*2.6. Data Analysis*

Chemical data were analyzed by means of multifactor analysis of covariance (MAN-COVA), considering the packaging material, EVOO cultivar, and storage temperature as main factors, and storage time as covariate. Two-way interaction effects were also evaluated. After checking the normal distribution of the responses, only TPC needed a squared transformation in order to fulfill the normality assumption. For significant factors, the Fisher's least significant difference (LSD) procedure was applied for mean comparisons (*p* < 0.05). Data were processed by Statgraphics Centurion software (v. 18.1; Statgraphics Technologies Inc., The Plains, VA, USA).

E-nose data were analyzed by principal component analysis (PCA), an unsupervised technique used to pre-process and reduce the dimensionality of high-dimensional datasets while preserving the original structure and relationships inherent to the original dataset. PCA reduces the number of the original variables into unobservable variables (principal components) that are linear combinations of the original ones. The main purpose of PCA is the explanation of the variability of the original dataset with as few principal components as possible, thus allowing one to visualize the data structure and the relationships between objects (score plot) and how strongly each variable influences a principal component (loading plot) [15].

PCA can be performed in covariance or correlation matrix: if the variables studied are measured using the same scale, it is reasonable to use the covariance matrix to obtain the PCs; on the other hand, if the variables are measured in different scales, the correlation matrix must be applied as the original variables are all standardized to unit variance [16]. In this work, PCA was performed in covariance matrix since the scale is the same for all the e-nose sensors.

E-nose data were elaborated by Minitab 17 (v. 1.0, Minitab Inc., State College, PA, USA) software package.

#### **3. Results and Discussion**

#### *3.1. Quality Parameters of Olive Oils Stored under Different Conditions*

Accelerated shelf-life tests were carried out in order to compare the ability of the proposed packaging materials to preserve the quality of EVOO by preventing oxidation, which causes loss of nutritional value and defects in the sensory properties [17,18].

Since the experimental factors were the packaging material, the oil cultivar, and the storage temperature, a multifactor analysis of variance was performed on quality parameters, using the storage time as covariate, as it was correlated to all the responses. Thus, the real effect of each experimental factor was assessed, after adjusting the storage time effect. Table 1 shows the results obtained, in terms of significance of the main and interaction effects. As expected, storage time significantly affected all the parameters, covariating with them. All the considered experimental factors were also significant, with the exception of storage temperature for acidity.

**Table 1.** Results of MANCOVA (F-ratio and significance level) for oil quality parameters.


PV, peroxide value; TPC, total phenolic content; n.s., not significant; \*, *p* < 0.05; \*\*, *p* < 0.01; \*\*\*, *p* < 0.001.

Actually, a similar and progressive, though limited, acidity increase was observed at both 40 and 60 ◦C (Figure 2), as a consequence of the hydrolysis of triglycerides due to the action of lipases present in olives and produced by yeasts [19]. Nera oil showed acidity values significantly higher than that of Bosana oil (*p* < 0.001; Table 1), probably caused by a different sanitary status of the olives at harvesting. Indeed, excessive free fatty acids are associated with large, fully ripened, and fungus-infected drupes obtained from trees with low fruit loads. Even a small amount of such olives can spoil the oil quality [20]. However, different polyphenol content can also affected the acidity value since a higher phenolic concentration inhibits the activity of the lipase-producer yeasts [21]. At 40 ◦C, a similar increase of free fatty acid percentage was observed for the three packaging materials, whereas at 60 ◦C the highest acidity (about 0.58% and 0.32% for Nera and Bosana EVOOs, respectively) was evidenced for the oils stored in brown-amber glass bottles (packaging A) and in the transparent plastic material (packaging B). Indeed, the packaging × storage temperature effect was significant (*p* < 0.001), evidencing lower values of acidity at 60 ◦C for packaging C (metallized material) (Table 1). Anyhow, all the collected values did not exceed the limit of 0.8% set by the European Legislation [22].

**Figure 2.** Trend of acidity in EVOOs stored in (A) brown-amber glass bottles, (B) transparent plastic material, and (C) metallized material; (**a**) Nera oil at 40 ◦C; (**b**) Nera oil at 60 ◦C; (**c**) Bosana oil at 40 ◦C; (**d**) Bosana oil at 60 ◦C. Error bars represent the standard deviation values (ranging from 0.001 to 0.024%).

Regarding the evolution of PV, after a first slight increase, a decrease was observed; thus, the legal limit of 20 meq O2kg−<sup>1</sup> [22] was never reached (Figure 3). This trend can be explained by considering that PV decreases with the appearance of secondary oxidation products. During oxidation, the hydroperoxides can form and at the same time decompose; when the decomposition rate prevails, PV is lowered even before exceeding the legal limit if the temperature is high and the oxygen concentration low [23].

All the considered experimental factors significantly affected PV (*p* < 0.001), whereas the two-way interactions were all not significant (Table 1). The lowest values were observed for oils stored in the metallized material (C), especially when stored at 60 ◦C. This means that with this packaging the oil is less protected toward oxidation since the formation of secondary oxidation products is faster. A plausible explanation for this result is the different oxygen barrier performance of the three packaging materials. Indeed, the metallized material had a permeability to gas higher than that of packaging B and A. Brown-amber glass bottles are impermeable to oxygen and packaging B had an OTR of 0.1 ± 0.02 cm3/m2 24 h, whereas OTR of packaging C was approximately double (0.23 ± 0.04 cm3/m2 24 h). Moreover, the possible contact of the oil with the metallized side of the inner layer could have catalyzed the oxidation reactions.

As expected, due to the acceleration of oxidation reactions, storage at 60 ◦C caused a significantly higher decrease of PV.

**Figure 3.** Trend of peroxide value (PV) in EVOOs stored in (A) brown-amber glass bottles, (B) transparent plastic material, and (C) metallized material; (**a**) Nera oil at 40 ◦C; (**b**) Nera oil at 60 ◦C; (**c**) Bosana oil at 40 ◦C; (**d**) Bosana oil at 60 ◦C. Error bars represent the standard deviation values (ranging from 0.01 to 1.51 meq O2kg<sup>−</sup>1).

The two EVOO cultivars showed significantly different PV (*p* < 0.001; Table 1), with the higher values in Bosana oil. For Nera oil, the initial PV was 12.9 meq O2kg<sup>−</sup>1; during storage at 40 ◦C, the maximum PV was reached at 21 days in the packaging A and 12 days in the packaging B and C; then, hydroperoxides readily decomposed to aldehydes, ketones, acids, esters, alcohols, and short-chain hydrocarbons [24], and PV gradually decreased to about 9.0 meq O2kg−<sup>1</sup> at the end of storage (Figure 3a). A similar trend was observed for Nera oil stored at 60 ◦C (Figure 3b), with lower PV at the end of storage (5.7–7.8 meq O2kg−1) due to a faster degradation of peroxides to secondary oxidation products. The same PV evolution was observed for Bosana EVOO at both 40 (Figure 3c) and 60 ◦C (Figure 3d). Starting from an initial value of 15.4 meq O2kg−1, this parameter first increased and then decreased; in particular, at 60 ◦C, the hydroperoxide decomposition was prevalent just after 5 days of storage, and the final PV was between 7.7 and 10.9 meq O2kg<sup>−</sup>1. On average, at the end of storage Nera oil showed lower PV values than Bosana oil, probably due to a different polyphenol content affecting the protection toward oxidation phenomena.

K270 is known to be a good marker of oxidation secondary stage because it is related to conjugated trienes and carbonyl compounds [25]. In a recent work, Conte et al. [23] found that K270 is the best index for allowing one to predict EVOO shelf life when an accelerated test is applied. For both EVOO cultivars, this parameter significantly (*p* < 0.001) increased during storage, with significantly higher values in Bosana oil at 60 ◦C (Table 1; Figure 4), as a logical consequence of the higher initial oxidation state of Bosana samples and the acceleration of chemical reactions at higher temperatures. The packaging material had a significant effect on K270 (*p* < 0.001); the oils stored in the two innovative packaging materials (B and C) showed lower values than the oils stored in the brown-amber glass. In particular, the significantly lowest values were observed for the oils packaged in the metal-

lized material stored at 60 ◦C; indeed, the interaction packaging × storage temperature had significant results (*p* < 0.01).

**Figure 4.** Trend of the spectrophotometric index K270 in EVOOs stored in (A) brown-amber glass bottles, (B) transparent plastic material, and (C) metallized material; (**a**) Nera oil at 40 ◦C; (**b**) Nera oil at 60 ◦C; (**c**) Bosana oil at 40 ◦C; (**d**) Bosana oil at 60 ◦C. Error bars represent the standard deviation values (ranging from 0.001 to 0.068).

For Nera oil stored in the brown-amber glass bottles (packaging A), a significant increase of K270 was observed, and the legal limit of 0.22 [22] was exceeded after 12 days at 40 ◦C and 11 days at 60 ◦C. For packaging B and C, the extinction at 270 nm increased considerably only at the end of storage at 40 ◦C, whereas at 60 ◦C, the legal limit was exceeded after 11 days for packaging B and 15 days for packaging C (Figure 4a,b). Similar results were obtained for Bosana oil (Figure 4c,d), with the samples packaged in brownamber glass bottles characterized by a higher index of secondary oxidation products and exceeding the legal limit for K270 after 21 and 5 days of storage at 40 and 60 ◦C, respectively. This result could seem in contrast with the well-known protection ability of glass toward oil oxidation phenomena. However, in this case, the higher K270 values could be related to the higher retention capacity of glass toward low-molecular-weight compounds with respect to the tested innovative packaging materials.

Phenolic compounds are naturally present in olive oils, and they are responsible for oil stability during storage [26]. Figure 5 shows the evolution of TPC in Nera and Bosana oil as affected by packaging material and storage temperature.

Nera and Bosana oils were characterized by significantly different values of TPC, and the cultivar was indeed a significant factor (*p* < 0.001; Table 1) affecting this quality parameter during storage. Nera cultivar had a medium/low content of phenolics (300 mgGAE kg−1), whereas Bosana was characterized by high polyphenol concentration (558 mgGAE kg<sup>−</sup>1). During storage at both temperatures, a progressive decrease in TPC was observed (Figure 5), with a significantly (*p* < 0.001; Table 1) different trend for the tested

packaging materials. In particular, the metallized material (packaging C) had a detrimental effect on phenolics, causing a higher TPC reduction, especially at 60 ◦C. In fact, the interaction packaging material x storage temperature was significant (*p* < 0.001). The average TPC reduction for the two EVOOs in packaging C at the end of storage was about 52% at 40 ◦C and 64% at 60 ◦C. The best-performing material in protecting oil phenolic content was the transparent plastic packaging (B), followed by the brown-amber glass (packaging A); this result can be again ascribed to the better oxygen barrier performance of the transparent pouches. During storage at 40 and 60 ◦C, TPC reduction was on average 21% and 30% for packaging B and 29% and 40% for packaging A. These results are in agreement with previous works showing that, during storage, phenolic compounds undergo quantitative modification due to oxidation and the temperature as well as the packaging material can have a notable influence on phenolic degradation [8,10].

**Figure 5.** Trend of total phenolic content (TPC) in EVOOs stored in (A) brown-amber glass bottles, (B) transparent plastic material, and (C) metallized material; (**a**) Nera oil at 40 ◦C; (**b**) Nera oil at 60 ◦C; (**c**) Bosana oil at 40 ◦C; (**d**) Bosana oil at 60 ◦C. Error bars represent the standard deviation values (ranging from 0.1 to 47.6 mgGAE kg<sup>−</sup>1).
