A Vegetable Fermentation Facility Hosts Distinct Microbiomes Reflecting the Production Environment

Abstract

Fermented vegetables are highly popular internationally in part due to their enhanced nutritional properties, cultural history, and desirable sensorial properties. In some instances, fermented foods provide a rich source of the beneficial microbial communities that could promote gastrointestinal health. The indigenous microbiota that colonize fermentation facilities may impact food quality, food safety, and spoilage risks and maintain the nutritive value of the product. Here, microbiomes within sauerkraut production facilities were profiled to characterize variance across surfaces and to determine the sources of these bacteria. Accordingly, we used high-throughput sequencing of the 16S rRNA gene in combination with whole-genome shotgun analyses to explore biogeographical patterns of microbial diversity and assembly within the production facility. Our results indicate that raw cabbage and vegetable handling surfaces exhibit more similar microbiomes relative to the fermentation room, processing area, and dry storage surfaces. We identified biomarker bacterial phyla and families that are likely to originate from the raw cabbage and vegetable handling surfaces. Raw cabbage was identified as the main source of bacteria to seed the facility, with human handling contributing a minor source of inoculation. Leuconostoc and Lactobacillaceae dominated all surfaces where spontaneous fermentation occurs, as these taxa are associated with the process. Wall, floor, ceiling, and barrel surfaces host unique microbial signatures. This study demonstrates that diverse bacterial communities are widely distributed within the production facility and that these communities assemble nonrandomly, depending on the surface type.IMPORTANCE Fermented vegetables play a major role in global food systems and are widely consumed by various global cultures. In this study, we investigated an industrial facility that produces spontaneous fermented sauerkraut without the aid of starter cultures. This provides a unique system to explore and track the origins of an "in-house" microbiome in an industrial environment. Raw vegetables and the surfaces on which they are handled were identified as the likely source of bacterial communities rather than human contamination. As fermented vegetables increase in popularity on a global scale, understanding their production environment may help maintain quality and safety goals.

Keywords: fermentation; food microbiology; lactic acid bacteria; microbiology of the built environment; microbiome; phylogenetic diversity.

FIG 1

Facility map with sampling surface key. Facility map is color-coded to show different surface types, sample names, and sampling areas. Raw vegetables enter the facility through a garage door on the left side of the map and are transported on wooden pallets to the processing room and hand-processed for fermentation.

FIG 2

Taxon abundance heatmap. Heatmap depicts the relative abundances of bacterial phyla across all surfaces. The surface type of the sample is indicated by the colored block on the top of the heatmap.

FIG 3

Box chart representation for phylum-level comparison. Relative abundances of bacterial phyla Proteobacteria (a), Firmicutes (b), and Actinobacteria (c) were compared across raw vegetables and environmental surfaces in the facility. Surfaces are organized by similar locations within the facility. Raw vegetable group includes the raw cabbage and vegetable handling surfaces, whereas the environmental group includes the processing room, fermentation room, and dry storage surfaces. Statistical analysis was computed using an unpaired t test with Welch's correction (*, P < 0.05; **, P < 0.01; ***, P < 0.001).

FIG 4

Relative abundances of bacterial families across surface types. Each colored bar represents a bacterial family percentage that comprises the microbiome sample. Surfaces are organized by similar location. HS, hand sink; CB, cutting board; S, shredder; MB, mixing bin; MS, mop sink (see Fig. 1 for surface descriptions).

FIG 5

Mean relative abundances of the bacterial families according to surface swab classification. One-way analysis of variance (ANOVA) with post hoc Tukey test was performed, with significance at a P value of <0.05. a to m, significant differences in mean relative abundance for bacterial family between two surface swabs. For example, “a, b” on the top of the bar represents that the family Moraxellaceae is significantly different between vegetable handling and dry storage surfaces, as well as between vegetable handling and processing room surfaces. a, vegetable handling versus dry storage; b, vegetable handling versus processing room; c, raw vegetable versus processing room; d, vegetable handling versus fermentation room; e, fermentation barrels versus fermentation room; f, fermentation barrels versus dry storage; g, processing room versus fermentation barrels; h, raw vegetable versus fermentation barrels; i, vegetable handling versus fermentation barrels; j, processing room versus dry storage; k, dry storage versus fermentation room; l, raw vegetable versus fermentation room; m, processing room versus fermentation room. “Other” includes collective abundance of all other bacterial families.

FIG 6

Spatial distribution heatmap of bacteria in the fermentation facility environment. (a) Plots indicate relative abundances of genera as a percentage of the entire community. The scale bar on top of each map normalizes the relative abundances of the defined taxa. In the Leuconostoc map, for example, the darkest green color indicates that the walk-in refrigerator community was 19% Leuconostoc. In Staphylococcus, the darkest green color represents 4% of the total community structure. (b and c) Predicted microbial contamination sources within the vegetable fermentation facility. The maps illustrate percentages of predicted sources for members of the community, as estimated by SourceTracker. The scales above the maps represent the percentage of the total community composition for which the source accounts; data are shown for raw vegetables (blue) (b) and unknown origins (orange) (c).

FIG 7

(a) Principal-coordinate analysis of environmental swab surfaces. Weighed UniFrac distances were used to assess beta diversity. Raw vegetable swabs (blue circles) are genetically distinct from environmental swabs (red circles). Approximately 67% of the variability can be explained by the first two principal components (PC1 and PC2). (b) Principal-coordinate analysis among pre- and postfermentation surfaces. A high degree of genetic dissimilarity between the prefermentation (blue circle) and postfermentation (red circles) environmental surfaces was observed. Approximately 50% of the variability can be explained by PC1 and PC2.

FIG 8

Family-level comparisons between prefermentation and postfermentation room environmental swab surfaces. Each stacked bar represents mean abundance of 3 surface swabs from same area. Vertical red line segregates the surfaces.

FIG 9

Mean relative abundances of the bacterial families in pre- (pre) and postfermentation (post) room surface swabs. One-way ANOVA with a post hoc Tukey test was performed, with significance at a P value of <0.05. a to p, significant difference of mean relative abundances for bacterial families between two surface swabs. For example, a, b on the top of bars represents that family Comamonadaceae is significantly different between wall postfermentation and prefermentation surfaces as well as ceiling postfermentation and prefermentation surfaces. a, wall post versus wall pre; b, ceiling post versus ceiling pre; c, barrel pre versus floor between barrels post; d, floor between barrels post versus floor post; e, floor post versus floor pre; f, barrel post versus barrel pre; g, floor between barrels post versus floor post; h, floor post versus floor pre; i, barrel pre versus floor between barrels post; j, barrel post versus barrel pre; k, ceiling post versus ceiling pre; l, barrel post versus barrel pre; m, barrel pre versus floor between barrels post; n, barrel post versus floor between barrels post; o, barrel pre versus floor between barrels post; p, barrel post versus barrel pre.