Fungal diversity on brewery filling hall surfaces and quality control samples

Abstract Breweries produce an increasing selection of beer and nonbeer beverages. Yeast and filamentous fungi may compromise quality and safety of these products in several ways. Recent studies on fungal communities in breweries are scarce and mostly conducted with culture‐dependent methods. We explored fungal diversity in the production of alcoholic and nonalcoholic beverages in four breweries. Samples were taken for next generation sequencing (NGS) at the key contamination sites in 10 filling lines. Moreover, fungal isolates were identified in 68 quality control samples taken from raw materials, filling line surfaces, air, and products. NGS gave a comprehensive view of fungal diversity on filling line surfaces. The surface‐attached communities mainly contained ascomycetous fungi. Depending on the site, the dominant genera included Candida, Saccharomyces, Torulaspora, Zygosaccharomyces, Alternaria, Didymella, and Exophiala. Sanger sequencing revealed 28 and 27 species of yeast and filamentous fungi, respectively, among 91 isolates. The most common species Saccharomyces cerevisiae , Zygosaccharomyces rouxii, and Wickerhamomuces anomalus were detected throughout production. Filling line surface and air samples showed the greatest diversity of yeast and filamentous fungi, respectively. The isolates of the most common yeast genera Candida, Pichia, Saccharomyces, and Wickerhamomyces showed low spoilage abilities in carbonated, chemically preserved drinks but could grow in products with reduced hurdles. Preservative resistant yeasts were rare, belonging to the species Dekkera bruxellensis, Pichia manschurica, and Zygosaccharomyces bailii. Penicillium spp. were dominant filamentous fungi. The results of this study help to evaluate spoilage risks caused by fungal contaminants detected in breweries.


| INTRODUCTION
Despite many advances in controlling microbiological spoilage of beverages, microbial contaminants occasionally cause various defects that are perceptible to a consumer, and liable to cause dissatisfaction, complaint, or rejection of the product. Economic and environmental consequences of the spoilage incidents may be substantial (Begrow, 2017;Stratford, 2006). Fungi, in particular yeasts, are among major spoilage organisms in acid beverages (pH < 4.5) due to their adaptation to acidic habitats containing sugar, alcohol, and/or chemical preservatives and having low oxygen tensions (Juvonen et al., 2011;Stratford, 2006). The category of acid beverages includes a range of alcoholic and nonalcoholic products such as beer, beer mix beverages, ciders, carbonated soft drinks, health and sports drinks, and enhanced waters. Riedl et al. (2019) recently estimated that yeasts cause more than 90% of spoilage incidents in low alcohol beers and beer-mix beverages. Typical signs of yeast spoilage include swelling and even explosion of packages, development of turbidity, sediments, flocs, or surface films and various off-flavors described as phenolic, fermented, floral, or vinegar (Hutzler et al., 2012;Stratford & James, 2003). Fermentative yeasts also produce ethanol, which may turn nonalcoholic into an alcoholic drink. The growth of filamentous fungi may lead to formation of hydrolytic enzymes, various off-flavors and odors, mycelial mats, and discoloration and even allergens and mycotoxins (Filtenborg et al., 2004;Juvonen et al., 2011).
Fungal contaminants may initially find their way into beverage production from various sources such as raw materials, packaging material, insects, and humans (Storgårds & Priha, 2009;Stratford, 2006). Many modern breweries produce a range of beverages, which increases the number of possible microbial sources entering the facility. Microorganisms, including fungi, typically colonize niches that contain moisture and nutrients and are difficult to clean and disinfect (Storgårds & Priha, 2009;Stratford & James, 2003). It has been estimated that 95% of soft drink yeast spoilage incidents are due to poor factory hygiene (van Esch, 1987). In breweries and soft drink factories not using in pack pasteurization, the origin of spoilage incidents can often be traced back to filling machines and the surrounding environment that provide suitable conditions for microbial growth and biofilm formation (Storgårds & Priha, 2009;Timke et al., 2007). Fungi have been found in brewery biofilms and they are among the first microorganisms to attach on filling line surfaces after cleaning (Storgårds et al., 2006;Timke et al., 2005aTimke et al., , 2005b. These pioneer organisms pave the way for other microbes, including spoilage species, but do not necessarily grow in beer themselves. The past surveys of yeast ecology in breweries and soft drink factories have revealed an extensive list of species (for review, see Hutzler et al., 2012;Stratford, 2006;Stratford & James, 2003). In practice, a limited number of species is able to grow during production or in the final drinks produced under good manufacturing practices.
The specific spoilage species are mainly dictated by intrinsic product properties and preservation treatments. In large breweries, beer is mostly brewed with pure yeast, and the term wild yeast is used to refer to any other yeast strain. Saccharomyces species, in particular Saccharomyces cerevisiae and other Saccharomyces sensu stricto species, pose the greatest threat to beer quality both during brewing and in the pack by being strongly fermentative under low oxygen tensions (Hutzler et al., 2012;Kühle & Jespersen, 1998;Pham et al., 2011;Timke et al., 2007). Many aerobic or weakly fermenting Candida, Pichia, and Wickerhamomyces (including some former Pichia) species have been identified in brewery raw materials and equipment surfaces (Kühle & Jespersen, 1998;Pham et al., 2011;Storgårds et al., 2006;Timke et al., 2007). They may proliferate during initial phases of fermentation or in final beer in cases of oxygen ingress.
Yeast contaminants associated with soft drink production have been traditionally classified according to spoilage potential (Davenport, 1996;Hutzler et al., 2012;Stratford & James, 2003). The most dangerous spoilage yeasts in carbonated products include approximately 10-12 fermentative preservative-tolerant species, the most important being Zygosaccharomyces spp. (especially Zygosaccharomyces bailii), Dekkera anomala, D. bruxellensis, Dekkera naardenenesis, S. cerevisiae, Kazahstania exigua (former Saccharomyces exiguus), and Schizosaccahromyces pombe. In practice, opportunistic species that grow in the product only if some of the intrinsic hurdles are lowered due to errors in manufacturing or high microbial load cause most of the spoilage incidents (Stratford, 2006).
Filamentous fungi contaminations are usually due to poor factory hygiene or due to growth of heat-resistant species in heat-processed beverages (Wareing, 2016). They usually originate from outdoor air or soil (Tribst et al., 2009;Wareing, 2016). Any airborne filamentous fungi can contaminate finished products, but vigorously sporulating species are the most common in the soft drink and juice industry (Wareing, 2016). These species belong to genera Penicillium, Aspergillus, Eurotium, Fusarium, Cladosporium, and Alternaria. Heat-resistant filamentous fungi able to cause spoilage of soft drinks include species from genera Aspergillus, Byssochlamys, Peacilomyces, Phialaphora, and Talaromyces (Wareing, 2016). They can even survive flash pasteurization of 90 C to 96 C for 30-90 s or tunnel pasteurization of 72 C to 80 C for 5-20 min and can thus grow in pasteurized products (Juvonen et al., 2011). Several spoilage filamentous fungi can grow in low pH, and although they usually require oxygen to grow, some species can grow under anaerobic conditions with fermentative metabolism (Filtenborg et al., 2004).
In response to consumer demands, the diversity of beer and nonbeer beverage products is continuously increasing and traditional hurdles for microorganisms (such as chemical preservatives and alcohol) are being reduced, which is creating new opportunities for microbial contaminations and spoilage and could also affect fungal diversity in this environment (Hutzler et al., 2008;Juvonen et al., 2011). Much of the current knowledge on fungal populations in breweries is relatively old and mainly gathered using cultivation and isolation techniques (Kühle & Jespersen, 1998;Pham et al., 2011;Storgårds et al., 2006;Timke et al., 2007). The present study was undertaken to examine fungal diversity at the key contamination sites of filling lines in four modern breweries with next generation sequencing (NGS). This technique could provide new insight into fungal diversity, providing an opportunity to get a good coverage of the communities with reasonable work (Priha et al., 2016). Moreover, a culture-dependent method was used to identify various types of fungi in quality control (QC) samples and spoilage potential of selected isolates was evaluated.
2 | MATERIALS AND METHODS 2.1 | Fungal community profiling of surface samples 2.1.1 | Surface samples Surface samples were obtained from filling lines in four breweries to assess fungal community composition on the surfaces. The breweries performed the sampling between February and April 2017. In total, 55 samples were collected during the production of various beer and nonbeer beverages. Each brewery provided 6-12 samples from three to four product categories (Table 1). At the time of sampling, the filling lines had been typically operating 1-5 days since last extensive washing. Some breweries performed mild washes during production.
Surface area of 10 cm Â 10 cm was swapped with sterile nonwoven gauzes (Mesoft, Mölnlycke Health Cara, Gothenburg, Sweden), which were placed immediately in 30 ml of sterile 0.9% sodium chloride (Merck, Darmstadt, Germany) solution. The samples were transported to the molecular biology laboratory in a cold box (+6 C) the same day or at the latest the next morning. Yeasts and filamentous fungi were detached from the gauzes by homogenization for 1 min in a Stomacher blender (Seward Laboratory System, Worthing, UK).
Before DNA extraction, the viable counts of filamentous fungi and yeasts in the homogenates were determined on YM agar medium with chloramphenicol (100 mg l À1 ) (PD/Difco, USA).

| DNA extraction
The homogenized swab sample suspensions were filtered through

| Quantitative fungal PCR analysis
The presence and the amount of fungal DNA was determined with a Other alcoholic beverage 4 0/0 1/0 2/1 1/0 4/1 with primers 5.8F1 5.8R1 and 5.8P1 (Haugland & Vesper, 2002 The primers were identified from the sequences and removed using a cutadapt tool (Martin, 2011). Quality of the sequence reads was checked according to DADA2 workflow. Next, the sequences were filtered and trimmed using DADA2 filterAndTrim function. Filtering parameters maxN = 0, maxEE = c(2, 2), truncQ = 2, and minLen = 50 were used. Minimum length 50 bp was used to remove spurious very low-length sequences. The maximum possible error rates were calculated using the learnErrors command. Identical reads were dereplicated (unique sequences). Amplicon sequence variants of the sequence data were identified using DADA2 pipelines core sample inference algorithm. Denoised paired reads were merged according to the DADA2 pipeline and an amplicon sequence variant table (ASV) was constructed. Subsequently, chimeric sequence reads were removed from the data set with remove BimeraDeNovo function, using the consensus option. Finally, taxonomy from domain to genuslevel was assigned to ASVs with DADA2's native implementation of the naive Bayesian classifier method. Taxonomy was assigned against UNITE database version 8.2 (2020-02-04) (Kõljalg et al., 2013). All images of the sequencing data were constructed with R using packages the phyloseq (McMurdie & Holmes, 2013) and ggplot2 (Wickham, 2015). Alpha diversity indexes chao1 (Chao, 1984) and Shannon diversity index (Shannon, 1948) (Morgulis et al., 2008) in Geneious software.

| Growth tests in commercial beverages
Spoilage ability of selected yeast isolates was studied in commercial beverage products. The products included a simple sugar-containing soft drink preserved with sorbate (pH 2.9), a preservative-free juicecontaining soft drink (pH 2.7), a still water product containing sugar and benzoate as major constituents (pH 4.3), a beer with 4.5% (v/v) alcohol (pH 4.2), an apple cider with 4.5% (v/v) alcohol and sulfite (pH 3.0) and a preservative-free nonbeer beverage with 5.5% (v/v) alcohol (pH 2.8). The products were aseptically distributed in 9-ml aliquots into 10-ml plastic screw capped tubes (Greiner bio-one, Austria Germany) as required. Each product was inoculated at 10 3 -10 4 cfu ml À1 in duplicate. Inoculated and uninoculated tubes of each product were incubated at 25 C for 6 weeks. Turbidity was assessed every week visually and by measuring turbidity at 620 nm wavelength (Multiskan EX instrument, Thermo Labsystems, Finland).
The growth result was scored as no, weak, moderate and intense when the turbidity increase was <1.5-fold, 1.5 to twofold, more than twofold but less than fourfold, and more than fourfold compared with the uninoculated control, respectively. Samples that were turbid by nature were cultivated on YM agar plates (BD/Difco). At the end of the follow up period, viability and purity of the inoculated yeast strains was confirmed by plate cultivation and final pH was measured.
Moreover, the inoculated and noninoculated samples were sniffed by a single person to detect any obvious off-flavors. In total, 203 ASVs were detected in the sample set. The number of observed ASVs in the samples varied between 4 and 48 ( Figure S2).
When comparing the Chao1 ASV richness estimate values to true observed ASV numbers, all of the estimated fungal ASVs were obtained from the sequence data ( Figure S2), meaning that sequencing depth was sufficient to fully characterize the fungal communities in all of the samples.
Fungi from phylum Ascomycota dominated fungal communities in all samples ( Figure S1). Fungi affiliated to Basidiomycota were found in approx. one third of the samples. Their relative abundance exceeded 20% only in four samples. Furthermore, low relative abundancies (<10%) of Mucoromycota or unaffiliated fungi were detected in 40% of the samples.
Saccharomycetes was the dominating fungal class (relative abundance >50%) in 72% of the samples (Figure 1). Filamentous fungi from the classes Dothideomycetes, Eurotiomycetes, Lecaronomycetes, Sordariomycetes, and Leotiomycetes were detected in varying relative abundancies, depending on the sample. Basidiomycota on the filling line surfaces mainly belonged to the classes Malasseziomycetes and Tremellomycetes.
In total, 13 and six ascomycetous yeast genera were detected in the surface samples at above 0.1% and 1% relative abundance, respectively. On average, the principal yeast genera detected were Saccharomyces (29%), Candida (11.4%), Wickerhamomyces (7.5%), Torulaspora (7.4%), and Zygosaccharomyces (  Fifty-nine independent filamentous fungi isolates were obtained from the QC samples ( Figure 4b, Table S1). In total, 32 isolates were identified from air samples, three from filling line surface samples, 10 from soft drink products, 13 from beer products, and one from other alcoholic products. All of the 27 filamentous fungal species isolated from QC samples belonged to Ascomycota phylum. Penicillium (61% of the isolates) was the most common genus in the samples ( Figure 4). Other genera isolated included Phoma (7.1%), Talaromyces

| Spoilage ability of fermentative yeast isolates
Spoilage ability of 28 isolates of fermentative yeast species was studied in different types of alcoholic and nonalcoholic beverage products using a challenge test (Table 2). Representative strains of each species identified from the various product samples and those raw material or process isolates not identified in the product samples were included in the study. Nonfermentative species and filamentous fungi were excluded due to their expected low spoilage potential.
Turbidity increase was the main sign of yeast spoilage in clear and slightly opaque products. Visual turbidity typically appeared within 1-3 weeks after inoculation. Obvious off-flavors, noticeable by sniffing of the products, or pH changes were not detected at the end of the 6-week follow-up period (data not shown). Overall, the strains of the species D. bruxellensis, P. manshurica, and Z. bailii showed the highest spoilage ability, growing in four out of the six studied products (

| DISCUSSION
The making of beer and nonbeer beverages in a brewery environment is not a fully aseptic process, and despite regular cleaning, microorganisms tend to accumulate on process equipment surfaces and in the surrounding environment from raw materials and other sources (Bokulich et al., 2015;Storgårds et al., 2006;Stratford & James, 2003). Filling line machines are especially favorable niches for microbial attachment and growth due to the presence of product residues and water at ambient temperatures (Storgårds et al., 2006).
In the present study, NGS was applied to characterize fungal communities building up at the key contamination sites of brewery filling lines during the production of beer or nonbeer beverages. Beer w À w w À À Note: All products except enhanced water were carbonated. They were distributed in 9-ml aliquots in 10-ml plastic tubes and inoculated at 10 3 -10 4 cfu ml À1 , followed by incubation at 25 C for 6 weeks. The growth was measured weekly using turbidometry at 620 nm. a Turbidity increase less than 1.5-fold. b Moderate growth (turbidity increase more than twofold but less than fourfold). c Intense growth (turbidity increase more than fourfold). d Weak growth (turbidity increase 1.5 to twofold). e One out of two replicates.
($1000 cfu/cm 2 ) compared with bacterial counts throughout the 8-week follow up period. Timke et al. (2005a) detected yeast-derived fatty acids in nearly all mature biofilm samples taken from beer bottling plants of two breweries, but in another study, FISH-signals for eukaryotic microorganisms at two bottle conveyers were extremely low (Timke et al., 2005b).
Much of the available knowledge of fungal populations on brewery filling lines derives from culture-dependent analyses. NGS gave a comprehensive overall picture of fungal diversity on the filling line surfaces. The fungal diversity at various sites was low (number of ASVs from 4 to 48) compared with bacterial diversity revealed in beer filling lines with NGS (number of OTUs from 71 to 376) (Priha et al., 2016) or other methods (Maifreni, Frigo, Bartolomeoli, Buiatti, Picon, et al., 2015;Timke et al., 2005b). The short and relatively conserved ITS marker gene region provides at best species group or genus level assignment and underestimates true species richness. In line with previous studies, fungal populations on the filling line surfaces were mainly composed of Ascomycetes (Bokulich et al., 2015;Hutzler et al., 2012;Kühle & Jespersen, 1998;Pham et al., 2011;Timke et al., 2007). NGS revealed a higher number of yeast genera than earlier reported with culture-based methods. In part this is due to changes in nomenclature and in part could reflect the wide range of beverages produced in the studied breweries and detection of dead and uncultivable organisms using NGS. However, only limited number of genera, including Candida, Saccharomyces, Torulaspora, Wickerhamomyces, and Zygosaccharomyces, were detected at above 1% abundancies. Apart from Zygosaccharomyces, these yeast genera have been frequently identified on brewery filling line surfaces with other methods (Storgårds et al., 2006;Timke et al., 2007).

NGS showed that filamentous fungi and black yeasts with
Ascomycota affiliation dominated the fungal communities in some of the filling line sites, indicating their possible role in the microbial communities. Filamentous fungi have been shown to contribute to ecology of drinking water distribution system biofilms, for example, by providing support to the colonization of bacteria (Douterelo et al., 2018) and many ascomycetous genera and species are capable of biofilm formation (Siqueira & Lima, 2013). Didymella, the most often detected genus with NGS, is a plant pathogen originating from soil and infecting various plant including barley grains. It has not been linked with food or beverage spoilage (Chen et al., 2017). The black yeast Exophiala detected on the filling line surfaces is a common environmental fungi often associated with decaying wood, plants, and soil (Matos et al., 2002), but also detected in dirty bottles and the brewing process. Exophiala spp. may produce exopolysaccharides, which could promote the adherence and survival of the cells on the surfaces (Matos et al., 2002). No beverage spoilage incidents have been linked with this genus (Pitt & Hocking, 2009). Genus Flavoplaca that was the dominant genus detected in sample P11_B is a lichen frequently detected in Finland and can also grow in buildings and concrete (Stenroos et al., 2016) but no association with food or beverage spoilage have been detected.
The most common fungi within Basidiomycota phylum detected in surface samples from filling lines were affiliated to yeast genera Hannaella and Malassezia. Malassezia is a dominant component of the mycobiota on human skin (Amend, 2014). Recently, molecular methods have revealed that these difficult to cultivate species can be found in a diversity of habitats (Amend, 2014). Strains of the genus Hannaella have been isolated on the external surfaces of plants, which is a common habitat for many basidiomycetous yeasts (Kaewwichian et al., 2015).
Microbial accumulation on brewery filling lines can be affected by a multitude of factors including the design of the filling machine and the specific location within the equipment, the cleaning regimes, the products being filled as well as microbial sources from raw materials and environment (Bokulich et al., 2015;Priha et al., 2016). Although the present study was not designed to explore factors affecting the fungal community composition on filling lines, some associations between the sampling location and fungal community structure were detected. The filling line surfaces in two breweries occupied fungal communities characteristic to each brewery according to PCoA analysis. Despite relatively low percentage of the variance explained by the first two PCs, biological patterns may still revealed (Goodrich et al., 2014;Kuczynski et al., 2010). As the other two breweries were  Timke et al. (2005aTimke et al. ( , 2005b concluded that there is no typical biofilm community, not even for distinct regions of the bottling plant. However, the major wild yeast species isolated from brewery bottling line biofilms did not show differences (Timke et al., 2007).
The diversity of yeast species identified in the QC samples with the culture-dependent method reflected the range of beer and nonbeer beverages produced in the studied breweries. It needs to be noted that each brewery used their own specific cultivation methods for isolation of fungi, which may have affected the observed species diversity. In breweries, S. cerevisiae, Z. rouxii, and W. anomalus were the most frequently isolated species. S. cerevisiae and W. anomalus have been reported as major contaminants in pitching yeast (Kühle & Jespersen, 1998), beer fermentation and conditioning vessels (Pham et al., 2011), and brewery filling lines (Storgårds et al., 2006;Timke et al., 2007). We detected W. anomalus and Z. rouxii from raw materials as well as final products, suggesting raw materials as an initial contamination source. W. anomalus is a ubiquitous species known to contaminate also beverage ingredients, whereas the association of Z. rouxii with spoilage of high sugar raw materials is well established (Laitila et al., 2010;Martorell et al., 2007;Stratford & James, 2003). Saccharomyces yeasts, on the other hand, occur widely in a brewery environment (Bokulich et al., 2015). The Saccharomyces isolates recovered from the QC samples grew well in beer, but showed at best limited growth in carbonated, chemically preserved beverages. Saccharomyces yeasts are well known beer spoilers (Hutzler et al., 2012;Kühle & Jespersen, 1998;Timke et al., 2007), whereas their ability to spoil chemically preserved beverages varies (Stratford & James, 2003). Our results are also in line with earlier findings that weakly fermenting W. anomalus species rarely spoils carbonated beverages or beer unless oxygen is available, but may grow in a variety of still beverages (Hutzler et al., 2012;Kühle & Jespersen, 1998;Pham et al., 2011;Timke et al., 2007). The Z. rouxii isolates characterized in the present study can also be considered harmless contaminants in the end products owing to their poor growth in acid and chemically preserved soft drinks and their inability to use the main carbohydrates of beer for growth (Krogerus et al., 2021). However, some strains of Z. rouxii can be resistant to food preservatives especially in high sugar media (Martorell et al., 2007;Stratford, 2006;Stratford & James, 2003).
The QC samples from filling line surfaces contained a greater diversity of yeast species compared with the raw material and product samples and extremophiles were not detected. This reflects the unselective nature of the filling line environment as the preservative factors are diluted in the product residues and oxygen is available for the growth of aerobic species (Maifreni et al., 2015;Timke et al., 2005a). Especially different aerobic and weakly fermenting Candida and Pichia species were common in line with previous studies (Storgårds et al., 2006;Timke et al., 2004). It is interesting to note  (Storgårds et al., 2006;Timke et al., 2007). W. anomalus has been considered a pioneer organism in brewery biofilms owing to its´frequent association with biofilms and ability to form biofilm. Similarly, many Candida species found in the production of beer, fruit juice and fermented foods have shown propensity for biofilm formation (Storgårds et al., 2006;Timke et al., 2007;Zara et al., 2020). Candida sojae isolated in the present study was also shown to form biofilm, whereas Trigonopsis cantarellii was not (Krogerus et al., 2021). S. cerevisiae isolates from beer bottling lines lacked biofilm forming ability and were thought to colonize more mature biofilms (Timke et al., 2007).
Zygosaccharomyces spp. have been found in the surface flocs of aging wines and in fruit juice processing plant biofilms and may also form biofilm on abiotic surfaces (Tristezza et al., 2010;Zara et al., 2020).
This study reinforces the previous findings that many yeast contaminants associated with beer and nonbeer beverage production are opportunistic spoilers that may grow in final products only in case of some process failure or if the contamination level overrides the efficacy of the preservative system (Stratford, 2006;Stratford & James, 2003). This was shown by the ability of many of the yeast isolates to grow in the preservative-free soft drink. Moreover, nearly all species isolated from the QC samples grew in beer under aerobic conditions, but the growth of weakly fermenting Candida, Pichia, and Wickerhamomyces spp. was prevented in oxygen limited conditions (data not shown). Of the species isolated in the present study, C. parapsilosis, C. sojae, Clavispora lusitaniae, P. membranifaciens, P. kudrivazevii, K. exiqua, T. delbrueckii, and W. anomalus have been classified as commonly encountered second division, group 2 spoilage/hygiene yeast species in soft drink factories (Stratford, 2006). The results of the present study also imply that some of the new types of beverage, such as the moderately acidic, still water product containing sugar and benzoate, support the growth of opportunistic spoilers. In particular, the ubiquitous W. anomalus species could be a threat in this kind of beverages. Furthermore, the addition of new extract sources into beer when producing beer mix beverages could render the products susceptible to spoilage by various Zygosaccharomyces species.
Our results support the previous findings that the preservative resistant yeasts are relatively rare in beverage production, but if access is gained, they can cause spoilage of a variety of alcoholic and nonalcoholic drinks (Stratford, 2006). Z. bailii and D. bruxellensis were the only species growing abundantly in the sorbate preserved beverage and sulfite was tolerated only by Pichia manschurica and D. bruxellensis isolates. D. bruxellensis and Z. bailii are well documented preservative resistant yeast spoiling various alcoholic and nonalcoholic beverages (Dimopoulou et al., 2019;Hutzler et al., 2012;Martorell et al., 2007;Smith & Divol, 2016). Although P. manshurica has been mostly linked with spontaneous wine fermentations and wine spoilage (Perpetuini et al., 2021), it has also been found in breweries (Turvey et al., 2016). Wine spoilage isolates have shown variable resistance to sulfites and some isolates readily formed biofilm (Perpetuini et al., 2021;Tristezza et al., 2010).
Common spoiling filamentous fungi originating usually from outdoor air are species from genus Aspergillus and Penicillium that were also the most common filamentous fungi detected in the QC samples together with species from genera Talaromyces and Phoma.
Talaromyces spp. are heat-resistant filamentous fungi usually of soil origin and are frequently detected in pasteurized fruit-based products including flavored mineral waters (Hocking & Pitt, 2001;Pitt & Hocking, 2009). Species from genus Phoma are common outdoor and soil fungi that have been previously detected also in soft drink manufacturing facilities (Aoyama & Miyamoto, 2016).
Filamentous fungi usually require oxygen to grow. One exception is Paecilomyctes variotii which was detected in QC samples and can grow in microanaerobic conditions (Wareing, 2016). However, it is rarely able to grow in carbonated products. Paecilomycetes genus is an anamorph of heat-resistant ascomycete genus Byssochlamys (Samson et al., 2009). Some filamentous fungi produce mycotoxins that cause a food safety concern for humans. P. variotii is able to produce mycotoxin viriditoxin and Fusarium oxysporum that was also detected in QC samples produces oxysporone that is commonly detected in treated orange juice (Wareing, 2016). In addition to mycotoxin production, raw material contamination with Fusarium species can also lead to production of gushing inducers and gushing of beer have been detected in beers produced from malts contaminated with Fusarium species (Sarlin et al., 2005;Sarlin et al., 2007).

| CONCLUSIONS
This study explored fungal diversity in modern breweries producing beer and nonbeer beverages. NGS analysis was found to be a good tool for obtaining a comprehensive view of fungal communities on filling line surfaces. It revealed higher number of fungal genera in association with brewery filling lines than earlier reported when applying other methods. The fungal species identified in QC samples reflected the range of beer and nonbeer beverages produced in the studied breweries. Majority of the isolated yeast contaminants did not appear to pose a spoilage threat in carbonated, chemically preserved beverages or in beer but could play a role in the establishment of biofilm on equipment surfaces or in the spoilage of new types of beverage products. Preservative resistant species were rare. This study underlies the importance of maintaining good process hygiene especially when producing beverages with reduced hurdles. The findings of the study may be applied to evaluate harmfulness of fungal contaminants detected in breweries.

ACKNOWLEDGMENTS
We thank Merja Salmijärvi, Jenni Limnell and Anne Heikkinen for technical assistance. We thank PBL Brewing Laboratory (Oy Panimolaboratorio -Bryggerilaboratorium Ab) and Business Finland for funding the study.

CONFLICT OF INTEREST
PBL Brewing Laboratory funded this research. The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.