:1, and the splitless sample injection mode. The volatiles were desorbed for 150 s at 40 • C. The column temperature was then increased to 110 • C at a rate of 0.17 • C s −1 , increased to 230 • C at a rate of 0.10 • C s −1 , and maintained for 600 s. The olfactory detector (OP275 Pro II, GL Sciences Inc., Japan) was connected at the outlet of the capillary column of GC, and the split ratio of MS and the olfactory detector was 1:1. The odor characteristics and intensities of detected volatiles were described and evaluated by five olfactory panelists at the outlet of the olfactometer (Majcher et al., 2020). The MS condition used as follows: ion source of EI, ion energy of 70 eV, full scan mode, scan range of 35-500 m z −1 , ion source, and transmission line temperature of 230 • C.
The RT of each normal alkane was measured after mixing the standard solutions of C4-C26 normal alkanes using GC-MS instrument. The retention index (RI) is known as the Kovats index, a parameter for the qualitative analysis of unknown compounds by gas chromatography (Matyushin et al., 2020). The RI of target volatile compound is calculated according to the RTs of the two normal alkanes adjacent to the target volatile compound (Matyushin et al., 2020). The RI of each volatile compound was calculated using the following equation : where t R is the RT, and Z and Z + 1 are the numbers of carbon atoms in the normal alkanes before and after the target volatiles (x) flow out, respectively. Note: t R (z) < t R (x) < t R (z + 1) . The qualitative analyses of volatiles were compared with the standard mass spectrum of the library (NIST2011, United States) and RI (Selli et al., 2014). Volatiles were assessed using mass spectrometry Pu et al., 2020), and only those with both positive and negative matches > 800 (maximum of 1,000) were selected. The peak area normalization method was used to calculate the relative concentration of the various volatiles (Topi, 2020), as follows: where m n is the concentration (10 −9 kg L −1 ) of the volatile compound (named "n"), S n is the peak area of the volatile compound (named "n"), m t is the internal standard (3nonanone) concentration (10 −9 kg L −1 ), S t is the internal standard peak area, and m 0 is the mass (kg) of the sample. Since tomato fruits were found to comprise up to 94.52% water (Petro-Turza, 1987), the mass of 1 L tomato homogenate is about 1 kg. Volatile compounds are partitioned differently in the headspace and have different affinities for the polymer on the SPME fiber (Chambers and Koppel, 2013); therefore, the calibration curves of key volatiles were essential for quantitative analysis using the GC-MS method , i.e., the measured volatile concentrations required correction according to their calibration curves. In this study, 60 chromatographically pure standards of volatile compounds (Meryer, China Chemical Technology Co., Ltd., Shanghai, China) were added to 0.005 kg water containing 10 −5 L 3-nonanone (as the internal standard). The concentration gradient was formed by adding 0, 0.5 × 10 −5 , 1 × 10 −5 , 1.5 × 10 −5 , 2 × 10 −5 , 2.5 × 10 −5 , or 3 × 10 −5 L standards, respectively. The compounds were measured under the same GC-MS conditions. Linear regression analysis was performed using the theoretical concentrations of volatile standards and the concentrations calculated by equation (1), as measured by GC-MS. The accurate concentration of each compound was calculated according to the corresponding calibration curve ( Table 2) and Equation (2).

Sensory Evaluation
Sensory evaluations of sweetness, sourness, characteristic flavor, and overall acceptability were conducted according to published methods (Tieman et al., 2012;Vallverdu-Queralt et al., 2013;Aisala et al., 2020), with slight modification. Briefly, different tomato accessions were numbered and cut into wedges . After 7 days of training in the College of Food Science and Engineering, Northwest A&F University, the taste panels (25 male and 25 female, aged 18-60 years old) had mastered the taste evaluation methods (Tieman et al., 2012). Then, they conducted the sensory evaluations of 71 tomato accessions. To reduce the influence of visual preference on sensory evaluation, the sensory evaluators wore eye masks throughout the process. The maximum score was tentatively set at 8.00 points (Zhang K. et al., 2019). The taste panels scored the sweetness, sourness, characteristic flavor, and overall acceptability according to the taste intensity, e.g., the stronger the taste, the higher the score. After tasting each sample, the panels rinsed their mouths three times with purified water. To reduce taste fatigue, the panels conducted evaluations for 2,700-s periods (evaluate four to six tomato samples) and then took breaks of 900 s.

DTOPSIS Analysis
DTOPSIS analysis (Zhao et al., 2012) was used to evaluate the flavor of each tomato accession according to the following formula: where S i + is the distance from the desirable flavor (X + j ), S i − is the distance from the undesirable flavor (X − j ), and C i is the closeness to the ideal fruit flavor.

Statistical Analysis
All test data were recorded using WPS Office 2019. The standard deviation (SD) and coefficient of variation (CV) of the 71 tomato accessions were calculated using SPSS 22.0 . At least three biological replicates were performed for all flavor factors for each sample. The Z-score was used to standardize the data of tomato taste compounds, volatiles concentrations, and sensory evaluation scores (Ronningen et al., 2018). The significant differences in flavor among PC, PL, RC, and RL accessions were analyzed by a one-way ANOVA using SPSS 22.0. Pearson's correlation of flavor compounds and sensory evaluation was analyzed using SPSS 22.0. The heatmap plots were prepared using the Heatmapper software 1 (Sasha et al., 2016). In the calibration curve (y = kx + b), "y" represents the theoretical concentration of the labeled standard, and "x" represents the peak area ratio of a compound in the internal standard. b Odor descriptions of volatiles come from the olfactory panelists and refer to the literature (Du et al., 2015;Kreissl and Schieberle, 2017;Zhu Y. et al., 2018). c Odor thresholds of volatiles determined in water (10 −9 kg L −1 ), reference to the literature (Kreissl and Schieberle, 2017;Zhu Y. et al., 2018). d Volatiles that could be detected by an artificial olfactory system, as indicated by " * " next to the compound name. e "-" indicates the absence of data.

Analysis of the Taste Compounds of Mature Tomato Fruits
The concentrations of taste compounds from the 71 tomato accessions are shown in Table 1, Supplementary Table S1, and Figure 1. The soluble solids, fructose, glucose, citric acid concentrations, and the sugar and acid ratio in cherry tomatoes were significantly higher than that in large-fruited tomatoes. RC had the lowest malic acid concentration. The soluble solids consisted mainly of fructose, glucose, citric acid, and malic acid. The concentration of fructose was higher (by 1.61-fold) than that of glucose, and the concentration of citric acid was higher (by 1.77-fold) than that of malic acid. Among the 71 tomato accessions, the concentration of soluble solids was higher in accessions PC6, RL16, RC7, RC10, and RC9 and ranged from 3.67 to 11.43%. Fructose concentrations ranged from 910 to 2,400 mg 100 g −1 and were highest in accessions PC6, PC8, RC9, PC5, and PC2. Glucose concentrations ranged from 560 to 1,600 mg 100 g −1 and were highest in accessions RC9, RC10, RC7, PC8, and PC6. Citric acid concentrations ranged from 120 to 540 mg 100 g −1 and were highest in accessions RL28, RC4, RC11, and RC10. Malic acid concentrations ranged from 60 to 390 mg 100 g −1 and were highest in accessions RL28, RC4, RC11, and PC4. The sugar and acid ratios ranged from 2.77 to 10.89, and accessions RL27, RC7, RC9, RC2, and PL10 had the highest ratios (above 10.00). Accessions RC9, PC6, RC7, and RC10 had the highest concentrations of taste compounds.

Analysis of the Volatiles in Mature Tomato Fruits
A total of 60 volatiles were detected in this study ( Table 2 and Figure 1). The concentrations of the total volatiles ranged from 3.67 to 53.37 × 10 −6 kg L −1 (mean: 15.54 × 10 −6 kg L −1 ).
FIGURE 1 | Heat map of concentrations of flavor compounds in mature tomato fruits. a PC, PL, RC, and RL indicate pink cherry tomato, pink large-fruited tomato, red cherry tomato, and red large-fruited tomato, respectively. b TSS represents the total soluble solids. c V1-V60 represents the volatiles from top to bottom in Table 2, respectively.

Sensory Evaluation
The sensory evaluation scores are shown in Supplementary  Table S4 and Figure 2. The taste factors included sweetness, sourness, the sweetness and sourness ratio, characteristic flavor, and overall acceptability. Sweetness, sweetness and sourness ratio, characteristic flavor, and overall acceptability were the strongest in PCs and the weakest in PLs. The sweetness scores ranged from 2.00 to 8.00 (mean ± CV: 4.33 ± 0.35) and were higher for accessions PC8, PC4, and RC10. The sourness scores ranged from 1.50 to 6.00 (4.13 ± 1.10) and were lower for accessions RL34, RL30, RL18, and PC4. The sweetness and sourness ratios ranged from 0.44 to 3.56 (1.16 ± 0.47) and were higher for accessions PC4, RC9, and RL34. The characteristic flavor scores ranged from 2.00 to 7.75 (5.00 ± 0.28) and were higher for accessions PC4, PC8, and RL2. The overall acceptability scores ranged from 1.75 to 8.00 (4.97 ± 0.33) and were higher for accessions PC4, PC8, RC10, and RL2. Accession PC4 had a high sweetness and sourness ratio and high characteristic flavor and overall acceptability scores. Accessions PC8, RC10, and RL2 had high characteristic flavor and overall acceptability scores.

DTOPSIS Analysis
The DTOPSIS analysis method was used to evaluate the flavors of the different tomato accessions (Supplementary Table S4 and Figure 2). Among all the sensory factors, the sourness and irritant odor were negative indicators of flavor, while the others were positive factors. To rank accessions according to taste and odor factors, each sensory factor was simplified as a unique score . The scores and rankings of the tomato flavor evaluation [i.e., results from Equation (3)] are shown in Supplementary Table S1. The taste evaluation [C i(taste) ], odor evaluation [C i(odor) ], and comprehensive flavor evaluation [C i(overall) ] values were -2.05-4.45, -0.64-6.85, and -2.69-6.70, respectively. The 71 tomato accessions can be divided into four classes according to [C i(overall) ] from high to low, contains 10, 20, 28, and 13 tomato accessions, respectively (Table 1 and  Supplementary Table S4). Of the 71 tomato accessions assessed, the flavor scores were higher in accessions PL11, PC4, PC2, PC8, RL35, RC6, and RC10; among these, accessions PC4, PC8, RC10, RL2, and RL35 had better tomato taste, and accessions PL11, PC2, and RC6 had better tomato odor. The intensities of the green and fatty odors were the strongest in PCs and the weakest in PLs. The [C i(taste) ] and [C i(overall) ] values were significantly higher in PCs than in the other types of tomatoes.

DISCUSSION
Flavor involves both taste and odor and is perceived by the binding of taste compounds and volatiles to sensory receptors. The human taste system can detect five to seven tastes (sweet, sour, salty, bitter, hemp, umami, and koukumi) and the olfactory system can detect thousands of odors. Sweetness and sourness were the basis of tomato flavor (Baldwin et al., 2008;D'angelo et al., 2018). In a previous study, tomato fruits were found to comprise up to 94.52% water, in addition to fructose, glucose, citric acid, and malic acid concentrations accounting for 25, 22, 9, and 4% of the dry weight of tomato, respectively (Petro-Turza, 1987). In addition, this previous study reported the concentrations of fructose, glucose, citric acid, and malic acid in fresh tomato fruits to      ,464.36, 911.72, 281.50, and 158.81 mg 100 g −1 , respectively . Compared to the wild-type Solanum pimpinellifolium (Çolaka et al., 2020), tomato fruits in this study have lower glucose and fructose (decreased by 75%) and higher malic acid and citric acid (by 40-fold). The decreased sugar and increased malic acid concentrations, and resulting higher sourness, are prominent issues in modern tomato fruits. The sweetness of tomato fruits are mainly attributed to their fructose and glucose concentrations. The (E)-6,10dimetyl-5,9-undecadien-2-one, ethyl octanoate, and 2-hydroxyethyl benzoate volatiles were perceived to be sweet. Baldwin et al. (1998) also found that sweetness was closely correlated with glucose, fructose, (E)-6,10-dimetyl-5,9-undecadien-2-one, hexanal, (Z)-3-hexenal, (E)-2-hexenal, and (Z)-3-hexenol. The sweetness and the sweetness and sourness ratio were significantly higher in PCs than in the other three types tomatoes, which resulted in the overall acceptability and tomato-like flavor of PCs being perceived as more delicious. Decrease in sweetness is a prominent issue in modern tomato fruits, which can be improved by increasing the concentrations of sugars and volatiles with sweetness perception. However, sugar concentrations are reportedly negatively correlated with fruit weight (Folta and Klee, 2016). The most promising way to improve the sweetness of tomatoes is via the promotion of certain volatiles. Such investigations have great potential and are worth undertaking because (1) the concentrations of volatiles in tomato fruits are very low and can be increased greatly without affecting the yield or fruit size; there is an urgent requirement for the concentrations of consumer-preferred volatiles to be increased (Tieman et al., 2017) to improve the flavor intensity and variation in modern tomatoes.
The sensory evaluation found a low correlation (only 0.06) between the taste score and odor intensity. The concentrations of taste compounds were not significantly correlated with odor intensity, and only several volatiles were significantly correlated with the taste score. These findings were consistent with those of a previous study, in which consumers largely considered sweet and sour to be important contributions to flavor (Andersen et al., 2019). Sugar and acid are considered to be the basic compounds for fruit flavor formation (Baldwin et al., 2008;Bastias et al., 2011), while the volatiles form the characteristic flavor compounds of different fruits (Baldwin et al., 2008;Du et al., 2015). Therefore, it is not scientifically valid to judge tomato flavor on taste alone; instead, taste and odor should both be assessed to comprehensively evaluate the flavor (Pavagadhi and Swarup, 2020). The results of the DTOPSIS evaluation in the present study indicated that tomato accessions with a preferred (i.e., "better") flavor either scored high in the taste evaluation or had a strong odor.
In the present study, only the free volatiles were analyzed, but the glycoside-bonded volatiles also account for a certain proportion of volatiles in tomato fruits. Glycoside-bonded volatiles serve as reserve odors and can be hydrolyzed by enzymes and acids to free volatiles (Schwab et al., 2015). Although the taste and odors of 71 tomato accessions were evaluated based on taste compounds and volatiles, the complexity of flavor and individual differences in the sensory evaluations complicates the process of improving tomato fruits flavor. Additionally, flavor factors can interact with each other; in which case both the concentration and proportion of the compounds affect the production of "good" flavors. Comprehensive investigations are still needed to effectively improve the flavor of tomato fruits. Driven by the motivation to breed delicious tomatoes and using multidisciplinary approaches that involve biological evolution, multiple omics, flavor chemistry, psychology, sociology, etc., researchers are presently exploring the flavor compounds that are absent in modern tomatoes, analyzing the reasons for these losses, studying the mechanisms of flavor formation, determining more scientifically valid and reasonable methods to evaluate flavor, clarifying the road map of tomato flavor improvement, and cultivating some tomato varieties with excellent flavor.

DATA AVAILABILITY STATEMENT
All data generated or analyzed during this study are included in this published article and its Supplementary Material. The mass spectronomy data can be found here: doi: 10.6084/m9.figshare.12744695.

ETHICS STATEMENT
The studies involving human participants were reviewed and approved by the Northwest A&F University Institutional Review Board. The patients/participants provided their written informed consent to participate in this study.

AUTHOR CONTRIBUTIONS
YL: conceptualization, resources, supervision, writing-review and editing, and project administration. GC: conceptualization, performing the experiments, data analysis, and writing-original draft preparation. FZ: supervision and writing-review and editing. PC: methodology, date analysis, and formula analysis. LW: visualization. YS: data analysis; AE-S: writing-review and editing. All authors contributed to the article and approved the submitted version.

ACKNOWLEDGMENTS
Thanks are due to the Tomato Genetic Breeding and Quality Improvement Team of Northwest A&F University for providing the tomato material. Thanks to Yan Zhang for his suggestions on the test. Thanks also to Jing Zhang and Jing Zhao for their assistance in the use of GC-MS instruments in this experiment. Thanks to Qi Lu for her advice on the application of the formulas. Thanks to Dr. Tayeb Muhammad for correcting the English in this manuscript. We would also like to thank Editage (www. editage.cn) for the English language editing.

SUPPLEMENTARY MATERIAL
The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fpls.2020. 586834/full#supplementary-material Supplementary Table 1 | Concentrations profiles of taste compounds of mature tomato fruits. a In calibration curve (y = kx + b), "y" represents the theoretical concentration of the labeled standard, and "x" represents the peak area ratio of a compound in the internal standard. b '-' indicates that a labeled standard does not exist and, thus, a calibration curve was could not be created.
Supplementary Table 4 | Hedonism scores and odor activity values in 71 tomato accessions. a PC, PL, RC, and RL indicated pink cherry tomato, pink large-fruited tomato, red cherry tomato and red large-fruited tomato, respectively. b C i values represent the scores of the DTOPSIS analysis.