3.1. Composition and Physicochemical Characteristics
Physicochemical analysis can provide useful information related to the quantification of the cheese’s basic components as well as its quality [
15,
16]. According to Lawrence et al. [
17,
18], pH, SM, FDM, and MNFS are important parameters regarding the composition of a cheese and can have a major effect on its quality.
The classification of cheeses is achieved in different countries according to various principles. Factors such as milk type, degree of ripening, method of manufacture, and physicochemical or microbiological characteristics are usually used for the grouping of a cheese variety [
4,
19]. In Greece, cheeses are classified primarily according to their moisture content. Specifically, the very hard cheeses have a maximum moisture of 32%, the hard have a maximum of 38%, the semi hard have a maximum of 46%, while the maximum moisture content of the soft cheeses must not exceed 58%. Fresh cheeses from milk, without ripening, contain less than 75% moisture and whey cheeses with or without ripening have less than 70% moisture content [
20]. The classification of a cheese based on the moisture content provides information regarding its nutritional value and is widely used [
4].
According to the Greek legislation, the term “quality” is accompanied with metric data on moisture and Fat-in-Dry-Matter content etc. [
20]. Although cheeses with different production methods and characteristics were collected and analyzed in the present work, these cheeses were grouped together, primarily, according to their moisture content, as defined by the Greek legislation. In each group (soft, hard, etc.), the moisture and fat-in-dry-matter content was presented, which, according to the Greek legislation, was used to classify cheeses in qualities (very good, excellent, etc.) (
Table 1,
Table 2 and
Table 3).
However, information on the label of the package or from the websites or from oral communication with the producer (when available) was also taken into consideration, in case needed. The general characteristics of Greek traditional cheese varieties have been previously studied [
3,
21,
22,
23,
24,
25]. In addition, the composition of commercial Greek cheese samples such as Kefalotyri, Anthotyros, Myzithra, Teleme, and Graviera [
15,
26,
27,
28,
29,
30] has previously been reported.
In the present study, 50 of the collected samples were classified as hard cheeses with a mean moisture of 33.8% and FDM of 49.9% (
Table 4).
According to the Greek Codex Alimentarius [
20], an excellent quality hard cheese has a moisture content < 35% and FDM > 47W; therefore, the hard cheeses of the present work could be classified as having an excellent quality. Kefalotyri and Graviera (excluding Graviera Kritis, Graviera Naxou, and Graviera Agrafon, which are PDO cheeses) are two well-known cheeses that are produced widely in Greece. Usually, the name of the local place where they are manufactured is used on the label of these cheeses. The composition of Kefalotyri cheese in the present study (
Table 1) generally was in accordance with that found for commercial samples by Andrikopoulos et al. [
28], while Graviera was in line with that of Vatavali et al. [
29,
30]. However, Zerfiridis et al. [
31] observed a higher FDM for commercial Gruyere cheese samples than in this study. The values of fat of Pecorino of Ios Island were higher than those reported for Pecorino Romano and Sicilliano [
32], whereas for the moisture and salt (
Table 1) they were lower than that found by Kasapian et al. [
15]. In general, these differences, when observed, may be attributed to variations in milk composition, manufacturing and storage conditions, degree of ripening, etc.
Six out of 102 cheeses belonged to the semi hard group, contributing to a mean moisture of 39.1% and FDM of 52.5% (
Table 4). The cheeses of this group were of an excellent quality following Greek Codex Alimentarius [
20], as an excellent quality semi hard cheese has a moisture content < 40% and FDM > 50%. The composition of semi hard Kashkaval cheese (
Table 2) is generally comparable with those reported for pasta filata cheeses [
23,
33] and fulfill (dry matter of 60.8% and FDM of 49.2%,
Table 2) the requirements (dry matter > 58%, FDM > 45%) of the EU [
34]. Under this context, Kashkavali Pindou applied for a PDI indication [
1].
Eighteen soft cheese samples were present in this study, with a mean moisture of 53.9% and FDM of 59.8% (
Table 4). These cheeses were characterized as having an excellent quality (moisture of < 54%, FDM of > 46%) [
20]. Teleme, a white brined cheese manufactured from sheep, goat, and cow milk or mixtures of them, had a generally similar composition to that determined by Andrikopoulos et al. [
28] (
Table 2). Touloumotyri had a higher moisture content than that found by Kasapian et al. [
15] (
Table 2). Kariki is a cheese that is matured in an empty gourd for 40 days up to seven months. This cheese, as well as Mpalaki soft of Tinos, had a low moisture content (37.4% and 36.9%, respectively) and could have been classified to the hard cheeses group, but both cheeses were characterized as soft, with a maximum moisture content of up to 55%, following the piece of information present on their label (
Table 2). Tsalafouti is a spread cheese seeking registration as a PDO Greek cheese [
35]. The composition of Tsalafouti cheese samples from a Greek market (
Table 3) was generally in line with the data reported for industrially produced Tsalafouti cheese by Pappa and Kondyli [
3]. The composition of whey cheeses is shown in
Table 3 and
Table 4. Different values were reported for a similar cheese produced in Turkey, namely Mud whey cheese [
36]. Some differences were observed between the composition of Myzithra and Anthotyros cheeses of the present study and other commercially similar cheeses [
25,
28], possibly due to differences in their manufacturing process.
Although the cheeses studied here are very much appreciated by consumers locally, for several of them, there is a lack of scientific information and, to the best of our knowledge, the identity and the composition of some cheeses was studied for the first time. Such cheeses are Malathouni of Tinos, tyri of Mpournias, volaki of Andros and skotyri of Ios. Furthermore, Tyromalama, Galomyzithra, and Sitaka of Kasos Island (
Table 2) were found to possess a very low salt content (0–1%) and can be suitable for consumers on a low sodium diet. However, in order to better establish their identity, more samples of each cheese need to be analyzed.
Sheep milk is widely preferred in cheesemaking as it results in the manufacture of a cheese with a higher yield than from cow. This is due to its higher protein, fat, and total solids content compared with cow milk [
37,
38]. In Greece, sheep and goat milk is mainly used for cheese production, while cow milk is mainly used for direct consumption, and IN the year 2020, 154,868 tonnes sheep, 36,495 tonnes goat, and 23,580 tonnes cow cheeses were produced [
39]. In the present study, 37 samples were produced from the mixture of sheep and goat milk; 8 samples from solely sheep milk; 11 samples from goat milk; 9 samples from cow milk; 9 samples from mixture of sheep, goat, and cow milk; 1 sample from a mixture of sheep and cow milk; and 1 from other milk namely buffalo milk, while in 26 samples, the kind of milk used was not established because it was not reported on the package.
3.2. Discriminant Analysis
Discriminant analysis was applied to the 13 physicochemical cheese attributes to evaluate if the 100 samples could be distinguished by milk type. Two samples were excluded from discriminant analysis on the basis of milk because there was only a single sample per milk type group (buffalo milk or a mixture of sheep and cow milk). The milk type, cheese type, and location (cheese produced in a Greek island or in mainland) effects on composition and physicochemical characteristics of non-PDO/non-PGI cheeses for the present study are presented in
Table 4. Neither location, nor milk type, had any effect on the physicochemical cheese attributes (
p > 0.05), except from the color of goat cheese samples that were less yellow (lower b* values,
p < 0.05) in comparison with the cheese samples made from cow milk. Color changes could be observed as an effect of diet, breed, lactation time, light exposure during processing, packaging, and distribution [
40]. This difference in cheese color may reflect the milk color differences between goat and cows, which is attributed to their different carotenoid contents [
41,
42]. Bovines show a higher efficiency in the conversion of carotenoids, particularly β-carotene, into retinol in the enterocytes in comparison with ovines; therefore, cow milk is more yellow than goat milk [
43]. In close agreement with the results of the present study, Lucas et al. [
44] reported a β-carotene content of 3.78 mg/kg fat for 301 cow milk cheese samples, whereas the respective mean for 106 goat milk cheese samples was 0.00 mg/kg fat.
On the other hand, it was cheese type that had a significant effect on the composition and physicochemical characteristics (p < 0.05), except for redness (a*), which was unaffected (p > 0.05). Hard and semi hard cheeses showed higher protein, MNFS (p < 0.05) in comparison to soft, spread and whey cheeses. The fat and salt contents were shown to be higher in hard and semi hard cheeses compared with spread and whey cheese (p < 0.05). However, when fat was expressed on a dry matter basis, soft cheeses showed the greatest fat percentage, (59.8%), whereas hard and whey cheeses showed the least, at 49.9 and 45.6%, respectively (p < 0.05). The composition differences observed can explain color differences, namely more yellow (higher b* values) and less light color (decreased L* values) for hard cheeses compared with the rest of those in the present study (p < 0.05). The salt content differences observed were not altered when salt was expressed in the moisture content.
Cheeses exposed to natural and artificial light throughout processing, packaging, distribution, and at retailers are likely to be subjected to peroxidation. Light exposure causes the formation of off flavors, loss in nutritional value, and color changes, which rapidly impair product quality and marketability [
11]. As a result, it is important to classify cheeses according to their susceptibility to oxidation and thus their market life. Cheese type had a significant effect on oxidative stability (
p < 0.05). The lowest MDA values, indicative of an improved oxidative stability, were present in the hard and semi hard cheese samples, possibly due to their lower fat content on a dry matter basis compared with that for soft cheeses. However, because of the high SEM and the conservative Bonferroni adjustment method applied, no significant pairwise differences were detected. In general, the MDA values assessed in the present study were in accordance with those reported in previous studies [
13,
45,
46]. In detail, the impact on the biochemical changes in cheese during ripening (proteolysis, lipolysis, and glycolysis) may influence its sensory characteristics. Under this context, proteolysis and lipolysis reactions had a significant important role in the sensory of Gouda-type cheese [
45,
46]. In Cheddar cheese, it was shown that the antioxidant activity of water-soluble extracts was correlated to the ripening period. More specifically, a decrease in antioxidant activity after the fifth month of ripening was reported, which might be indicative that the antioxidant peptides were not resistant to further proteolysis. It seems that proteolytic enzymes of adjunct cultures may contribute significantly to the increased production of peptides related to the antioxidant activity and sensory properties [
47]. Similarly, in white brined goat milk cheese, the antioxidant activities of all of the examined samples increased until day 60, but then decreased at the end of the ripening period, indicating the role of using adjunct cultures in order to release bioactive peptides [
48].
Differences in MDA values may be correlated not only with the type of cheese and its manufacture procedure, but also the milk type used during cheese making. In the present study, milk type did not have a significant effect on MDA values (
p = 0.13), although higher values of MDA were noted in cheeses produced from sheep compared with cow milk (91.8 vs. 24.5%,
Table 4). Nevertheless, studies examining milk showed that sheep milk contained a higher PUFA content (~25%) compared with cow milk, leading to a higher risk for lipid peroxidation and consequently to a higher MDA content. Indeed, the MDA content in sheep compared with cow milk was higher (13.40 vs. 8.07%) after 24h of storage at room temperature [
49]. Moreover, the same researchers observed a strong positive correlation between PUFA and the index of lipid oxidation (MDA) [
49]. Additionally, the total antioxidant capacity, determined using ORAC assay, of low-fat pasteurized milk was significantly lower (13.624–13.984 μmol TE/l) than that of milk with a higher fat content (14.124–14.216 μmol TE/l) [
50]. A significant higher total antioxidant capacity, as determined by ABTS assay, was also found in cow milk with 3% fat than cow milk with 0.5–1.5% fat and skimmed milk [
51].
Two discriminant functions were found to be significant (
p < 0.01) for distinguishing the samples among the different milk types. A graph of the two discriminant functions is shown in
Figure 2.
The separation between milk type groups was not so clear and this was also evident by the close placement of the group centroids. As shown in
Table 5, the sample percentage that was correctly classified into the appropriate milk type group was estimated to be low, at 62.8%.
Cow, goat, and sheep milk group samples were less misclassified, as 33.3, 18.2, and 25%, respectively, in comparison with their mixtures that were misclassified to a greater extent, as 49.6, 55.6 and 42.3%, for sheep and goat, sheep and goat and cow, and not defined milk groups, respectively. The stepwise discriminant analysis showed that 9 out of 13 characteristics were primary responsible for the discrimination of samples into milk-type groups. These elements were color attributes L, a and b, salt, ash, FDM, MNFS, SM, and MDA contents. A plausible explanation may be related to the differences in milk chemical composition between the three animal species, namely cow, sheep, and goat, and manufacture procedure and ripening. Sheep milk has been reported to be preferable by several cheese producers due its higher protein, fat, and total solids content compared with cow milk [
37,
38]. Under this scope, it should be note that previous work by our team on PDO cheeses [
9] revealed that using physicochemical parameters and fatty acids during discriminant analysis, the percentage of the samples that were classified into the correct group according to the milk type, used for cheese manufacture, was 99.1%.
The discriminant analysis of the physicochemical cheese attributes according to the type of cheese is presented in
Figure 3.
There were three discriminant functions significant (P<0.01) for distinguishing the samples among the cheese-type groups. A plot of the first two discriminant functions is presented in
Figure 3. A clear separation between hard, soft, spread, and whey cheeses was observed, whereas semi hard cheese samples were clustered into the hard cheese group, confirmed from the hard and semi hard group centroids that were placed close together. As shown in
Table 6, 82.1% of the samples were correctly classified into the appropriate cheese type group, whereas a 17.9% were misclassified.
Hard cheese samples were misclassified as semi hard cheese at a rate of 14% (7 out of 50 samples), whereas one semi hard cheese sample was misclassified as hard cheese. Two soft cheese samples, out of 17, were misclassified as spread cheese and hard cheese, respectively. Two spread-type samples out of 11 were misclassified as soft-type cheeses, whereas the lowest rate of misclassification (11.8%) was observed for whey cheese samples. The stepwise discriminant analysis showed that 7 out of 13 characteristics were mainly responsible for the discrimination of samples into cheese-type groups. These elements were color attributes a and b, moisture, ash, fat, MNFS contents, and pH. A plausible explanation may be the differences in milk chemical composition between the three animal species, namely cows, sheep, and goats, and the manufacture procedure and ripening. Previously, using the total elemental fingerprint of PDO cheeses during discriminant analysis based on different milk types, 94.6% of cases were classified correctly [
10]. A reasonable explanation might be connected to various parameters used at discriminant analysis (physicochemical, elements, and the fatty acids) or the difference between PDO and cheeses of limited production, such as those used in the present study. Differences between PDO and cheeses of limited production not listed in the EU Geographical Indications Registers indicate the potential benefits of a registration of an agricultural product. Potential benefits of a registration may be related to standardized production technology (type of milk used, specific details during manufacture, etc.) that provide high and constant quality and safety of a product [
52]. As a certification of geographical indication is not applicable for those locally produced cheeses, the results of the present study can support and valorize local production. It is necessary to ensure that cheeses of limited productions are available to consumers all year round, as some of them are produced regionally and are hardly found throughout the country. Events, such as local festivals, agritourism, and regional celebrations, may be a way for manufacturers to promote their cheeses. On farm shops, e-commerce, local markets, and supermarkets may be appropriate actions to build a close relationship between purchaser and producer.