International Journal of Food Microbiology 95 (2004) 157 168 www.elsevier.

com/locate/ijfoodmicro

Application of central composite design and response surface methodology to the fermentation of olive juice by Lactobacillus plantarum and Debaryomyces hansenii
S. Tsapatsaris, P. Kotzekidou *
Laboratory of Food Microbiology and Hygiene, Department of Food Science and Technology, Faculty of Agriculture, Aristotle University of Thessaloniki, P.B. 250, 54124 Thessaloniki, Greece Accepted 6 February 2004

Abstract The objectives of this study were to investigate the effects of NaCl, calcium acetate and calcium lactate in concentrations corresponding to ionic strengths equivalent to 2 10%, w/v salt brines as well as the 50% replacement of NaCl contained in the above mixture by KCl. A central composite design and response surface methodology were used to optimize the maximum specific growth rate of Lactobacillus plantarum ATCC 8014 and Debaryomyces hansenii 2114. The fermentation was carried out in olive juice obtained from Kalamon black olives at 30 jC with initial pH 5.0. Mathematical models describing the combined effects of these factors on the maximum specific growth rate of L. plantarum or D. hansenii were established. Both strains in single cultures showed higher maximum specific growth rate in olive juice supplemented with NaCl/KCl, Ca-acetate and Ca-lactate. But in mixed culture fermentations of olive juice supplemented with NaCl, Ca-acetate and Ca-lactate, higher specific growth rates were obtained. Under the optimum growth conditions determined for the single culture fermentations, i.e. 378.4 mM NaCl, 34.1 mM Ca-acetate and 39.9 mM Ca-lactate, mixed culture fermentation was undertaken by varying the time of inoculation of the yeast strain. When D. hansenii was inoculated 48 h before L. plantarum the maximum specific growth rate of L. plantarum was increased to 0.247 per hour, which was significantly higher compared to L. plantarum alone (0.211 per hour). In mixed culture fermentation of olive juice supplemented with the mixture of NaCl/KCl under similar conditions as above, a maximum specific growth rate of L. plantarum of 0.218 per hour was determined. The optimum conditions determined for mixed culture fermentation are useful in fermentation of black olives Kalamon variety under lower salt content. D 2004 Elsevier B.V. All rights reserved.
Keywords: Olive juice fermentation; Response surface methodology; Central composite design; Maximum specific growth rate; Lactobacillus plantarum; Debaryomyces hansenii

1. Introduction Olive processing is the largest agro-industrial sector in Mediterranean countries such as Spain, Italy and
* Corresponding author. Fax: +30-2310998791. E-mail address: kotzekid@agro.auth.gr (P. Kotzekidou). 0168-1605/$ - see front matter D 2004 Elsevier B.V. All rights reserved. doi:10.1016/j.ijfoodmicro.2004.02.011

Greece (Leone, 2000). The different processing methods applied in order to remove the natural bitterness of the fruit, due to oleuropein, a secoiridoid glucoside, are related to the stage of ripeness of the fruits. The transformation of table olives occurs during fermentation in the brining stage. According to the Greekstyle process, completely ripe olives, i.e. when the

158

S. Tsapatsaris, P. Kotzekidou / International Journal of Food Microbiology 95 (2004) 157168

skin has a colour between violet and black, but before they are over-ripe, are placed in brine and the debittering process is slow and sometimes partial. The fruits are then put into tanks and covered with a 6 10% brine. A mixed population of Gram negative bacteria of the coliform aerogenes group, lactic acid bacteria (mainly Lactobacillus spp.) and yeasts are present if the NaCl concentration is V 7%. The growth of lactic acid bacteria throughout the fermentation depends on various factors such as the olive cultivar, NaCl concentration, initial pH correction of brines, etc. (Quintana et al., 1997). If the concentration of salt in the brine stays at 10%, the yeasts are dominant. In this case, an acetic fermentation caused by acid forming, salt tolerant yeasts occurs. When fermentative yeasts predominate, the final product has pH values between 4.5 and 4.8. Although the contribution of individual yeasts to the fermentation is unclear, it is believed that they contribute to the organoleptic characteristics of fermented olives. Yeasts strains isolated from spontaneous naturally black olive fermentations are mainly Torulaspora delbrueckii, Debaryomyces hansenii, and Cryptococcus laurentii (Kotzekidou, 1997). In spontaneous fermentations, the predominant microorganisms are yeasts, whereas in reduced salt concentrations, a lactic acid fermentation occurs in addition. A more active role of lactic acid bacteria is desirable to limit the negative effects of spoilage yeasts on the product quality, as yeasts strains are responsible for deterioration, i.e. gas pocket formation and softening of the fruit (Ozay and Borcakli, 1996). Olive processing remains an empirical process and is far from being controlled. However, interest in developing starter cultures to be used in table olive fermentation is increasing. The determination of the conditions for bacterial growth in olive brines is of great importance for the development of a new biotechnological process for removing the bitter-tasting oleuropein (Marsilio and Lanza, 1998). Lactobacilli have an important role in the biological debittering process of naturally ripened olives. As Lactobacillus plantarum dominates the epiphytic microflora, the use of oleuropeinolytic L. plantarum strains as starter culture is of great importance. Response surface methodology, a statistically designed experimental protocol, in which several factors are simultaneously varied, is finding progres-

sive applications in the optimization of microbial growth (Roebuck et al., 1995). The multi-variate approach reduces the number of experiments, improves statistical interpretation possibilities and indicates whether parameters interact. Central composite design is the most widely used response surface design. Although rotatability is a desirable property of a central composite design, in the case of difficulty to extending the star points1 beyond the experimental region defined by the upper and lower limits of each factor a face-centered design can be used (Neter et al., 1996). A multiresponse optimization to minimize salt concentration in natural cucumber fermentation and storage (Guillou and Floros, 1993) and to establish conditions for green table olive fermentation at low temperature (Duran Quintana et al., 1999) using response surface methodology have been reported. Increasing concern about environmental pollution problems created by disposal of high salt brines (Guillou and Floros, 1993) as well as the development of hypertension in humans (Naewbanji et al., 1990), which is correlated with cardiovascular diseases, have prompted food processors to develop low sodium products. The negative effects related to high salt concentration can be reduced by replacing NaCl with an equivalent amount of other salts. One often-used approach has been to substitute sodium chloride with chloride salts of potassium or calcium. Substitution of NaCl by 50% KCl, resulted in more selective growth conditions for lactic acid bacteria in cucumber extracts (Naewbanji et al., 1990). In order to overcome the negative effects correlated with olive fermentation and storage under reduced salt concentrations, i.e. shrivelling and softening of the fruits, the substitution of NaCl has to be with a mixture of salts with an equivalent ionic strength to that of the control fermentations as the ionic strength is a very important factor accounting for the behaviour of ions in a solution (Pitzer, 1995). In preliminary studies, the effect of substitution of NaCl by KCl, calcium acetate and calcium lactate on the texture, flavour and colour characteristics of fermented olives was evaluated. The aim of the present investigation was to employ response surface methodology to determine
1 Also called axial points; a star point is one in which all factors but one are set at their mid-levels.

S. Tsapatsaris, P. Kotzekidou / International Journal of Food Microbiology 95 (2004) 157168

159

the optimum concentrations of calcium acetate and calcium lactate that minimize either the NaCl or the combination of NaCl/KCl concentration (in the case of 50% replacement of NaCl by KCl) necessary for achieving maximum specific growth rate of L. plantarum and D. hansenii in olive juice. Olive juice was used as a model fermentation medium to represent conditions of the Greek-style black olive fermentation. The mean values of the optimum concentrations of the factors were also applied in mixed culture fermentations. The predictive model for growth of L. plantarum and D. hansenii in olive juice will be useful in modeling the mixed culture (lactic acid bacteria and yeast) fermentation of brined black olives.

2. Materials and methods 2.1. Microorganisms, media and preparation of inocula The strains used in this study are L. plantarum ATCC 8014, a reference oleuropeinolytic microorganism as reported by Marsilio et al. (1996), and D. hansenii 2114 isolated from Greek-style fermented black olives by Kotzekidou (1997). L. plantarum (ATCC 8014) was maintained in MRS agar (MRSA, Merck, Darmstadt, Germany) and D. hansenii in Yeast Malt agar (Difco). For propagation of inocula, MRS broth and YM broth were inoculated by transferring a colony of L. plantarum or D. hansenii from MRS agar and YM agar, respectively, and incubated at 30 jC for 24 h to reach ca. 108 cfu/ml. Cells were harvested by centrifugation (5000 g, 10 min), washed twice in saline (0.9% NaCl, w/v), and finally resuspended in sterile distilled water. 2.2. Preparation of olive juice broth and fermentation conditions Traditionally debitted black olives of Kalamon variety were purchased from a local plant (Tzounakos AE, Gythion, Lakonia). The fruits were pitted off, tap water was added (1:1, w/v), and the olives were blended. The slurry was allowed to equilibrate at 4 jC overnight and then filtered until a pulp-free liquid was obtained. To determine the effect of NaCl,

KCl, calcium lactate (Ca-lactate) and calcium acetate (Ca-acetate), the medium was supplemented with NaCl (256.6 to 855 mM), whereas in the case of replacement of NaCl by KCl the substitution corresponded to 50% molar substitution of NaCl with KCl, calcium acetate (19.1 to 190 mM in fermentations carried by L. plantarum, whereas in fermentations with D. hansenii from 19.1 to 100 mM) calcium lactate (9.6 to 95 mM in fermentations carried by L. plantarum, whereas in fermentations with D. hansenii from 9.6 to 185 mM) as indicated in Table 1. All ions present in the fermentation medium were taken into account to calculate the ionic strength of the solution according to the equation I = 1/2Simi z2 where mi the molality of the i ith ion in mM of the solute and zi its charge (Pitzer, 1995). The ionic strength values of the media used in each treatment were between 343 and 1710, which correspond to 2 10%, w/v NaCl in fermentation brines. The pH of olive juice was adjusted to 5.0 with 5 N NaOH or HCl and then sterilized by autoclaving at 115 jC for 15 min. The fermentation was carried out in 250 ml screw cap flasks containing 100 ml olive juice. After inoculation with L. plantarum ATCC 8014 or D. hansenii 2214 to obtain an inoculation level of 3.0 log cfu/ml in olive juice, the fermentation flasks were incubated at 30 jC for 7 days. 2.3. Microbiological analysis Samples (1 ml) were removed from the fermentation media at the following times: 0, 3, 5, 6, 7, 8, 9, 10, 12, 14, 16, 23, 32, 40, 48, 72, 96, 120, 144, 168 h. The samples were serially diluted in sterile saline and plated onto MRS agar for enumeration of L. plantarum incubated anaerobically (GasPak, BBL, Cockeysville, MD, USA) or on YM agar for enumeration of D. hansenii aerobically, both incubated at 30 jC for 3 days. In the case of mixed culture fermentations of olive juice by L. plantarum and D. hansenii the population of the microorganisms was enumerated on MRS agar supplemented with 0.02% w/v sodium azide and 0.05%, w/v cyclohexamide, or on Plate Count agar (Merck) supplemented with 0.01%, w/v chloramphenicol (Sigma) and 1%, w/v chlortetracycline-hydrochloride (Sigma), respectively, both incubated as mentioned above.

160

S. Tsapatsaris, P. Kotzekidou / International Journal of Food Microbiology 95 (2004) 157168

Table 1 Process variables and levels corresponding to ionic strength values (I)a between 343 and 1710 of the brine (343 = the ionic strength of 343 mM or 2% NaCl solution; 1710 = the ionic strength of 1710 mM or 10% NaCl solution) during fermentation of olive juice broth by L. plantarum ATCC 8014 and D. hansenii Fermentation (strain) Process variables Symbol Level Coded x1 1 0 +1 1 0 +1 1 0 +1 1 0 +1 1 0 +1 1 0 +1 1 0 +1 1 0 +1 1 0 +1 1 0 +1 1 0 +1 1 0 +1 Uncoded (mM) 256.6 555.8 855 19.1 104.6 190 9.6 52.3 95 128.3/128.3 277.9/277.9 427.5/427.5 19.1 104.5 190 9.6 52.3 95 256.6 555.8 855 19.1 59.5 100 9.6 97.3 185 128.3/128.3 277.9/277.9 427.5/427.5 19.1 59.5 100 9.6 97.3 185

From each growth curve, the maximum specific growth rate (lm per hour) was obtained by fitting the viable cell numbers to the Gompertz equation modified by Zwietering et al. (1990). TableCurvek 2D software (SPSS, Chicago, IL) was used for generation of growth curves from experimental data. 2.5. Experimental design of maximum specific growth rate The two microorganisms (L. plantarum ATCC 8014 and D. hansenii 2114) were studied for maximum specific growth rate in olive juice using a central composite (face-centered) design with three variables and a response surface methodology (Neter et al., 1996). A central composite design was chosen as a 23 full factorial design with star point and six trials at the center. Based on preliminary studies, the processing variables (factors) NaCl or NaCl/KCl, calcium acetate and calcium lactate were chosen according to a central composite design. In this experimental design, there were three coded factor levels: 1, 0, + 1 in which 1 corresponds to the low level of each factor, 1 to the high level and 0 to the mid-level. The actual level of each factor was calculated using the following equation (Neter et al., 1996): Actual level Coded value*high level low level =2 high level low level=2 The factors and respective coded and uncoded levels are given in Table 1. The effect of the three independent variables on the maximum growth rate ( Y) was modeled using a polynomial response surface. The second-order response function for our experiments was predicted by the following equation: Y b0 b1 x1 b2 x2 b3 x3 b11 x2 b22 x2 1 2 b33 x2 b12 x1 x2 b13 x1 x3 b23 x2 x3 3 where x1, x2 and x3 represented the coded values of the factors according to Table 1, and the parameters b0, b1,. . .. . ., b23 are constant coefficients. The point where the response surface is at the maximum is denoted by the vector Xs = 1/2B 1b*, where the elements of the matrix B are the estimated coefficients at the second-order terms and the elements of the vector b* are the estimated coefficients of the first-

Fermentation 1 (by L. plantarum ATCC 8014)

NaCl

Ca-acetate

x2

Ca-lactate

x3

Fermentation 2 (by L. plantarum ATCC 8014)

NaCl/KCl

x4

Ca-acetate

x2

Ca-lactate

x3

Fermentation 3 (by D. hansenii)

NaCl

x1

Ca-acetate

x5

Ca-lactate

x6

Fermentation 4 (by D. hansenii)

NaCl/KCl

x4

Ca-acetate

x5

Ca-lactate

x6

I = 1/2Simi z i2, where mi the molality of the ith ion in mmol l 1 of the solute and zi its charge.
a

2.4. Determination of the growth rates For modeling the growth of food-fermenting organisms, the emphasis is exclusively on the growth conditions that allow growth to a high population density.

S. Tsapatsaris, P. Kotzekidou / International Journal of Food Microbiology 95 (2004) 157168

161

order terms (Neter et al., 1996). In the Pareto plot of standardised effects, the important effects are visually identified. The bars correspond to the absolute magnitudes of the estimated effect coefficients, and the vertical line indicates the minimum magnitude of statistically significant effects. 2.6. Mixed culture fermentation Using the optimized factor values of NaCl or the NaCl/KCl mixture, Ca-acetate and Ca-lactate for olive juice fermentation, the maximum specific growth rate of L. plantarum ATCC 8014 under different inoculation times was calculated, i.e. simultaneous inoculation with D. hansenii or inoculation of L. plantarum 12, 24 and 48 h after D. hansenii. 2.7. Analysis of data Data were analyzed using the STATISTICA software package (StatSoft, Tulsa, OK).

From each growth curve, the lag time, the asymptotic value (maximum population density) and the maximum specific growth rate can be estimated (Zwietering et al., 1990). In the case of olive fermentation, when starter cultures are applied, the minimum inoculum size for significant improvement of the fermentation process ranges from 105 to 107 cfu/ml (RuizBarba et al., 1994; Etchells et al., 1966). As favourable conditions during fermentation result in an accelerated growth phase, the maximum specific growth rate is the response that defines the fermentation process. The maximum specific growth rates estimated at the different fermentation conditions of olive juice by L. plantarum ATCC 8014 are presented in Table 2. In Fermentation 1, the maximum specific growth rate response to the combined effects of salt (x1), Caacetate (x2) and Ca-lactate (x3) (each at three levels and in the combinations as indicated in Table 2) in concentrations corresponding to ionic strength values between 343 and 1710 were studied at 30 jC and initial pH 5.0. The response was represented by the following polynomial quadratic equation, containing 10 estimated coefficients all significant at P < 0.05. Y1 0:1607 1:8 104 x1 2 104 x2 8:3 104 x3 1:8 107 x2 1 1:1 106 x2 5:4 106 x2 2 3 1:3 107 x1 x2 1 106 x1 x3 2:4 106 x2 x3 The R2 value of 0.995 and the F-test for the regression was significant at a level of 5% ( P < 0.05), whereas the lack of fit was not significant at the 5% level ( P>0.05). The derived model fit the experimental data. A Pareto chart of standardised effects, showed significant effects of all variables (linear and quadratic as well as of the interactions of the variables). The most important significant effect was due to the interaction of salt and Ca-lactate (Fig. 1), followed by salt (quadratic), the interaction of Ca-acetate and Ca-lactate, as well as salt and Ca-lactate (linear). The quadratic effect of Ca-lactate and Ca-acetate, the linear effect of Ca-acetate as well as the interaction of salt with Ca-acetate were also significant, but with lower absolute values (Fig. 1). The maximum value of the response (0.211 h 1) is observed at 376.5 mM NaCl, 32.3 mM Ca-acetate and 33.9 mM Ca-lactate as

3. Results and discussion 3.1. Optimization of olive juice broth fermentation by L. plantarum and D. hansenii To optimize the maximum specific growth rates of L. plantarum or D. hansenii during olive juice fermentation, response surface methodology was used. The process variables investigated during fermentations were salt, mixture of salt and KCl, Caacetate and Ca-lactate as indicated in Table 1. Utilization of Ca2 + in fermentation brines provides the potential for reducing losses caused by pectinolytic softening as well as reducing levels of NaCl used for fermentation (Buescher et al., 1981). In a study on the firmness of Spanish green olives it was observed that calcium displaced sodium from the cell wall structure to a great extent resulting in cell wall stabilization by two different ways: one by formation of coordination complexes, due only to calcium and the other by electrostatic means due to both calcium and sodium (Jimenez et al., 1997). Lactate and acetate are present in a mixture in fermented vegetables and synergistic antimicrobial properties are often observed.

162

S. Tsapatsaris, P. Kotzekidou / International Journal of Food Microbiology 95 (2004) 157168

Table 2 Experimental design, ionic strength values corresponding to each combination and data during fermentation of olive juice broth by L. plantarum ATCC 8014 incubated at 30 jC (initial pH: 5.0; inoculum: 3.0 log10 cfu ml 1) Treatment no. Factors NaCl or NaCl/KCl (x1 or x4) 1 +1 1 +1 1 +1 1 +1 1 +1 0 0 0 0 0 0 0 0 0 0
a b

Ca-acetate (x2)

Ca-lactate (x3)

Ionic strength (I)a

Responsesb Maximum specific growth rate, lm (h 1) Fermentation 1 (x1, x2, x3) Fermentation 2 (x4, x2, x3) 0.264 0.258 0.267 0.215 0.277 0.193 0.222 0.110 0.274 0.215 0.278 0.226 0.280 0.223 0.258 0.259 0.260 0.257 0.265 0.257

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20

1 1 +1 +1 1 1 +1 +1 0 0 1 +1 0 0 0 0 0 0 0 0

1 1 1 1 +1 +1 +1 +1 0 0 0 0 1 +1 0 0 0 0 0 0

343 941 855 1454 599 1197 1111 1710 726 1325 770 1283 898 1154 1026 1026 1026 1026 1026 1026

0.199 0.177 0.189 0.150 0.198 0.120 0.150 0.062 0.200 0.145 0.205 0.157 0.205 0.153 0.192 0.192 0.188 0.187 0.190 0.192

As indicated in Table 1. Determined from growth data by the modified Gompertz equation by Zwietering et al. (1990): y = Aexp{ exp [(lme/A) (k t) + 1]}.

indicated in Table 4. The response surface was used for the purpose of finding the combination of salt and Ca-lactate at the optimum Ca-acetate concentration of 32.3 mM that lead to an optimum maximum specific growth rate (Fig. 2a). As the concentrations of salt and

Fig. 1. Standardised effects of NaCl, Ca-acetate and Ca-lactate concentration on maximum specific growth rate of L. plantarum ATCC 8014 in olive juice fermentation.

Ca-lactate increase, the growth response increases to a maximum value and then decreases to a minimum at the combination of high concentrations of the variables. Similar stimulatory-inhibitory effects of NaCl were observed in cucumber extract for L. plantarum by Passos et al. (1993). In our experiment beyond the initial stimulatory effect, a further increase in the salt and Ca-lactate concentrations resulted in an inhibitory effect on the maximum specific growth rate of L. plantarum. The addition of lactate to MRS broth resulted in linear decrease in the growth rate of the culture, whereas all the metabolic activities were completely inhibited at 110 g/l lactate (Giraud et al., 1991). A delay in the growth of L. plantarum DSM 10492 was observed in an unconventional medium (pH 5.5) containing 4 and 6% w/v NaCl. Furthermore, at 8%, w/v NaCl there was no growth, but the survival of cells was not greatly reduced during 120 h of incubation. In the present study, L. plantarum ATCC 8014 grew vigorously in NaCl concentrations up to 376.5 mM. In low concentration, NaCl stimulated the

S. Tsapatsaris, P. Kotzekidou / International Journal of Food Microbiology 95 (2004) 157168

163

retaining the same values of the other variables (as indicated in Table 1, Fermentation 2) in order to have equivalent ionic strengths as in the treatments of Fermentation 1. The analysis of the overall data set indicated that all variables were significant while composite x4x3 and x2x3 variables showed the most pronounced effects on the response (Fig. 3). The variable NaCl/KCl (quadratic and linear) had also a significant effect on the maximum specific growth rate, but with lower absolute values (Fig. 3). The data obtained in Fermentation 2 under the combined effects of the three process variables (x4, x2, x3 as indicated in Table 1) are explained by the following second-order polynomial equation: Y2 0:2172 4 104 x4 3:5 104 x2 1 103 x3 7:2 107 x2 1:2 106 x2 4 2 5 106 x2 7:2 107 x4 x2 2:7 106 x4 x3 3 3:4 106 x2 x3 The R2 value of 0.995 and the F-test for the regression was significant at a level of 5% ( P < 0.05), whereas the lack of fit was not significant at the 5% level ( P>0.05), indicating the good predictability of the model. The maximum specific growth rate of 0.291 (h 1) is observed at 197.2/197.2 mM NaCl/KCl, 39.5 mM Ca-acetate and 35.3 mM Ca-lactate (Table 4). The above combination is a good substitution for NaCl from the microbiological point of view concerning the increase of the maximum specific growth rate of L. plantarum ATCC 8014 estimated in olive juice at 30

Fig. 2. Response surface plot for maximum specific growth rate of L. plantarum ATCC 8014 in olive juice fermentation as a function of: (a) NaCl and Ca-lactate concentration (Ca-acetate at the optimum concentration of 32.3 mM); (b) Ca-acetate and Ca-lactate concentration (NaCl at the optimum concentration of 376.5 mM).

growth of L. plantarum in olive juice, probably due to the small decrease in the water activity value as reported for Staphylococcus xylosus, which showed a higher specific growth rate in medium containing 3.5% NaCl compared with no salt (McMeekin et al., 1987). Low concentrations of Ca-acetate and Calactate act synergistically and increase the maximum specific growth rate of L. plantarum in olive juice as indicated in Fig. 2b. In an effort to further reduce the concentration of NaCl used in the previous fermentation, a 50% replacement of NaCl by KCl (x4) was undertaken,

Fig. 3. Standardised effects of NaCl/KCl, Ca-acetate and Ca-lactate concentration on maximum specific growth rate of L. plantarum ATCC 8014 in olive juice fermentation.

164

S. Tsapatsaris, P. Kotzekidou / International Journal of Food Microbiology 95 (2004) 157168

jC and initial pH 5.0 in comparison to the value estimated in Fermentation 1 (in which NaCl was partially substituted by Ca-acetate and Ca-lactate). As indicated in Table 1, the maximum level of Caacetate used in the central composite design during fermentation of olive juice by D. hansenii 2114 (Fermentations 3 and 4) was lower than in fermentation with L. plantarum. High concentrations of Caacetate were toxic for D. hansenii and no growth was observed in preliminary studies. Indeed higher concentrations of Ca-lactate were applied, in an effort to obtain ionic strengths equivalent to 2 10% NaCl without increasing the concentration of salt. The high concentrations of Ca-lactate did not affect the growth of the microorganism, because D. hansenii consumes lactate oxidatively as reported by Barnett et al. (1990). The maximum specific growth rate values estimated at the different fermentation conditions are presented in Table 3. The model developed had no significant lack of fit at the 5% level ( P>0.05), the R2 value of 0.994 and the F-test for the regression was significant at a

level of 5% ( P < 0.05). Therefore, it represents the data adequately. The data obtained in Fermentation 3 under the combined effects of the three process variables (x1, x5, x6 as indicated in Table 1) are explained by the following equation: Y3 0:1214 9:8 105 x1 1:2 103 x5 3:5 105 x6 6:7 108 x2 7:8 106 x2 1 5 1 106 x2 1:1 106 x1 x5 1:4 107 x1 x6 6 5:5 106 x5 x6 The linear coefficient sign was significant and the positive sign indicated increase of the maximum specific growth rate. The negative quadratic coefficient values indicated that treatment concentrations higher than the optimum values decrease the maximum specific growth rate. Similar results were obtained by Passos et al. (1997), working with Saccharomyces rosei in cucumber fermentation. Low concentrations (1 2%) of NaCl stimulated the

Table 3 Experimental design, ionic strength values corresponding to each combination and data during fermentation of olive juice broth by D. hansenii incubated at 30 jC (initial pH: 5.0; inoculum: 3.0 log10 cfu ml 1) Treatment no. Factors NaCl or NaCl/KCl (x1 or x4) 1 +1 1 +1 1 +1 1 +1 1 +1 0 0 0 0 0 0 0 0 0 0
a b

Ca-acetate (x5)

Ca-lactate (x6)

Ionic strength (I)a

Responsesb Maximum specific growth rate, lm (h 1) Fermentation 3 (x1, x5, x6) Fermentation 4 (x4, x5, x6) 0.181 0.191 0.189 0.132 0.179 0.175 0.097 0.019 0.185 0.155 0.198 0.122 0.190 0.133 0.178 0.180 0.176 0.178 0.179 0.184

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20

1 1 +1 +1 1 1 +1 +1 0 0 1 +1 0 0 0 0 0 0 0 0

1 1 1 1 +1 +1 +1 +1 0 0 0 0 1 +1 0 0 0 0 0 0

343 941 585 1184 868 1467 1111 1710 728 1327 906 1148 765 1291 1028 1028 1028 1028 1028 1028

0.159 0.160 0.158 0.099 0.163 0.144 0.079 0.010 0.155 0.128 0.168 0.102 0.165 0.114 0.149 0.154 0.151 0.143 0.145 0.146

As indicated in Table 1. As indicated in Table 2.

S. Tsapatsaris, P. Kotzekidou / International Journal of Food Microbiology 95 (2004) 157168

165

growth of S. rosei, probably due to the small decrease in the water activity value (Passos et al., 1997). Beyond the initial stimulatory effect, a further increase in the NaCl concentration resulted in an inhibitory effect on the specific growth rate of S. rosei (Passos et al., 1997). Composite Ca-acetate Ca-lactate and NaCl Ca-acetate variables showed the most pronounced effects on the response (Fig. 4). The maximum value of the specific growth rate (0.17 h 1) was observed at 380.2 mM NaCl, 35.8 mM Caacetate and 45.8 mM Ca-lactate (Table 4). Fig. 5 shows that increasing concentrations of Caacetate, Ca-lactate and the mixture of NaCl/KCl had a positive influence on maximum specific growth rate until an optimum value, whereas high concentrations of the above variables had a significantly negative influence on the response. The interaction effect between Ca-acetate and Ca-lactate was negative and had the most significant effect on maximum specific growth rate. This means that raised concentrations of Ca-acetate had a greater negative impact on L. plantarum growth at high Ca-lactate concentrations than at low Ca-lactate concentrations. The same phenomenon was seen for the interaction between NaCl/KCl and Ca-acetate. At high NaCl/KCl concentrations, Caacetate had a negative impact, as opposed to low NaCl/KCl concentrations. The impact of the above interaction effects is illustrated in Fig. 6a,b, which shows maximum specific growth rate as a function of Ca-acetate, Ca-lactate and the mixture NaCl/KCl. It is clearly seen in Fig. 6 that there is strong synergy between Ca-acetate and Ca-lactate, as well as between

Table 4 Optimum values for process variables during fermentation of olive juice broth by L. plantarum ATCC 8014 and D. hansenii Fermentation (strain) Fermentation 1 (by L. plantarum ATCC 8014) Fermentation 2 (by L. plantarum ATCC 8014) Fermentation 3 (by D. hansenii) Fermentation 4 (by D. hansenii) Process variables NaCl Ca-acetate Ca-lactate NaCl/KCl Ca-acetate Ca-lactate NaCl Ca-acetate Ca-lactate NaCl/KCl Ca-acetate Ca-lactate Symbol x1 x2 x3 x4 x2 x3 x1 x5 x6 x4 x5 x6 Optimum value 376.5 32.3 33.9 197.2/197.2 39.5 35.3 380.2 35.8 45.8 199.7/199.7 39.9 50.1

NaCl/KCl and Ca-acetate. The response was represented by the following polynomial quadratic equation, containing 10 estimated coefficients all significant at P V 0.05. Y4 0:1313 2:2 104 x4 1:6 103 x5 5 104 x6 2:1 107 x2 9:2 106 x2 4 5 1:7 106 x2 2:9 106 x4 x5 6 3:3 107 x4 x6 6:6 106 x5 x6 The R2 value of 0.997 and the F-test for the regression was significant at a level of 5% ( P < 0.05), whereas the lack of fit was not significant at the 5% level ( P>0.05), indicating how well the derived model fit the experimental data. The maximum specific growth rate of 0.206 (h 1) was observed at 199.7/199.7 mM NaCl/

Fig. 4. Standardised effects of NaCl, Ca-acetate and Ca-lactate concentration on maximum specific growth rate of D. hansenii in olive juice fermentation.

Fig. 5. Standardised effects of NaCl/KCl, Ca-acetate and Ca-lactate concentration on maximum specific growth rate of D. hansenii in olive juice fermentation.

166

S. Tsapatsaris, P. Kotzekidou / International Journal of Food Microbiology 95 (2004) 157168

3.2. Effect of mixed culture on maximum specific rate of L. plantarum and D. hansenii The predictive models for growth of L. plantarum and D. hansenii are useful in modeling the mixed culture fermentation of olive juice. Taking into account the mean optimum values of the process variables determined from the models applied during Fermentations 1 and 3, a mixed culture fermentation was carried out in olive juice broth containing 378.4 mM NaCl, 34.1 mM Ca-acetate and 39.9 mM Ca-lactate to determine the maximum specific growth rate of L. plantarum ATCC 8014 and the final pH values obtained at 30 jC and initial pH 5.0 (Table 5). In these experiments the time of inoculation of each species was varied. Varying the time of inoculation of the yeast and L. plantarum had a significant effect on the maximum specific growth rate of L. plantarum. When D. hansenii was inoculated 48 h before L. plantarum, a significant increase of the maximum specific growth rate of L. plantarum was observed as indicated in Table 5 in comparison to 0.211 h 1 observed when L. plantarum is cultivated alone. The
Table 5 Maximum specific growth rate valuesa of L. plantarum ATCC 8014 during fermentation of olive juice brothb incubated at 30 jC and final pH (initial pH 5.0; inoculum level: L. plantarum 3.0 log10 cfu ml 1; D. hansenii 3.0 log10 cfu ml 1) D. hansenii inoculum added Fermentation carried out with NaCl, Ca-acetate, Ca-lactate loptc Fig. 6. Response surface plot for maximum specific growth rate of D. hansenii in olive juice fermentation as a function of: (a) Ca-acetate and Ca-lactate concentration (NaCl/KCl at the optimum concentration of 199.7/199.7 M); (b) NaCl/KCl and Ca-acetate concentration (Ca-lactate at the optimum concentration of 50.1 mM). 48 h before L. plantarum 24 h before L. plantarum 12 h before L. plantarum Same time as L. plantarum L. plantarum only
a b

Fermentation carried out with NaCl/KCl, Ca-acetate, Ca-lactate lopt Final pH 4.64 4.62 4.46 4.32 3.95

Final pH 4.56 4.52 4.34 4.17 3.8

0.247 0.211 0.208 0.200 0.211

0.218 0.204 0.198 0.176 0.291

KCl, 39.9 mM Ca-acetate and 50.1 mM Ca-lactate (Table 4). The above combination is a good substitution for NaCl from the microbiological point of view concerning the increase of the maximum specific growth rate of D. hansenii estimated in olive juice at 30 jC and initial pH 5.0 as a higher maximum specific growth rate was obtained in comparison to that of Fermentation 3 (where NaCl was partially substituted by Ca-acetate and Ca-lactate).

Values are means of triplicate fermentations. Olive juice broth was supplemented with the average of the optimum values, i.e. for the fermentation carried out with NaCl: NaCl 378.4 mM, Ca-acetate 34.1 mM, Ca-lactate 39.9 mM and for the fermentation with 50% replacement of NaCl by KCl: NaCl/KCl 198.5/198.5 mM, Ca-acetate 39.7 mM, Ca-lactate 42.7 mM). c lopt equals lm at the growth conditions with the combination of factors at the average optimum values.

S. Tsapatsaris, P. Kotzekidou / International Journal of Food Microbiology 95 (2004) 157168

167

yeast strain does play a role in encouraging the growth of L. plantarum. As reported by Ruiz-Barba and Jimenez-Diaz (1995), some yeast strains are able to produce large amounts of nicotinic and pantothenic acids, biotin, or vitamin B6, which are essential for growth of L. plantarum in olive brines. The high maximum specific growth rate obtained indicates that the microbial instability of olives fermented under reduced salt concentrations can be overcome. In addition, the final pH of the medium was reduced with a higher final pH (4.17), when D. hansenii was added with L. plantarum, as compared to L. plantarum alone (3.8). A similar pattern was observed when S. rosei was added with L. plantarum in cucumber juice (Daeschel et al., 1988). This would have a positive effect on the colour of the fermented whole olives, as an acidic environment results in alteration of the natural black colour of the olives. In the case of mixed culture fermentation carried out under the mean optimum values of Fermentations 2 and 4 (i.e. NaCl/KCl, Ca-acetate, Ca-lactate), the maximum specific growth rate of L. plantarum as indicated in Table 5 was lower compared to L. plantarum alone (0.291 h 1) under the same fermentation conditions, which correlates with the higher terminal pH values observed. During the processing of green table olives, a brine containing CaCl2 and KCl was used, which allowed a product with good organoleptic characteristics to be obtained and permitted the olives to lose their bitter taste (Mule et al., 2000).

fermentation of olive juice is affected by the variation of the time of inoculation of the yeast strain. When D. hansenii is inoculated 48 h before L. plantarum, the growth rate of L. plantarum increases significantly in comparison to the simultaneous inoculation of both microorganisms.

Acknowledgements We are grateful to J.L. Ruiz-Barba of the Departamento de Biotechnologia de Alimentos, Instituto de la Grasa (C.S.I.C.), Seville, Spain, for providing the strain of L. plantarum ATCC 8014.

References
Barnett, J.A., Payne, R.W., Yarrow, D., 1990. Yeasts: Characteristics and Identification. C.U. Press. Press Syndicate of the University of Cambridge, Cambridge. Buescher, R.W., Hudson, J.M., Adams, J.R., 1981. Utilization of Calcium to reduce pectinolytic softening of cucumber pickles in low salt conditions. Lebensm.-Wiss. Technol. 14, 65 69. Daeschel, M.A., Fleming, H.P., McFeeters, R.F., 1988. Mixed culture fermentation of cucumber juice with Lactobacillus plantarum and yeast. J. Food Sci. 53, 862 864. Duran Quintana, M.C., Garcia Garcia, P., Garrido Fernandez, A., 1999. Establishment of conditions for green table olive fermentation at low temperature. Int. J. Food Microbiol. 51, 133 143. Etchells, J.L., Borg, A.F., Kittel, I.D., Bell, T.A., Fleming, H.P., 1966. Pure culture fermentation of green olives. Appl. Microbiol. 14, 1027 1041. Giraud, E., Lelong, B., Raimbault, M., 1991. Influence of pH and initial lactate concentration on the growth of Lactobacillus plantarum. Appl. Microbiol. Biotechnol. 36, 96 99. Guillou, A.A., Floros, J.D., 1993. Multiresponse optimization minimizes salt in natural cucumber fermentation and storage. J. Food Sci. 58, 1381 1389. Jimenez, A., Heredia, A., Guillen, R., Fernandez Bolanos, J., 1997. Correlation between soaking conditions, cation content of cell wall, and olive firmness during Spanish green olive processing. J. Agric. Food Chem. 45, 1653 1658. Kotzekidou, P., 1997. Identification of yeasts from black olives in rapid system microtitre plates. Food Microbiol. 14, 609 616. Leone, F.G., 2000. The globalisation of the olive oil market and the competitive position of the sector in Italy: an international comparison. Olivae 83, 10 14. Marsilio, V., Lanza, B., 1998. Characterisation of an oleuropein degrading strain of Lactobacillus plantarum. Combined effects of compounds present in olive fermenting brines (phenols, glucose and NaCl) on bacterial activity. J. Sci. Food Agric. 76, 520 524. Marsilio, V., Lanza, B., Pozzi, N., 1996. Progress in table olive

4. Conclusions Central composite design and response surface methodology can be used for the purpose of finding the combination of the process variables (salt or mixture of salt and KCl, calcium acetate and calcium lactate) that lead to an optimum maximum specific growth rate of L. plantarum and D. hansenii during olive juice fermentation. In the case of 50% replacement of NaCl by KCl, there is strong synergy between Ca-acetate and Ca-lactate and a higher maximum specific growth rate of single cultures of both microorganisms is estimated in comparison to fermentations where NaCl is partially substituted by Ca-acetate and Ca-lactate. The application of the predictive models for growth of the single cultures in mixed culture

168

S. Tsapatsaris, P. Kotzekidou / International Journal of Food Microbiology 95 (2004) 157168 Passos, F.V., Fleming, H.P., Felder, R.M., Ollis, D.F., 1997. Modeling growth of Saccharomyces rosei in cucumber fermentation. Food Microbiol. 14, 533 542. Pitzer, K.S., 1995. Electrolyte solutions: introduction and simple examples. Thermodynamics. McGraw-Hill, New York, pp. 254 274. Quintana, M.C.D., Barranco, C.R., Garcia, P.G., Balbuena, M.B., Fernandez, A.G., 1997. Lactic acid bacteria in table olive fermentations. Grasas Aceites 48, 297 311. Roebuck, K., Brundin, A., John, M., 1995. Response surface optimization of temperature and pH for the growth of Pachysolen tannophilus. Enzyme Microb. Technol. 17, 75 78. Ruiz-Barba, J.L., Jimenez-Diaz, R., 1995. Availability of essential B-group vitamins to Lactobacillus plantarum in green olive fermentation brines. Appl. Environ. Microbiol. 61, 1294 1297. Ruiz-Barba, J.L., Cathcart, D.P., Warner, P.J., Jimenez-Diaz, R., 1994. Use of Lactobacillus plantarum LPCO10, a bacteriocin producer, as a starter culture in Spanish-style green olive fermentations. Appl. Environ. Microbiol. 60, 2059 2064. Zwietering, M.H., Jongenburger, I., Rombouts, F.M., vant Riet, K., 1990. Modeling of the bacterial growth curve. Appl. Environ. Microbiol. 56, 1875 1881.

debittering: degradation in vitro of oleuropein and its derivatives by Lactobacillus plantarum. J. Am. Oil Chem. Soc. 73, 593 597. McMeekin, T.A., Chandler, R.E., Doe, P.E., Garland, C.D., Olley, J., Putro, S., Ratkowsky, D.A., 1987. Model for combined effect of temperature and salt concentration/water activity on the growth rate of Staphylococcus xylosus. J. Appl. Bacteriol. 62, 543 550. Mule, R., Fodale, A.S., Bati, C.B., Tucci, A., Di Pisa, A., 2000. Preliminary results of a new processing in order to obtain green table olives with a low sodium content. Industrie Alimentaria 39, 844 847. Naewbanji, J.O., Stone, M.B., Chambers, E.I.V., 1990. Lactobacillus plantarum and Enterobacter cloacae growth in cucumber extracts containing various salts. J. Food Sci. 55, 1634 1637. Neter, J., Kutner, M.H., Nachtsheim, C.J., Wasserman, W., 1996. Applied Linear Statistical Models, 4th ed. McGrawHill, Chicago. Ozay, G., Borcakli, M., 1996. Effect of brine replacement and salt concentration on the fermentation of naturally black olives. Food Res. Int. 28, 553 559. Passos, F.V., Fleming, H.P., Ollis, D.F., Hassan, H.M., Felder, R.M., 1993. Modeling the specific growth rate of Lactobacillus plantarum in cucumber extract. Appl. Microbiol. Biotechnol. 40, 143 150.