Red Wine and Model Wine Astringency as Affected by Malic and Lactic Acid S. KALLITHRAKA, J. BAKKER, and M.N. CLIFFORD ABSTRACT Red wines and model solutions containing grape seed phenols were assessed for perceived astringency, bitterness and sourness by a panel of trained assessors using time-intensity (T-I) methodology. The effect of lactic and malic acid on perception of these attributes was evaluated at two pH values, by adding different amounts of acids. Maximum intensity (Imax) and total duration (Ttot) of astringency increased with decreasing pH in model solutions and red wine, but no differences were found between malic and lactic acid on perception of astringency. The pH and the acid type did not affect any of the bitterness T-I parameters. However, sourness showed an increase in Imax and Ttot with decreasing pH. Key Words: astringency, bitterness, sourness, grape seed phenols, timeintensity INTRODUCTION ASTRINGENCY is an essential characteristic of red wine produced by tannins which affect the perceived ‘mouthfeel’. Astringency is considered to be the dry and puckery sensation and is chemically defined as the ability to precipitate proteins (Bate-Smith, 1954). It is a tactile rather than gustatory stimulus, believed to be caused primarily by increased friction between the oral membranes (Breslin et al., 1993). The intensity of astringency increases with repeated stimulation rather than exhibiting a decreasing pattern, typical of adaptation (Guinard et al., 1986a). Binding between tannins and proteins has been largely attributed to multiple hydrogen bond formation between orthodihydroxylphenyl groups of tannins and carbonyl groups of protein peptides (Haslam, 1974). There is also evidence of hydrophobic interactions between the aromatic ring structure of phenolic compounds and hydrophobic regions of proteins (Hagerman and Butler, 1981) and of covalent interactions above pH 9.0. Ionic bonds between the cation site of the protein molecule and the phenolate anion might also be involved, depending on pH of the solution and pI of the protein (Clifford, 1986). However, hydrogen bonding and hydrophobic interactions are hypothesized to be the most likely mechanisms of interaction between phenol and protein under physiological conditions (Clifford, 1986). NMR studies have established that proline residues in synthetic proline-rich peptides form a good binding site, providing an open, rigid, hydrophobic surface favorable for association with hydrophobic molecules, such as the aromatic rings of phenols (Murray et al, 1994). Although sourness is the predominant sensation of organic acids, dryness or astringency of acids has also been reported (Lee and Lawless, 1991; Rubico and McDaniel, 1992). Astringency of acids is attributed either to the direct contribution of H+ ions or to the hydrogen bonding capabilities of the hydroxyl groups on the anion or undissociated acid (Lawless et al., 1994). Five organic acids and one inorganic elicited astringency and astringent subqualities (Thomas and Lawless, 1995). In order to explain the sensory properties of organic acids they suggested that acids without hydroxyl groups might act by other mechaAuthors Kallithraka and Bakker are with the Institute of Food Research, Earley Gate, Whiteknights Road, Reading RG6 6BZ, U.K. Author Clifford is with the University of Surrey, School of Biological Science, Guildford, Surrey GU2 5XH, U.K. nisms such as the denaturation of proteins in the saliva or direct attack on the mucous layer and oral epithelium. The presence of acids in astringent media has been shown to affect the intensity of perceived astringency. According to Amerine (1980), the astringent sensation in wine was more pronounced in the presence of a moderate amount of tartaric acid along with the polyphenolic compounds. Guinard et al. (1986b) showed that acidity increased astringency of model solutions and white wines. They found a linear increase in intensity of astringency as a function of pH reduction (from 3.75 to 2.59). Astringency elicited by phenolic compounds in wine could be affected by organic acids naturally present. The contribution of malic and lactic acids to sensory attributes of wine is of great interest. In many red wines the malo-lactic fermentation is considered desirable, involving the decarboxylation of malic to lactic acid and resulting in a decrease of wine acidity. Our objective was to apply temporal methology to evaluate the interactions between acidity and astringency, bitterness and sourness in wine and in a model wine solution. The effects of malic and lactic acids (which differ only in the number of carboxyl groups) at different pH values were also determined. MATERIALS & METHODS Subjects Twelve healthy subjects (female) were paid to participate as part of a trained taste panel. All were experienced in sensory assessments of a range of foods and had used the T-I method for evaluation of sensory attributes. Preparation of samples A model solution of grape seed extract in 300 mg/L potassium bitartrate solution containing 10% v/v ethanol was prepared. Phenolic compounds were extracted from grape seeds of Vitis vinifera cv. Roriz grapes (Douro valley, Portugal). Seeds (72g) were crushed and immediately extracted for 5 min with 800 mL 75% ethanol (Hayman LTD). After 10 min the seeds were removed using a Buchner funnel (filter paper Whatman No. 4). The filtrate was stored at 2207C. The morning before the experiment, 266.7 mL were thawed and diluted to 2L with a solution of 346.2 mg/L potassium bitartrate (Peter Whiting Chemicals) in deionized water. The pH of 400 mL of the model solution was reduced from 3.8 to 3.5 and 3.2 by the addition of 300 mg/L and 525 mg/L of malic acid respectively, or by addition of 250 mg/L and 450 mg/L lactic acid respectively. The remaining 400 mL of solution was used as a control without addition of acid (Table 1). We used an experimental wine, made from Tinta Roriz grapes from the Douro valley (Portugal). The normal red wine making procedure was used, and 75 mg/L SO2 was added to the crushed grapes prior to ferTable 1—Description of model and wine samples Wine Model Sample description pH Titrable aciditya 1 2 3 4 5 3.8 3.5 3.2 3.5 3.2 0.11 0.42 0.78 0.36 0.70 Control After addition After addition After addition After addition of of of of malic acid malic acid lactic acid lactic acid a Acidity expressed as g tartaric acid/100 mL. 416—JOURNAL OF FOOD SCIENCE—Volume 62, No. 2, 1997 pH Titrable aciditya 4.0 3.7 3.5 3.7 3.5 0.70 1.18 1.80 1.01 1.60 Table 2—Chemical analysis of wine and model solution Red wine pH Titratable acidity (g/100 mL as tartaric acid) Total phenols (absorbance units at 280 nm) Total pigments (absorbance at 520 nm) Organic acids (mg/L) Tartaric Malic Succinic Lactic Acetic Noncolored phenols Gallic acid (mg/L) 3,4-dihydroxy-benzoic acid (mg/L gallic acid) 4-hydroxyphenethyl alcohol (mg/L gallic acid) cis-caftaric acid (mg/L gallic acid) (1)-catechin (mg/L) Vanillic acid (mg/L gallic acid) Caffeic acid (mg/L gallic acid) Syringic acid (mg/L gallic acid) p-coumaric acid (mg/L gallic acid) (2)-epicatechin (mg/L catechin) Myricetin (mg/L catechin) Quercetin (mg/L catechin) Kaempferol (mg/L catechin) Isorhamnetin (mg/L catechin) Procyanidin B1 (mg/L) Procyanidin B2 (mg/L) Procyanidin C1 (mg/L) Epicatechin gallate (mg/L catechin) 4.0 0.70 46.4 5.4 2340 44 1510 2102 416 6.3 1.9 15.4 0.2 7.2 1.7 12.2 6.3 2.6 41.4 22.1 45.7 8.1 3.0 0.0 0.0 0.0 0.0 Model solution 3.8 0.11 12.0 2264 0 0 80 37 0.7 0.0 0.0 0.0 7.7 0.0 0.0 0.0 0.0 8.5 0.0 0.0 0.0 0.0 7.6 9.5 21.9 3.0 assessed samples. To overcome the build-up of astringent sensation over time and to balance the effects of order of presentation, samples were served using a balanced block design (MacFie et al., 1989). Sample presentation and assessment. Judges were presented with 10 mL samples at room temperature in 30 mL plastic cups, randomly coded with three digits. They were asked to place the sample in the mouth at exactly the start of assessment, to swirl the sample for 15 s in the mouth and then to expectorate. Sampling and expectorating intervals were signalled orally by the experimenter. A 4-min break was taken between samples, during which they were required to eat a cracker and rinse the mouth thoroughly with spring water. A computerized T-I method was used in which the three attributes were rated by manipulating a marker on an unstructured line scale of 150 mm length, anchored at each end by 0 5 none and 100 5 extreme using the mouse. Data collection was performed via TASTE software. Judges rated the intensity of the specified attribute continuously over time from tasting the sample, through expectoration after 15s to the end of assessment after 2 min. Data analysis. Time to maximum intensity (Tmax), total duration (Ttot) (both measured in sec) and maximum intensity (Imax) (mm on a 150 mm line scale) for astringency, bitterness and sourness were extracted from T-I curves which were plotted using Genstat Software. Each variable was analyzed by Genstat using analysis of variance in which judges were treated as a random effect. The least significant differences (LSD) were calculated at p , 0.05. Chemical analysis of the wine and model solution mentation. To increase the wine astringency, 600 mg/L (1)-catechin (Senn chemicals AG) was added. The pH (4.0) of two aliquots of 400 mL was reduced to 3.7 and 3.5 by addition of 1100 mg/L and 2575 mg/L of malic acid respectively. Concentrations of 775 mg/L and 1525 mg/L of lactic acid were added to two other samples to reduce the pH to 3.7 and 3.5. A sample without added acids was used as control (Table 1). Training All subjects had participated in a previous study, involving astringency, bitterness and sourness in wines and model wines using the scalar method. In addition the panel had received extensive T-I training during previous experiments not related to astringency. Two training sessions were conducted prior to this experiment. During the first training, panelists were presented with aqueous solutions of 1500 mg/L tannic acid as an example of astringency. They were trained to differentiate astringency from bitterness and sourness using 1000 mg/L caffeine and 1500 mg/L citric acid as reference standards (Peter Whiting Chemicals). All panelists indicated their ability to discriminate between these taste stimuli. The control and low pH samples of the model solution and the wine were presented to subjects in order to give them an experience of the concentrations of the stimulus they would assess. The panelists were asked to taste the samples and to suggest terms which describe the sensory attributes of these samples. Astringency, bitterness and sourness were selected to describe the samples and hence they were used to assess the samples by the T-I method in the experiment. During the second training session, panelists who were already familiar with the computerized system for recording T-I information were asked to rate the above samples for the three selected attributes in order to practice use of the technique for these specific attributes. The experimental conditions during this session were exactly the same as in the six sets of the study. Recorded information was checked for each panelist to ensure they understood instructions. Experimental design Six sets (three replications of model solution and three for wine) were assessed during the study, which lasted for a month (including the training sessions). Tests were conducted 2 days/wk from 10:00 am to 1:00 pm in individual booths under incandescent light. In each set five samples (two concentrations of each of the two acids and one control) were evaluated for three attributes: astringency, bitterness and sourness. Judges were divided into two groups of six, the maximum that could be served at any one time, but data were treated as from one panel. Each group of panelists participated in the experiment for 30 min and then had a break for the following 30 min, during which the other group Chemical analysis of the wine and the model solution included pH, titratable acidity, total phenols, organic acid analysis and analysis of noncolored phenols. Determination of total pigments was done only for the wine. The pH was measured with a glass electrode of a Beckman digital pH meter, model 3500, that had been standardized to pH 4.00 and 7.00 with standard buffer solutions. Acidity was determined according to Ough and Amerine (1988), and expressed as g/100 mL tartaric acid. Total phenols and total pigments were determined by spectrophotometer at 280 and 520 nm, by measuring absorbance on a 1003 dilution in 0.1M HCl in a 10 mm cell (Bakker and Timberlake, 1985; Bakker et al., 1986). Concentrations of organic acids were measured by ion chromatography (Dionex 4500 model) with chemical eluent suppression (2 mM octane sulphonic acid in 2% isopropanol as eluent, anion micro membrane as suppressor and 5 mM tetra-n-butylammonium hydroxide as regenerator) and conductivity detection (conductivity cell). Samples were membrane filtered (0.45µ), phenols were adsorbed on a C18 Sep-Pak cartridge and eluted samples were diluted 25-fold in water. A 50 µL prepared sample was injected on a single ion-exclusion column (ASI Dionex) and eluted using a flow rate 0.8 mL/min. Noncolored phenolic compounds were analyzed by HPLC (HP 1090M model) with an auto injector (25 µL injection volume) and a diode array detector at 280 nm. A reversed-phase ODS Hypersil column (100 mm 3 2.1 mm; particle size 5 µm) at 407C was used, with a flow rate of 0.3 mL/min. Using 0.6% aqueous perchloric acid and methanol as eluants the following linear gradient was used: in 30 min (for model solution) and 50 min (for wine) from 20% to 50% methanol, in 1 min to 98% methanol, hold for 3 min at 98% methanol to wash the column and then returned to initial conditions to re-equilibrate for 10 min. Concentrations of phenolic acids were expressed as mg/L gallic acid. The concentration of (1)-catechin, (2)-epicatechin, epicatechin gallate, myricetin, quercetin, kaempferol and isorhamnetin were expressed as mg/L (1)catechin. B1, B2 and C1 were expressed as mg/L. The concentrations were calculated using external standards of gallic acid (60 mg/L) (Sigma) (1)-catechin (63 mg/L) (Sigma), B1 (65 mg/L), B2 (62 mg/L) and C1 (64 mg/l) (all the procyanidins were prepared in the laboratory) in 95% 0.6% aqueous perchloric acid and 5% methanol. RESULTS & DISCUSSION Chemical analysis of wine and model solution Results of the chemical analysis of the model solution and wine were compared (Tables 1 and 2). Using HPLC, the monomeric phenolic compounds and the procyanidin dimers and trimers were analyzed. The composition of the non-colored phenols showed lower concentrations of gallic acid, (1)-catechin and (2)-epicatechin in the grape seed extract than in the wine. Procyanidin dimers B1, B2 and trimer C1 were present in low concentrations in the model solution, but absent in the wine. However, the anthocyanins and flavan-3-ols in red wines are Volume 62, No. 2, 1997—JOURNAL OF FOOD SCIENCE—417 ACIDS AND ASTRINGENCY IN WINES . . . Table 3—F ratios and significance levels (sig.) for the Time-Intensity parameters [time to maximum intensity (Tmax), maximum intensity (Imax) and total duration (Ttot)] of astringency, bitterness and sourness for all sources of variation in the model wine solution (degrees of freedom : assessors511; samples54; assessors 3 samples544) Sources of variation Assessors Tmax Ttot Imax Samples Tmax Ttot Imax Assessors 3 samples Tmax Ttot Imax Astringency F sig. F Bitterness sig. F Sourness sig. 11.48 5.81 11.49 ,0.001 ,0.001 ,0.001 9.61 2.75 10.67 ,0.001 ,0.001 ,0.001 9.84 3.29 17.71 ,0.001 ,0.001 ,0.001 0.57 5.84 33.12 0.685 ,0.001 ,0.001 0.96 2.00 0.31 0.437 0.112 0.869 2.03 7.73 37.89 0.106 ,0.001 ,0.001 1.41 1.14 1.16 0.073 0.286 0.266 0.80 1.33 1.18 0.797 0.117 0.239 0.94 1.60 1.31 0.585 0.024 0.127 Table 4—Least significant differences (LSD) * P,0.05 (degrees of freedom54,44) and mean values for the T-I parameters for astringency, bitterness and sourness of : model wine solution (pH 3.8) (sample 1), model solution with added malic acid (pH 3.5 and 3.2) (samples 2 and 3 respectively) and model solution with added lactic acid (pH 3.5 and 3.2) (samples 4 and 5 respectively). Tmax and Ttot were measured as seconds and Imax as mm in a 150 mm line scale T-I parameters Sample 1 Sample 2 Sample 3 Sample 4 Sample 5 3.2 3.5 3.5 3.2 3.8 pH lactic LSD lactic malic malic control acid Tmax Ttot Imax 19.94a 39.06a 35.39a 21.06a 46.56b 54.14b Tmax Ttot Imax 23.83a 57.39a 60.08a 21.25a 48.47a 56.19a Tmax Ttot Imax 17.42a 37.42a 27.89a 21.06a 52.58b 58.06b Astringency 18.94a 20.83a 52.98c 45.69b 73.42c 53.75b Bitterness 21.69a 22.03a 49.08a 50.56a 55.28a 59.11a Sourness 18.86a 21.06a 56.39b 50.31b 66.19c 55.79b Table 5—F ratios and significance levels (sig.) for the Time-Intensity parameters [time to maximum intensity (Tmax), maximum intensity (Imax) and total duration (Ttot)] of astringency, bitterness and sourness for all sources of variation in the wine (degrees of freedom: assessors511; samples54; assessors 3 samples544) 19.61a 51.17bc 78.50c — 6.38 8.57 20.86a 46.92a 56.36a — — — 20.06a 56.89b 71.78c — 8.12 7.94 a-c Means with same letter in each row are not significantly different at p,0.05. ** LSD only provided where means differed p,0.05. known to polymerize (Bakker et al., 1986). HPLC analysis of the wine and the model solution showed a broad, late eluting band present in the HPLC chromatograms. These humps in the visible trace of the chromatogram of a red wine have been attributed to polymerized anthocyanins (Bakker et al., 1986), while such broad late eluting bands in the grape seed model solution were attributed to polymeric procyanidins (Ricardo da Silva et al., 1991) which were not identified or quantified. The total phenol measurement reflects the concentrations of all phenolic compounds determined by HPLC, in addition to a contribution made by polymerized compounds. Moreover, anthocyanins present in red wine but absent from grape seed extract contribute to the wine total phenols. This value was within the range reported by Somers and Evans (1977), of 20 to 100 total phenols as absorbance units. The sensory sensations in the model wine solutions were expected to be due to the dimeric, trimeric and polymeric procyanidins present above the reported threshold concentration of 4.1 mg/L for a mixture of trimeric and tetrameric procyanidins in water (Delcour et al., 1984). The other monomeric compounds were present at low concentrations and were not expected to make a notable contribution. For wine, the main contributors to sensory properties were the polymerized phenolic compounds, in addition to the added 600 mg/L (1)-catechin. Sensory studies (Vérétte et al., 1988) have shown that the contribution of hydroxycinnamates present in wine make no direct contribution to its astringency. The contribution of the other compounds to its sensory properties still remains to be studied. The titratable acidity of wine was higher than that of the model (Table 1), due to naturally occurring acids present in the Sources of variation Assessors Tmax Ttot Imax Samples Tmax Ttot Imax Assessors 3 samples Tmax Ttot Imax Astringency F sig. F Bitterness sig. F Sourness sig. 23.07 9.81 10.01 ,0.001 ,0.001 ,0.001 28.61 7.45 11.87 ,0.001 ,0.001 ,0.001 15.09 8.36 20.38 ,0.001 ,0.001 ,0.001 1.17 3.04 14.22 0.336 ,0.001 ,0.001 0.45 0.73 0.50 0.771 0.578 0.735 0.75 5.83 24.01 0.564 ,0.001 ,0.001 1.33 1.13 1.61 0.114 0.302 0.023 0.66 0.62 1.15 0.944 0.966 0.271 1.15 0.83 0.80 0.276 0.758 0.798 wine (see Table 2). As expected, samples adjusted with malic acid showed a higher titratable acidity than those adjusted with lactic acid, reflecting the larger concentrations of malic acid added to adjust pH. Model solution The significant F ratios after analysis of variance (Table 3), showed highly significant differences for Ttot and Imax for astringency and sourness. There were no significant differences for Tmax for any of attributes. The means of T-I parameters (Table 4, showed significant increases in Imax and Ttot for astringency as a function of pH reduction. Time to maximum intensity of astringency (Tmax) remained unaffected by pH reduction. At any given pH value there were no differences in astringency as a result of acid type. The observed increase in intensity of astringency with pH reduction confirms the findings of Guinard et al. (1986b). They found a linear increase in astringency intensity as a function of pH reduction. In our experiment for the model solution samples we also found a linear increase in astringency Imax for both malic acid (r 5 1.000, 1 d.f., P , 0.001) and lactic acid (r 5 0.996, 1 d.f., P , 0.02) acid, although we only had 3 measurement points. Total duration of astringency Ttot also showed a linear increase as a function of pH reduction (r 5 0.999, 1 d.f., P , 0.01 for malic acid samples and r 5 0.998, 1 d.f., P , 0.02 for lactic acid samples). There were no significant differences for bitterness. Binding of phenolic compounds to salivary proteins may prevent their binding to bitter taste receptors. It has been demonstrated (Kock et al., 1994), that transgenic mice with proline rich proteins (PRP) secreted by Von Ebner’s gland, had much greater tolerance against bitterness than mice lacking those proteins. This suggested that PRP transported the bitter compound away from the bitter receptor. Imax scores increased significantly for sourness related to pH reduction in agreement with Fischer and Noble (1994). Ttot of this attribute also increased significantly with decreasing pH from 3.8 to 3.5. Tmax remained unaffected by pH changes. Again there were no significant differences between the two acids at the same pH. Wine Significant F ratios after analysis of variance were compared (Table 5), and showed highly significant differences for Ttot and Imax for astringency and sourness. There were no significant differences for Tmax for any of the attributes. Means of T-I parameters were compared (Table 6). For astringency, there were significant increases in Imax and to a lesser extent for Ttot related to pH reduction. A similar pH reduction did not have a significant effect on astringency in red 418—JOURNAL OF FOOD SCIENCE—Volume 62, No. 2, 1997 Table 6—Least significant differences (LSD) * P,0.05 (degrees of freedom5 4,44) and mean values for the T-I parameters for astringency, bitterness and sourness of : red wine (pH 4.0) (sample 1), wine with added malic acid (pH 3.7 and 3.5) (samples 2 and 3 respectively) and wine with added lactic acid (pH 3.7 and 3.5) (samples 4 and 5 respectively). Tmax and Ttot were measured as seconds and Imax as mm in a 150 mm line scale T-I parameters Sample 1 Sample 2 Sample 3 Sample 4 Sample 5 3.5 3.7 3.5 3.7 4.0 pH lactic LSD lactic malic malic control acid Tmax Ttot Imax 19.17a 36.36a 41.53a 19.83a 40.36ab 62.03b Tmax Ttot Imax 21.22a 42.39a 51.58a 21.07a 41.24a 54.24a Tmax Ttot Imax 18.67a 37.81a 41.26a 17.42a 44.28b 67.33bc Astringency 19.03a 19.39a 44.92b 42.19b 68.89bc 54.06b Bitterness 20.14a 20.31a 39.47a 41.67a 52.22a 52.14a Sourness 19.00a 18.61a 48.40c 41.94b 73.81c 61.39b 17.47a 42.00b 73.64c — 5.14 9.60 20.36a 40.67a 57.06a — — — 17.36a 43.67b 68.64bc — 4.54 7.36 a-c Means with same letter in each row are not significantly different p,0.05. ** LSD only provided where means differed P,0.05. wine samples that Guinard et al., (1986b) examined. This difference in results may be due to different wine samples, the lower pH, the different added acid (citric), or the different method of measuring Imax. They used the scalar method, thus Imax ratings were not collected over a period of time but at a defined time. For Ttot significant differences were observed between high (4.0) and low (3.5) and (only for lactic acid samples) the high and medium (3.7) pH samples. The two T-I parameters of the wine samples did not show a linear relationship with pH, except for Ttot of malic acid samples where there was an increase in Ttot with decreasing pH (r 5 0.988, 1 d.f., P , 0.1). Other compounds present in wine might have affected or masked perceived astringency. Tmax values were also not significantly different across the range of pH values. At any given pH value, there was no effect of acid on astringency. For bitterness parameters, there were no significant differences among samples, suggesting that this attribute remained unaffected by addition of acids. In contrast, Fischer and Noble (1994) reported a significant increase in bitterness by reducing wine pH from 3.8 to 3.2 with citric acid. Imax for sourness showed a significant increase with decreasing pH, but at the same pH, the type of acid had no effect (Table 6). Ttot showed an increase for both added acids which was significant between high and medium and high and low pH values for both acids, and between the medium and low pH for malic acid. Generally, malic acid samples gave higher scores for Ttot than samples with lactic acid, which were significantly different only at the low pH. General It has been shown that subjects with larger mouths gave higher ratings to astringent samples than subjects with smaller mouths (Thomas and Lawless, 1995b). Since males usually have larger mouths than females, one could expect that the female panel of our experiment gave lower ratings for astringency than would be obtained from a mixed panel. In an attempt to ascertain if gender was a predominant factor affecting perception of astringency, Thomas and Lawless (1995b) selected a group of panelists evenly matched for height and found that height or body size was more influential than gender. Although no height and body size information was collected, the panelists had a range of heights and body sizes. Previous T-I studies have shown that the Imax of an attribute was correlated with its Ttot (Robichaud and Noble, 1990; Peleg and Noble, 1995) and our results were consistent with this observation. When an increase in Imax was observed (astringency and sourness) Ttot increased. In contrast, the Tmax ratings were generally independent of the Imax of the attribute. It has been suggested (Leach, 1980) that the Tmax of an attribute depends on concentration of stimuli. Since Tmax of astringency showed no differences, it is conceivable that concentrations of stimuli were all quite similar, and that perception differences in Imax were not directly due to concentration differences. Thus small changes in ionization of phenolic compounds produced by pH reduction had only a small effect on the concentration. Hence differences in Imax could have been due to other effects, such as changes in salivary proteins. For sourness Tmax did not increase with an increase in acid concentration and possibly high levels of astringency and bitterness confused the judges. A comparison of Imax values for astringency of control samples (No 1) of wine (41.53) and model wine (35.39) revealed little difference between them, given the difference in total phenol values (Table 2). One possible explanation was that the late eluting band of polymeric procyanidins observed in the HPLC chromatograms of the grape seed solution contributed greatly to astringency. Indeed, Lea and Arnold (1978) reported that among the dimeric, trimeric, tetrameric, pentameric and polymeric procyanidin fractions, there was an increase in astringency with degree of polymerization. However, wine polymers are not expected to have the same effect on astringency. Since the wine had aged for 2 years, polymerization reactions of anthocyanins and procyanidins could lead to a reduced availability of procyanidins able to bind with salivary proteins. Polymerization is believed to lead to precipitation of polymeric material and at the same time to loss of astringency in red wines (Haslam, 1989). Additionally, dimeric (B1 and B2) and trimeric (C1) procyanidins as well as epicatechin gallate, that were only present in the model solution, contributed considerably (Lea and Arnold, 1978) to its astringency. Samples 3 and 5 (Table 4) of model solution were perceived to be more astringent than the corresponding ones of wine (Table 6) possibly due to the lower pH of the model solution (3.2) compared to wine (3.5). In addition, the difference in pH values between samples 2 and 3 and 4 and 5 was 0.3 for the model wine and 0.2 for the wine. This may have resulted in the greater increase in astringency of the model wine samples with decreased pH. Astringency Imax increased with decreasing pH of both acids, confirming results of Fisher et al. (1994) who found that astringency was higher at pH 3.0 than 3.6 in white wine. Guinard et al. (1986b) observed a similar effect of pH on astringency of model solutions. Addition of acids reduced the pH of the solution, increased the percentage of tannins in the phenol form and therefore increased the likelihood of hydrogen bonding and binding to salivary proteins. According to Clifford (1995), the above explanation is not plausible, since the pKa of the phenolic hydroxyl groups is at a much higher pH (.pH 8.0). Since the pH of the solutions that Guinard et al. (1986b) used was in the range of 2.5 to 3.8, this ion effect was expected to be too small to account for the observed increase in astringency. Alternatively, a change in charge of salivary proteins involved in adsorption and desorption of the tastant/irritant could affect the binding and dissociation of phenolic compounds, or precipitation of salivary proteins could increase as their pI was approached. This explanation is in agreement with the lack of differences in Tmax values, indicating no differences in stimuli concentrations since the concentration of phenolic compounds is expected to remain unaffected by pH change. Thomas and Lawless (1995a) suggested that the presence of acids could affect the salivary or mucous layer and the epithelium proteins by denaturing them thus increasing the sensation of perceived astringency. This explanation fits with our observations. A similar pH reduction did not have a significant effect on astringency at high tannin concentrations such as found in red wines. This could be because the red wine was so astringent that changes induced by acidity adjustments were not detectable. The limiting factor to tannin-protein complex formation may also be no longer the tannin concentration but the quantity of proteins in the mouth. In their view, the effect of acidity on astringency Volume 62, No. 2, 1997—JOURNAL OF FOOD SCIENCE—419 ACIDS AND ASTRINGENCY IN WINES . . . was less pronounced in wine than in model solutions due to the higher buffer capacity of wine. In our experiment, it is possible that astringency was enhanced by direct contribution, elicited by the added acids. Several reports (Peleg and Noble, 1995; Rubico and McDaniel, 1992) have confirmed that acids can elicit astringency. Thomas and Lawless (1995a) reported that malic and lactic acids were rated as astringent in addition to their sour taste. Acids have also been shown to increase saliva output (Norris et al., 1984; Fischer et al., 1994). Fischer et al. (1994) demonstrated that subjects with low saliva flow rate perceived astringency more intensely and with longer duration than those with high flow rate. It was suggested that higher saliva flow corresponded to a greater supply of saliva proteins and thus higher level of oral lubrication. However, it has not been demonstrated that increase in saliva output was necessarily accompanied by increased secretion of salivary proteins (Clifford, 1995). In our results the possible increase in saliva output of panelists due to the presence of acids in the mouth for 15 sec prior to expectoration, could have resulted in lower absolute values for astringency. Samples with malic and samples with lactic acid at the same pH did not differ significantly in astringency despite the difference in titratable acidity. This suggests that it was the free [H+], that is the pH, which affected astringency and not the potential [H1] (titratable acidity) which depends on the number of undissociated carboxyl groups in the molecule of the acid. According to Noble (1995), little difference was observed between evaluation of bitterness or sourness by expectorating vs swallowing of the wines or solutions. Additionally, Peleg and Noble (1995), Fischer et al. (1994) and Robichaud and Noble (1990), asked panelists to expectorate samples evaluated for bitterness. Hence during our study samples were expectorated because we thought that this would not notably influence results and would prevent panelists from swallowing the alcohol in the samples. The effect of pH on bitterness parameters was not significant in wine or model solutions. The high levels of astringency masked possible changes in bitterness or this attribute did not change significantly since concentration of bitter phenolic compounds remained the same. Sourness Imax and Tmax increased on lowering pH. This was expected since sourness is associated with acids. Several researchers suggested that there is no simple relationship between pH and perceived sourness (Makhlouf and Blum, 1972; Norris et al., 1984) and that the intensity of that attribute depends also on titratable acidity and the anion of the acid (Rubico and McDaniel, 1992). At equal pH value and titratable acidity malic acid is more sour than lactic (CoSeteng et al., 1989). In our results, wines and model solutions with malic and lactic acid showed no differences in sourness parameters. CONCLUSIONS THE PERCEIVED Imax and Ttot of astringency were increased by pH reduction both in model solution and red wine, possibly due to increased precipitation of salivary proteins in the mouth as their pI was approached. Tmax of the same attribute remained unaffected by pH reduction. At a given pH, there were no significant differences between samples containing malic or lactic acid for any of astringency T-I parameters. The same variation in pH had no effect on bitterness parameters in model solution and in red wine. The effect observed for sourness Imax and Ttot in both model solution and wine was due to differences in pH. At a given pH there were no differences between malic and lactic acid samples for any of the T-I parameters. REFERENCES Amerine, M.A. 1980. The words used to describe abnormal appearance, taste and tactile sensations of wines. In The Analysis and Control of Less De- sirable Flavors in Foods and Beverages. G. Charalambous (Ed.), p. 319– 351. Academic Press, New York. Bakker J. and Timberlake, C.F. 1985. The distribution and content of anthocyanins in young port wines as determined by high performance liquid chromatography. J. Sci. Food Agric. 36: 1325–1333. Bakker, J., Bridle, P., Timberlake, C.F., and Arnold, G.M. 1986. The colour, pigment and phenol content of young port wines; effects of cultivar, season and site. Vitis 25: 40–52. Bate-Smith, E.C. 1954. Astringency in foods. Food Process. Pack. 23: 124– 135. Breslin, P.A., Gilmore, M., Beauchamp, G.K., and Green, B.G. 1993. Psychological evidence that oral astringency is a tactile sensation. Chem. Senses 18: 273–284. Clifford, M.N. 1986. Phenol-protein interactions and their possible significance for astringency. In Interaction of Food Components. G.G. Birch, and M.G. Lindley (Ed.), p. 143–163. Elsevier Science Publishers, London. Clifford, M.N. 1995. Astringency. Presented at the phytochemistry of fruit and vegetables symposium, Murcia, Spain, Sept. 1995. CoSeteng, M.Y., McLellan, M.R., and Downing, D.L. 1989. Influence of titrable acidity and pH on intensity of sourness of citric, malic, tartaric, lactic and acetic acids solutions and on the overall acceptability of imitation apple juice. Con. Inst. Food Sci. Technol. 22: 46–51. Delcour, J.A., Vandenberghe, M.M., Corten, P.F., and Dondeyne, P. 1984. Flavor thresholds of polyphenolics in water. Am. J. Enol. Vitic. 35(3): 134– 136. Fischer, U., Boulton, R.B., and Noble, A.C. 1994. Physiological factors contributing to the variability of sensory assessments: Relationship between salivary flow rate and temporal perception of gustatory stimuli. Food Quality and Preference 5: 55–64. Fischer, U. and Noble, A.C. 1994. The effect of ethanol, catechin concentration, and pH on sourness and bitterness of wine. Am. J. Enol. Vitic. 45: 6– 10. Guinard, J.X., Pangborn, R.M., and Lewis, M.J. 1986a. The time-course of astringency in wine upon repeated ingestion. Am. J. Enol. Vitic. 37: 184– 189. Guinard, J.X., Pangborn, R.M., and Lewis, M.J. 1986b. Preliminary studies on acidity-astringency interactions in model solutions and wines. J. Sci. Food Agric. 37: 811–817. Hagerman, A.E. and Butler, L.G. 1981. The specificity of proanthocyanidinprotein interactions. J. Biol. Chem. 256: 4494–4497. Haslam, E. 1974. Polyphenol-protein interactions. Biochem. J. 139: 285–288. Haslam, E. 1989. Polyphenol complexation. In Plant Polyphenols. Vegetable Tannins Revisited, University of Cambridge (Ed.), p. 155–219. Cambridge University Press, Cambridge. Kock, K., Morley, S.D., Mullins, J.J., and Schmale, H. 1994. Denatonium bitter tasting among transgenic mice expressing rat von Ebner’s gland protein. Physiol. Behav. 56: 1173–1177. Lawless, H.T., Corrigan, C.J., and Lee, C.B. 1994. Interaction of astringent substances. Chem. Senses 19: 141–154. Lea, A.G.H. and Arnold, G.M. 1978. The phenolics of ciders: Bitterness and astringency. J. Sci. Food Agric. 29: 478–483. Leach, E.J. 1984. Evaluation of astringency and bitterness by scalar and time-intensity procedures. Master thesis, University of California, Davis. Lee, C.B. and Lawless, H.T. 1991. Time-course of astringent sensations. Chem. Senses 16: 225–238. MacFie, H., Bratchell, N., Greenhoff, K., and Lloyd, V.V. 1989. Designs to balance the effect of order of presentation and first-order carry-over effects in Hall tests. J. Sensory Studies 4: 129–148. Makhlouf, G.M. and Blum, A.L. 1972. Kinetics of the taste response to chemical stimulation; a theory of acid taste in man. Gastroenterology 63: 67– 75. Murray, N.J., Williamson, M.P., Lilley, T.H., and Haslam, E. 1994. Conformational study of a salivary proline-rich protein repeat sequence. Eur. J. Biochem. 219: 923–935. Noble, A.C. 1995. Application of time-intensity procedures for the evaluation of taste and mouthfeel. Am. J. Enol. Vitic. 46(1): 128–133. Norris, M.B., Noble, A.C., and Pangborn, R.M. 1984. Human saliva and taste responses to acids varying in anions, titrable acidity and pH. Physiology & Behavior 32: 237–244. Ough, C.S. and Amerine, M.A. 1988. In Methods for Analyses of Musts and Wines. R. Rouseff (Ed.). Elsevier Science Publishers, New York. Peleg, H. and Noble, A.C. 1995. Perceptual properties of benzoic acid derivatives. Chem. Senses 20: 393–400. Ricardo da Silva, J.M., Rigaud, J., Cheynier, V., Cheminat, A., and Moutounet, M. 1991. Procyanidin dimers and trimers from grape seeds. Phytochemistry 30: 1259–1264. Robichaud, J.L. and Noble, A.C. 1990. Astringency and bitterness of selected phenolics in wine. J. Sci. Food Agric. 53: 343–353. Rubico, S.M. and McDaniel, M.R. 1992. Sensory evaluation of acids by freechoice profiling. Chem. Senses 17: 273–289. Somers, T.C. and Evans, M.E. 1977. Spectral evaluation of young red wines: Anthocyanin equilibria, total phenols, free and molecular SO2, chemical age. J. Sci. Food Agric. 28: 279–287. Thomas, C.J.C. and Lawless, H.T. 1995a. Astringent subqualities in acids. Chem. Senses 20: 593–600. Thomas, C.J.C. and Lawless, H.T. 1995b. Stimulus quantity, mouth size and perception of astringency intensity. Abstract presented at the second Pangborn sensory science symposium, Davis, August, 1995. Vérette, E., Noble, A.C., and Somers, T.C. 1988. Hydroxycinnamates of Vitis vinifera: Sensory assessment in relation to bitterness in white wines. J. Sci. Food Agric. 45: 267–272. Ms received 3/19/96; revised 7/16/96; accepted 10/7/96. We thank the EC fo r funding this pro ject with the Human Training and Mo bility Grant (SK). We also thank Co ckburn Smithes & CIA, LDA, Vila No va da Gaia, Po rtugal, fo r making the wine as part o f the FLAIR pro ject (no 89053). 420—JOURNAL OF FOOD SCIENCE—Volume 62, No. 2, 1997