*3.3. Principal Component Analysis*

PCA was used to observe any possible groups within the commercial and prepared mayonnaises as well as to reveal the interrelationships among variables (QUENCHER-DPPH, QUENCHER-ABTS, PV, AnV, TOTOX, AV, area of ODs, L\*, a\* and b\*) that mainly influence the similarities and differences of the analyzed samples. The first two principal components took into account 85.86% (PC1 = 57.80% and PC2 = 28.06%, respectively) of the total variation.

The entire data set was visualized by a bi-plot (combined scores and loadings plot for two components) and presented in Figure 4.

As can be seen, the QUENCHER-DPPH (0.8486), QUENCHER-ABTS (0.8497), AV (0.0010), L\* (0.8625), and a\* (0.0146) of mayonnaise samples had positive loadings on the PC1, whereas the PV (−0.9169), AnV (−0.9388), TOTOX indexes (−0.9656), ODs size (−0.9123), and b\* (−0.3271) were the variables with negative loadings on PC1. However, all variables (except AnV) revealed loadings on the negative dimension of PC2 (−0.0045–−0.9819). It can be noted that QUENCHER-DPPH, QUENCHER-ABTS, AV, L\* and a\* were the variables with positive loadings on PC1 and negative loadings on PC2. On the contrary, AnV revealed loadings on the negative dimension of PC1 and positive on PC2. However, b\*, PV, TOTOX and ODs were the features with negative loadings on PC1 and PC2.

The PCA graph depicted that the two vegan mayonnaises (the commercial MV and MLO sample made from a blend of RRO and CPLO) with the lowest RSA (QUENCHER-DPPH = 589–626 μmol TE/100 g, QUENCHER-ABTS = 1371–1674 μmol TE/100 g) and L\* (41.4–41.5) values and the highest oxidative parameters (PV = 2.27–4.61 mEq O2/kg, AnV = 2.95–3.07, TOTOX = 7.49–12.29) and area of ODs (56.76–101.40 μm2) were located to the left in the score bi-plot and had negative values for PC1. Three prepared vegan emulsions (MSO, MCO and MRO) and two commercial egg yolk mayonnaises (MT1 and MT2) with high radical scavenging properties (QUENCHER-DPPH = 643–828 μmol TE/100 g, QUENCHER-ABTS = 1836–3320 μmol TE/100 g) and more lightness (L\* = 44.5–48.7) were situated at the right in the score bi-plot and had positive values for PC1. Consequently, the studied samples could be divided into three groups based on their distribution on the PCA graph. The yellowest commercial plant-based mayonnaise (MV) with the longest distance from other samples had the highest of all the oxidative parameters (PV, AnV, TOTOX, AV), ODs area and the lowest radical scavenging properties determined by two analytical methods (QUENCHER-DPPH and QUENCHER-ABTS). Two traditional mayonnaises containing egg yolk (MT1 and MT2) characterized by the highest RSA (QUENCHER-DPPH = 692–828 μmol TE/100 g, QUENCHER-ABTS = 2808–3320 μmol TE/100 g), the lowest ODs (4.44–4.47 μm2)

and with the same AV results (0.49 mg KOH/g) created a distinct cluster. Additionally, four egg-free mayonnaises made from blends of RRO and cold-pressed vegetable oils with moderate radical scavenging properties (QUENCHER-DPPH = 626–690 μmol TE/100 g, QUENCHER-ABTS = 1674–2121 μmol TE/100 g), oxidative status (PV = 1.19–2.27 mEq O2/kg, AnV = 1.12–2.95, TOTOX = 3.93–7.49), and ODs areas (14.29–56.75 μm2) as well as the lowest AV (0.11–0.26 mg KOH/g) and a\* (−0.3–0.9) values were separated from the other studied samples and located at the upper the A1 axis.

**Figure 4.** Biplot of scores and loadings of data obtained from radical scavenging properties (QUENCHER-DPPH, QUENCHER-ABTS), oxidation stability (PV—peroxide value; AnV—anisidine value; TOTOX—total oxidation index; AV—acid value), areas of oil droplets (ODs) and color parameters (L\*—lightness; a\*—redness; b\*—yellowness) of aquafaba-based mayonnaise samples with blends of refined rapeseed oil and cold-pressed rapeseed oil (MRO), cold-pressed sunflower oil (MSO), cold-pressed linseed oil (MLO), and cold-pressed camelina oil (MCO); MT1—commercial egg yolk mayonnaise from producer 1; MT2—commercial egg yolk mayonnaise from producer 2; MV—commercial vegan mayonnaise.

It is noteworthy that there was a high correlation between the QUENCHER-DPPH and QUENCHER-ABTS results for all studied mayonnaises (r = 0.9427). This suggests that antioxidants present in the studied samples have the ability to scavenge both stable DPPH radical and ABTS cation radical.

Moreover, RSA values of the discussed mayonnaises were positively correlated with their AV data (r = 0.3177 and 0.4319) as well as color parameters, such as lightness (L\*, r = 0.6238 and 0.6894) and redness (a\*, r = 0.2615 and 0.4544), but negatively associated with oxidative parameters (PV, AnV and TOTOX, r = −0.6453–−0.7549) and ODs sizes (r = −0.6432–−0.7315). These high negative correlations between oxidative parameters and radical scavenging properties of the studied emulsions indicated the significant contribution of the antioxidant potential of samples to enhance their oxidation status. The lower PV values can be explained by the hindered oxidation of the lipid fraction in the mayonnaise samples, and thus the delay in their deterioration, due to the presence antioxidants and higher RSA values.

As expected, the PV results for seven mayonnaises contributed significantly and positively with the AnV (0.7800) and TOTOX indexes (0.9843). Additionally, a high correlation coefficient (r = 0.8781) was found between AnV and TOTOX data for these samples.

Interestingly, there were high positive correlations between the amounts of primary and secondary oxidation products in the analyzed samples and ODs (r = 0.8162–0.9117). These positive correlations indicate that emulsions with smaller areas of ODs had a better oxidative stability. Nevertheless, the area of ODs for mayonnaises was negatively associated with their lightness (L\*, r = −0.7916). However, negative correlation coefficients (r = −0.7470– −0.8583) for the relationships between PV, AnV, TOTOX and L\* were observed.

In contrast, there were positive correlations between the free fatty acids content and color parameters of the investigated samples (AV–a\*, r = 0.8540 and a\*–b\*, r = 0.7187).

The results obtained by PCA indicated the differences in radical scavenging and physicochemical properties of two yolk egg mayonnaises, one commercial vegan mayonnaise, and four aquafaba-based samples made from blends of RRO and cold-pressed vegetable oils. This grouping of the samples suggests that the used ingredients in mayonnaise samples and technological conditions were responsible for their quality.
