Bacterial Communities Related to Aroma Formation during Spontaneous Fermentation of 'Cabernet Sauvignon' Wine in Ningxia, China

Abstract

Bacteria are an important part of wine 'microbial terroir' and contribute to the formation of wine flavor. Based on high-throughput sequencing and non-targeted metabonomic technology, this study first explored the bacterial composition and its effect on the aroma formation of spontaneously fermented 'Cabernet Sauvignon' (CS) wine in the Eastern Foot of Helan Mountain (EFHM), Ningxia. The results showed that there were significant differences in bacterial communities during fermentation of CS grapes harvested from different sub-regions of EFHM, with the earlier-established vineyard obtaining more species. The level of bacterial diversity initially decreased and then increased as the fermentation proceeded. Malolactic fermentation (MLF) was spontaneously initiated during alcohol fermentation (AF). Pantoea, Lactobacillus, Rhodococcus, Fructobacillus, and Komagataeibacter were the core bacterial genera in the fermentation mixture. Lactobacillus contributed to the synthesis of methyl and isobutyl esters and the formation of red and black fruity fragrances of wine. Fructobacillus was closely related to the synthesis of aromatic alcohols and the generation of floral flavors.

Keywords: Cabernet Sauvignon; Ningxia; bacteria; flavor; spontaneous fermentation; wine.

Figure 1

The fermentation process and sequencing evaluation. (a) The release of CO₂ of three spontaneous fermentations. Samples were collected at six points marked with A (0 d), B (1 d), C (2 d), D (4 d), E (8 d), and F (12 d); (b) the rarefaction curve: X-axis, sequenced reads per sub-region; Y-axis, the percentage of discovered bacteria to predicted total bacteria per sub-region; (c) the observed numbers of bacteria per sub-region; (d) the shared numbers of bacteria among three sub-regions. OTUs, operational taxonomic units. Sequences with more than 97% similarity were classified as one OTU.

Figure 2

NMDS plots of samples from different sub-regions. Discrimination by (a) taxonomic distance, and (b) phylogenetic distance. The NMDS stress value below 0.1 and simultaneously p-value below 0.05 indicate significant discrimination among different sub-regions. The color of each square/dot/triangle corresponds to the color of each sub-region in the legend.

Figure 3

Bacterial diversity of different sub-regions. Alpha diversity at genus level: (a) Chao 1 diversity, (b) Shannon diversity, (c) Shannoneven diversity, and (d) Faith’s phylogenetic diversity; Alpha diversity values were calculated based on rarefied data established using 16S sequencing reads from 18 samples per sub-region; the statistical significance analysis was based on Wilcoxon rank-sum test, ** p < 0.01, *** p < 0.001; (e) the Venn plot of bacteria at the genus level; (f) the shared top 10 bacteria with a relative abundance of more than 1% at the genus level among three sub-regions.

Figure 4

Core bacteria at the genus level of the Eastern Foot of Helan Mountain. (a) A RF model based on a 10-fold cross-validation test was constructed with 500 trees, using taxonomic assignments of genera as predictors and regional origin as class labels. When the bacterial number at the genus level was 7, the RF error reached the lowest value (0.001). The bar plot of variable importance at genus level: X-axis, bacterial importance measurement/standard deviation; Y-axis, bacterial name; (b) the percentages of the five core bacteria in three sub-regions. The color of each dot corresponds to the color of each bar in Figure 4a. The size of the circle represents the proportion of each bacterial genus in different sub-regions.

Figure 5

Bacterial communities and their functional annotation in different fermentation stages. The successions of bacteria at genus level during fermentation: (a) top 11 genera with relative abundance greater than 0.1%, and (b) 5 core genera (darker red squares represent higher relative abundance); (c) top 10 functions annotated by the FAPROTAX database.

Figure 6

Relative concentrations of VOCs in different fermentation stages of different sub-regions.

Figure 7

Correlation analysis between core bacteria and VOCs. RDA was used to summarize linear relationships between core bacteria and VOCs in (a) YC sub-region, (c) YQY sub-region, and (e) QTX sub-region. Arrow angle indicates the strength of the association between variables: Acute angle indicates strong concordance; obtuse angle, weak concordance. The different colors of arrows represent VOCs and genera. The color of each point shape corresponds to the color of each fermentation stage in the legend. A correlation heatmap was used to reveal significant linear correlations between each VOC and each genus in (b) YC sub-region, (d) YQY sub-region, and (f) QTX sub-region. A red ellipse with an asterisk represents a significantly positive correlation (Spearman’s rank test, p < 0.05), while a blue ellipse with an asterisk represents a significantly negative correlation (Spearman’s rank test, p < 0.05). AE: acetic esters; ME: methyl esters; IBE: isobutyl esters; IAE: isoamyl esters; SCFA: straight-chain fatty alcohols; BCFA: branched-chain fatty alcohols; AA: aromatic alcohols; FAL: fatty aldehydes; AAL: aromatic aldehydes; FA: fatty acids; T: terpenes; AALK: aromatic alkenes; S: sulfides. Pan., Pantoea; Lac., Lactobacillus; Rho., Rhodococcus; Fru., Fructobacillus; Kom., Komagataeibacter.

Figure 8

Correlation analysis between core bacteria and aroma profiles. (a) Aroma profile of spontaneously fermented grape wines from three sub-regions analyzed by QDA. (b) A correlation heatmap was used to identify significant linear correlations between each aroma descriptor and each core bacterial genus. A red ellipse with an asterisk represents a significantly positive correlation (Spearman’s rank test, p < 0.05), while a blue ellipse with an asterisk represents a significantly negative correlation (Spearman’s rank test, p < 0.05). Pan., Pantoea; Lac., Lactobacillus; Rho., Rhodococcus; Fru., Fructobacillus; Kom., Komagataeibacter.

Figure 9

Correlation analysis between core bacteria and ethanol in (a) YC sub-region, (b) YQY sub-region, and (c) QTX sub-region. A red ellipse with an asterisk represents a significantly positive correlation (Spearman’s rank test, p < 0.05), while a blue ellipse with an asterisk represents a significantly negative correlation (Spearman’s rank test, p < 0.05). Et., ethanol; Pan., Pantoea; Lac., Lactobacillus; Rho., Rhodococcus; Fru., Fructobacillus; Kom., Komagataeibacter.