In this study, we extracted DNA from insect-pollinated floral samples (Lilium japonicum, Gypsophila elegans, Camellia japonica) and grape berry surfaces (both fresh and dried) to detect S. cerevisiae using specific primers. The S. cerevisiae strain IFO10217 and grape berry surfaces, known to harbor Saccharomyces, served as positive controls to validate this methods. Lilium and Gypsophila were collected from the KOSEN, Kure campus (E132˚36'21" N34˚13'44"). Camellia samples were collected from the Irifuneyama Memorial Museum, Kure city (E132˚33'50" N34˚14'25").

Enrichment was performed using both batch (conditions 1 to 3) and chemostat (conditions 4 to 6) cultivation. Condition-1: involved a medium primarily composed of koji saccharification broth. Rice koji (Masuyamiso Co., Ltd., Hiroshima, Japan) was saccharified overnight at 56˚C with four times its volume in distilled water. The solution was then decanted to obtain the broth, which contains Yeastcidin produced by Aspergilus oryzae, known for its antifungal and antiyeastal action against non-S. cerevisiae species [14] . The saccharification liquid was supplemented with 2 g/L sodium propionate and 100 mg/L chloramphenicol, and the pH was adjusted to 4.0 with lactic acid. Condition-2: utilized S. cerevisiae’s high tolerance for sugar and osmotic pressure [15] , comprising a medium of 500 g/L glucose, 10 g/L yeast extract, 20 g/L hypolypepton (Shiotani MS, Japan), 2 g/L sodium propionate, and 100 mg/L chloramphenicol. Condition-3: leveraged S. cerevisiae’s unique ability to assimilate raffinose [16] , consisting of a medium with 10 g/L raffinose, 6.7 g/L yeast nitrogen base (BD, USA), 80 mL/L ethanol, 2 g/L sodium propionate, and 100 mg/L chloramphenicol. For batch cultivation, the culture media under these conditions were left undisturbed for two weeks before being used for isolation cultivation.

Condition-4 had a similar medium composition to Condition-2 but was used in chemostat cultivation to minimize changes in medium components. Condition-5 was based on the knowledge that lactic acid produced by Lactobacillus during sugar decomposition is a factor in S. cerevisiae’s dominance [17] . The medium was adjusted with lactic acid to a pH of 5.0, containing 400 g/L glucose, 10 g/L yeast extract, 20 g/L hypolypepton (Shiotani MS, Japan), 2 g/L sodium propionate, and 100 mg/L chloramphenicol. Condition-6 was designed based on S. cerevisiae’s high ethanol tolerance as a dominance factor [18] , creating a medium with an elevated ethanol concentration: 400 g/L glucose, 100 mL/L ethanol, 10 g/L yeast extract (Difco, USA), 20 g/L hy-polypepton (Shiotani MS, Japan), 2 g/L sodium propionate, and 100 mg/L chloramphenicol. The chemostat cultivation vessels were modified medium bottles with a working volume of 200 mL. The culture medium was stirred at a low speed to maintain homogeneity, and the system was continuously supplied with fresh medium to maintain a dilution rate (D) of 2.0 day⁻¹, corresponding to a hydraulic retention time of 0.5 days. This rate was intentionally set lower than the specific growth rate of S. cerevisiae to impose steady selective pressure. The enriched material, similar to that in the batch system, was harvested after two weeks of cultivation for isolation.

Given the expectation that the enriched material contains multiple yeast species, employing selective pressures beyond substrate assimilation capabilities could potentially enable the isolation of S. cerevisiae strains with a broader range of characteristics. In Step-3.5, we adopted a microbial isolation technique based on this patent [19] , using the species-specific plasmids pAUR101 and pAUR112, both purchased from Takara Bio Inc. (Shiga, Japan) [20] . These plasmids confer resistance to Aureobasidin A specifically in S. cerevisiae and have been previously validated in laboratory strains. No further modifications or re-cloning were performed prior to use. The enriched culture material was washed with sorbitol buffer, then diluted to an O.D. 600 of 0.1 for electroporation using the Gene Pulser Xcell^TM Electroporation System (Bio-Rad Laboratories) and 0.2 cm gap cuvettes, under conditions of 1.5 kV, 500 Ω, and 50 μF. Two micrograms of each plasmid were added per transformation. Following electroporation, the cells were recovered in YPD medium for 2 hours and then plated on PDA medium containing 0.1 mg/L Aureobasidin A for selection.

For general yeast culture, Potato Dextrose Agar (Fuji film wako pure chemical Co., Osaka, Japan) medium was prepared. The enriched culture material was colonized on this medium. In particular, for the samples that underwent Step 3.5, PDA medium supplemented with 0.5 mg/L Aureobasidin A (PDA + AUR) was used. Two methods were employed to achieve colonization: direct streaking of the enriched samples onto the medium and dilution plating [21] . All colonies obtained, irrespective of the method used, were preserved as individual strains.

In this study, DNA was extracted from all petal samples (Maple, Lilium, Gypsophila), as well as from dried and fresh grape samples, and the IFO10217 strain culture, using the CTAB method [22] . Additionally, DNA from clones isolated from the enriched material was crudely extracted using the cellysate method [23] . To verify the presence of Saccharomyces cerevisiae in the environmental samples, PCR was performed with specific primers SC1 (5'-AACGGTGAGAGATTTCTGTGC-3') and SC2 (5'-AGCTGGCAGTATTCCCACAG-3') [24] . The PCR reactions were conducted using KOD one (TOYOBO, Japan), and the resulting amplicons were purified using AMPure (Beckman Coulter, Germany). For the identification of the isolated strains at the genus and species levels, a similar PCR process was employed using different primers for the ITS region of 16S-23SrRNA. The primers ITS1F (GTAACAAGGTYTCCGT) and ITS1R (CGTTCTTCATCGATG) were used [25] . The sequencing was performed on the ABI 3100avant Sanger sequencer to determine the microbial communities of the colonies originating from the enrichment systems. The obtained sequences were then analyzed for nucleotide sequence homology using BLASTn [26] .

In this study, Saccharomyces cerevisiae was not detected on grape surfaces, which

contradicts our initial expectations of grapes as a reliable positive control ( Figure 2 ). This finding diverges from the results reported in [27] , where S. cerevisiae was successfully identified in similar samples. Employing a potent KOD one polymerase and executing 35 PCR cycles, this study suggests that the presence of S. cerevisiae in natural environments might be more sporadic and less abundant than previously assumed. Successful detection was achieved only with the control strain IFO10217, showing the necessity of enriching sufficient yeast densities for detection and isolation. Given these challenges in detecting S. cerevisiae even on grape surfaces, this study indicates that flower petals may not be the ideal source for isolating wild S. cerevisiae. Investigating more specific microhabitats within floral environments, such as nectar, could be crucial, as these may harbor higher concentrations of the yeast. However, the absence of S. cerevisiae on grape surfaces also implies that even these localized microbial communities might not demonstrate a predominance of S. cerevisiae. This study, therefore, underscores the importance of targeting specific microhabitats and the need for more nuanced approaches to effectively isolate S. cerevisiae from natural environments.

Considering these observations, while it’s unlikely that a significant amount of Saccharomyces cerevisiae adheres to the bulk of Lilium, Gypsophila, and Camellia petals, past studies suggest that isolation becomes possible through enrichment cultivation. Therefore, in light of these insights, Camellia petals were selected as the enrichment source for the experiments detailed in the following sections.

Currently, there is no reported medium that selectively cultivates Saccharomyces cerevisiae directly from environmental samples. Most media capable of supporting the growth of S. cerevisiae and other yeasts also permit the proliferation of other fungi, such as molds. Therefore, antifungal components like propionic acid are commonly added to yeast isolation media. However, these additives do not completely inhibit the growth of competing mold species. Consequently, when environmental samples are directly applied to such isolation media, molds tend to dominate due to their inherent microbial community structure, preventing effective yeast isolation. For bacteria, strong inhibitors like chloramphenicol can be used to significantly reduce their presence [28] .

As revealed in the previous section, S. erevisiae is extraordinarily rare in environmental microbial communities, even in places like grape surfaces, traditionally thought to be rich in brewing seed yeasts. This rarity necessitates enrichment culture as a key strategy for isolation. Indeed, several enrichment media have been proposed previously. In this study, we used a koji extract-based enrichment medium (condition-1), leveraging the fact that Yeasticidin, produced by koji mold during sake brewing’s saccharification process, inhibits non-S. cerevisiae yeasts [14] . We also utilized a medium tailored for S. cerevisiae’s resilience to the osmotic pressure of high glucose concentrations, containing over 50% glucose (condition-2) [15] . Additionally, a medium was prepared based on S. cerevisiae’s unique ability to metabolize raffinose, a sugar most other yeast species cannot utilize (condition-3) [16] .

The identification of isolated strains, as depicted in Table 1 , showed similar fungal communities across all three conditions, predominantly consisting of Lachancea, Metschnikowia, Hanseniaspora, and Candida genera. This indicates that the specific medium used, whether influenced by Yeasticidin, high glucose concentration, or raffinose utilization, did not significantly alter the diversity of yeast and fungal species isolated from the Camellia petals. A particularly noteworthy outcome from condition-3 was the isolation of BR-38-1, which is closely related to, and perhaps a strain of, S. cerevisiae. This marked a unique instance in this study where S. cerevisiae was successfully isolated, an observation not replicated in this chemostat cultivations, which will be discussed in a later section. The batch enrichment system (denoted by “B” in strain names in Table 1 ) yielded a relatively diverse group of colonies, including one Saccharomyces cerevisiae strain. While species-level resolution based on ITS sequencing can sometimes be limited, especially among closely related taxa, the fact that BR-38-1 was the only isolate assigned to the genus Saccharomyces—whereas all others belonged to distinct genera—supports the robustness of this identification at least at the genus level. Further genomic analysis would still be necessary to confirm its precise species classification within the Saccharomyces genus.

While comprehensive reports detailing yeast colony profiles from environmental samples are scarce, Iturritxa’s study on wild yeast strains isolated from tree bark, soil, and acorns in northern Spanish forests provides a valuable comparison [29] . They utilized a Sniegowski enrichment medium containing moderate sugar and ethanol concentrations, along with chloramphenicol to inhibit bacterial growth. Notably, the sugar concentration in their medium was much lower than in our high-glucose condition-2, suggesting that a lower concentration of around 1% sugar is effective for yeast enrichment. Parallel to Iturritxa’s findings, S. cerevisiae did not emerge as the dominant species in our enriched material across conditions 1 - 3. This observation underlines that while certain media compositions are known to support S. cerevisiae growth, they may not exclusively promote its enrichment, showing the need for more targeted approaches to isolate S. cerevisiae specifically.

The batch enrichment system (denoted by “B” in strain names in Table 1 ) yielded a relatively diverse group of colonies, including one Saccharomyces cerevisiae strain. Over time, batch enrichment involves the fermentation of sugars into ethanol, lactic acid, and acetic acid, leading to dynamic substrate composition changes and consequent shifts in the fungal community. For instance, raffinose, a trisaccharide composed of glucose, fructose, and galactose, may not exclusively favor S. cerevisiae during selective enrichment as its hydrolyzed components can serve as substrates for other yeast species. This raises a question about the effectiveness of raffinose as a selective factor for S. cerevisiae enrichment. If the failure to achieve S. cerevisiae dominance in batch enrichment is attributed to dynamic

changes in medium composition, then stabilizing these changes through chemostat cultivation might enable easier and more pronounced dominance of a species well-adapted to the medium, ideally S. cerevisiae. Based on this hypothesis, conditions 4 - 6 utilized chemostat cultivation for enrichment, resulting in 23 isolates. Intriguingly, all isolates from these systems, except for one (Torulaspora sp. strain CE-4), were identified as Lachancea thermotolerans with 100% sequence match. This unprecedented result challenges the traditional understanding of S. cerevisiae’s “final” dominance in brewing, attributed to high sugar osmotic pressure [15] , pH reduction by lactic acid [17] , and the bactericidal effect of ethanol [18] . It suggests that these factors may not necessarily lead to the dominance or exclusion of S. cerevisiae and other yeast species under constant selective pressure. Instead, Lachancea species seem more adaptable to these specific media conditions.

These findings clearly indicate that S. cerevisiae dominance is not solely achieved through selective pressure by medium composition. This raises a crucial question: how is S. cerevisiae dominance achieved in the brewing process? While this study does not provide a conclusive answer, the occasional isolation of S. cerevisiae in the “B” series suggests a complex interplay of dynamic substrate changes, secondary metabolites produced by other yeast species, and the source of inoculation. These results imply that a combination of multiple factors, rather than just medium composition, influences the ecological dynamics and dominance of S. cerevisiae in natural and industrial settings.

In the batch enrichment systems, as identified by the “B” series in Table 1 , we obtained a diverse array of colonies, including a strain of Saccharomyces cerevisiae. This diversity prompts a crucial question: which is more effective for isolating non-dominant yeast species such as S. cerevisiae—streaking or dilution plating? Our literature review revealed a lack of documentation on the use of dilution plating for wild yeast isolation, a method commonly employed in bacterial isolation. This suggests that dilution plating, despite its potential effectiveness, has been underutilized in yeast isolation, particularly for species without mycelial growth. Given that streaking is traditionally used for purifying dominant species within a culture, its effectiveness for isolating non-dominant species like S. cerevisiae is questionable. In contrast, dilution plating allows for the isolation of less dominant species, provided their presence in the culture is around 0.5% or more. However, it’s important to note that the probability of isolating less prevalent species proportionally decreases with their occupancy rate in the culture. These findings indicate that, considering S. cerevisiae’s status as a non-dominant species in enrichment cultures, dilution plating is the preferable method.

Remarkably, no prior reports on wild yeast isolation have discussed the limitations of current batch cultivation methods or the significance of dilution plating. Most literature vaguely describes the process as “spreading the enrichment culture on agar media to form colonies,” without specifying the method used (e.g. [9] [10] and [27] ). We propose that the lack of emphasis on S. cerevisiae’s non-dominance and the underuse of dilution plating have been significant barriers to successfully isolating wild strains of S. cerevisiae in subsequent research. Supporting this hypothesis, we obtained some strains directly through streaking from the enrichment cultures (e.g., BG-09-1, BG15-3, BR-11-1), all of which were identified as belonging to dominant genera Metschnikowia and Hanseniaspora. In contrast, the closely S. cerevisiae-related strain BR-38-1 was isolated using dilution plating, reinforcing the effectiveness of this method for isolating non-dominant species.

In our final exploration, we attempted to isolate Saccharomyces cerevisiae using a selection pressure different from substrate utilization, albeit unsuccessfully ( Figure 3 ). As Liti et al. have demonstrated [8] , the environmental presence of S. cerevisiae is broader than previously assumed, and humanity has likely accessed only a fraction of its genetic diversity. This raises the question: does substrate utilization-based selection merely isolate a limited subset of S. cerevisiae strains? To advance this discussion, future genomic sequencing of our isolated strain BR-38-1 and further analysis of the genetic diversity among flower-derived S. cerevisiae strains are warranted. To access a broader range of S. cerevisiae bioresources, we explored a novel selection method based on genetic transformability—another hallmark trait of yeast biology alongside fermentative metabolism.

Specifically, following the method reported by Kimura [19] , we introduced plasmids pAUR112 and pAUR101—both conferring aureobasidin A resistance—into the enrichment culture under condition-3. These plasmids require either autonomous replication or genomic integration, and their function hinges on the successful intracellular expression of resistance genes. Consequently, this system imposes dual selection pressures: plasmid compatibility and drug resistance.