Research Article Received: 29 October 2013

Revised: 21 August 2014

Accepted article published: 10 September 2014

Published online in Wiley Online Library: 10 October 2014

(wileyonlinelibrary.com) DOI 10.1002/jsfa.6907

Sweetness and other sensory properties of model fruit drinks: does viscosity have an impact?† Cai VS Brandenstein,a* Mechthild Busch-Stockfischa and Markus Fischerb Abstract BACKGROUND: The impact of thickening agents and viscosity levels on sensory perception was studied in model fruit drinks. Four formulations were prepared that varied in the sweetener blend (erythritol, maltitol and/or steviol glycosides). Locust bean gum and its blends with either xanthan or carrageenan were used to adjust viscosity levels (20, 40, and 70 mPa s). The ranges of viscosity and sweetness level were selected to represent a typical concentration range in commercially available beverages. RESULTS: An increase in viscosity resulted in significant increases in pulpiness, sliminess and perceived viscosity (P-values ≤ 0.001), which were not dependent on sweeteners or hydrocolloid type. Taste perception remained largely unchanged irrespective of the hydrocolloid used. CONCLUSION: The impact of viscosity on sweetness and taste perception was much smaller in the concentrations used than has been generally reported. The effect of the type of hydrocolloid on the perception of taste attributes was greater than that of viscosity. © 2014 Society of Chemical Industry Keywords: steviol glycosides; erythritol; locust bean gum; xanthan; carrageenan; descriptive analysis

INTRODUCTION Consuming a beverage is a multi-modal experience for consumers. Consumers simultaneously perceive taste, aroma and trigeminal sensations, as well as the texture of the product.1 Replacing sucrose in beverages can cause changes in these perceptions. For example, the use of high-intensity sweeteners that do not increase viscosity and density results in thin, watery products. By adding hydrocolloids, these changes can be compensated. Hydrocolloids are functional ingredients that are widely used in processed foods and beverages. At relatively low concentrations, they are used to prevent sedimentation of fruit particles or to adjust viscosity.2 – 4 The latter is probably the most common of their applications.5 Among other things, increasing viscosity is industrially important to build body in beverages. However, the addition of hydrocolloids can change a product’s rheological properties and thus modify its texture, which can lead to changes in taste perception.1,6 – 8

J Sci Food Agric 2015; 95: 809–818

∗

Correspondence to: Cai VS Brandenstein, Department of Nutritional Sciences, Faculty of Life Sciences, Hamburg University of Applied Sciences, Lohbrügger Kirchstr. 65, 21033 Hamburg, Germany. E-mail: [email protected]

† Part of this study was presented at 10th Pangborn Sensory Science Symposium 2013 and at 42. Deutscher Lebensmittelchemikertag. a Hamburg University of Applied Sciences, Faculty Life Sciences, Department of Nutritional Sciences, 21033, Hamburg Germany b Institute of Food Chemistry, Hamburg School of Food Science, University of Hamburg, Hamburg, Germany

www.soci.org

© 2014 Society of Chemical Industry

809

Therefore, in addition to sweeteners, hydrocolloids are one of the key factors governing taste perception of a product. Pangborn et al.9 suggested, among other things, that the effects of hydrocolloids are influenced by the nature of the sweetener(s). Since it is of practical importance to know if and how the taste perception of a certain product is changed, a number of researchers have sought an understanding of the impact of hydrocolloids on sensory characteristics and their possible interactions with sweeteners.4,9,10 In general, increased viscosity reportedly leads to a decrease in taste perception, and therefore in reduced sweetness perception.6,8,11 – 13 However, there are also opposing opinions about the correlations between sweetness, texture and viscosity.

Some authors have reported that an increase in viscosity may enhance sweetness and other taste attributes.14 – 17 As different are the statements about changes in taste perception, as manifold are the test conditions in the literature. Holm et al.16 observed an increase in sweetness with an increasing amount of pectin; Kanemaru et al.17 found the sweetness-enhancing effects of soluble starch; and Burns et al.15 studied the effect of Polycose®, a commercially available glucose polymer, as a thickening agent. The contradictory findings in the literature are considered to be due to the high number of experimental variables (e.g. hydrocolloid type, viscosity range, food matrix, choice of sensory evaluation technique), which often limits the comparison of published results.8 Most of the literature has focused on dairy dessert system models1,7,18 or on gelled products.6,19,20 Only a few studies have investigated the effects of hydrocolloids on taste perception of

www.soci.org

CVS Brandenstein, M Busch-Stockfisch, M Fischer

thickened beverages.4,21 – 25 Here, also, test conditions varied considerably among studies investigating milk beverages24 or carbonated beverages.25 Because of different textural characteristics and varied nature of the data (as mentioned above), it is difficult to draw definite and general conclusions. Therefore, the validity of extrapolating the reported conclusions to thickened beverages must be investigated. In our study, we focused on alternatively sweetened model fruit drinks, which have received only limited attention to date.21,26 The objective of this study was to investigate the effects of viscosity and different hydrocolloids on the texture and taste perceptions of low-viscosity model fruit drinks. We also sought to clarify whether sucrose replacement was connected with detrimental effects on taste perception when combined with hydrocolloids. Different viscosity levels (20, 40, and 70 mPa s) were obtained by the use of locust bean gum (LBG) alone or in combination with either xanthan (Xan) or carrageenan (Car). Blends of alternative sweeteners (erythritol (Ery), maltitol (Mal) and/or steviol glycosides (Stev)) were used as sucrose (Suc) substitutes. The ranges of viscosity and sweetness levels were selected to represent typical concentration ranges in commercially available beverages. Paired comparison tests were conducted primarily to construct an overview of changes in the sweetness perceptions of model fruit drinks. During the paired comparison tests, three viscosities within one formulation were compared with each other. To obtain in-depth information about changes in taste and perceived texture, descriptive analysis according to the quantitative descriptive analysis (QDA) method was subsequently conducted.

a Sum of sweetener concentration is approximately equisweet to a 5 g kg−1 sucrose solution.

EXPERIMENTAL

Table 3. Basic formulation of model fruit drinka

Materials Samples were prepared with unsweetened concentrated apple and orange juices (PepsiCo Deutschland GmbH, Hamburg, Germany). Sucrose (commercial grade) was purchased from the local market. Maltitol (Maltidex CH 16385) was obtained from Cargill Europe BVBA, Mechelen, Belgium; erythritol (Erylite F8030) from Jungbunzlauer Suisse AG, Basel, Switzerland; and steviol glycosides (Reb A 97) from NP Sweet, Copenhagen, Denmark. Locust bean gum (Viscogum FA), 𝜅-carrageenan (Satiagum BDC 20) and xanthan gum (Satiaxane CX910) were provided by Cargill Inc., Minneapolis, MN, USA. Sweeteners Different blends of alternative sweeteners were chosen based on preliminary tests (Table 1). The required amounts of each sweetener were calculated by dividing the amount of sucrose by the sweetening power of each sweetener.27 The sweetening powers of the sweeteners relative to sucrose, calculated as the ratio between the sucrose concentration and the equisweet concentration of the sweeteners, are shown in Table 2.

810

Preparation of model fruit drinks Following a basic formulation (Table 3), samples were prepared in 1000 mL batches. Sweeteners were dispersed in a mixture of concentrated apple and orange juices. The mixture was diluted afterwards with hot (85 ∘ C), filtered water to 250 mL. Hydrocolloids, at different levels, were dispersed in 750 mL hot water and were stirred with a hand mixer (Bosch, Munich, Germany) for 10 s at speed 1 (600 W). Then, the pre-mixture was homogenized for

wileyonlinelibrary.com/jsfa

Table 1. Blends of sweeteners and their corresponding amounts used in this study Amount of sweetenersa (g kg−1 ) Sweetener Sucrose (Suc) Erythritol + Maltitol Erythritol + Steviol glycosides Erythritol + Maltitol + Steviol glycosides

Blend ratio – 50:50 80:20 30:30:40

Suc

Ery

Mal

50.0 80.0

41.7

30.0

25.0

Stev

50.0 0.03 0.07

Table 2. Degrees of sweetness of the sweeteners relative to 3 g kg−1 aqueous sucrose concentration, as verified by paired comparison tests Sweetener

Degree of sweetnessa

Sucrose Erythritol Maltitol Steviol glycosides

1 0.5 0.6 300

a Equisweetness was the concentration at which 50% of the panelists perceived the alternative sweetener to be sweeter than sucrose.

Ingredient Apple juice Orange juice Sucrose Waterb

(g kg−1 ) 100 75 50 775

a

Hydrocolloids were also added. The amount of water was changed with sweetener blends – water was added until a total mass of 1000 g was reached. b

2 min at speed 2 using an ULTRA-TURRAX dispersing machine (IKA Werke GmbH & Co. KG, Staufen, Germany). The final mixture was prepared by mixing the sweetener solution with the hydrocolloid solution and homogenizing for 30 s. The model drink was cooled to room temperature within 1 h. Model fruit drinks were prepared 1 day before panel evaluation and stored for 24 h at 4 ∘ C. Before sensory analysis or viscosity measurements, samples were homogenized for 2 min at speed 5. Adjustment of viscosity and rheological measurements The model fruit drinks were adjusted in viscosity steps of 20, 40 and 70 mPa s. Concentrations of hydrocolloids were selected to produce viscosity properties as similar as possible. Rheological measurements were carried out at 20 ∘ C and a shear rate of 100 s−1 with a rotational viscometer (Haake Viscotester VT 550, Thermo Electron (Karlsruhe) GmbH, Germany). To control viscosity, two batches of each formulation were prepared and 10 mL aliquots of each batch were measured in duplicate. Deviations of ±5 mPa s

© 2014 Society of Chemical Industry

J Sci Food Agric 2015; 95: 809–818

Sweetness and other sensory properties of model fruit drinks

Table 4. Concentrations of hydrocolloids (g kg−1 ) giving equivalent viscosities in the presence of the sweeteners tested Viscosity levela (mPa s) LBG 20 40 70 Xan/LBG 60:40 20 40 70 LBG/Car 80:20 20 40 70

Ery_Stev

Ery_Mal

Ery_Mal_Stev

2.3

1.3 2.1 2.9

1.5 2.1 2.8

1.6 2.3 3.0

1.9

1.1 1.7 2.3

1.1 1.9 2.5

1.1 1.9 2.5

2.5

1.5 2.4 3.2

2.2 2.7 3.5

2.0 2.8 3.6

a

The chosen ratios exhibited the strongest attractions between the corresponding hydrocolloids.27,28 b Sucrose-sweetened samples were used as controls.

were deemed acceptable. The resulting amounts of hydrocolloids are shown in Table 4.

Foamy Bubbly Pulpy Fruity Sweet Sour Bitter Metallic Slimy Viscous Lingering sweetness

Definition Foaming on surface Amount of bubbles (in liquid) Homogeneous consistency Fruitiness Basic taste Basic taste Basic taste Taste sensation Sticky consistency, difficult to swallow Expansion in mouth, thickness of sample Persistent sweetness after swallowing

as a control in each session to ensure comparability of the samples. Each session started with the assessment of a warm-up sample, which was the same for all sessions and was excluded from analysis. The presentation order of the three-digit coded samples was randomly designed over the panelists, based on a balanced Latin square design. Eleven-point scales anchored with ’nil’ and ’high’ were used to describe attribute intensity. Filtered water was provided to rinse between each sample. Between the two sessions, cream cheese, cucumbers and crackers could be taken ad libitum to neutralize the palate. Statistical analyses Data acquisition was conducted with Fizz software (v2.47B, 2012, Biosystèmes, France). Data analyses were carried out using XLSTAT (Version 2012.3.01, Addinsoft, Paris, France). Analysis of variance (ANOVA) and Tukey’s multiple comparison tests were performed to determine the significance of differences (𝛼 < 0.05).

RESULTS AND DISCUSSION Perception of sweetness The sweetness of most of the samples was not significantly influenced by viscosity (Table 6). Samples of the same formulation thickened either with LBG alone or with LBG in combination with Car did not significantly differ from each other. Only two exceptions were found between samples containing the combination of LBG and Xan. For samples containing the blend of Ery and Mal, those with lower viscosity (40 mPa s) were significantly perceived as sweeter than samples with high viscosity (70 mPa s). This finding confirmed the general opinion given in the literature that an increase in viscosity resulted in a reduction of perceived sweetness. However, the significant difference between the samples sweetened with erythritol and steviol glycosides was antithetical. Panelists perceived the sample with the medium viscosity (40 mPa s) as sweeter than the sample with the lowest viscosity (20 mPa s). In contrast to most findings in the literature describing a decrease in sweetness with increasing viscosity,6,8,11 – 13 no reliable trend could be observed. As already mentioned, the test conditions of previous studies were quite different. Boland et al.6 and Bayarri et al.12 investigated the effects on sweetness in hydrocolloid gels. Although Cook et al.8 also studied the impact of hydrocolloids in solution, they selected different hydrocolloids and a larger concentration range (2.0 and 10 g kg−1 ).

© 2014 Society of Chemical Industry

wileyonlinelibrary.com/jsfa

811

Sensory analysis The effect of viscosity on sweetness perception was evaluated through paired comparison tests28 between samples with equal sweetener and hydrocolloid type but with different viscosities. Twenty panelists (17 female and three male, all students from Hamburg University of Applied Sciences, Germany), well trained and selected according to their ability to distinguish between concentrations of the same stimulus,29,30 participated in these tests. The tests took place at the sensory facilities of Hamburg University of Applied Sciences, Germany, according to ISO 8589.31 Samples (20 mL) were served at room temperature (20 ± 1 ∘ C) and presented to panelists in odor-free plastic cups coded with randomly selected three-digit numbers and in random order. Panelists were asked to judge (force-choice mode) which sample tasted sweeter within each pair. Each panelist participated in two sessions per day, separated by a 15 min break, and received three pairs per session. Panelists were instructed to carefully rinse their mouths with water between each pair of samples. In addition, crackers and cream cheese were provided as palate cleansers between the two sessions. Paired comparison tests were considered two-tailed. The results were analyzed in the usual way by obtaining the total distribution of choice for each sample within each pair. Significant differences (𝛼 ≤ 0.05) were established by comparing the experimental results with the corresponding Table A.2 in ISO 5495.28 Descriptive analysis of the model fruit drinks according to the principles of the QDA Method® was conducted by the same 20 panelists. All panelists were familiar with the testing procedure and the product. Table 5 gives an overview of the examined attributes. Panelists participated in two profiling sessions per day separated by a 30 min break. Three viscosities of the same thickening agent and the same sweetener or blend were served per session. Sensory analysis was carried out in duplicate for each formulation on separate testing days. A sucrose-containing sample with the same hydrocolloid and medium viscosity (40 mPa s) was also examined J Sci Food Agric 2015; 95: 809–818

Table 5. Attributes and their definitions of sensory analysis of flavored drinks Attribute

Sweetener Sucb

www.soci.org

www.soci.org

CVS Brandenstein, M Busch-Stockfisch, M Fischer

Table 6. Results of paired comparison tests between samples within a formulation with different viscosities. Pairwise comparisons with significant differences in sweetness are indicated with asterisksa

Table 7. P-values of attributes obtained from analysis of variance followed by Tukey’s multiple comparison tests (𝛼 < 0.05) Attribute

Sweetener

Viscosity

Suc

20 40 20 20 40 20 20 40 20 20 40 20

Ery_Mal

Ery_Mal_Stev

Ery_Stev

40 70 70 40 70 70 40 70 70 40 70 70

LBG NS NS NS NS NS NS NS NS NS NS NS NS

LBG + Xan NS NS NS NS * NS NS NS NS (*) NS NS

LBG + Car NS NS NS NS NS NS NS NS NS NS NS NS

a Significant differences for which the more viscous sample was rated as sweeter are marked by parenthetical asterisks. *P value ≤ 0.05; NS, not significant.

Descriptive analysis No significant differences were found between the control samples within one thickening agent. Therefore, the control samples were treated as one sample in the analysis and the data from the different formulations within one thickening agent can be considered together.

812

Perceived texture attributes Although in the present study the differences in measured viscosity were small, the panelists were still able to discriminate the higher hydrocolloid-containing model fruit drinks. Analysis of the data obtained from the profiling tests showed that the panelists distinguished significant differences (𝛼 < 0.05) in pulpiness, sliminess and the perceived viscosity of the samples. In all cases, P-values were less than or equal to 0.001 (Table 7). As can be seen in Fig. 1, increasing viscosity hardly affected texture perception for all thickening systems and resulted in increased perceived pulpiness, sliminess and viscosity. Consequently, for each formulation, the more pulpy, slimy and viscous samples were clearly those of higher viscosity, whatever type of hydrocolloid or sweetener was used. As expected, these findings were in agreement with other studies that reported an increase in apparent viscosity when hydrocolloids were added.24,32 – 34 When adding hydrocolloids (2.0 g kg−1 ) in amounts similar to our study, Chung et al.32 observed an increase in viscosity in a model sauce that was in good agreement with the perceived ’mouthfeel viscosity/thickness’. Yanes et al.,24 who investigated flavored milk beverages, also pointed out that the more viscous sample was perceived to be thicker than the less viscous one. This correlation between viscosity data and sensory perception was determined as relatively high by several authors.33,35,36 Comparing the results of the three thickening systems, it was notable that samples thickened with the combination of LBG and Xan obtained the highest ratings. The data confirmed the observations obtained for other models by Makri et al.37 that the combination of Xan and LBG exhibited higher viscosity values than LBG alone. Similar findings were made by Sahin and Ozdemir,2 who reported that the presence of Xan caused a slimy texture. Further studies found that Xan had the largest impact on the batter viscosity of cakes, compared to other hydrocolloids.38,39

wileyonlinelibrary.com/jsfa

Foamy Bubbly Pulpy Fruity Sweet Sour Bitter Metallic Slimy Viscous Lingering sweetness

LBG