3.1. Evolution of Dried Tomatoes Preserved in Oil Quality Parameters during Shelf-Life
The acidity of EVOOs was used as a reference parameter for the overall evaluation of the product quality. The highest quality EVOOs must feature a free acidity lower than 0.8%. In the first month, the samples maintain constant acidity values (0.68%), while after the fourth month, a significant increase was observed. In particular, at 12 months, values of 0.98%, 0.97%, and 0.94% were found, for EVOO, and EVOO_DTL, and EVOO_DTD, respectively (
Table 2). It is clear that the presence of the tomatoes slices into EVOO and storage in the dark helped protect oil from degradation.
Oxidation causes the product to become rancid, forming unpleasant odors and flavors. According to current regulations, the maximum number of peroxides is fixed at 20 milliequivalents of oxygen per kg of oil (meq O2/kg). Reduced levels of peroxides are associated with an increase in the shelf-life of the product and a delay of rancidity. Peroxide formation is related to the action of oxygen, high temperatures, light, and the catalytic action of some enzymes capable of binding oxygen to the fatty acids.
A large number of peroxides indicate the start of an irreversible oxidative process. After 12 months, an increase in the levels of peroxides was observed, with values of 22.54 and 26.24 meq O
2/kg in EVOO and EVOO_DTL, respectively (
Table 2). Less variation from 15.55 to 20.71 meqO
2/kg was observed for EVOO_DTD, suggesting that the presence of tomatoes and storage in the dark reduced degradation and improved the shelf life of the products. All EVOO samples exhibited initial ΔK values of 0.01. This parameter was maintained (0.17 and 0.16, respectively) with a minimal variation in EVOO_DTL and EVOO_DTD.
Concerning dried tomatoes, water activity (aw) and moisture content were also considered (
Table S1). Water activity expresses the relationship between the vapor pressure of the water in the food and the vapor pressure of pure water. It is a dimensionless value that indicates the amount of water contained in the product, free from bonds with other components, and capable of giving chemical and biological reactions, and available for food enzyme activity and microbial growth [
31].
When the aw value increases, the shelf life of the product decreases. In general, foods present aw levels in the range of 0.2 (very dry foods) to 0.99 (moist fresh foods). Micro-organisms maintained their viability regardless of the aw, but to grow bacteria requires an aw > 0.8, whereas yeasts and moulds require an aw > 0.6 [
31]. Obtained data relating to moisture content and aw showed that when stored in the light, moisture content varies from 31.66 to 40.4%, and aw from 0.63 to 0.71. However, when the product was stored in the dark, a less significant increase was observed. After 12 months, a moisture content of 37.63% and an aw of 0.69 were found (
Table S1). These results demonstrated that the absence of light is a fundamental parameter to extend the shelf life of the product.
Recently, Owureku-Asare et al. [
31] compared the physicochemical characteristics of cabinet dryer (called solar cabinet dryer) and sun-dried tomato powder. Results showed that the moisture content of the solar dried tomato (13.94–14.57%) was lower than the sundried tomato (19.38–21.63%). Additionally, the same authors reported aw values lower for solar cabinet dried tomato powder (0.35–0.38) compared to sundried tomato powder (0.53–0.57).
During storage, EVOO loses its quality (expressed as free acidity and peroxide content) with respect to samples containing dry tomatoes (EVOO_DTL and EVOO_DTD). Particularly EVOO_DTD showed a lesser increase in these two parameters.
3.2. Evolution of Phytochemicals in Dried Tomatoes Preserved in Oil
Carotenoids and chlorophylls, extracted from the drupe during the transformation process, are natural pigments that influence the EVOO color. For this reason, we decided to monitor the chlorophyll content, total carotenoids content (TCC), and β-carotene and lycopene contents of investigated samples. The CIELab parameters were also investigated.
The most significant chlorophyll content change was observed in EVOO_DTL with variation from 66.17 to 23.10 mg/Kg, followed by EVOO with a chlorophyll content ranging from 66.17 to 28.65 mg/Kg (
Table S2). A similar situation was observed in EVOO_DTD where a reduction of the content in this pigment was verified, and the storage conditions in the dark better preserved the matrix from degradation. In fact, chlorophyll content decreased by −38%, −43%, and −37%, respectively for EVOO, EVOO_DTL, and EVOO_DTD.
A significant reduction in TCC was observed in EVOO with values in the range from 22.45 to 11.78 mg/Kg, while the TCC level remained quite constant in EVOO_DTL (22.45–21.35 mg/Kg) (
Table S2). After 12 months, a reduction in the carotenoid content of 10.7 and 1.1% was observed for EVOO and EVOO_DTL, respectively. On the contrary, there is a notable increase (+2.5%) in EVOO_DTD due to the contribution provided by the dried preserved tomato. Indeed, the carotenoid content in DTD decreased from values of 24.57 to 19.96 mg/Kg. A significant reduction of −6% in the TCC was observed in DTL.
The lycopene content of EVOO_DTL and EVOO_DTD found at month 0 was probably caused by the applying of manual pressure to stack the dried tomatoes inside the jar containing the oil used for preservation. This pressure may facilitate the extraction of lycopene by the EVOO (solvent) as a consequence of the maceration process, taking into account that analysis was carried out 72 h after jar preparation [
32]. Analysis of oils (EVOO_DTL and EVOO_DTD) evidenced that in the first 6 months of storage, a decrease in lycopene content was observed. Successively, this content increased due to the passage of this pigment from the dried pressed tomatoes to the EVOO. After 12 months, an increase in the lycopene content of 0.70% and 0.20% was observed, for EVOO_DTD (0.91–1.59 mg/100 g) and EVOO_DTD (0.91–1.10 mg/100 g), respectively. On the contrary, the lycopene content in DTL and DTD decreased by 0.31% and 0.49%, respectively, with a content variation from 0.98 to 0.67 mg/100 g and from 0.98 to 0.4 mg/100 g, respectively. Moreover, we have to consider that tomato’s lycopene content could be influenced by several factors, such as the tomato genotype, plant nutrition, the environment, and the period of collection [
33]. Previously, Karakaya and Yılmaz [
34] investigated the lycopene content in fresh and processed tomatoes and found values of 1.74, 5.51, and 3.55 mg/100 g in fresh, sun-dried, and canned tomatoes, respectively. The lycopene content of fresh “Roma” tomatoes found by Owureku-Asare et al. [
31] was 1.49 mg/100 g. De Abreu et al. [
35] reported a lycopene content in the range of 19.97–33.95 mg/100 g dried material (DM) in eight preserved (in oil) dried tomatoes. These results are agreement with Toor and Savage [
36]. Similar lycopene values were found in cherry tomatoes dehydrated at different temperatures [
37]. Values in the range of 23.05 to 29.77 mg/100 g DM were observed with the highest value in tomatoes subjected to dehydration at a low temperature (40 °C). Additionally, Mwende et al. [
38] suggested that lycopene content is correlated to the tomato variety. These authors studied four varieties of fresh tomatoes, grown under the same conditions, and found lycopene contents of 174.86, 108.46, 135.80, and 198.25 mg/100 g DW for “Anna F1”, “Kilele”, “Prostar F1”, and “Riogrande”, respectively. Different factors, such as oxygen, high temperatures, and light, reduced the TCC in this food matrix [
39].
A great variability in the β-carotene content was found (
Table 3). In EVOO, the content decreased from 90.26 to 38.61 mg/100 g of oil, while in EVOO_DTL and EVOO_DTD, less significant reduction (−17% and −4%) occurred after 12 months due to the protective action exerted by tomatoes. Indeed, a reduction of 5% in β-carotene content was found in both DTD and DTL (
Table 4). De Abreu et al. [
35] reported a β-carotene content in the range of 210–447 mg/100 g DM found in eight preserved (in oil) dried tomatoes.
The variation in pigment content also influenced CIELab parameters (
Table 5). After 12 months of storage, the L* parameter was highest in the EVOO samples followed by EVOO_DTL which could be due to auto-oxidation and subsequently to a greater degradation of chlorophylls in these samples. On the contrary, the C* value was highest in EVOO_DTD and EVOO_DTL due to a red color conferred by the lycopene contained in dried tomatoes (
Table 5). No significant differences were recorded in Hue (H*) values.
Interesting Pearson’s correlations were found. In EVOO, C* positively correlated with chlorophylls, lycopene, β-carotene, and TCC (r = 0.95, 0.85, 0.95, and 0.90). In EVOO_DTL and EVOO_DTS, C* positively correlated only with TCC (r = 0.71 and 0.55) and lycopene content (r = 1.00).
The overall change of color during 12 months of storage was assessed by the ∆E*ab calculation (
Figure 1). This color parameter was used to characterize the variation of colors in foods during processing. Data evidenced significant differences between ∆E*ab values in all samples (
p < 0.05) (
Table S3). From eight months of storage, the ∆E*ab parameter for EVOO_DTD increased to reach 15.03 after 12 months storage.
EVOO phenolic compounds are responsible for the pungent taste as well as for the resistance to the oxidative process that reduces EVOO quality. Significant differences were observed in EVOO, EVOO_DTL, and DTD samples regarding TPC and TFC samples. EVOO_DTD showed values ranging from 110.80 to 142.24 mg CAE/g for TPC, and from 56.40 to 64.89 mg QE/g for TFC (
Table 6). TPC values in the range of 110.80–90.21 mg CAE/g and TFC values in the range of 56.40 and 25.19 μg QE/g were observed in EVOO_DTL. Less content was found in EVOO ranging from 110.80 to 77.13 mg CAE/g for TPC and 56.40 to 22.31 mg QE/g for TFC.
Conversely, the enrichment of EVOO_DTD in TPC and TFC (31% and 8%, respectively) was related to the reduction of these phytochemicals in DTD (−13% and −20% for TPC and TFC, respectively) (
Table 7). In comparison to EVOO, EVOO_DTL showed an initial increase in TPC and TFC (+7 and +2%, respectively). This enrichment is surely due to the release of these phytochemicals by the dried preserved tomatoes. Subsequently, a progressive reduction in these components was recorded in the tomato matrix.
De Abreu et al. [
35] reported the TPC in the range from 337.8 to 835.7 mg GAE/100 g DM in eight preserved (in oil) dried tomatoes, whereas lower values were found by Toor and Savage [
36] in semi-dehydrated tomatoes with a TPC ranging from 44.6 to 74.3 mg/100 g. This observation can be justified because the dried tomatoes are subjected to thermal treatment, which favors the liberation of phenolic compounds and lycopene from the cellular matrix. For this reason, lycopene and phenolic compounds are more bioavailable in dried tomatoes compared to semi-dehydrated with a consequent improvement of its functional and nutritional properties [
33].
The determination of the presence of the microorganism in food above a level is considered an indication that food is produced and could be consumed under safe conditions. For these reasons, at the end of storage, all samples were subjected to microbiological investigation of mesophilic aerobic bacteria, yeast and moulds. All samples presented a lower bacterial population <102 CFU\g as well as the growth of yeast and moulds at undetectable levels (<101) (data not shown).
3.4. Antioxidant Activity
The determination of antioxidant capacity is an essential parameter for the shelf-life evaluation of foods. Several antioxidants may act in vivo through different mechanisms, and consequently no single test can fully define the antioxidant potential of a matrix.
For this reason, the goal of this work was to screen the antioxidant ability of samples using three methods: 2,2-azinobis (3-ethylbenzothiazoline-6-sulfonic acid) (ABTS), 1,1-diphenyl-2-picrylhydrazyl (DPPH), and β-carotene bleaching tests. This scientific evidence justifies the different results obtained in relation to the applied test. The radical scavenging activity of EVOOs and dried preserved tomatoes slices was examined using the DPPH radical and ABTS radical cation that possess a diverse stereochemistry and different training mechanisms.
The potential of samples to inhibit lipid peroxidation was assessed using the β-carotene bleaching test. A concentration-dependent activity was observed for all samples (
Table 9). Interestingly, the antioxidant potential gradually increases in both EVOO_DTL and EVOO_DTD during storage compared to EVOO. This evidence may be explained by the enrichment of the oil with bioactive compounds from tomatoes. During the period of observation (12 months), a reduction in EVOO radical scavenging activity was observed for DPPH and ABTS tests with IC
50 values from 90.17 to 110.69 μg/mL and from 85.43 to 98.46 μg/mL, respectively. A similar consideration was made for β-carotene bleaching tests, in which a reduction in lipid peroxidation protection was found with IC
50 variations from 105.81 to 112.89 μg/mL and from 110.25 to 119.83 μg/mL, after 30 and 60 min of incubation, respectively.
Conversely, in EVOO_DTD, the antioxidant potential calculated as the percentage on IC50 value data increases by +43, +47, +48, and +38%, for the DPPH, ABTS, and β-carotene bleaching test after 30 and 60 min of incubation, respectively in comparison to EVOO alone. For EVOO_DTL, an increase of +30, 34, +32, +35% was found. Additionally, EVOO_DTD presented an antioxidant activity 2.4, 2.6, and 1.95 times higher in the DPPH, ABTS, and β-carotene bleaching test (30 min) in comparison to EVOO, respectively. Similarly, EVOO_DTD was 1.3, 1.4, and 1.3 times more active compared to EVOO_DTL in the DPPH, ABTS tests, and β-carotene bleaching test, respectively. These results demonstrated that tomatoes preserved and increased the antioxidant capacity of the EVOO_DTL and EVOO_DTD samples.
Interesting correlations were observed using Pearson’s statistical analysis. From data analysis, r values of 0.98, 0.97, 0.90 and 0.92 were found between the oleic acid (OA)/linoleic acid (LA) ratio and the DPPH, ABTS tests, and β-carotene bleaching test after 30 and 60 min, respectively.
Values of r of 0.97, 0.97, 0.90, and 0.93 were found between the MUFA/PUFA ratio and the DPPH, ABTS test, and β-carotene bleaching test after both incubation times. In EVOO_DTL, r values > 0.90 were observed when the MUFA or PUFA content was correlated with the DPPH, ABTS tests, and β-carotene bleaching test after 30 and 60 min.
In addition, the Pearson’s correlation coefficient was positive between TFC and the DPPH, ABTS tests, and β-carotene bleaching test after 30 and 60 min, with r = 0.78, 0.81, 0.74, and 0.81, respectively. A similar positive correlation between the MUFA and PUFA content and the antioxidant assay was observed in EVOO_DTD.
As expected, dried tomato samples showed a reduction in antioxidant activity with a particular decrease in DTD (
Table 10). Already from the first month, there was a reduction of 10%, 8%, 4%, and 17% for DTL up to a 23%, 22%, 17%, and 30% decrease in the twelfth month, respectively, in the DPPH, ABTS, and β-carotene bleaching tests after 30 and 60 min of incubation. Indeed, IC
50 variations in the range 35.23–58.44, 23.84–46.23, 55.68–72.57, and 58.37–88.54 μg/mL were found for the DPPH, ABTS tests, and β-carotene bleaching test after 30 and 60 min of incubation, respectively.
A more significant decrease was observed in both DTD and DTL samples but with main impact on DTD with IC50 values of 58.44, 46.23, 72.57, and 88.51 μg/mL (after 12 months) and 35.23, 23.84, 55.68, and 58.37 μg/mL (at 0 months storage) in the DPPH, ABTS, and β-carotene bleaching tests after 30 and 60 min of incubation, respectively. The radical scavenging activity decreased by −13% in both tests, while a reduction in lipid peroxidation protection of −16% and −24% after 30 and 60 min incubation, respectively, was observed. Interestingly, at the beginning of the observation period, both dried tomatoes had the greatest antioxidant activity compared to all EVOOs. After 12 months of storage, EVOO_DTD presented the greatest activity. Except for the radical scavenging activity, EVOO_DTL displayed an important protection of lipid peroxidation. In contrast, EVOO exhibited less antioxidant activity in all assays.
The RACI values of EVOOs and dried “San Marzano” tomato samples were calculated separately as the mean of standard scores transformed from the raw data generated with diverse antioxidant methods. Therefore, RACI was used to extrapolate samples with the highest antioxidant potential. Based on RACI data, the following antioxidant rank order was found: EVOO_DTD > EVOO_DTL > EVOO in each month (
Figure S1a,b).
Concerning dried tomatoes, DTD presented the highest antioxidant potential in each month of observation.
Just few studies investigated the antioxidant activity of dried tomatoes. Arslan et al. [
40] investigated the effect of different drying processes (sun, oven, and microwave oven) on the antioxidant activity of tomato slices. A significant variability was observed in samples treated with different drying conditions. In particular, microwave oven drying at 700 W resulted in the most active effect independently by the applied antioxidant test. The application of osmotic pre-treatment with 10% saline solution to tomato slices before the oven drying process at 105 °C, for different times of exposure from 60 to 300 min, resulted in an efficient methodology to preserve bioactive compounds with DPPH radical scavenging potential [
41]. More recently, Al Maiman et al. [
42] demonstrated that storage of dried tomato slices resulted in the increase in antioxidant activity measured as radical scavenging potential (2.71 mg trolox/g DM for fresh tomato slice vs. 3.55 mg trolox/g DM for fresh tomato slice). Additionally, a similar finding was reported by Martinez-Valverde et al.’s [
43] method. De Abreu et al. [
35] analysed the antioxidant activity of hydrophilic and hydrophobic extracts from eight preserved (in oil) dried tomatoes. The radical scavenging activity was investigated using the DPPH test at a concentration 1.25 mg/mL, finding that the hydrophilic extracts had a greater capacity to sequester the DPPH radical than the hydrophobic ones (ranging 16–39.15% vs. 13.13–15.56%). On the contrary, in the case of the β-carotene/linoleic acid system, the hydrophobic extracts (from 72.5 to 86.65%) had a higher antioxidant capacity than the hydrophilic ones (from 72.5 to 86.65% vs. <22 to >60%). Furthermore, it should be considered that the total antioxidant activity of tomatoes is the result of the synergistic effect of the different bioactive compounds contained in the tomato matrix (carotenoids, phenols, and vitamins).
3.5. Principal Component Analysis
Data were analysed by means of PCA to determine the systematic variation and underlying relationships between bioactive compounds in the samples and antioxidant properties. All the determinations previously described were used to develop the PCA model. The first two PCs explained 91.3% of the variance in the data (PC1 = 60.37 and PC2 = 30.93), which was high enough to represent all the variables. The score plot for PC1 versus PC2 clearly distinguished three groups defined by length of storage, indicating that storage time had a major influence on the quality.
Figure 2a shows the PCA data concerning the content of active compounds in the samples and their antioxidant activity. The first component mainly correlated with the chlorophylls-EVOO, carotenoids-EVOO, β-carotene-EVO, chlorophylls-EVOO-DTL, β-carotene-DTL, chlorophylls-EVOO-DTD, lycopene-DTL, β-carotene-DTL, carotenoids-DTL, lycopene-DTD, β-carotene-DTD, carotenoids-DTD, and TPC-EVOO. It was negatively correlated with the acidity-EVOO, NP-EVOO, ΔK-EVOO, acidity-EVOO-DTL, NP-EVOO-DTL, ΔK-EVOO-DTL, acidity-EVOO-DTD, NP-EVOO-DTD, and ΔK-EVOO-DTD.
The second principal component correlated with the acidity-EVOO, NP-EVOO, ΔK-EVOO, acidity-EVOO-DTL, NP-EVOO-DTL, ΔK-EVOO-DTL, acidity-EVOO-DTD, NP-EVOO-DTD, ΔK-EVOO-DTD, acidity-EVOO-DTD, NP-EVOO-DTD, ΔK-E-DTD, carotenoids-EVOO-DTD, and lycopene-EVOO-DTD, while it negatively correlated with chlorophylls-EVOO, carotenoids-EVOO, β-carotene-EVOO, chlorophylls-EVOO-DTL, lycopene-DTL, β-carotene-DTL, carotenoids-DTL, lycopene-DTD, β-carotene-DTD, carotenoids-DTD, and TPC-EVOO. Two major groups based on the storage time were observed (
Figure 2a).
Samples, which lie close to each other, are similar, while those away from the origin are extreme samples. Apart from the regular groupings of all storage, in intervals there were certain samples lying outside of these clusters (T2, T3, T4, T5, T6, and T7); moreover, they were negatively correlated to PC1. The control samples (T0) and (T1) exhibited a completely different behavior, showing extreme values and not a part of any cluster. The samples DTL and EVOO-DTL with a high content of carotenoids and β-carotene belonging to the first and second sampling (T0 and T1) were mostly located above the x axis. Principal component analysis of the fatty acids percentage gives two linear combinations which explain, overall, 90.64% of the variance, in particular 67.78% for the first component and 22.86% for the second.
Figure 2b shows the vectors of each variable and the distribution of the oil samples in the plane defined by the values of the two principal components. The first component was mainly correlated with C16:0, C16:1, C17:1, C18:0, C18:1, C18:2, C20:0, C18:3, C20:1, OA/LA, SFA, and MUFA for EVO, EVOO-DTL, and EVOO-DTD, while it correlates negatively with MUFA/PUFA for EVOO, EVOO-DTL, and EVOO-DTD. The second component was correlated positively with C14:0 and C17:0 for EVOO, EVOO-DTL, and EVOO_DTD. Three groups can be observed based on the similarities between the samples (T0-T1, T2-T10, and T11-T12). In each group, as shown in the
Figure 2b, the samples are very similar to each other.