Assessing the Influence of Fat Content upon the Sensory and Textural Properties of Margarine to Guide the Development of Healthier Products

1. Introduction

Margarine is a water-in-oil emulsion consisting of an aqueous phase dispersed as fine droplets in liquid oil, usually derived from vegetable crops such as soybean, sunflower, corn, and palm, stabilized within a network of solid fat crystals [1]. During its production, the oil and aqueous phases are independently prepared and carefully mixed by means of dosing systems to form an emulsion, which is subsequently pasteurized and cooled to partially crystallize the fat and stabilize the system. In addition, other ingredients and additives may be added, i.e., salt, colorants, flavorings, and emulsifiers [2]. Initially, margarine was developed as a more affordable and versatile alternative to cow milk’s butter, as it resembles the latter in terms of texture, flavor, mouthfeel, and use. It is a globally appreciated product that is commonly used as a butter substitute in cooking, as an ingredient in baking, and as a table spread itself. Preferences for margarine over butter can differ due to cultural factors, personal tastes, culinary habits, and individual dietary requirements, as some people still prefer traditional butter [3].

The composition of margarines can vary substantially, leading to products with specific characteristics like low-fat, vitamin-enriched, or lactose-free options. In addition to appearance, flavor and aroma, the consistency, mouthfeel, and texture of margarines are intrinsic properties that significantly affect their consumer acceptance [4]. Consistency is often a measure of smoothness, evenness, and plasticity, whereas texture refers mostly to structure, which can range from smooth to mealy, granular, crumbly, sandy, coarse, or lumpy. Margarine consistency and texture depend heavily on composition, fat content, fatty-acids profile of the oil phase, and processing techniques [4,5]. These properties are key to consumer acceptance as they significantly impact mouthfeel [5].

Currently, there is a growing consumer concern about improving their diets for better health and well-being. As a result, a growing demand for well-balanced diets and functional foods that provide specific health benefits has been observed. Healthier versions of common foods include lower levels of sodium, sugar, and saturated fat, along with significantly reduced calorie density compared to their regular counterparts [6,7]. Consumer interest in low-fat margarines (containing less than 80% fat) has surged due to the detrimental impact of high saturated and trans-fat intake on cardiovascular health [8]. According to the latest Dietary Guideline for Americans [9], partially hydrogenated oils (PHOs), once a common ingredient of margarines and the major source of artificial trans-fat in foods, are no longer Generally Recognized as Safe (GRAS) by the Food and Drug Administration (FDA), as of 2018. In a previous publication, the US Department of Agriculture [10] advised American consumers to limit their daily consumption of fat spreads, especially hard or stick varieties, due to their high trans fatty acid content. Nonetheless, Weber et al. [11] found that most margarine brands in the US marketplace have already been reformulated to comply with the current daily trans-fat intake recommendation post-FDA ban on PHOs.

Manufacturers are constantly changing food products to meet the requirements of regulatory organizations and follow consumer trends. However, reformulations aiming to reduce the fat content in margarines may significantly affect their physical and structural properties, leading to undesirable outcomes to their sensory profiles and consumer acceptance thereby. For instance, margarines with higher lipid contents are often associated with increased values of instrumental firmness and consumer acceptance scores for flavor and overall impression, in comparison to those with lower lipid contents. Additionally, margarines with lower lipid contents generally tend to exhibit negative sensory attributes more often, such as vegetable oil aroma, oily flavor, and rancid flavor, leading to lower consumer acceptance scores [4,12].

Sensory science comprises a set of systematic techniques aiming at accurately and objectively measuring human responses to consumer goods through their senses. As such, determining the sensory properties of the food itself provides key information for product development [13]. With the advancement of sensory analysis from the mid-twentieth century onwards, several sensory profiling methods have been developed. Some of them focus on training subjects based on standard sensory lexicons, whereas others enable subjects to more freely describe differences between samples (Dehlholm et al., 2012).

Lately, rapid descriptive methods, i.e., Projective Mapping and its variants, Napping [14] and Check-All-That-Apply (CATA) [15] have gained popularity in the realm of sensory science. Among the advantages of these analyses is the possibility of tracing the sensory profiles of a large number of samples in significantly less time, in comparison to classical descriptive techniques such as Quantitative Descriptive Analysis (QDA®) [16].

Projective mapping enables untrained subjects to group samples based on their similarities and differences by placing them on a two-dimensional surface using a piece of paper [17,18]. Subjects select their own criteria and determine their relative importance. Projective mapping has been applied to various food products, including apples and cheeses [19], chocolate milk desserts [15], granola bars [20], wines [21,22], and spirits [23]. CATA enables sensory profile assessment using consumers, without the need for repetitions. It has been widely used because it is quick and simple to apply, in addition to presenting results similar to classical descriptive methods such as QDA®. This technique consists of selecting terms from a list previously defined descriptors that best describe the product under evaluation. Data are analyzed according to the frequency of selection of each term through correspondence analysis and Cochran's Q Test [18,24].

Given the difficulties of reducing lipid content in margarine without compromising some of its characteristic properties, the development of low-fat alternatives conforming to consumers’ expectations and regulatory standards often becomes a challenge for the food industry [25]. In this context, sensory science techniques emerge as valuable tools to monitor consumers’ reaction to margarines with varying lipid contents. The present work aimed to determine the textural properties, sensory profiles, and hedonic ratings of commercial margarine with varying lipid contents from Brazilian market, by means of rapid sensory descriptive techniques (CATA and Napping), consumer acceptance test and instrumental analysis.

2. Materials and Methods

2.1. Samples

Twelve commercially available margarine samples from different manufacturers, yet sharing the same production date, were analyzed, including light, creamy, and regular versions. As shown in Table 1, samples had differing fat contents, representing the various characteristics of margarines available in supermarkets from Campinas, Brazil. Noticeably, margarines with high fat content are more commonly found in Brazilian market, as five of the samples had 80% lipids or more. All samples contained salt in their formulations.

2.2. Instrumental texture

Samples (4°C) were evaluated, in triplicate, for instrumental texture using a TA XT2 texture analyzer (Systems, Godalming, EUA) equipped with a 0.5 cm diameter cylindrical probe. The texture profile analysis (TPA) method was used, through two sequential penetrations with speed of 2mm/s, distance of 15mm, and a force of 5g. The following physical properties were evaluated: Hardness, adhesiveness, cohesiveness, and resilience.

Hardness is defined as the force required to deform or penetrate a food matrix. Regarding semi-solid foods, it refers to the force exerted by the tongue against the palate. Adhesiveness is related to the force required to break the force of attraction between the food matrix and other surfaces that come into contact with them, such as teeth. Cohesiveness refers to the strength of the internal connections that keep the integrity of a food matrix after a deformation event, as the extent of deformation of the food prior to its rupture is evaluated by measuring the difference between the forces of the first and second penetrations. Finally, resilience evaluates the recovery of the food structure after deformation, being the ratio between the area before and after deformation [26,27].

2.3. Sensory analysis

Sensory analyses took place at the Laboratory of Sensory Science and Consumer Research (LSCCR), at the University of Campinas (UNICAMP), Campinas, SP, Brazil. The research was approved by the ethics and research committee of UNICAMP, under the number CAAE 50531121.9.0000.5404. Subjects were invited to participate in the margarine study through specific social media groups created for this purpose and posters on university bulletin boards. They evaluated samples in individual booths with white light in a controlled environment at 21 ± 2°C, following ISO guidelines [28]. Before any evaluation, subjects were asked about their willingness to participate in the tests as volunteers and those choosing to collaborate signed an informed consent form.

2.3.1. Sample preparation and presentation

Samples were stored under refrigeration conditions at 4 ± 2°C until the sensory analyses. Immediately before evaluation, samples were retrieved from the cold storage and served to subjects. Moreover, subjects were provided with filtered water and Bauducco® cracker biscuits (Extrema, MG, Brazil) to be consumed as palate cleansers between each sample. Samples were served in plastic cups labeled with unique three-digit codes, with each sample portioning at 20 g. Sample presentation followed a sequential monadic order, arranged within a complete balanced blocks design as described by Meilgaard et al. [29]. For the evaluation of spreadability, subjects used disposable plastic spatulas to smear the margarine samples on slices of crustless white bread (Bimbo Brasil, Mogi das Cruzes, SP, Brazil), cut in dimensions of 5cm x 3cm x 1cm.

Acceptance test and CATA were performed in two separate sessions on two consecutive days, as six samples were evaluated in the first session and the remaining six samples in the second session, amounting to twelve samples presented to subjects in a completely balanced blocks design with sequential monadic order [29]. As for Projective Mapping, all twelve samples were presented simultaneously to each subject in a single session, who were asked to compare and group them according to their similarities and differences [30]. Among existing methods, Check All That Apply (CATA) and Projective Mapping (Napping) are often considered quicker alternatives to sensory profiling than traditional descriptive methods, which require lengthy training sessions [31].

2.3.2. Acceptance analysis

Acceptance analysis was performed with 112 subjects who declared themselves to be consumers of margarine, thus representative of the target audience. They were recruited based on their frequency of margarine consumption (at least three times per week). Subjects were asked to evaluate how they liked samples in relation to their appearance, aroma, flavor, texture, and overall impression, using a 9-cm unstructured scale, anchored at the left and right extremes by the terms “I disliked it very much” and “I liked it very much”, respectively [13,16]. The purchase intention of consumers was also evaluated using a 5-point scale, ranging from “I certainly would not buy” to “I certainly would buy” [29].

2.3.3. Check-All-That-Apply (CATA)

Check-all-that-apply analysis was applied to the same 112 subjects who participated in the acceptance analysis questionnaire [32,33]. They were asked to evaluate samples and choose the terms that best described them from a list of 20 predefined descriptors [34], which were: yellow color, light yellow color, aerated, bright, milk aroma, rancid aroma, vegetable oil aroma, butter flavor, milk flavor, rancid flavor, sweet taste, grass flavor, oil taste, metallic taste, salty taste, bitter taste, sour cream taste, light texture, homogeneous, and consistent. Descriptive terms were previously determined through the repertory grid method [35,36]. The sequence of terms in the list was balanced for each subject and kept the same for all samples [24].

2.3.4. Projective Mapping (Napping)

Projective Mapping was carried out with another group of 62 subjects, specifically recruited for this analysis, which is considered enough in the literature [37]. They were aged between 28 and 54 years old and 50% declared themselves as females. Subjects were selected based on their frequency of margarine consumption (more than twice a week) and were previously familiarized with the analysis procedure. Each session lasted approximately 40 minutes.

All 12 samples were presented simultaneously to each subject. Data were collected through Fizz software program. They were asked to position samples in a two-dimensional plane on the computer screen, so that samples perceived as being similar were grouped closer to each other, whereas samples perceived to be different were grouped farther from each other [38]. Subjects were also asked to list the characteristics of each group of samples to describe the sensory differences and similarities between the margarines [39].

As a result, each subject provided two coordinate vectors of 1 × 1 dimension each (one for the X axis, one for the Y axis), where they denoted the number of stimuli to be positioned in the rectangle. Thus, the final analyzed data set (denoted by X) was obtained by merging the N pairs of coordinate vectors, where N represented the number of subjects. In other words, X can be seen as a set of data structured in N groups of two variables each. Typically, the statistical analysis of data set X considers the “natural” partition of the variables.

2.4. Statistical analyses

Statistical analysis of data from acceptance test and instrumental texture was performed by ANOVA and Tukey’s Honestly Significant Difference (HSD) Test of Means (P ≤ 0.05), using SAS software version 9.4 (Statistical Analysis System, Raleigh, NC, USA). Data from purchase intention were plotted in frequency histograms for each sample [29]. Data from CATA and Projective Mapping were analyzed using XLSTAT software version 2023 (Addinsoft, Paris, IF, France). Cochran’s Q test (P ≤ 0.05) and correspondence analysis (CA) were also performed for data from CATA and correlated to acceptance test data.

Furthermore, data from CATA and acceptance test were correlated by means of Partial Least Squares (PLS) regression analysis [24]. Overall impression was considered the dependent variable (Y-matrix), while CATA parameters were the independent variables (X-matrix) [40]. Data from Napping were analyzed using Multiple Factor Analysis (MFA) and Hierarchical Cluster Analysis (HCA) [38,41].

3. Results and Discussion

3.1. Instrumental texture analysis

According to the results of instrumental texture analysis displayed in Table 2, samples with low fat contents were overall softer than those with high fat contents, although the latter did not differ significantly for hardness from samples D70 and C80, both having high lipid levels (70 and 80% respectively). The hardest samples were A60, A80 and B80. Margarine quality is generally defined by its texture, consistency, hardness and plasticity, with hardness being the most significant textural property perceived by consumers [42].

Adhesiveness is defined as the ease of removing margarine from the mouth or any surface [43]. Low-fat samples were significantly less adhesive (p ≤ 0.05) than those with high lipid content, except for B35 that did not statistically differ from the formers. Among high-fat samples, only A80 and F80 significantly differed (p ≤ 0.05) in cohesiveness from low-fat samples, those being the least cohesive of all samples. Regarding resilience, samples were segmented into two groups. High-fat samples showed lower values (A80, B80, C80, D82, A60, D70 and E50), whereas samples with the lowest lipid levels displayed significantly higher values (A30, B35, C38 and D55). Likewise, Silva et al. [44] verified similar behavior in relation to hardness and adhesiveness when analyzing margarines with different lipid contents. On the other hand, Ergönül [42] did not find statistical differences for the textural properties of hardness, adhesiveness or cohesiveness when analyzing margarines sold in the Turkish market.

3.2. Acceptance test

Acceptance ratings for appearance, aroma, flavor, texture, and overall impression are presented in Table 3. Means marked with the same letters in the same column do not significantly differ from each other according to Tukey's test. For attributes of appearance, aroma and texture, no significant difference (p ≤ 0.05) was observed between samples regarding their fat contents. When considering means of overall impression, which is often indicative of subjects' general hedonic response, high-fat samples (F80, D82, B80, A80, and D70) were preferred over those with lower lipid levels, with the exception of sample C38 did not significantly differ (p ≤ 0.05) from the latter.

According to Andersen et al. [45], flavor presents the strongest correlation with the overall impression among all attributes of a product, which was also observed in this study, as samples that were preferred also scored the highest ratings for flavor. Exceptionally, sample C38 was significantly different from the sample with the highest mean for flavor (F80), despite being among the most preferred. Moreover, it is possible to confirm consumers’ preference for samples C80 and F80 by analyzing the histograms of purchase intention (Figure 1), while samples A30 and E50 scored the lowest purchase intention ratings by subjects.

Acceptance test results are in line with the observations of Pădure [5], since there was a significant positive linear correlation between fat content and overall impression ratings, with 75.6% of the increase in acceptance being explained by the increase in the percentage of lipids of margarines (p<0.05). Acceptance means relative to flavor exhibited a similar trend, as 74.5% of the increase in this variable was explained by the increase in the fat content of the margarine samples studied (p<0.05).

3.3. Check-All-That-Apply (CATA)

Table 4 shows the frequency in which CATA terms were used to describe samples, as well as Cochran's Q test p-values. According to Cochran's Q test, 18 of the 20 attributes presented were considered significant (p < 0.05) to differentiate samples. Only the attributes “aerated” and “sweet taste” did not present any differences, that is to say, they did not contribute to the discrimination between samples. Furthermore, samples with lower lipid percentages exhibited a reduced frequency of mentions for “butter flavor”, while samples with higher lipid percentages exhibited a higher frequency of mentions for both “milk flavor” and “butter flavor”. These findings are key for understanding the overall acceptance of the product. The information obtained from this study could be valuable for the development of innovative products within the margarine and vegetable fat spreads industry, as well as in other segments that incorporate these products into their formulations.

The principal coordinates analysis of CATA in Figure 2 shows the grouping of samples according to their similarities and differences, as well as the descriptors that characterize them the most by proximity, with 81.87% explanation. The sample with the highest lipid content (D82) was described by terms “texture”, “aeration”, “shine” and “softness”. The other high-fat samples (A80, C80, F80, B80) were correlated with descriptors “consistent”, “butter flavor”, “milk aroma”, “milk cream flavor”, “milk flavor”, and “homogeneous”. Samples with intermediate lipid contents were characterized by “yellow color” and “salty taste”, whereas those with lower lipid levels were more related to terms “rancid aroma”, “rancid flavor”, “metal flavor” and “bitter taste”. Similar findings were observed by Foguel et al. [46] in commercial samples of cream cheese. The authors point out that the sample with the highest fat content was related to descriptors “milky”, “mild flavor” and “firm”, and that with the lowest fat content was more related to “sandy”, “matte”, “smooth”, “spreadable” and “white”.

The Partial Least Squares (PLS) regression analysis showed in Figure 3 correlates data from CATA with the hedonic data obtained in acceptance test. Figure 3a displays which descriptors are directly related to the greater acceptance of the product, located close to the overall impression, those being “consistent”, “homogeneous”, “soft”, “bright”, “milk aroma”, “milk flavor”, “milk cream flavor”, and “butter flavor”, which are the same terms that best describe high-fat samples. Figure 3b, which illustrates the descriptors impacting positively (blue) and negatively (red) overall impression ratings, confirmed that “vegetable oil aroma”, “oily flavor” and “rancid flavor” had a negative impact upon acceptance, as in the case of sample A30, the least preferred and best described by these descriptors. Overall, fat plays a key role in flavor release by conveying taste- and odor-active compounds to chemosensory receptors as it melts during oral processing [46]. This may explain why samples with lower percentage of lipids were the least preferred by subjects.

3.4. Projective Mapping (Napping)

Projective Mapping was used as a complementary tool in sensory profiling, considering the vast number of samples. By means of this method, samples were characterized by descriptors defined by subjects themselves during evaluation. The first factorial plane issued by the Multiple Factor Analysis (MFA) and showed in Figure 4 indicates the similarities and differences between samples as perceived by subjects. Similarly, the dendrogram in Figure 5 shows results of Hierarchical Cluster Analysis (HCA), which segmented samples in three distinct groups according to their similarities and differences. Consumers generally differ in their sensory likes and dislikes, and this inter-individual variation can be reduced by organizing the population by the sensory preference segmentation method [47]. HCA results corroborated with the sensory profiles obtained in CATA analysis, in which samples D55, A60, B80, D70, F80 and A80 presented similarities according to their proximity, and sample D82 was best described by the texture descriptor of “creaminess”. Samples C80 and B35 were characterized by the absence of “salty taste” and “milk flavor”, whereas sample A30 was discriminated from the others and characterized by descriptors “watery”, “residual” and “artificial flavor”.

CATA and Napping yielded different results, yet complementary to each other. However, it is important to emphasize that the results of these two methods rely on sensory lexicons sourced from differing databases. In CATA, participants receive the same predefined list of words, whereas in Napping, the set of analyzed words is formed spontaneously and individually by subjects based on their own perceptions and sensory references.

4. Conclusion

Results suggest that high-fat margarines tend to be harder and more adhesive, as well as more preferred, than their counterparts with lower fat contents, as shown by instrumental texture analysis and acceptance test. However, the texture attribute seemed to be less critical than the flavor attribute to explain overall impression ratings. Furthermore, the sample containing only 30% of lipids, which was mostly associated with negative flavor descriptors, i.e., vegetable oil aroma, oily flavor, and rancid flavor, was also the least preferred among all samples, scoring the lowest averages for flavor, texture, and overall impression. These findings highlight the key role of fat in flavor release, as lipids act as carriers for tastants and odorants during the oral processing of fat-based products, especially edible water-in-oil emulsions. Several studies have dealt with the effects of lipids composition and concentration on the release rate of odor-active compounds that characterize flavor in various food systems [48,49,50,51,52].

This study also shows that it is essential to conduct focused sensory research aimed at obtaining margarines with reduced lipid levels, while still meeting consumers' acceptance criteria. Additionally, conducting instrumental studies to evaluate the physicochemical characteristics of margarines in conjunction with sensory studies is recommended. This integrated approach can offer comprehensive insights to guide formulation adjustments and process enhancement, facilitating the development of nutritionally improved margarines while considering the potential effects of lipid reduction upon consumer preferences.