Oaxaca cheese: flavour, texture and their interaction in a Mexican traditional pasta filata type cheese

Abstract

Oaxaca cheese is classified as a pasta filatacheese and it is considered as an unavoidable ingredient in broad varieties of Mexican recipes. It is produced in many regions by industrial companies but mainly by handmade procedures. It is thought that its flavour and texture profiles could be recognized by consumers even if traditional production processes are not standardized, so this study is intended to define its flavour and texture profiles and the relationship between them considering the variability given by its handmade character. The origin of the milk has a major contribution to define the flavour explaining about 51.9% of the variance. Sensory hardness, cohesiveness, chewiness and springiness were highly correlated (0.6–0.83). Texture and flavour, as multivariate variables, are linearly dependent with a high canonical correlation (0.966) as well.

El queso Oaxaca se clasifica como un queso de pasta filatay es considerado como un ingrediente que no puede faltar en una gran variedad de recetas de comida mexicana. Se produce en muchas regiones de México a nivel industrial y tradicional. Se considera que su sabor y textura se reconocen por el consumidor habitual aún cuando en la producción tradicional su proceso de elaboración no está estandarizado. En el presente trabajo se pretende definir el perfil de sabor y textura del queso Oaxaca y la relación entre estas variables, considerando la variabilidad debido a su carácter artesanal. El origen de la leche influye considerable en la definición del sabor, explicando el 51,9% de la varianza. La dureza, cohesividad, masticabilidad y elasticidad sensoriales muestran una alta correlación (0,6–0,83). La textura y el sabor, como variables multivariadas, son linealmente dependientes y también muestran una alta correlación (0,966).

Keywords:

• Oaxaca cheese
• pasta filata
• flavour
• texture
• traditional production
• non-pasteurized milk

Palabras clave:

• Queso Oaxaca
• pasta filata
• sabor
• textura
• producción tradicional
• leche no pasteurizada

Introduction

Mexican cuisine is characterized by its diversity in flavours, fragrances and textures. Moreover, it is considered as a living gastronomy that has been exported to a wide range of regions all around the world, developing special traits according to the preferences of the culture where it is adopted. Oaxaca cheese, “quesillo” or “thread cheese” is an unavoidable ingredient in broad varieties of Mexican recipes and it is considered as a foodstuff itself. It is also one of the most consumed milk products all over Mexico being produced in many regions by big industrial companies but mainly by handmade procedures (Villegas de Gante, 2004). According to the Instituto Nacional de Estadística Geografía e Informática (INEGI, 2010; the Mexican National Institute of Statistics, Geography and Information Technology), the volume of fresh cheese sold in Mexico in 2009 was of 76,696 tons with a total sales value of over 2.7 billion Mexican pesos, while in the same year the volume sold of Oaxaca cheese was 16,723 tons with a total sales value of nearly 800 million Mexican pesos, representing 21.8% and 29.4%, respectively. However, these figures relate to industrialized Oaxaca cheese marketed through formal channels. Reliable data related neither production nor sales of artisan-made cheese is not available.

Oaxaca cheese is manufactured from cow's milk and it is classified as a pasta filatacheese because during its elaboration procedure the acidified curd is kneaded with hot water leading to stretchable and plasticized dough which is able to form stripes (Figure 1). The handmade production process includes the use of raw cow's milk and requires knowledge of some critical points and the ability of the artisan to identify them, such as the acidity needed to obtain the stretchable dough or the quantity of salt that must be added (Esquivel & Santos-Moreno, 1996; Villegas de Gante, 1993). Those observations mean that sensory and structural characteristics of Oaxaca cheese are artisan skills dependent and that in order to minimize variations between production lots it is required a standardized repeatability of the manufacturing process. This case is well exemplified by the traditional Oaxaca cheese production process of a town named Aculco, located in the central highlands of Mexico, and according to the last population report, with a population of 40,492 men and women older than 5 years old (INEGI, 2005). This town is known as a traditional cheese producer, where Oaxaca cheese is one of the most produced. The area has 37 cheese factories, of which 34 produce Oaxaca cheese.

Figure 1. Oaxaca cheese sample.

Figura 1. Muestra de queso Oaxaca.

In spite of its high consumption, just a few investigation reports regarding Oaxaca cheese have been published focusing on the optimization of its production process. De Oca-Flores, Castelan-Ortega, Estrada-Flores, and Espinoza-Ortega (2009) described the highly variable handmade production process of Oaxaca cheese (flow chart reproduced in this paper with permission from the authors as Figure 2) in the same town, concluding that there are huge variations in the physicochemical characteristics of the milk used and in the cheese protein content (150.3–241.5 g kg⁻¹), fat (170–253.3 g kg⁻¹), pH (4.9–5.8), salt (8.4–36.2 g kg⁻¹), moisture (124.1–610.3 g kg⁻¹) and ashes (18–41.8 g kg⁻¹). In another paper, Aguilar-Uscanga, Montero-Lagunes, De la Cruz, Solís-Pacheco, and García (2006) incorporated fermented whey to the milk used in the manufacture of a controlled Oaxaca cheese production process, obtaining no significant differences in yield, physicochemical and sensory characteristics when comparing treatments with the controlled regular procedure, reducing production time in about 38%. In another study, Totosaus and Guemes-Vera (2008) evaluated the effect of carrageenan as a fat substitute on some quality characteristics of the cheese.

Figure 2. Manufacture process of Oaxaca cheese flow chart made in Aculco, Mexico (Courtesy of M. De Oca-Flores et al., 2009).

Figura 2. Diagrama de flujo del proceso de elaboración de queso Oaxaca en Aculco, México (Cortesía de M. De Oca-Flores et al., 2009).

It is important to consider that because of its fresh cheese character, Oaxaca cheese does not develop ripened type flavours but in spite of this absence of aromatic products, it is thought that its flavour and texture profiles could be recognized by usual consumers. Moreover, handmade Oaxaca cheese has different sensory properties given by its manufacture procedure and the characteristics of the milk used. That is why we can find in Mexico a huge variety of handmade Oaxaca cheeses with the only basic requirement of a stretchable structure like the one shown in Figure 1. According to these last comments, it is possible that we could find more than one flavour profile for Oaxaca cheese manufactured by traditional handmade procedures but just one unique texture profile. Besides, there are no reports related to the description of Oaxaca cheese flavour and/or texture, so this study is intended to define the flavour profile and the relationship between flavour and sensory texture for this type of cheese considering its intrinsic variability given by its handmade character.

Materials and methods

Eleven cheese producers were chosen from those 34 (32.35%) that produce Oaxaca cheese with raw milk. One cheese per producer (1 kg per piece) was sampled, and sensory and some physicochemical characteristics were determined at room temperature 1 day after being stored at 4°C. Physicochemical analysis included moisture content (%), protein (%), acidity (g lactic acid kg⁻¹), fat (%) and salt (%), following official methods according to the Association of Official Analytical Chemists (AOAC, 1990) and were done in duplicate.

Sensory evaluation (flavour and texture) was done by panellists trained for 60 h during 45 working days (4–5 h/day) 1 day after collecting the samples.

Instrumental texture analysis

Three cheese cubes of 2 cm side (Antoniou, Petridis, Raphaelides, Ben Omar, & Kesteloot, 2000) with a perpendicular orientation to the threads were cut from the centre of the cheese (to limit the effects of surface drying). The samples underwent uniaxial compression tests of two consecutive cycles, according to texture profile analysis (TPA) technique (Breuil, & Meullenet, 2001; Osorio, Ciro, & Guillermo, 2005; Tunick, & Van Hekken, 2002; Zúñiga, Ciro, & Osorio, 2007). A test cylinder of 10 mm in diameter was used (Benedito, González, Rossello, & Mulet, 2000) in order to compare the instrumental penetration of the probe with the sensory chewiness value, applying a compression of 50% of the sample side length at a rate of 1.0 mm s⁻¹and a force threshold of 0.0495 N in triplicate to the samples already standardized at a temperature of 21°C. Hardness, cohesiveness, chewiness, springiness and adhesiveness were obtained for the first and second cycles of the compression test using a Stable Micro Systems TA-XT2 Texturometer (Stable Microsystems, NY, USA), according with Meullenet and Gross (1999).

Panellists training procedure

Panellists were selected from a population of students (Agronomic Sciences Faculty, Universidad Autónoma del Estado de México) between 20 and 23 years old adapting the selection questionnaire to Oaxaca cheese (Meilgaard, Civille, & Carr, 1999). Personal interviews were conducted as assistance with the selection of the candidates. Fourteen students were selected and the training procedure began with two introductory sessions explaining main objectives of a trained panel, measuring scales, methodology, evaluation sheets and the rules that must be observed during training and evaluation sessions. During training and evaluation sessions, all of the panellists were adequately separated using individual screens adapted with white/red light control and were provided with crackers and water to rinse. A 15-cm scale with anchors at 0 and 15 was used, according to Meilgaard et al. (1999). No difficulty was found to use neither the scale nor the terminology.

Flavour training

After this two first sessions, the candidates were trained in identification and ranking of the four basic tastes (sweet, sour, acid and salty) according to Jellinek (1995). After that, same four tastes were evaluated according to Gallerani, Gasperi, and Monetti (2000) using cottage cheese as carrier instead of ricotta cheese. Eight different identification and ranking tests were conducted in duplicate to evaluate each panellist performance and confidence using the sequential analysis (Montgomery, 2002) and processing the data in Microsoft Excel®. In order to select the best trained judges, five different tests (in accordance with Gallerani et al., 2000) were conducted in duplicate (combinations of different tastes and intensities). Those tests allowed the selection of five accurate judges to be included in the evaluation of the flavour profile.

Texture training

Texture analysis was carried out in accordance with Szczesniak (1963) with modifications reported by Bourne (2002), Meilgaard et al. (1999) and Muñoz (1986) and four texture characteristics were measured by the panellists: hardness, cohesiveness, springiness and chewiness. Each sample was prepared as described by Antoniou et al. (2000), Muñoz (1986), Szczesniak (1963) and Wium, Gross, and Qvist (1997).

Generation of the descriptive terms

Flavour terminologywas generated using as reference some descriptors reported by Drake, McIngvale, Gerard, Cadwallader, and Civille (2001) that could be applied to a fresh cheese elaborated with non-pasteurized milk. The descriptors were proposed by the judges previously selected during a group session where the following samples were evaluated:

	– Non-pasteurized whole milk (obtained from the Veterinary School at Universidad Autónoma del Estado de México)
	– Pasteurized whole milk (bought at retail)
	– Ultra-high temperature (UHT) pasteurized whole milk (bought at retail)
	– Natural yogurt no additives added (bought at retail)
	– Milk butter (bought at retail)
	– Milk cream (25% fat, bought at retail)
	– Gouda type cheese (25% fat, bought at retail)
	– Oaxaca type cheese (18% fat, bought at retail)
	– “Light” Oaxaca type cheese (8% fat, bought at retail)
	– Oaxaca type cheese elaborated with non-pasteurized milk (artisanal).

During the group discussion, 10 descriptors were generated and accepted by the judges (creamy, fatty, buttery, fermented, smoked, burnt milk, caramelized, watery, mouldy and cowy) in addition to the four basic tastes (sweet, salty, acid and sour) making a total of 14 flavour descriptors. In order to validate the terminology two different identification and intensity tests were conducted in duplicate using the 15-cm scale and the following samples:

	– Whole milk (non-pasteurized, pasteurized and UHT)
	– Oaxaca cheese elaborated with different types of milk (whole milk-non-pasteurized, whole milk-pasteurized and skimmed milk-pasteurized).

After scale and terminology validation, the judges evaluated (under a red light environment, monadically presented in a three-digit code lidded plastic cups) the 11 cheese samples collected and results were processed by means of principal component (PC) analysis using Hair, Anderson, Tatham, and Black (1998) methodology so that redundant terms could be avoided, interdependency between them could be elucidated and their contribution to the explanation of the variance could be defined. Quantitative values have been reported using Microsoft Excel 2000®, radar graphs and canonical correlations (Hair et al., 1998).

Texture terminologywas developed during a group session presenting to the panel of well-known cheese samples that represented different points on the textural scale for each term (hardness, cohesiveness, springiness and chewiness). Through a panel consensus, scores were assigned to these cheeses. In order to validate the scores, four unknown cheese samples (different than those used to generate them) were evaluated in duplicate by each judge under the same methodology used to evaluate Oaxaca cheese samples. During Oaxaca cheese evaluation, the 11 cheese samples were cut into 8 cm³cubes, and each panellist was provided with three cubes per sample per replication. The samples were presented monadically in lidded 4 oz plastic sample cups with three-digit codes in a red light environment.

Results and discussion

Table 1shows the average and variability for salt, acidity, protein, fat and moisture content of the cheeses sampled. These results coincide with those reported by De Oca-Flores et al. (2009) where the huge variation suggests a heterogeneous behaviour between cheese producers, and in some cases there are also differences between production lots of the same producer.

Table 1. Physicochemical composition of the 11 Oaxaca cheese samples (results are mean values from duplicate analysis).
Tabla 1. Composición fisicoquímica de las once muestras de queso Oaxaca (los resultados son el promedio de duplicados).

Parameter	Salt (%)	Acidity (g lactic acid kg^−1)	Protein (%)	Fat (%)	Moisture (%)
Average	1.75	4.44	28.40	22.17	35.99
Standard error	0.38	1.59	2.52	1.39	1.43
Coefficient of variation	21.52	35.88	8.88	6.28	3.99
Upper confidence interval (α = 0.05)	2.49	7.56	33.34	24.90	38.81
Lower confidence interval (α = 0.05)	1.01	1.32	23.45	19.44	33.18

The main difference is related to acidity (reported as lactic acid) and salt content with variability coefficients of 35.88% and 21.52%, respectively, pointing out large differences between production processes. Those parameters are determinant in the cheese flavour. The same variables will not affect texture properties because once the required acidity has been reached and threads have been formed the recently produced cheese will keep its structure. Protein content has significance in flavour profiles of ripen cheeses but has no considerable influence on a fresh cheese flavour profile (Beuvier, & Buchin, 2004; Van Hekken, Drake, Molina-Corral, Guerrero-Prieto, & Gardea, 2006). Regarding fat content, it is expected to be considered as a significant influence on flavour profile too (Fenelon, Guinee, Delahunty, Murray, & Crowe, 2000). Figure 3represents the flavour profile of Oaxaca cheese, showing mean values for each descriptor as well as the area between the mean and one added and subtracted standard deviation for each case. This chart illustrates that even if Oaxaca cheese does not show a regular flavour profile it is still considered as Oaxaca cheese because of its shape, appearance and texture. The flavour of Oaxaca cheese elaborated under a traditional production process in central Mexico is defined within an interval that allows the flavour itself to keep its Oaxaca cheese identity even if traditional production process is highly variable.

Figure 3. Flavour profile of traditional Oaxaca cheese elaborated with non-pasteurized milk.

Figura 3. Perfil de sabor de queso Oaxaca tradicional elaborado con leche no pasteurizada.

Flavour descriptors have been analysed by PC analysis; thus Table 2shows the obtained matrix of component weights (after Varimax rotation). Four PCs which account for 78.3% of the variability in the original data were extracted. These PC were named: fermented, caramel-cooked, salty-creamy and acid. It could be emphasized that PC 1 is composed of six descriptors related to flavours associated with the origin of milk (buttery, fermented, smoked, mouldy and cowy) being more related to sweet than to salty taste. PC 2 is highly related with caramel and cooked-sugar like flavours, including sweet, sour, burnt milk and caramelized. Watery descriptor is more related to this PC than to “fatty” notes grouped in PC 3 where creamy, fatty and salty are considered as values with a relatively high importance.

Table 2. Principal component weight matrix (after Varimax rotation) for terms generated to describe Oaxaca cheese flavour.
Tabla 2. Matriz de componentes principales (después de aplicar rotación Varimax) para los términos generados para describir el sabor del queso Oaxaca.

	PC
1	2	3	4
Descriptor	Fermented	Caramel-cooked	Salty-creamy	Acid	Estimated communality
Sweet	0.666*	0.601*	0.207	−0.132	0.865
Salty	−0.043	−0.064	0.684*	0.523	0.747
Sour	0.338	0.661*	0.044	0.257	0.619
Acid	0.055	0.068	−0.020	0.919*	0.853
Creamy	0.228	0.562	0.698*	0.013	0.855
Fatty	0.382	0.059	0.647*	−0.155	0.592
Buttery	0.766*	0.513	0.188	0.016	0.886
Fermented	0.860*	0.225	0.041	0.088	0.800
Smoked	0.836*	0.298	0.159	0.124	0.828
Burnt milk	0.386	0.699*	−0.033	−0.074	0.644
Caramelized	0.322	0.831*	0.034	0.010	0.796
Watery	0.146	0.831*	0.414	0.017	0.884
Mouldy	0.622*	0.232	0.518	−0.053	0.712
Cowy	0.827*	0.328	0.278	−0.052	0.872
Cumulative explained variance (%)	51.881	62.228	70.729	78.311
Note: *Values higher than 0.6 are considered statistically significant.Nota: valores superiores a 0.6 se consideran estadísticamente significativos.

PC 4 describes specifically the acid taste not being related to any other flavour, so this taste could be considered as a characteristic that identifies Oaxaca cheese flavour. Knowing that the acid note is important to define Oaxaca cheese flavour, it could be interesting to get deeper in finding the acids responsible of this characteristic.

Neither the flavour nor the texture profile of Oaxaca cheese has been reported, even if there are reports on Mozzarella cheese flavour and texture. It is important to point out that although Mozzarella and Oaxaca cheeses are manufactured through a pasta filataproduction process, the origin of the milk is not the same and the last steps in the manufacture, pressing and kneading, respectively, give to the end product different characteristics that influence their flavour and texture. Regarding flavour, Pagliarini, Monteleone, and Wakeling (1997) stated in their paper about the sensory profile of Mozzarella cheese that fat was determinant in the flavour and texture and that the three cheeses analysed were sensory well differentiated including flavours like sweet, milky, creamy, acid and salty. Besides, Gasperi et al. (2000) compared two different methods to obtain a flavour profile and concluded that some of the notes that describe Mozzarella cheese flavour are acid, salty, yoghurt, creamy, sour, milky and olive. The flavour profile obtained from different Oaxaca cheese samples manufactured by a traditional process using non-pasteurized milk added some interesting notes to the already reported flavour characteristics of Mozzarella cheese like the strong cowy note perceived, lacking of the sour, and yoghurt and olive notes reported by Gasperi et al. (2000).

Sensory and instrumental analyses of hard and semi-hard cheeses have widely been reported (Antoniou et al., 2000; Benedito et al., 2000; Breuil and Meullenet, 2001; Osorio et al., 2005), and talking about pasta filatacheeses, Mozzarella texture has been extensively studied (Ak & Gunasekaran, 1995; Bhaskaracharya & Shah, 1999; Brown, Foegeding, Daubert, Drake, & Gumpertz, 2003; Tunick, 2000; Tunick, Malin, Smith, & Holsinger, 1995), but reports about the relationship between sensory and instrumental analyses nor documents regarding the relationship between flavour and texture of a fresh cheese like Oaxaca are not common even if both sensory properties define the overall liking or disliking of a fresh cheese to the consumer. Weinrichter, Rohm, and Jaros (2000) have concluded in their document about mechanical properties of unpressed semi-hard cheeses that it is complex to understand the behaviour of an open anisotropic structure, and that is the case of Oaxaca cheese. Considering the fact previously mentioned, developing of equivalence between a well-known instrumental method to characterize cheese texture (TPA) and a descriptive sensory analysis could help to a better understand of those kinds of cheeses. Figure 4shows that sensory hardness and cohesiveness are highly correlated with instrumental adhesiveness, cohesiveness and springiness. Not other correlations are significant (with the exception of those that subsist within instrumental and sensory variables). Drake and Gerard (1999) neither reported significant correlations between instrumental and sensory adhesiveness. Lower correlations that those reported here between instrumental and sensory cohesiveness in food products including cheeses have been reported by Meullenet and Gross (1999) and Muñoz (1986).

Figure 4. Significant linear correlations between sensory and instrumental textures.

Figura 4. Correlación lineal significativa entre la textura sensorial e instrumental.

According to these results, the sensory panel expressed the cheese cohesiveness in terms of its mechanical deformation and stickiness, as reported by Drake and Gerard (1999), since instrumental cohesiveness describes how well the cheese withstands a second deformation relative to how it behaved under the first deformation. Nevertheless, Meullenet and Gross (1999) pointed out that description of sensory perception of cohesiveness may require more than one physical measurement of work energy ratio such as TPA cohesiveness. This also means that the cheese considered as hard has been the most elastic and cohesive.

Since flavour and the relationship between instrumental and sensory textures have been previously described, the relationship between flavour and texture has also been explored. Results are shown in Figure 5(A) by a canonical correlation where it could be seen that flavour and texture, as multivariate variables, are related with a correlation coefficient of 0.9659 (p < 0.001). This effect could be the result of a modified perception of the texture by some flavour notes because both evaluations were done by the same judges on the same samples, even if both analyses were done separately (different days of evaluation). Rega, Guichard, and Voilley (2002) concluded that the mobility of volatile compounds through the matrix of a pectin gel favoured the flavour release, supporting the idea of a combined perception flavour–texture. As a result of the canonical correlation, some sub-variables that contribute more to the overall correlation could be extracted: buttery, cowy, burnt milk and caramelized (flavour) and chewiness, springiness and adhesiveness (texture). Considering only those sub-variables the correlation is still significant (p < 0.01) but not so high [0.728; Figure 5(B)] and this contribution may be explained because an increase in chewiness, springiness and adhesiveness may allow some volatile compounds to be released more easily from the cheese matrix. This explanation is consistent with the report of Kim, Kim, Yoo, and Kim (2010) where the intensity of the perceived flavour showed strong negative correlations with the firmness of four different pectin gels.

Figure 5. Canonical correlation of Oaxaca cheese (A: all the sensory attributes included; B: only significant sensory attributes included).

Figura 5. Correlación canónica de queso Oaxaca (A: incluyendo todos los atributos sensoriales; B: incluyendo únicamente los atributos sensoriales de mayor significancia).

Conclusions

The flavour of Oaxaca cheese elaborated under a traditional production process in central Mexico is defined within an interval that allows the flavour itself to keep its Oaxaca cheese identity even if traditional production process is highly variable. The acidity is an independent variable that is identifiable by the judges and that contribute to the flavour because of its importance during cheese manufacture. The origin of the milk is also important to define the flavour of this cheese that is elaborated with non-pasteurized milk (cowy, buttery, smoked and fermented flavour) explaining about 51.9% of the variance. The flavour profile obtained added some interesting notes to the already reported flavour characteristics of pasta filatacheese (Mozzarella) like the strong cowy note perceived, lacking of the sour, and yoghurt and olive notes already reported. Regarding texture, Oaxaca cheese has low hardness and adhesiveness and high springiness, cohesiveness and chewiness. Sensory hardness, cohesiveness and springiness were highly correlated. On the other hand, the process of mechanical deformation and stickiness of the cheese during its consumption is interpreted by the panellists as its hardness and cohesiveness. Although sensory attributes are a common issue to be investigated in ripened cheeses, the relationship between flavour and texture in a fresh cheese turned out interesting since both characteristics are evaluated simultaneously by a consumer. It could be concluded that texture and flavour, as multivariate variables, are linearly dependent but although this perception is more influenced by some attributes or sub-variables, the overall impact is the result of the interaction of all of them.