Dissemination and survival of commercial wine yeast in the vineyard: A large-scale, three-years study

Abstract

The use of commercial wine yeast strains as starters has been extensively generalised over the past two decades. In this study, a large-scale sampling plan was devised over a period of three years in six different vineyards to evaluate the dynamics and survival of industrial yeast strains in the vineyard. A total of 198 grape samples were collected at various distances from the wineries, before and after harvest, and yeast strains isolated after spontaneous fermentation were subsequently identified by molecular methods. Among 3780 yeast strains identified, 296 isolates had a genetic profile identical to that of commercial yeast strains. For a large majority (94%), these strains were recovered at very close proximity to the winery (10–200 m). Commercial strains were mostly found in the post-harvest samples, reflecting immediate dissemination. Analysis of population variations from year to year indicated that permanent implantation of commercial strains in the vineyard did not occur, but instead that these strains were subject to natural fluctuations of periodical appearance/disappearance like autochthonous strains. Our data show that dissemination of commercial yeast in the vineyard is restricted to short distances and limited periods of time and is largely favoured by the presence of water run-off.

Wine yeast dissemination, Vineyard ecosystems, Molecular identification methods

1 Introduction

The predominant yeast species used in the production of wine is Saccharomyces cerevisiae, universally known as “wine yeast”. Under selective conditions of grape must fermentation, yeasts efficiently compete with other microorganisms present in musts, such as moulds and lactic- and acetic-acid bacteria. A succession of various yeast species – the apiculate yeasts Hanseniaspora uvarum (=Kloeckera apiculata) and other yeasts of the genera Metschnikowia, Candida or Pichia– is found in the early stages of fermentation [1]. As the concentration of ethanol increases, these species are rapidly outgrown by S. cerevisiae and related species, which invariably dominate the later stages of the process.

Since the beginning of the 1980s, the use of active dried S. cerevisiae yeast starters has been extensively generalised. Today, the majority of wine production is based on the use of active dried yeast, which ensures rapid and reliable fermentation, and reduces the risk of sluggish or stuck fermentation and of microbial contamination. Most commercial wine yeast strains available today have been selected in the vineyard for enological traits such as fermentation performance, ethanol tolerance, absence of off-flavors and production of desirable metabolites. These and other technological developments have contributed to an improvement in the quality of wine, and have enhanced the ability of winemakers to control the fermentation process and achieve specific outcomes.

As a result of modern winemaking practices and diversification of wine products, there is an increasing quest for specialised wine yeast strains. During the last two decades a considerable knowledge of S. cerevisiae genetics and physiology has been generated as well as numerous genetics tools. Recombinant-DNA technologies have been successfully applied to wine yeast, generating specialized wine yeast strains which have been engineered for specific traits, such as improved fermentation performance and process efficiency, wine sensory quality and health benefits for consumers [2–[8].

In view of a possible future use of genetically modified wine yeasts, a sound evaluation of the potential environmental impact of genetically modified wine yeast is absolutely required. In this context, industrial yeasts used as fermentation starters are a good study model to evaluate the competition and the influence of inoculated strains on the fermentations of the following years, especially those performed according to traditional practices which rely on spontaneous fermentations. Commercial yeasts are classically used in winemaking without any special containment and are annually released in large quantities, together with liquid and solid wine-making residues, in the environment around the winery. The behaviour of these yeasts in the ecosystem of the vineyard is totally unknown, as is their potential impact on the natural microbiota. In particular, it is not known if commercial strains are able to survive in nature and to become members of the vineyard microbiota.

Limited data are available that could contribute to an evaluation of the importance of starter yeast dissemination and permanence in the vineyard [9–[11]. Recently, a large-scale biogeographical study in South-African vineyards has been carried out over four years. In five areas, situated in the coastal region vineyards of the Western Cape, 13 samples were collected and commercial yeasts were recovered from three samples [12,[13]. These studies have made it necessary to carry out this type of study on a larger scale, with the aim of increasing the statistic significance of the results obtained.

The present large-scale study, carried out at different geographical localizations in France and Portugal, aims at evaluating the industrial starter yeasts' ability to spread and survive in nature.

2 Materials and methods

2.1 Sampling plan and selection of wineries

Grapes were harvested during three consecutive years (2001–2003) in six vineyards, three of which were located in the south of France and three in the northwest of Portugal, as shown in Fig. 1. In France, the wineries were located in the Languedoc-Roussillon Region, around the Mediterranean city of Montpellier, and the vineyards were situated at a distance of 30 and 80 km from each other. In Portugal, the three wineries were located in the north, centre and south of the Região Demarcada dos Vinhos Verdes, the distance between each being approximately 50 km. In each vineyard, six sampling points were defined according to the predominating wind direction at a distance between 20 and 1000 m from the winery, as shown in Fig. 1.

1

Geographic localization of the vineyards belonging to the Languedoc (A–C) and Vinho Verde (D–F) wine regions, with an indication of the sampling sites in each of the six vineyards. In each site, two samples (pre- and post-harvest campaign) were collected. Factors that may influence the dissemination of the yeasts are indicated in the figure.

In order to evaluate the survival of commercial yeast over years, a first sampling campaign was performed before the winery started wine production with the use of commercial yeast strains (pre-harvest samples). In a second post-harvest sampling campaign, the grapes were collected after the onset of wine production, in order to evaluate the immediate commercial yeast dissemination from the winery. The gap between the pre- and post-harvest campaigns was ten days, during which time waste water was released from the wineries. In the consecutive years, samples were always collected from the same area at a radius of 5 m. With the present experimental design, 72 grape samples were collected each year. The selected wineries have used one or more commercial yeast strains consecutively during at least the last five years. Tables 1 and 2 show the commercial yeasts used in each winery during the period studied (2001–2003) and their geographic origin, respectively.

1

Commercial yeast strains used in each winery during 2001–2003

Year	Winery A	Winery B	Winery C	Winery D	Winery E	Winery F
2001	K1M ICV-INRAa	K1M ICV-INRA	K1M ICV-INRA	ZymafloreVL1	Zymaflore VL1	Zymaflore VL1
	ICV D254	ICV D254	Zymaflore VL3	ZymafloreVL3	Lalvin EC 1118
	Enolevure K34	ICV D80	Maurivin PDM	Zymaflore F10
	Lalvin QA23	Uvaline BL	ICV D254	Zymaflore F15
	ICV D47	Lalvin BM45	ICV D47	Uvaferm L2056
		Maurivin AWRI2	Uvaline arôme	Lalvin CY 3079
			Vitilevure-Chardonnay	Uvaferm ALB
			Anchor VIN 13	Uvaferm 2287
2002	K1M ICV-INRA	K1M ICV-INRA	K1M ICV-INRA	ZymafloreVL1	Zymaflore VL1	Zymaflore VL1
	ICV D254	ICV D80	Maurivin PDM	ZymafloreVL3	Lalvin EC 1118	Lalvin EC 1118
	Lalvin QA23	Uvaline BL	ICV D47	Zymaflore F10	Levuline BRG
		Lalvin BM45	Anchor VIN 13	Zymaflore F15	Fermichamp
		Maurivin AWRI2	Zymaflore VL3	Uvaferm ALB
		Uvaline CVR	Anchor NT 116	Uvaferm 228
			Vitilevure-Sauvignon	Uvaferm CS2
2003	K1M ICV-INRA	K1M ICV-INRA	K1M ICV-INRA	ZymafloreVL1	Zymaflore VL1	Zymaflore VL1
	ICV D254	Uvaline BL	Zymaflore VL3	ZymafloreVL3	Fermafine	Lalvin CY 3079
	Enolevure K34	Lalvin BM45	Maurivin PDM	Zymaflore F10	Fermafruit
		Anchor NT 45	Vitilevure-Chardonnay	Zymaflore F15	IOC 18-2007
		Anchor NT 50	Vitilevure-Sauvignon	Zymaflore VL2	Lalvin CY 3079
	ICV D80		Uvaferm ALB
		Uvaline CVR
		Enolevure K34
		Maurivin PDM

All strains belong to S. cerevisiae.

^aThe strains shown in bold were used for at least the last five years prior to the study.

2

Geographic origin of the commercial yeast strains used in the wineries studied

Strains	Origin
Anchor NT 45	South Africa
Anchor NT 50	South Africa
Anchor NT 116	South Africa
Anchor VIN 13	Stellenbosch, South Africa
Enolevure K34	Valencia, Spain
Fermafine	Not known
Fermafruit	Not known
Fermichamp	Alsace, France
ICV D 47	Rhône, France
ICV D 80	Rhône, France
ICV D 254	Languedoc, France
IOC 18-2007	Not known
K1M ICV-INRA	Languedoc, France
Lalvin BM 45	Sangiovese, Italy
Lalvin EC1118	Champagne, France
Lalvin QA23	Portugal
Lalvin Cy 3079	Bourgogne, France
Levuline BRG	Not known
Maurivin AWR12	Bordelais, France
Maurivin PDM	Champagne, France
Uvaferm 228	France
Uvaferm ALB	Not known
Uvaferm CS2	Alsace, France
Uvaferm L 2056	Rhône, France
Uvaline arôme	Loire, France
Uvaline BL	Champagne, France
Uvaline CVR	Not known
Vitilevure Chardonnay	Languedoc, France
Vitilevure Sauvignon	Sauvignon, France
Zymaflore F10	Bordelais, France
Zymaflore F15	Gironde, France
Zymaflore VL1	Gironde, France
Zymaflore VL2	Burgundy, France
Zymaflore VL3	Gironde, France

2.2 Sample collection and yeast isolation

From each sampling point, approximately 2 kg of grapes were collected aseptically and directly placed into sterile plastic bags, which were transported to the laboratory in cool bags. At the laboratory, grapes were crushed by hand in the plastic bags, which were opened and 180 ml of juice was poured into sterile 250-ml fermentors. The fermentors were placed in a temperature-controlled room at 20 °C with mechanical agitation. Daily weight determinations allowed the monitoring of the fermentation progress. The yeast community was analysed when the must weight was reduced by 70 g l⁻¹, corresponding to the consumption of about 2/3 of the sugar content. Must samples were diluted and spread on plates with YEPD medium (yeast extract 1% w/v, peptone 1% w/v, glucose 2% w/v, agar 2% w/v), and after 2 days of incubation 30 randomly selected colonies were collected from each spontaneous fermentation.

2.3 Selection of Saccharomyces

To rapidly discriminate between Saccharomyces and non-Saccharomyces yeast, every isolate was evaluated according to its ability to grow on l-lysine [14]. All isolates that were not able to grow on the YNB medium with l-lysine as the sole nitrogen source, but grew on the control medium YNB with ammonium sulphate were considered as Saccharomyces and selected for further molecular identification.

2.4 Molecular identification methods

DNA was extracted from yeast cells cultivated in 1 ml YEPD medium (36 h, 28 °C, 160 rpm) as described previously [15], with a modified cell lysis procedure, using 25 U of Zymolase (Sigma, St. Louis, MO). Cell lysis was dependent on the strain and lasted between 20 min and 1 h (37 °C).

Mitochondrial-DNA restriction profiles were established as described previously [16]. Digestions (HinfI) were performed overnight at 37 °C in a final volume of 20 μl [17].

Microsatellite analysis was performed using six loci (ScAAT1–ScAAT6), previously described by Pérez et al. [18], that were amplified (iCycler thermal cycler Bio-Rad, Marnes-La Coquette, France) in two multiplex reactions. The samples were denatured and separated by capillary electrophoresis in an ABI Prism 310 DNA sequencer (Applied Biosystems, Foster City, CA) and analysed using Genescan (Metairie, LA) software. The complete method was described by Schuller et al. [17].

Chromosomal profiles were established by pulsed-field gel electrophoresis (PFGE) using yeast chromosomal DNA prepared in plugs and the TAFE (transverse alternating field electrophoresis) system (Geneline, Beckman, Villepinte, France) as previously described [19]. The gels were run for 6 h at 250 V with a 35-s pulse time, followed by 20 h at 275 V with a 55-s pulse time, at 14 °C.

3 Results and discussion

3.1 Sampling sites and isolation of Saccharomyces

A large sampling plan was followed: a total of 198 samples were collected during three consecutive campaigns (2001–2003), 108 of which were taken in France and 90 in Portugal. It is to be noted, as can be observed in Fig. 1, that due to geographical constraints the samples in Portugal were collected much closer to the winery than those in France. In the French wineries (A–C), the sample sites were located at a distance of between 100 and 1000 m from the winery, whereas in the Portuguese wineries (D–F) half of the sampling sites were located at a distance of less than 70 m from the winery and none was located further than 400 m.

Table 3 shows the global data in each country, broken down into years. Of the 198 samples, 126 musts (64%) produced spontaneous fermentations, 20% and 44% by musts from pre-harvest and post-harvest campaigns, respectively. The percentages of spontaneous fermentations in both countries were similar, 66% in France and 60% in Portugal. A total of 3780 colonies were isolated from these fermentations (2160 and 1620 in France and Portugal, respectively).

3

Distribution of global data by country and year

	2001		2002		2003		Total
	France	Portugal	France	Portugal	France	Portugal
Samples	36	36	36	18	36	36	198
Spontaneous fermentations	24	19	33	12	15	23	126
Isolates	720	570	990	360	450	690	3780
Saccharomyces strains	406	570	120	360	209	690	2355

Discrimination between Saccharomyces and non-Saccharomyces strains isolated in Languedoc was performed using a selective medium with l-lysine as the sole nitrogen source [14]. By this method only two species of the genus Saccharomyces (i.e., S. kluyveri and S. unisporus), which do not occur in enological environments, are capable of growing with l-lysine. From this we concluded that yeasts, isolated after fermentation, that can utilize l-lysine, do not belong to the genus Saccharomyces. To confirm this hypothesis, isolates from the fastest fermentations that grew in l-lysine medium were identified by PCR-RFLP analysis of the rDNA ITS region [20]. The results confirmed that they were non-Saccharomyces strains, belonging mainly to the genus Kloeckera (data not shown). All isolates not able to grow on the l-lysine medium were therefore selected for molecular identification. In Portugal, all isolates were assigned to different groups according to their mtDNA RFLP pattern. One representative strain from each group was randomly withdrawn, and its ability to grow on l-lysine was tested. Based on these methods, 2355 Saccharomyces strains were selected from the 3780 isolates collected during the three years.

3.2 Geographic distribution of recovered commercial yeast strains

The global composition of the yeast population isolated after fermentation from the six wineries over the three years studied, in pre- and post-harvest campaigns, is shown in Fig. 2. Table 4 shows the distribution and frequency of commercial yeasts in each vineyard.

2

Global composition of the yeast communities isolated from each site at the six wineries during the pre- and post-harvest sampling campaigns over the three years. The motifs show the presence of commercial yeasts, light grey indicates other Saccharomyces strains and dark grey the non-Saccharomyces strains. Nf: no fermentation; Nc: not collected.

4

Commercial yeast strains recovered in each vineyard over the three years studied

Vineyards	A	B	C	D	E	F	Total
Spontaneous fermentations	19	24	29	16	23	15	126
Spontaneous fermentations with ≥ 1 commercial yeast strains	0	2	1	11	9	2	25
Isolates	570	720	870	480	690	450	3780
Commercial yeast strains	0	15a	1	206	54 + 18a	2	296
% Commercial yeast/number of isolates	0	2	0.1	43	10	0.5	7.8

^aCommercial yeasts initially isolated in the same region.

Identification of Saccharomyces strains was performed by different molecular-typing methods, depending on the specific resources of each laboratory. An example of genetic profiles, both of natural isolates and commercial yeast strains from France and Portugal, is shown in Fig. 3. Chromosomal-pattern analysis of 735 Saccharomyces isolates from France (wineries A–C) was performed, and compared with that of the 19 commercial yeasts used in the three wineries. In Portugal, all 1620 isolates were analysed by mtDNA RFLP (HinfI), and their patterns compared to those of a strain collection including all strains used by the three wineries. At least one representative isolate of each group of strains showing identical mtDNA RFLP patterns identical to those of commercial strains was further confirmed by microsatellite analysis. In order to evaluate the discriminatory power of these three methods, we had previously performed a survey of the genetic polymorphisms generated by distinct methods on a total of 23 commercial yeast strains used in the wineries of the two countries [17]. The results had shown that the discriminatory power of microsatellite typing using these six different loci, and that of mtDNA RFLP patterns generated by the enzyme HinfI, was the same and similar to that of karyotype analysis. Among the 23 commercial yeast strains analysed, 21 different patterns were obtained using the first two methods and 22 using the last. Due to the verified similarity of the discriminatory power of these methods, any of them can be used for our study, with comparable results.

3

Examples of molecular fingerprinting of commercial yeasts and natural isolates (indicated by numbers). (a) Chromosomal profiles of commercial yeasts and natural isolates from France. Profiles 4–6 are identical to ICV-D254. (b) mtDNA RFLP profiles of commercial yeasts and natural isolates from Portugal. Profiles 245, 13, 5, 105, 157, 12 and 32 are identical to Zymaflore VL3, F10, F15, Lalvin QA23, ICV-D254, Zymaflore VL1 and Uvaferm BDX, respectively.

The analysis of genetic profiles of 2355 out of 3780 Saccharomyces isolates resulted in the identification of 296 commercial yeasts, representing 7.8% of the fermentative yeast community (Table 4), the majority of which (5.8%) were recovered in post-harvest campaigns (Fig. 2). It should be noted that in this study fermentation is used as an enrichment tool for Saccharomyces strains. Therefore, the present results do not allow conclusions about the number of strains occurring on the surface of the grape, which is in fact very low, but reflect only those strains that could possibly have some enological use. Instead, the number of fermentations with at least one commercial yeast strain gives a better picture of the situation as it occurs in vineyards; commercial yeast strains were recovered from 12% of the samples (Tables 3 and 4).

These global data reflect very different situations. In the vineyards where the sampling sites were placed at a greater distance from the winery, i.e., vineyard F in Portugal and the three French vineyards (A–C), the occurrence of commercial yeasts was very low, representing between 0% and 2% of the fermentative community, and these strains were isolated from only five samples (Table 4). In France, the genetic profile of 16 clones out of 735 Saccharomyces isolates (2%) was identical to that of commercial yeasts (Fig. 2). These strains corresponded to 0.8% of the yeast strains isolated after fermentation. With only one exception, these strains (fifteen isolates) had a profile identical to that of the autochthonous strain ICV D254 and were found at the same site (winery B), in pre-harvest samples taken in 2001 (Fig. 2). This fact could indicate previous dissemination, but this cannot be confirmed, since the commercial yeast strain ICV D254 was initially isolated from the same region in the south of France (Table 2) where the study was carried out. No commercial yeasts were found from winery A and one colony, isolated in 2003 from winery C (site 3), had the same profile as K1M ICV-INRA, used in the three French wineries for the last 5–15 years. It is noteworthy that this yeast, which has been used extensively for a considerable length of time, has never been found in the vineyard, except in this case. The same situation occurs in the Portuguese winery F, where only two isolates were found with the same profile as the extensively used commercial yeast, Zymaflore VL1, in use for five years. Since strain ICV D254 was initially isolated in the region from which it was recovered, dissemination in these four vineyards was proved only by the presence of three isolates (0.1% of the fermentative flora), one of K1M ICV-INRA and two of Zymaflore VL1. Their presence might be considered to be due to immediate dissemination, probably mediated by insects or another occasional dissemination vector. It is, in any case, evident that the presence of the most-widely used commercial yeasts for the last 5–10 years in French (i.e., K1M ICV-INRA) and in Portuguese wineries (i.e., Zymaflora VL1) was incidental and that these yeasts hardly ever dominated the microflora of any of these four vineyards. These results, in accordance with those obtained previously in South-African vineyards [12,[13], indicate a very low level of dissemination/implantation of commercial yeasts in the vineyard ecosystem.

The results were very different in the Portuguese wineries D and E, for which a high number of commercial strains was isolated from 20 spontaneous fermentations, representing 43% and 10% of the fermentative yeast community, respectively. Indeed, the large majority (94%) of the commercial strains isolated within the six vineyards were recovered from these two vineyards only, and 70% solely from vineyard D. It can be observed from Figs. 1 and 2 that the majority of the commercial strains in these two vineyards were isolated from sites closest to the winery, namely sites 4–6 in vineyard D and sites 1, 2 and 6 in vineyard E. The major difference between these two vineyards and the four others is that the sample sites in the first two were located in close proximity to the winery (Fig. 1). In addition, the presence of water run-off at these sites indicates that dissemination is probably largely favoured by liquid effluents. In vineyard D, due to the ground inclination, water run-off flowing from the winery to the vine may contribute to the frequent occurrence of commercial strains at these sites. It is also noteworthy that site 1 of winery E, where the highest number of VL1 strains was recovered, is located close to a rill that transports run-off water from the winery, emphasizing the importance of water as a vehicle for yeast strain dissemination. Furthermore, the dumping site of macerated grape skins also is adjacent to site 1, constituting a fermenting sugary substrate harbouring large amounts of yeasts that can be distributed throughout the vineyard.

An overview of the dissemination of commercial strains in relation to their distance from the winery is shown in Fig. 4. Of the commercial strains, 94% were found in a radius of around 10–200 m from the winery and a large majority (78%) was recovered at sites very close (10–50 m) to the wineries (vineyards D and E). A major proportion (73%) was collected in post-harvest campaigns, indicating immediate dissemination. With the exception of the autochthonous ICV D254 strain collected in French winery B, commercial yeasts in pre-harvest campaigns were only collected at sites very close to winery D (10–50 m) and the strain found in the greatest quantity (87%) was Zymaflore F15. In the post-harvest samples, strain VL1 represented 49% of the commercial strains recovered. This strain was derived from sites close to the area where macerated grape skin was deposited or water run-off occurred, and never further than 10–20 m from the winery (Figs. 1 and 2). A lower percentage of the other predominant strains: Zymaflore F10 and F15, the formerly used minority strains Uvaferm BDX and ICV D254, and the autochthonous strain Lalvin QA23, were found at sites closer to the winery (10–50 m). Zymaflore F15, F10 and ICV D254 were also found at about 100 m from the winery. The occurrence of several isolates found at 200 m (site 1, winery D) can be attributed to the presence of a small building for storage of harvest transport equipment. Two samples taken in France at a distance of 400 and 1000 m contained yeasts with a karyotype identical to that of the indigenous strain ICV D254. In very rare cases, dissemination to sites located further away from the winery (i.e., two isolates at 400 m and one at 1000 m from a total of 3780 strains) was revealed and may be attributable to other factors, such as insects or wind.

4

Overall distribution during three years of commercial yeast strains according to the distance from the wineries in pre-harvest (a) and in post-harvest (b) campaigns.

3.3 Dissemination of commercial strains as a function of their utilisation

As shown in Table 5, the 296 strains collected had a genetic profile identical to that of only nine commercial yeast strains from a total of 34 strains used in the six wineries. In most instances, the strains with a profile similar to that of a commercial strain were recovered from a vineyard in which the same commercial yeast was used, except for ICV D254 and Uvaferm BDX, which were collected in vineyard D and not used there during the study. However, these strains had been used previously (1998–2000) in the same vineyard. The other exception was strain Lalvin QA23, which was used in vineyard A and collected only from vineyard E. Since this strain had initially been isolated in this Portuguese region, the most likely explanation is that the strain isolated in vineyard E is not the result of dissemination, but that it is a member of the indigenous yeast community.

5

Origin of 296 strains with genetic patterns identical to commercial yeast strains used in the wineries

Commercial wine strains	Number of strains with identical genetic pattern	Wineries were these strains were used	Vineyard were these strains were collected	Utilization level during the 3 years
Zymaflore VL1	99	D, E, F	D, E, F	+++
Zymaflore F15	74	D	D	+
ICV D254	68	A, B, C, Da	Bb, D	++
Zymaflore F10	24	D	D	++
Lalvin QA23	19	A	Eb	+
Uvaferm BDX	9	Da	D	–
K1M ICV-INRA	1	A, B, C	C	+++
Zymaflore VL3	1	C, D	D	+
Lalvin CY 3079	1	D, E, F	E	+

^aUsed before the study

^bIsolated in the same region.

The industrial yeasts most commonly used in the wineries were usually collected in great abundance in the vineyard. However, this was not always the case, because the strain K1M ICV-INRA was the most widely used in the three French wineries and only one isolate out of 2160 isolates collected in France had a genotypic pattern identical to that of this strain. In Portuguese wineries, Zymaflore VL1 was predominantly and continuously used for more than 10 years, followed by Zymaflore F10 and VL3 (Table 1). The strains VL1 and F10 were frequently recovered, but this could be due to the fact that the sites where they had been collected were located close to the winery (i.e., vineyards D or E, Fig. 2). The strain Zymaflore F15, although frequently collected in the same vineyard D, was used to a lesser extent. Over the period of the study, Zymaflore VL3 was also widely used in Portuguese wineries, and the genetic profile of only one isolate was identical to this commercial yeast. As a whole, these data indicate that there is no strict correlation between the utilisation level and the frequency of dissemination.

3.4 Evolution of the fermentative yeast community

The evolution of the total yeast community isolated after fermentation in the different wineries of France and Portugal during the three years studied is shown in Fig. 5. From a total of 296 commercial yeasts recovered during this period in the six vineyards, 76% were found in 2001, in pre- and post-harvest samples collected in vineyard D and in post-harvest samples collected in vineyard E. In the following two years commercial yeasts were detected only in certain post-harvest but not in pre-harvest samples. As can be observed in Fig. 5, five different commercial yeast strains were found in the pre-harvest campaign of winery D in 2001, namely the predominantly used strains VL1, F10 and F15, and in much smaller quantities the strains Uvaferm BDX and ICV D254, used from 1998 to 2000, thus showing their survival in the vineyard from one year to another. However, given that the two latter strains appeared in 2001 only, their permanence was limited.

5

Evolution of the total fermentative yeast communities from each of the wineries (A–F) during the three years in pre-harvest and in post-harvest campaigns.

The commercial yeasts collected at each site were nevertheless different. The highest number of Zymaflore VL1 isolates was obtained from grapes collected after harvest at site 4, whereas in samples collected before harvest, VL1 and F10 rarely occurred. This contrasts to the abundance of the strain F15 in the pre-harvest campaign of 2001 at sites 4 and 6 (Fig. 2). These data suggest a better ability of strain Zymaflore F15 to remain in the vineyard, although no isolate of strain F15 was found in 2002 and only one in 2003 (Figs. 2 and 4). In this winery no samples were collected in the post-harvest campaign of 2002, and a lower quantity of commercial yeasts was found in 2003. In addition, the presence of one isolate of Zymaflore VL3, not present in 2001, was detected. In the post-harvest campaign of 2001, two commercial yeast strains, Zymaflore VL1 and the autochthonous yeast Lalvin QA23, were isolated in winery E. This last strain was the only commercial yeast found in the same winery in 2002, but it was not present in 2003. On the other hand, Zymaflore VL1 was not found in this winery in 2002, but was present in 2003, although in lower proportions. The situation observed in Portuguese winery F, as described previously, was similar to that in French wineries. No commercial yeasts were detected in 2001 and 2002, and only two isolates of Zymaflora VL1 were found in 2003. In winery B, the autochthonous strain ICV D254 was found in the pre-harvest campaign in 2001, but did not occur in the following years. Only one isolate of K1M ICV-INRA was found in 2003 in winery C. As a whole, the evolution of the fermentative yeast communities over the three years studied showed that the same strains were not found at the same sites from one year to another. This indicates that, if some of these strains are able to remain in the ecosystem, as suggested by the presence of commercial yeasts in pre-harvest samples taken in 2001 in Portugal, they are not capable of dominating the natural yeast community of the vineyard.

In conclusion, this systematic study has provided new insights in relation to the impact of commercial yeasts on the communities of fermentative yeasts that inhabit surrounding vineyards. The methodology used, based on analysis of the yeast community after spontaneous fermentation, permitted the isolation of a very large number of Saccharomyces wine yeasts, which are poorly found on the grapes. A significant number of non-Saccharomyces strains was also found in the spontaneous fermentations from the French samples, but not from the Portuguese grape musts (Table 3). Climatic factors and differences in phytosanitary treatment may be the cause of these discrepancies. In future studies, the occurrence of non-Saccharomyces strains during fermentation could be reduced by adding SO₂ to the grape musts prior to fermentation. It is important to mention that among the 30 colonies analyzed per fermentation, the number of different genetic profiles varied from 1 to 21, with an average of about five different Saccharomyces biotypes per sample ([22] and unpublished data). This reflects great differences in the samples regarding the presence of Saccharomyces. These data also demonstrate that the number of colonies analysed per sample was high enough to show the initial biodiversity. For future studies, increasing the initial amount of grapes collected may increase the number of spontaneous fermentations and therefore of S. cerevisiae strains isolated.

Data obtained in the present study show that dissemination of commercial yeasts in the vineyard is restricted to short distances and limited periods of time. More than 90% of commercial yeasts were found at a radius between 10 and 200 m from the winery and did not become implanted in the ecosystem in a systematic way. Dispersal of commercial strains seems to be mainly mediated by water run-off and may also derive from macerated grape skin at dumping sites. This situation was observed during the habitual functioning of a winery, where commercial strains are used without any containment. Avoiding grape skin deposition and canalisation of water run-off are low-cost measures, which can significantly reduce the population size of commercial yeast strains around the winery.

Given that they are used in large quantities, commercial strains tend to out-compete autochthonous strains inside the winery [21]. In contrast, they do not seem to settle in the vineyard. Rather, they show natural fluctuations of periodical appearance and disappearance, just like autochthonous strains do. Moreover, vine-associated autochthonous Saccharomyces biodiversity is not affected by long-term use of commercial yeasts [22]. Considering commercial yeast strains as an appropriate model system for genetically modified yeast strains, our data can contribute to the in-depth environmental-risk assessment concerning the use of such strains in the wine industry.

Acknowledgements

This study was supported by the project ENOSAFE (No. 762, Programa AGRO, medida 8.1) and the grant No. 657 C2 from the cooperation agreement between the Portuguese Institute for International Scientific and Technological Cooperation (ICCTI) and the French Embassy in Lisbon and the Marie Curie Fellowship of the European Community programme of Quality of Life under Contract QLK4-CT-2001-51873. The authors wish to thank A. Ribeiro, H. Alves, L. Quintas, C. Camarasa and C. Machado for their support during grape collection and sample processing. M.S. Graça is gratefully acknowledged for the operation of the DNA sequencer. We would like to express our sincere gratitude to D. Delteil from ICV-Montpellier for his help in the selection of sample sites in France. We also appreciate the kind assistance of the enologists of the Portuguese wineries (R. Cunha, A. Mendes, E. Rodrigues and J. Domingues) as well as the enologists (B. Agay, E. Feneuil and E. Bru) and directors the (J.L. Refle, J. Combette and R. Bruno) of the French wineries, facilitating sampling campaigns in the vineyards.

Author notes

1

Contributed equally to this work.