Physicochemical, Microbiological, Proximate, and Consumer Characterization of Traditional Tenate Cheese in Two Mexican Regions

Abstract

Tenate cheese is a traditional Mexican pressed semi-hard cheese made from raw cow’s milk and wrapped in palm fiber. The characterization of this cheese remains scarce. This study presents an exploratory characterization of a single production batch of traditional Tenate cheese obtained from one artisanal producer, providing preliminary information on its physicochemical, microbiological, and proximate analyses, combined with consumer evaluation. The latter was analyzed using hierarchical cluster analysis (HCA) as an exploratory segmentation tool. Tenate cheese was characterized as a semi-hard cheese with active lactic fermentation, a lactic aroma, acidic and umami flavors, and a firm, granular texture. Microbiological analyses showed the absence of coliforms, enterobacteria, and Staphylococcus aureus among the microorganisms evaluated, whereas yeast counts exceeded the regulatory limit. As major foodborne pathogens were not included in the microbiological assessment, the overall microbiological safety of the product could not be confirmed. A total of 318 consumers from Aguascalientes (AGS, n = 149) and the Guadalajara Metropolitan Area (GMA, n = 169) evaluated the product using hedonic and Just-About-Right scales. Consumers from AGS reported significantly higher liking scores than those from GMA. Penalty analysis identified insufficient softness as the main attribute associated with lower liking in AGS, whereas low flavor intensity and weak aftertaste reduced acceptance in GMA. Hierarchical cluster analysis (HCA) identified three consumer segments in each location, revealing distinct preference patterns linked to regional expectations. The main contributions of this study are threefold. First, it contributes to the limited scientific knowledge available on Tenate cheese by providing a comprehensive characterization of the analyzed sample. Second, it shows that consumer acceptance differed between two regional markets, comparing two university-affiliated consumer groups, highlighting the value of consumer segmentation for product positioning. Third, it proposes and applies an integrated framework combining physicochemical, microbiological, nutritional, sensory, and consumer segmentation analyses that can be applied to the study of other artisanal cheeses.

1. Introduction

Tenate cheese is a traditional Mexican cheese that originated in the haciendas of Tulancingo, Hidalgo, and Tlaxco, Tlaxcala, at the end of the nineteenth century. It is produced from raw cow’s milk and takes its name from the tenate, a woven palm basket traditionally used as a mould during cheese manufacture. The cheese is wrapped in this palm basket, which contributes to its distinctive aroma and flavor (Figure 1). It is a pressed, semi-hard cheese that can be preserved for more than eight days at room temperature and is recognized as one of Mexico’s genuine cheeses [1].

Traditional cheeses are closely linked to local culture and to cheesemaking practices that have been preserved for generations. Their sensory attributes, particularly aroma and flavor, contribute to their typicality and are highly valued by consumers [2,3]. Furthermore, these products constitute part of Mexico’s gastronomic and cultural heritage, where approximately 40 varieties of authentic cheeses have been identified [4].

Despite their cultural importance, traditional cheeses produced from raw milk face challenges related to microbiological safety and the inherent variability of the manufacturing process. Several studies have indicated that the safety of these products depends largely on the implementation of good manufacturing practices rather than exclusively on the use of pasteurized milk [5,6]. In addition, indigenous lactic acid bacteria naturally present in traditional cheeses may contribute to both microbiological safety and the development of their characteristic sensory properties [7]. Furthermore, standardizing quality attributes and characterizing sensory properties are important for preserving and valorizing these products.

The characterization of traditional cheeses requires a comprehensive approach encompassing physicochemical, proximate, microbiological, and sensory aspects. In addition, hierarchical cluster analysis techniques have emerged as useful tools for analyzing complex consumer-generated data and identifying groups with similar characteristics and preferences based on multidimensional information [8,9].

Therefore, the aim of this study was to characterize Tenate cheese through physicochemical, proximate, and microbiological analyses, complemented by consumer tests and hierarchical cluster analysis. The study contributes to the scientific characterization of Tenate cheese and provides an integrated framework that may be applied to other traditional artisanal cheeses.

2. Materials and Methods

2.1. Cheese Sampling and Transportation

Tenate cheese samples were obtained from a traditional artisanal producer located in Tulancingo, Hidalgo, Mexico. The producer provided sixteen cheese units from a single production batch. Eight cheese units were shipped to Aguascalientes and eight to the Guadalajara Metropolitan Area (GMA) for consumer evaluation. Samples were transported in insulated coolers containing frozen refrigerant packs to maintain a temperature of approximately 4 °C. Upon arrival, all samples were refrigerated until analysis. Cheese from a batch was selected for physicochemical, microbiological, and proximate analyses, which were conducted in Aguascalientes. All determinations were performed in analytical triplicate using subsamples obtained from a single production batch and producer. The remaining cheese units were used for sensory evaluations in both study locations.

2.2. Physicochemical Analyses

Several physicochemical analyses were performed, including pH, titratable acidity, moisture content, and color. A single cheese unit from one production batch was used for all analyses. For pH determination, 10 g of cheese were homogenized with 10 mL of distilled water until a uniform mixture was obtained, and the homogenate was measured immediately at room temperature (approximately 25 °C) using the potentiometric method according to NMX-F-317-NORMEX-2013 [10] with a LAQUA PH1200 potentiometer (HORIBA Advanced Techno Co., Ltd., Kyoto, Japan). Titratable acidity was determined by titration according to NOM-243-SSA1-2010 [11].

Color was evaluated using a portable Pantone CAPSURE™ model RM200+B spectrocolorimeter (X-Rite Incorporated, Grand Rapids, MI, USA). Prior to analysis, the palm wrapping was removed, and measurements were taken directly at different locations on the external cheese surface. The instrument was calibrated according to the manufacturer’s instructions before use. Measurements were performed under a D65 illuminant and a 10° standard observer angle. Color parameters were expressed as L*, a*, and b* coordinates of the CIELAB color system.

All physicochemical analyses were performed in analytical triplicate using samples obtained from a single production batch and producer; the results were expressed as mean ± standard deviation. Therefore, the reported values represent analytical replicates rather than biological or production replicates.

2.3. Microbiological Analyses

The evaluated microbial groups included both technological microorganisms and hygiene indicator microorganisms commonly used for food quality assessment. Aerobic mesophilic bacteria provide an overall estimate of the microbial load present in the product, whereas coliforms, Enterobacteriaceae, and Staphylococcus aureus are widely recognized as indicators of hygiene and sanitation conditions during food processing. In Mexico, these microbial indicators are commonly used within the framework of Good Manufacturing Practices (GMP) and sanitary quality evaluation to verify the effectiveness of cleaning, handling, and processing operations in dairy products. Their use is also supported by NOM-243-SSA1-2010 [11], which establishes the microbiological specifications and sanitary requirements for milk, dairy products, and processed foods, and considers these microbial groups as indicators of hygienic quality and potential contamination. Lactic acid bacteria (LAB) were quantified for their role in fermentation and sensory development, while molds and yeasts were evaluated for their potential influence on product quality and shelf life.

The material was sterilized in a vertical autoclave model EV-24 (Novatech, San Pedro Tlaquepaque, Jalisco, Mexico) at 121 °C, 15 min. Microbiological analyses were performed in a laminar flow hood (Novatech, Querétaro, Mexico). Ten grams of cheese were homogenized with 90 mL of sterile distilled water to obtain the initial dilution (10⁻¹). Subsequently, serial decimal dilutions ranging from 10⁻¹ to 10⁻⁷ were prepared.

The growth of Mesophilic aerobic bacteria, coliforms, Enterobacteriaceae, Staphylococcus aureus, Lactic Acid Bacteria, molds, and yeasts was evaluated using NEOGEN Petrifilm^® Aerobic Count (AC), Coliform Count (CC), Enterobacteriaceae Count (EB), Staph Express Count (STX), Yeast and Mold Count (YM), and Lactic Acid Bacteria Count (LAB) plates (NEOGEN Corporation, Lansing, MI, USA). One milliliter of the appropriate dilutions was inoculated onto the corresponding Petrifilm^® plates according to the manufacturer’s instructions.

Aerobic Count plates were incubated at 35 ± 1 °C for 48 ± 3 h; Coliform Count plates at 35 ± 1 °C for 24 ± 2 h; Enterobacteriaceae Count and Staph Express Count plates at 37 ± 1 °C for 24 ± 2 h; Yeast and Mold Count plates at 25 ± 1 °C for 48–72 h; and Lactic Acid Bacteria Count plates at 30 ± 1 °C for 48 ± 3 h under anaerobic conditions. Viable cell counting and result interpretation were performed according to the manufacturer’s instructions. Microbial counts were calculated from plates within the countable range established by the corresponding Petrifilm^® method. Samples in which no colonies were detected were reported as ND (not detected), indicating counts below the detection limit of the method.

All microbiological analyses were performed in analytical triplicate using samples obtained from the cheese unit of Tenate cheese of a single production batch and producer, and the results were expressed as mean ± standard deviation.

2.4. Proximate Analysis

Proximate analysis was conducted at the Food Technology Laboratory of the Universidad Autónoma de Aguascalientes, Mexico. A 500 g Tenate cheese unit was submitted in its original commercial presentation, including the palm-fiber Tenate wrapping. Prior to analysis, the latter was removed, and only the cheese was used for the determinations. The results were expressed on a wet basis. Moisture content was determined according to NMX-F-083-1986 [12], ash content according to NMX-F-066-S-1978 [13], ether extract by the Goldfish method according to NMX-F-089-S-1978 [14], protein content by the Dumas method using a LECO FP-528 analyzer, crude fiber according to NMX-F-090-S-1978 [15], and carbohydrate content by difference. Sodium content was estimated from the producer’s formulation, based on the amount of salt added per liter of milk during cheese manufacture. All proximate analyses were performed in analytical triplicate using samples obtained from the cheese unit of a single production batch and producer, and the results were expressed as mean ± standard deviation.

2.5. Consumer Test

2.5.1. Participants

Two consumer panels were conducted simultaneously in March 2026, with participants recruited from two campuses of Universidad Panamericana, each located in a different metropolitan area of Mexico: Guadalajara Metropolitan Area (GMA, Guadalajara, Mexico) and Aguascalientes (AGS, Aguascalientes, Mexico). Importantly, this study is a two-location consumer evaluation: both groups shared the same institutional environment and differences between the two consumer samples, but they were from distinct geographic regions. The participants were students or teachers from Universidad Panamericana. This University offers a degree related to Gastronomy matters.

The selection of these locations was based on clear demographic and market differences: the Guadalajara Metropolitan Area (GMA) is one of the largest, most urbanized, and diverse food markets in Mexico, while Aguascalientes (AGS) is a smaller metropolitan area with a stronger regional identity and lower urbanization. Interpretations regarding cultural influence are exploratory and limited to the locations’ demographic and market contrasts.

The GMA panel comprised 169 consumers (137 women, 31 men, and 1 participant who preferred not to disclose gender), while the AGS panel comprised 149 consumers (84 women, 64 men, and 1 participant who preferred not to disclose gender). Eligibility criteria included being at least 18 years old and reporting regular cheese consumption. Prior to participation, all consumers provided written informed consent.

Because participants were recruited from university communities, the sample should be considered a convenience sample. It is not a representative sample of the general population in either region. Consequently, the results should not be generalized to all consumers in GMA or AGS. There were also differences in the demographic composition of the panels, particularly the higher proportion of women in the GMA sample. This may have influenced hedonic responses and should be taken into account when interpreting regional differences.

The deliberate focus on university-affiliated young adults was based on the study’s objective: evaluating Tenate cheese acceptance among a young consumer segment relevant for future traditional product markets. This population was selected for its significance in the potential preservation and commercialization of traditional foods. Thus, the study is designed to compare young adults from two different geographic regions, not to represent entire regional populations or broader national trends.

2.5.2. Sample

Each participant evaluated a standardized portion of Tenate cheese (approximately 3cm × 3 cm and 35 g). All samples originated from the same production batch to ensure product uniformity throughout the study. Following manufacture, cheeses were stored under refrigeration (4 ± 1 °C) until sensory evaluation. Samples destined for both study locations were transported under refrigerated conditions in insulated coolers equipped with ice packs, maintaining a temperature below 8 °C during transit. Consequently, GMA and AGS evaluated identical product lots.

Prior to serving, samples were removed from refrigeration and allowed to equilibrate for approximately 30 min to reach a serving temperature of 20–22 °C. Individual portions were placed in plastic containers with lids and identified using random three-digit codes. Water and low-salt whole-wheat crackers were provided as palate cleansers between evaluations. The sensory sessions were conducted under similar environmental conditions at both locations.

2.5.3. Procedure

The evaluation questionnaire consisted of the following sections.

The first section collected demographic information, including gender, age group, and place of residence. These data were used to characterize the study population and verify the comparability of the consumer panels.

The second section consisted of a hedonic acceptance test. Consumers evaluated overall liking, color, appearance, odor, flavor, flavor intensity, texture, freshness, and overall quality using a 7-point hedonic scale, where 1 = “dislike very much”, 4 = “neither like nor dislike”, and 7 = “like very much”.

The third section consisted of a Just-About-Right (JAR) evaluation. Consumers assessed the adequacy of selected sensory attributes using a 5-point JAR scale for seven attributes: color, traditional appearance, aroma, flavor, salt, softness, and aftertaste. The anchors for all attributes ranged from “much less than desired” (1) to “much more than desired” (5), with the midpoint (3) corresponding to “just-about-right” (JAR).

The fourth section included sensory characterization of the cheese through consumer perception of selected sensory descriptors, whereas the fifth section included attitudinal and purchase-intent questions related to authenticity, cultural heritage, and commercialization potential. Data from these sections were collected as part of a broader research project but are not included in the analyses presented in this manuscript.

Sensory descriptors included in the questionnaire, including umami flavor, were evaluated according to consumers’ own perceptions and prior sensory experience. No formal sensory training or reference standards were provided, as the study was designed as a consumer test rather than a descriptive analysis with trained panelists.

All evaluations were completed individually under controlled sensory testing conditions. The study was conducted in accordance with the Declaration of Helsinki and approved by the Institutional Review Board of Universidad Panamericana (ING-DNA-01-2025-2026).

2.6. Data Analysis

Physicochemical, microbiological, and proximate analyses. All analyses were performed in triplicate, and the mean and standard deviation are presented.

Consumer data. To evaluate the differences in hedonic scores between the two study locations, AGS and GMA, an independent samples t-test was used, since the sample sizes were n ≥ 30, so the sample means tend toward a normal distribution, according to the Central Limit Theorem [16].

In Just-About-Right (JAR) analysis, first consumer responses are grouped into three categories: “Too less”, “Just about right” (JAR), and “Too high”. In this case, the category “Too less” groups “Much less than desired” and “Less than desired,” and the category “Too high” groups “Much more than desired” and “More than desired”

For each attribute, the mean overall liking score is calculated for consumers within each category. The Mean Drop (Penalty) is then determined as the difference between the average liking score of the JAR group and that of the non-optimal groups (“Too less” or “Too high”). This metric quantifies the reduction in overall liking associated with deviations from the ideal attribute intensity [17]. Each attribute generates two independent penalties, one for each side of the JAR scale. The p-values corresponded to comparisons of the mean JAR scores with the means of the two non-optimal groups [18].

The practical relevance of each sensory attribute was evaluated using two commonly accepted criteria: (i) at least 20% of consumers selecting a non-JAR category, and (ii) a mean drop close to or greater than 1. Attributes meeting both criteria were considered potential drivers of disliking and opportunities for product optimization. Because penalty analysis is primarily exploratory and diagnostic, greater emphasis was placed on practical significance rather than on statistical significance alone.

All analyses were performed using XLSTAT Sensory software (version 2025.2.0, Addinsoft, Paris, France).

JAR data were analyzed using penalty analysis to identify attributes associated with reduced overall liking. Mean drop (MD) was the difference between mean liking scores for “just-about-right” and non-optimal attribute ratings. Positive MD values indicated decreased liking for non-optimal perceptions. Both MD size and the proportion of affected consumers determined each attribute’s practical importance.

Differences in hedonic ratings between consumers from AGS and the GMA were assessed using independent-samples t-tests. Although hedonic ratings are collected using ordinal scales, they are commonly treated as interval-level data in sensory and consumer research, particularly when sample sizes are large. Previous studies have demonstrated that parametric methods, including t-tests and analysis of variance (ANOVA), are robust for analyzing hedonic data and generally yield conclusions comparable to those obtained with nonparametric approaches. Therefore, independent-samples t-tests were considered appropriate for comparing mean liking scores between the two consumer groups. Statistical significance was established at p < 0.05.

Exploratory multivariate techniques applied to data analysis. The second approach of the JAR test-based data analysis was developed using exploratory multivariate techniques. The goal of this process was to identify patterns in the data collected from consumer interviews. Agglomerative Hierarchical Cluster Analysis (HCA) was applied to group consumers based on multidimensional similarities [19]. HCA facilitates the visualization of proximity relationships through dendrograms, enabling the identification not only of specific market clusters but also of the subjective “distance” between the preferences of diverse consumer niches.

Consumer clusterization through Agglomerative Hierarchical Cluster Analysis (HCA). HCA was implemented using an agglomerative approach to identify preference patterns and organize consumers based on their multidimensional sensory perceptions. This process consisted of three fundamental technical stages.

The applied clustering method (HCA) constructs a cluster hierarchy. The steps followed in this analysis are: data loading and preprocessing; dimensionality reduction; determination of the optimal k value (based on the Silhouette score); determination of the final k value; training the k-means model with the optimal k; and assigning each instance in the dataset to a cluster.

During data loading and preprocessing, numerical columns were selected, null values were filled with the mean, and the data were scaled. StandardScaler was used for this purpose. This technique’s main function is to standardize the features of a dataset so that each feature has a mean of 0 and a standard deviation of 1.

For each characteristic (column) in the data, StandardScaler performs the following transformation:

x_scaled = (x − μ)/σ

where

x: Is the original value of an observation for a specific characteristic.

μ: This is the average value of that feature across the entire training dataset.

σ: This is the standard deviation of that feature across the entire training dataset.

Standardization ensures that all features contribute equally by allowing the comparison of features that originally had completely different units or scales.

After standardizing the data, the values for each feature are transformed: positive values indicate that the original value was above the feature’s average, and negative values indicate that it was below the average. The magnitude of the value (how far it is from 0) indicates how atypical or extreme the observation is for that feature. The larger the absolute value, the more atypical the observation.

To reduce noise and dimensionality, Principal Component Analysis (PCA) was applied. By retaining 80% of the explained variance, the scaled data were transformed into a lower-dimensional space. The distance matrix was then calculated to determine which points are ‘close’ based on the Euclidean distance in the PCA-reduced space.

To execute the hierarchical clustering process, the link function was used alongside Ward’s criterion to establish the cluster hierarchy. This method aims to reduce intra-cluster variance by integrating groups that yield the minimal increase in cumulative variance, thereby generating the linkage matrix.

After creating the linkage matrix, a dendrogram is constructed to visualize the cluster hierarchy. In this plot, individual observations appear on the X-axis. The Y-axis shows the dissimilarity distance where groupings merge. Longer vertical branches indicate higher dissimilarity between the merged clusters.

To delineate specific consumer niches, the established cluster hierarchy is truncated at a designated vertical distance. Several criteria govern the selection of this cut-off point, which ultimately defines the total number of groupings. First, visual assessment of the dendrogram remains a prevalent approach; here, the dissimilarity threshold on the Y-axis is identified, where a horizontal intersection yields a relevant set of clusters. This involves detecting substantial “jumps” or elbows in the vertical branches, indicating that the merging groups are highly dissimilar. Second, the desired k value may be pre-determined through preliminary K-means analysis, facilitating the calculation of the precise distance required to achieve that count. Finally, statistical metrics, such as the Silhouette score or the Davies-Bouldin index, provide a quantitative evaluation of cluster quality across different values of k.

Choosing the optimal number of consumer groups involves more than specific statistical metrics. It also requires blending quantitative factors with the subjective meaning and interpretability of the sensory profiles produced.

3. Results and Discussion

3.1. Physicochemical Analyses

In

Table 1. Physicochemical properties of Tenate cheese, expressed as mean ± standard deviation.

Parameter	Result (Mean ± SD)
pH	5.74 ± 0.142
Titratable acidity (% lactic acid)	0.91 ± 0.03
Colorimetry (CIELAB coordinates)	L* 82.73 ± 1.83 a = −0.36 ± 1.48 b = 20.97 ± 4.95

, we observe the physicochemical analysis, including pH, titratable acidity, moisture, and colorimetry. Physicochemical analysis revealed a low pH (5.74) and a high titratable acidity (0.91%) in Tenate cheese, which can be attributed to lactose fermentation by native Lactic Acid Bacteria (LAB) associated with the use of unpasteurized milk [20], resulting in the production of lactic acid as a metabolic product. The pH value falls within the range reported for Mexican cheeses, such as Chihuahua cheese, which is 5.5–5.8 [21]. Based on its moisture on a fat-free basis (MFFB = 59.5%), calculated from the moisture and fat contents determined in this study, Tenate cheese can be classified as a firm (semi-hard) cheese according to the General Standard for Cheese CXS 283-1978 of the Codex Alimentarius [22]. Regarding colorimetry, the luminosity value obtained was L* = 82.73 ± 1.83, which was slightly lower than that reported for other artisan cheeses, likely due to the lower moisture content of Tenate cheese [2]. In addition, the positive b* value (20.97 ± 4.95) suggests the predominance of yellow tones associated with carotenoids and milk fat content.

3.2. Microbiological Analyses

Microbiological analysis of Tenate cheese (

Table 2. Microbiological analyses of Tenate cheese are expressed as mean ± standard deviation, and microbiological limits were established according to NOM-243-SSA1-2010.

Microorganism	Result log CFU/g (Mean ± SD)	Limit log CFU/g (NOM-243-SSA1-2010)
Mesophilic aerobic bacteria	9 ± 0.1	Not specified
Coliforms	ND	Not specified
Enterobacteria	ND	Not required
Staphylococcus aureus	ND	3
Molds	1.5 ± 0.5	3
Yeasts	5.4 ± 0.1	3
Lactic acid bacteria	8.9 ± 0.01	Not required

ND: no colonies were observed on countable plates at the lowest dilution analyzed and were therefore considered below the detection limit of the method.

) showed high aerobic counts and a LAB concentration of 9 log CFU/g, consistent with those reported for artisan cheeses [23,24]. These results indicate active fermentation, in which LAB plays a key role by producing compounds that contribute to the characteristic sensory properties of cheese and support pathogen inhibition [25]. The absence of detectable coliforms, Enterobacteriaceae, and Staphylococcus aureus indicates low levels of these microbial groups and may reflect acceptable hygienic conditions during processing. However, these results should be interpreted only for the microorganisms analyzed in the present study and do not constitute a comprehensive microbiological safety assessment of the product. This may also be linked to the high LAB content, as these microorganisms produce lactic acid, which promotes product acidification and preservation and generates antimicrobial compounds [26]. Mold counts were within the microbiological specifications established by NOM-243-SSA1-2010, with a value of 2 log CFU/g [11]. However, yeast counts exceeded the established limit, reaching 5 log CFU/g. This may be attributed to the presence of yeasts in the curd of cheeses produced from raw milk, whose high concentrations are associated with artisanal production and storage conditions [27]. Although the evaluated microbial indicators provide useful information regarding product quality and hygiene, additional analyses, including relevant foodborne pathogens such as Salmonella spp., Listeria monocytogenes, and Shiga toxin-producing Escherichia coli, would be necessary to achieve a more comprehensive assessment of the microbiological safety of Tenate cheese.

Another potential source of cross-contamination is the Tenate container used for the cheese, since, being made from palm fibers, it may act as a porous surface that promotes the retention and transfer of microorganisms [28]. Nevertheless, this container undergoes an artisanal thermal treatment with hot water; therefore, this procedure should be microbiologically validated by establishing specific time and temperature conditions. To reduce yeast counts and comply with the limits established by the standard, good manufacturing practices (GMP) should be rigorously strengthened within the processing plant, and the Hazard Analysis and Critical Control Points (HACCP) system should be implemented to ensure food safety [6,29].

The high aerobic count of 9 log CFU/g may be associated with the indigenous lactic microbiota naturally present in artisanal fermented cheeses during the ripening process [30].

3.3. Proximate Analysis

Proximate analysis (

Table 3. Proximate composition of Tenate cheese per 100 g.

Proximate Composition Content (g/100 g)	(Mean ± SD)
Moisture	46.44 ± 1.52
Protein	25.37 ± 0.05
Fat	21.89 ± 0.2
Dietary Fiber	0.00 ± 0.00
Carbohydrates	2.20 ± 0.27
Ash	3.70 ± 0.17

, Figure 2 revealed 25.37% protein, 21.89% fat, and 46.44% moisture. These values comply with the specifications established in NOM-223-SCFI/SAGARPA-2018, which states that cheeses without a specific standard must contain a minimum of 10% protein, a maximum of 80% moisture, and a fat percentage consistent with that declared on the labeling [31].

The low carbohydrate content observed (2.2 g/100 g) may be associated with lactose utilization by lactic acid bacteria during fermentation, leading to lactic acid production [26].

Based on the estimated sodium and saturated fat contents and according to the criteria established in NOM-051-SCFI/SSA1-2010, Tenate cheese would be expected to display the warning seals “Excess Calories,” “Excess Sodium,” and “Excess Saturated Fat” (Figure 3) [32]. The product contained 307.29 kcal per 100 g, an estimated sodium content of 786 mg/100 g, and an estimated 44.87% of its total caloric content was derived from saturated fat. Sodium content was estimated from the amount of salt added during cheese manufacture, whereas saturated fat content was calculated from the total fat content and the typical fatty acid composition of milk fat reported by Rashidimehr et al. [33]. Therefore, these values and their corresponding labeling implications should be considered preliminary and require analytical confirmation before definitive nutritional labeling conclusions can be established.

3.4. Consumer Tests

The study included consumers from only two Mexican regions, limiting generalizability to other cultural or international contexts. The panels were mainly university-affiliated and referred to the same university. However, this study was performed in the same University, Universidad Panamericana. The GMA and AGS panels comprised 169 participants (81.1% women, 18.3% men, and 0.6% who preferred not to disclose) and 149 participants (56.4% women, 43% men, and 0.7% who preferred not to disclose), respectively. The age groups were 18–25 years old: 82.2% in GMA and 92% in AGS; 26–34 years old: 3.6% in GMA and 2% in AGS; and more than 35 years old: 14.2% in GMA and 6% in AGS. All students are very similar in socioeconomic status across the two regions; the only difference is where they live: Aguascalientes (AGS) versus Guadalajara (GMA), a medium city (1.4 million inhabitants) vs. (5.5 million inhabitants).

Our results show that consumer acceptance differed significantly across regions; consumers in AGS showed higher overall liking scores than those in GMA, particularly for the texture and flavor attributes.

3.4.1. JAR Tables (Just-About-Right)

In Figure 3a,b, the Just-About-Right (JAR) graphs are used to show the method to assess whether consumers perceive the product’s sensory attributes at an optimal level. In these graphs, we can observe whether a specific attribute is perceived as “too low,” “just-about-right (JAR),” and “too high.” The main purpose of JAR graphs is to identify which attributes require adjustment to improve product acceptance [34]. In the case of AGS, the cheese was lacking sufficient softness, and in the case of GMA, the flavor intensity and aftertaste were insufficient.

3.4.2. Penalty Analysis

Penalty analysis identifies which sensory attributes reduce overall consumer liking when perceived as either “too high” or “too much” compared to the Just-About-Right (JAR) level. For Aguascalientes (AGS), analysis showed texture softness was the only attribute that significantly decreased overall liking (p < 0.05). Specifically, 30.87% of consumers perceived the cheese as less soft than their JAR level, indicating a firmer-than-preferred texture (Figure 4a,b). This deviation from optimal softness measurably lowered consumer acceptance.

Penalty analysis revealed that different consumer groups have specific sensory optimization needs. For AGS consumers, overall liking dropped significantly when the cheese lacked softness, highlighting the central role of texture in product acceptance in this region. Although Tenate cheese traditionally has a semi-firm, compact structure due to pressing and reduced moisture, excessive firmness may compromise palatability and reduce perceived freshness [35].

Conversely, a relatively large proportion of GMA consumers perceived flavor intensity and aftertaste as lower than desired. These attributes may represent opportunities for product optimization, although only aftertaste reached statistical significance, and both mean drop values remained below the conventional threshold of one. This suggests they prefer cheeses with more aromatic persistence and flavor complexity. Previous studies have shown that flavor intensity is a major driver of artisan cheese preference and purchasing decisions. It often surpasses price and packaging attributes in importance [3].

Both consumer groups showed relatively low penalties for excessive aroma or saltiness. This suggests that the strong sensory intensity of Tenate cheese is generally well tolerated. It may contribute positively to its perceived typicity. Similar behavior has been described for traditional cheeses. Consumers often associate pronounced sensory attributes with authenticity, territorial identity, and artisan production systems [2].

The observed differences across regions support the need to tailor commercialization strategies to consumer segmentation. Texture optimization may improve acceptance in AGS. For GMA consumers, enhancing flavor persistence and aromatic complexity may be more effective.

3.4.3. Attributes Associated with Lower Liking

Referring to

Table 4. Penalty analysis with mean overall liking score (7-point hedonic scale) of AGS.

Variable	Level	%	Mean (Overall Liking)	Mean Drops	Penalties	p-Value
	Too Less	5.37%	4.000	1.068
Color	JAR	79.19%	5.068		−0.384	0.115
	Too High	15.44%	5.957	−0.889
	Too Less	9.40%	4.071	1.063
Traditional appearance	JAR	69.80%	5.135		−0.043	0.842
	Too High	20.81%	5.677	−0.543
	Too Less	25.50%	4.632	0.556
Aroma	JAR	53.69%	5.188		0.086	0.665
	Too High	20.81%	5.677	−0.490
	Too Less	14.09%	4.143	1.037
Flavor	JAR	59.73%	5.180		0.080	0.693
	Too High	26.17%	5.615	−0.436
	Too Less	19.46%	4.655	0.590
Salt	JAR	65.77%	5.245		0.284	0.173
	Too High	14.77%	5.364	−0.119
	Too Less	30.87%	4.609	0.763
Softness	JAR	52.35%	5.372		0.470	0.017
	Too High	16.78%	5.440	−0.068
	Too Less	13.42%	4.300	0.929
Aftertaste	JAR	70.47%	5.229		0.274	0.206
	Too High	16.11%	5.500	−0.271

and

Table 5. Penalty analysis with mean overall liking score (7-point hedonic scale) of GMA.

Variable	Level	%	Mean (Overall Liking)	Mean Drops	Penalties	p-Value
	Too Less	11.83%	4.250	0.645
Color	JAR	67.46%	4.895		0.131	0.488
	Too High	20.71%	5.057	−0.162
	Too Less	21.89%	4.459	0.437
Traditional appearance	JAR	62.72%	4.896		0.118	0.518
	Too High	15.38%	5.231	−0.335
	Too Less	39.05%	4.652	0.234
Aroma	JAR	41.42%	4.886		0.057	0.750
	Too High	19.53%	5.182	−0.296
	Too Less	27.22%	4.043	0.957
Flavor	JAR	49.11%	5.000		0.291	0.100
	Too High	23.67%	5.475	−0.475
	Too Less	25.44%	4.767	0.095
Salt	JAR	60.36%	4.863		0.027	0.882
	Too High	14.20%	4.958	−0.096
	Too Less	47.93%	4.716	0.324
Softness	JAR	43.79%	5.041		0.335	0.059
	Too High	8.28%	4.643	0.398
	Too Less	22.49%	4.158	0.852
Aftertaste	JAR	57.99%	5.010		0.376	0.035
	Too High	19.53%	5.182	−0.172

, which establish the mean drop vs. % of consumers, it can be observed how specific sensory attributes affect overall liking and how many consumers are impacted. This table comes from the JAR and penalty analysis. It mainly measures the mean drop (MD), which indicates how much the overall liking score decreases when consumers perceive an attribute as “too less” or “too high” relative to the JAR level. A larger MD indicates a stronger negative impact on acceptance. Commonly, a mean drop value of 1 is considered the threshold for a significant decrease in liking. The % of consumers represents those who perceived the attribute as non-optimal. Often, at least 20% of consumers are required in a category for the penalty to be considered relevant [34]. Penalty analysis showed that sensory drivers of disliking differed between the two regional consumer samples evaluated in this study. In AGS, insufficient softness reduced acceptance. In GMA, low flavor intensity and weak aftertaste lowered consumer preference. These findings should be interpreted as exploratory differences between two regional convenience samples rather than evidence of sociocultural effects.

Penalty analysis revealed that the sensory drivers of disliking differed between the two consumer groups, suggesting that consumer expectations may differ between the two regional consumer samples evaluated. In AGS, Table 4 shows that too little softness affected 30.87% of consumers (MD = 0.763), and consumers perceived that Tenate cheese had a hard texture, which contributed to lower acceptance (p < 0.05). Texture-related characteristics have been recognized as important determinants of cheese liking because they influence mouthfeel, perceived freshness, and overall quality. Similar findings have been reported for traditional and artisanal Mexican cheeses, whose sensory identity is strongly associated with specific textural attributes developed through traditional manufacturing practices [30].

In contrast, consumers from GMA identified flavor intensity and aftertaste as attributes that may warrant further optimization (Table 5). Specifically, 27.22% of consumers perceived flavor as “too little” (MD = 0.957), whereas 22.49% considered the aftertaste insufficient (MD = 0.852).

Although the MD values in both regions were slightly below the conventional threshold of 1.0, Popper [34] noted that the appropriate cutoff depends on the liking scale used. When mean drops are calculated from a seven-point hedonic scale, lower cutoff values may still be considered meaningful. Furthermore, these attributes exceeded the 20% consumer criterion, indicating relevant opportunities for product improvement. These results indicate that a considerable proportion of GMA consumers expected a more intense and persistent sensory profile than that provided by the evaluated cheese.

The importance of flavor intensity and aftertaste aligns with previous studies on traditional cheeses, which have identified flavor-related attributes as major drivers of consumer acceptance. Traditional Mexican cheeses such as Cotija, Chihuahua, Bola de Ocosingo, and Poro cheese are characterized by distinctive sensory profiles, including pronounced dairy notes, acidity, and persistent flavor, attributes that consumers frequently associate with authenticity and product quality [30]. Consequently, the product perceived as lacking flavor intensity may receive lower hedonic scores, particularly among consumers familiar with traditional artisanal dairy products.

The differences observed between AGS and GMA consumers may be explained by differences in cultural familiarity and consumption habits. Hidalgo-Milpa et al. [36] reported that consumers with stronger ties to traditional food systems tend to value authenticity-related sensory characteristics and show greater acceptance of traditional cheeses than consumers whose preferences are shaped by industrialized food markets. Likewise, studies on artisanal cheeses have shown that previous experience, cultural background, and product familiarity significantly influence acceptance and preference patterns.

This interpretation is consistent with the broader literature on traditional Mexican cheeses, which emphasizes that the preservation of their sensory identity is closely linked to local culture, regional consumption habits, and consumer familiarity with artisanal products [30].

Attributes that significantly (p < 0.05) decreased overall liking, flavor, and aftertaste were the most notable. Specifically, 39% of consumers rated flavor as “too low,” indicating that the product was perceived as less intense than their JAR level. Similarly, 22% of consumers evaluated aftertaste as “too low”. In both cases, the mean drop in liking (penalty) was close to 1 unit.

3.4.4. Overall Liking

When comparing the two cities, AGS and GMA (Figure 5), Tenate cheese was rated higher in terms of overall liking in AGS than in GMA. Significant differences (p < 0.05) were observed across all evaluated liking attributes, except for appearance and flavor intensity, for which no significant differences were found between the two cities, indicating similar levels of consumer acceptance for these attributes in both cities.

When comparing the two cities, Tenate cheese received significantly higher overall liking scores AGS than in GMA. Significant differences (p < 0.05) were observed for most hedonic attributes. However, appearance and flavor intensity did not differ significantly between regions. This suggests that visual perception and perceived flavor strength were similarly accepted by consumers in both locations. The differences in overall liking indicate that other sensory dimensions—such as texture, aroma integration, and flavor persistence—may have contributed to the distinct consumer responses observed between regions.

One possible explanation is differences in familiarity with traditional dairy products; however, familiarity was not measured in this study, and therefore this interpretation remains speculative. Consumer studies on traditional cheeses have shown that familiarity and habitual consumption positively influence hedonic responses. Consumers develop sensory expectations that align with the characteristic attributes of local products [36]. Familiar consumers are generally more tolerant of sensory variability. They are also more likely to associate distinctive sensory characteristics with authenticity and quality. This phenomenon has been described as the “familiarity effect,” whereby repeated exposure increases product acceptance and perceived authenticity.

The present findings are consistent with studies conducted on traditional Mexican cheeses. These studies reported that consumers with greater cultural proximity to artisanal products exhibit higher acceptance of sensory attributes such as acidity, firmness, pronounced lactic notes, and persistent flavor [30]. Similar observations have been reported for traditional European cheeses with protected geographical identity. In these studies, consumer liking is strongly associated with familiarity and recognition of product typicity rather than with conformity to standardized industrial sensory profiles [37]. In this context, Tenate cheese is perceived not only as a food product but also as a culturally embedded product. Its acceptance depends on prior exposure and sensory expectations.

In contrast, consumers from GMA exhibited lower liking scores across several attributes. This finding may reflect differences in food environments and consumption habits associated with urbanization. Previous studies have shown that consumers living in highly urbanized areas are more frequently exposed to industrial dairy products. These products are characterized by milder flavor profiles, lower acidity, softer textures, and greater sensory standardization [30]. Consequently, the more distinctive sensory characteristics of artisanal cheeses may be perceived as less familiar or less desirable by consumers whose preferences have been shaped by industrial food systems.

The absence of significant differences in appearance and flavor intensity is also noteworthy. Visual attributes are often the first determinants of consumer acceptance. They can serve as indicators of product quality across regional differences. Likewise, similar ratings of flavor intensity suggest that differences in overall liking were not driven by perceptions of. Instead, they were driven by the interpretation and appreciation of the flavor’s sensory profile. This observation supports previous findings indicating that consumer acceptance of traditional cheeses depends not only on the intensity of sensory attributes but also on their congruence with culturally learned expectations and perceptions of authenticity [36,37].

Overall, these results reinforce the notion that acceptance of Tenate cheese is influenced by both intrinsic sensory properties and differences between the two regional convenience samples evaluated in this study. The higher liking scores observed among AGS consumers suggest that cultural familiarity and previous exposure to traditional dairy products enhance consumer acceptance. In contrast, lower acceptance among GMA consumers may reflect differences in sensory expectations. These differences are often derived from more industrialized food environments. Such findings highlight the importance of considering regional consumer segmentation when promoting and preserving traditional cheeses. Cultural identity and familiarity appear to be key determinants of market acceptance.

3.4.5. Unsupervised Learning

Several computational experiments employed clustering techniques, including k-means and hierarchical clustering, on both datasets. Hierarchical cluster analysis (HCA) was most effective, delineating three well-defined consumer groups based on sensory attributes. The present investigation evaluated K-Means, Hierarchical Clustering, and DBSCAN to delineate the multidimensional structures within the consumer data.

K-Means Clustering

The silhouette coefficient indicated that a two-cluster configuration was optimal for the participants.

Quantitative metrics were used to validate the robustness of the clustering. For the GAM dataset, the Calinski-Harabasz Index (72.719) was high, indicating well-differentiated and distinct groupings. Conversely, the Davies-Bouldin Index (1.416) remained low, suggesting higher group compactness and superior separation. Furthermore, the average silhouette scores per segment were recorded at 0.364 for cluster 1 and 0.161 for cluster 2.

For the AGS dataset, the Calinski-Harabasz Index (72.719) yielded a high value, indicating well-differentiated and distinct groupings. Conversely, the Davies-Bouldin Index (1.273) remained low, suggesting higher group compactness and superior separation. Furthermore, the average silhouette scores per segment were recorded at 0.34 for cluster 1 and 0.025 for cluster 2.

The K-Means sensory profiles revealed two very distinct consumer types for both datasets:

Cluster 1: This group comprises individuals who prioritize this product and exhibit milder or less pronounced sensory intensities.

Cluster 2: This segment comprises consumers who prefer complex, highly intense sensory attributes.

DBSCAN (Density-Based Spatial Clustering of Applications with Noise)

Regarding the technical implementation of this algorithm, the designated parameters were eps = 4 and min_samples = 15, determined after several iterative adjustments to the spatial and sample-size dimensions.

DBSCAN grouped 169 observations into a single main cluster and identified 7 as noise (−1). It is important to note that the Silhouette score remained uncalculable because the density-based approach did not meet the requirement of identifying at least 2 distinct cluster structures. DBSCAN did not give a relevant segmentation. Most points clustered together, suggesting the data’s density is not suited for this method.

Hierarchical Cluster Analysis (HCA)

To achieve higher granularity in consumer segmentation, a k-value of 3 was determined from the resulting dendrogram (

Table 6. p-values for AGS dataset clusters.

p-Value
Characteristic	Cluster_1 vs. Cluster_2	Cluster_1 vs. Cluster_3	Cluster_2 vs. Cluster_3
Acidity	2.65 × 10−3	9.97 × 10−15	1.84 × 10−7
Adhesiveness	3.03 × 10−9	7.57 × 10−14	1.53 × 10−3
Bitter	9.33 × 10−4	4.75 × 10−14	3.75 × 10−6
Body odor	1.34 × 10−5	1.13 × 10−8	5.90 × 10−2
Chemical odor	1.97 × 10−4	1.38 × 10−11	8.38 × 10−5
Elasticidad	3.26 × 10−8	1.06 × 10−19	1.01 × 10−5
Firmness	2.85 × 10−11	2.31 × 10−7	3.70 × 10−1
Floral odor	1.10 × 10−4	9.63 × 10−25	8.21 × 10−13
Fruity odor	3.12 × 10−3	3.95 × 10−22	6.50 × 10−12
Granularity	8.20 × 10−15	3.57 × 10−27	1.57 × 10−3
Herbal odor	8.13 × 10−5	2.49 × 10−26	7.63 × 10−14
Homogenous appearance	6.09 × 10−25	5.40 × 10−17	8.99 × 10−1
Lactic odor	1.61 × 10−6	6.52 × 10−7	2.81 × 10−1
Metallic	5.68 × 10−2	2.55 × 10−11	3.17 × 10−8
Putrid odor	5.35 × 10−2	1.83 × 10−3	2.01 × 10−1
Saltiness	9.49 × 10−8	8.34 × 10−10	7.75 × 10−2
Softness	1.66 × 10−8	2.54 × 10−11	8.13 × 10−3
Sweet	2.25 × 10−5	1.75 × 10−17	9.12 × 10−7
Umami	1.84 × 10−4	6.82 × 10−17	1.80 × 10−7
White colour	4.54 × 10−11	1.88 × 10−7	8.20 × 10−1

and

Table 7. p-values for GAM dataset clusters.

p-Value
Characteristic	Cluster1_vs._Cluster2	Cluster1_vs._Cluster3	Cluster2_vs._Cluster3
White Color	2.09 × 10−6	2.55 × 10−6	4.43 × 10−1
Homogeneous Appearance	3.50 × 10−19	1.52 × 10−12	2.05 × 10−1
Fruity Odor	6.36 × 10−8	1.23 × 10−32	5.67 × 10−7
Floral Odor	6.33 × 10−4	1.28 × 10−31	3.96 × 10−11
Lactic Odor	4.38 × 10−7	6.40 × 10−3	5.38 × 10−2
Putrid Odor	6.42 × 10−1	1.84 × 10−11	9.69 × 10−6
Body Odor	8.59 × 10−1	3.56 × 10−9	6.18 × 10−8
Chemical Odor	6.36 × 10−1	1.16 × 10−14	2.60 × 10−8
Herbal Odor	9.90 × 10−5	3.07 × 10−20	7.14 × 10−5
Umami Taste	5.67 × 10−12	9.53 × 10−17	1.12 × 10−1
Sour Taste	5.81 × 10−5	4.42 × 10−14	5.74 × 10−5
Bitter Taste	2.79 × 10−3	1.26 × 10−12	7.82 × 10−5
Sweet Taste	1.43 × 10−12	2.05 × 10−19	1.79 × 10−2
Salty Taste	2.74 × 10−3	3.33 × 10−6	1.28 × 10−2
Metallic	4.50 × 10−1	8.49 × 10−13	2.71 × 10−6
Firmness	2.81 × 10−6	4.53 × 10−4	3.99 × 10−1
Adhesiveness	3.99 × 10−9	2.00 × 10−15	1.51 × 10−2
Granularity	2.41 × 10−8	1.21 × 10−12	8.78 × 10−2
Smoothness	8.17 × 10−4	2.15 × 10−6	5.12 × 10−2

).

Furthermore, the Davies-Bouldin Index was 1.644 for the GAM dataset and 1.531 for the AGS dataset, indicating a partition somewhat less optimal than that achieved by K-Means. In addition, the average silhouette scores per segment were 0.199 for Cluster_1, 0.222 for Cluster_2, and 0.124 for Cluster_3 for the GAM dataset, and the silhouette magnitudes were 0.297, 0.166, and 0.156543 for clusters 1, 2, and 3, respectively, for the AGS dataset.

Hierarchical clustering helped define three main consumer groups:

Cluster 1 includes consumers with intermediate sensory evaluations, prioritizing “lactic odor” and “salty taste.”

Cluster 2 includes those preferring the lowest intensity across most variables, favoring a highly neutral sensory experience.

Cluster 3 groups consumers who provide the highest intensity ratings, seeking complex, pronounced sensory profiles across the full range of attributes.

K-Means provided the most robust clustering solution, as indicated by global metrics such as the Calinski-Harabasz and Davies-Bouldin indices. It generated two well-differentiated segments in sensory intensity. Hierarchical Clustering, while slightly inferior in global metrics, produced three clusters. This approach offered a more granular and psychologically rich interpretation of consumer profiles.

Quantitative evaluation of the grouping robustness revealed average silhouette scores of 0.199 for Cluster 1, 0.222 for Cluster 2, and 0.124 for Cluster 3 for the GAM dataset; meanwhile, the AGS dataset recorded silhouette magnitudes of 0.297, 0.166, and 0.156543 for clusters 1, 2, and 3, respectively.

In conventional data science, these metrics may suggest geometric overlap. However, they are considered robust and representative within sensory science and consumer research, since human perception is highly variable and subjective. All recorded silhouette coefficients were positive, showing that each instance was properly categorized inside its group. The technical relevance of this segmentation was supported by later statistical evaluations, which found significant differences (p < 0.05) in sensory profiles. This means that the identified niches have unique cultural perceptions and distinct behaviors, even when the dissimilarity distances at the boundaries are small.

To further differentiate the consumer niches, Student’s t-tests were conducted to determine p-values (Table 6 and Table 7) by comparing the means across each pair of identified clusters. The resulting statistical data illustrate the comparative p-values for the dyads ‘Cluster1_vs._Cluster2’, ‘Cluster1_vs._Cluster3’, and ‘Cluster2_vs._Cluster3’ relative to each sensory variable for the GAM and AGS datasets (Table 6 and Table 7). Significance is established when a p-value falls below the 0.05 threshold (denoted by light green shading), indicating a statistically significant difference in the mean perception of that attribute between the compared groups.

For reasons mentioned above, HCA was used to analyze the datasets for both AGS and GMA. This analysis identified three distinct clusters for each dataset. The results of the hierarchical clustering are visualized using a dendrogram (Figure 6a,b). This tree-like graph shows the sequence of cluster mergers. The X-axis of the dendrogram represents the individual samples or clusters. The Y-axis indicates the distance (or dissimilarity) between the clusters as they merge.

To determine the subgroups, a horizontal line is drawn across the dendrogram [38]. The number of intersections between this line and the vertical lines shows the number of clusters at that level of dissimilarity. The most ‘natural’ number of clusters is identified by finding large jumps in the merger distances on the Y-axis. This threshold was determined programmatically using the linkage matrix, which records each merge and the distance at which it occurred, along with the desired number of clusters (3). The program finds the distance in the linkage matrix that, when used as the cutoff, produces exactly the desired cluster count. To ensure the cutoff is inclusive, a small value (+0.001) is added to the calculated threshold. Thus, the threshold for 3 clusters is the critical distance that divides the continuous hierarchical structure of the dendrogram into 3 distinct groups, based on the data’s internal similarity.

The observation that this distance is significantly greater (20.82 vs. 14.12) indicates that the variability among consumer profiles in Aguascalientes is notably broader before being consolidated into three well-defined clusters. This higher cutoff distance implies greater heterogeneity in consumer preferences in AGS than in GMA, where the same number of clusters was achieved with a lower dissimilarity distance.

The resulting dendrograms (Figure 6a,b), for both the AGS and GMA datasets, visualize the hierarchical clustering structure with colored branches. In the AGS dendrogram, the horizontal lines represent cluster mergers. The lower the merger on the Y-axis, the more similar the elements or clusters that merge. Identifying the number of clusters: The horizontal dashed red line at y = 20.82 is a cut line. When the dendrogram is cut at this height, the vertical lines intersect, and the number of intersections indicates the number of clusters. In this case, the cutoff line is designed to identify three clusters, shown in different colors (orange, green, and dark red/brown) below the line, while the branches above the threshold are gray. The long branches on the Y-axis before a merger (especially the one joining the large orange/green clusters and the red cluster) indicate that these clusters are distinct from each other. Thus, the upper blue cluster shows that the two large clusters that merge are the most dissimilar.

The GMA dendrogram visualizes the results of hierarchical clustering, with particular emphasis on the identification of three clusters. This tree-like graph shows how individual observations group into clusters. The branches represent cluster mergers, and the height of the merger on the Y-axis indicates the distance between the merged clusters. The higher the branch, the greater the dissimilarity between the merging groups. The dashed red horizontal line indicates the cut made in the dendrogram. The height of this cut was determined to obtain exactly three clusters. The printed value, 14.12 distance units, is the distance at which the tree is cut to define these three main groups. This means that any mergers occurring below this line will result in sub-clusters, and mergers above this line are already uniting more dissimilar groups. The dendrogram uses different colors (orange, green, and red) to highlight the branches that belong to the three main clusters defined by this cut. Branches that have not yet merged with any of these three main clusters (i.e., branches that cross the cut line above it) are shown in gray. This allows you to visually see the composition of each of the three large groups and the distance at which they formed.

Comparative Analysis of Consumer Clusters

A comparative analysis of clusters in AGS and GMA finds notable statistical, structural, and cultural differences. This study highlights the varied ways different segments judge the sensory features of traditional Tenate cheese.

The HCA algorithm created three consumer groups in each region. However, participant distributions within each cluster show key structural differences:

• Guadalajara Metropolitan Area (GMA, n = 169): This population demonstrates a highly concentrated and polarized market structure. Cluster 1 accounts for the vast majority of consumers at 63.31% (n_0 = 107), followed by Cluster 2 at 21.30% (n_1 = 36), and Cluster 3 at 15.38% (n_2 = 26) of the sample.

In AGS, the relatively even distribution across clusters implies the need for more diversified marketing strategies to address the broader range of consumer preferences within this region.

Upon calculating the mean consumer evaluations across both regions, participants in AGS assigned higher absolute scores to all 20 sensory characteristics than those in GMA. The cheese exhibited greater global intensity in AGS, with specific mathematical variances showing statistically significant differences.

Δ = X⁻_AGS − X⁻_GDL

Softness: Δ = +0.59 (X⁻_AGS = 2.21 vs. X⁻_GDL = 1.63)

Floral odor: Δ = +0.51 (X⁻_AGS = 0.89 vs. X⁻_GDL = 0.37)

Bitter: Δ = +0.51 (X⁻_AGS = 1.34 vs. X⁻_GDL = 0.83)

Elasticity: Δ = +0.50 (X⁻_AGS = 1.44 vs. X⁻_GDL = 0.95)

Lactic odor: Δ = +0.48 (X⁻_AGS = 2.74 vs. X⁻_GDL = 2.26)

Comparing the segment centroids (means) between the two regions reveals significant quantitative differences. Specifically, in GMA, Cluster 1, the majority group, perceives appearance as twice as homogeneous (0.79) as Cluster 1 in AGS (0.42). In contrast, AGS participants detect a lower level of saltiness (1.10) than those in GMA (1.64). Shifting to Cluster 2, in AGS, this segment represents a large core (36.91%) that strongly rates Firmness at 3.71 and Dairy Odor at 3.29. Meanwhile, in GMA, Cluster 2 forms a minority group (21.30%) with similar firmness descriptors (3.58) but with a significantly higher Umami perception (1.83 vs. 1.21 in AGS). Finally, for Cluster 3, the group’s behavior changes direction depending on the region: in AGS, Cluster 3 is oriented towards Fruity odor (2.74), Floral odor (2.71), and Herbal odor (2.71), whereas in GMA, Cluster 3 is drastically oriented towards Putrid odor (2.08 in GMA vs. 1.09 in AGS) and Body odor (2.35 in GMA vs. 1.66 in AGS).

In GMA, the concentration of nearly two-thirds of the participants (63.31%) within Cluster 1 suggests a significant cultural influence on sensory perception. Given that GMA is a highly urbanized and industrialized center, its consumers often prefer standardized, pasteurized dairy products with subtle sensory profiles. As a result, when evaluating the rustic and authentic characteristics of Tenate cheese, many participants demonstrated sensory adaptation or hedonic saturation, resulting in lower intensity ratings. In contrast, in AGS, approximately 60% of the consumer base was distributed across the high-intensity groupings (Clusters 2 and 3).

This distribution in AGS suggests that its consumers possess a more diverse sensory reference framework and greater familiarity with the traditional attributes of artisan dairy heritage. As a result, such cultural proximity enables a superior capacity to differentiate complex textural dimensions—including firmness and softness—and the subtle secondary aromatic compounds inherent to this traditional cheese.

Regardless of their geographic origin, consumers of Tenate Cheese can nevertheless be qualitatively divided into three sensory groups as follows:

2. Cluster 2: Focuses on the core, positive attributes of a good artisanal cheese (high firmness, whiteness, visual uniformity, and a dominant dairy profile), remaining completely unaffected by any extraneous notes or defects.

3. Cluster 3: Registers and amplifies the entire biochemical spectrum of raw milk cheese, detecting both secondary aromatic complexity and aggressive volatile compounds.

4. Discussion

4.1. Findings

4.1.1. Findings on the Scientific Characterization of Tenate Cheese

The present study provides one of the first integrated scientific characterizations of Tenate cheese, combining analyses of physicochemical, microbiological, nutritional, sensory, and consumer characteristics. The findings are based on a single production batch from one artisanal producer. Consequently, the results should not be considered representative of all Tenate cheese produced in the region but rather as an exploratory characterization. The results indicate that Tenate cheese can be classified as a semi-hard raw-milk cheese with active lactic fermentation, high concentrations of lactic acid bacteria, and a distinctive sensory profile characterized by lactic aroma, acidic and umami flavors, and a firm granular texture. Together, these characteristics contribute to the product’s identity as a traditional Mexican cheese and provide baseline information that may support future preservation, quality assurance, and valorization initiatives. These characteristics are consistent with those reported for other traditional Mexican cheeses, in which the use of raw milk and artisanal manufacturing practices contribute to the development of distinctive sensory attributes and the presence of lactic acid bacteria [2,39,40].

4.1.2. Findings on Consumer Acceptance

The consumer study demonstrated that acceptance of Tenate cheese varied significantly between locations despite participants sharing similar institutional and educational backgrounds. This study compares two university-affiliated consumer groups from two Mexican regions. Consumers from AGS showed higher liking scores than consumers from the GMA, while penalty analysis revealed different sensory drivers of disliking in each location. These findings suggest that regional market differences may influence perceptions of traditional cheeses and underscore the importance of consumer segmentation when developing commercialization strategies. The HCA results further demonstrated distinct preference clusters, indicating that traditional cheeses may appeal to different consumer niches rather than to a homogeneous market. A comparative analysis of variable importance across GMA and AGS datasets indicates that while a shared descriptive profile exists, there are marked distinctions in the composition and intensity of the sensory profiles that consumers prioritize in each region. The results indicate that consumers in both cities perceived a similar attribute profile for Tenate cheese, characterized by a white color, a homogeneous appearance, a lactic odor, and acidic, umami, and salty flavors, as well as a firm, granular texture with a certain degree of softness. However, differences between the two cities in Mexico were observed in the perceived intensity of some attributes. Consumers in AGS reported higher intensity ratings for specific attributes, particularly bitterness, as well as texture-related attributes such as granularity, firmness, and elasticity. Similar findings have been reported by Pereira et al. [41], who identified different consumer segments for artisanal cheeses and highlighted the importance of market segmentation when designing commercialization strategies. Likewise, Freire et al. [17] observed that consumers may differ in the intensity ratings they assign to sensory attributes such as firmness, saltiness, and acidity.

The sensory characteristics perceived by consumers may be associated with the complex biochemical and microbiological processes occurring during cheese manufacture and ripening. The lactic odor identified by consumers is commonly associated with organic acids and fermentation-derived compounds, particularly lactic acid, acetoin, diacetyl, and acetaldehyde, which contribute fresh, fermented, and buttery notes to dairy products [42,43]. Recent studies have highlighted the central role of microbial metabolism in the formation of these compounds, demonstrating that interactions among lactic acid bacteria, yeasts, and other members of the cheese microbiota significantly influence aroma development and flavor complexity [44]. Likewise, the acidic flavor perceived in Tenate cheese is mainly associated with the accumulation of lactic acid resulting from lactose fermentation by lactic acid bacteria, whereas umami perception is often related to the release of free amino acids, particularly glutamic acid, generated during proteolysis [42].

Differences in bitterness perception between consumers from AGS and the GMA may be associated with variations in the concentration of hydrophobic peptides formed during casein degradation. Recent peptidomic studies have identified numerous bitter peptides derived from αs1- and β-caseins and demonstrated that their accumulation can negatively affect consumer acceptance when present above sensory threshold levels [45]. Furthermore, the sensory complexity of traditional raw-milk cheeses is strongly influenced by volatile compounds, including ketones, aldehydes, esters, alcohols, sulfur-containing compounds, and short-chain free fatty acids, which are produced by lipolysis, proteolysis, and microbial metabolism during cheese maturation [43]. In particular, compounds such as butanoic, hexanoic, and octanoic acids contribute characteristic cheesy, pungent, rancid, and animal-like notes that may influence consumer liking depending on their concentration and consumers’ familiarity with traditional cheese flavors. Recent research on artisanal cheeses has further demonstrated that the abundance of these volatile compounds is closely associated with the composition of the indigenous microbiota and traditional manufacturing practices, thereby contributing to product typicity and regional identity [46].

Additionally, recent studies on artisanal and traditional cheeses have shown that consumer perceptions are influenced by familiarity with characteristic flavor compounds and expectations associated with local food cultures, resulting in distinct preference patterns across market segments [46]. These findings reinforce the importance of considering both sensory drivers and consumer segmentation when developing commercialization strategies for traditional cheeses such as Tenate cheese.

4.1.3. Transferability of the Integrated Framework

Beyond the specific findings for Tenate cheese, the present study demonstrates a multidisciplinary framework for characterizing traditional dairy products. The integration of physicochemical, microbiological, nutritional, sensory, and consumer segmentation analyses provides a more comprehensive understanding of artisanal cheeses than any single analytical approach alone.

Although the specific results obtained for Tenate cheese should not be generalized to other products, the methodological approach is transferable. The framework may be applied to other traditional cheeses to document product identity, evaluate quality attributes, identify consumer segments, and support preservation and commercialization efforts. This is particularly relevant for artisanal cheeses whose production systems and consumer acceptance remain insufficiently documented.

4.2. Limitations and Future Research

Several limitations should be considered when interpreting the present findings. First, the physicochemical, microbiological, and nutritional analyses were conducted using cheese from a single production batch and producer, limiting the assessment of production variability. Therefore, the results should not be considered as representative of all Tenate cheese production; future studies include multiple producers, production batches and seasons.

Future studies should determine sodium and saturated fat contents using validated analytical methods and evaluate strategies to reduce their levels. Such strategies could include the use of sodium substitutes and reduction in the fat fraction in the product; however, the latter could affect texture, one of the most highly valued attributes of artisanal cheeses [47,48].

Second, the consumer study was based on university communities in two metropolitan areas with a gastronomic background, which may not represent the broader Mexican population. Finally, although relevant hygienic indicators were evaluated, the microbiological findings should be interpreted only for the microorganisms included in the present study. Major foodborne pathogens were beyond the scope of this study; therefore, no conclusions can be drawn regarding the overall microbiological safety of the product.

Future research should apply the proposed multidisciplinary framework to multiple producers, production seasons, and traditional cheese varieties to assess product variability and validate its broader applicability. Studies involving more diverse consumer populations across different Mexican regions and international markets are needed to better understand the influence of cultural familiarity, food neophobia, purchasing behavior, and repeated exposure on consumer acceptance. Comprehensive microbiological safety assessments, together with advanced analytical techniques such as volatile compound profiling, texture profile analysis, microbiome characterization, and metabolomics, would provide a more complete understanding of cheese quality, microbial ecology, and sensory perception. In addition, exploratory consumer segmentation approaches integrating physicochemical, microbiological, and sensory data could support quality optimization while preserving the authenticity and distinctive characteristics of traditional artisanal cheeses.

5. Conclusions

Tenate cheese is a traditional raw-milk Mexican cheese wrapped in palm fiber. This study provides an exploratory characterization of one production batch of traditional Tenate cheese from a single artisanal producer. Although the findings offer valuable preliminary information regarding its physicochemical, microbiological, nutritional, and sensory properties, they should not be generalized to all Tenate cheese. This batch is characterized by a white, homogeneous appearance, a lactic aroma, acidic and umami flavors, and a firm, granular texture typical of artisan cheeses. Physicochemical analysis classified this Tenate cheese as a semi-hard cheese with active lactic fermentation and physicochemical characteristics consistent with good-quality artisanal cheese. Combining physicochemical characterization, microbiological evaluation, sensory analysis, and exploratory consumer segmentation techniques offers a comprehensive framework for identifying consumer niches and supporting artisanal cheese valorization.

Consumer evaluations showed regional differences in acceptance, with consumers from AGS regions reporting higher liking scores than those from GMA. Hierarchical clustering identified distinct consumer segments, indicating that regional preferences should be considered when developing marketing strategies or promoting Tenate cheese across different regions of Mexico and other countries.

This study shows that preserving traditional cheeses requires safeguarding artisanal production and understanding modern consumer expectations through interdisciplinary approaches. This methodology could have a potential application to other artisanal cheeses and serve as a roadmap. Integrating sensory science, microbiological characterization, and exploratory multivariate techniques offers a promising path for preserving the authenticity, competitiveness, and cultural value of traditional Mexican cheeses in contemporary markets. Agglomerative hierarchical cluster analysis (HCA) highlights the exploratory consumer segmentation’s role in identifying consumer segments and understanding differences between regional consumer segments in traditional dairy products. These findings support the preservation, commercialization, and cultural valorization of Mexico’s traditional cheese heritage.