The Effect of Co-Culture with Different Pichia kluyveri and Saccharomyces cerevisiae on Volatile Compound and Characteristic Fingerprints of Mulberry Wine

In this study, changes in volatile compounds co-fermented by different Pichia kluyveri with Saccharomyces cerevisiae were analyzed using GC-IMS and compared with S. cerevisiae fermentation, to investigate the production of aroma in mulberry wine during the fermentation process. A total of 61 compounds were accurately identified, including 21 esters, 10 alcohols, 8 aldehydes, 6 ketones, and 19 other volatiles. Compared with the single strain fermentation (S. cerevisiae), the content of 2-methylpropyl acetate, allyl Isothiocyanate, ethyl crotonate, isobutyl propanoate, and butyl 2-methylbutanoate, co-fermentation groups (S. cerevisiae with different P. kluyveri) showed a significant decrease. Alcohols, aldehydes, ketones, and organic acid were lower in both the F(S-P1) and F(S-P2) groups than in the F(S) group throughout fermentation. The 2-methylpentanoic acid only was contained in the F(S) group. The co-fermentation with different P. kluyveri could also be well distinguished. The content of Benzaldehyde and 4-methylphenol in the F(S-P1) group was significantly lower than that in the F(S-P2) group. The PCA results revealed effective differentiation of mulberry wine fermented by different fermentation strains from GC-IMS. The result showed that P. kluyveri could establish a new flavor system for mulberry wine, which plays a crucial role in enhancing the flavor of fruit wine.


Introduction
Mulberry fruit (Mores alba L.) is one of the most popular fruits of the mulberry genus for their nutritional value, flavor, and ability to improve health [1,2].Mulberry fruit is rich in nutrients including fatty acids, amino acids, vitamins, minerals, and bioactive compounds such as anthocyanins, rutin, quercetin, chlorogenic acid, and polysaccharides [3].Due to the abundance of natural compounds it contains, previous studies have shown that mulberry fruit has anti-inflammatory, antioxidant, antibiotic, and hepatoprotective properties [4].
However, the storage and transport performance of fresh mulberries has not been satisfactory [5].Plenty of mulberries are used for fresh consumption and primary processing, such as making juice, mulberry jam, and dried fruit [6].To enrich mulberry products and adjust consumer needs, fruit wine, as a good value-added product, has gained increasing attention in recent years in China.Mulberry wine is one of the most popular fruit wines consumed worldwide [7,8].As a traditional fermented alcoholic beverage with substantial territorial and socio-cultural significance, wine has a long history that is patterned by a tradition of innovation [9].Presently, fruit wine, a different fruit-based fermented alcoholic beverage, has been accepted by people because of its low alcohol content and special flavor characteristics.
Foods 2024, 13, 422 2 of 14 Generally, traditional fruit winemaking often involves fermenting a single yeast [10].The Saccharomyces cerevisiae is a traditional ferment used to produce fruit wine because of its strong alcohol fermentation capacity [11].However, single strain fermentation could make for a similar flavor and less characteristic profiles of the products.In recent years, non-Saccharomyces have received increasing attention.A cohort of scholars alternated to traditional fermentation performed by co-fermentation of S. cerevisiae and non-Saccharomyces, which lead to significant changes in the chemical composition and flavor characteristics of fruit wine [12].Among the non-Saccharomyces yeasts, Pichia kluyveri has been considered a potential yeast in fruit wine processing because of its excellent aroma production ability.It has been reported that P. kluyveri increases the production of glycerol and esters [13].And P. kluyveri could contribute to the production of aroma compounds, such as 2-phenylethyl acetate and 3-sulfanylhexan-1-ol acetate [14,15].In addition, Pichia fermentans, either dead or alive, can make a positive contribution to promote the growth of Oenococcus oeni.And the co-fermentation with Pichia fermentans, improved the production of yeast-derived and grape-derived volatiles, which enhances the fresh fruity aroma of wines [16].Therefore, it is believed that co-fermentation with S. cerevisiae and P. kluyveri could be a promising brewing method to introduce new flavors to mulberry wines.We intended to ferment mulberry juice by co-culturing of S. cerevisiae and P. kluyveri, thereby comparing the flavor differences of mulberry wine between single and co-culture fermentation.
For fruit wine, volatile compounds of flavor play a critical role in wine quality, which directly affects the quality and grade of wine.Typically, GC-MS (gas chromatographymass spectrometry) can separate and identify volatile compounds from complex samples relatively accurately for the purpose of evaluating the flavor of products.However, the detection performance of GC-MS on low molecular weight (below 300 daltons) compounds is not ideal.Small molecules (C2-C10) with a low content of volatile components below 300 daltons, such as some ketones, acids, furans, and other compounds, can be key components in the modification of flavor.Previous studies have shown that these substances can be characteristic of different types of wine aroma.By monitoring these substances, the fermentation process can be precisely controlled.This will improve the quality of the product [17,18].
At present, GC-IMS (gas chromatography-ion mobility spectrometry) has a good detection effect on such small molecule volatile substances [19].Ion mobility spectrometry (IMS) is an analytical technique to detect trace gases and to characterize chemical ionic substances based on the difference in the rate of migration of gas phase ions in an electric field [20].The advantages of gas chromatography ion mobility spectrometry (GC-IMS) is that it requires no sample pretreatment, enables fast analysis timeframes, and has a low detection limit [21], which is a well-established technology that has been widely used in airport security, military security, and drug detection fields nowadays [22].Especially, due to its lower testing and incubating temperature requirements, GC-IMS has promising prospects for food flavor analysis as it does not damage the original compounds of the food [23].Previous research has confirmed that IMS is an effective method for the analysis and characterization of volatile compounds with different properties [24,25].Many studies have also confirmed that GC-IMS technology minimizes loss of flavor substances, which is suitable for food volatile components analysis during storage [26] and processing [27].
Therefore, the aim of this study was to ferment mulberry juice by co-culturing S. cerevisiae and P. kluyveri, and monitor the release of VOCs in mulberry wine during fermentation; then, comparatively analyze the differences of the VOCs between single and co-culture fermentation; and finally, establish the characteristic volatile fingerprints of the VOCs in mulberry wine.On that basis, we can appraise the aroma influence of marker volatile composition in co-fermentation by difference strains so as to understand the immense potential of P. kluyveri in applications.At the same time, the creation of fingerprints of flavor compounds at different stages of fermentation will also contribute to the targeted regulation of flavor during the fermentation process.

Mulberry Juice Fermentation
Squeezed and standardized mulberry juice (Brix, 25 • Bx).The strains S. cerevisia (S), P. kluyveri (P1), and P. kluyveri (P2) were separately pre-cultured in YPD medium at 28 • C with 170 rpm.After 40 h incubation, the yeast cells were separately collected, washed twice, and then resuspended in the 30 mL sterile saline solution.Finally, approximately 10 6 CFU/mL of each yeast strain was transferred into the 500 mL of pasteurized mulberry juice.Fermentation tests were set up as follows: (1) F(S): single inoculation with S. cerevisiae, (2) F(S-P1): co-inoculation of S. cerevisiae (S) and P. kluyveri (P1), (3) F(S-P2): co-inoculation of S. cerevisiae (S) and P. kluyveri (P2), and the inoculum ratio was 1:1 [10].We set the natural fermentation group as the control (CK).The control group and each fermentation broth were fermented at 16 • C for 18 days.Then, the samples were injected into the centrifuge tube and stored at −80 • C in frozen storage every 6 days (0, 6, 12, 18 days) to prepare for the subsequent analyses.

Methods
All samples were analyzed for the presence of volatile compounds using a gas chromatography-ion mobility spectrometry (GC-IMS) instrument (FlavourSpec ® , Gesellschaft für Analytische Sensorsysteme mbH, Dortmund, Germany), which connected with a MXT-WAX capillary column (30 m × 0.53 mm × 1.00 um, RESTEK, Bellefonte, PA, USA).The n-ketones C4-C9 (Sinopharm Chemical Reagent Beijing Co., Ltd., Beijing, China) were used as external references to calculate the retention index (RI) of volatile compounds.The VOCs were identified by comparing the RI and the drift time (in milliseconds) of the standard in the GC-IMS library.
A total of 3 mL of mulberry wine was absorbed, then placed into a 20 mL headspace vial.Subsequently, samples were incubated at 45 • C with 500 rpm for 10 min.After incubation, 500 µL of volatile composition was automatically injected into the injector by means of a heated syringe at 50 • C with splitless mode.The oven temperature was set as 60 • C with isothermal conditions.The samples were driven into the column by nitrogen at a programmed flow as follows: 2 mL/min for 10 min, 15 mL/min for 5 min, 50 mL/min for 5 min, 100 mL/min for 5 min.Then, they were ionized in the IMS ionization chamber of 45 • C. The drift gas flow was set at a constant flow of 150 mL/min.
All analyses were performed in triplicate.Identification of volatiles were identified by comparing retention indices (RI) and the drift time (in milliseconds) of standard in the GC-IMS library [19].

Statistical Analysis
The instrumental analysis software LAV version 2.0.0,G.A.S. (Laboratory Analytical Viewer) and three attached plug-ins were used for sample analysis.Particularly, the program GC-IMS Library Search was applied to analyze the characteristic flavor substances.Then, the Reporter and Gallery plug-ins (G.A.S., Dortmund, Germany) were used to characteristic fingerprint the volatile flavor compounds, and the multivariate data analysis (PCA) was performed using the Dynamic PCA plug-in (G.A.S., Dortmund, Germany).Correlation analysis among volatile compounds was analyzed using Origin 2023 b software (Origin-Lab Corporation, Northampton, MA, USA).

GC-IMS Topographic Plots in Mulberry Wine during Fermentation
The differences of volatile compounds in mulberry wine samples during the fermentation process were analyzed by GC-IMS.As shown in Figure 1, the GC-IMS analysis of sample volatiles resulted in a 3D-topographic plot, where X, Y, and Z-axes, respectively, represent the ion migration time (DT) for identification, the retention time (RT) of the gas chromatograph, and the peak intensity for quantification.The 2D-topographic spectra of volatile compounds during fermentation with treatment samples are shown in Figure 2.
In Figures 1 and 2, the overall blue background was the GC-IMS spectra, and the remarkable red vertical line at abscissa 1.0 was the reactive ion peak (RIP) after normalization.Each heterochromatic or homochromatic point on both sides of the RIP line represented a volatile compound, which indicated the concentration of compound by different levels of color saturation.In general, light color represents lower concentration and redness reflects higher concentration, which means, the darker color indicates the higher concentration of the volatile compound [28].The effect of fermentation with different strains on the volatile compounds is a significant difference.And there are also varying degrees of differences in peak intensity.

GC-IMS Topographic Plots in Mulberry Wine during Fermentation
The differences of volatile compounds in mulberry wine samples during the fermentation process were analyzed by GC-IMS.As shown in Figure 1, the GC-IMS analysis of sample volatiles resulted in a 3D-topographic plot, where X, Y, and Z-axes, respectively, represent the ion migration time (DT) for identification, the retention time (RT) of the gas chromatograph, and the peak intensity for quantification.The 2D-topographic spectra of volatile compounds during fermentation with treatment samples are shown in Figure 2.  In Figures 1 and 2, the overall blue background was the GC-IMS spectra, and the remarkable red vertical line at abscissa 1.0 was the reactive ion peak (RIP) after normalization.Each heterochromatic or homochromatic point on both sides of the RIP line represented a volatile compound, which indicated the concentration of compound by different levels of color saturation.In general, light color represents lower concentration and redness reflects higher concentration, which means, the darker color indicates the higher concentration of the volatile compound [28].The effect of fermentation with different strains on the volatile compounds is a significant difference.And there are also varying degrees of differences in peak intensity.The difference comparison model was applied to compare the aroma differences among the samples, conveniently.The topographical plot of F(S) 6d was taken as the reference and the topographical plots of other samples (including F(S) 12d, F(S) 18d, F(S-P1) 6d, F(S-P1) 12d, and F(S-P1) 18d) were deducted from the reference (Figure 3A).For each dot after deduction appearing in the figure, the red color means a higher concentration of a volatile compound than the reference, blue color indicates a lower concentration of a volatile compound than the reference, and the white dot implies the volatile compounds are consistent with the reference.Similarly, Figure 3B also displayed on this principle, and F(S) 6d was taken as the reference as well.As shown in Figure 3, significant differences between F(S) and F(S-P1) groups were observed, which indicated the difference in molecule structure and concentration of volatile compounds [29].This appearance should be caused by the participation of non-Saccharomyces (P1 and P2).Notably, there were also significant differences among the fermentation broth in different periods.As shown in Figure 3, some volatiles showed significant color differences in the spectra at different stages of fermentation.It is able to show the changes in the content of the substance as the fermentation process takes place.In Figures 1 and 2, the overall blue background was the GC-IMS spectra, and the remarkable red vertical line at abscissa 1.0 was the reactive ion peak (RIP) after normalization.Each heterochromatic or homochromatic point on both sides of the RIP line represented a volatile compound, which indicated the concentration of compound by different levels of color saturation.In general, light color represents lower concentration and redness reflects higher concentration, which means, the darker color indicates the higher concentration of the volatile compound [28].The effect of fermentation with different strains on the volatile compounds is a significant difference.And there are also varying degrees of differences in peak intensity.
(A) The difference comparison model was applied to compare the aroma differences among the samples, conveniently.The topographical plot of F(S) 6d was taken as the reference and the topographical plots of other samples (including F(S) 12d, F(S) 18d, F(S-P1) 6d, F(S-P1) 12d, and F(S-P1) 18d) were deducted from the reference (Figure 3A).For each dot after deduction appearing in the figure, the red color means a higher concentration of a volatile compound than the reference, blue color indicates a lower concentration of a volatile compound than the reference, and the white dot implies the volatile compounds are consistent with the reference.Similarly, Figure 3B also displayed on this principle, and F(S) 6d was taken as the reference as well.As shown in Figure 3, significant differences between F(S) and F(S-P1) groups were observed, which indicated the difference in molecule structure and concentration of volatile compounds [29].This appearance should be caused by the participation of non-Saccharomyces (P1 and P2).Notably, there were also significant differences among the fermentation broth in different periods.As shown in Figure 3, some volatiles showed significant color differences in the spectra at different

Fingerprint Analysis of Volatile Compounds during the Fermentation Process of Mulberry Wine with Different Strains
To further identify the differences of volatile compound profiles in different mulberry wines during fermentation, all information provided to qualitatively characterize by the fingerprint analysis technique.It means to give an overall profile of the aroma composition, not only based on the identification of one or two volatile compounds [30].Therefore, the Gallery Plot plug-in could be used for previewing the volatile fingerprints by showing the color change trend of each point in the 2D plot, which included all volatile substances identified by the GC-IMS instrument (Figure 4).As shown in Figure 4, Gallery Plot listed volatile components together for intuitive comparison, all bright points of each row represented the entire signal peak of a sample, and each column showed the signal strength of the same substance detected in different periods of fermentation.
Therefore, the Gallery Plot plug-in could be used for previewing the volatile fingerprints by showing the color change trend of each point in the 2D plot, which included all volatile substances identified by the GC-IMS instrument (Figure 4).As shown in Figure 4, Gallery Plot listed volatile components together for intuitive comparison, all bright points of each row represented the entire signal peak of a sample, and each column showed the signal strength of the same substance detected in different periods of fermentation.IMS drift times and retention indices were compared with authentic reference compounds to characterize these compounds.In Figure 4, generally, a total of 64 flavor substance signals were measured.Among these peaks, 61 compounds were accurately identified (based on the GC-IMS library, NIST database).Among these volatiles, as shown in Table 1, included 21 esters, 10 alcohols, 8 aldehydes, 6 ketones, and 19 other volatiles such as terpenes and acids (Table 1).It should be noted that there were 3 compounds, which had monomer (M) and dimer (D) form, produced two peak signals by one volatile (Figure 4 and Table 1).
The content of some ester substances in all experimental groups was relatively high in the early stage of fermentation, but showed an overall decreasing trend as fermentation IMS drift times and retention indices were compared with authentic reference compounds to characterize these compounds.In Figure 4, generally, a total of 64 flavor substance signals were measured.Among these peaks, 61 compounds were accurately identified (based on the GC-IMS library, NIST database).Among these volatiles, as shown in  1).It should be noted that there were 3 compounds, which had monomer (M) and dimer (D) form, produced two peak signals by one volatile (Figure 4 and Table 1).The content of some ester substances in all experimental groups was relatively high in the early stage of fermentation, but showed an overall decreasing trend as fermentation progressed.That was probably because the rapid proliferation of yeast during fermentation contributes to the formation of esters [28,31].Compared to the F(S) group, the F(S-P1) and F(S-P2) groups showed a significant decrease in this study.However, no significant difference was found between the F(S-P1) and F(S-P2) groups.As shown in Figure 4A, there were five identified ester substances, including 2-methylpropyl acetate, allyl Isothiocyanate, ethyl crotonate, isobutyl propanoate, and butyl 2-methylbutanoate.The F(S-P1) group substances were lower than those of the F(S) group during fermentation.It was also observed that the incorporation of Pichia (P1 and P2) had a greater effect on the formation of small molecule alcohols (Figure 4A,B).High levels of alcohol were observed in the F(S) group at different stages of fermentation.Alcohols were lower in both the F(S-P1) and F(S-P2) groups than in the F(S) group throughout fermentation.In this study, the levels of ester substances such as butyl 2-methylbutytanoate were negatively correlated with alcoholic substances.This phenomenon might be due to the fact that Pichia could significantly enhance the enzymatic reaction between higher alcohols and acetyl-CoA, thereby producing more esters [32].Overall, the results of this study suggested that small molecule alcohols may also influence the formation of the corresponding esters.This was in line with the previous reports [28].
There were relatively few types of aldehydes and ketones in this study.Two types of carbonyl compounds gradually accumulated during the fermentation process.The F(S-P1) and F(S-P2) groups were significantly lower than the F(S) group.Benzaldehyde had a strong bitter almond and nutty aroma with a slight fruity sweetness [33].There was no significant difference in the content of benzaldehyde between the F(S) group and the F(S-P2) group (Figure 4B), but the production of the F(S-P1) group was lower than that of the other groups throughout the fermentation process (Figure 4A,C).The results suggested that during the process of metabolism of Pichia (P1), the yield of benzaldehyde was consistently lower than that of the other two types of strain.In addition, a certain amount of benzaldehyde was produced during the metabolism of strain S. cerevisiae (S) itself (Figure 4A,B).The lower overall levels in the F(S-P1) group could also be due to the fact that the process of metabolism of P. kluyveri (P1) consumes a large amount of benzaldehyde, resulting in the absence of corresponding odor.
The organic acid substances F(S-P1) and F(S-P2) groups were significantly lower than the F(S) group, and there was no significant difference between the F(S-P1) and F(S-P2) groups.This could prevent the wine from developing an overly sharp acidic taste.But improper control could also make the wine taste monotonous.As shown in Figure 4C, most of the acids accumulated gradually with the fermentation process, while 2-methylpropanoic acid decreased gradually with the fermentation process.The content of 2-methylpropanoic acid in the F(S) group was consistently higher than that in the F(S-P1) and F(S-P2) groups.And the F(S-P1) group showed a significant decrease in content after 6 days, whereas the F(S-P2) group showed the same situation after 12 days.Notably, 2-methylpentanoic acid appeared only in the F(S) group and accumulated with fermentation.It did not appear in the F(S-P1) and F(S-P2) groups.It should be noted that the F(S) group has a distinct grassy taste compared to the other two groups in this study.The 2-methylpentanoic acid was probably partly responsible for this significant difference.This result is also consistent with previous reports [34].
The content of other substances, including ethers, pyrazines and others, was not significantly different among the experimental groups, and there was little change during the fermentation process.These substances were mostly formed in the early stages of fermentation and their levels did not change significantly as the fermentation process progresses (Figure 4A-C).It is noteworthy that both the F(S-P1) and the F(S-P2) groups showed a certain amount of dimethyl disulfide at the late stage of the fermentation process.Sulfur-containing substances, such as dimethyl disulfide, have an odor of their own.It would bring the special bouquet that the sulfur-containing substances had the right amount.However, its high levels could be very detrimental to fruit wines [35].
In addition, the only phenolic substance 4-Methylphenol is described as caramel flavor (threshold, 20 µg/L), which may become a potential hazardous substance [36].In this study, 4-methylphenol levels remained consistently low in the F(S-P1) group (Figure 4A,C).It was higher in the F(S-P2) group than in the F(S) and F(S-P1) groups (Figure 4B,C).The results indicated that Pichia (P1) had a better inhibitory effect on the production of 4-methylphenol.

PCA Analysis of Different Mulberry Wine during Fermentation
In order to distinguish and present the differences among the different mulberry wine samples during fermentation, the principal component analysis (PCA) technique was used to evaluate the differences.In this study, the signal intensity of different volatile compounds identified by GC-IMS was used to structure the PCA model.The PCA of volatile compounds in different mulberry wine samples during fermentation is presented in Figure 5.The distribution map indicated that the total contribution ratio of the first two principal components was 64.6%, which collected 47.5% (PC 1) and 17.1% (PC 2) of the cumulative variance contribution rate.The PCA results exactly indicated that the fermentation with single and co-culture of yeast would be distinguished in the distribution map.The same sample also showed significant stage differences at different stages of fermentation.As show in Figure 5, all samples of F(S) could be separated by the positive score values of PC1, and the F(S-P1) group could be distinguished according to the negative scores of PC1.On the other hand, the score values of PC2 could be used to separate the different fermentation stages of fruit wine.On this foundation, together with biplot of PCA analysis, the result showed that the non-Saccharomyces Pichia kluyveri P1 and P2 could enhance the characteristic flavor of mulberry wine by improving the composition and content of characteristic aroma compounds, leading to the creation of new aromatic mulberry wines.As a result of differential metabolism of non-Saccharomyces, it would provide flavors with different characteristics.This also results in significant differences in wine flavor type.
tation.As show in Figure 5, all samples of F(S) could be separated by the positive score values of PC1, and the F(S-P1) group could be distinguished according to the negative scores of PC1.On the other hand, the score values of PC2 could be used to separate the different fermentation stages of fruit wine.On this foundation, together with biplot of PCA analysis, the result showed that the non-Saccharomyces Pichia kluyveri P1 and P2 could enhance the characteristic flavor of mulberry wine by improving the composition and content of characteristic aroma compounds, leading to the creation of new aromatic mulberry wines.As a result of differential metabolism of non-Saccharomyces, it would provide flavors with different characteristics.This also results in significant differences in wine flavor type.

Conclusions
Aroma is one of the most important means of evaluating and identifying a fruit wine.Fruit wines are frequently made with a particular type of Saccharomyces cerevisiae, which means that the raw materials significantly impact the flavor of the product.It should be noted that traditional methods are advantageous in molding product flavour, but the extension of fruit wine varieties is significantly restricted.Appropriate application of non-Saccharomyces could help fruit wines develop new and more attractive flavors.In this study, a total of 64 signal peaks from topographic plots were identified in different mulberry wine samples.The specificity of P. kluyveri (P1, P2) in regulating the production of 2-methylpentanoic acid, benzaldehyde and, conversely, 4-methylphenol is reported for the first time in this article, and deserves further investigation in future research.Additionally, the VOCs fingerprint also demonstrates the process of change in the content of different volatile substances at different stages of fermentation.In addition, the results of PCA showed that the different mulberry wines were well distinguished.Therefore, non-Saccharomyces will provide more options for creating fruit wine flavors and has great research potential for producing new types of fruit wine.
By constructing effective fingerprints, key flavor compounds can be identified.This allows rapid differentiation of the flavor characteristics of wines fermented by different non-Saccharomyces.This feature can be used to create fingerprint maps during fermentation to monitor changes in the levels of the key flavor compounds.It also means that by regulating the fermentation process based on changes in key flavor compounds, the quality of wine can be improved.

ds 2024 ,
13, x FOR PEER REVIEW 4 of 15Correlation analysis among volatile compounds was analyzed using Origin 2023 b software (Origin-Lab Corporation, Northampton, MA, USA).

Figure 1 .
Figure 1.3D-topographic plots in different mulberry wine samples during fermentation.(A) group F(S); (B) group F(S-P1); (C) group F(S-P2).In Figure2, all the mulberry wine samples are rich in volatile compounds.It can be seen that most of the peak signals appeared in the retention time of 100-900 s, and a few signals appeared in the retention time of 900-1400 s.The peak signals were observed in ranges of drift time of 1.0-2.0ms.The difference comparison model was applied to compare the aroma differences among the samples, conveniently.The topographical plot of F(S) 6d was taken as the reference and the topographical plots of other samples (including F(S) 12d, F(S) 18d, F(S-P1) 6d, F(S-P1) 12d, and F(S-P1) 18d) were deducted from the reference (Figure3A).For each dot after deduction appearing in the figure, the red color means a higher concentration of a volatile compound than the reference, blue color indicates a lower concentration of a volatile compound than the reference, and the white dot implies the volatile compounds are consistent with the reference.Similarly, Figure3Balso displayed on this principle, and F(S) 6d was taken as the reference as well.As shown in Figure3, significant differences between F(S) and F(S-P1) groups were observed, which indicated the difference in molecule structure and concentration of volatile compounds[29].This appearance should be caused by the participation of non-Saccharomyces (P1 and P2).Notably, there were also significant differences among the fermentation broth in different periods.As shown in Figure3, some volatiles showed significant color differences in the spectra at different stages of fermentation.It is able to show the changes in the content of the substance as the fermentation process takes place.

Figure 4 .
Figure 4. Fingerprints of volatile compounds in different mulberry wine samples during fermentation.(A) group F(S) vs. group F(S-P1); (B) group F(S) vs. group F(S-P2); (C) group F(S-P1) vs. group F(S-P2).I refers to esters, II refers to alcohols, III refers to aldehydes, IV refers to ketones, V refers to acids, VI refers to ethers, VII refers to sulpho compounds, VIII refers to pyrazine, Ⅸ refers to terpenes, Ⅹ refers to phenol, Ⅺ refers to thiazole, and Ⅻ refers to nitrogen compound.

Figure 4 .
Figure 4. Fingerprints of volatile compounds in different mulberry wine samples during fermentation.(A) group F(S) vs. group F(S-P1); (B) group F(S) vs. group F(S-P2); (C) group F(S-P1) vs. group F(S-P2).I refers to esters, II refers to alcohols, III refers to aldehydes, IV refers to ketones, V refers to acids, VI refers to ethers, VII refers to sulpho compounds, VIII refers to pyrazine, IX refers to terpenes, X refers to phenol, XI refers to thiazole, and XII refers to nitrogen compound.

Figure 5 .
Figure 5. PCA analysis of flavor compounds mulberry wine with different strains during the fermentation process.(A,B) The score plot and biplot of PCA analysis (with CK).(C,D) The score plot and biplot of PCA analysis (without CK).

Table 1 .
GC-IMS integration parameters of volatile compounds obtained from three different mulberry wine samples.
a MW: molecular mass; b RI: retention index; c RT: retention time; d Dt: drift time.