The different samples were purchased from a local supermarket (Tottus), located in Santiago, Chile. The dressings were as follows: VEG (Notmayo, vegan dressing-type mayonnaise) as a control, HOM (Hellmans, traditional homemade recipe mayonnaise), CRE (Kraft, high-fat-mayonnaise), and LIG (JB, light-mayonnaise). Table 1 and Table 2 show the nutritional information and ingredient list for each of them, reported in labeled brands.

Physical properties such as water activity, average droplet size, rheological properties, and color measurement were measured along with the chemical properties FT-IR infrared spectroscopy and oxidative stability.

All measurements were performed in triplicate. The results were reported using their average value and standard deviation of at least three measurements. The statistical analysis of these results was evaluated using the one-way ANOVA analysis of variance. For significant differences among samples, multiple comparisons were made using the Tukey test, with a significance level of 95%, using the GraphPad Prism 5.01 software (GraphPad Prism Ink, USA).

Water activity (aw) is the water available for microbial development and/or development of biochemical reactions within a food. This property is related to how much food is perishable or not (Fenoglio et al., 2020). Regarding this property, Table 3 shows the results for the different samples, wherein in all cases, their values are high (> 0.9) due to the nature of the mayonnaise which corresponds to an oil emulsion with high water content. Where the CRE sample shows a slight significant (p < 0.05) decrease in its water activity (0.938 ± 0.003), which may be because it is the sample with the highest fat content (Table 1). Besides, in the case of VEG, CAS, and LIG, the main ingredient is water (aw = 0.973 ± 0.003); while for CRE, it is oil, so decreasing the aw value. This is corroborated in table 2, where the list of ingredients, which are in decreasing order of initial weight can be seen. The results obtained are comparable to the study carried out by Amin et al. (2014), where a low-fat mayonnaise was developed with different hydrocolloid gum types and levels. The results showed that water activity depended on oil concentration, where at 75% oil has a lower water activity (0.89 ± 0.01), while the sample with 45% oil increased its water activity (0.94 ± 0.02). Considering that all samples showed high aw, it was necessary to add ingredients such as acetic acid, salt, and EDTA (Ethylenediaminetetraacetic acid) to control microbial and biochemical reaction developments. These contents can be observed in Table 2.

Color is one of the most important properties which affects the appearance and acceptability of mayonnaise and the traditional mayonnaise recipe is characterized by a bright yellow appearance (Drozłowska et al., 2020). Table 4 shows the L* a* b* parameters, the color differences between traditional mayonnaise compared to VEG (∆E (00)), and the yellowness index (Y) of the different samples. For all samples, the color was significantly different (p < 0.05) among samples. The highest L* mean value corresponded to HOM, being the sample with the highest lightness, and VEG, the one with the least lightness. According to Mun et al. (2009), who mentions that the decrease in L* parameter can be attributed to the presence of solid protein particles, the studied VEG sample showed the highest protein content (Table 1) and therefore, the lowest lightness (Table 4). At the same time, the lower L* parameter of the LIG sample is explained by the higher content of thickeners used in its formulation compared to the other samples (Amin et al., 2014). Low-fat mayonnaise obtained similar results to flaxseed meal extract (Drozłowska et al., 2020). The a* and b* parameters indicate the tendency towards red or green and yellow or blue, respectively, if it is a positive or negative value. In all samples, a* and b* showed significant differences (p < 0.05) among samples, where positive values were obtained for both parameters, so a tendency towards red and yellow, respectively. In the VEG sample, b* value and yellowness index (Y) were lower than the other samples because the chickpea’s protein content controls these color parameters. Similar results were reported by Raikos et al. (2020) for a vegan mayonnaise using canned chickpea aquafaba, where the sample that contained a greater quantity of aquafaba indicated a lower value of b* and therefore of Y. On the other hand, the HOM sample showed the highest a* value = 22.70 ± 2.58, but CRE samples showed the highest b* value = 17.17 ± 0.59. Therefore, HOM and CRE samples showed the highest lightness and yellowness index (Table 4). These characteristics belong to a type of traditional homemade mayonnaise with high oil contents (Chang et al., 2017). The color difference (∆E00) is also observed in Table 4, according to the perceived scale of the color difference among samples proposed by (Yang et al., 2012). A significant color difference in VEG (control) with other samples is between VEG and HOM with an ∆E00 value of 22.34 ± 1.36, which can be due to the highest Y value in sample HOM (31±1). In contrast, a lower value for ∆E00 (9.34 ± 0.30) was determined between VEG and LIG, although notable differences were observed. Finally, the color change and color perception of each formulation can be attributed to the presence of thickeners, fat, and protein contents.

The microstructure of mayonnaise was determined by different factors, including the type and concentration of emulsifiers used to form the emulsion, the viscosity of the aqueous phase, the oil content, and the droplet size (Laca et al., 2010). Droplet size is an important parameter to predict the physical stability of emulsions (Arancibia et al., 2017), with high stability in small droplet size and significant impact on the rheological properties of the final product (McClements, 2012). Table 3 shows the droplet size results, and Figure 1 presents the optical micrographs of the different samples, where significant differences (p < 0.05) between samples were observed. For VEG, the largest droplet sizes were obtained with a homogenous monomodal droplet population of 27.3 ± 3.7 µm. Between HOM and CRE the mean sizes were 15.7 ± 3.2 and 13.8 ± 2.1 µm, respectively, with spherical shapes of the oil droplets and different populations; while the smallest size was for the LIG sample of 8.4 ± 1.7 um. In the case of the LIG sample, the droplet size was similar to a study conducted by Drozłowska et al. (2020). for a light mayonnaise with substitution of oil for flaxseed protein (5, 10, and 15%), where the droplet sizes of the analyzed samples were 9.4, 9.6 and 8.2 µm, respectively. Similar values were also reported by Worrasinchai et al. (2006) (1-9 μm) and Laca et al. (2010) (3 to 12 μm). The smaller droplet size can be attributed to two factors, a low oil concentration and the presence of a thickener. In the LIG sample, the main ingredients were water, oil, and modified corn starch (thickener). In the study by Drozłowska et al. (2020), it was shown that the droplet size decreased with decreasing oil concentration and when replacing this ingredient with thickeners (proteins or polysaccharides) because the thickeners seemed to affect the development of the system due to their high emulsifying potential. Therefore, the thickener granules kert the oil droplets separated and prevented coalescence (Di Mattia et al., 2015). The stabilization of the mayonnaise’s oil-water interface was mainly due to microparticles formed from the phosphoprotein and fused low-density lipoprotein constituents of egg yolk (Drozłowska et al., 2020).

In addition, Cornelia et al. (2015) related the size of the drops in mayonnaise made with egg protein to larger sizes than for vegan dressing-type mayonnaise due to the structure of animal origin proteins; therefore, similar results were expected in this study. However, the unexpected behavior of the VEG sample (the largest droplet size) could be attributed to the technology (type and processing conditions) used for the preparation of this dressing. Drozłowska et al. (2020) pointed out that it must be taken into account that the size of the particles and the differences in the colloidal state can be attributed to the different ingredients and preparation processes that are used in the formulation of mayonnaise. Schultz et al. (2004) also indicated that the amount of energy required to produce such fine emulsions can generally only be achieved using high-pressure systems. Therefore, the droplet size parameter also depends on technological and process factors during manufacturing. For example, the homogenization technology used for the VEG emulsion might not be sufficient to decrease the droplet size in this mayonnaise. Therefore, greater physical instability, such as droplet agglomeration, is expected in this sample. However, regardless of the droplet size, the observed aggregation droplets (CRE> VEG> HOM> LIG) (Figure 1) increased as the fat content increased (76% > 64% > 41% > 11% (w / w), respectively) in the samples (Table 1).

Figure 2 shows the upward flow curves (upper curve) and downward flow curves (lower curve) of the different samples. The rheological parameters, such as yield stress, consistency index, flow behavior index, coefficient of determination, and relative hysteresis area are detailed in Table 5. All the curves were fitted to the Herschel - Bulkley model and showed an excellent fit to the experimental data (R² > 0.995). Flow index values were lower than 1, indicating that all the samples had a pseudoplastic flow behavior, typical of mayonnaise, which suggests that the samples were in semi-solid condition with breakable networks, which is in agreement with findings reported by (Liu et al., 2007; Li et al., 2014; Drozłowska et al., 2020). The pseudoplastic properties of mayonnaise may result from the flocculation and deflocculation of oil droplets, leading to the formation of two different size aggregates, which also affects the flow of mayonnaise (Sun et al., 2018). As the shear rate grows during the flow, the deflocculation phenomenon intensifies, thus decreasing the system’s viscosity (Juszczak et al., 2003). Regarding the consistency coefficient, the highest value can be observed for the LIG sample. This behavior is explained by the presence of thickeners in its formulation, where the fat is replaced by a substitute based on carbohydrates such as modified corn starch (see Table 2). Hydrocolloid forms a gel-like structure, which traps oil droplets, slows down their movements, and increases viscosity, which is why this product can resemble the texture of traditional mayonnaise (Amin et al., 2014). This same phenomenon is related to the consistency index of the VEG sample corresponding to 10.86 ± 3.42, where this sample does not contain thickeners in its formulation (see table 2). However, this high consistency index is attributed to chickpea protein, a hydrocolloid that causes the same phenomenon as the thickeners.

The yield stress increased when the fat content increased (Peressini et al., 1998; Ma and Barbosa-Cánovas, 1995). In this study, LIG had less fat and a lower yield stress value (85.5 ± 1.1 Pa). Ma and Barbosa-Canvas (1995) showed that yield stress for mayonnaise ranged from 23 to 235 Pa, with similar data on yield stress ranging from this study (85.5-154.4 Pa). The comparison between the ascending and descending flow curves determines the thixotropic hysteresis area of mayonnaise (Štern et al., 2001). Therefore, the viscosity changes were dependent on time, and related to the structural re-destruction degree of the product due to the shear (Primacella et al., 2019). The sample with the largest area was HOM (116). At the same time, VEG (105) had the smallest area, which may be advantageous in technological processes such as pumping, transport, or storage of the sample due to greater capacity to reconstruct damaged structure after removing shear forces. Studies indicated a relationship between fat content with a greater thixotropic area (Juszczak et al., 2003). However, this relationship was not found in this study because, in the formulation of the mayonnaises analyzed, the consistency was given mainly by the thickeners used to replace high-fat content. The stress threshold was associated with a three-dimensional network rupture of product structure. The obtained values were similar to the studies by Juszczak et al. (2003) for a Polish mayonnaise. The highest threshold stresses correspond to samples VEG 154.4 ± 5.1, HOM 143.9 ± 0.7, and CRE 136.5 ± 4.1, and the lowest stress for LIG 85.5 ± 1.1, which is correlated with the lower fat content and smaller size of oil drops.

The FT-IR spectra are presented in Figure 3. The characteristic peak of the H-O-H bands of the water at 1650 cm^-1 is observed (Daoud et al., 2019), where it can be seen that in decreasing order, the peaks correspond to LIG > HOM > CRE > VEG, which is related to the water content of the samples (Table 1). In contrast, CRE and VEG have water as the second-largest component and oil as the first, which is why it showed lower peak height.

The mono-unsaturated oil characteristic peaks of the bands -C=O at 1750 cm^-1 (Rohman et al., 2011), attributed to the triacylglycerides ester bonds, can be observed in figure 3. The decreasing peak area corresponds to VEG> CRE> HOM> LIG. This relationship was supported by the nutritional information (Table 1) of the different samples of monounsaturated fats, where they contain 41% > 17% > 14% > 4% (w / w), respectively. Similar studies of quantitative analysis of fat and water content in mayonnaise using the FT-IR technique were performed by Chippie et al. (2002), and demonstrated a linear relationship between its composition and the height of the peaks.

Oxidative stability is the resistance of samples to be oxidized by the content of fats or other lipids when they come into contact with atmospheric oxygen (Miguel et al., 2019). Mayonnaise is a high-fat food (70% - 80% vegetable oil) and therefore is susceptible to oxidative deterioration through the auto-oxidation of the unsaturated and polyunsaturated fats in the oil, which depending on the extent, is likely to have a negative impact on flavor, aroma, color and nutritional value of food (Depree and Savage, 2001). Several strategies can be effective against the lipid oxidation of mayonnaise, such as the addition of antioxidants or the use of a lipid source that is naturally rich in compounds with potent antioxidant activity (Li et al., 2014; Di Mattia et al., 2015). Table 3 shows the results of the lipid oxidation induction time. These values may be influenced by the antioxidant content used in the different samples. Although all the samples contain EDTA as an antioxidant and preservative component, the labels do not specify the amount incorporated into them. However, it was possible to relate the results obtained with the content of polyunsaturated fats that are available to be oxidized due to instability and greater resonance of the double bonds exposed to oxygen (Roman et al., 2019). The LIG sample had the highest induction time of 216 ± 7 min and the lowest content of these fats with only 6% w / w, followed by HOM with a time of 201 ± 4 min with 22% w / w. Similar results were found for the VEG sample with time and content of 184 ± 2 min and 19% w / w. Finally, CRE, with 144 ± 3 min and the highest polyunsaturated fats content, with 43% w / w. Therefore, as the increasing content of polyunsaturated fats in the samples, they will oxidize faster, one factor for this type of product’s shelf-life.