YEAST RESEARCH. Not your ordinary yeast: non-saccharomyces yeasts in wine production uncovered. Natural yeast and natural wine a rather unnatural tale

Transcription

1 MINIREVIEW Not your ordinary yeast: non-saccharomyces yeasts in wine production uncovered Neil P. Jolly 1, Cristian Varela 2 & Isak S. Pretorius 3 1 ARC Infruitec-Nietvoorbij, Stellenbosch, South Africa; 2 The Australian Wine Research Institute, Adelaide, SA, Australia; and 3 Macquarie University, Sydney, NSW, Australia YEAST RESEARCH Correspondence: Isak S. Pretorius, Macquarie University, Balaclava Road, North Ryde, Sydney, NSW 2109, Australia. Tel.: ; fax: ; sakkie.pretorius@mq.edu.au Received 27 August 2013; accepted 7 October Final version published online 11 November DOI: / Editor: Jens Nielsen Keywords non-saccharomyces; yeasts; wine; aroma; flavour; fermentation. Abstract Natural yeast and natural wine a rather unnatural tale When grapes and yeast combine, wine emerges; however, when wine and people mix, opinions diverge. Sometimes these opinions are based on fact, and sometimes not, but turn to the Roman philosopher, Gaius Plinius Secundus better known as Pliny the Elder in his First Century encyclopaedic work, Naturalis Historia, and we find in vino veritas, that is, the truth is in the wine! So, how can we find the truth through the debates of wine bloggers over the past decade or so first about organic, then biodynamic, and now natural wines when some journalists, importers and retailers are turning this Saccharomyces cerevisiae and grape juice are natural companions and make a happy wine marriage. However, this relationship can be enriched by allowing wild non-saccharomyces yeast to participate in a sequential manner in the early phases of grape must fermentation. However, such a triangular relationship is complex and can only be taken to the next level if there are no spoilage yeast present and if the wine yeast S. cerevisiae is able to exert its dominance in time to successfully complete the alcoholic fermentation. Winemakers apply various matchmaking strategies (e.g. cellar hygiene, ph, SO 2, temperature and nutrient management) to keep spoilers (e.g. Dekkera bruxellensis) at bay, and allow compatible wild yeast (e.g. Torulaspora delbrueckii, Pichia kluyveri, Lachancea thermotolerans and Candida/Metschnikowia pulcherrima) to harmonize with potent S. cerevisiae wine yeast and bring the best out in wine. Mismatching can lead to a two is company, three is a crowd scenario. More than 40 of the 1500 known yeast species have been isolated from grape must. In this article, we review the specific flavour-active characteristics of those non-saccharomyces species that might play a positive role in both spontaneous and inoculated wine ferments. We seek to present single-species and multi-species ferments in a new light and a new context, and we raise important questions about the direction of mixed-fermentation research to address market trends regarding so-called natural wines. This review also highlights that, despite the fact that most frontier research and technological developments are often focussed primarily on S. cerevisiae, non-saccharomyces research can benefit from the techniques and knowledge developed by research on the former. naked-as-nature-intended approach of wine production into an ideological crusade? When did wine become unnatural? It is a question worth asking, given today s debate about natural yeasts and natural winemaking practices, and claims by some commentators that natural wine is now the hottest category in the wine industry. Confusing messages leave many producers and consumers baffled. The answer requires a brief review of 7000 years of winemaking history. The first fermentation, for example, was more likely the result of serendipity rather than design. Spontaneously, ambient yeasts fermented damaged grapes in harvesting pots which mystified hunter-gatherers who

2 216 N.P. Jolly et al. established agriculture and the first great civilization in Mesopotamia around the Tigris-Euphrates river system and who tasted wine for the first time (Chambers & Pretorius, 2010). Enjoying the taste and psychotropic effects of their discovery both a pleasurable and storable drink they went on to harness natural events in repeated yearly experiments. However, even during those early vintages, it was clear that, without human intervention, the result of naturally fermenting grapes is variable, unreliable and can be undrinkable. It did not take long before the ancients realized that the completely natural end-result of fermenting grapes is vinegar. Throughout history, wine has retained a mythic aura a natural product cloaked in mystique. Could this be a contributing factor why some wine enthusiasts are so concerned that today s winemakers backed by frontier science and rigorous research have so much influence over the production process and so much opportunity to direct viticulture and vinification to shape wine according to consumer preferences? What is clear, however, that the pressure is on. There is heated argument as to whether today s wine is of higher quality due to the contribution of scientific knowledge, technology and research or whether so-called natural wine is better. There is a newfound nostalgia for the wine of yesteryear made with a minimalist approach and variable outcomes. Some traditionalists and proponents of natural wine reject, for example, the interventionist practice of inoculating grape must with selected cultured yeasts to avoid the risk of stuck ferments and off-flavours or to produce wine according to predetermined definable flavour specifications and styles. These are the hallmarks of industrial products, they say, not natural wine (for a review see Lewin, 2010, and references therein). On the other hand, there is a group of inventive winemakers and yeast researchers who are frustrated by such arguments, waiting impatiently to uncork their artistic creativity and artisanal craftsmanship alongside the next generation of technical innovation. As they have done throughout history, wine s innovators are keen to assist in the crafting of unique, stand-out wines that meet ever-shifting consumer expectations while underpinning profitability and sustainability. The reality is that winemaking is both art and science and always had been. The supposed dichotomy between natural and unnatural wine is a false one. History taught us that the best outcome for both winemaker and consumer is achieved when the wine industry harnesses what nature, human ingenuity and cutting-edge science offer in harmony with the unique artistic nature of wine. Here, we take stock of what nature s treasure trove of wild yeasts has on offer and how inventive winemakers can use them in a scientifically controlled manner to craft wine styles that match consumer expectations in a diverse range of market segments. Non-Saccharomyces yeasts a doubleedged sword worth investigating Since 1866 when Louis Pasteur first elucidated the bioconversion of grape juice into wine, this complex biochemical process and the role of the yeast therein has been studied continuously (Figs 1 and 2). The role of the primary yeast, Saccharomyces cerevisiae, often simply referred to as the wine yeast has received the most attention. This yeast is not only responsible for the metabolism of grape sugar to alcohol and CO 2 but has an equally important role to play in the formation of secondary metabolites, as well as in conversion of grape aroma precursors to varietal wine aromas (Reed & Peppler, 1973; Fleet, 1993, 2008; Darriet et al., 1995; Dubourdieu, 1996; Pretorius et al., 1999; Ribereau-Gayon et al., 2000; Pretorius, 2003; Howell et al., 2004; Swiegers & Pretorius, 2005; Swiegers et al., 2005). Grape musts naturally contain a mixture of yeast species and wine fermentation is not a single-species fermentation (Fleet, 1990). The dominance of S. cerevisiae (inoculated or indigenous) in the fermentation is expected and desired. However, the indigenous non-saccharomyces yeasts, already present in the must, and often in greater numbers than S. cerevisiae, are adapted to the specific environment and in an active growth state, which gives them a competitive edge (Cray et al., 2013). Consumption Packaging Labelling Bottling Membrane filtration Cold stabilisation Fining Blending Grape growing WHITE WINE MAKING Battonage Racking Clarification Malolactic fermentation (only for certain wine styles) Natural yeast or yeast addition Grape harvesting Sorting & triage Cooling Destemming Fermentation Crushing Pressing Clarification Maceration Fig. 1. A schematic outline of the main steps in white wine production. Some steps and the sequence thereof differ between the production of white and red wine (compare Fig. 2).

3 Non-Saccharomyces wine yeasts 217 Consumption Packaging Labelling Bottling Membrane filtration Cold stabilisation Fining Blending Battonage Racking Clarification Grape growing RED WINE MAKING Grape harvesting Sorting & triage Cooling Destemming Crushing Fermentation Pumping over Maceration Punch-down Malolactic Draining of gross lees Natural yeast fermentation Pressing or yeast addition Maceration Clarification Fig. 2. A schematic outline of the main steps in red wine production. Some steps and the sequence thereof differ between the production of red and white wine (compare Fig. 1). Non-Saccharomyces yeasts were originally seen as responsible for microbial-related problems in wine production due to their isolation from spoiled wines (Van der Walt & Van Kerken, 1958; Amerine & Cruess, 1960; Van Zyl & Du Plessis, 1961; Van Kerken, 1963; Rankine, 1972; Le Roux et al., 1973). Although it was known that some non-saccharomyces yeasts could form beneficial metabolites for wine quality (Castor, 1954; Amerine & Cruess, 1960; Van Zyl et al., 1963), this was outweighed by the high levels of volatile acidity and other negative compounds produced (Castor, 1954; Amerine & Cruess, 1960; Van Zyl et al., 1963; Amerine et al., 1967, 1972). This caused a blanket distaste for all non-saccharomyces yeasts. Authors of earlier publications considered non-saccharomyces yeasts to be sensitive to SO 2 added during wine production, to control their growth and that of spoilage bacteria (Amerine & Cruess, 1960; Van Zyl & Du Plessis, 1961; Amerine et al., 1972). Non-Saccharomyces yeasts were also known to be poor fermenters of grape must and intolerant to ethanol (Castor, 1954), especially in the presence of SO 2 (Amerine & Cruess, 1960; Amerine et al., 1972). It was therefore accepted that those non-saccharomyces yeasts, not initially inhibited by the SO 2, died during fermentation due to the combined toxicity of the SO 2 and alcohol. In contrast, winemakers conducting spontaneous fermentations (comprising mixed and sequential dominance of non-saccharomyces and Saccharomyces yeasts), viewed indigenous yeasts as integral to the authenticity of their wines by imparting desired and distinct superior regional characteristics (Amerine et al., 1972). Spontaneous fermented wines, although carrying a higher risk of spoilage, are generally regarded as having improved characteristics, such as complexity, mouth-feel (texture) and integration of flavours relative to inoculated wines (Heard & Fleet, 1985; Fleet, 1990; Bisson & Kunkee, 1991; Gil et al., 1996; Lema et al., 1996; Grbin, 1999; Heard, 1999; Soden et al., 2000; Varela et al., 2009). Later, research highlighted the high numbers (10 6 to 10 8 cells ml 1 ), and sustained presence of non-saccharomyces yeasts in modern wine fermentations, resulting in wine microbiologists revisiting the role of these yeasts. Consequently, their role in wine production has been debated extensively (Fleet et al., 1984; Heard & Fleet, 1985; Fleet, 1990, 2003; Herraiz et al., 1990; Longo et al., 1991; Romano et al., 1992; Todd, 1995; Gafner et al., 1996; Gil et al., 1996; Lema et al., 1996; Granchi et al., 1998; Henick-Kling et al., 1998; Lambrechts & Pretorius, 2000; Rementeria et al., 2003; Combina et al., 2005; Xufre et al., 2006; Varela et al., 2009; Ciani et al., 2010; Ciani & Comitini, 2011). Non-Saccharomyces yeasts, as the name suggests, refers to all yeast species found in wine production barring S. cerevisiae, with the proviso that this only includes yeast with a positive role in wine production. Recognized spoilage yeasts, such as Dekkera/Brettanomyces, are normally left out of this description. Although most fields of research are often focussed primarily on S. cerevisiae, non-saccharomyces research can benefit from the techniques and knowledge developed by the S. cerevisiae and other yeast researchers (Cray et al., 2013). Yeast classification Non-Saccharomyces yeast is a loose colloquial term used among wine microbiologists and in wine industries, which includes many different yeast species. These yeasts are either ascomycetous or basidiomycetous that have vegetative states which predominantly reproduce by budding or fission and which do not form their sexual states within or on a fruiting body (Kurtzman et al., 2011a). Current taxonomies recognize 149 yeast genera comprising nearly 1500 species (Kurtzman et al., 2011b). Of these, more than 40 species have been isolated from grape must (Jolly et al., 2006; Ciani et al., 2010). Yeasts may be known by two valid names, the teleomorphic name referring to the sexual state producing ascospores (Kurtzman et al., 2011a), and the anamorphic name referring to the asexual state that does not form ascospores. Yeast classification can be difficult because some yeasts do not sporulate easily and the ability to form ascospores can be lost during long-term storage (Kurtzman et al., 2011c). Delays between isolation and

4 218 N.P. Jolly et al. Table 1. Teleomorphs, anamorphs and synonyms (Kurtzman et al., 2011b) of some of the non-saccharomyces yeasts in the Ascomycetous genera reported on grapes and in wine fermentations (Jolly et al., 2006; Ciani & Comitini, 2011; Tofalo et al., 2012) Teleomorphic form Anamorphic form Synonyms* Citeromyces matritensis Candida globosa Debaryomyces hansenii Candida famata Pichia hansenii Dekkera bruxellensis Brettanomyces bruxellensis Hanseniaspora guilliermondii Kloeckera apis Hanseniaspora occidentalis Kloeckera javanica Hanseniaspora osmophila Kloeckera corticis Hanseniaspora uvarum Kloeckera apiculata Hanseniaspora vineae Kloeckera africana Lachancea kluyveri Saccharomyces kluyveri Lachancea thermotolerans Kluyveromyces thermotolerans; Candida dattlia Metschnikowia pulcherrima Candida pulcherrima Torulopsis pulcherrima Meyerozyma guilliermondii Candida guilliermondii Pichia guilliermondii Milleronzyma farinosa Pichia farinosa Pichia fermentans Candida lambica Pichia kluyveri Hansenula kluyveri Pichia membranifaciens Candida valida Pichia occidentalis Candida sorbosa Issatchenkia occidentalis Pichia terricola Issatchenkia terricola Saccharomycodes ludwigii Starmerela bombicola Candida bombicola Torulopsis bombicola Torulaspora delbrueckii Candida colliculosa Saccharomyces rosei Wickerhamomyces anomalus Candida pelliculosa Pichia anomala; Hansenula anomala Zygoascus meyerae Candida hellenica Zygosaccharomyces bailii Saccharomyces bailii Candida zemplinina Possibly Candida stellata in older literature Candida stellata Torulopsis stellata *Names sometimes found in older literature. No teleomorphic form. No anamorphic form. identification can lead to a newly isolated yeast being identified as either teleomorphic or anamorphic if culture-based techniques are being followed. On-going changes in yeast taxonomy (Kreger-van Rij, 1984; Kurtzman & Fell, 1998; Kurtzman et al., 2011b) also results in confusion for nontaxonomists. Especially when citing older literature, it is not always clear what yeasts were actually investigated. Fortunately, DNA-based approaches have largely helped to clarify modern taxonomy. Some of the more commonly encountered teleomorphic yeasts and their anamorphic counterparts in must and wine are given in Table 1. The biology of non-saccharomyces yeast Origin of non-saccharomyces yeasts during wine production Yeasts are found throughout nature typically forming communities within specific habitats (Starmer & Lachance, 2011). Within the winemaking environment (habitat), grape berry surfaces, cellar equipment surfaces and grape can be considered specialized niches where wine-related yeasts form communities (Polsinelli et al., 1996; Goddard & Anfang, 2010; Gayevskiy & Goddard, 2012). These niches differ broadly. The surface of the unripe grape berry presents nutrient limitations that are alleviated as berries ripen and/or are damaged. Due to constant contact with grape must, cellar surfaces can harbour yeasts, but this is highly dependent on the cellar-hygiene practices followed. Although grape must is a rich nutritive environment, low ph, high osmotic pressure and the presence of SO 2 detract from this otherwise ideal yeast niche. Many external factors affect populations both on grapes and in must (Martini et al., 1980, 1996; Rosini et al., 1982; Sharf & Margalith, 1983; Monteil et al., 1987; Gao & Fleet, 1988; Bisson & Kunkee, 1991; Regueiro et al., 1993; Boulton et al., 1996; Cabras et al., 1999; Epifanio et al., 1999; Guerra et al., 1999; Pretorius et al., 1999; Pretorius, 2000; Jawich et al., 2005; Hierro et al., 2006).

5 Non-Saccharomyces wine yeasts 219 During crushing, the non-saccharomyces yeasts on the grapes, on cellar equipment and in the cellar environment (air- and insectborne) are carried over to the must (Peynaud & Domercq, 1959; Bisson & Kunkee, 1991; Boulton et al., 1996; Lonvaud-Funel, 1996; T or ok et al., 1996; Constantı et al., 1997; Mortimer & Polsinelli, 1999; Fleet, 2003). However, cellar surfaces play a smaller role than grapes as a source of non-saccharomyces yeasts, as S. cerevisiae is the predominant yeast inhabiting such surfaces (Peynaud & Domercq, 1959; Rosini, 1984; Lonvaud- Funel, 1996; Pretorius, 2000). Furthermore, hygienic procedures used in most modern cellars should minimize contamination of must by resident cellar flora (Pretorius, 2000). Dominant yeasts in must after crushing should therefore be the same as are found on grapes (Rementeria et al., 2003). Despite all the variables in grape harvest and wine production, the yeast species generally found on grapes and in wines are similar throughout the world (Amerine et al., 1967; Longo et al., 1991; Yanagida et al., 1992; Constantı et al., 1997; Zahavi et al., 2002; Jolly et al., 2006). However, the proportion or population profile of yeasts in various regions shows distinct differences. Importance of non-saccharomyces yeast The contribution by non-saccharomyces yeasts to wine flavour will depend on the concentration of metabolites formed. This in turn is affected by how active the non- Saccharomyces yeasts are. The specific environmental conditions in the must, that is, high osmotic pressure; equimolar mixture of glucose and fructose; the presence of SO 2; nonoptimal growth temperature; increasing alcohol concentrations and anaerobic conditions; and decreasing nutrients all play a role in determining what species can survive and grow (Bisson & Kunkee, 1991; Longo et al., 1991). The clarification of white must (centrifugation, enzyme treatments, cold settling) can also reduce the initial population of yeasts (Fleet, 1990; Lonvaud- Funel, 1996; Pretorius, 2000). The initial belief that all non-saccharomyces yeasts died soon after the commencement of an alcoholic fermentation due to the rising ethanol concentration and added SO 2 has not been sustained by later research (Fleet et al., 1984; Heard & Fleet, 1985; Fleet, 1990, 2003; Querol et al., 1990; Longo et al., 1991; Todd, 1995; Gafner et al., 1996; Granchi et al., 1998; Zohre & Erten, 2002; Jolly et al., 2003c; Combina et al., 2005; Renault et al., 2009). The higher numbers of non-saccharomyces yeasts reported in recent literature might be the result of improved cellar technology and hygiene in modern cellars. This has led to a reduction in SO 2 usage, which presumably results in the survival of a greater number and diversity of non-saccharomyces yeasts. In parallel, the use of modern laboratory techniques has made the detection of non- Saccharomyces yeasts easier. Non-Saccharomyces yeasts found in grape must and during fermentation can be divided into three groups: (1) yeasts that are largely aerobic, for example, Pichia spp., Debaryomyces spp., Rhodotorula spp., Candida spp., and Cryptococcus albidus; (2) apiculate yeasts with low fermentative activity, for example, Hanseniaspora uvarum (Kloeckera apiculata), Hanseniaspora guilliermondii (Kloeckera apis), Hanseniaspora occidentalis (Kloeckera javanica); and (3) yeasts with fermentative metabolism, for example, Kluyveromyces marxianus (Candida kefyr), Torulaspora delbrueckii (Candida colliculosa), Metschnikowia pulcherrima (Candida pulcherrima) and Zygosaccharomyces bailii (Fleet et al., 1984; Querol et al., 1990; Bisson & Kunkee, 1991; Longo et al., 1991; Lonvaud- Funel, 1996; Lorenzini, 1999; Torija et al., 2001; Combina et al., 2005). During fermentation, and more evident in spontaneous fermentations, which lack the initial high-density inoculum of S. cerevisiae, there is a sequential succession of yeasts. Initially, species of Hanseniaspora (Kloeckera), Rhodotorula, Pichia, Candida, Metschnikowia and Cryptococcus are found at low levels in fresh must (Parish & Caroll, 1985; Bisson & Kunkee, 1991; Frezier & Dubourdieu, 1992; Granchi et al., 1998; Fleet, 2003; Combina et al., 2005). Of these, H. uvarum is usually present in the highest numbers, followed by various Candida spp. This is usually more apparent in red must than white, possibly due to the higher ph of the former. However, exceptions do occur and Hanseniaspora can also be absent or present at low levels (Van Zyl & Du Plessis, 1961; Parish & Caroll, 1985; Jolly et al., 2003a; Jolly, 2006). Despite the sustained presence of certain non-saccharomyces yeasts, the majority do disappear during the early stages of a vigorous fermentation (Fleet et al., 1984; Henick-Kling et al., 1998). This might be due to their slow growth and inhibition by the combined effects of SO 2, low ph, high ethanol and oxygen deficiency (Heard & Fleet, 1988; Combina et al., 2005). This is consistent with their oxidative or weak fermentative metabolism. Nutrient limitation and size or dominance of S. cerevisiae inoculum can also have a suppressive effect, sometimes separate from temperature or ethanol concentration (Granchi et al., 1998). It has been reported that T. delbrueckii and Kluyveromyces thermotolerans (now classified as Lachancea thermotolerans) are less tolerant to low oxygen levels and this, rather than ethanol toxicity, affects their growth and leads to their death during fermentation (Holm Hansen et al., 2001; Lachance & Kurtzman, 2011). It was also shown that a cell cell contact mechanism in the presence of high concentrations of

6 220 N.P. Jolly et al. viable S. cerevisiae yeasts played a role in the inhibition of these two non-saccharomyces species (Nissen et al., 2003). The non-saccharomyces spp. that do survive and are present until the end of fermentation may also have a higher tolerance to ethanol which would account for their sustained presence (Pina et al., 2004; Combina et al., 2005). Other species reported throughout fermentation are Saccharomyces acidifaciens (now classified as Z. bailii; Peynaud & Domercq, 1959) and Pichia sp. (Bisson & Kunkee, 1991). Characteristics of the individual species will affect the extent to which they are present. Growth parameters for one species will not necessarily be the same for others, while strains within a species can also show different growth kinetics. The standard practice of di-ammonium hydrogen phosphate (DAP) addition to grape must, higher ph values and increased temperatures can all lead to increased fermentation ability of non- Saccharomyces yeast (Jolly et al., 2003c). Besides affecting wine flavour, the metabolism of non- Saccharomyces yeast can also influence the growth and activity of wine bacteria. In the initial phases of fermentation, non-saccharomyces yeast can deplete essential nutrients that, combined with toxic metabolites formed, can inhibit the growth of lactic acid bacteria essential for the secondary malolactic fermentation in wine (Fornachon, 1968; Costello et al., 2003; Ribereau-Gayon et al., 2006). Conversely, other by-products formed by non-saccharomyces yeast can have a stimulating effect on lactic acid bacteria. Contribution by non-saccharomyces yeast (specific metabolites) Ethanol is the main product of alcoholic fermentation. Currently, consumer and market demand for wines containing lower ethanol has shaped research to develop and evaluate strategies to generate reduced- or low-ethanol wines (Kutyna et al., 2010). Several studies have reported lower ethanol yields when using non-saccharomyces yeast (Ciani & Ferraro, 1996; Ferraro et al., 2000; Soden et al., 2000; Ciani et al., 2006; Comitini et al., 2011; Magyar & Toth, 2011; Di Maio et al., 2012; Sadoudi et al., 2012). Unfortunately, lower ethanol yields are sometimes the result of wines with high residual sugar (> 5gL 1 ; Ciani & Ferraro, 1996; Ciani et al., 2006; Magyar & Toth, 2011). Nevertheless, statistically significant differences in ethanol concentration between wines obtained by mixed fermentation and wines produced by S. cerevisiae monocultures ranged from 0.2% v/v to 0.7% v/v (Ferraro et al., 2000; Soden et al., 2000; Comitini et al., 2011; Izquierdo Canas et al., 2011; Di Maio et al., 2012; Sadoudi et al., 2012; Benito et al., 2013; Gobbi et al., 2013). Another alternative to lower ethanol concentration in wine is to exploit the oxidative metabolism observed in some non-saccharomyces species (Gonzalez et al., 2013). However, only one study has reported the use of aerobic yeast for the production of reducedalcohol wine (Erten & Campbell, 2001). Wines containing 3% v/v ethanol were obtained after fermentation of grape must by Williopsis saturnus and Pichia subpelliculosa under intensive aerobic conditions. These reducedalcohol wines were judged to be of an acceptable quality (Erten & Campbell, 2001). The range of flavour compounds produced by different non-saccharomyces yeasts is well documented (Castor, 1954; Suomalainen & Lehtonen, 1979; Soles et al., 1982; Nyk anen, 1986; Herraiz et al., 1990; Rauhut, 1993; Romano & Suzzi, 1993a; Lema et al., 1996; Lambrechts & Pretorius, 2000; Rojas et al., 2003; Romano et al., 2003; Moreira et al., 2005; Swiegers & Pretorius, 2005; Swiegers et al., 2005). The metabolic products resulting from non- Saccharomyces growth include terpenoids, esters, higher alcohols, glycerol, acetaldehyde, acetic acid and succinic acid (Fleet et al., 1984; Bisson & Kunkee, 1991; Boulton et al., 1996; Lonvaud-Funel, 1996; Heard, 1999; King & Dickson, 2000; Zohre & Erten, 2002; Clemente-Jimenez et al., 2004). Although far less studied, wine colour can also be affected by non-saccharomyces yeast (Benito et al., 2011; Morata et al., 2012). Sequential fermentation of grape juice enriched with anthocyanins using P. guilliermondii and S. cerevisiae has been shown to increase the formation of vinylphenolic pyranoanthocyanins molecules which show greater colour stability (Benito et al., 2011). The role of other non-saccharomyces strains on wine colour remains to be established. The primary flavour of wine is derived from the grapes, while secondary flavours are derived from ester formation by yeasts during wine fermentation (Nyk anen, 1986; Lambrechts & Pretorius, 2000). Several flavour and aroma compounds in grapes are present as glycosylated flavourless precursors (Todd, 1995; Pretorius, 2003). These compounds may be hydrolysed by the enzyme b-glucosidase to form free volatiles that can improve the flavour and aroma of wine, but this enzyme is not encoded by the S. cerevisiae genome (Ubeda-Iranzo et al., 1998; Van Rensburg et al., 2005). In contrast, non- Saccharomyces yeasts belonging to the genera Debaryomyces, Hansenula, Candida, Pichia and Kloeckera possess various degrees of b glucosidase activity and can play a role in releasing volatile compounds from non-volatile precursors (Rosi et al., 1994; Todd, 1995; Spagna et al., 2002; Fernandez-Gonzalez et al., 2003; Rodrıguez et al., 2004; Hernandez-Orte et al., 2008). Cofermentation of Chardonnay grape juice with Debaryomyces pseudopolymorphus and S. cerevisiae resulted in an increased concentration of the terpenols: citronellol, nerol and geraniol in wine

7 Non-Saccharomyces wine yeasts 221 (Cordero Otero et al., 2003). Similarly, cofermentation of Muscat grape juice with Debaryomyces vanriji and S. cerevisiae produced wines with increased concentration of several terpenols (Garcia et al., 2002). Equally, mixed cultures of Sauvignon Blanc grape juice with C. zemplinina/ S. cerevisiae and T. delbrueckii/s. cerevisiae generated wines with high concentrations of terpenols compared to wines fermented with S. cerevisiae (Sadoudi et al., 2012). Another strategy to increase the release of bound volatile compounds is to exogenously add enzyme preparations that can act on nonvolatile precursors. Several studies have characterized and described the effect of b-glucosidase addition on grape juice or wine, focusing particularly in the inhibition of b-glucosidase activity by sugar, alcohol, ph and/or temperature. An intracellular b-glucosidase from Debaryomyces hansenii, which is not inhibited by glucose and ethanol, was used during fermentation of Muscat grape juice resulting in an increase in concentration of monoterpenols in the wine (Yanai & Sato, 1999). Similarly, intracellular b-glucosidases from Hanseniaspora sp. and Pichia anomala have been shown to increase the concentration of volatile compounds after treatment of Traminette grape juice and Traminette wine, respectively (Swangkeaw et al., 2011). A b-glucosidase from Sporidiobolus pararoseus has also been shown to increase the release of volatile terpenols in white and red wine (Baffi et al., 2011), whereas b-glucosidase from Issatchenkia terricola was able to increase the amount of free monoterpenes and norisoprenoids in white Muscat wine (Gonzalez-Pombo et al., 2011). The concentration of volatile terpenes in Arien, Riesling and Muscat wines was also increased following addition of an enzyme extract from Debaryomyces pseudopolymorphus. Consequently, sensory differences were found between treatments (Arevalo-Villena et al., 2007). Over 160 esters have been distinguished in wine (Jackson, 2000). These esters can have a positive effect on wine quality, especially in wine from varieties with neutral flavours that are consumed shortly after production (Lambrechts & Pretorius, 2000; Sumby et al., 2010). Non- Saccharomyces can be divided into two groups, viz. neutral yeasts (producing little or no flavour compounds) and flavour-producing species (both desired and undesired; Van Zyl et al., 1963). Flavour-producing yeasts included P. anomala (Hansenula anomala) and K. apiculata. Candida pulcherrima is also known to be a high producer of esters (Bisson & Kunkee, 1991; Clemente- Jimenez et al., 2004). The net accumulation of esters in wine is determined by the balance between the yeast s ester-synthesizing enzymes and esterases (responsible for cleavage and in some cases, formation of ester bonds; Swiegers & Pretorius, 2005). Although extracellular esterases are known to occur in S. cerevisiae (Ubeda-Iranzo et al., 1998), the situation for non-saccharomyces needs further investigation. Different non-saccharomyces yeasts produce different levels of higher alcohols (n-propanol, isobutanol, isoamyl alcohol, active amyl alcohol; Romano et al., 1992; Lambrechts & Pretorius, 2000). This is important during wine production, as high concentrations of higher alcohols are generally not desired, whereas lower values can add to wine complexity (Romano & Suzzi, 1993b). Non-Saccharomyces yeasts often form lower levels of these alcohols than S. cerevisiae, but there is great strain variability (Romano et al., 1992, 1993; Zironi et al., 1993). Glycerol, the next major yeast metabolite produced during wine fermentation after ethanol, is important in yeast metabolism for regulating redox potential in the cell (Scanes et al., 1998; Prior et al., 2000). Glycerol contributes to smoothness (mouth-feel), sweetness and complexity in wines (Ciani & Maccarelli, 1998), but the grape variety and wine style will determine the extent to which glycerol impacts on these properties. Although the quality of Chardonnay, Sauvignon Blanc and Chenin Blanc is not improved by increased glycerol concentrations (Nieuwoudt et al., 2002), some wines might benefit from increased glycerol levels. Several non-saccharomyces yeasts, particularly L. thermotolerans and C. zemplinina, can consistently produce high glycerol concentrations during wine fermentation (Ciani & Ferraro, 1998; Soden et al., 2000; Comitini et al., 2011). Unfortunately, increased glycerol production is usually linked to increased acetic acid production (Prior et al., 2000), which can be detrimental to wine quality. Spontaneously fermented wines have higher glycerol levels, indicating a possible contribution by non-saccharomyces yeasts (Romano et al., 1997a; Henick-Kling et al., 1998). Nevertheless, the use of some non-saccharomyces yeast in mixed fermentations with S. cerevisiae can generate wines with decreased volatile acidity and acetic acid concentration (Bely et al., 2008; Comitini et al., 2011; Domizio et al., 2011a). Some non-saccharomyces yeasts are able to form succinic acid (Ciani & Maccarelli, 1998; Ferraro et al., 2000). This correlates with high ethanol production and ethanol tolerance. Succinic acid production could positively influence the analytical profile of wines by contributing to the total acidity in wines with insufficient acidity. However, succinic acid has a salt-bitter-acid taste (Amerine et al., 1972) and excessive levels will negatively influence wine quality. Other non-saccharomyces metabolites can act as intermediaries in aroma metabolic pathways. Acetoin is considered a relatively odourless compound in wine with a threshold value of c. 150 mg L 1 (Romano & Suzzi, 1996). However, diacetyl and 2,3-butanediol (potentially

8 222 N.P. Jolly et al. off-flavours in wine) can be derived from acetoin by chemical oxidation and yeast-mediated reduction, respectively. This indicates that acetoin can play a role in offflavour formation in wines. Indeed, high concentrations of acetoin produced by non-saccharomyces yeasts can be utilized by S. cerevisiae in mixed and sequential culture fermentations (Zironi et al., 1993). However, Zironi et al. (1993) could not confirm what metabolites were formed from acetoin by S. cerevisiae. Other compounds that are known to play a role in the sensory quality of wine include volatile fatty acids, carbonyl and sulphur compounds (Nyk anen, 1986; Lambrechts & Pretorius, 2000; Moreira et al., 2005). However, as stated by Guth (1997), there are over 680 documented compounds in wine and a large number of these can, depending on concentration, contribute either positively or negatively to wine aroma and flavour. Volatile thiols greatly contribute to the varietal character of some grape varieties, particularly Sauvignon Blanc (Swiegers et al., 2009). Some non-saccharomyces strains, specifically isolates from C. zemplinina and Pichia kluyveri can produce significant amounts of the volatile thiols 3-mercaptohexan-1-ol (3MH) and 3-mercaptohexan-1-ol acetate (3MHA), respectively, in Sauvignon Blanc wines (Anfang et al., 2009). Similarly, T. delbrueckii, M. pulcherrima and L. thermotolerans have also been described as able to release important quantities of 3MH from its precursor during Sauvignon Blanc fermentation (Zott et al., 2011). Other non-saccharomyces extracellular enzymatic activities, such as proteolytic and pectinolytic (polygalacturonase) enzymes, might also be beneficial to winemaking (Strauss et al., 2001). For example, proteolytic activity of some non-saccharomyces yeast could lead to a reduction in protein levels with accompanying increase in protein stability of the end-product. However, Dizy & Bisson (2000) reported to the contrary that increased yeast proteolytic activity did not lead to a reduction in haze formation in white wine. Species found to produce the greatest number of extracellular enzymes are C. stellata, H. uvarum and M. pulcherrima. Non-Saccharomyces yeasts have also been reported to affect the concentration of polysaccharides in wine (Domizio et al., 2011a, b). Two-strain mixed cultures of S. cerevisiae and Hanseniaspora osmophila, Pichia fermentans, Saccharomycodes ludwigii, Zygosaccharomyces bailii and/or Zygosaccharomyces florentinus were found to produced wines with increased concentration of polysaccharides (Domizio et al., 2011a, b). Polysaccharides can positively influence wine taste and mouth-feel by increasing the perception of wine viscosity and fullness on the palate (Vidal et al., 2004). The early death of some non-saccharomyces yeasts during fermentation can also be a source of specific nutrients for S. cerevisiae enabling it to ferment optimally. These nutrients include cellular constituents such as cell wall polysaccharides (mannoproteins). For this method of nutrient supply to be effective, any killer or other inhibitory effects by the non-saccharomyces yeasts against S. cerevisiae should be known (Herraiz et al., 1990; Panon, 1997; Nguyen & Panon, 1998; Fleet, 2003) so that the subsequent S. cerevisiae fermentation is not adversely affected. The deliberate use of non- Saccharomyces yeast in wine production Various authors have reported on deliberate inoculation of selected non-saccharomyces yeasts for wine production. These included Torulaspora, Candida, Hanseniaspora, Zygosaccharomyces, Schizosaccharomyces, Lachancea (formerly Kluyveromyces; Lachance & Kurtzman, 2011) and Pichia species. All those yeasts are poor fermenters; therefore, S. cerevisiae (either indigenous or inoculated) is always needed to complete wine fermentation. Typically, non-saccharomyces yeasts have been used in sequential fermentation where these yeasts are allowed to grow or ferment between one hour and fifteen days before inoculation with S. cerevisiae (Ciani & Ferraro, 1998; Ferraro et al., 2000; Herraiz et al., 1990; Zironi et al., 1993; Jolly et al., 2003b,c). Many of these trials were conducted on a laboratory-scale utilizing small volumes of grape juice and the results may not necessarily be the same as what could be expected in larger commercial fermentations. Factors such as small amounts of air that can enter small volume fermentations (e.g. during sampling), and rapid sedimentation of yeast cells that can reduce the fermentation rate, can affect the final results (Henschke, 1990). Torulaspora delbrueckii Torulaspora delbrueckii (anamorph: C. colliculosa), was one of the first commercial non-saccharomyces yeast to be released. Torulaspora delbrueckii, formerly classified as Saccharomyces rosei, was previously suggested for vinification of musts low in sugar and acid was used for the commercial production of red and rose wines in Italy (Castelli, 1955). Recently, pure cultures of T. delbrueckii have been shown to produce lower levels of volatile acidity than S. cerevisiae strains (Moreno et al., 1991; Renault et al., 2009). Thus, T. delbrueckii has been useful in the production of wines from high sugar musts derived from botrytized grapes (Bely et al., 2008). Other metabolites produced by T. delbrueckii include succinic acid (Ciani & Maccarelli, 1998) and, for particular strains, linalool, which is derived from monoterpene alcohols and adds to

9 Non-Saccharomyces wine yeasts 223 the varietal aroma of Muscat type wines (King & Dickson, 2000). As T. delbrueckii affects wine composition it also modulates wine flavour and aroma. Following a coinoculated strategy, with T. delbrueckii and S. cerevisiae, Sauvignon Blanc and Chenin Blanc wines were both judged to be better than their respective S. cerevisiae reference wines five and 18 months after production (Jolly et al., 2003b). Similarly, Amarone wines produced by sequential inoculation with T. delbrueckii and S. cerevisiae were judged to have increased aroma intensity, including ripe red fruit aroma, increased sweetness and astringency and decreased intensity for vegetal attributes (Azzolini et al., 2012). In 2003, the first commercial release of T. delbrueckii was as a component of a yeast blend (Vinoflora â Melody.nsac and Vinoflora â Harmony.nsac) with S. cerevisiae and K. thermotolerans (Anonymous, 2004a; CHR Hansen, 2013a, b). Subsequently, the T. delbrueckii component was released on its own (CHR Hansen, 2013a, b). A further two T. delbrueckii strains from other commercial yeast manufacturers are also available (Lallemand, 2012; Laffort, 2013), indicating that some winemakers are eager to experiment with carefully selected and tested non- Saccharomyces yeasts. Candida pulcherrima Metschnikowia pulcherrima (anamorph C. pulcherrima) is another yeast commercially available. This commercial strain produces an extra-cellular a-arabinofuranosidase that impacts on the concentration of varietal aromas such as terpenes and volatile thiols (Lallemand, 2012). This yeast species is also known to produce high concentrations of esters (Bisson & Kunkee, 1991; Rodrıguez et al., 2010; Sadoudi et al., 2012), especially the pear-associated ester, ethyl octanoate (Lambrechts & Pretorius, 2000; Clemente-Jimenez et al., 2004) and can occur in high numbers in grape must (Sch utz & Gafner, 1993; Jolly et al., 2003a). Wines of the grape varieties Sauvignon Blanc, Chenin Blanc and Muscat d Alexandrie obtained by sequential fermentation with C. pulcherrima and S. cerevisiae showed higher quality scores than control wines (obtained by fermentation with S. cerevisiae; Jolly et al., 2003b; Rodrıguez et al., 2010). Similarly, an indigenous C. pulcherrima strain has been reported to increase wine flavour and aroma of Debina wines following sequential inoculation (Parapouli et al., 2010). However, a Chardonnay wine produced by sequential inoculation with C. pulcherrima and S. cerevisiae was judged to be of an inferior quality than the control wine (S. cerevisiae only) implying that specific non-saccharomyces/grape variety combinations lead to increased wine quality scores (Jolly et al., 2003b). It has also been reported that C. pulcherrima can have an antagonistic effect on several yeasts including S. cerevisiae which leads to delays in fermentation (Panon, 1997; Nguyen & Panon, 1998). This phenomenon was due to a killer effect, although not the same as the classical S. cerevisiae killer phenomenon, and was linked to pulcherrimin pigment produced by C. pulcherrima. Differing reports on the interactions between C. pulcherrima and other yeasts may be due to different distinct biotypes within the C. pulcherrima species (Pallmann et al., 2001). Candida zemplinina/candida stellata In 2011, specific strains of Candida stellata were reclassified to C. zemplinina (Kurtzman et al., 2011b). It can therefore be surmised that older literature references to C. stellata, may probably be C. zemplinina and that true C. stellata may not be associated with grapes and wine. In this review the original taxonomic names as published, are used. Candida stellata is known as a high glycerol producer with concentrations reported in wine up to 14 g L 1 (Ciani & Picciotti, 1995; Ciani & Ferraro, 1998; Ciani & Maccarelli, 1998). In contrast, S. cerevisiae has been reported to produce between 4 and 10.4 g L 1 of glycerol (Radler & Sch utz, 1982; Ciani & Maccarelli, 1998; Prior et al., 2000). Glycerol concentrations over 5.2 g L 1 can produce a sweet taste (Noble & Bursick, 1984). Glycerol is also thought to contribute to the mouth-feel and complexity of wine flavour at lower levels (Scanes et al., 1998; Prior et al., 2000). Unlike S. cerevisiae, which favours glucose utilization, C. stellata consumes fructose preferentially to glucose and is therefore considered a fructophilic yeast (Soden et al., 2000; Magyar & Toth, 2011; Di Maio et al., 2012). As S. cerevisiae is a glucophilic yeast, it is not unusual to observe high residual fructose after fermentation of grape musts containing high concentrations of initial sugar. However, after a sequential inoculation strategy of Pinot Grigio grape must containing high sugar concentration (270 g L 1 ), wines obtained by mixed cultures of C. stellata and S. cerevisiae showed no residual sugar, due to the complementary utilization of fructose and glucose by both strains (Ciani & Ferraro, 1998). Hence, fermentation kinetics were faster, shortening fermentation length. Resulting wines showed increased concentrations of glycerol and succinic acid and reduced concentrations of acetic acid and higher alcohols (Ciani & Ferraro, 1998). Similar findings were observed following a sequential inoculation strategy with C. stellata/s. cerevisiae using Trebbiano Toscano grape juice (Ferraro et al., 2000). Sauvignon Blanc wines produced by sequential inoculation with C. zemplinina and S. cerevisiae showed very

10 224 N.P. Jolly et al. different volatile profiles than wines fermented with S. cerevisiae monocultures (Sadoudi et al., 2012). Specifically, C. zemplinina/s. cerevisiae wines showed significantly increased concentrations of terpenols (linalool, citronellol, geraniol, nerolidol and farnesol) and decreased concentrations of aldehydes and acetate esters (Sadoudi et al., 2012). Conversely, coinoculation of Macabeo grape juice with C. zemplinina and S. cerevisiae produced wines with increased concentration of higher alcohols, ethyl esters and short-chain fatty acids (Andorra et al., 2010), indicating that yeast strain and/or grape variety affect the volatile profile of wines fermented with C. zemplinina. Wines exhibiting different compositions of volatile compounds will show a different flavour profile; however, the effect of volatile composition, either positive or negative, on wine flavour is not simple to predict. Chardonnay wines produced by both coinoculation and sequential inoculation with C. stellata and S. cerevisiae showed low aroma intensity for desirable sensory attributes, or exhibited high intensities for undesirable sensory descriptors (Soden et al., 1998, 2000). Compared to wines fermented with S. cerevisiae monoculture, coinoculated wine was scored lower for floral and banana aromas while other sensory descriptors were similar. Wine produced by sequential fermentation showed lower scores for banana, floral and lime aromas, but it was similar in honey, apricot and sauerkraut aromas attributed to the C. stellata yeast. This wine also showed a high ethyl acetate aroma, had the highest concentrations of glycerol and succinic acid, and a lower concentration of ethanol. Wine produced by monoculture of C. stellata was scored particularly high for apricot, honey and sauerkraut aromas. The sauerkraut and ethyl acetate nuances could be considered to detract from wine quality as they are listed under microbiological and oxidized according to wine evaluation terminology (Noble et al., 1987). Similarly, Chardonnay wines produced by sequential inoculation with C. stellata and S. cerevisiae were judged to be of lesser quality than reference wines produced with monocultures of S. cerevisiae, even though reference wines showed lower concentrations of total esters (Jolly et al., 2003b). It seems that the use of C. zemplinina for wine production might involve a role for increasing wine complexity rather than increasing the perception of particular desirable sensory attributes. Hanseniaspora species The apiculate yeasts Hanseniaspora uvarum (anamorph Kloeckera apiculata) are the non-saccharomyces yeasts found in the highest numbers in grape must. Therefore, they should be in the best position to make a contribution to wine quality. Hanseniaspora spp. generally show low fermentative power but are important in the production of wine volatile compounds, and the chemical composition of wines made with Hanseniaspora spp./ S. cerevisiae combinations differ from reference wines produced with S. cerevisiae monoculture (Herraiz et al., 1990; Mateo et al., 1991; Zironi et al., 1993; Gil et al., 1996). The low frequency of Hanseniaspora spp. during fermentation has also been suggested as a reason for the lack of aroma complexity of Folle Blanche wines in the Basque region in Spain (Rementeria et al., 2003). Hanseniaspora vineae (formerly H. osmophila) and H. guilliermondii have been reported to produce increased amounts of 2-phenyl-ethyl acetate during fermentation (Rojas et al., 2003; Viana et al., 2009). This acetate ester is associated with rose, honey, fruity and flowery aroma descriptors (Lambrechts & Pretorius, 2000; Swiegers & Pretorius, 2005; Swiegers et al., 2005), and as part of the fermentation bouquet, it can contribute to the overall flavour of young wines. Cofermentation of Bobal grape must with H. vineae and S. cerevisiae produced wines that not only showed an increased concentration of 2-phenylethyl acetate but also exhibited higher fruity sensory scores than wines produced with S. cerevisiae monoculture (Viana et al., 2009). The amount of 2-phenylethyl acetate produced, however, depended on the proportion of H. vineae/s. cerevisiae (Viana et al., 2009). The same authors reported a higher production of 2-phenylethyl acetate in Tempranillo wines produced by sequential inoculation with H. vineae/s. cerevisiae compared with wines produced by cofermentation (Viana et al., 2011). In addition to 2-phenylethyl acetate, wines produced with H. guilliermondii and S. cerevisiae have shown higher concentrations of hexyl acetate, ethyl acetate and isoamyl acetate than wines produced with S. cerevisiae (Moreira et al., 2008). In these wines, the production of heavy sulphur compounds was also affected by H. guilliermondii. Thus, wines obtained by mixed fermentation showed increased concentrations of 3-(ethylthio)-1-propanol (associated with rancid and sweaty sensory descriptors), 3-mercapto-1-propanol (associated with sweaty and potato ), trans-2-methyltetrahydrothiophen-3-ol ( onion, chive-garlic ) and decreased concentrations of 2-(methylthio)-ethanol + 2-methyltetrahydrothiophen-3-one, the former associated with French bean and cauliflower descriptors, while the latter is described by the attributes metallic and natural gas (Moreira et al., 2008, 2010). Although some of these compounds are associated with unpleasant sensory descriptors, they might have a role increasing wine complexity. Hanseniaspora uvarum has also been used in mixed fermentations with S. cerevisiae for wine production. Macabeo wines fermented with H. uvarum/s. cerevisiae showed increased concentrations of higher alcohols, acetate and