Yeast heterogeneity during spontaneous fermentation of black Conservolea olives in different brine solutions

Authors

  • A.A. Nisiotou,

    1. Laboratory of Microbiology and Biotechnology of Foods, Department of Food Science & Technology, Agricultural University of Athens, Iera Odos, Votanikos, Athens, Greece
    Search for more papers by this author
  • N. Chorianopoulos,

    1. Laboratory of Microbiology and Biotechnology of Foods, Department of Food Science & Technology, Agricultural University of Athens, Iera Odos, Votanikos, Athens, Greece
    Search for more papers by this author
  • G.-J.E. Nychas,

    1. Laboratory of Microbiology and Biotechnology of Foods, Department of Food Science & Technology, Agricultural University of Athens, Iera Odos, Votanikos, Athens, Greece
    Search for more papers by this author
  • E.Z. Panagou

    1. Laboratory of Microbiology and Biotechnology of Foods, Department of Food Science & Technology, Agricultural University of Athens, Iera Odos, Votanikos, Athens, Greece
    Search for more papers by this author

Efstathios Z. Panagou, Laboratory of Microbiology and Biotechnology of Foods, Department of Food Science and Technology, Agricultural University of Athens, Iera Odos 75, Athens, Greece, GR-118 55.
E-mail: stathispanagou@aua.gr

Abstract

Aims:  To assess the yeast community structure and dynamics during Greek-style processing of natural black Conservolea olives in different brine solutions.

Methods and Results:  Black olives were subjected to spontaneous fermentation in 6% (w/v) NaCl brine solution or brine supplemented with (i) 0·5% (w/v) glucose, (ii) 0·2% (v/v) lactic acid and (iii) both glucose and lactic acid. Yeast species diversity was evaluated at the early (2 days), middle (17 days) and final (35 days) stages of fermentation by restriction fragment length polymorphism and sequence analyses of the 5·8S internal transcribed spacer and the D1/D2 ribosomal DNA (rDNA) regions of isolates. Analysis revealed a relatively broad range of biodiversity composed of 10 genera and 17 species. In all treatments, yeasts were the main micro-organisms involved in fermentation together with lactic acid bacteria that coexisted throughout the processes. Metschnikowia pulcherrima was the dominant yeast species at the onset of fermentation, followed by Debaryomyces hansenii and Aureobasidium pullulans. Species heterogeneity changed as fermentations proceeded and Pichia membranifaciens along with Pichia anomala evolved as the main yeasts of olive elaboration, prevailing at 17 and 35 days of the process. Molecular techniques allowed for the identification of five yeast species, namely A. pullulans, Candida sp., Candida silvae, Cystofilobasidium capitatum and M. pulcherrima, which have not been reported previously in black olive fermentation.

Conclusions:  By using molecular techniques, a rich yeast community was identified from Conservolea black olive fermentations. Metschnikowia pulcherrima was reported for the first time to dominate in different brines at the onset of fermentation, whereas Pichia anomala and P. membranifaciens evolved during the course. The addition of glucose and/or lactic acid perturbed yeast succession and dominance during fermentation.

Significance and Impact of the Study:  Yeasts have an important role in black olive fermentation and contribute to the development of the organoleptic characteristics of the final product. At the same time, certain species can cause significant spoilage. The present study adds to a better knowledge of yeast communities residing in olive fermentations towards a well-controlled process with minimization of product’s losses.

Introduction

Natural black olives in brine is one of the three most important commercial preparations of table olives in the international market, the other two are Spanish-style green olives and black oxidized olives also known as Californian-style olives (Garrido Fernández et al. 1997). This kind of preparation is especially important in Greece, where almost 50% of the 90 000–100 000 tonnes of drupes processed annually produce natural black olives. Thus, it is not surprising that this type of olives is worldwide known as Greek-style black olives in brine. The bulk volume of olives is processed according to a traditional anaerobic method in which the drupes after harvest, sorting and washing, are immersed in 8–10% (w/v) NaCl brine, where they undergo spontaneous fermentation by a mixed microbiota of Gram-negative bacteria, lactic acid bacteria and yeasts (Balatsouras 1990). After an initial stage of vigorous fermentation, where the diverse microbial groups compete for nutrients, the process is dominated by lactic acid bacteria and yeasts that coexist throughout the process (Durán Quintana et al. 1986; Tassou et al. 2002). The relevant population of each group over others depends on several technological factors, such as salt concentration, initial pH adjustment, oxygen availability, diffusion of nutrients from the drupes and fermentation temperature (Durán Quintana et al. 1986, 1997; Özay and Borcakli 1996; Nychas et al. 2002; Tassou et al. 2002).

The contribution of yeasts in table olive processing has positive and negative aspects in both fermentation and packing of the final product (Arroyo-López et al. 2008). During fermentation, the beneficial role of yeasts lies in the production of desirable volatile compounds and metabolites that improve the organoleptic properties (Garrido Fernández et al. 1995), enhancement of the growth of lactic acid bacteria (Tsapatsaris and Kotzekidou 2004; Segovia Bravo et al. 2007), killer activity (Marquina et al. 1997; Psani and Kotzekidou 2006) and biodegradation of phenolic compounds (Ettayebi et al. 2003). On the other hand, yeasts may cause gas pocket formation because of excessive CO2 production at the early stage of fermentation (Durán Quintana et al. 1986; Lamzira et al. 2005) and softening of the olive tissue (Hernández et al. 2007). Finally, at the stage of packing, yeast activity may be detrimental resulting in package bulging caused by CO2 accumulation, clouding of the brines, softening, production of off flavours and odours and resistance to preservatives (Turantaşet al. 1999).

Assessment of the various populations within yeast communities is fundamental in natural black olive fermentation, where yeasts have a central role in the fermentation process and the development of the final organoleptic characteristics. In the past, characterization and identification of yeasts associated with table olives has been attempted through morphological and biochemical approaches based on taxonomic keys (Barnett et al. 1990). Recently, molecular techniques are widely employed providing higher accuracy in the identification of yeasts in fermented foods (Guillamón et al. 1998; Vasdinyei and Deak 2003; Arroyo-López et al. 2006; Nisiotou et al. 2007; Hurtado et al. 2008). However, in the case of Conservolea natural black olives, the yeast microflora has been barely explored (Kotzekidou 1997), while its succession during fermentation has not been described as yet. The aim of the present study was to evaluate the indigenous yeast population during spontaneous fermentations of natural black olives of cv. Conservolea. Fermentations were carried out in different brine solutions industrially applied in black olive preparations. Yeast species diversity at distinct phases of fermentation (early, middle and final stages) were determined through PCR-restriction fragment length polymorphism (PCR-RFLP) of the 5·8S internal transcribed spacer (ITS) region combined with sequence analysis of the D1/D2 domain of ribosomal DNA (rDNA).

Materials and methods

Olive samples and fermentation procedures

Natural black olives cv. Conservolea from the 2006/2007 crop were harvested in November from an olive tree orchard in Central Greece and transported to the laboratory within 24 h. Olives were harvested at the stage of full ripeness, when 3/4 of the mesocarp had attained black colour, which is appropriate for natural black olive processing (Balatsouras 1990). Fermentations were carried out in 3·5 l total volume plastic vessels, containing c. 2·0 kg of olives and 1·5 l of 6% NaCl brine (standard treatment) or brine supplemented with (i) 0·5% (w/v) glucose, (ii) 0·2% (v/v) lactic acid or (iii) both glucose and lactic acid at the same concentrations. All treatments were performed in duplicate. The different processes were selected on the basis that they are extensively employed by the Greek table olive industry. Fermentation vessels were maintained at controlled temperature (20°C) for an overall period of 35 days. During the process, salt level was maintained constant at 6% by periodic additions of coarse salt in the brine.

Microbiological analysis and determination of kinetic parameters

Brine samples were analysed for yeasts and lactic acid bacteria. A total of 11 sampling points at days 0, 2, 4, 6, 8, 10, 13, 17, 20, 27 and 35 were used for the determination of microbial kinetics. Decimal dilutions in 1/4 strength Ringer’s solution were prepared, and duplicate 1 or 0·1 ml samples of the appropriate dilutions were mixed or spread on the following agar media: de Man-Rogosa-Sharp medium (MRS; Merck 1·10660, Darmstadt, Germany) for lactic acid bacteria, overlaid with the same medium and incubated at 25°C for 72 h; Rose Bengal Chloramphenicol agar (RBC; Oxoid CM 549, Basingstoke, Hampshire, UK, with selective supplement SR 78) for yeasts and fungi, incubated at 25°C for 48 h.

Microbial data from plate counts vs time were modelled with the Churchill model (Membréet al. 1997) as both growth and decline in microbial counts were observed during the processes. A simplified version of the model is the following:

image(1)

where Nt is the microbial population at time t; K1 and K2 are constants; λ1 and λ2 are the growth and decline rates of the micro-organism respectively. In the case where the decline of the micro-organism was followed by a tailing phase, the Churchill model was inappropriate to describe the whole biokinetic responses. In this case, a two-term Gompertz equation was used as proposed previously by Bello and Sánchez Fuertes (1995):

image(2)

where Nt is the number of micro-organisms at time t, N0 the initial population of micro-organisms, k1 the increase of micro-organisms from the initial level to the maximum level reached, k2 the relative growth rate of the micro-organism, k3 the time at which growth rate is maximum, k4 the decrease of micro-organisms from the maximum level to a minimum level, k5 the relative death rate of the micro-organism and k6 the time at which death rate is maximum. The calculated biological parameters for yeast responses obtained in each fermentation were averaged, and the corresponding standard deviations (SD) calculated.

Yeast isolation

Yeast colonies were randomly selected from RBC plates, and isolates were purified by successive streaking onto the same medium. Pure cultures were stored at −80°C in YEPD (yeast extract/peptone/dextrose) broth with 20% glycerol until further analysis. A total of 308 isolates were picked at three different sampling times, namely 2 days after brining (T2), 17 (T17) and 35 (T35) days of fermentation. The total number of isolates corresponded to c. 20–25 colonies per agar plate, sampling time and fermentation process. Comparison between species distribution of different brine solutions at each sampling time was conducted by chi-square test at P < 0·05.

DNA extraction

Genomic DNA was extracted essentially as described by Burke et al. (2000) with minor modification. Briefly, yeast cells were grown overnight in YEPD broth (1% yeast extract, 2% bacteriological peptone, 2% glucose, pH 6·2) at 30°C in a rotary shaker. Cells were collected by centrifugation at 8000 g for 1 min, resuspended in 300 μl of breaking buffer (2% Triton X-100, 1% sodium dodecyl sulfate, 100 mmol l−1 NaCl, 10 mmol l−1 Tris pH 8, 1 mmol l−1 EDTA pH 8) and transferred to 2-ml tubes containing 0·3 g of 0·5-mm-diameter glass beads (Sigma). Cell suspension was subjected to vortex for 2 min after the addition of 300 μl phenol/chloroform/isoamyl alcohol (25 : 24 : 1 v/v). Three hundred microlitres of TE buffer (10 mmol l−1 Tris, 1 mmol l−1 EDTA, pH 7·6) was added, and the bead/cell mixture was centrifuged at 12 000 g for 10 min. The supernatant was transferred to a fresh 1·5-ml tube, and DNA was precipitated by the addition of 2·5 volumes ethanol, followed by centrifugation at 12 000 g for 10 min. The DNA pellet was washed with 70% ethanol and resuspended in water.

PCR amplification

The 5·8S-ITS rDNA region was amplified using the primers ITS1 (5′-TCCGTAGGTGAACCTGCGG-3′) and ITS4 (5′-TCCTCCGCTTATTGATATGC-3′) (White et al. 1990) and the D1/D2 domain of 26S rDNA gene using the primer-pair NL1 (5′-GCATATCAATAAGCGGAGGAAAAG-3′) and NL4 (5′-GGTCCGTGTTTCAAGACGG-3′) (Kurtzman and Robnett 1998). PCRs were performed in 50 μl containing 10 ng of template DNA, 20 pmol of each primer, 100 μmol l−1 of each dNTP and 1 U of DyNAzyme™ EXT DNA Polymerase (Finnzymes, Oy, Finland) in the incubation buffer provided by the manufacturer of the enzyme. All amplifications were achieved in a PTC-200 Peltier thermal cycler (MJ Research, Waltham, MA, USA) programmed as follows: 94°C for 3 min and 35 cycles at 94°C for 30 s, 52°C for 30 s and 74°C for 2 min, followed by 74°C for 10 min. PCR products were separated by gel electrophoresis in 1·0% (w/v) agarose gel, detected by ethidium bromide staining and photographed under UV light using a DCC camera (Sony, Japan). Sizes of fragments were determined using a standard molecular weight marker (100-bp ladder; Fermentas, Vilnius, Lithuania).

Restriction analysis

For restriction reactions of the 5·8S-ITS region, c. 500 ng of PCR products was separately incubated for 1 h at 37°C with 10 U of HinfI, HaeIII or HhaI (Takara, Japan) restriction endonucleases. Restriction fragments were separated by gel electrophoresis in 3% (w/v) agarose gel, detected by ethidium bromide staining and photographed. Sizes of fragments were estimated using standard molecular weight markers (50- and 100-bp ladders, Fermentas). For yeast species assignment, comparisons were conducted among restriction profiles of isolates and reference strains or other published profiles (Esteve-Zarzoso et al. 1999; de Llanos Frutos et al. 2004; Nisiotou and Nychas 2007).

Sequence analysis

PCR products of the 5·8S-ITS region and the D1/D2 domain of one to six randomly selected isolates per distinct restriction pattern were purified using the QIAquick® PCR purification kit (Qiagen, Germany) according to the manufacturer’s instructions. By using forward (NL1 or ITS1) and reverse (NL4 or ITS4) primers, both DNA strands were directly sequenced with an ABI 3730 XL automatic DNA sequencer by Macrogen (http://www.macrogen.com). Blast searches of sequences were performed at the National Centre for Biotechnology Information (NCBI) GenBank data library.

Nucleotide sequence accession numbers

Nucleotide sequences have been deposited in the NCBI GenBank data library under accession numbers FJ715430, FJ649188 to FJ649197 and GQ140297 to GQ140301.

Results

Population dynamics

The population dynamics of the prevailing microbial groups during the different fermentation processes is presented in Fig. 1. The growth pattern of lactic acid bacteria presented an increase within the first 5 days of the process, regardless of the treatment applied, reaching a maximum at c. 6·0–7·5 log10 CFU ml−1, followed by a steady decline thereafter. Yeasts occurred together with lactic acid bacteria throughout fermentation and followed a similar growth profile. The average initial yeast population was c. 3·0–4·0 log10 CFU ml−1 and increased to c. 5·0–6·0 log10 CFU ml−1 within the first 5–7 days of the process, with the exception of the standard brine where a maximum in yeast population was observed after 12–15 days. The kinetic parameters of yeasts and lactic acid bacteria growth/decline calculated using the Churchill and two-term Gompertz equations are presented in Table 1. Generally, lactic acid bacteria presented higher growth rate values compared to yeasts in all fermentations studied, indicating the dominance of this microbial group in low salt brines (6% w/v, NaCl). Yeasts presented an average value of growth rate for all treatments of 0·53 day−1, and an average decline rate of 0·27 day−1. It is characteristic that the highest growth rate in yeasts was observed in the brine supplemented with glucose (0·59 day−1) or glucose plus lactic acid (0·89 day−1), followed by the acidified brine (0·45 day−1) and the control treatment (0·19 day−1).

Figure 1.

 Population dynamics of lactic acid bacteria (♦) and yeasts ( bsl00001 ) during fermentation of cv. Conservolea natural black olives at 20°C with the traditional anaerobic method: (a) 6% NaCl brine, (b) brine supplemented with 0·5% (w/v) glucose, (c) brine acidified with 0·2% (v/v) lactic acid and (d) brine supplemented with 0·5% glucose and acidified with 0·2% lactic acid (lines are growth/decline curves fitted with the Churchill equation, or the two-term Gompertz equation when decline was followed by tailing. Points are average values of duplicate fermentations ± standard deviation).

Table 1.   Estimated kinetic parameters for yeasts and lactic acid bacteria during fermentation of naturally black Conservolea olives using the Churchill and two-terms Gompertz equation
EquationFermentation process
Control (6% NaCl brine)Brine with 0·5% (w/v) glucoseAcidified brine with 0·2% (v/v) lactic acidBrine with 0·5% glucose and 0·2% lactic acid
ChurchillYeastsLABYeastsLABYeastsLABYeastsLAB
  1. Data are average values from duplicate fermentations ± standard deviation.

K15·16 ± 0·827·08 ± 0·82 7·04 ± 0·13 8·82 ± 1·108·90 ± 0·847·64 ± 1·03
K26·95 ± 1·167·78 ± 0·16 7·82 ± 0·25 6·31 ± 0·146·49 ± 0·427·12 ± 0·18
λ1 (day−1)0·19 ± 0·061·08 ± 0·17 1·28 ± 0·38 0·87 ± 0·170·89 ± 0·321·11 ± 0·24
λ2 (day−1)0·013 ± 0·0060·006 ± 0·001 0·005 ± 0·001 0·002 ± 0·0010·015 ± 0·0030·009 ± 0·001
Two-term Gompertz
 k1 (log CFU ml−1)  2·59 ± 0·42 1·49 ± 0·13   
 k2 (day−1)  0·59 ± 0·13 0·45 ± 0·09   
 k3 (day)  2·17 ± 0·40 1·97 ± 0·78   
 k4 (log CFU ml−1)  1·84 ± 0·44 1·38 ± 0·20   
 k5 (day−1)  0·52 ± 0·17 0·53 ± 0·28   
 k6 (day)  9·29 ± 0·77 13·43 ± 0·9   

Identification of isolates based on RFLP and sequence analysis of rDNA

A total of 308 yeasts isolated from T2, T17 and T35 days of fermentation were subjected to PCR-RFLP analysis of the 5·8S ITS region of rDNA (Fig. 2). Using the restriction endonucleases HinfI, HaeIII and HhaI, seventeen different banding profiles were generated (profiles I–XVII, Table 2). Profile comparisons between isolates and published or reference strains directly assigned eleven groups of isolates to the species Aureobasidium pullulans, Candida aaseri, Candida boidinii, Debaryomyces hansenii, Metschnikowia pulcherrima, Pichia anomala, Pichia guilliermondii, Pichia kluyveri, Pichia membranifaciens, Rhodotorula mucilaginosa and Saccharomyces cerevisiae. Groups II, V, VI, X, XIV and XVII did not match any of the published data sets, and identification was based on sequencing analysis of the D1/D2 domain.

Figure 2.

 Representative restriction patterns of the 5·8S-ITS region of yeast isolates obtained with HhaI. M: 50-bp molecular marker; 1–2 and 4–5: Pichia membranifaciens; 3 and 6–10: Candida boidinii; 11: Debaryomyces hansenii; 12: Pichia anomala.

Table 2.   Restriction analysis of the 5·8S-ITS rDNA amplicons and sequence information for the D1/D2 region of 26S rRNA gene of yeasts isolated from fermenting Conservolea olives
ProfilePCR product (bp)Restriction fragments (bp)Closest relativeMatching nucleotides (Identity %)*
HhaIHaeIIIHinfI
  1. *Sequence identity in the D1/D2 region of isolates and the closest relative species in the GenBank.

I600185, 180, 90, 90, 65450, 150290, 170, 140Aureobasidium pullulans556/557 (99·8%)
II750500, 70, 70200, 190, 150, 90, 90310, 300Candida blattariae525/532 (98·7%)
III630290, 290, 50410, 150, 90310, 170, 150Candida aaseri513/513 (100%)
IV750350, 310, 90700390, 190, 160Candida boidinii549/549 (100%)
V620290, 240, 100410, 90, 70, 60260, 190, 160Cystofilobasidium capitatum577/578 (99·8%)
VI460210, 170, 80280, 120230, 190Candida silvae521/521 (100%)
VII650300, 300, 50420, 150, 90330, 320Debaryomyces hansenii548/548 (100%)
VIII400205, 100, 95280, 100200, 190Metschnikowia pulcherrima419/420 (99·8%)
IX650580, 70500, 150310, 3100Pichia anomala547/550 (100%)
X500260 + 120 + 80320, 90, 50280, 220Pichia manshurica540/540 (100%)
XI590295, 255390, 100, 80315, 270Pichia guilliermondii560/561 (99·8%)
XII450175, 115, 80, 80370, 80250, 200Pichia kluyveri547/548 (99·8%)
XIII500175, 110, 90, 75330, 90, 50275, 200Pichia membranifaciens481/481 (100%)
XIV620310, 210, 100620220, 130, 130, 90, 50Rhodotorula diobovatum555/555 (100%)
XV640320, 240, 80420, 220340, 220, 80Rhodotorula mucilaginosa554/554 (100%)
XVI880385, 365, 130320, 230, 180, 150365, 365, 150Saccharomyces cerevisiae554/554 (100%)
XVII640320, 270, 50600, 40290, 290, 60Zygowilliopsis californica572/572 (100%)

The D1/D2 domain of 26S rDNA gene of representative isolates from groups I to XVI was amplified and determined by sequence analysis. Previous identification based on restriction analysis of the 5·8S ITS region was confirmed. The D1/D2 domain of groups V, VI, X, XIV and XVII assigned the isolates to Candida silvae, Cystofilobasidium capitatum, Pichia manshurica, Rhodosporidium diobovatum and Zygowilliopsis californica respectively. In the case of group II, sequence alignments and further phylogenetic analysis (data not shown) clearly placed isolates within the genus Candida, being most closely related to the recently described Candida blattariae (Suh et al. 2005). However, the level of sequence divergence detected was relatively high (6 noncontiguous substitutions and 1 indel in 532 nucleotides) to assign isolates to the above species (Kurtzman and Fell 2006). Therefore, this group of isolates is hereafter referred to as Candida sp.

Yeast species heterogeneity

The different yeast species identified in the fermentations, together with their frequencies, are listed in Table 3. In all cases, Metschnikowia pulcherrima was the dominant yeast species at early fermentation (T2), with a frequency of isolates ranging from 31–56% depending on the process, followed by Debaryomyces hansenii (10–25%), Aureobasidium pullulans (3–10%) and R. mucilaginosa (3·5–9·5%). Rhodosporidium diobovatum was also at relatively high percentages (17–25%), except for the brine solution supplemented with glucose. Instead, in that treatment three fermentative yeast species, i.e. Pichia guilliermondii (27·5%), P. anomala (13·5%) and Candida sp., evolved. Other species were encountered at rather low relative proportions (≤3·5%).

Table 3.   Yeast species (% prevalence) isolated from brines of natural black olives under different fermentation processes
Yeast speciesFermentation process
6% (w/v) NaCl brineBrine supplemented with 0·5% (w/v) glucoseAcidified brine with 0·2% (v/v) lactic acidBrine supplemented with 0·5% glucose and 0·2% lactic acid
T2a*T17bT35aT2bT17aT35bT2aT17aT35cT2aT17aT35b
  1. –, not detected; T2: 2 days of fermentation; T17: 17 days of fermentation; T35: 35 days of fermentation.

  2. *Samples within the same fermentation stage with different letters showed significantly different species distribution (P < 0·05).

Aureobasidium pullulans10·0%3·5%3·0%6·5%
Candida sp.7·0%
Candida aaseri10·0%5·0%6·0%
Candida boidinii10·0%37·5%4·5%3·0%9·5%9·5%
Candida silvae3·0%
Cystofilobasidium capitatum3·0%
Debaryomyces hansenii10·0%10·5%12·5%4·5%25·0%3·0%
Metschnikowia pulcherrima47·0%31·0%56·5%31·0%3·0%
Pichia anomala40·0%8·5%13·5%95·0%61·0%86·5%39·0%72·0%71·5%
Pichia guilliermondii27·5%4·5%
Pichia kluyveri3·0%13·0%
Pichia manshurica13·0%
Pichia membranifaciens40·0%50·0%26·0%35·0%6·5%19·0%
Rhodotorula diobovatum17·0%22·0%25·0%
Rhodotorula mucilaginosa7·0%3·5%6·0%9·5%
Saccharomyces cerevisiae4·0%13·0%
Zygowilliopsis californica3·5%

Species heterogeneity changed as the fermentation proceeded and Pichia membranifaciens along with P. anomala evolved as central micro-organisms in olive fermentation, prevailing at 17 and 35 days of the course. Candida boidinii also counted for a significant proportion of the total yeast population at the respective stages, except for the brine supplemented with glucose. At the end of the process, P. anomala (61%), P. membranifaciens (26%) and P. kluyveri (13%) were the prevailing species in the glucose-supplemented process. The former two species also dominated in the acidified process with 39% and 35% isolation frequencies respectively. However, two additional species were isolated in this process, namely S. cerevisiae (13%) and P. manshurica (13%). In the acidified and glucose reinforced fermentation, apart from P. anomala (71·5%) and P. membranifaciens (19%), C. boidinii was also isolated in <10%. The structure of yeast community was different in the case of standard brine fermentation, where P. membranifaciens (19%) and C. boidinii (37·5%) were the prevailing species, followed by P. anomala (8·5%) and S. cerevisiae (4%).

Discussion

The anaerobic fermentation of natural black olives is a traditional process carried out by coexisting communities of lactic acid bacteria and yeasts, and the relevant abundance of one microbial group over the other strongly affects the organoleptic and physicochemical characteristics of the final product. In the past years, high salt levels (>10%) were used in Greece, especially in fermentations carried out at farmer’s level, resulting in a process that was dominated by yeasts and occasionally by lactic acid bacteria, rendering a product with milder taste and less self-preservation characteristics (final pH 4·5–4·8). Nowadays, table olive industries have reduced salt level to 6–8% favouring a mixed fermentation by lactic acid bacteria and yeasts that coexist until the end of fermentation resulting in a product with better characteristics (pH, acidity) compared to the traditional process (Tassou et al. 2002). More recently, the enrichment of brines with glucose and/or lactic acid is employed by the Greek table olive industry. However, the impact of this practice on microbial association of olive fermentation has not been evaluated as yet.

The growth/decay curves of lactic acid bacteria and yeasts indicated some interesting trends among the treatments. Specifically, the use of glucose enhanced the growth of yeasts compared to the standard brine, whereas the growth pattern of lactic acid bacteria seemed to be unaffected (Fig. 1a,b). This finding is in agreement with other researchers (Chorianopoulos et al. 2005) who reported that increasing concentrations of glucose in the brine from 0–1% (w/v) resulted in increased growth rates of yeasts from 0·25–0·77 days−1, whereas the corresponding changes in lactic acid bacteria growth rates did not vary considerably (0·98–1·12 day−1). The acidification of the brines with lactic acid had a pronounced effect on lactic acid bacteria, as their maximum population was decreased by c. 1·5 log10 CFU ml−1, whereas yeasts were slightly affected (Fig. 1c). This observation is in line with the findings of Durán Quintana et al. (2005), who reported that the addition of lactic acid maintained or slightly increased the population of yeasts with time. The combined use of glucose and lactic acid also favoured the growth of yeasts (Fig. 1d) which presented the highest growth rate (Table 1). This could be explained by the fermentative capacity of certain yeast species from the brine environment and their ability to assimilate the lactic acid provided for acidification (Hernández et al. 2007). Taken together, these results indicate that the treatments employed by many table olive industries in Greece in black olive fermentation have a tendency to favour the growth of yeasts over lactic acid bacteria compared to the standard brining process.

Identification of yeast isolates at the species level was initially attempted by RFLP analysis of the 5·8S-ITS. This method provides a fast means for accurate discrimination of yeast species and is commonly applied in yeast biodiversity explorations. Yet the descriptions of olive yeast biota by molecular means are very scarce (Arroyo-López et al. 2006; Coton et al. 2006; Hurtado et al. 2008). RFLP identified most of the isolates at the species level, but it had to be combined with sequence analysis to reveal Candida sp., C. silvae, Cystofilobasidium capitatum, P. manshurica, Rhodosporidium diobovatum and Zygowilliopsis californica. Informative restriction patterns for the species mentioned could be useful for species identification in olive studies or other yeast diversity explorations.

A relatively broad range of biodiversity was revealed composed of 10 genera and 17 species. The presence of Candida sp., A. pullulans, C. silvae, M. pulcherrima, R. diobovatum and C. capitatum is reported in black olive fermentation for the first time. Although stable development of yeast population was evident by applying standard plating techniques, structural alterations of yeast communities were recorded at the different stages of fermentation, indicating a dynamic succession of species.

A rich yeast community, largely composed of species unfamiliar to the olive elaboration, was isolated at the early stage of fermentation (T2). It must be emphasized that the species encountered at the beginning of fermentation originated from the initial microbiota adhered to the olive surface and could be different to those present in an industrial process where the resident yeast flora may have an important contribution in the process. For instance, the presence of M. pulcherrima, which dominated at this stage, has never been reported before. Although at lower populations, the presence of A. pullulans, Candida sp., C. silvae and C. capitatum is also reported here for the first time. Several of these species possess important technological characteristics that may influence the process and the organoleptic characteristics of the final product. For example, strains of A. pullulans can grow at salt concentration of 17% or more (Kogej et al. 2005), while C. capitatum may produce pectinolytic enzymes. However, their role and possible implications in the fermentation process need further investigation. Future research will show whether these are cosmopolitan species or constitute part of the native yeast flora. Rhodotorula mucilaginosa and R. diobovatum were also recovered at stage T2. Their presence has been previously associated with directly brined green or turning colour olives (Pelagatti 1978; Marquina et al. 1992). Species of Rhodotorula are of considerable concern in olive elaboration, because they are known to cause softening of olive tissue because of polygalacturonases excretion and pellicle formation in brine (Vaughn et al. 1972). Debaryomyces hansenii was also recovered from the brines at this stage at a rate of 10–25%, depending on the treatment. This species has been previously reported in olive brines during fermentation (Balatsouras 1990; Marquina et al. 1992; Borcakli et al. 1993; Fernández González et al. 1993; Kotzekidou 1997). It has been characterized as killer yeast with high pectolytic activity (Psani and Kotzekidou 2006; Hernández et al. 2007), and its presence in the brines has been associated with an increase in lactic acid bacteria population (Tsapatsaris and Kotzekidou 2004). Zygowilliopsis californica is a species with high lipolytic activity and has been mainly associated with extra virgin olive oil (Ciafardini et al. 2006). Its presence in brine may affect the organoleptic and nutritional value of table olives, as it can increase the free fatty acid content of drupes during processing (Ciafardini et al. 2006; Hernández et al. 2007).

In the middle (T17) and final (T35) stages of fermentation, P. anomala along with P. membranifaciens became the climax yeasts, overwhelming the oxidative or weak fermentative species prevailing at the onset of the course. Both yeasts are rather common members of olive fermenting flora (Arroyo-López et al. 2008), and they have been documented to dominate in black olive preparations (Durán Quintana et al. 1986; Fernández González et al. 1992; Kotzekidou 1997; Arroyo-López et al. 2006; Coton et al. 2006). Pichia anomala plays a central role in the fermentation process through antioxidant activity (Arroyo-López et al. 2008) or excessive production of CO2 causing olive bloater spoilage (Faid et al. 1994). Pichia membranifaciens may also affect the process by shaping yeast association through the production of killer toxins (Marquina et al. 1997; Psani and Kotzekidou 2006). In the present study, it was shown that the addition of glucose and/or lactic acid may favour the growth of P. anomala at the midstage of fermentations over P. membranifaciens. This could be attributed to the strong fermentative character of P. anomala and its capacity to assimilate lactic acid. On the other hand, P. membranifaciens exhibits variability in the fermentation of glucose and lactic acid (Kurtzman 1998).

Candida aaseri and C. boidinii were encountered at relatively low percentages (<10%) in different brine solutions both at the middle and at the final stages of fermentation. Both species constitute part of the olive-associated yeast community, as it has been characterized so far. The former yeast has been reported previously from the mid-fermentation stage of directly brined green olives of Arbequina variety (Hurtado et al. 2008) and has been associated with olive bloater spoilage (Faid et al. 1994). Candida boidinii has been isolated from black olives processed by either aerobic or anaerobic method (Arroyo-López et al. 2006) and from directly brined black olives of Hojiblanca variety as well as directly brined green Arbequina olives (Hurtado et al. 2008). Other yeast species detected in minor quantities were S. cerevisiae, P. kluyveri and P. manshurica. Saccharomyces cerevisiae is a fermentative yeast associated with olive fermentation since the first scientific studies of the process. The low occurrence of this yeast in the brines is in agreement with previous reports in different kinds of olive preparations (Marquina et al. 1992; Fernández González et al. 1993; Hernández et al. 2007), whereas other researchers have reported higher isolation frequencies for directly brined black olives processed with the traditional anaerobic method (Arroyo-López et al. 2006).

In conclusion, by applying molecular techniques, a rich yeast community was identified from Conservolea traditional anaerobic black olive fermentation. Metschnikowia pulcherrima was reported for the first time to dominate in different brines at the onset of fermentation. During the course, P. anomala and P. membranifaciens evolved, while the presence of glucose and/or lactic acid favoured the dominance of the former species. Discrepancies were also encountered among quantitatively minor species in different brine solutions. These structural differences may influence the succession of species and determine, at least partially, the final species dominance and product quality.

Acknowledgements

The authors acknowledge the TRUEFOOD–‘Traditional United Europe Food’, an Integrated Project financed by the European Commission under the 6th Framework Programme for RTD (contract number FOOD-CT-2006-016264). The information in this document reflects only the authors’ views, and the Community is not liable for any use that may be made of the information contained therein. The technical assistance of Miss M. Sourri and Mr. G. Sgouros is also greatly appreciated.

Ancillary