Obtaining and characterizing “dulce de leche” prepared with sheep’s and cow’s milk in different proportions

The “dulce de leche”, a viscous milk candy spread, is one of the most produced Brazilian dairy products. However, it is still mostly made from cow’s milk and the sheep’s milk rarely used in its production. This study aimed to obtain and physicochemically characterize the whole milks and the “dulce de leche” (DL) made from cow’s (CM) and sheep’s milk (SM), and CMI:SM mixing rations of 100I:0, 75I:25, 50I:50, 25I:75 and 0I:100. Milk’s physicochemical analyzes were pH, titratable acidity, density, moisture, total solids, proteins, lactose, fats, and ash. DL physicochemical analyzes were the same of the milk plus yield, color, and texture profile analysis. From inclusion of sheep’s milk, the DL presented higher values of total solids, ashes and protein, with lipids increased only from the proportion 50I:50. Lactose decreased at the proportion 0I:100, however, was similar to the 25I:75 proportion. The moisture decreased as the inclusion of sheep’s milk on the product. Regarding color, the difference from the standard sample was verified with the inclusion of higher proportions of sheep milk. The combination of sheep’s milk and cow’s milk is a technically viable alternative, as it considerably increases the nutritional value and yield of the product.


Introduction
The increase in the demand for products with healthy and sustainable appeal has created opportunities for poorly exploited markets, such as the use of sheep's milk in dairy products. This raw material has been highlighted in Brazil as a sustainable alternative, with advantages for small and medium-sized rural producers, and the potential of adding value to industrialized dairy products (Peruzzi et al., 2016). Sheep's milk has great nutritional value due to its high content of proteins, lipids, minerals and essential vitamins. Physicochemical characteristics such as high viscosity, small fat globules and expressive amount of short chain fatty acids (Revers et al., 2016;Balthazar, et al., 2017b), which make it increasingly popular in the market due to its peculiar taste and higher nutrient content and functional properties. For this reason, it can be used as a raw material in cheese making (Albenzio et al., 2015), dulce de leche , yogurt (Balthazar et al., 2015) and ice cream (Balthazar et al., 2017b).
The "dulce de leche" is one of the most produced dairy product in Brazil, which consists of a concentrated milk obtained by heat treatment with or without negative pressure, with the addition of ingredients, mainly sucrose, providing differential sensory and physicochemical characteristics compared to other dairy products and highly consumed in Latin America. On addition, some countries such as Brazil have exported to other economic blocs, such as the European Union and Asia Pacific Economic Cooperation, highlighting the process of expansion of the dulce de leche exportation market . "Dulce de leche" is still mostly made from cow's milk, while sheep's milk (which is commonly present in the production of cheeses) is rarely, if ever, used as a raw material (Chacón Villalobos et al., 2013). Furthermore, sheep's milk is significantly more expensive, and therefore less attractive from the standpoint of adding value to industrial products. However, combining both types of milk could be an alternative for adding nutritional value without considerably increasing the final product's price. The objective of this work was to obtain and physicochemically characterize the milks and "dulce de leche" made from the combination of cow's and sheep's whole milks, in different proportions.

Raw material
Sheep's milk (SM) and cow's milk (CM) used in this research were purchased from regional producers and transported in thermal boxes containing ice. Both in natura milks were pasteurized slowly (65 °C for 30 minutes) using a Thermomix (Vorweker  ), and then packed in 900 mL, duly sanitized plastic packages, for storage at -20 °C in a freezer (Consul  ).
Fractions of the milks were thawed in a water bath (Tecnal  ) at 60 °C, to prepare the mixtures, and standardized to 0.13 g.L -1 sodium bicarbonate (Vetec  ), followed by the production of the "dulce de leche" samples.

Milks' microbiological analyses
To confirm the efficiency of the milk pasteurization process, microbiological analyzes for mesophiles, psychrotrophs, molds, yeasts, Salmonella, Staphylococcus aureus and thermotolerant coliforms were carried out, following Association of Dfficial Analytical Chemists (2000).

Characterization of "dulce de leche" samples
Characterization of the samples followed Protocol nº 354/1997, which establishes the following parameters (g/100 g)I: moisture (max. 30.0); fat (6.0 to 9.0); ash content (max. 2.0) and proteins (min. 5.0). These analyses followed Association of Dfficial Analytical Chemists (2000) methodologies. Lactose analysis was also performed by redox titration using alkaline CuSD 4 (Fehling's solution) (Onstituto Adolfo Lutz, 2008). Color analysis was performed using a colorimeter system (Mini Scan EZ Hunterlab 4500L), in three scalesI: L* a* b*. The color difference between the formulations is given by ΔE*, according to Equation 1. The 100I:0 (CMI:SM) ratio was used as the standard. When ΔE* < 1, no color difference is perceived by the human eye; when 1 < ΔE* < 3, small color differences may be perceived, depending on the tonality; and when ΔE* > 3, color differences are obvious to the human eye (Bodart et al., 2008).

Statistical analysis
The "dulce de leche" formulations were performed in duplicate and the analyzes in triplicate. Samples were submitted to analysis of variance (ANDVA) and Tukey's test at 95% significance (p ≤ 0.05), using R software.

Milks' characterization
The results of the physicochemical analysis of CM, SM and their mixtures (Table 1) indicated higher values of density, titratable acidity, total solids, ashes, proteins, lipids and lactose when 0I:100 CMI:SM proportion was used. For instance, the analyzed parameters increased gradually along with the proportion of SM in the mixtures.
Sheep's milk has higher solid contents and, therefore, greater density values. The results showed that there were significant differences (p ≤ 0.05) among 100I:0, 25I:75 and 0I:100 samples. As shown, this study found a total solids value of 1.036 g.mL -1 in the 0I:100 CMI:SM mixture, in comparison to 1.028 g.mL -1 in the 100I:0 mixture. This value is line with the current legislation, which establishes total solids values between 1.028 and 1.034 g.mL -1 (Brasil, 2002) for cow's milk. Dur results to sheep's milk are similar to those found by Park et al. (2007) (1.034 to 1.038 g.mL -1 ).
The higher titratable acidity found in the 0I:100 CMI:SM formulation may be explained by the availability of lactose in this milk, given that titratable acidity is a function of the relationship between lactose availability and lactic acid production by microorganisms, leading to an increase in lactic acid (Revers et al., 2016). Results also showed that there was a significant difference (p ≤ 0.05) among samples 100I:0, 25I:75 and 0I:100. Park et al. (2007) reported sheep's milk had an acidity between 16 and 28 °D, while the titratable acidity of cow's milk was between 15 and 18 °D. Based on that, the values found on the present research are in accordance with published data.
The 100I:0 CMI:SM formulation had a higher pH when compared the 0I:100 CMI:SM formulation, with a gradual decrease (p ≤ 0.05) in pH along the incorporation of SM in higher proportions. Sheep's milk's pH varies from 6.63 to 6.68, according to Park et al. (2007). The amounts of casein, phosphates and other acid components in the milk's dry matter cause this oscillation, with acidity being higher in milks with more protein content.
The total dry extract reflects the constituents of milk (fat, proteins, lactose and minerals), and sheep's milk has higher solids concentration than other species of milk (Fava, 2012). This explains the 19.20% of total solids (0I:100 CMI:SM), which was statistically different (p ≤ 0.05) from the values found for 100I:0 CMI:SM proportion (11.12%).
The ash value in the 100I:0 CMI:SM proportion was 0.71%, which differed statistically from the others (p ≤ 0.05). The values reported in the literature vary between 0.53-0.8 and 0.85-1.0 for cow's milk and sheep's milk, respectively (Guerra et al., 2008;Park et al., 2007).
The protein levels reported by Guerra et al. (2008) and Park et al. (2007) ranged from 3.2 to 7.0% for sheep's milk, and from 3.0 to 3.9% for cow's milk. These values were similar to those found here for the 0I:100 and 100I:0 CMI:SM proportions. Ot can be noticed that as the inclusion of SM on the formulations, protein concentration increased, and all treatments were statistically different (p ≤ 0.05).
According to the analyses, the 100I:0 CMI:SM proportion had a 45.5% fat value, lower than in the 0I:100 CMI:SM proportion. Brito et al. (2006) found 5.79% fat in milk from Lacaune sheep, which is below the values found here (6.0%). This may have occurred due to seasonal variations, lactation period and milking shift (Guerra et al., 2008).
Lactose contents in all proportions were statistically different (p ≤ 0.05), and the values found here for cow's milk were lower than those reported in the literatureI: 4.3% according to Dliveira & Timm (2006). Lactose values for sheep's milk were lower than those found by Czarnobay et al. (2017), of 4.71%.

Process of solids concentration for "dulce de leche" production
The yield of the 0I:100 CMI:SM formulation was higher when compared to the other formulations (Table 2), which can be explained by the sheep's milk solid's concentration. This amount increased together with the sheep's milk content in the mixture. Formulation 100I:0 CMI:SM had the lowest yield, confirming the importance of total solids for the cost-effectiveness and value conferred to this product. The 50I:50, 25I:75, and 0I:100 CMI:SM formulations showed no significant differences (p ≤ 0.05) for yield, pointing that the substitution of up to 50% of cow's milk by sheep's milk may be advantageous in terms of product yield.

Characterization of "dulce de leche" samples
According to Table 3, moisture values decreased as the SM was included on the formulation, possibly due to the high levels of total solids present in sheep's milk, i.e., "dulce de leche"   (Brasil, 1997).
Ot is worth noting that low moisture improves the conservation of the "dulce de leche" (Demiate et al., 2001).
Concerning ashes analysis, there were statistical differences between the 100I:0 CMI:SM and 0I:100 CMI:SM formulations. These ash contents may be related to milk density, given that the higher the amount of minerals and organic substances, the greater the density. The increase in ash contents occurs during the "dulce de leche" manufacturing process, when milks loose water, leading to a higher concentration of minerals (Pellegrini et al., 2013). The protein contents of the "dulce de leche" formulations followed the same milk-composition pattern seen elsewhere, and protein values were statistically different for all formulations (p ≤ 0.05). The 100I:0 CMI:SM formulation had the lowest protein content, while the 0I:100 CMI:SM formulation had the highest protein content.
Lipid contents in the formulations were 6.00% for the 100I:0 CMI:SM proportion and 11.50% for the 0I:100 CMI:SM proportion and differed statistically (p≤0.05). The physicochemical fat standard established by the Technical Regulation for the Odentification and Quality of "dulce de leche" (Brasil, 1997) is a maximum of 9.0%. The 25I:75 and 0I:100 CMI:SM formulations presented fat values above this standard, due to the naturally higher fat contents of the sheep's milk's, not yet regulated by Brazilian legislation. For ash, moisture and proteins, all formulations were in accordance with current legislation (Brasil, 1997).
The results obtained in this research were similar to those reported by Laguna (2000) for "dulce de leche" made from goat's milkI: average moisture content (19.44%) lower than the maximum standard stipulated by the legislation for cow-milk "dulce de leche"; and average protein contents (13.14%) and lipids (11.5%) higher than official requirements (also for cow's milk), evidencing the differences between each type of milk.
The lactose concentration was higher in the 100I:0 CMI:SM than in the 0I:100 formulation, decreasing along with the decrease in cow's milk contents. The lower lactose value of sheep's milk is possibly due to the relationship between lactose and total solids, providing better digestibility and can be considered an ideal cow's milk substitute for allergic people (Balthazar et al., 2017a).
Studies carried out by Niro et al. (2014) with different proportions of sheep's milk and cow's milk in cheese production showed that the addition of sheep's milk (18%) influenced nutritional composition, with higher protein (4.0%), fat (4.6%) and lactose (4.3%) values when compared to cheese made from 100% cow's milk (3.6% protein, 4.4% fat and 4.0% lactose). The increase in the proportion of sheep's milk in the formulations resulted in an increase of important constituents, such as proteins, and a decrease in lactose when compared to standard "dulce de leche" (100I:0 CMI:SM). As such, sheep's milk proves to be a viable alternative to improve the product's nutritional characteristics.
Results for the texture profile analysis of "dulce de leche" formulations are presented in Table 4. There were no significant hardness differences (p ≤ 0.05) between the formulations, that is, the different proportions of sheep's milk did not influence this parameter. On regard to adhesiveness, only the 100I:0 CMI:SM formulation presented a significant difference (p ≤ 0.05). Adhesiveness is defined as the energy required to overcome the attractive forces between the food surface and other materials in contact with it (Szczesniak et al., 1963). Formulations containing sheep's milk in different proportions led to a decrease in the adhesiveness, possibly due to the higher lipid contents of sheep's milk (Table 1).
Cohesiveness values in the studies conducted by Francisquini et al. (2016) were between 0.44 and 0.93 for commercial "dulce de leche" based on cow's milk, close to those seen in this study (0.82 to 0.96). Cohesiveness-wise, only the 75I:25 CMI:SM sample differed statistically from the others (p ≤ 0.05).
Elasticity of the samples increased along with the proportion of sheep's milk in the formulations, which is probably due to the higher protein and lipid content of sheep's milk (Tables 1). Elasticity values were similar to those found by Francisquini et al. (2016), which ranged from 9.94 to 25.11 in commercial "dulce de leche".
Gumminess decreased along with the increase of sheep's milk in the mixtures (Table 4). The higher fat content (Table 1) in sheep's milk facilitates the disintegration of the sample. Thus, the formulations with the highest proportions of sheep's milk had the lowest gumminess values. Chewiness, hardness, adhesiveness, gumminess, cohesiveness and elasticity vary according to solids concentration, temperature and process time (Carvalho et al., 2017). The 75I:25 CMI:SM formulation had the smallest amount of color deviation (ΔE*) from the standard formulation (100I:0 CMI:SM), with a barely perceivable tonal difference. The 0I:100, 50I:50 and 25I:75 formulations, however, presented greater, easily perceived differences. Among the formulations, none had color differences imperceptible to the naked eye (ΔE* < 1) when compared to the standard. This was due to the distinct characteristics of the different milk compositions, which influence color, flavor and aroma of the final product.
According to Shibao & Bastos (2011), among the existing browning reactions, the Maillard reaction, which occurs in a wide variety of foods including "dulce de leche", causes the most important changes in color, taste, nutritional value, antioxidant properties, and food texture. On fact, determinants for the course of the reaction are processing at temperatures above 40 °C, pH between 6 and 8, and water activity between 0.4 and 0.7. Sheep's milk has a higher lactose content than cow's milk, probably hindering the development of the Maillard reaction during the concentration stages. According to Fennema et al. (2010), the Maillard's start depends on the reaction of the aldehyde group of lactose with the ε-amino group of lysine. The light color is due to the low contents of reducing sugars (glucose) in the formulations used in this study.
Mixtures of milk from different species have a potential for dairy market, because together, they can improve the sensory and physicochemical properties, adding value to these products Queiroga et al., 2013). Further sensory analysis of "dulce de leche" formulations is recommended by qualitative methods. On the study performed by Pinto et al. (2018), with different fermented milk formulations, the authors demonstrated viable and effective application of sensory analysis, to clarify consumer perceptions and commercial viability.

Conclusion
The results obtained in this study show the greater nutritional richness of sheep's milk in comparison to cow's milk, pointing to the potential of this milk's use, both as exclusive raw material and mixture component, in the manufacturing of "dulce de leche". The combination of sheep's milk and cow's milk is a technical and nutritionally viable alternative, as it considerably increases the yield of the product. However, sensorial properties and the cost-effectiveness of this product should be considered.