Article Figures & Data
- FIG 1
Map of sampling sites across Napa and Sonoma Counties. Each point represents an individual vineyard from which grapes were harvested for the fermentations monitored in this study. Points are colored by AVA designation, as indicated in the key. The inset illustrates the position of this sampling area within California.
- FIG 2
Microbiota exhibits regional variation in musts and wines. (A) Chardonnay. (B) Cabernet Sauvignon. Shown are PCoA comparisons of bacterial weighted UniFrac distance (left two columns) and fungal Bray-Curtis dissimilarity (right two columns) in musts and wines (see column labels), categorized by vineyard (color) and AVA source (shape). Each point represents an individual sample, and sample proximity on the plot is a function of similarity in bacterial and fungal community composition.
- FIG 3
Stage of fermentation influences microbial richness. Shown are the mean ± standard deviation (SD) bacterial (A) and fungal (B) richness (observed OTU) in Cabernet Sauvignon and Chardonnay by stage of fermentation. Different lowercase letters indicate significantly different means.
- FIG 4
Stage of fermentation influences microbial composition of Cabernet Sauvignon. (A and B) Bacterial weighted UniFrac (A) and fungal Bray-Curtis (B) PCoA comparing similarity among all Cabernet samples, colored by stage of fermentation. (C and D) Relative abundance of bacteria (C) and fungi (D) that differed significantly by stage of fermentation. Only taxa detected at >1% relative abundance are shown in panels C and D. False discovery rate (FDR)-corrected P values are listed for each taxon. The taxon “Leuconostocaceae” represents O. oeni, which belongs to this bacterial family, and all OTU assigned to this taxonomy matched O. oeni by BLASTn.
- FIG 5
Must microbiome is correlated with the wine metabolome. Shown are the results from multifactorial analysis (MFA) of must microbiome and wine metabolome profiles of Chardonnay (A to D) and Cabernet Sauvignon (E to G). (A and E) An MFA sample ordination plot demonstrates regional and vineyard segregation of Chardonnay (A) and Cabernet Sauvignon (E). (B and F) A partial-axis MFA plot illustrates category correspondence between metabolite (orange), bacterial (blue), and fungal (red) subcategory coordinates. Partial mean individuals (means of sample ordination for bacterial, fungal, or metabolite profiles) are linked to the subcategory common mean (centroid of all samples for a given region of origin). (C and G) MFA group representations illustrate the relationship between bacterial, fungal, and metabolite profiles with wine origin (region and vineyard). (D) An MFA correlation circle depicts correlations in normalized abundance between all Chardonnay must bacterial taxa (blue), fungal taxa (red), and wine metabolites (orange) along the MFA axes. Metabolite nominal masses are used for clarity; accurate masses are provided in Table S5 in the supplemental material. To improve readability of the plots, only the top correlations in each dimension are shown.
- FIG 6
Microbial composition accurately predicts metabolite abundance of select metabolites. Shown are the results from random forest regression of predicted versus observed metabolite intensity in validation samples using a sparse set of predictors (inset, model using all taxa as predictors) to predict abundance of a C6 keto acid in Cabernet Sauvignon wines (A) and a C6 acid, ester, or lactone in Chardonnay wines (B). Trend lines indicate a true 1:1 ratio. Bar plots were used to display the ranked feature importance of bacterial (blue bars) and fungal (red bars) taxa used to train the optimized models. MW, molecular weight; MDA, mean decrease in model accuracy if that feature is removed from the model.
- FIG 7
Microbiota correlations with must/juice and fermentation characteristics. Shown is the Spearman correlation between must/juice chemistry, fermentation characteristics, and microbiota in musts (left columns) and end of fermentations (EOF [right columns]). Only significant correlations (FDR-corrected P value of <0.05) are shown. As chemical composition was only measured in musts and juices, no data appear for must/juice correlations at end of fermentation (gray boxes at top right corner). NH3, ammonia concentration; NOPA, total nitrogen by o-phthaldialdehyde assay; YAN, yeast assimilable nitrogen.
- TABLE 1
Fermentation stages and sample collection schematic
Stagea Process Chardonnay Cabernet Sauvignon Statusb Sitec Status Site Crush Crush Juice PP Must Prior to inoculation Juice T Must T Early Early fermentation ~22 °Brix T Mid Mid-fermentation ~20 °Brix T Late Near end of fermentation ~1 °Brix B Pressed T End End of fermentation pre-SO2 Dry (wine) B MLF End of MLF ML—dry B Wine Aged Prior to bottling B Prior to racking B ↵a Stage name indicates the name used in subsequent figures.
↵b Status indicates the material type and sugar concentration (°Brix) at that stage. Musts are unfermented crushed grapes containing both pomace and juice. Juice is unfermented, pressed grape must. The product is considered wine after the end of fermentation (“end” stage). Empty entries indicate no sample collected at that stage, as red and white wines are processed differently.
↵c PP, press pan; T, fermentation tank; B, barrel; MLF, malolactic fermentation.
Supplemental Material
TABLE S1
Number of must and juice samples collected from each vineyard. Table S1, DOCX file, 0.01 MB.
Copyright © 2016 Bokulich et al.
This is an open-access article distributed under the terms of the Creative Commons Attribution 4.0 International license.
TABLE S2
Permutational MANOVA comparisons of microbial diversity in musts and wines. Table S2, DOCX file, 0.01 MB.
Copyright © 2016 Bokulich et al.
This is an open-access article distributed under the terms of the Creative Commons Attribution 4.0 International license.
TABLE S3
Random forest models predict vineyard origin of grape musts. Table S3, DOCX file, 0.05 MB.
Copyright © 2016 Bokulich et al.
This is an open-access article distributed under the terms of the Creative Commons Attribution 4.0 International license.
FIGURE S1
Fermentation stage exhibits winery-specific influences on Chardonnay microbial profiles. (A and B) Bacterial weighted UniFrac PCoA of Far Niente Chardonnay fermentations, color coded by stage (A), and Nickel & Nickel Chardonnay color coded and labeled by vineyard of origin (B). (C) Fungal Bray-Curtis dissimilarity of all Chardonnay samples colored by stage indicates that fungal profiles change by stage of fermentation and follow the same pattern in both wineries. (D and E) Relative abundance of bacteria (D) and fungi (E) that differed significantly by stage of fermentation. Only taxa detected at >1% relative abundance are shown. False discovery rate (FDR)-corrected P values are listed for each taxon. C, crush stage; M, must; E, early fermentation. Download Figure S1, JPG file, 0.5 MB.
Copyright © 2016 Bokulich et al.
This is an open-access article distributed under the terms of the Creative Commons Attribution 4.0 International license.
TABLE S4
Regionally differential Cabernet Sauvignon wine mass features and putative metabolites. Table S4, DOCX file, 0.01 MB.
Copyright © 2016 Bokulich et al.
This is an open-access article distributed under the terms of the Creative Commons Attribution 4.0 International license.
FIGURE S2
Metabolite profiles of Chardonnay and Cabernet Sauvignon wines cluster by vineyard and AVA. PCA of metabolite profiles of Chardonnay (A) and Cabernet Sauvignon (B) wines categorized by vineyard (color) and AVA source (shape). Download Figure S2, JPG file, 0.3 MB.
Copyright © 2016 Bokulich et al.
This is an open-access article distributed under the terms of the Creative Commons Attribution 4.0 International license.
TABLE S5
Regionally differential chardonnay wine mass features and putative metabolites. Table S5, DOCX file, 0.01 MB.
Copyright © 2016 Bokulich et al.
This is an open-access article distributed under the terms of the Creative Commons Attribution 4.0 International license.
TABLE S6
Chardonnay metabolite random forest model summaries. Table S6, DOCX file, 0.01 MB.
Copyright © 2016 Bokulich et al.
This is an open-access article distributed under the terms of the Creative Commons Attribution 4.0 International license.
TABLE S7
Cabernet Sauvignon metabolite random forest model summaries. Table S7, DOCX file, 0.01 MB.
Copyright © 2016 Bokulich et al.
This is an open-access article distributed under the terms of the Creative Commons Attribution 4.0 International license.
Additional Files
Supplementary Data
Supplementary Data
- Figure sf1, JPG - Figure sf1, JPG
- Figure sf2, JPG - Figure sf2, JPG
- Table st1, DOCX - Table st1, DOCX
- Table st2, DOCX - Table st2, DOCX
- Table st3, DOCX - Table st3, DOCX
- Table st4, DOCX - Table st4, DOCX
- Table st5, DOCX - Table st5, DOCX
- Table st6, DOCX - Table st6, DOCX
- Table st7, DOCX - Table st7, DOCX
