Comparison of bayberry fermented wine aroma from different cultivars by GC‐MS combined with electronic nose analysis

Abstract Four bayberry cultivars (Biqi, Dongkui, Wandao, and Dingao) in eastern China were selected to produce the fermented bayberry wine. The volatile flavor compounds in different bayberry wine were compared by gas chromatography–mass spectrometry (GC‐MS) and electronic nose. The results showed that 46 volatile flavor compounds were found in bayberry wine, including 19 esters, 7 alcohols, 6 acids, 2 aldehydes, 2 ketones, 3 terpenes, and 7 others compounds. The most important contribution to the aroma of bayberry wine was esters and alcohols, respectively. Differentiation of four kinds of bayberry wine was conducted analysis by E‐nose. Sensory evaluation showed that Biqi bayberry wine was highly evaluated for its highest score in color, floral aroma, overall acceptability, and fruity aroma. Our results suggest that there were differences in the flavor characteristics of bayberry wine brewed from different varieties of bayberry. The results of this study will provide valuable information for bayberry wine makers to select raw materials.


| INTRODUC TI ON
Bayberry (Myrica rubra) originating from China is one of the most popular fruits on the market (Fang, Zhang, Sun, & Sun, 2006). It is cultivated in China for more than 2000 years (Chen, 1996). Bayberry is a favorable and profitable fruit with abundance in carbohydrate, organic acids, soluble sugars, minerals, vitamins, and phenolics (Cheng et al., 2016;Xu, Zhang, Fang, Sun, & Wang, 2014).
Dongkui bayberry (DK), Biqi bayberry (BQ), Dingao bayberry (DA), and Wandao bayberry (WD) are the four main cultivars in Zhejiang province, China, accounting for more than 60% of the total yield of bayberry in China. Because bayberry was harvested during the hot and rainy season from June to July, it was susceptible to mechanical injury and microbiological decay (Fang et al., 2009;Yu, Lin, Zhan, He, & Zhu, 2013), which greatly affects the commercial value of the bayberry (Zhang et al., 2005). With the increase in yield of bayberry, the bayberry has been further processed into juice drinks (Shao & He, 2007), canned bayberry (Ya-Mei et al., 2007), and dried bayberry  and bayberry wines in order to increase its consumption and extend the shelf life. Bayberry wine is produced by fermenting methods using bayberry as raw material has extremely high nutritional value and medicinal effect (Zhang, Li, & Fan, 2019).
Aroma is an important indicator that influences the intrinsic quality of bayberry fruit and its deep processed products. In addition, aroma was one of the most valuable attributes of wines that determines the sensory quality and value of wine (Mamede, 2005).
The quality of aroma directly affects the flavor quality of bayberry wine and the consumers' acceptance and preference. Chinese bayberry cultivars grown in different locations have different flavors, which affect the flavor and quality of bayberry wine. Xu et al. (Xu et al., 2014) have studied the flavor changes during processing and storage of bayberry juice using the headspace solid-phase microextraction (HS-SPME) coupled with gas chromatography-mass spectrometry (GC-MS). The results showed that the bayberry juice produced fermentation-like flavors with an increase in alcohols (11.45%) and decreases in esters (14.91%) after 9 months of storage.
Cheng et al. (Cheng, Chen, Li, et al., 2015) used HS-SPME-GC-MS combined with principal component analysis to identify the volatile flavor components of bayberry during storage, indicating that different varieties of bayberry have different flavor characteristics. Kang et al. (Kang, Li, Xu, Jiang, & Tao, 2012) studied the aroma components of immature and mature bayberry fruit by HS-SPME/ GC-MS. The results showed that terpenoids (such as caryophyllene) were the most abundant, and alcohol, aldehydes, ketones, esters, and acids were less abundant. However, it has not yet been fully reported investigating the volatile flavor compounds of bayberry wines by GC-MS coupled with electronic nose (E-nose). GC-MS studies have mainly focused on the measurement of certain volatile compounds, while the E-nose is an instrument that uses chemical sensors to detect volatiles and then provides a holistic view of the volatile compounds of the sample through a powerful mathematical software analysis system that helps determine the odor. The detection data can be analyzed using principal component analysis (PCA), cluster analysis (CA), and linear discriminant analysis (LDA) (Wilson & Baietto, 2009). Owing to the advantages of rapid, accurate, and effective determination, well verification, and complement to each other, E-nose combined with GC-MS had already been used in the analysis of Goji berries (Li, Yu, Xu, & Gao, 2017). In addition, E-nose has been used to evaluate the quality of beverages (Banerjee, Tudu, Bandyopadhyay, & Bhattacharyya, 2019), including the identification of alcohol brands, the quality identification of distilled white spirits, and the identification of different types of red wines (García et al., 2006;Lozano, Arroyo, Santos, Cabellos, & Horrillo, 2008). However, little information has been reported in aroma analysis of different bayberry wines by using E-nose combined with HS-SPME/GC-MS.
The fruit wine industry has a famous saying that "wines success for brewing, more important is raw materials," which shows the importance of raw materials in fruit wine brewing (González-Mas et al., 2009). There are differences in the characteristics and composition of different varieties of bayberry fruit, such as color, aroma, and taste (Cheng et al., 2016). The influence of bayberry cultivars on the volatile flavor compounds of bayberry wines has not been reported.
The choice of yeast is a key step in the production of fruit wine, which directly affects the flavor quality of the bayberry wine. The bayberry variety in China contains a large amount of malic acid to negatively affect the wine quality. Issatchenkia orientalis can degrade malic acid efficiently (Kim, Hong, & Park, 2008;Negi & Dey, 2013).
In our previous study, two strains of Saccharomyces cerevisiae 131 (Sc 131) and Issatchenkia orientalis 166 (Io 166) were selected for mixed fermentation of fruit wines, which are the most suitable to produce flavor and alcohol components, respectively (Wenwen, Peifang, & Zufang, 2019). The present study aims to elucidate the flavor characteristics and key volatile components of four varieties of bayberry wines fermented by Io 166 and Sc 131. GC-MS and E-nose were used to analyze the volatile flavor components of four major mainplanted bayberry cultivars in eastern Zhejiang. At the same time, its components and sensory quality were also analyzed. This study results provided an important basis for the selection of bayberry cultivars and raw materials for fermenting wine.

| Sample preparation
Bayberry cultivars including Dongkui (DK), Biqi (BQ), Dingao (DA), and Wandao (WD) were used as raw materials for wine fermentation. These bayberries were purchased at a mature stage from their main production areas in Ningbo, Zhejiang Province, China, on the same day during June 2018. Bayberries were packaged in ice bags and transported to the laboratory as soon as possible, where they were preserved at 4°C no more than an hour. Then, four cultivars of bayberry juices were produced with a juicer extractor and filtered through gauze within 1 hr.

| Chemicals and reagents
Ethanol Assay Kit K-ETOH was purchased from Megazyme, Ireland; YPD medium, phenol, sodium hydroxide, potassium metabisulfite, 3,5-dinitrosalicylic acid, sodium potassium tartrate, and sodium bisulfite are of analytical grade and they are purchased from Sinopharm Chemical Reagent Co., Ltd.

| Winemaking process
The total soluble solid (TSS) content of bayberry juice was adjusted to 22.5 °Brix using sucrose. Then, potassium metabisulfite was added to make the concentration of sulfur dioxide at 40 ~ 100 mg/L. Each cultivar was well mixed before winemaking and separated into three replicates to avoid compositional variation (Liu, Li, Gao, Cheng, & Yuan, 2019). The mixture was pasteurized at 75°C for 15 min and cooled to 20°C and inoculated with Sc 131 at approximately 10 5 cfu/ml and Io 166 at approximately 10 6 cfu/ml. The main fermentation was conducted at 27°C for about 4 to 6 days until the total sugar content less than 8 g/L. Then, the fermented mash was placed at 18°C for 15 days for post-fermentation. At the end of fermentation, the wine was clarified using 0.08 g/L of chitosan for 2 hr and racked for 1 day at 4°C.
After centrifugation, 70 mg/L of potassium metabisulfite was added to the wines. Then, bayberry wines were bottled with equal headspace volume. Finally, they were labeled and stored at room temperature, respectively, for 3 months in the dark before analysis was carried out.
The concentration of each important composition was detected according to references including total sugars (Liu et al., 2013), total soluble solids (Koshita, Yamane, Yakushiji, Azuma, & Mitani, 2011), and total anthocyanin (Giusti & Wrolstad, 2001), as well as alcohol, titratable acid, and pH. Alcohol content was detected by Ethanol Assay Kit K-ETOH. The analysis was conducted in quadruplicate for each parameter investigated.

| GC-MS analysis
According to the previously published method, the volatile compounds in the bayberry wine were extracted by headspace solidphase microextraction, and the method was slightly modified (Liu et al., 2018;Yu, Xie, Xie, Ai, & Tian, 2019). Bayberry wine samples (5 ml) were added into 20-ml headspace glass vials (18 mm

| Electronic nose analysis
The electronic nose (Germany Airsense PEN 3.5) was used to tentatively estimate the aroma profile similarity after fermentation.
The E-nose analysis was based on previous reports (Hong, Wang, & Qi, 2015;Li et al., 2017) and with modifications. The procedures were as follows: 5 ml of each bayberry wine sample was added in a 20-ml glass vial and capped with a Teflon rubber cap. The vial with the bayberry wine sample was allowed to stand at room temperature for 30 min, while the headspace collected the volatiles from the wines. During the measurement process, the headspace gaseous compounds were pumped into the sensor arrays through a tube connected to a needle in the Teflon rubber cap at a flow rate of 400 ml/min, resulting in the ratio of conductance G/G0 (G and G0 are conductance of the sensors exposed to wine gas and zero gas, respectively) of each sensor changed.
The measurement time was 220 s, which was long enough for the sensors to reach stable signal values. When the measurement was completed, the data were stored by electronic nose software for later PCA and LDA analysis. After each samples, zero gas (air filtered by active carbon) was pumped into the sample gas path from the other port of the instrument for 120 s (flush time). The 10 metal oxide sensors of the PEN 3.5 electronic nose are described in Table 1. Different sensors respond to different volatile substances.
Each wine sample was measured for five repeats.

| Statistical analysis
Data from the characterization of the bayberry wines are reported as mean ± standard deviation for quadruplicate determinations.
Electronic nose measurements of bayberry wine sample were performed using WinMuster software (Winmuster1.6.2) for PCA and LDA. All the data were analyzed using the one-way analysis of variance (ANOVA) using SPSS, version 22.0.

| Analysis of physicochemical properties of different bayberry wines
The content of the total sugar, soluble solids, anthocyanins, alcohol, titratable acidity, and pH of four bayberry wines was analyzed, and the data are listed in Table 3.
It can be found that in the BQ wine, titratable acidity content was lowest, and the total sugar and alcohol content were moderate.
Statistical analysis showed that there were significant differences (p < .05) in the total anthocyanin content between the bayberry wines made from the four cultivars. WD wine had the highest total anthocyanin content (116.6 mg/L) among the four kinds of   Note: The symbol "-" represents nondetected; different superscript letters in the same rows mean significant differences (p < .05); Aroma Description Reference Website http://www.thego odsce ntsco mpany.com/.

TA B L E 4 (Continued)
bayberry wines. The other bayberry wines, in descending order by total anthocyanin content, were BQ wine (88.5 mg/L), DA wine (72.8 mg/L), and DK wine (51.1 mg/L). Anthocyanins are highly correlated to antioxidant capacity of the most fruits, which may have potential benefits for human health and disease prevention (Wallace, 2011;Wang et al., 2016). The present result showed that the variety of bayberry influenced the quality of bayberry wines.

| Comparison of volatile flavor compounds of different bayberry wines
The volatile flavor compounds of the four kinds of bayberry wines are shown in Table 4. As can be seen from Table 4, a total of 46 volatile flavor compounds were found in the four kinds of bayberry wines, including 19 esters, 7 alcohols, 6 acids, 2 aldehydes, 2 ketones, 3 terpenes, and 7 others compounds. Most of the compounds were detected with high detection frequency in all four bayberry wines. Therefore, these compounds might play an important role in the characterization of bayberry wines.
During fermentation, yeasts convert sugar to ethanol, producing a variety of by-products such as higher esters, alcohols, acids, aldehydes, ketones, terpenes, and other volatile compounds which contribute to wine aroma (Styger, Prior, & Bauer, 2011). As shown in Figure 1, esters and alcohols were the largest groups, the main aroma compounds in bayberry wines. They were produced during alcoholic fermentation and played an essential role in wine flavor, depending on types of compounds and their concentrations (Valero et al., 2002).  (Liu et al., 2019). It has been suggested that esters are formed mainly through the esterification of alcohols with organic acids during the fermentation and storage processes (Erten, Tanguler, & Cakiroz, 2007). Aldehydes and ketones were another key aroma group in bayberry wine. Phenylacetaldehyde, 3-hydroxy-2-butanone, and 6-methyl-5hepten-2-one have sweet and fruity aroma. Benzaldehyde possesses an almond aroma. Terpenes have great benefits for the human body (Petrović, Stojković, & Soković, 2019). Biqi bayberry wine contained F I G U R E 3 LDA of four kinds of bayberry wines F I G U R E 4 Graph of the mean sensory scores of the four bayberry wines studied more amounts of terpenes (total 2.700 mg/L) than DK wine, DA wine, and WD wine. Caryophyllene has a sweet woody and with a citrus background aroma.

| Electronic nose analysis
Principal component analysis (PCA) is a method for studying the similarities and differences between various measurement data (Huang, Wu, Chen, Weng, & Zhang, 2018). to the high variance contribution rate, sufficient representativeness can be observed (Li et al., 2017). The results show that the volatile flavor compounds of the four kinds of bayberry wines had obvious differences. This is consistent with the results detected by GC-MS.

| Sensory evaluation
The results of the sensory evaluation analysis are shown in Figure 4.
The result analysis demonstrated that "fruity aroma," "floral aroma," "alcoholic aroma," "sour," "color," and "overall acceptability" descriptors showed significant differences between the four kinds of bayberry wines in the sensory evaluation scores. Dongkui bayberry wine was intense in "sour" descriptors, whereas "floral aroma" and "color" exhibited lower level. Dongkui bayberry wine tasted the sourest due to a lot of acetic acid. Acetic acid played an important role in the formation of fruit wine flavor substances, but excessive acetic acid causes the wine too acidic when tasted.
In addition to alcoholic aroma, BQ wine exhibited obviously fruity aroma and floral aroma, and it may be related to the high relative content of esters and higher alcohols. Moreover, BQ wine has beautiful color and good overall acceptability. From the sensory evaluation, the sensory quality of the BQ wine was the best among the four bayberry wines.

| CON CLUS IONS
The analysis of sensory, physicochemical properties and volatile flavor compounds of the different bayberry wines by GC-MS coupled with E-nose have shown that the variety of bayberry has a greater effect on the quality of bayberry wines under the same brewing process. Esters and alcohols were the main aroma compounds in bayberry wines.
Moreover, the different bayberry wines aroma feature could well be distinguished based on GC-MS results and PCA and LDA of E-nose data. Based on the comprehensive results, it is advisable to select the BQ in the eastern Zhejiang province for the fermentation of the bayberry wine. The fact of this study may provide an important basis for selection of brewing materials of high-quality bayberry wines.

ACK N OWLED G M ENTS
The authors are grateful to the financial support of National Natural Science Foundation, China (NNSF No. 31471709), Public Welfare Project of Zhejiang (LGN18C200018), and K.C. Wong Magna Fund in Ningbo University.

CO N FLI C T O F I NTE R E S T
The authors have no conflicts of interest.

E TH I C A L A PPROVA L
This study does not involve any human or animal testing.