Chemical, physicochemical, microbiological and sensory characterization of cow and buffalo ghee

carolina.pena@udea.edu.co Abstract Due to the increasing consumption of ghee in the Western countries, a complete characterization of buffalo and cow ghee was performed to complement and update the available literature. Ghee is a lipophilic dairy product with 98.9% lipids, 0.3% water and less than 0.9% nonfat solids. Fatty acids are the major lipid fraction and represent 85.1% and 83.65% for buffalo and cow ghee, respectively. More than 52% of the fatty acids were saturated, and palmitic (24-28.8%), stearic (9.4-14%) and myristic (8.5-10%) acids were predominant. Monounsaturated fatty acids were approximately 23.8% and the major component was oleic acid. Polyunsaturated fatty acid content was 2.45% (buffalo) and 4% (cow). The vaccenic acid (2.18%) and the conjugated linoleic acid (CLA cis-9, trans-11) with a concentration of 0.77% in buffalo and 1% in cow ghee, were the main ruminant trans fatty acids. The physicochemical and microbiological characteristics of cow and buffalo ghee complied with the literature and national regulation. Finally, the sensory profile of buffalo and cow ghee was defined with a predominantly lactic odor, followed by cooked and fatty notes. The taste was characterized as fatty, lactic, sweet and cooked; and the texture was described as fatty with fatty mouthfeel, lumpy and greasy acids. Practical Application: This study intends to be a basis for the analysis and discussion of ghee intake and the benefits and risks to human


Introduction
Ghee also known as clarified butter, is an ancient dairy product prepared by heating milk, cream or butter over 100 °C to evaporate water and precipitate the nonfat solids (Andrewes, 2012;Antony et al., 2018;Sharma et al., 2010;Sieber, 2005). Ghee is widely produced and consumed in India, Sudan, Ethiopia and the Middle East (Antony et al., 2018;El-Shourbagy & El-Zahar, 2014;Kumar et al., 2000); nevertheless, in the last decade, the American continent, with USA, Argentina and Paraguay, as the main producers, has increased the production of cow ghee between 3000 and 12000 tons per year (Food and Agriculture Organization of the United Nations, 2019). In the same way, Western countries have displayed increasing ghee intake (Antony et al., 2018) as a result of the globalization process as well as the replacement of the consumption of margarine due to the high content of industrial trans fatty acids (iTFA). According to the current scientific evidence, iTFA has exhibited a higher negative impact on cardiovascular disease, diabetes and even depression than saturated fatty acids (Ford et al., 2016;Mozaffarian et al., 2009;Park, 2009;Vučić et al., 2015). Galvín et al. (2016) reported that one gram of iTFA has 15-fold higher risk of coronary disease than one gram of saturated fatty acids.
Due to the lipophilic composition of ghee, as well as the scientific evidence of the association between fats intake and the risk of coronary disease (Ford et al., 2016;Galvín et al., 2016;Vučić et al., 2015), and the increasing trend on ghee consumption exhibited by the Western countries (Antony et al., 2018), the current study aims to characterize buffalo and cow ghee in terms of the proximate composition, and the physicochemical, microbiological and sensory characteristics to be a basis for the analysis and discussion regarding ghee consumption and the benefits or risks to human health and nutrition.

Materials and methods
Three batches of each type of ghee were produced using 100 L of milk. Three samples of each batch, for a total of nine samples, were taken and analyzed.

Materials
Buffalo milk was collected from Vegas de la Clara farm located in Gomez Plata, Antioquia, Colombia, and cow milk was obtained from La Montaña farm located in San Pedro de los Milagros, Antioquia, Colombia. Both farms are property from the University of Antioquia. Chemicals for chromatographic analyses, cis/trans linoleic acid methyl ester mix, FAME mix food industry of 37 components fatty acids from C4 to C24 (Restek), triglyceride triundecanoin, hexane, boron trifluoride and NaOH/methanol were purchased from Millipore-Sigma. Chemical reagents and solvents for physicochemical and proximate composition (NaOH, HCl, ether, ethanol, KOH, KI, sodium thiosulfate and Wij´s solution) were of analytical grade and were purchased from Millipore-Sigma.

Extraction of cream and butter production
Buffalo or cow milk (100 L) was warmed at 40 ± 2 °C and then skimmed using a Slavic beauty centrifugal cream separator. Cream was pasteurized at 85 ± 3 °C for 10 min and stored at 4 °C for 24 h. Buffalo and cow butters were obtained by churning cold cream (< 10 °C) in a butter-making machine and then, these two were packaged in vacuum packaging and stored at 4 °C until ghee production. Both processes were carried out at the dairy plant located at Vegas de la Clara farm.

Ghee production
Ghee was prepared by the creamery butter (CB) method, where the butter was heated at a temperature between 85-100 °C with periodic stirring until water evaporation and the sedimentation and development of golden-colored nonfat solids (NFS). Thereafter, the heating temperature increased to 113 °C for 3 min, and finally, the ghee was cooled to 60 °C, filtered, packed in glass containers with metal lids and stored in darkness at room temperature (24 ± 2 °C) until analytical procedures. The production of ghee was also performed at the dairy plant located in Vegas de la Clara farm.

Chemical analyses
Ghee samples were analyzed by gas chromatography to identify and quantify the fatty acid profile. Chemical analyses were performed to assess proximate composition and physicochemical characteristics of ghee.

Lipid extraction and methyl derivative preparation
150-200 mg of ghee sample was weighed and extracted with 50 mL of hexane using the Soxtec 2050 equipment following the established program (20 min boiling, 40 min rinsing, 10 min recovery and 10 min pre-drying) and then hexane was evaporated at 70 °C. The extracted lipid material was combined with 1 mL of the internal standard, triundecanoin (> 98%) with concentration of 5 mg/mL, and 4 mL of 0.5M NaOH/methanol. The mixture was heated to 100 °C/10 min and 5 mL of 7% boron trifluoride in methanol was added and heated for another 2 min. Then, 4 mL of hexane was added, and heated for 1 min. Thereafter, it was cooled to room temperature and a saturated sodium chloride solution was added. The upper layer that contained the fatty acid methyl ester derivative (FAME) was transferred to an Eppendorf vial and stored at -18 °C until chromatographic analyses.

FAME identification and quantification
The saturated (SFA), monounsaturated (MUFA) and polyunsaturated (PUFA) fatty acids contained in buffalo and cow ghee were assessed by a gas chromatograph with flame ionization detector-GC/FID (Agilent Mod. 7890B) equipped with a TR-CN100 column (60 m × 250 μm ID × 0.20 μm film thickness). GC/FID conditions were as follows: injector temperature 260 °C, detector temperature 300 °C, split ratio 100:1, helium carrier gas with a flow rate of 1.1 mL/min, injection volume 1 μL and temperature program: 90 °C/7 min followed by heating at 5 °C/min to 240 °C/15 min. GC-FID conditions for ruminant trans fatty acids (rTFA) quantification were as follows: injector temperature 220 °C, detector temperature 250 °C, split ratio 50:1, helium carrier gas with a flow rate of 1 mL/min, injection volume 1 μL and the temperature program 170 °C/5 min followed by heating at 5 °C/min to 240 °C/5 min. FAME identification was performed by comparing the retention times with those of chromatographic standards, and quantification was carried out according to the standard method 996.06 (Association of Official Analytical Chemists, 2016).

Proximate composition
The protein content of cow and buffalo ghee was assessed following the Kjeldahl standard method 954.01 (Association of Official Analytical Chemists, 2016) converting the nitrogen to protein using the factor 6.38 and the water content was determined by drying the samples at 105 ± 1 °C to constant weight according to the standard method 920.116 (Association of Official Analytical Chemists, 2016). The ash concentration was quantified by calcination at 550 °C until the residue was white following the standard method 920.117 (Association of Official Analytical Chemists, 2016). Lipid concentration was measured according to the direct method used for butter, standard method 938.06 (Association of Official Analytical Chemists, 2016), and the energy content was evaluated by an automated adiabatic calorimeter.

Physicochemical characteristics
The acid, peroxide, saponification and iodine values of cow and buffalo samples were assessed. The acid value was determined by mixing 20 g of ghee and 50 mL of neutralized ethanol, then the mixture was titrated against 0.1N NaOH according to the standard method 940.28 (Association of Official Analytical Chemists, 2016). The peroxide value was assessed by mixing 5 g of ghee, 0.5 mL of saturated KI solution and 30 mL of a solvent mixture of acetic acid and chloroform (3:2), followed by titration against 0.1N Na 2 S 2 O 3 using 1% starch as indicator according to standard method 965.33 (Association of Official Analytical Chemists, 2016). The saponification value was quantified following the standard method 920.160 (Association of Official Analytical Chemists, 2016). 5 g of ghee and 50 mL of alcoholic KOH solution were combined in a flask, thereafter, the flask was connected to a reflux condenser and boiled for an hour, then, the cooled mixture was titrated against 0.5N HCl using phenolphthalein as indicator. Finally, the iodine value was measured according to the Wijs method 920.159 (Association of Official Analytical Chemists, 2016).

Microbiological analyses
The mesophilic aerobic bacteria and Staphylococcus coagulase positive were quantified (CFU/g), the total and fecal coliforms were estimated (MPN/g), and Salmonella sp. and Listeria monocytogenes were detected. The mesophilic aerobic bacteria were enumerated on plate count agar incubated aerobically at 30 °C for 72 h. Staphylococcus coagulase + were enumerated on Baird-Parker agar with egg yolk tellurite incubated at 36 °C for 48 h. Coliform bacteria were estimated on Fluorocult® Brillant Green 2%-Bile (BRILA) broth, incubated at 37 °C for 24-48 h. Tubes were checked under UV light (λ=366 nm), and when observed a positive response, the fecal coliforms were confirmed by adding KOVACS indole reagent until the appearance of a red ring. The presence of L. motocytogenes was determined in 25 g of sample incubated on VIDAS® LMX broth at 37 °C for 26-30 h, and Salmonella sp. was detected in 25 g of ghee sample according to the culture enrichment procedure 6579-1 (International Organization for Standardization, 2017).

Sensory analysis
Buffalo and cow ghees were evaluated using the method of sensory profile by a multidimensional approach (Instituto Colombiano de Normas Técnicas y Certificación, 1996) which is a descriptive analysis widely used in dairy products (Kwak et al., 2016;Silva et al., 2018;Torres et al., 2017). Sensory tests were performed after the confirmation of the microbiological safety of ghee samples. The sensory evaluation was carried out by a trained panel of six members (both sexes, ages ranging between 25 and 55 years old) in the sensory room of the Nutrition and Diet School from University of Antioquia (Medellin, Colombia). Ghee samples were placed in plastic cups coded with random 3-digit codes and presented in sequential order to panelists under white fluorescent light at room temperature 24 ± 2 °C and relative humidity of 48 ± 1%. First, the panelists identified and selected the descriptors for establishing the sensory profile of ghee, and thereafter, the intensities of the descriptors were evaluated in a scale from 0 to 5 (where 0 = low and 5 = high) and the global quality was assessed in a scale from 1 to 3 (where 1 = low and 3 = high). The descriptors were established for the sensory attributes of appearance, odor, flavor and texture.

Statistical analyses
The data were subjected to multivariate analysis of variance -MANOVA in order to evaluate the overall statistical difference between ghee samples considering all the correlated response variables. Thereafter, the analysis of canonical correlation -ACC allowed us to establish multidimensional differences between cow and buffalo ghee. The biplot analysis based on Pythagorean dissimilarities and central transformation was also carried out. The statistical analyses were performed with the free software SAS University edition and R-project v 3.4.4.

Results and discussion
A complete characterization of buffalo and cow ghee was performed. The analyses included the evaluation of the proximate composition and the fatty acid profile, as well as the physicochemical, microbiological and sensory characteristics of ghee.
The major component of ghee samples was the lipid fraction comprising approximately 98.9% of the product, less than 0.9% nonfat solids and 0.3% of moisture. Buffalo and cow ghees exhibited highly statistical differences (p < 0.0001) on the ash content while the protein, moisture, energy and lipid compositions were similar (p > 0.0001) in both samples (Table 1). When all the nutritional components were evaluated simultaneously, the analysis of the canonical correlation (ACC) allowed to establish that buffalo and cow ghee displayed similar gross composition. According to Ganguli & Jain (1972) and Sserunjogi et al. (1998), the moisture and lipid content of ghee should be below 0.3% and above 96%, respectively; thus, the proximate composition of cow and buffalo samples complied with the reported values.
Moreover, ghee samples displayed no oxidation after the production since the peroxide value was 0 meq-g/kg; nonetheless, cow ghee showed higher acid value than buffalo ghee (Table 1). Based on that result, it is likely that cow ghee undergoes higher oxidation and rancidity throughout the time. On the other hand, the saponification value of the samples exhibited no difference which means that the molecular weight of the fatty acids that are present in the buffalo and cow ghee was similar and corresponded to long-chain fatty acids above 14-carbons as can be seen in Table 2. According to the iodine value, cow ghee contained a higher amount of unsaturated fatty acids than the buffalo ghee. This result was also confirmed with the higher concentration of MUFA and PUFA contained in cow ghee (28.31%) regarding 25.19% exhibited by the buffalo sample ( Table 2). The MANOVA analysis that was performed to the physicochemical characteristics of the two products showed highly statistical difference (p < 0.0001) and the ACC analysis that was applied for evaluating simultaneously all physicochemical characteristics of ghee samples indicated that buffalo and cow samples displayed differences. Sserunjogi et al. (1998) reported a maximum concentration of free fatty acids of 0.3% and a peroxide value below 1. Similarly, Ganguli & Jain (1972) described the saponification value of ghee above 220 and the iodine value between 26 and 38. Regarding the content of free fatty acids and the peroxide value, the buffalo and cow ghees are in accordance with the values reported in the literature; nonetheless, they displayed iodine values outside the reported range and as mentioned above, it was a consequence of the higher concentration of unsaturated fatty acids contained in cow ghee and the lower content in the buffalo sample. Likewise, cow ghee exhibited lower saponification number than informed by Ganguli & Jain (1972). Despite of the differences exhibited by the ghee samples evaluated in this study and the reported values for the iodine and saponification indices, the authors consider that such variation cause no negative effect on the quality of ghee, and it is likely a result of a different fatty acid profile produced by the animal feeding (Ferlay et al., 2017).
Due to triacylglycerols (TAG) are the main fraction of the lipids contained in bovine milk, and comprise approximately 98% of the lipids (Liu et al., 2018), TAG are also the dominant component in ghee (Antony et al., 2018). The main element of triacylglycerol is the fatty acid, and their concentration in buffalo and cow ghee were 85.1% and 83.65%, respectively. Saturated fatty acids (SFA) were predominant and comprised above 55% of the total fatty acids, followed by approximately 23.5% of monounsaturated fatty acids (MUFA) and polyunsaturated fatty acids (PUFA) up to 4% ( Table 2). The MANOVA analysis established a highly statistical difference (p < 0.0001) between the fatty acid profiles of the two products. Moreover, when all the SFAs were evaluated simultaneously, the ACC established that buffalo and cow ghee had different SFA profiles. Similarly, the polyunsaturated and monounsaturated fatty acids exhibited differences between samples.
Buffalo and cow ghee exhibited higher SFA content than the reported by Dwivedi et al. (2002) and Sserunjogi et al. (1998) with SFA concentrations of 47.8% and 46%, respectively. Nonetheless, it was similar to the proportion determined by Manickavasagan & Al-Sabahi (2014) of 59%.
For several years, margarine was recognized by the scientific community as well as the consumers, as a healthier source of lipids due to the higher PUFA and lower SFA content than that in butter. Despite that, a study performed by Vučić et al. (2015) on 13 Serbian margarines showed SFA contents between 22.76% and 51.17%. Hence, margarines might contain similar or even higher concentration of SFA than that in butter and/or ghee. Some studies have reported a content of MUFA and PUFA in ghee of 32.8% and 2.9% (Manickavasagan & Al-Sabahi, 2014) and 36% and 5% (Sserunjogi et al., 1998), respectively. Both studies displayed higher MUFA and PUFA content than the ghee samples that were evaluated in this study (Table 2).
Currently, Colombia has no microbiological regulation for ghee, as well as there is no international microbiological standard from Codex Alimentarius; hence, microbiological characterization of ghee (Table 3) was based on butter. According to Colombian regulation, Resolución 2310/1986 (Colombia, 1986), butter must comply with total coliform below 150 MPN/g, fecal coliform < 3 MPN/g, Staphylococcus coagulase positive < 200 CFU/g and no presence of Salmonella. Due to Listeria monocytogenes is a pathogenic bacterium, its presence was also verified. As observed in Table 3, cow ghee exhibited higher content of microorganisms than buffalo ghee; nonetheless, ghee  samples complied with national regulation for butter, and thus, both products were claimed as safe for the sensory evaluation.
The sensory profile of buffalo and cow ghees were established by a trained panel of six members and the results are presented in Figure 1. To date, and to the best of our knowledge, no study has been accomplished on the organoleptic characteristics of ghee.
The statistical analysis MANOVA indicated a highly statistical difference (p < 0.0001) between the sensory profiles of cow and buffalo ghees. According to the results, the odor profile of ghee was defined as predominantly lactic followed by cooked and fatty odor with sweet and butyric notes (Figure 1a). The lactic odor was perceived in a higher proportion in buffalo than in cow ghee. The taste profile was characterized mainly by fatty followed by lactic, cooked and sweet notes (Figure 1b). Both samples displayed differences mainly in the cooked and fatty descriptors.
The texture profile was described as predominantly fatty, and exhibited fatty mouthfeel, lumpy and greasy texture. The cow and buffalo samples displayed also slight sandy texture ( Figure 1c); nonetheless, cow ghee displayed the highest presence of particles (Figure 1d). The texture profile of cow and buffalo samples showed high difference in the fatty mouthfeel as well as the sandiness. The global quality of buffalo ghee was evaluated with a higher score by the panel (Figure 1e) due to the best flavor of the sample. According to the biplot analysis (Figure 1f), the sensory evaluators estimated more accurately buffalo ghee than cow ghee since the latter exhibited the highest dispersion.
The sensory profile of cow and buffalo ghees determined in this study may be a reliable tool for the dairy industry of the Western countries that are currently standardizing the production process. The organoleptic profile allows the dairy industry to adjust the process conditions as well as to establish the physicochemical and proximate characteristics of milk that are required in order to guarantee the best quality of the final product (Torres et al., 2017).
In the same way, the authors consider that further affective analyses with consumers may be a good alternative to study their product perception Pinto et al., 2019). Qualitative techniques will allow the dairy industry to improve and/or adjust sensory attributes of ghee according to customers´ preferences and needs (Oliveira et al., 2017). Moreover, they will also promote the dairy product among new users, enabling the growth of the dairy sector Pinto et al., 2019).

Conclusions
Despite the existence and consumption of ghee for a long time, the scientific literature is limited and dated. Based on that, as well as the increasing trend exhibited by the Western countries on the product intake, the current study performed a complete characterization of buffalo and cow ghee. Ghee exhibited 0.3% water and 98.9% lipids, where triacylglycerols were the main lipid component. Saturated fatty acids (SFA) were present over 52% of the total fatty acid content, followed by monounsaturated (MUFA) and polyunsaturated (PUFA) fatty acids. Buffalo ghee displayed more SFA content while cow ghee showed higher MUFA and PUFA concentration. The main fatty acids contained in ghee were palmitic, oleic, stearic and myristic acids. Vaccenic acid (2.18%) and conjugated linoleic acid (0.77% in buffalo and 1% in cow ghee) were the main ruminant trans fatty acids. The physicochemical and microbiological quality of buffalo and cow ghees were in accordance with literature and/or national or international regulation. The sensory profile of buffalo and cow ghee was characterized by an odor described as predominantly lactic, cooked and fatty, while the taste was defined mainly as fatty, lactic, sweet and cooked. The texture was identified as fatty with fatty mouthfeel, lumpy and greasy notes, and finally, the appearance was described by no uniform color and the presence of particles, exhibiting higher content in cow than in buffalo ghee.