Juice turbidity and fermentation method: key drivers of aroma complexity in Italian Riesling grape spirit production

Abstract

Juice turbidity and fermentation mode are key factors affecting aroma formation in grape spirit production, but their combined effects remain unclear. In this study, Italian Riesling juice under two turbidity conditions (clear and turbid) was fermented either spontaneously or with commercial yeast, generating four treatments. Physicochemical properties, volatile compounds, odor activity values, and sensory characteristics of base wines and grape spirits were compared. Both factors significantly influenced base-wine composition and these differences were reflected in the distillates. Spontaneous fermentation increased aroma diversity and complexity, whereas inoculated fermentation produced a more uniform volatile profile. Clear juice favored floral and fruity esters, while turbid juice promoted fermentation-derived aroma precursors. Distillation further amplified treatment-dependent volatile differences, resulting in distinct aromatic styles. These results show that juice turbidity and fermentation mode jointly shape aroma development from fermentation to distillation and provide guidance for targeted aroma modulation in Italian Riesling grape spirit production.

Graphical abstract

Highlights

• •Juice turbidity and fermentation mode jointly shaped the aroma profile of grape spirits.
• •Spontaneous fermentation increased aroma diversity and overall compositional complexity.
• •Clear juice favored the formation of floral and fruity aroma-active esters.
• •Turbid juice promoted fermentation-derived precursors associated with aroma richness.
• •Distillation amplified treatment-dependent differences in volatile composition.

1. Introduction

Grape spirit is a distilled alcoholic beverage produced from fermented grape juice or wine. Its aroma quality is strongly influenced by the volatile composition of the base wine before distillation, especially in unaged or lightly aged spirits where oak maturation contributes little to aroma modification. Therefore, regulating aroma formation during base-wine fermentation is essential for improving the sensory quality of grape spirits (Balák et al., 2024; Basile et al., 2025; Pinu et al., 2023).

Fermentation method is a major determinant of base-wine aroma. Spontaneous fermentation involves indigenous microorganisms, whose succession and interactions can contribute to aroma complexity and regional typicity, but it is often associated with greater variability and lower predictability (Boban et al., 2024; S. Li et al., 2025; Mas & Portillo, 2022; Tufariello et al., 2021; Zhu et al., 2023). In contrast, inoculation with commercial Saccharomyces cerevisiae strains provides better fermentation control and more stable aroma formation through strain-dependent metabolic pathways, although single-strain fermentation may reduce aroma diversity under certain conditions (L. Wang et al., 2024).

Juice turbidity is another important pre-fermentation parameter. By altering the levels of suspended solids, grape cell wall fragments (Balák et al., 2024), lipids, phenolic compounds, and nutrients, clarification can modify the physicochemical and nutritional environment of fermentation and consequently influence the formation of esters, higher alcohols, fatty acids, terpenes, and other volatile compounds (Z. Liang et al., 2021; Z. Zhang et al., 2022). However, turbidity effects are not uniform and may vary with grape cultivar, yeast strain, nutrient status, and fermentation conditions (Y. Wang et al., 2025).

Despite these findings, most previous studies have examined fermentation method or juice turbidity separately in wine systems (X. Zhang, Zhao, et al., 2025; Santos et al., 2023). Their interactive effects in grape spirit production, particularly their effects on volatile formation in base wine and on the transfer, enrichment, or attenuation of aroma compounds during distillation, remain unclear. In addition, traditional grape spirits are commonly produced from varieties such as ‘Ugni Blanc’, ‘Colombard’, ‘Folle Blanche’, and ‘Montils’ (Lončarić et al., 2022), whereas Italian Riesling, an increasingly cultivated variety in Chinese wine regions such as Ningxia, has rarely been evaluated for spirit production (Padilla et al., 2016; L. Wang et al., 2023; Xia et al., 2023).

Therefore, the novelty of this study lies in evaluating the combined effects of fermentation method and juice turbidity on Italian Riesling grape spirit production within an integrated base-wine–distillate framework. Italian Riesling grapes from Ningxia were fermented using commercial yeast inoculation or spontaneous fermentation under clear-juice or turbid-juice conditions, followed by double distillation. Volatile compounds in base wines and spirits were analyzed by HS-SPME-GC–MS, and multivariate statistics, two-way ANOVA, and sensory evaluation were used to assess chemical and sensory differences. This study aimed to determine how fermentation method and juice turbidity regulate volatile compound formation in base wines and their subsequent transfer to grape spirits, and to infer possible metabolic implications of juice turbidity in modulating yeast-associated aroma synthesis. The findings may support aroma-targeted process optimization and the development of distinctive regional grape spirits in China.

2. Materials and methods

2.1. Raw material collection and fermentation process

Italian Riesling grapes (Vitis vinifera L. cv. Italian Riesling) were collected from the Ganchengzi area of Qinglongxia, Ningxia, China (38°06′25.19″ N, 105°56′11.11″ E). The region has a temperate continental arid climate, with an annual precipitation of 173–179 mm, an altitude of 1100–1200 m, and grey calcareous and light grey calcareous soils. Grapes were hand-harvested at technological maturity on September 1, 2022, carefully selected, and transported at ambient temperature to Yuma International Wine Industry (Ningxia) Co., Ltd. for fermentation. The basic parameters of the grapes were as follows: soluble solids, 203 g/L; pH, 3.24; and total acidity, 8.4 g/L.

After destemming and crushing, no sulfur dioxide (SO₂) was added. The grape must was naturally settled at 15 °C for 24 h. After sedimentation, the upper relatively clear fraction was separated and defined as clear juice (Q), whereas the lower fraction containing more suspended particles and grape-derived solids was defined as turbid juice (Z). In this study, clear juice and turbid juice were operationally defined based on natural settling behavior rather than on a predefined turbidity threshold. After separation, pH, total acidity, reducing sugar, alcohol content, and volatile acidity were measured to characterize the compositional differences between the two juice conditions before fermentation.

A two-factor experimental design was adopted, including two fermentation modes and two juice conditions. The fermentation modes were spontaneous fermentation (ZR) and inoculated fermentation with commercial yeast FC9 (FC), while the juice conditions were clear juice (Q) and turbid juice (Z). Accordingly, four experimental groups were established: ZR-Q, ZR-Z, FC-Q, and FC-Z. Fermentations were carried out in 20 L food-grade polyethylene tanks filled to 80% of their total volume, and each treatment was performed in triplicate using independent fermentation vessels.

For the inoculated fermentation groups, Lalvin FC9® active dry yeast (Saccharomyces cerevisiae; CICC 33077, Lallemand) was added at 200 mg/L. Before inoculation, the yeast was activated with grape juice according to winery practice and then added to the corresponding fermentation tanks. For spontaneous fermentation, no commercial yeast was added, and fermentation proceeded with the indigenous microbiota naturally present in the grape material and winery environment. Alcoholic fermentation was conducted at 14–21 °C without SO₂ addition. Fermentation progress was monitored daily by measuring reducing sugar using the Lane–Eynon method, and primary fermentation was considered complete when the residual sugar concentration decreased to 4 g/L. This residual sugar level was selected as a near-dry endpoint to indicate that alcoholic fermentation was essentially complete and to reduce the potential influence of remaining sugars on subsequent distillation and aroma analysis.

2.2. Distillation procedure

The commercial yeast used in the inoculated fermentation treatment was Lalvin FC9® (Saccharomyces cerevisiae; CICC 33077, Lallemand). This strain was originally selected for producing base wines intended for Cognac and brandy distillation. In this study, it was used as the starter culture for the FC treatments, and its effect was evaluated by comparison with the corresponding spontaneous fermentation treatments.

After completion of alcoholic fermentation, the base wines were distilled using 20 L copper Charente pot stills. Distillation was carried out under identical operating conditions for all treatments. For the first distillation, 20 L of base wine with an alcohol content of approximately 11.5%–12.5% (v/v) was introduced into the still and heated by direct steam. The distillate began to flow at approximately 65 °C, and the temperature was gradually raised to 85 °C and maintained until distillation ceased.

The collection of the heart fraction was performed strictly according to the procedures described in Brandy Technology by Gongtang Wang. Before distillation, the expected volume of each fraction was estimated according to the base wine volume and the preset fraction proportion. During distillation, fraction separation was further guided and verified by real-time monitoring of alcohol content. Therefore, the separation of heads, hearts, and tails was based on both the pre-estimated distillate volume and the change in alcohol content during distillation.

During the first distillation, the distillate was divided into heads, hearts, and tails. The heads fraction, corresponding to the initial distillate enriched in low-boiling compounds, was discarded. The hearts fraction was retained as the target distillate, while the tails fraction, enriched in high-boiling compounds, was either reused or discarded according to production practice. For the second distillation, 2 L of primary distillate with an alcohol content of approximately 27% (v/v) was subjected to fractional distillation. The fractions were collected as follows: heads, approximately 70%–58% (v/v) alcohol content and collected over approximately 30 min; hearts 1, approximately 58%–20% (v/v) alcohol content and collected over approximately 6 h; hearts 2, approximately 20%–0% (v/v) alcohol content and collected over approximately 4 h; and tails, corresponding to the residual liquid in the still and discarded.

The samples were coded according to processing stage: base wine samples were prefixed as YJ, primary distillation samples as YC, and secondary distillation samples as EC. These samples were used for subsequent physicochemical and volatile compound analyses.

2.3. Physicochemical analysis and HPLC determination of non-volatile compounds

The basic physicochemical parameters of grape juice, base wine, and distillate samples were determined according to routine analytical methods. Soluble solids, pH, total acidity, reducing sugar, alcohol content, and volatile acidity were measured to characterize the fermentation substrates and products.

Major non-volatile compounds were quantified by high-performance liquid chromatography (HPLC) using a Shimadzu LC-2050C 3D system (Shimadzu, Japan), following previously reported methods with minor modifications (Guo et al., 2024; Pu et al., 2023). Chromatographic separation was performed using an Aminex HPX-87H column (300 mm × 7.8 mm; Bio-Rad, USA). The detector system consisted of a UV detector and a refractive index detector (RID-20 A) connected in series. The UV detector was set at 210 nm for organic acid quantification, whereas the refractive index detector was used for the quantification of sugars and ethanol. The mobile phase was 5 mM H₂SO₄ aqueous solution under isocratic elution at a flow rate of 0.6 mL/min, and the column temperature was maintained at 60 °C (Hranilovic et al., 2020). All samples were filtered through a 0.22 μm membrane before analysis, and each sample was analyzed in triplicate.

The quantified non-volatile compounds included glucose, fructose, glycerol, ethanol, tartaric acid, malic acid, lactic acid, acetic acid, succinic acid, and citric acid. Compound identification and quantification were performed using external standards.

2.4. HS-SPME-GC–MS analysis of volatile aroma compounds

Volatile aroma compounds in wine and distillate samples were analyzed by headspace solid-phase microextraction combined with gas chromatography–mass spectrometry (HS-SPME-GC–MS), using a Thermo Scientific Trace 1610 gas chromatograph coupled with an ISQ 7610 mass spectrometer (Thermo Scientific, USA).

For sample preparation, 5 mL of sample was transferred into a 20 mL headspace vial, followed by the addition of 1.2 g NaCl and 10 μL of internal standard solution (4-methyl-2-pentanol, 1.0031 g/L). A DVB/CAR/PDMS fiber (50/30 μm) was used for extraction. The sample was equilibrated at 40 °C for 30 min and then extracted for 30 min under magnetic stirring at 500 rpm.

After extraction, the fiber was immediately inserted into the GC injection port and desorbed at 250 °C for 8 min. Separation was performed on a TG-WAXMS capillary column (60 m × 0.32 mm × 0.25 μm). High-purity helium (99.999%) was used as the carrier gas at a constant flow rate of 1.0 mL/min. The oven temperature program was as follows: 50 °C for 1 min, increased to 220 °C at 3 °C/min, and held at 220 °C for 5 min (Chen et al., 2025).

Mass spectrometric detection was performed in electron ionization mode. The ion source temperature was 230 °C, the transfer line temperature was 250 °C, and the scan range was m/z 35–550. Compounds were identified by comparison with the NIST mass spectral library, retention index matching, and authentic standards when available. Quantification was based on calibration curves prepared in a synthetic wine matrix containing 8% ethanol, 3.25 g/L tartaric acid, and 1.5 g/L citric acid at pH 3.3. Odor activity values (OAVs) were calculated as the ratio of compound concentration to the corresponding odor threshold.

2.5. Sensory aroma evaluation

Sensory aroma evaluation was performed after completion of alcoholic fermentation and preliminary clarification of the wine samples. The sensory panel consisted of 12 trained assessors, including 6 females and 6 males. Before formal evaluation, all panelists underwent 3 months of intensive aroma recognition training using the 54-aroma standard kit Le Nez du Vin, and their recognition accuracy for characteristic aroma descriptors was maintained above 95%.

Sensory evaluation was conducted in a standardized tasting room maintained at 23 °C. Approximately 30 mL of each sample was poured into a black professional tasting glass and presented in a completely randomized blind manner. Only orthonasal aroma evaluation was performed, and no tasting was conducted.

The evaluation procedure consisted of static smelling for 5–8 s, followed by swirling and smelling again for 5–10 s after aroma release. Based on a predefined aroma descriptor list, each assessor selected 4–6 terms that best matched the aroma characteristics of the sample and rated the intensity of each aroma attribute, such as fruity, floral, fermentative, and mature notes, using a 5-point scale, where 1 indicated very weak, 3 indicated moderate, and 5 indicated extremely strong intensity.

Ethical approval was not required for this sensory evaluation according to the institutional and national regulations applicable to this type of study. All procedures were conducted in accordance with appropriate protocols to protect the rights, privacy, and well-being of the participants. All participants were informed about the study and provided their voluntary informed consent prior to participation.

2.6. Statistical analysis

All experiments were performed with three replicates, and the results are expressed as mean ± standard deviation. Basic statistical analyses were conducted using Microsoft Excel 2021. Significant differences among treatments were evaluated by one-way analysis of variance (ANOVA) followed by Tukey's honestly significant difference test at p < 0.05 using SPSS version 27.0 (IBM Corp., Armonk, NY, USA). Two-way ANOVA was used to evaluate the main effects of fermentation mode and juice condition, as well as their interaction effects.

Principal component analysis (PCA) was performed using OriginPro 2024b (OriginLab Corporation, Northampton, MA, USA). Orthogonal partial least squares-discriminant analysis (OPLS-DA) was performed using SIMCA version 14.1. Model validity was assessed by 200 permutation tests. Variables with variable importance in projection (VIP) values greater than 1 were regarded as important contributors to sample discrimination. The VIP > 1 criterion was used because it indicates that the corresponding variable contributes more than the average level to the classification model (Ge et al., 2026; Han et al., 2026; Milheiro et al., 2026; X. Zhang, Zhang, et al., 2025).

3. Results and discussion

3.1. Physicochemical properties of grape base spirits

The physicochemical properties of grape base spirits produced under different fermentation modes and juice turbidity conditions are shown in Table 1. Significant differences were observed among treatments in residual sugars, ethanol, glycerol, and organic acid composition.

Table 1.

Basic Physicochemical Indicators of Grape Base Spirit.

Compounds	FC-Z		FC-Q		ZR-Z		ZR-Q
glucose	0.25	a	0.27	a	0.44	b	0.56	b
fructose	0.14	b	0.02	a	0.05	a	nd	a
ethanol	12.23	a	12.31	b	12.35	a	11.96	a
glycerol	6.29	b	4.34	a	7.15	bc	7.80	c
acetic acid	0.12	a	0.21	a	0.04	a	0.62	b
lactic acid	1.78	a	1.26	b	1.72	a	1.22	b
tartaric acid	0.05	bc	0.04	ab	0.05	c	0.04	a
succinic acid	0.06	b	0.03	a	0.06	b	0.06	b
citric acid	4.64	a	4.45	a	4.64	a	5.92	b
malic acid	0.09	bc	nd	a	0.15	c	0.04	ab

Note: Alcohol in % volume (% v/v) and the rest of the substance in g/L. Data are means of four parallel sets of samples determined independently; different letters in the rows indicate significant differences (p < 0.05) between the raw wines of different process treatments, which were analyzed by ANOVA.

Glucose concentrations ranged from 0.25 to 0.56 g/L. The inoculated fermentation treatments showed lower glucose contents, with 0.25 g/L in FC-Z and 0.27 g/L in FC-Q, whereas the spontaneously fermented treatments showed relatively higher levels, reaching 0.44 g/L in ZR-Z and 0.56 g/L in ZR-Q. Fructose contents were generally low, with 0.14 g/L in FC-Z, 0.02 g/L in FC-Q, and 0.05 g/L in ZR-Z, while no fructose was detected in ZR-Q.

Ethanol contents ranged from 11.96% to 12.35% (v/v). The highest ethanol concentration was observed in ZR-Z (12.35% v/v), followed by FC-Q (12.31% v/v) and FC-Z (12.23% v/v), while ZR-Q had the lowest value (11.96% v/v). Glycerol concentrations differed markedly among treatments, with the highest value in ZR-Q (7.80 g/L), followed by ZR-Z (7.15 g/L) and FC-Z (6.29 g/L), whereas FC-Q showed the lowest level (4.34 g/L).

Among the organic acids, acetic acid was highest in ZR-Q (0.62 g/L), exceeding the levels in FC-Z (0.12 g/L), FC-Q (0.21 g/L), and ZR-Z (0.04 g/L). Lactic acid was higher in FC-Z (1.78 g/L) and ZR-Z (1.72 g/L) than in FC-Q (1.26 g/L) and ZR-Q (1.22 g/L). Citric acid reached its highest level in ZR-Q (5.92 g/L), while the other three treatments showed similar values between 4.45 and 4.64 g/L. Malic acid was not detected in FC-Q, but was present at 0.09 g/L in FC-Z, 0.15 g/L in ZR-Z, and 0.04 g/L in ZR-Q.

Overall, both fermentation mode and juice turbidity were associated with differences in the basic composition of the base wines before distillation.

3.2. Volatile composition of grape base spirits and distillates

The volatile compounds identified in grape base spirits and distillates are summarized in Table S1. These compounds were classified into acetates, ethyl esters, other esters, phenols, aldehydes and ketones, fatty acids, higher alcohols, C6 alcohols, and terpenes. Their concentrations varied substantially among treatments and across the three processing periods.

Among the acetate esters, ethyl acetate was one of the most abundant compounds throughout the process. In the base spirit stage, its concentration ranged from 38,701.42 μg/L in FC-Q to 125,226.40 μg/L in ZR-Q. After primary distillation, it remained at relatively high levels, particularly in ZR-Q (111,046.10 μg/L) and ZR-Z (87,930.62 μg/L). In the secondary distillates, the highest concentration was observed in FC-Q (593,791.75 μg/L), whereas FC-Z remained much lower (6732.48 μg/L).

Several ethyl esters, including ethyl butyrate, ethyl hexanoate, and ethyl octanoate, showed marked increases in distilled samples. For example, ethyl hexanoate increased from 230.85 to 421.78 μg/L in the base spirits to 5860.04 μg/L in the secondary distillate of FC-Q. Ethyl octanoate reached 10,572.28 μg/L in the secondary distillate of FC-Q, and ethyl butyrate reached 21,273.16 μg/L in the same sample.

Stage-related differences were also observed for other chemical groups. Some terpene compounds, such as geraniol, maintained relatively high levels in certain base spirits and primary distillates, especially in ZR-Z and ZR-Q. 2,4-Di-tert-butylphenol was detected at comparatively high levels in several samples and showed higher concentrations in some distilled fractions. In addition, acetoin was present at notably high levels in FC-Q, particularly after secondary distillation.

Overall, the volatile composition changed substantially across processing periods, and the magnitude and direction of these changes depended on treatment.

3.3. Odor activity values of volatile compounds

To evaluate the potential aroma contribution of individual volatile compounds, odor activity values (OAVs) were calculated for compounds with available odor thresholds, and the results are shown in Tables S2–S4. Compounds with OAV > 1 were considered odor-active (W. Gao et al., 2014; Huang et al., 2023; L. Liang et al., 2023; X. Wang et al., 2024; Willner et al., 2013; Xiang et al., 2020; B. Zhang et al., 2022).

In the base spirit samples (Table S2), geraniol showed consistently high OAVs across all treatments, ranging from 11.60 to 18.38, indicating an important floral contribution at this stage. Ethyl octanoate and ethyl decanoate also showed OAVs above 1 in several treatments. By comparison, linalool, α-terpineol, and citronellol generally remained below the odor threshold in most base spirit samples.

In the primary distillates (Table S3), the number of odor-active compounds increased. Ethyl acetate showed OAVs of 14.81 in ZR-Q, 11.72 in ZR-Z, and 9.06 in FC-Q, but only 0.68 in FC-Z. Hexyl acetate became odor-active in ZR-Z and FC-Q, while propionic acid also showed OAVs above 1 in ZR-Q, ZR-Z, and FC-Q. Benzoic acid maintained relatively high OAVs across treatments, and geraniol remained an important odor-active compound in ZR-Q and ZR-Z.

In the secondary distillates (Table S4), treatment-dependent differences became more evident. FC-Q showed particularly high OAVs for several esters, including ethyl acetate (79.17), ethyl butyrate (607.80), ethyl hexanoate (418.57), and hexyl acetate (33.84). By contrast, FC-Z showed a relatively high OAV for phenylethyl acetate (27.45), suggesting a different odor-active profile. ZR-Q and ZR-Z also retained several odor-active esters, but their overall ester-related OAV levels were lower than those of FC-Q.

Taken together, the OAV results indicate that the sensory relevance of volatile compounds differed not only among treatments but also among processing periods, with the secondary distillates showing the clearest differences.

3.4. Changes in major volatile classes across the three processing periods

The concentrations of major volatile classes in grape base spirits, primary distillates, and secondary distillates are shown in Fig. 1, including ethyl esters (A), acetates (B), other esters (C), C6 alcohols (D), aldehydes and ketones (E), fatty acids (F), higher alcohols (G), phenols (H), and terpenes (I). According to Tukey's test (p < 0.05), values marked with different letters differ significantly.

Fig. 1.

Concentrations of volatile compounds in grape distillates from three processing periods, including ethyl esters (A), acetates (B), other esters (C), C6 alcohols (D), aldehydes and ketones (E), fatty acids (F), higher alcohols (G), phenols (H) and terpenes (I, according to Tukey's test (p < 0.05), values labeled with different letters in the chart are statistically significant.

Clear differences were observed among both treatments and processing periods. In general, several classes, especially esters, aldehydes and ketones, phenols, fatty acids, and higher alcohols, showed higher overall levels in distilled samples than in base spirits. However, the extent of change differed across treatments.

The most pronounced treatment differences were observed in the secondary distillates. FC-Q showed notably higher levels for several volatile classes, particularly acetate esters, ethyl esters, aldehydes and ketones, fatty acids, higher alcohols, and phenols. In contrast, some naturally fermented treatments showed relatively stronger terpene representation in earlier stages. C6 alcohols varied less dramatically than several other classes, while terpenes showed a more treatment-dependent pattern across the three periods.

Overall, Fig. 1 indicates that both processing stage and treatment influenced the distribution of volatile classes, and that the secondary distillation stage contributed strongly to the differentiation of the final distillate profiles.

3.5. Principal component analysis of odor-active compounds

Principal component analysis (PCA) was performed using 25 odor-active compounds (OAV > 1), and the results are shown in Fig. 2. Fig. 2A–C present the score plots for grape base spirits, primary distillates, and secondary distillates, respectively, while Fig. 2D–F show the corresponding loading plots.

Fig. 2.

Principal Component Analysis (PCA) diagram of 25 volatile aromatic compounds (OAV > 1) in grape base spirit and grape distillate samples. Fig. A, B, and C correspond to samples from different periods, labeled with distinct colors. Fig. D, E, and F respectively show the distribution of the 25 aroma compounds (principal component loadings) within their respective periods.

For the grape base spirits (Fig. 2A), PC1 and PC2 explained 40.0% and 24.2% of the total variance, respectively. The four treatments were separated in the score plot, indicating differences in odor-active composition before distillation.

For the primary distillates (Fig. 2B), PC1 and PC2 explained 43.4% and 22.3% of the total variance, respectively. The relative positions of the treatments changed compared with the base spirit stage, indicating that the first distillation step altered the overall distribution of odor-active compounds.

For the secondary distillates (Fig. 2C), PC1 explained 52.3% and PC2 explained 15.9% of the total variance. Separation among treatments remained evident, with FC-Q showing clear distinction from the other treatments. According to the loading plots (Fig. 2D–F), multiple compound groups, including esters, acids, terpene alcohols, and phenolic compounds, contributed jointly to the sample separation patterns.

These PCA results indicate that odor-active composition differed among treatments at all three processing periods, with stronger separation observed after distillation.

3.6. OPLS-DA discrimination and model validation

To further examine differences among treatments based on odor-active compounds, orthogonal partial least squares-discriminant analysis (OPLS-DA) was performed, and the results are shown in Fig. 3. Fig. 3A–C present the score plots for the three processing periods, and Fig. 3D–F show the corresponding 200-permutation tests.

Fig. 3.

Orthogonal Partial Least Squares Discriminant Analysis (OPLS-DA) of 25 aroma compounds with OAV > 1 in grape base spirit and grape distillate samples. Fig. A, B, and C correspond to samples from different periods, with samples labeled in different colors. Fig. D, E, and F show the results of 200 permutation tests for the corresponding periods.

At the grape base spirit, primary distillation, and secondary distillation stages, the score plots showed clustering patterns corresponding to treatment groups, indicating that the selected odor-active compounds were able to distinguish samples from different treatments. This trend was consistent with the PCA results.

The permutation tests were used to assess model validity. For grape base spirits, the model yielded R² = 0.499 and Q² = −0.139; for primary distillates, R² = 0.261 and Q² = −0.777; and for secondary distillates, R² = 0.334 and Q² = −0.57. The negative Q² intercepts suggest that the models were not obviously overfitted. However, because the Q² values were relatively low, the predictive performance of the models was limited. Therefore, the OPLS-DA results were used only as an auxiliary visualization tool, and the screening of differential co'mpounds should be interpreted with caution.

Overall, the OPLS-DA results support the existence of treatment-related differences in odor-active profiles, particularly after distillation, but they should be interpreted cautiously.

3.7. VIP-based screening of differential odor-active compounds in secondary distillates

The top 15 odor-active compounds with the highest VIP values in the secondary distillate samples are shown in Fig. 4. All selected compounds had VIP > 1, indicating that they contributed substantially to the discrimination among treatments in the secondary distillation stage.

Fig. 4.

Selection of the top 15 odor-active compounds with VIP values in secondary distilled wine samples, with red squares indicating higher concentrations and blue squares indicating lower concentrations. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

The selected compounds included citronellol, diethyl succinate, trans-β-ionone, 1-propanol, acetic acid, phenylethanol, phenylethyl acetate, ethyl butyrate, propionic acid, ethyl 3-methylbutyrate, hexyl acetate, ethyl caproate, 1-octen-3-ol, 2,4-di-tert-butylphenol, and ethyl carbamate. These compounds belong to several chemical classes, indicating that the differences among final distillates were not associated with a single aroma group.

Based on the heatmap pattern, FC-Q showed relatively higher levels of several ester-related compounds, including ethyl butyrate, ethyl 3-methylbutyrate, hexyl acetate, and ethyl caproate. In contrast, FC-Z showed relatively higher levels of diethyl succinate, trans-β-ionone, phenylethanol, and phenylethyl acetate. ZR-Q was characterized by relatively higher levels of citronellol, 1-propanol, acetic acid, and propionic acid, whereas ZR-Z generally showed lower relative abundance for several of the selected VIP compounds.

Thus, the VIP analysis highlighted a set of differential odor-active compounds associated with treatment separation in the secondary distillates.

3.8. Contribution of fermentation mode, juice turbidity, and their interaction to odor-active compounds

The effects of fermentation mode, juice turbidity, and their interaction on odor-active compounds are summarized in Fig. 5. The left panel presents the normalized Z-score distribution of compounds with OAV > 1 among the sample groups, whereas the right panel shows the proportions of variance explained by fermentation mode, juice turbidity, their interaction, and residuals (RS) based on two-way ANOVA.

Fig. 5.

Analysis of Volatile Characteristics in Grape Base Spirit and Grape Distillate. Compounds with OAV values >1 were selected from volatile compounds, and their concentrations in grape base spirit and grape distillate were determined (left figure). A two-way ANOVA was employed to determine the percentage variance explained by yeast strain, process selection, their interaction, and residuals (RS) (right figure).

The Z-score distribution indicates that the relative abundance patterns of odor-active compounds differed across treatments and processing stages. The secondary distillate samples showed more pronounced divergence for several compounds than the earlier stages, especially for ester-related variables.

The variance partitioning further showed that different compounds were influenced to different extents by the tested factors. For several esters, including ethyl acetate, hexyl acetate, phenylethyl acetate, ethyl butyrate, ethyl 3-methylbutyrate, ethyl caproate, and ethyl caprylate, a substantial proportion of variance was associated with process selection and the interaction term. In comparison, compounds such as 1-propanol and phenylethanol appeared to show a relatively stronger association with the yeast strain factor. Some compounds, including capric acid and benzoic acid, showed relatively high residual contributions, indicating that additional unmeasured factors may also have contributed to their variation.

Overall, Fig. 5 suggests that the variation of odor-active compounds was associated with fermentation mode, juice turbidity, and their interaction, although the relative importance of these factors differed among compounds.

3.9. Sensory profiles of the distilled spirit samples

The sensory profiles of the four distilled spirit samples are shown in Fig. 6, which includes the attributes fruity, flower, woody, nutty, spice, complexity, and pleasure. The radar plot indicates perceptible differences among the four treatments.

Fig. 6.

Radar chart of sensory analysis of four distilled spirits samples.

Among the samples, FC-Q showed relatively high scores for fruity, complexity, and pleasure, and also displayed a comparatively strong woody character. ZR-Q also showed relatively high intensities for fruity, flower, nutty, and pleasure. ZR-Z showed intermediate intensities for most attributes. By contrast, FC-Z generally exhibited lower scores for several attributes, particularly fruity, spice, and complexity.

Overall, the sensory results indicate that the four distilled spirits differed in aroma perception, and that FC-Q and ZR-Q showed comparatively stronger overall sensory expression than the other two treatments.

4. Discussion

4.1. Effects of fermentation mode on aroma composition

Fermentation mode influenced both the basic physicochemical composition and the volatile profile of the base wines and distilled spirits. In the present study, spontaneous fermentation generally retained higher residual glucose and glycerol concentrations than inoculated fermentation, while treatment-dependent differences were also observed for acetic acid, lactic acid, and citric acid. These results indicate that different fermentation modes were associated with different substrate utilization patterns and fermentation outcomes, which is consistent with previous reports showing that spontaneous and inoculated fermentations often diverge in sugar consumption behavior and by-product formation due to differences in microbial succession and fermentation kinetics (Y. Gao et al., 2026; L. Li et al., 2025; Luo et al., 2024).

Differences related to fermentation mode were also reflected in the volatile composition. In particular, the OAV and multivariate analyses showed that odor-active profiles differed between spontaneous and inoculated fermentations across all three processing periods. In the secondary distillates, some compounds such as phenylethanol and 1-propanol appeared to be more closely associated with the fermentation mode factor, whereas several esters were more strongly affected by process-related factors and interaction effects. These findings suggest that fermentation mode contributed to the differentiation of aroma expression, but its influence was compound-dependent rather than uniform across all volatile classes (Foss et al., 2023).

It should be noted, however, that the current study did not characterize microbial community structure in the spontaneous fermentation treatments, nor did it measure enzyme activities related to ester formation or terpene release. Therefore, although the observed differences are consistent with the view that microbial composition and fermentation behavior can influence aroma development, the present data do not allow direct attribution of specific volatile changes to particular microorganisms or metabolic pathways (Çelebi Uzkuç et al., 2020; Sugihara & Kamada, 2024).

4.2. Juice turbidity and its influence on volatile expression

Juice turbidity also played an important role in shaping the aroma profile of the resulting products (Y. Gao et al., 2026; L. Li et al., 2025; Ryona et al., 2012). Differences between clear and turbid juice treatments were already evident in the base wine composition and became more pronounced after distillation. In particular, the secondary distillate obtained from inoculated fermentation with clear juice (FC-Q) showed notably high OAVs for several esters, including ethyl acetate, ethyl butyrate, ethyl hexanoate, and hexyl acetate, whereas other treatments displayed distinct odor-active signatures (Stanzer et al., 2023).

These results indicate that clarification level affected the formation or retention of volatile compounds relevant to aroma quality. Previous studies have suggested that juice solids can alter the fermentation microenvironment by modifying nutrient availability, lipid composition, suspended particles, and oxygen transfer, thereby influencing yeast performance and metabolite production. The present results are generally compatible with such a view. However, dissolved oxygen, assimilable nitrogen, lipid fractions, and related physicochemical parameters were not quantified in this study. As a result, the current work can demonstrate an association between juice turbidity and volatile composition, but it cannot resolve the specific physicochemical drivers underlying these differences (Foss et al., 2023; F. Wang et al., 2024).

Another noteworthy result is that the influence of juice turbidity was not expressed in the same way across all treatments. For example, clear juice under inoculated fermentation favored the accumulation of several odor-active esters in the final distillate, whereas spontaneous fermentation under clear juice was more closely associated with compounds such as citronellol, acetic acid, and propionic acid. This suggests that juice turbidity should not be interpreted as an isolated factor; rather, its aroma-related effects depended on the fermentation context in which it operated (Balák et al., 2024; Sugihara & Kamada, 2024).

4.3. Distillation-associated changes in odor-active compounds

Marked differences in volatile composition were observed among the base wines, primary distillates, and secondary distillates, indicating that distillation substantially altered the distribution of aroma compounds. Several volatile classes, particularly acetate esters, ethyl esters, aldehydes and ketones, phenols, fatty acids, and higher alcohols, showed clear stage-related variation, and the secondary distillates exhibited the most distinct treatment separation in both PCA and OPLS-DA (Santos et al., 2023).

The OAV results further demonstrated that distillation changed not only compound concentrations but also their potential sensory relevance. Some compounds that contributed modestly at the base wine stage became more odor-active in the distilled fractions, while the relative contribution of others declined. In particular, the final distillate of FC-Q showed very high OAVs for several fruity esters, indicating a stronger ester-driven aroma profile than the other treatments. This observation was also in agreement with the sensory data, in which FC-Q received relatively high scores for fruity character, complexity, and overall pleasure (Z. Liang et al., 2021).

Nevertheless, these results should not be interpreted as evidence of real-time or mechanistically resolved aroma transformation during distillation. The present study compared discrete processing periods rather than continuously monitoring distillation fractions. Therefore, the data support the conclusion that distillation altered the composition and sensory relevance of volatile compounds, but they do not directly describe the kinetics of aroma migration throughout the distillation run.

4.4. Interaction effects of fermentation mode and juice turbidity

One of the main objectives of this study was to evaluate whether fermentation mode and juice turbidity acted independently or interactively in regulating the aroma profile of grape spirits. The two-way ANOVA results indicated that interaction effects contributed substantially to the variation of several odor-active compounds, especially esters such as ethyl acetate, ethyl butyrate, ethyl 3-methylbutyrate, ethyl caproate, and phenylethyl acetate. This means that the effect of one factor depended, at least for some compounds, on the level of the other factor.

The importance of these interaction effects was also reflected in the sample-specific odor-active signatures. FC-Q was characterized by relatively high levels of multiple ester-related compounds, FC-Z showed higher relative abundance of compounds such as phenylethanol and phenylethyl acetate, and ZR-Q was more closely associated with citronellol and volatile acids. Such patterns indicate that neither fermentation mode nor turbidity alone can fully explain the aroma differentiation observed in the final spirits.

Compared with previous studies that focused on a single pretreatment factor or only on fermented wine rather than distilled products, the present work provides evidence that aroma differences in grape spirits emerge from the combined action of upstream processing factors and subsequent distillation. In this sense, the novelty of this study lies not in demonstrating a single dominant factor, but in showing that treatment interactions can be retained and further expressed in the final distilled product.

4.5. Sensory relevance, practical implications, and study limitations

The sensory analysis provided practical support for the compositional results. Among the four treatments, FC-Q and ZR-Q showed relatively stronger overall sensory expression, whereas FC-Z generally received lower scores for several aroma-related attributes. In particular, the strong fruity and complex profile of FC-Q was consistent with its elevated OAVs for several esters in the final distillate. This agreement between chemical and sensory data suggests that treatment selection can influence not only analytical aroma composition but also the perceptual quality of grape spirits (Garbay et al., 2026; R. Li et al., 2026; Nicolli et al., 2018; Zhu et al., 2021; Zou et al., 2026).

From a practical perspective, the results indicate that the combination of fermentation mode and juice clarification strategy can be used as a process-level tool to modulate aroma style in Italian Riesling grape spirits. For example, inoculated fermentation with clear juice favored a more ester-driven and sensorially expressive final distillate under the present experimental conditions. However, this should not be generalized beyond the current material and process system without further validation, because cultivar, vintage, microbial background, and distillation regime may all affect the final outcome.

Several limitations should also be acknowledged. First, microbial community data were not available for the spontaneous fermentation treatments, limiting interpretation of microbiological contributions. Second, no enzyme activity measurements were conducted, which restricts mechanistic discussion of aroma formation pathways. Third, juice turbidity-related physicochemical variables such as dissolved oxygen, nitrogen availability, and lipid composition were not quantified. Finally, the distillation process was assessed using stage-based sampling rather than continuous fraction monitoring. Future studies combining microbial profiling, enzyme assays, targeted nutrient analysis, and refined distillation fraction tracking would help clarify the mechanisms underlying the compositional patterns observed here.

5. Conclusion

This study demonstrated that fermentation mode and juice turbidity, together with their interaction, were associated with significant differences in the physicochemical composition, volatile profile, odor-active compounds, and sensory properties of Italian Riesling grape spirits. Clear differences were observed among base wines, primary distillates, and secondary distillates, indicating that distillation substantially influenced the expression of aroma-active compounds and amplified treatment-dependent differences in the final products.

Among the treatments, inoculated fermentation with clear juice (FC-Q) produced a final distillate characterized by relatively high OAVs for several esters and by stronger fruity, complex, and pleasant sensory attributes. In contrast, the other treatments showed different odor-active signatures, indicating that fermentation mode and clarification strategy can modulate aroma style in distinct ways.

Importantly, the effects of fermentation mode and juice turbidity were not fully independent. Their interaction contributed substantially to the variation of several key odor-active compounds, especially esters, highlighting the need to evaluate these factors in combination rather than isolation when designing processing strategies for grape spirits.

However, the present study was based on compositional, multivariate, and sensory evidence only. Because microbial community structure, enzyme activities, and turbidity-related physicochemical variables were not measured, the mechanistic basis of the observed differences remains to be further clarified. Future studies integrating microbial ecology, biochemical measurements, and refined distillation monitoring will help establish a more comprehensive understanding of aroma regulation in grape spirit production.

CRediT authorship contribution statement

Huiyan Wang: Writing – original draft, Formal analysis, Data curation. Jingying Bi: Visualization, Investigation, Formal analysis. Guantong Liu: Methodology. Tian Gao: Investigation. Wenhong Bai: Investigation. Yanying Liang: Investigation. Yuyang Song: Writing – review & editing, Supervision, Formal analysis, Conceptualization. Yi Qin: Writing – review & editing, Supervision, Formal analysis, Conceptualization.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Data availability

Data will be made available on request.