The Sensory Quality Improvement of Citrus Wine through Co-Fermentations with Selected Non-Saccharomyces Yeast Strains and Saccharomyces cerevisiae

Co-fermentation of selected non-Saccharomyces yeast strain with Saccharomyces cerevisiae is regarded as a promising approach to improve the sensory quality of fruit wine. To evaluate the effects of co-fermentations between the selected non-Saccharomyces yeast strains (Hanseniaspora opuntiae, Hanseniaspora uvarum and Torulaspora delbrueckii) and S. cerevisiae on the sensory quality of citrus wine, the fermentation processes, the chemical compositions, and the sensory evaluations of citrus wines were analyzed. Compared with those of S. cerevisiae fermentation, co-fermentations produced high sensory qualities, and S. cerevisiae/H. opuntiae co-fermentation had the best sensory quality followed by Sc-Hu and Sc-Td co-fermentations. Additionally, all the co-fermentations had a lower amount of ethanol and total acidity, higher pH value, and higher content of volatile aroma compounds, especially the content of higher alcohol and ester compounds, than those of S. cerevisiae fermentation. Therefore, co-fermentations of the non-Saccharomyces yeast strains and S. cerevisiae could be employed to improve the sensory quality of citrus wines. These results would provide not only methods to improve the sensory quality of citrus wine, but also a valuable reference for the selection of non-Saccharomyces yeast strains for fruit wine fermentation.


Introduction
Citrus is one of the most abundant fruit crops in China, with a production of 8.56 million tons, accounting for 11.8% of the global production in 2017 (FAO, 2017). Mandarin fruit cv. Ponkan (Citrus reticulata Blanco cv. Ponkan) is widely cultivated in Asian countries for its high quality [1]. Ponkan has a thin skin, and the quality of the fruit during storage at room temperature is highly prone to deterioration and decay over time. Improving storage performance and reducing fruit decay during storage and transportation are major issues for the citrus industry [2,3]. In addition to being consumed as fresh fruit, citrus can also be processed into citrus wine to extend its shelf life and increase its added value [4]. However, the insufficient research and development of fermented citrus wine, as well as the poor flavor and quality of citrus wine, decreased its competitiveness on the fruit wine market [5].
Yeasts are primarily responsible for the alcoholic fermentation of fruit juice. Industrial fruit wine is usually fermented by using commercial Saccharomyces cerevisiae with the advantages of the controllable processes and stable quality products, but its sensory quality is inferior to those of successful starters of S. cerevisiae/H. uvarum (Sc-Hu), S. cerevisiae/T. delbrueckii (Sc-Td) or S. cerevisiae/H. opuntiae (Sc-Hop) in 1.0 L sterile bottles at 25 • C. The co-fermentations were performed with co-cultures of 10 7 CFU/mL non-Saccharomyces yeast strain and 10 6 CFU/mL S. cerevisiae with sequence inoculation strategy. The sequence inoculation strategy meant that S. cerevisiae was inoculated into citrus juice 24 h later than H. opuntiae and T. delbrueckii, or 72 h later than H. uvarum. Pure fermentation of S. cerevisiae with 10 6 CFU/mL inoculation was used as a control. The fermentation process of citrus wine was monitored by analyzing the residual sugar concentration and yeast cell counts daily. Residual sugar concentration of citrus wine was assessed by dinitrosalicylic acid (DNS) method according to the International Organization of the Vine and Wine (OIV, 2005). Yeast cell count was determined by successive dilution method on the WL medium and LYS medium. All reagents were obtained from Sigma-Aldrich (Sigma-Aldrich, Shanghai, China).

Physicochemical Analysis of Citrus Wine
The content of residual sugar, total acidity (expressed as g/L of malic acid) and volatile acid (expressed as g/L of acetic acid) of citrus wine were analyzed by the International Organization of the Vine and Wine (OIV, 2005). Bromothymol blue titration method was used to analyze the total acidity of citrus wine. Distillation and titration method was applied to evaluate the volatile acid of citrus wine. Rapid oxidation of potassium permanganate method was employed to detect the alcohol content of citrus wine. The pH value of citrus wine was determined by pH meter (Mettler-Toledo, Shanghai, China). All experiments were determined in triplicate.

HS-SPME/GC-MS Analysis of Volatile Aroma Compounds
The volatile aroma compounds were extracted by headspace solid-phase microextraction (HS-SPME) method with a 50/30 µm divinylbenzene/carboxen/polydimethylsiloxane (DVB/CAR/PDMS) fiber (Supelco, Bellefonte PA, USA). The Agilent 6890N gas chromatography (Agilent 6890N, Agilent Technologies Inc., Shanghai, China) on an HP-5 capillary column (30 m × 0.32 mm × 0.25 µm) coupled to an Agilent 5975B mass spectrometer was used to analyze the extracted volatile aroma compounds. The extraction, analysis, and identification of volatile aroma compounds were conducted as described by Hu et al. [24]. The odor active value (OAV), calculated as the ratio between the concentration of flavor compound to its odor threshold (OT), was used to obtain odor patterns. Volatile aroma compounds with OAV≥1 were considered as odor-active compounds [26].

Sensory Evaluation of Citrus Wine
The sensory evaluation of citrus wine was conducted by a trained panel consisting of nine panelists (five females and four males) from Huazhong Agricultural University. The sensory evaluation of citrus wine was performed as described by [24].

Statistical Analyses
One-way ANOVA and Duncan test of all indicated significant differences were conducted by SPSS 19.0 (SPSS Inc., Chicago, IL, USA). Principal component analysis (PCA) was performed to identify the most influential volatile aroma compounds in different fermentations by SIMCA-P 14.1 (Umetrics AB, Umea, Sweden). Hierarchical cluster analysis and heat map visualization of volatile aroma compounds with the Z-score standardization in different fermentations were analyzed by MultiExperiment Viewer 4.9.0 (TIGR, America).

Growth Kinetics and Sugar Consumption Kinetics of Yeast Strains during Fermentations
As charted in Figure 1, these non-Saccharomyces yeast strains shortly adapted to the fermentation environment, then grew normally within 4-6 days and reached their maximum biomass of 1.0 × Microorganisms 2020, 8, 323 4 of 16 10 8 CFU/mL. However, the growth of H. uvarum was immediately inhibited by S. cerevisiae with sharp decrease of biomass during fermentation ( Figure 1A). The growth inhibition phenomenon of non-Saccharomyces yeast strains were also reported in other studies [27,28]. Additionally, the early inoculation of these non-Saccharomyces yeast strains negligibly affected the maximum biomass of S. cerevisiae (1.4 × 10 8 -1.9 × 10 8 CFU/mL) in co-fermentations compared with that of pure S. cerevisiae fermentation (2.3 × 10 8 CFU/mL) ( Figure 1A-D). These results indicated the selected non-Saccharomyces yeast strains and S. cerevisiae could grow normally during their co-fermentation with the exception of H. uvarum which was inhibited by S. cerevisiae. The inhibition of non-Saccharomyces yeast strains was probably ascribed to the fierce nutrients competition or killer factors from S. cerevisiae [29].

Growth Kinetics and Sugar Consumption Kinetics of Yeast Strains during Fermentations
As charted in Figure 1, these non-Saccharomyces yeast strains shortly adapted to the fermentation environment, then grew normally within 4-6 days and reached their maximum biomass of 1.0 × 10 8 CFU/mL. However, the growth of H. uvarum was immediately inhibited by S. cerevisiae with sharp decrease of biomass during fermentation ( Figure 1A). The growth inhibition phenomenon of non-Saccharomyces yeast strains were also reported in other studies [27,28]. Additionally, the early inoculation of these non-Saccharomyces yeast strains negligibly affected the maximum biomass of S. cerevisiae (1.4 × 10 8 -1.9 × 10 8 CFU/mL) in co-fermentations compared with that of pure S. cerevisiae fermentation (2.3 × 10 8 CFU/mL) ( Figure 1A-D). These results indicated the selected non-Saccharomyces yeast strains and S. cerevisiae could grow normally during their co-fermentation with the exception of H. uvarum which was inhibited by S. cerevisiae. The inhibition of non-Saccharomyces yeast strains was probably ascribed to the fierce nutrients competition or killer factors from S. cerevisiae [29].

Physicochemical Parameters of Citrus Wine
As presented in Table 1, co-fermentations of non-Saccharomyces yeast strains and S. cerevisiae contained lower concentrations of ethanol (9.74%-10.24%) and total acidity (7.68-8.63 g/L) than pure S. cerevisiae fermentation did (11.29% and 9.44 g/L). The Sc-Hu co-fermentation possessed a lower concentration of residual sugar (1.38 g/L), while others contained a higher level of residual sugar (2.03-2.81 g/L) than pure S. cerevisiae fermentation did (2.03 g/L), which might be caused by different sugar consumption abilities of these yeast strains [30,31]. The pH value of co-fermentations (3.37-

Physicochemical Parameters of Citrus Wine
As presented in Table 1, co-fermentations of non-Saccharomyces yeast strains and S. cerevisiae contained lower concentrations of ethanol (9.74%-10.24%) and total acidity (7.68-8.63 g/L) than pure S. cerevisiae fermentation did (11.29% and 9.44 g/L). The Sc-Hu co-fermentation possessed a lower concentration of residual sugar (1.38 g/L), while others contained a higher level of residual sugar (2.03-2.81 g/L) than pure S. cerevisiae fermentation did (2.03 g/L), which might be caused by different sugar consumption abilities of these yeast strains [30,31]. The pH value of co-fermentations (3.37-3.47) increased compared with that of pure S. cerevisiae fermentation (3.35), which was also found in co-fermented bilberry wine [32]. Additionally, the volatile acid concentrations in co-fermentations (0.11-0.12 g/L) with the exception of Sc-Td co-fermentation (0.15 g/L) displayed insignificant difference from that of pure S. cerevisiae fermentation (0.11 g/L), while other researchers reported that co-fermentations had a lower content of volatile acid than pure S. cerevisiae fermentation had [8,33]. The different research results in volatile acid content might be caused by different fruit juice, fermentation strategies or starters. These results indicated these co-fermentations produced citrus wine with lower content of ethanol and total acidity, higher pH value and with no significant difference in volatile acid concentration. Data show the mean value of triplicate, and the significant differences in each row are respect to the control Sc. a Expressed as g/L of malic acid; b Expressed as g/L of acetic acid ** p < 0.01; * p < 0.05. Sc-Hu, S. cerevisiae/H. uvarum co-fermentation; Sc-Td, S. cerevisiae/T. delbrueckii co-fermentation; Sc-Hop, S. cerevisiae/H. opuntiae co-fermentation; Sc, S. cerevisiae fermentation.

Volatile Aroma Compounds in Citrus Wines
Forty-eight kinds of volatile aroma compounds in citrus wines, including 8 kinds of higher alcohols, 6 kinds of fatty acids, 24 kinds of esters and 10 kinds of other compounds, were detected. Compared with the S. cerevisiae fermentation, co-fermentations of these non-Saccharomyces yeast strains and S. cerevisiae produced higher content of volatile aroma compounds, and Sc-Td co-fermentation produced the maximum amount of volatile aroma compounds (819.46 mg/L), followed by Sc-Hop (689.32 mg/L), Sc-Hu (519.97 mg/L) and Sc fermentations (187.58 mg/L). Moreover, co-fermentations produced higher content of higher alcohol, acetate and terpene, and lower content of volatile fatty acids with the exception of Sc-Td fermentation ( Table 2).   Higher alcohols contributed positively to the "fresh fruity", "vegetal" notes, and aroma complexity in young red wine [34]. Higher alcohols could react with organic acids to form esters with a pleasant flavor, while excessive levels (≥500 mg/L) may lead to unpleasant flavor in alcohol beverage [35]. As tabulated in Tables 2 and 3, co-fermentations produced higher concentrations of higher alcohols (235.52 mg/L-580.72 mg/L) than that of pure S. cerevisiae fermentation did (87.97 mg/L). The Sc-Td co-fermentation produced the maximum concentration of higher alcohols (580.72 mg/L), followed by Sc-Hop (410.48 mg/L) and Sc-Hu (235.52 mg/L) fermentations. However, excessive concentration of higher alcohols (580.72 mg/L) in Sc-Td co-fermentation might result in undesirable flavor. Previous research also showed that S. cerevisiae/T. delbrueckii and S. cerevisiae/H. opuntiae co-fermentations generated large amounts of higher alcohols compared with pure S. cerevisiae fermentation did, while the differences in production of higher alcohols among different non-Saccharomyces strains were significant [36][37][38].
Among the detected alcohol groups, 1-pentanol and phenylethanol were the active odor compounds (OAV > 1), and the phenylethanol concentration in co-fermentations (183.00 mg/L-306.33 mg/L) was greatly higher than that of pure S. cerevisiae fermentation (33.21 mg/L), which was also reported by Sun et al. [33]. Phenylethanol usually presented on pleasant honey, floral aroma (rose petals), and spicy flavor [39,40]. Therefore, a higher amount of phenylethanol in co-fermentations would be positive to the flavor of citrus wines. Besides, the content of 1-pentanol with sweet and vanilla odor was higher in co-fermentation with the exceptions of Sc-Hu fermentation than that of pure S. cerevisiae fermentation did, which would strengthen the fruity and balsamic of citrus wine. These results revealed that co-fermentations of the non-Saccharomyces yeast strains and S. cerevisiae significantly contributed to the biosynthesis of higher alcohols, especially phenylethanol, and the higher alcohols production in fruit wine was determined by non-Saccharomyces yeast strain and the fermentation process.

Fatty Acids
As viewed in Tables 2 and 3, the Sc-Td co-fermentation produced higher concentration of fatty acids (20.04 mg/L) than S. cerevisiae fermentation did (2.54 mg/L), while other co-fermentations hardly produced fatty acids (0.00 mg/L-0.16 mg/L). Volatile fatty acids are essential precursors of ester productions that provide fruity aromas to wines [47], but excessive content of fatty acids (≥20 mg/L) in Sc-Td co-fermentation might produce rancid flavor [48,49]. Octatonic acid was the only odor-active fatty acid and presented in Sc-Td co-fermentation (OAV = 40.08) and pure S. cerevisiae fermentation (OAV = 4.45). The high content of fatty acids, especially octatonic acid, in Sc-Td co-fermentation was inconsistent with previous report which indicated that co-fermentation of T. delbrueckii and S. cerevisiae greatly decreased the fatty acids concentration, especially octatonic acid [50], which might be caused by different fruit juice and fermentative process. These results indicated that co-fermentations of the selected non-Saccharomyces yeast strains and S. cerevisiae have no regular effects on the amount and the kinds of fatty acids in citrus wine, which might be determined by yeast strains, fruit juice, and fermentation process.

Esters
Esters compounds including acetates and ethyl esters are important aroma compounds with a positive contribution to the desired fruit aroma characters in wine [51]. The total ester concentrations in all the co-fermentations (210.52 mg/L-281.61 mg/L) with significant differences (p < 0.05) were higher than that of pure S. cerevisiae fermentation (96.84 mg/L) ( Table 2), which was also found in bilberry wine co-fermented by T. delbrueckii and S. cerevisiae [39].
As for the acetates, all the co-fermentations also generated higher acetates amount than pure S. cerevisiae fermentation did, and Sc-Hu fermentation produced the maximum acetates content (236.82 mg/L), followed by Sc-Hop (199.59 mg/L) and Sc-Td fermentations (126.36 mg/L). Among the acetate compounds, ethyl acetate, isoamyl acetate, and phenethyl acetate were the odor active compounds (OAV > 1) and greatly increased in all the co-fermentations compared with those of S. cerevisiae fermentation. Ethyl acetate may improve aroma complexity of wine at low level (approximately 50 mg/L), but it is associated with negative sensory descriptors (nail polish and solvent etc.) at concentrations above 150 mg/L. Therefore, the ethyl acetate concentration in Sc-Td and Sc-Hop fermentations (32.04 mg/L and 81.64 mg/L, respectively) would have positive effect on the flavor of citrus wine, while that in Sc-Hu fermentation (153.50 mg/L) might has negative effect on the sensory quality of citrus wine and should be evaluated through sensory evaluation furtherly. However, Mingorance-Cazorla et al. reported that 9.67 mg/L-163.18 mg/L ethyl acetate were also detected from citrus wine, and a little over threshold of ethyl acetate in citrus wine (163.18 mg/L) had no negative effect on its flavor [52]. Isoamyl acetate and phenethyl acetate are recognized as an important flavor compound in wine and contribute to the fruity notes of wine [39]. Moreira et al. [53] and Rojas et al. [54] also revealed that non-Saccharomyces yeast, such as H. uvarum, could increase the content of isoamyl acetate and the phenethyl acetate in wine. These results suggested that co-fermentations could improve the acetate content, especially the content of ethyl acetate, isoamyl acetate and phenethyl acetates in citrus wine compared with those of pure S. cerevisiae fermentation, which would be favorable to the flavor of citrus wine.
As for ethyl esters, Sc-Td and Sc-Hop fermentations displayed distinct advantage in ethyl esters producing (77.17 mg/L and 73.31 mg/L, respectively), while Sc-Hu fermentation (43.99 mg/L) was slightly inferior to S. cerevisiae fermentation (66.62 mg/L) ( Table 2). However, Hu et al. [55] reported co-culture of H. uvarum/S. cerevisiae produced more ethyl ester in wine than S. cerevisiae did. These differences in producing abilities of ethyl ester among different yeast strains might be caused by their different producing ability of organic acids, because the organic acids were the precursors of ethyl esters [56]. Among the ethyl ester compounds, ethyl caproate, ethyl octanoate and ethyl caprate were the odor active compounds, which were characterized by "pineapple, fruity and floral", "fruity, pineapple, pear and floral", and "fruity and fresh", respectively (Table 3). These results suggested that co-fermentations of the non-Saccharomyces yeast strains and S. cerevisiae exhibited great potential in ethyl esters compounds production in citrus wine.
The content of other esters in Sc-Td co-fermentation was higher than that of other fermentations. These results suggested that different co-fermentations had no regular effects on other esters content.

Volatile Terpenes, Aldehydes, Ketones, and Phenols
Major volatile compounds in citrus wine were higher alcohols, esters and fatty acids, but the volatile aldehydes, ketones, terpenes and phenols were also identified. Various types of terpene and aldehyde compound, such as D-limonene, linalool, octanal, and decanal etc. have been reported in citrus juice [57,58]. Terpenes are important varietal aroma compounds and can impact on floral aroma of wines [59]. Co-fermentations significantly enhanced the amount and the kinds of terpene compounds in citrus wine (2.12 mg/L-12.98 mg/L) compared with those of pure S. cerevisiae fermentation (0.00 mg/L) ( Table 3). The level of terpene compounds was also enhanced in wine and mango wine by non-Saccharomyces yeast strains [60,61]. Only D-limonene in Sc-Hu and Sc-Hop fermentations and 1-pentene in Sc-Hop fermentation had the odor activities (Table 2). Aldehydes and ketones were slightly produced in Sc-Hu (0.56 mg/L) and Sc-Hop (0.55 mg/L) fermentations, while they were not detected in other fermentations. Only phenylacetaldehyde in Sc-Hu and Sc-Hop fermentation was odor active, and would present on unpleasant green notes [62]. Sc-Td co-fermentation contained higher phenol content (5.20 mg/L) than other fermentations did (0.00 mg/L-0.23 mg/L), and 4-vinyl guaiacol in Sc-Td fermentation was the only odor active phenol compound which would present on coffee, beer, apple aroma (Tables 2 and 3). Different content of terpenes, aldehydes, ketones, and phenols in different fermentations would take on different variety of flavor characteristics in citrus wines and determined by different yeast strains.

Principal Component Analysis of Volatile Compounds in Citrus Wine
The principal component analysis (PCA) demonstrated the correlation and segregation of odor active compounds (OAV≥1) with fermentations ( Figure 2). PCA explained 89.7% of the total variation. The first principal component (PC1) accounted for 54.5% of the total variation, while PC2 explained 35.2%. Sc-Td co-fermentation significantly clustered with several odor active compounds including ethyl caproate, 1-pentanol, ethyl butyrate, 4-vinyl guaicol and octanoic acid, at the upper right corner, while Sc-Hop and Sc-Hu fermentations grouped together with odor active compounds such as benzene acetaldehyde, 1-pentene, D-limonene and ethyl acetate, at the upper left corner. However, S. cerevisiae fermentation located at the lower right corner with ethyl octanoate and ethyl caprate. Similarly, Zhang et al. indicated that the S. cerevisiae monoculture mainly gathered with ethyl octanoate [11]. These results highlighted that different fermentation strategies produced distinctive odor active compounds and would present on different flavor characteristics in citrus wine.
Microorganisms 2019, 7, x FOR PEER REVIEW 9 of 17 fermentation (0.00 mg/L) ( Table 3). The level of terpene compounds was also enhanced in wine and mango wine by non-Saccharomyces yeast strains [60,61]. Only D-limonene in Sc-Hu and Sc-Hop fermentations and 1-pentene in Sc-Hop fermentation had the odor activities (Table 2). Aldehydes and ketones were slightly produced in Sc-Hu (0.56 mg/L) and Sc-Hop (0.55 mg/L) fermentations, while they were not detected in other fermentations. Only phenylacetaldehyde in Sc-Hu and Sc-Hop fermentation was odor active, and would present on unpleasant green notes [62]. Sc-Td cofermentation contained higher phenol content (5.20 mg/L) than other fermentations did (0.00 mg/L-0.23 mg/L), and 4-vinyl guaiacol in Sc-Td fermentation was the only odor active phenol compound which would present on coffee, beer, apple aroma (Tables 2 and 3). Different content of terpenes, aldehydes, ketones, and phenols in different fermentations would take on different variety of flavor characteristics in citrus wines and determined by different yeast strains.

Principal Component Analysis of Volatile Compounds in Citrus Wine
The principal component analysis (PCA) demonstrated the correlation and segregation of odor active compounds (OAV≥1) with fermentations ( Figure 2). PCA explained 89.7% of the total variation. The first principal component (PC1) accounted for 54.5% of the total variation, while PC2 explained 35.2%. Sc-Td co-fermentation significantly clustered with several odor active compounds including ethyl caproate, 1-pentanol, ethyl butyrate, 4-vinyl guaicol and octanoic acid, at the upper right corner, while Sc-Hop and Sc-Hu fermentations grouped together with odor active compounds such as benzene acetaldehyde, 1-pentene, D-limonene and ethyl acetate, at the upper left corner. However, S. cerevisiae fermentation located at the lower right corner with ethyl octanoate and ethyl caprate. Similarly, Zhang et al. indicated that the S. cerevisiae monoculture mainly gathered with ethyl octanoate [11]. These results highlighted that different fermentation strategies produced distinctive odor active compounds and would present on different flavor characteristics in citrus wine.

Sensory Evaluations of Citrus Wines
The sensory evaluations of citrus wines were shown in Table 4. Compared with those of pure S. cerevisiae fermentation, the sensory evaluation scores of co-fermentations (13.00-15.00) were higher, Ⅰ Ⅱ Ⅲ Ⅳ Ⅰ Ⅱ Ⅲ Ⅳ

Sensory Evaluations of Citrus Wines
The sensory evaluations of citrus wines were shown in Table 4. Compared with those of pure S. cerevisiae fermentation, the sensory evaluation scores of co-fermentations (13.00-15.00) were higher, and the Sc-Hop co-fermentation received the highest one (15.00), followed by Sc-Hu (14.00) and Sc-Td co-fermentations (13.00). More importantly, the clarity, aroma, taste, taste lasting and overall acceptability of co-fermentations were significantly improved. In detail, aroma and taste of Sc-Hop and Sc-Hu fermentations were greatly improved, followed by Sc-Td fermentations. Meanwhile, the appearance, taste lasting attribute and overall acceptability of co-fermentations were strongly enhanced. Previous researches reported that co-fermentations of non-Saccharomyces yeast strains, such as H. opuntiae, P. kudriavzevii, H. uvarum, T. delbrueckii etc., and S. cerevisiae could improve the fruity, floral, and mouth-feeling of wine [63][64][65]. Therefore, H. opuntiae, H. uvarum, and T. delbrueckii could be employed to co-ferment with S. cerevisiae to improve the sensory quality of citrus wine. The Sc-Hop fermentation was the best method to produce citrus wine with outstanding aroma, taste, taste lasting, and overall acceptability, followed by fermentations of Sc-Hu and Sc-Td, respectively. The correlation analysis was employed to study the correlation between sensory evaluation scores and aroma substances. These results showed that the main indicators of sensory evaluation were more significantly related to the esters, terpenes, aldehydes, and ketones (p < 0.01) than the other volatile compounds (Table 5). Therefore, the sensory evaluation score of citrus wine was mainly related to the content of ester, terpenes, aldehydes and ketones. Low content of terpenes and aldehydes and ketones also contributed positively to the sensory quality of citrus wine. Generally, many aroma compounds in wine will have a synergistic effect on its flavor, an unbalanced aroma compounds level would have adverse effects on the sensory quality of wine.
In addition, biocontrol strategies for the limitation of undesired microbial growth in foods and beverages by the application of non-Saccharomyces yeasts have been highlighted in recent years [66,67]. Berbegal et al. [68] reported that the use of a mixed starter of different strains of S. cerevisiae or a mix of specific strains of non-Saccharomyces yeast with S. cerevisiae are efficient strategies in the control of the spoilage yeast like Brettanomyces bruxellensis and the volatile phenols' production in wine. Interestingly, the inoculation of different non-Saccharomyces and the inoculation time of the non-Saccharomyces yeast with respect to S. cerevisiae resources (co-inoculated and sequentially inoculated) influence the composition of the connected malolactic fermentation consortia, modulating malolactic fermentation performance [69]. Bold value indicated significant different, and *, ** indicated significant difference at 5% and 1%, respectively.

Conclusions
The selected non-Saccharomyces yeast strains and S. cerevisiae could grow normally during their co-fermentations with the exception of H. uvarum, which was inhibited by S. cerevisiae, and produced citrus wines with low amount of ethanol and total acidity, and high residual sugar content and pH value, and with no significant difference in volatile acid concentration compared with those of S. cerevisiae fermentation. The co-fermentations produced citrus wine with higher content of volatile aroma compounds, especially higher alcohols and esters, than pure S. cerevisiae fermentation did. However, they had no regular effects on the content of fatty acids, terpenes, aldehydes, ketones, and phenols. At the same time, different fermentations including co-fermentations and pure fermentation, produced different aroma compound profiles, and odor active compound profiles, and would present on different flavor complexities and characteristics in citrus wines. More importantly, all the co-fermentations could improve the sensory quality of citrus wine, Sc-Hop co-fermentation produced the best sensory quality of citrus wine, followed by Sc-Hu and Sc-Td co-fermentations. All the results indicated these co-fermentations, producing high content of volatile aroma compounds, especially higher alcohols and esters, benefitted the sensory quality improvement of citrus wine. However, the mechanism that co-fermentation can increase the content of higher alcohols in citrus wine is not known.