Marselan itself is a relatively young variety, a cross between Cabernet Sauvignon and Grenache that produces deeply colored wines with soft tannins and a balanced floral-fruity aromatic signature. In Ningxia, a temperate continental arid zone defined by abundant sunshine, wide diurnal temperature swings, scant rainfall, and well-drained soils, the variety has become a regional flagship. Grapes ripen fully in this environment, accumulating the aroma precursors that fermentation later converts into the volatile compounds perceived as banana, pineapple, rose, and citrus. Yet while terroir is traditionally framed as climate, soil, and topography, the research team, led by Zicheng Yi, Tengwen Chang, and colleagues at the Ningxia Institute for Agricultural Product Quality Standards and Testing Technology, argued that the fermentation ecosystem itself deserves equal billing. Yeast communities associated with grape surfaces and cellar environments metabolize grape-derived substrates into aroma-active molecules, meaning that the microbial residents of a vineyard may be as distinctive as its geology.

The problem, the authors noted, is that most earlier studies of microbial terroir stopped at cataloguing diversity or correlating community composition with broad wine characteristics. Few had traced a complete evidence chain from a specific indigenous yeast, through identified aroma-active compounds, to sensory outcomes validated by trained human panels. Meanwhile, the global wine industry’s reliance on a narrow pool of commercial Saccharomyces cerevisiae starters, prized for reliability and consistency, has arguably narrowed the aromatic range of wines worldwide. To close this gap, the team isolated eleven indigenous non-Saccharomyces yeasts from grape surfaces and spontaneous ferments in the Helan Mountain foothills in 2023, cryopreserved them, and reactivated them for controlled experiments with Marselan grapes harvested in September 2024 from a single vineyard, ensuring a consistent single-vintage system. Every strain was re-identified by sequencing the 26S rRNA D1/D2 domain before use.

The experimental design was elegant in its simplicity. Each of the eleven indigenous strains was co-inoculated with RC212, a widely used commercial S. cerevisiae reference in red winemaking, at a 1:1 volume ratio, with the non-Saccharomyces yeast introduced at a higher initial cell density. A control ferment received RC212 alone. Twelve treatments, each in triplicate, proceeded at 22 degrees Celsius in semi-sealed stainless steel fermenters, with daily residual sugar monitoring until dryness below 4 grams per liter was reached. All wines completed fermentation successfully, with alcohol between 9.89 and 13.47 percent by volume, and standard physicochemical parameters varied only modestly. The real story emerged when the volatile chemistry was examined. Headspace solid-phase microextraction coupled to gas chromatography–mass spectrometry identified 148 volatile compounds, dominated by 50 esters and 41 alcohols, the two families responsible for most fruity and floral notes. Total volatile concentrations ranged from 158.45 to 274.43 milligrams per liter, with most mixed fermentations exceeding the control’s 189.80 milligrams per liter.

Three co-fermentations stood out sharply. Wines made with Metschnikowia pulcherrima (MP-RC), Clavispora lusitaniae (CL-RC), and Pichia kluyveri (PK-RC) posted the highest total odor activity values, at 8026.85, 8332.64, and 6605.82 respectively, compared with 5592.91 for the control, driven largely by medium-chain esters such as ethyl octanoate, ethyl hexanoate, and ethyl decanoate, alongside the potent terpene β-damascenone. MP-RC achieved the highest ester concentration of any treatment, 829.73 micrograms per liter against the control’s 540.32, and both MP-RC and PK-RC scored above 8 out of 10 from a trained sensory panel for fruity and floral intensity. Principal component analysis separated these three wines from the pack, positioning them near loading vectors for ethyl octanoate, ethyl isovalerate, and β-damascenone, while the control and several weaker treatments clustered near the centroid with simpler, less distinctive profiles.

To move beyond correlation, the researchers deployed the full sensomics toolkit. Gas chromatography–olfactometry combined with aroma extract dilution analysis allowed trained assessors to sniff effluent from the chromatograph at successive dilutions, assigning flavor dilution factors that flag compounds perceptible even at extreme dilution. Fifty-two candidate aroma compounds with odor activity values above 0.1 were screened, and five emerged as the shared aromatic backbone of all three standout wines: isoamyl acetate, contributing fresh banana top notes; ethyl octanoate, delivering rounded pineapple-like tropical fruit; ethyl hexanoate, the most intense fruity ester, evoking apple and pear; β-damascenone, an extraordinarily potent norisoprenoid with an odor threshold of just 0.05 micrograms per liter, supplying a persistent citrus-floral background; and isoamyl alcohol, a higher alcohol adding mouthfeel depth. Their odor activity values in the finished wines were striking, with ethyl octanoate exceeding 3600 and β-damascenone exceeding 2100 in every one of the three treatments.

The decisive test came from aroma recombination and omission experiments. The team reconstructed each wine’s aroma in a standardized model matrix of tartaric acid, ethanol, and water adjusted to wine-like pH, adding the selected compounds at concentrations measured in the original wines. Sixteen ISO-qualified assessors then performed triangle tests in which a single compound was systematically removed from the recombination. Omitting isoamyl acetate, ethyl octanoate, ethyl hexanoate, β-damascenone, or isoamyl alcohol produced highly significant perceptual differences in all three wines, confirming these five odorants as the functional pillars of the fruit-floral structure. The complete recombination models best reproduced the original wines’ profiles, validating the analytical pipeline end to end. The work demonstrates, in effect, that wine aroma can be decomposed into a short list of demonstrable molecular drivers, and that different yeasts tune the relative weightings of that list.

Those differing weightings translated into genuinely distinct wine styles. MP-RC, the M. pulcherrima co-ferment, earned the highest floral score of 9, supported by partial least squares regression coefficients above 0.9 for terpenes and alcohols, and was enriched in isoamyl acetate, ethyl octanoate, and β-damascenone, yielding a floral-dominant, multidimensional profile with mid-palate fullness and a clean finish. The authors attribute this partly to M. pulcherrima’s reported β-glucosidase activity, which can hydrolyze glycosidically bound terpene and norisoprenoid precursors on the grape, releasing bound linalool and β-damascenone, and partly to complementary amino acid metabolism feeding the Ehrlich pathway and subsequent acetate ester formation. CL-RC, by contrast, accumulated the highest total volatile concentration and overall odor activity yet scored lower on florals, a cautionary result showing that abundance is not perception. Elevated isoamyl alcohol apparently masked floral compounds, while mid-chain esters like ethyl isovalerate and ethyl butyrate produced a fruit-and-acid tension style. PK-RC occupied a middle ground, fresh and balanced but less layered, with somewhat elevated isovaleric acid that may have curtailed its complexity.

The implications reach well beyond Ningxia. If specific indigenous yeasts reproducibly generate specific aroma outcomes, wineries could cultivate regional starter cultures that express local identity deliberately rather than leaving it to chance, offering an alternative to the aromatic homogenization that single commercial strains can impose. The study also refines how the industry should screen non-Saccharomyces candidates: sensory relevance, compound balance, and odor thresholds matter more than raw volatile yield, as CL-RC’s high production but flat sensory performance made clear. The authors are appropriately measured about limitations, noting that conclusions rest on one vineyard, one vintage, and laboratory-scale fermentation, and that dynamic yeast population monitoring, enzyme assays, transcriptomics, and metabolic flux analysis will be needed to confirm the underlying mechanisms. Still, the core message is compelling and commercially resonant: the wild yeasts clinging to a grape’s skin are not contamination to be suppressed but a fermentable archive of place, one that, when paired thoughtfully with S. cerevisiae, can write a region’s signature directly into the glass.

Subject of Research: Mixed fermentation with indigenous non-Saccharomyces yeasts and its effects on the aroma profile of Marselan wine, characterized through integrated sensomics analysis.

Article Title: Aroma differentiation of Marselan wine through mixed fermentation with indigenous non- Saccharomyces yeasts revealed by Sensomics analysis

Article References: Yi, Z., Chang, T., Huang, L., Hou, X., Cheng, H., Zhang, J., Peng, B., & Ge, Q. (2026). Aroma differentiation of Marselan wine through mixed fermentation with indigenous non-Saccharomyces yeasts revealed by Sensomics analysis. Food Chemistry: X, 39, Article 104457. https://doi.org/10.1016/j.fochx.2026.104457

Image Credits: AI Generated

DOI: 10.1016/j.fochx.2026.104457

Keywords: Marselan wine, non-Saccharomyces yeasts, mixed fermentation, aroma compounds, sensomics, Metschnikowia pulcherrima, Pichia kluyveri, Clavispora lusitaniae, microbial terroir, gas chromatography-olfactometry, aroma recombination, Ningxia wine region