OUP user menu

The effect of fungicides on yeast communities associated with grape berries

Neža Čadež, Jure Zupan, Peter Raspor
DOI: http://dx.doi.org/10.1111/j.1567-1364.2010.00635.x 619-630 First published online: 1 August 2010

Abstract

The influence of three commonly used fungicides (iprodione, pyrimethanil and fludioxonil plus cyprodinil) on the density and diversity of yeast populations present on grape berries was evaluated. At the time of harvest, the fungicide residues on grapes were below the maximum permitted levels. In general, larger yeast counts were found on the treated grapes than on the control samples. Among 23 species identified, Cryptococcus magnus, Rhodotorula glutinis and Sporidiobolus pararoseus dominated on sound grape berries. The results showed that the tested fungicides had only a minor impact on the composition of grape berry communities in comparison with the effect of weather conditions and the mode of grape berry sampling. Halo assays using filter discs loaded with fungicides were used as in vitro tests of the sensitivity of grape berry isolates. The fungicide containing pyrimethanil suppressed the growth of all basidiomycetous yeast species, while the sporadically occurring fermentative yeasts were unaffected. Fungicides with fludioxonil plus cyprodinil and iprodione as active substances showed specificity for certain species. Our results suggest that after the safety interval, the presence of fungicides has a minor impact on the composition of grape berry communities, although at the time of fungicide applications, the yeast species composition changes.

Keywords
  • fungicide
  • grape berry
  • wine yeast
  • basidiomycetes
  • polyphasic identification
  • halo assay

Introduction

Grapes are a primary source of indigenous yeast microbial communities that plays an important role in alcoholic fermentation. Although Saccharomyces cerevisiae is the main organism responsible for the conversion of grape juice into wine, indigenous yeasts were found to be important contributors to the chemosensory properties of the wine. They can either contribute to a more complex and better aroma or they can cause wine spoilage (Heard & Fleet, 1985; Granchi, 2002). There are two types of detrimental effects of natural yeast flora on wine fermentation: by the production of off-flavours or by the prevention of the normal predominance of S. cerevisiae (Bisson, 1999). For these reasons, the ecology of natural yeast flora is an important factor influencing wine quality.

Grape berry surfaces provide a physical environment suitable for the growth of microbial communities that depend on the grape vine for nutrients, water and protection (Schreiber, 2004). Yeast populations are spatially distributed over the grape berries (Belin, 1972) and grape bunches (Rosini, 1982), and are dynamic during the course of grape berry development (Rosini, 1982; Prakitchaiwattana, 2004). The community dynamics is also influenced by external factors such as geographical location, climatic conditions, grape cultivar, vine canopy and the use of agrichemicals (reviewed by Fleet, 2002).

The application of chemical fungicides to prevent the growth of the fungus Botrytis cinerea (teleomorph: Botryotinia fuckeliana), a causal agent of grey mould, is a routine viticultural practice in areas where weather conditions favour the disease (Rosslenbroich & Stuebler, 2000). Although the application of fungicides is an effective control against B. cinerea, their mode of action might be nonspecific and might act on organisms other than the target fungus (Calhelha, 2006). Based on these predictions, we examined the influence of three commonly used fungicides (iprodione, pyrimethanil and fludioxonil+cyprodinil) on the density and diversity of indigenous yeast populations present on grape berries. Because we cannot avoid the effect of weather conditions in open field experiments, we evaluated them along with the influence of the grape berry sampling methodology. Furthermore, the sensitivities of grape yeast species to chemical fungicides were determined under in vitro conditions.

Materials and methods

Grape vine treatments

The experiment was performed in the years 2002 and 2003 in a vineyard with the Rebula white grapevine variety (Vitis vinifera L.) in the Goriška Brda wine-growing region, Slovenia (46°00′49″N, 13°30′12″E, altitude 117 m). For this region, the climatological data for a 30-year period are as follows: an annual mean air temperature of 11.6 °C and a total annual precipitation of 1456 mm. The experimental field comprised four rows of vines within which a randomized trial of four blocks with eight vines per block was used. Between the blocks, the gaps consisting of approximately 16 untreated vines were present. The fungicides were applied in four blocks of the vines as follows: (1) iprodione/Kidan® (BASF AG, Ludwigshafen, Germany; 0.765 kg a.i. ha−1), (2) pyrimethanil/Mythos® (BASF AG; 0.750 kg a.i. ha−1), (3) cyprodinil in combination with fludioxonil/Switch® (Syngenta Crop Protection AG, Basel, Switzerland; 0.200 and 0.300 kg a.i. ha−1, respectively). In a control treatment (4), the grapes were not sprayed against grey mould. The fungicides were applied at two growth stages of grapevine development: at closure of the berries and at the beginning of grape ripening, considering the safety intervals for the fungicides used in the experiment (Kidan 28 days; Mythos and Switch 21 days). Regular canopy management was carried out before each application. The vines were sprayed using a knapsack sprayer (SOLO Port Type 423, VA). The quantity of water used was 3 L per treatment.

Determination of fungicide residues

GC/MS was applied to determine the fungicide residues on grapes as described by Baša-Česnik (2008). Extractions of fungicide residues were made from 20 g of the homogenized grape samples. The analytical set-up consisted of an HP-6890 (Hewlett-Packard, CA) GC combined with a GCMS 5973 (Hewlett-Packard) and an HP-5MS column (30 m × 0.25 mm i.d.; Agilent Technologies J&W, CA).

Sampling procedures

At harvest (17 September 2002 and 11 September 2003), 160 healthy grape berries with their pedicles (five grape berries per vine, 32 vines of one block trial) were aseptically collected in Stomacher bags in triplicate per treatment (three treatments and an untreated control). The grape berries were cut from different parts of several grape bunches located in various parts of the vines. All 12 samples of grape barries were transported on ice to the laboratory and processed approximately 3 h after harvest. The samples were crushed and homogenized manually. In the year 2002, we alternatively sampled whole grape bunches (5 kg per treatment in triplicate) that were crushed by a sterilized destemmer/crusher. The grape juice was sampled directly from the mash.

Tenfold dilutions of grape juice were transferred in triplicate onto YM agar plates (0.3% malt extract, 0.3% yeast extract, 0.5% Bacto peptone, 1% glucose and 2% agar). The plates were incubated at 26 °C for 5 days, after which colonies were counted and grouped according to their morphology (colony colour and shape of vegetative cells). Microbial density was expressed as CFU cm−2 of grape berry surface. Representative colonies from each countable plate were purified and preserved in 10% glycerol at −80 °C for identification.

Identification of isolates

Genomic DNA was isolated from the cultures according to the method of Möller (1992). The internal transcribed spacers (ITS) (ITS1 and ITS2) and 5.8S rRNA gene regions were amplified using ITS1 and ITS4 primers (White, 1990) as described previously (Cadez, 2002) and were digested with restriction enzymes HaeIII, CfoI, HinfI and MspI (Roche Diagnostics, Mannheim, Germany), or alternatively, by MboI and DraI (Promega, Madison, WI) according to the manufacturer's instructions. The digests were separated on 2.5% agarose gels. The molecular sizes of the ITS digests were determined by comparison with a DNA molecular marker using quantity one 4.0.3 (Bio-Rad, Hercules, CA). The isolates sharing identical restriction patterns were grouped and only representatives were characterized by traditional physiological testing in liquid media as recommended by Yarrow (1998). Phenetic similarities to described yeast species were examined using the biolomics computer program (Robert & Szoke, 2006) from the CBS database (http://www.cbs.knaw.nl). The identity of the species was confirmed by sequencing of the D1/D2 domain of the large subunit rRNA gene and ITS1–5.8S rRNA gene–ITS2 regions, as described previously by Cadez (2003). Further, for the type strains of the identified species, their ITS sequences were obtained from the GenBank/EMBL/DDBJ database and in silico restriction analyses were performed using a pdraw32 computer program (Kjeld, 2006). Identified strains were deposited in the ZIM Collection of Industrial Microorganisms and their nucleotide sequences were deposited in the GenBank/EMBL/DDBJ database. Their accession numbers are listed in Table 1.

View this table:
Table 1

Yeast species identified using traditional identification, PCR-RFLP of ITS regions using four restriction enzymes (HaeIII, CfoI, HinfI, MspI) and sequencing of the D1/D2 region of the LSU rRNA gene and ITS1–5.8S rRNA gene–ITS2 regions

SpeciesStrain designationPCR-RFLP of ITS regionsAccession numbers
D1/D2 regionITS regions
Aureobasidium pullulansZIM 625HaeIII: 430+150 CfoI: 190+170+90+90 HinfI: 300+190+130 MspI: 290+190+160ND
Auriculibuller fuscusZIM 609HaeIII: 470+80 CfoI: 290+250 HinfI: 270+240 MspI: 330+200AM748525
ZIM 2294
Bulleromyces albusZIM 608HaeIII: 530 CfoI: 260+250 HinfI: 300+130+80 MspI: 520AM748526
ZIM 643ND
Candida zemplininaZIM 677HaeIII: 460 CfoI: 210+110+60 MboI: 290+130 DraI: 310+120+50ND
Cryptococcus adeliensisZIM 600HaeIII: 490+70 CfoI: 310+290 HinfI: 340+270 MspI: 600AM748527FN400760
ZIM 638ND
Cryptococcus flavescensZIM 667HaeIII: 480 CfoI: 270+200 HinfI: 300+170 MspI: 400+90AM748528
ZIM 641AM748548
Cryptococcus magnusZIM 596HaeIII: 490+80 CfoI: 340+290 HinfI: 260+220+140 MspI: 400+240AM748529
ZIM 634ND
Cryptococcus sp.ZIM 2300HaeIII: 290+130+50 CfoI: 240+170+90 HinfI: 240+230 MspI: 250+210FN400761
Cryptococcus wieringaeZIM 602HaeIII: 420+80+60 CfoI: 330+270 HinfI: 240+220+140 MspI: 550AM748532
ZIM 640ND
Cryptococcus zeaeZIM 607HaeIII: 460 CfoI: 240+220 HinfI: 260+230 MspI: 440+60AM748533
Erythrobasidium hasegawianumZIM 664HaeIII: 610 CfoI: 550+60 HinfI: 240+140+120+60 MspI: 610AM748534
Filobasiduim floriformeZIM 606HaeIII: 490+80 CfoI: 320+280 HinfI: 360+260 MspI: 400+240AM748535FN400759
Hanseniaspora guilliermondiiZIM 623HaeIII: 760 CfoI: 310+310+100 HinfI: 360+200+160+70 MspI: 760AM748540
Hanseniaspora uvarumZIM 668HaeIII: 760 CfoI: 310+310+100 HinfI: 360+200+160+70 MspI: 760ND
Holtermanniella takashimaeZIM 629HaeIII: 350+70+70 CfoI: 270+220+150+140 HinfI: 430+370 MspI: NDAM748531
Kwoniella mangroviensisZIM 605HaeIII: 360+105+75 CfoI: 310+175+110 HinfI: 300+270 MspI: 380+210AM748530
Metschnikowia pulcherrimaZIM 621HaeIII: 270+100 CfoI: 200+90+80 HinfI: 200+180 MspI: 210+120+70AM748541
ZIM 669ND
Pichia guilliermondiiZIM 624HaeIII: 385+110+85 CfoI: 290+260 HinfI: 335+300 MspI: 390+170+100AM748542
Pseudozyma prolificaZIM 630HaeIII: 360+310+210 CfoI: 260+210+150 HinfI: 430+370 MspI: NDFN397675
Rhodotorula bacarumZIM 666HaeIII: 400+170 CfoI: 280+270 HinfI: 160+100+50 MspI: 540+50AM748543
Rhodotorula glutinis var. glutinisZIM 615HaeIII: 600 CfoI: 300+210+100 HinfI: 210+120+110 MspI: 600AM748544
ZIM 618ND
ZIM 653AM748550
Rhodotorula nothofagiZIM 620HaeIII: 390+100+100 CfoI: 310+280 HinfI: 260+150+100+80 MspI: 300+270+60AM748545
ZIM 657ND
Rhodotorula phylloplanaZIM 662HaeIII: 440+160 CfoI: 270+270 HinfI: 190+140+100+60 MspI: 570+60AM748546
Sporidiobolus pararoseusZIM 611HaeIII: 600 CfoI: 290+270 HinfI: 260+120+110+100 MspI: 280+280AM748547
ZIM 631AM748549
  • * ZIM, Collection of Industrial Microorganisms, Slovenia.

  • ND, not determined.

The identity of the Mythos® fungicide contaminant was determined by plating the isolate on Pseudomonas-selective agar with nalidixic acid (Biolife, Milan, Italy) and by Gram staining.

In vitro fungicide sensitivity

Selected yeast strains isolated from the grape berries were tested for their sensitivity to fungicides using halo assays (Dunn, 2005). The overnight cultures grown in yeast extract-peptone-dextrose (YPD) (Sigma-Aldrich, St. Louis, MO) were resuspended in 4 mL of top agar (0.7% agar, 0.82 × YPD) to a final concentration of 1 × 106 cells mL−1. The top agar was poured onto YPD plates. Filter discs (6 mm) (BD, Franklin Lakes, NJ) were soaked in fungicides prepared in the concentrations recommended for application on grape vines by the suppliers: Kidan® containing iprodione as an active substance (30 μL mL−1 solution in water), Mythos® containing pyrimethanil (10 μL mL−1 solution in water) and Switch® containing fludioxonil+cyprodinil (1 mg mL−1 solution in water). The control discs were soaked in sterile-distilled water. The plates were incubated at 26 °C for 48 h. The annular area of the halo was calculated by measuring the total clear zone diameter (including the filter disc), and then subtracting the area of the filter disc. The data were calculated as averages of four replicates for each strain.

Statistical analysis of the data

The significance of the differences between control and fungicide-treated samples was determined on the yeast density and relative abundance data by single-factor anova (Microsoft excel 2002).

Results and discussion

Determination of fungicide residues on grapes at harvest

In the experimental vineyard, four plots consisting of eight vines were treated with three commercially available fungicides used against grey mould disease on grapes in the 2 consecutive years of 2002 and 2003. The fungicides were applied according to the recommendations of the manufacturers considering the safety intervals. Nevertheless, at harvest, the concentrations of fungicide residues on grapes were determined and were as follows in the years 2002 and 2003, respectively: cyprodinil 0.7 and 0.21, fludioxonil 0.25 and 0.06, iprodione 2.6 and 2.03 and pyrimethanil 1.1 and 0.58. These results showed that all fungicide residues on wine grapes were below the maximum residue limits (MRL) established by the European legislation (EU pesticide database, 2008). The fungicides were not detected on untreated control samples. Because the fungicide residues were still present on grapes, their effect on yeast density and diversity was further studied.

Effect of fungicide treatments on yeast density

The total yeast counts of fungicide-treated grape berries and untreated controls are presented in Fig. 1. The yeast populations of the grape berry surface ranged from 5.1 × 104 to 7.4 × 105 CFU cm−2, corresponding to the values generally reported for mature grapes (Fleet, 2002). The yeast densities differed significantly between the two seasons (P=0.002, one-way anova), as in the years 2002 and 2003, the climatic conditions before harvest were extreme. In the year 2002, precipitation exceeded the normal amount by >200% (Cegnar, 2002, Fig. 1b), whereas in the year 2003, high temperatures, sunny weather and lack of rainfall caused drought (Cegnar, 2003, Fig. 1b). Because of such weather conditions, the yeast density declined by one-third in the dry season. This is in agreement with the results of other studies (Longo, 1991; de la Torre, 1999; Combina, 2005). The latter authors also explained that humidity promotes exosmosis and consequently nutrient release from grapes, which enables the growth of yeasts on the grape berry surface.

Figure 1

Mean yeast densities on grape berries treated with fungicides iprodione, cyprodinil plus fludioxonil and pyrimethanil at harvest in the years 2002 and 2003 (a). Untreated grape berries were used as a control. The results were averaged from triplicate dilutions and are expressed as means±SDs (error bars) of three samples. (b) The sum of precipitations and average temperatures after last applications of the fungicides between August and the first decade of September in 2002 and 2003.

The fungicide treatments exerted statistically less significant effects on the yeast counts in 2002 and 2003 (P=0.03 and 0.008, respectively). Contrary to previous reports on the influence of fungicides on population size (Andrews & Kenerley, 1978; Buck & Burpee, 2002; Gildemacher, 2004; Comitini & Ciani, 2008), we found higher counts on the treated grapes than on the control samples in general. In particular, we isolated significantly larger populations from grapes treated with iprodione, where the yeast counts in both years reached their highest values (Fig. 1; 7.4 × 105 CFU cm−2 in 2002 and 3.2 × 105 CFU cm−2 in 2003). Such an effect of iprodione on yeast load could be explained by the chemical composition of the fungicide preparation, which, besides the active ingredient, also consists of other adjuvants such as surfactants, emulsifiers, dispersants, solubilizers, solvents, diluents, preservatives and thickeners that enhance the efficacy and stability of the agrichemical (Crowdy, 1971). These chemicals may be a source of energy for some bacteria as it was confirmed by Ng (2005) and assumingly also for yeasts inhabiting grape berries.

Ecological factors affecting yeast diversity

We aseptically collected 160 sound grape berries with their pedicels. From these samples treated with three fungicides and from untreated control, during two vintages, we isolated a total of 2995 yeast colonies. The colonies were morphologically classified into groups on each countable YM agar plate (nine plates per sampling), representatives of which were then further subjected to PCR-RFLP analysis of the ITS regions of the rRNA gene. On the basis of the restriction patterns generated using four restriction enzymes, 317 isolates in 2002 and 195 isolates in 2003 were grouped into 16 and 20 groups, respectively. Some strains of each group were further identified phenotypically, using traditional morphological and physiological tests. Because these tests yielded ambiguous identifications for basidiomycetous yeast isolates, sequencing of the D1/D2 domain of the LSU rRNA gene was necessary for reliable identifications. However, some basidiomycetous yeast species cannot be discriminated based only on D1/D2 sequences [e.g. Cryptococcus magnus (accession no. AF181851), Filobasidium elegans (accession no. AF181548) and Filobasidium floriforme (accession no. AF075498); Fell, 2000], and for those, the sequencing of the ITS1–5.8S rRNA gene–ITS2 regions was required. Further, the species identities were confirmed by in silico RFLP of the ITS regions using the pdraw 1.0 software (Kjeld, 2006) as suggested by Raspor (2007). We isolated 24 species belonging to 13 genera, of which nine belonged to basidiomycetous yeasts, mainly red-pigmented species of Rhodotorula and Sporidiobolus and nonpigmented or lightly pigmented Cryptococcus sp. (Table 1). These yeast species are typical phylloplane colonists (Fonseca & Inacio, 2006) and are not usually found as the prevailing species of ripe grape berry communities (reviewed by Fleet, 2002). However, some studies showed that on mature grapes, basidiomycetous yeasts are predominant over fermentative ascomycetous yeast flora that is representative of overripe or damaged grapes (Prakitchaiwattana, 2004; Barata, 2008a, b). On the other hand, Comitini & Ciani (2008) suggested that the fungicides cause the elimination of the fermenting ascomycetous yeasts from grapes. To test whether the prevailing basidiomycetous yeast communities were present because of the sampling methodology used or because of applications of fungicides, we alternatively sampled visibly healthy whole grape bunches from the same vines in the year 2002 and processed them in a manner similar to grape berries. As expected, the ascomycetous yeasts of Candida zemplinina and Hanseniaspora uvarum prevailed over Sporidiobolus pararoseus, C. magnus and Rhodotorula glutinis on whole grape bunches (Fig. 2). This clearly indicates that whole grape bunches must contain some damaged or overripe grape berries on which fermentative and rapidly proliferating ascomycetous yeast flora predominates. Based on this, we showed that grape health is the most important factor determining the yeast quantity and composition.

Figure 2

Ascomycetous (light grey) and basidiomycetous (dark grey) yeast species isolated from sound grape berries (Ber) and from grape bunches (Bun) treated with fungicides iprodione, K; pyrimethanil, M; fludioxonil plus cyprodinil, S; and an untreated control, O. Those with a frequency of isolation of <6% were considered as minor species.

Further, only the samples consisting of sound grape berries with their pedicles were used for studying the effects of fungicides on naturally occurring grape microbial communities (Table 2). Using a polyphasic approach, a high diversity of minor species was revealed. Their frequency of isolation was <6% or <10 colonies present on nine countable plates of one sample (Table 2). Among them was a wide variety of cream or lightly pigmented Cryptococcus sp. or their teleomorphic genera of Auriculibuller, Bulleromyces, Filobasidium, Kwoniella and Holtermanniella gen. nov. Most of them were species that have been described recently [Cryptococcus adeliensis (Scorzetti, 2000), Cryptococcus zeae (Molnar & Prillinger, 2006), Auriculibuller fuscus (Sampaio, 2004) and Kwoniella mangroviensis (Statzell-Tallman, 2008)] or recognized as distinct species using the molecular approach [Cryptococcus flavescens (Takashima, 2003) and Cryptococcus wieringae (Fonseca, 2000)]. In the group of new or undescribed species, isolate ZIM 629 showed the highest similarity in the D1/D2 region with Holtermanniella takashimae sp. nov. VTT C-04546 (Wuczkowski, 2010) and isolate ZIM 2300 differed from its closest relative, Cryptococcus sp. CBS 10174, by 15 nucleotides in the D1/D2 region. However, none of these species have been associated with grape berries as yet. Interestingly, C. adeliensis and C. flavescens were reported as causative agents of cryptococcosis in immunocompromised patients (Rimek, 2004; Tintelnot & Losert, 2005) and the former species have also been frequently isolated from glacial environments (Scorzetti, 2000; Butinar, 2007; de Garcia, 2007). Nevertheless, these minor species appear to be widespread in nature according to the records in the GenBank, but due to their physiological resemblance to the species complexes of Cryptococcus albidus, Cryptococcus laurentii or C. magnus, their proper affinity might be missed.

Table 2

Numbers of yeast species and their frequency of occurrence (<5% white; 5–20% light gray; 20–30% grey; 30–50% dark grey; >50% black) on sound grape berries treated with fungicides and untreated control in the years 2002 and 2003

The second group of basidiomycetes was represented by red-pigmented species of Rhodotorula, its teleomorphic genera Erythrobasidium and Rhodosporidium and also Sporidiobolus sp. Even though S. pararoseus and R. glutinis prevailed on the grapes as pigmented species in all samples, the diversity of the minor species of Rhodotorula bacarum, Rhodotorula nothofagi, Rhodotorula phylloplana and Erythrobasidium hasegawianum depended on the identity of the predominant species. In the year 2002, when the predominant species was S. pararoseus, all four minor species were detected. In contrast, in the year 2003, when the predominant species were Aureobasidium pullulans and C. magnus, R. nothofagi was the only minor pigmented species isolated. The reported antagonistic activity of dimorphic hyphomycetes A. pullulans (Castoria, 2001) might influence the species composition of grape berries. However, additional studies are needed to confirm this hypothesis.

Effect of fungicide treatments on predominant yeast populations

In order to assess the effect of fungicides on nontarget organisms, we compared the relative abundances of the dominant yeast species present on treated grape berries with the untreated control (Fig. 3). Those with a frequency of isolation of >6% were considered to be dominant yeast species (Table 2). The species abundance was expressed as a percentage of the colonies belonging to the dominant species relative to the total number of colonies recovered from three countable plates of each sample. The samples were taken in triplicate per treatment.

Figure 3

Relative abundances of dominant yeast species on grape berries treated with fungicides iprodione, pyrimethanil, cyprodinil plus fludioxonil and an untreated control. The results were averaged from triplicate dilutions and are expressed as means±SDs (error bars) of three samples.

None of the fungicides used, iprodione, fludioxonil+cyprodinil or pyrimethanil, affected the abundances of A. pullulans, C. magnus, R. glutinis or S. pararoseus significantly (P>0.05, one-way anova) compared with the untreated control. These findings were in sharp contrast with our expectations, as well as with published findings on the impact of fungicides on wine-related yeasts (Guerra, 1999; Calhelha, 2006; Comitini & Ciani, 2008; Cus & Raspor, 2008). However, of these reports, only Comitini & Ciani (2008) systematically studied the impact of fungicide on the yeast diversity of grape berries. They showed a significant reduction in yeast density and a loss of fermentative microbial communities on grapes treated with cyprodinil+fludioxonil fungicide. In view of our findings on the influence of grape berry sampling methodologies (Fig. 2), their results might depend on the quality of the grape berries sampled and therefore the dominance of basidiomycetous yeast flora on the treated grapes might only be a consequence of grape berry soundness and not of spraying the grapes with fungicide.

A reduction in the abundance and diversity of nontargeted epiphytic yeasts on various fruits or leaves was also reported by several other authors (Andrews & Kenerley, 1978; Buck & Burpee, 2002; Gildemacher, 2004, 2006; Walter, 2007). Notably, the detrimental effect of fungicides on phylloplane yeasts was clearly shown for broad-spectrum fungicides, for example captan, dithianon, dodine (Gildemacher, 2004; Walter, 2007) and for many of the ergosterol-biosynthesis inhibitors, for example propiconazole (Buck & Burpee, 2002). Among the latter is iprodione, which was used in this study as a representative of an old-generation fungicide. Field experiments on the toxic effects of iprodione were not performed, but Buck & Burpee (2002) observed that the in vitro growth of phylloplane yeast isolates was not suppressed by iprodione due to acquired resistance by yeasts exposed to fungicide treatments. In our case this could be one of the explanations why the yeast flora of the vine-growing region is resistant to the fungicides used in the vineyards. However, additional in vitro testing of strains isolated from other sources for fungicide resistance would be necessary to confirm this hypothesis.

Representatives of the ‘last-generation’ fungicides used in our study were cyprodinil, fludioxonil and pyrimethanil, for which the manufacturers had to demonstrate their ineffectiveness against fermentative yeasts at their registration (Cabras, 1999). Their specificity against plant pathogens was confirmed in our study.

The effects of fungicides on naturally occurring grape microbial communities are difficult to establish directly in the field because fungicides are not the only factor generating differences in the population densities and the diversity of phylloplane yeasts. First, the influence of meteorological factors cannot be excluded and our results showed that the drought in the year 2003 accounted for a decrease in the yeast densities and for an increased abundance of A. pullulans in favour of S. pararoseus. To minimize the effect of weather conditions, sampling over a longer period of time (e.g. 10 years) would be required. Second, the influence of fungicide treatments on sporadically occurring species cannot be determined under field conditions as their incidence is too low to be statistically evaluated. And third, the high variability among samples of the same treatment (Fig. 3) makes interpretation of the data difficult. Hence, to avoid bias in the data interpretation, we performed in vitro sensitivity tests of selected isolates.

In vitro fungicide sensitivity of grape isolates

In order to evaluate yeast species and strain sensitivity to fungicides in vitro, we performed halo assays using filter discs loaded with fungicides at concentrations recommended by the manufacturers for grape berry applications. We tested 71 strains belonging to all 22 yeast species. The representatives of each species were isolated from the sound grape berries treated with different fungicides in both years (one representative per treatment). The results are shown in Fig. 4 as a graphical and photographic presentation of the area of the halo of different species. The most notable feature of the chart is that pyrimethanil suppressed the growth of all basidiomycetous yeast species, while the ascomycetous species of Hanseniaspora guilliermondii, H. uvarum and Metschnikowia pulcherrima were unaffected. Pichia guilliermondii was only slightly affected by pyrimethanil. Notably, the discs loaded with Mythos® fungicide (active substance pyrimethanil) were contaminated by the bacteria Pseudomonas sp. (Fig. 4, photos of halos). The contamination was observed even when a new batch of the fungicide was purchased. Addition of Pseudomonas sp. is not specified in the patented formulation of the commercial suspension of Mythos® (Dimitrova, 2000). In view of this, the hypothesis of Ng (2005) that the pesticide solutions might be a source of microbial contaminants has proven to be correct, even though they did not confirm the presence of viable microorganisms in any of the pesticides tested. However, they showed that some pesticides supported the growth of Pseudomonas sp. to a final concentration of 106–107 CFU mL−1. On this basis, we assume that the contamination occurred during the manufacture of this fungicide and that the active substance, surfactants, dispersants or wetters enabled the survival/growth of these microorganisms. This finding is important evidence that some pesticides are a source of microbial contamination that can be a risk to public health, can cause grape rot or affect the natural composition of grape berry communities that are important biological control agents.

Figure 4

Effects of iprodione, pyrimethanil and fludioxonil plus cyprodinil fungicides on the in vitro growth of selected yeasts isolated from treated and control grape berries. The upper part shows a graphical representation of the annular areas (log2 of the area of the total halo minus the area of the disc). The sizes of four annular areas were averaged and are expressed as means±SDs (error bars) of n number of strains belonging to the same species. The lower part shows photographs of the halos. A.p., Aureobasidium pullulans; A.f., Auriculibuller fuscus; B.a., Bulleromyces albus; C.a., Cryptococcus adeliensis; C.f., Cryptococcus flavescens; C.m., Cryptococcus magnus; C.VTT, Holtermanniella takashimae; C.sp., Cryptococcus sp.; C.w., Cryptococcus wieringae; C.z., Cryptococcus zeae; E.h., Erythrobasidium hasegawianum; F.f., Filobasidium floriforme; H.g., Hanseniaspora guilliermondii; H.u., Hanseniaspora uvarum; K.m., Kwoniella mangroviensis; M.p., Metschnikowia pulcherrima; P.g., Pichia guilliermondii; P.p., Pseudozyma prolifica; R.b., Rhodotorula bacarum; R.g., Rhodotorula glutinis; R.n., Rhodotorula nothofagi; R.p., Rhodotorula phylloplana; and S.p., Sporidiobolus pararoseus.

Interestingly, we observed an antagonistic interaction between yeasts unaffected by pyrimethanil (H. uvarum, H. guilliermondii and M. pulcherrima) and between Pseudomonas sp. This can be seen as a lack of bacterial growth around the discs on plates inoculated with these yeasts (Fig. 4, photos of halos). The Hanseniaspora sp. were also unaffected by iprodione and fludioxonil+cyprodinil, but M. pulcherrima was sensitive to the former fungicide. On the other hand, the asexual counterpart of smut fungi Pseudozyma prolifica and the rose-coloured yeasts R. bacarum and R. phylloplana were suppressed by all three fungicides.

In general, the in vitro experiment showed that application of fungicides in the concentrations recommended by the manufacturers selectively reduced grape berry microbial communities, but beyond the safety interval, as their concentrations decreased below the MRL, recolonization of these yeasts must have occurred. This is evident from the comparisons between the annular areas of the halos of the strains and the fungicide-treated or control samples from which these strains were isolated (e.g. the single strain of P. prolifica that was affected by all three fungicides was isolated from iprodione-treated grapes; Table 2).

The results of this study showed that beyond the safety interval, the presence of fungicides has a minor impact on the composition of grape berry communities, although at the time of applications of the fungicides pyrimethanil and fludioxonil+cyprodinil, the basidiomycetous yeast flora changed. We assume that beyond the safety intervals considered recolonization of grape berries occurred. On sound grape berries, the fermentative yeast species were present in extremely low concentrations, but in this study, we showed that this was not due to the applications of fungicides by in vitro experiments.

Acknowledgements

The authors wish to thank to Dr F. Čuš, Dr A. Gregorčič and Dr H. Baša-Česnik and acknowledge the technical assistance of Katja Jug. This work was supported by the Ministry of Higher Education, Science and Technology and by the Ministry of Agriculture, Forestry and Food of the Republic of Slovenia (CRP project no. V4-0591).

Footnotes

  • Editor: Teun Boekhout

References

View Abstract