Characterization of Key Aroma Compounds and Construction of Flavor Base Module of Chinese Sweet Oranges

Sweet orange flavor, with its refreshing, joyful and attractive aroma, is favored by the majority of consumers all over the world. However, the industry terminology between flavorists for flavor evaluation is a bit vague and not intuitive for customers. Therefore, the study focused on analysis of sweet orange aroma and establishment of base module of orange flavor. The approach to the research involves screening key aroma compounds, identifying the attributes aroma and building base module of sweet orange. The notes of sweet orange flavor were determined by GC-O olfaction and sensory evaluation. 25 key aroma compounds with OAV ≥ 1 were screened and divided into eight notes: citrus, fruity, fresh, green, peely, woody, fatty, floral. Partial least squares regression (PLSR) was used to further verify the corresponding relationship between the volatile substances and notes. Terpenes, esters, aldehydes and alcohols compounds can provide these notes. Based on the notes, 8 base modules of sweet orange were built by selecting and matching aroma ingredients. Through this study, beginners could be trained according to the 8 notes of base modules and flavorists can engage in dialogue with different raw material sourcing teams or providers.


Introduction
Oranges have become more and more popular in recent years due to their preferable flavors and the nutrient value. The annual world production of oranges has been estimated to be 47.8 million tons [1]. Oranges are produced in more than 140 countries around the word, at latitudes between approximately 40 • N and 40 • S [2]; the main orange-producing countries are USA (California, Florida and Arizona), Mexico, Brazil, Spain, Italy, Israel, Australia, South Africa, Japan and China [3,4]. According to previous reports [5,6], oranges are rich in a variety of metabolically active substances, such as vitamin C, carotenoids, flavonoid, coumarin and phenolic compounds; these components are very important for human health and provide protection against harmful free radicals. Sweet oranges are mostly consumed as fresh fruit, juice, canned orange segments, and wine, and even the peel can be used in essential oils, which are widely used in the fragrance and flavor industries [7,8].
Aroma is an important character of sweet oranges and a key indicator for evaluating sweet oranges quality. Up until now, approximately 80 aroma compounds have been determined in orange

Volatile Compounds of Sweet Orange Determined by GC-MS
A total of 49 volatile compounds were identified and quantified after the GC-MS analysis of orange juice samples, by identification of retention index (RI), odor descriptors with authentic standard, as shown in Table 1. There were 35, 31, 30, 28 and 29 volatile compounds corresponding to the different orange samples, respectively. Hydrocarbons were predominant in the headspace gas of orange juice, followed by esters, aldehydes, alcohols ketones and acids. It has been reported that sweet oranges displayed higher levels of monoterpenes, especially d-limonene [18]. Our results generally agreed with that data; limonene was the most abundant monoterpene hydrocarbon tested in this study, followed by o-cymene, p-cymene, myrcene and valencene. Limonene was the main constituent of all samples. ZG samples contained the highest percentage of limonene (24.4 mg/L). ZG and MS samples contained a high percentage o-Cymene and 4-isopropyltoluene, and a high percentage of myrcene was present in ZG and YX samples. Of the esters, a large amount of ethyl butyrate was present in QY. In addition, the ZG and QY samples contained a large amount of acetaldehyde; additionally, ZG and MS samples also contained a large amount of linalool. Other studies revealed that linalool was the most abundant alcohols in Archi citrus juice [21].

Characterizations of Odor-Active Compounds by GC-O Analysis
The results of the olfactometry analysis are summarized in Table 2. Ethyl butyrate, myrcene, limonene, p-cymene, and linalool were the major odor-active compounds according to their high aroma intensity; these compounds were generally associated with pineapple, floral, woody, lemon, green and leaf aromas in the sensory descriptions of the panelists ( Table 2). Among the hydrocarbon compounds, limonene (7.5-7.8) was present with the highest aroma intensity in all five samples, having a typical lemon aroma. Meanwhile, α-pinene (1.6-2.8) and myrcene (1. 3-6.3) were also present in all samples. 4-isopropyltoluene (1.6-7.5) had the highest aroma intensity in ZG, MS, and JY, and was an important volatile compound contributing to the green and herbal aroma in orange samples. In addition to hydrocarbons, esters were another important class of odor-active compounds, with apple, pineapple, pear, etc., fruity aromas. Ethyl butyrate (5.3-7.5) possessed the highest aroma intensity among these compounds. Ethyl isobutyrate (1.7), ethyl isovalerate (2.5) and propyl butyrate (1.9) were only detected in QY.

Calculating OAV Values of Volatile Compounds in Sweet Orange and Percentage of Notes Contribution
As is shown in Table 3, there were 13, 16, 10, 12 and 16 compounds (OAVs ≥ 1) corresponding to the ZG, MS, JY, YX and QY samples that were considered to contribute significantly to the aroma of the samples. In this study, ethyl butyrate (5-233), myrcene , limonene (96-122), 4-isopropyltoluene (46-146), linalool (10-136) and decanal (22-66) had relatively higher OAVs, and were powerful aroma compounds in the five sweet orange samples. According to the GC-O olfaction of sensory evaluation panelists and the scores of notes in five sweet orange samples, 25 key aroma substances with OAVs ≥ 1 were classified into 8 notes. The fresh note of sweet orange was all derived from acetaldehyde; hexanol and hexanal had a strong aroma of vegetables and grass, so they were classified as green note; all esters contributed to the characteristic fruits note; all terpene compounds, including limonene and terpinene, provided the characteristic aroma for the citrus note; linalool and hedion had the fragrance of roses and jasmine, which were classified as the note of flowers; decanal with long chain carbon had the peculiar aroma of citrus peel; trans-caryophyllene and germacrene had distinct aroma of wood and herbs, which were attributed to the woody note.    We can intuitively see the percentage contribution of each note in Figure 1. The percentage contribution of citrus note was 45.57%, accounting for almost half of the whole flavor; the second was the fruity note, with a percentage of 31.39%. Both percentage contributions of the citrus and fruit notes totaled more than 70%, which were the main aroma in sweet orange juice; the floral, peely, fresh, woody, green and fatty notes accounted for 9.48%, 6.53%, 6.02%, 1.15%, 1.06% and 0.31%, respectively. Although they had a smaller percentage contribution of notes in the aroma of sweet orange, they were indispensable. They played roles in modifying the aroma of orange juice and made the aroma more full and rich. The establishment of the note contour of the sweet orange aroma can help flavorists to innovate and develop products related to sweet orange flavor, and help the fragrance companies and customers to quickly and efficiently evaluate and discuss sweet orange flavor in terms of its aroma.

Partial Least Squares Analysis Further Verifies the Relationship between Aroma Compounds and Notes
As shown in Table 4, generic descriptive analysis and a ten-point line scale of sweet orange according to the aroma intensity were performed by six sensory panelists. Aromas of sweet orange were profiled with fresh, green, fruity, fatty, citrus, floral, peely and woody notes. Through the three replicates of sensory evaluation, obvious differences were found among the notes in the five sweet oranges. PLSR was applied to certify the correlation between the concentrations of aroma compounds detected by GC-MS and mean score of notes by sensory evaluation. The 49 odor-active compounds were used as X-matrix, and the 8 sensory notes were Y-matrix, which generated the correlation load diagram of PLSR, as shown in Figure 2. respectively. Although they had a smaller percentage contribution of notes in the aroma of sweet orange, they were indispensable. They played roles in modifying the aroma of orange juice and made the aroma more full and rich. The establishment of the note contour of the sweet orange aroma can help flavorists to innovate and develop products related to sweet orange flavor, and help the fragrance companies and customers to quickly and efficiently evaluate and discuss sweet orange flavor in terms of its aroma.

Partial Least Squares Analysis Further Verifies the Relationship between Aroma Compounds and Notes
As shown in Table 4, generic descriptive analysis and a ten-point line scale of sweet orange according to the aroma intensity were performed by six sensory panelists. Aromas of sweet orange were profiled with fresh, green, fruity, fatty, citrus, floral, peely and woody notes. Through the three replicates of sensory evaluation, obvious differences were found among the notes in the five sweet oranges. PLSR was applied to certify the correlation between the concentrations of aroma compounds detected by GC-MS and mean score of notes by sensory evaluation. The 49 odor-active compounds were used as X-matrix, and the 8 sensory notes were Y-matrix, which generated the correlation load diagram of PLSR, as shown in Figure 2. From Figure 2, the blue dots are clustered near the red dots, indicating that these volatile compounds contributed to the note. Points within the circle indicate that the contribution of aroma was less than 50%, and points on the circle indicate that the contribution of the aroma was between   50% and 100%. The first quadrant basically distributes the fruity note and representative substances A7-A20 (including ethyl butyrate, ethyl 2-methylbutyrate, ethyl isovalerate), fresh note A1 (acetaldehyde) and fatty note A21,A28 (nonanal and octanal); the second quadrant distributes the peely note A31 (decanal), floral note A33, A38, A42, A47(linalool, hedion, neral, citral) and citrus note A11-A25 (α-pinene, 3-carene, myrcene, a-terpinene, limonene, 4-terpineol, o-Cymene,α-Terpinolene, 4-Isopropyltoluene), the scattered distribution is more intensive, and their aroma contribution is larger; the third quadrant is the woody note A35 (trans-caryophyllene); the fourth quadrant distributes the green note A10, A26, A27 (hexanol, hexanal, cis-3-hexen-1-ol). PLSR data analysis revalidates the high correlation between aroma substances detected by GC-O and notes of sensory evaluation. Figure 2. Analysis of correlation between aroma active substances and sensory properties in sweet orange juices (the compound name of A1~A49 is listed in Table 2).

Composition and Proportion of Sweet Orange Flavor
On the basis of Table 3, representative volatile compounds with high OAVs were selected from each note. Based on the selection and matching of these characteristic substances with OAVs > 1, sweet orange flavor base modules were established according to the notes. The complex sweet orange flavor was split into a single note base module. Each note was composed of representative  Table 2). From Figure 2, the blue dots are clustered near the red dots, indicating that these volatile compounds contributed to the note. Points within the circle indicate that the contribution of aroma was less than 50%, and points on the circle indicate that the contribution of the aroma was between 50% and 100%. The first quadrant basically distributes the fruity note and representative substances A7-A20 (including ethyl butyrate, ethyl 2-methylbutyrate, ethyl isovalerate), fresh note A1 (acetaldehyde) and fatty note A21,A28 (nonanal and octanal); the second quadrant distributes the peely note A31 (decanal), floral note A33, A38, A42, A47(linalool, hedion, neral, citral) and citrus note A11-A25 (α-pinene, 3-carene, myrcene, a-terpinene, limonene, 4-terpineol, o-Cymene,α-Terpinolene, 4-Isopropyltoluene), the scattered distribution is more intensive, and their aroma contribution is larger; the third quadrant is the woody note A35 (trans-caryophyllene); the fourth quadrant distributes the green note A10, A26, A27 (hexanol, hexanal, cis-3-hexen-1-ol). PLSR data analysis revalidates the high correlation between aroma substances detected by GC-O and notes of sensory evaluation.

Composition and Proportion of Sweet Orange Flavor
On the basis of Table 3, representative volatile compounds with high OAVs were selected from each note. Based on the selection and matching of these characteristic substances with OAVs > 1, sweet orange flavor base modules were established according to the notes. The complex sweet orange flavor was split into a single note base module. Each note was composed of representative characteristic aroma substances and their usages. Through the sniffing training of the 8 base modules, different sweet orange flavors can be quickly evaluated ( Figure 3). It could also be used by companies to communicate with customers or suppliers. Meanwhile, the sweet orange flavor can be optimized and innovated on the basis of the module by flavorists, and develop sweet orange tastes which are more popular with consumers.

Materials
Sweet oranges from five different origins in China, including Zigui (ZG), Meishan (MS), Jiangyong (JY), Yaoxiang (YX) and Qingyang (QY) were selected for the experiment. 'ZG' were harvested from Hubei province, 'MS' were harvested Sichuan province, 'JY' were harvested from Hunan province, 'YX' were harvested from Guangxi province and 'QY' were harvested fromGansu province. These oranges were harvested at maturity period. The orange cores were removed and squeezed into juices by the JYL-C051 kitchen blender (Joyoung, Shandong, China). The filtered juices were kept in a refrigerator (−10 °C) until analyzed.

Materials
Sweet oranges from five different origins in China, including Zigui (ZG), Meishan (MS), Jiangyong (JY), Yaoxiang (YX) and Qingyang (QY) were selected for the experiment. 'ZG' were harvested from Hubei province, 'MS' were harvested Sichuan province, 'JY' were harvested from Hunan province, 'YX' were harvested from Guangxi province and 'QY' were harvested fromGansu province. These oranges were harvested at maturity period. The orange cores were removed and squeezed into juices by the JYL-C051 kitchen blender (Joyoung, Shandong, China). The filtered juices were kept in a refrigerator (−10 • C) until analyzed.

Solid Phase Microextraction (SPME) Adsorption of Aroma Compounds
10g orange juice was demulsified by adding 2 g NaCl and then quantitatively analyzed by adding 20 µL 2-octanol (100 mg/L) as internal standard. The juice was then placed in a 20 mL sealed solid microextraction vials and shaken evenly. Vials were kept at 45 • C in a water bath with 10 min of equilibration time. A SPME semi-quantification of volatile compounds was conducted according to some researches [23][24][25].
A 50/30 µm divinylbenzene/Carboxen/polydimethylsiloxane (DVB-CAR-PDMS) fiber (Supelco, Bellefonte, PA, USA) with a length of 1 cm was used. Extraction time was 45 min. Before chemical absorption, the fiber was preconditioned for 30 min on an Agilent 7890 gas chromatograph (Agilent Technologies, Santa Clara, CA, USA) with the injector temperature of 250 • C.

SPME-GC-MS of Volatile Compounds in Orange Juice
The volatile compounds were analyzed by a 7890A gas chromatograph with SPME and a 5973 mass selective detector (MSD) (Agilent Technologies). Samples were conducted using DB-Wax analytical fused silica capillary column (60 m × 0.25 mm × 0.25 µm; Agilent, Santa Clara, CA) and a DB-5 fused-silica capillary column (60 m × 0.25 mm × 0.25 µm, Agilent). Conditions for GC-MS analysis were as follows: the carrier gas was helium at a constant flow rate of 1 mL/min. The oven temperature was held at 50 • C, ramped at a rate of 10 • C/min to 100 • C for 5 min, and then ramped to 140 • C at a rate of 3 • C/min for 10 min, finally it reached 230 • C at a rate of 2 • C/min for 5 min. The desorption time was 5 min.
The MSD was used for chemical identification. Its electron ionization energy was 70 eV. The ion source temperature was set at 230 • C. The compounds were identified by matching retention time of authentic standards, retention indices (RIs), and mass spectra in the NIST 11 database. The RIs of unknown compounds were determined by alkanes C4-C30.

SPME-GC-FID-O Analysis of Orange Juice
GC-O analysis was conducted using an Agilent 7890A gas chromatograph equipped with a flame ionization detector (FID) and an ODP-2 olfactory port (Gerstel, Mulheim an der Ruhr, Germany). Separation of gas chromatographic was split into 1:1 (v/v) between flame ionization detector and sniffing. The purified and damp air flowing was transported to the olfactory assessor at a speed of 1.2 mL/min. Conditions for GC-O analysis and heating procedure referred to those of GC-MS.
The olfactory experiment was performed by 6 trained panelists (three females and three males, age: 22-40). Panelists were very sensitive to aroma identification due to olfactory training to reference compounds, sweet orange sample matrices in sniffing bottles, and experience in GC-O. Aroma characteristics and aroma intensity were recorded by the assessors based on 45 min of sniffing time. The intensity was calculated as the average of all panelists' scores for identified aroma. The odor intensities were evaluated on a 10-point intensity line scale, where 0 meant a compound had a slight odor, 5 represented a moderate intensity and 10 was for an extremely strong sensation.

Sensory Evaluation
Based on the previous studies [26][27][28], the method of sensory analysis was generic descriptive analysis. 20 g orange juice was prepared in a 100 mL plastic cup with a Teflon cover for evaluation. At the beginning, the aroma of the orange juice was evaluated by a well-trained panel of 6 members (3 males and 3 females). Then, through the three preliminary consensus sessions (each 2 h), the panelists eventually reached a final agreement about the aroma description of the orange juice. Based on these discussions, the sweet orange juice was divided into eight notes: fresh, green, fruity, fatty, citrus, floral, peely and woody. A ten-point line scale, from 0 (not perceivable) to 3.0, 4.0, 5.0 (moderately perceivable), to 10.0 (very strongly perceivable), was given to the respective notes of orange juice according to the aroma intensity.

Odor Activity Values (OAV)
By using the formula of olfactory activity value, OAV = C/T, in which OAV represents the olfactory activity value of each flavor substance, C represents the concentration of the each compound, T represents the each compound of detection threshold in water. It is generally believed that aromatic compounds with high OAV are most likely to be the main contributors to the overall aroma. OAV > 1 indicates that the compounds have a direct impact on the aroma of sweet orange [29]. According to GC-O olfaction and the note classification of the sensory group, all volatile substances detected by GC-MS were classified into eight notes, and the OAV value of each note was the sum of the OAV values of all volatile substances in this note. Contribution percentage of a note in sweet orange juice = the OAV value of the note/the sum of all the OAV values of notes.

Statistical Analysis
In this study, relative standard deviation (RSD) better reflected the precision of GC-MS and GC-O tests data. Aroma intensity of GC-O and quantitative contents of volatile compounds were performed by analysis of variance (ANOVA). When there were significant differences between samples, Duncan's multiple range tests were used with the level of significance set at p < 0.05. Both ANOVA and Duncan's multiple range tests were conducted by PASW Statistics 18 (IBM, Chicago, IL, USA). Partial least squares regression (PLSR) was conducted by Unscrambler version 9.8 (CAMO ASA, Oslo, Norway). PLSR was used to further verify the relationship between flavor notes and corresponding aroma compounds.

Conclusions
Based on the results of GC-MS analysis and sensory evaluation, the percentage contribution of note OAVs in sweet orange were calculated and the proportion of aroma components in sweet orange could be visually displayed by the aroma contour diagram. By this method, the aroma quality of sweet orange flavor of the same type can be precisely and meticulously compared and evaluated, so as to develop more innovative and competitive orange flavor products. Meanwhile, the aroma module of orange flavor is conducive to communication between the fragrances and customers. A training system for sensory evaluation of orange flavor has been established to enable trainees to distinguish and identify the quality of sweet orange flavors from perspective of aroma.