Fermented-food metagenomics reveals substrate-associated differences
TL;DR
Shotgun metagenomics of 58 artisanal fermented foods found that substrate was the strongest measured driver of microbial composition. Dairy foods were less diverse than brined vegetables or sugar-based ferments, and spontaneous fermentations were more diverse than starter-culture fermentations. This establishes that “fermented food” is not one microbial exposure; it does not test ingestion or health outcomes.
Design
- 58 consumable food samples from artisanal producers: 11 dairy foods, 26 brine foods (including seven sauerkrauts and five kimchis), 18 sugar-based foods (including kombucha and water kefir), and three other foods.
- Shotgun metagenomic sequencing characterized DNA-based taxonomy and predicted functional genes. Analyses compared substrate, starter use, physical state, producer, and country.
- This was a cross-sectional food-microbiome survey, not a human feeding study.
- There were no biological replicates of the individual foods, so food-specific conclusions cannot be separated from producer and batch effects.
Key findings
- The 58 samples contained 476 species above 0.1% relative abundance among taxonomically assigned reads. Brine, sugar, and dairy groups contained 301, 242, and 70 species in aggregate, averaging 11.5, 13.5, and 6.4 species per sample, respectively.
- Dairy foods had significantly lower alpha diversity than brine or sugar foods; spontaneous ferments were more diverse than starter-culture ferments.
- Brine foods were dominated by lactic-acid bacteria. Mean relative abundances included Lactiplantibacillus plantarum 9.6%, Levilactobacillus brevis 7.9%, Leuconostoc mesenteroides 4.7%, and Pediococcus parvulus 4.3%. Kimchi and sauerkraut also contained yeasts including Pichia and Kazachstania.
- Dairy foods were comparatively homogeneous and dominated by Lactococcus lactis (44.8% mean relative abundance) and Streptococcus thermophilus (16%); Kluyveromyces marxianus represented 3.7%.
- Sugar-based fermentations contained more fungal reads and included yeasts plus acetic-acid bacteria such as Gluconobacter, Acetobacter, and Komagataeibacter.
Interpretation and limitations
The study directly supports microbiological differences among the sampled kefir- and cheese-like dairy foods, kimchi/sauerkraut-like brine foods, and kombucha-like sugar ferments. It included no yogurt samples and therefore does not directly characterize commercial Greek yogurt. It also does not show that a more diverse food microbiome is healthier, that all commercial products resemble these artisanal samples, or that detected organisms remain viable after storage and ingestion. Percentages describe assigned DNA reads rather than absolute viable-cell counts, and predicted “health-associated” genes are not clinical outcomes. gap/needs-human-replication