Phenolic compounds were extracted according to Mc Donald et al. (2001). Freeze dried olive pulps (5 g) were homogenized in 25 mL of methanol/water (80:20, v/v). The residue was extracted twice; extracts were combined, and washed with hexane. The extracts were filtered, and then kept cold until analysis.

The total phenolic content of the extracts was determined with Folin Ciocalteu reagent according to Borzello et al. (2000). Total phenol values were expressed as mg gallic acid equivalents per 100 g Dry Weight (mg GAEq·100 g⁻¹ DW).

The presence and amount of phenolic compounds in the olive extracts were studied by reversed phase HPLC analysis using a binary gradient elution. The analysis was performed by reversed phase HPLC on a LC-10ADVP Shimadzu (Milan, Italy) liquid chromatography equipped with an SPD M10AVP diode array detector (Shimadzu). The chromatographic separation was achieved on a Spherisorb S5 ODS-3 (250 mm×4.6 mm i.d.) reversed-phase column (Phenomenex, Macclesfield, UK). The solvent system used was a gradient of solvent A (water: trifluoroacetic acid, 97:3, v/v), and solvent B (acetonitrile: methanol, 80:20, v/v). A step gradient from 5% to 98% B (45 min) was applied at a flow rate of 1 mL·min⁻¹. Peak quantification was carried out at 279 nm. The main phenolic compounds were identified by comparison with relative retention times of pure compounds, when available, or by comparing the relative elution order and UV spectra with those reported in the literature (Brenes et al., 2000; Rovellini and Cortesi, 2002). The identity of each peak was confirmed by LC-MS, performed on an LC-10AD VP Shimadzu (Milan, Italy) liquid chromatograph on-line with an LCMS-2010EV Shimadzu (Milan, Italy) mass spectrometer, equipped with an electrospray ionization (ESI) interface. A Discovery HS C18 column (5 µm, 150 mm×2.1 mm i.d., Supelco, St. Louis, MO, USA), at a flow rate of 0.35 mL·min⁻¹, was used. The solvent system used was a gradient of solvent A (water+formic acid 0.25%), and solvent B (methanol+formic acid 0.25%), with a step gradient from 5% to 55% B (45 min). The ESI mass spectra (m/z 60–900) in the negative ion mode were obtained under the following conditions: interface voltage 4 kV; nebulizer gas flow 1.5 L·min⁻¹; block heater temperature 250 °C; curved desolvation line temperature and voltage of 300 °C and −5 V, respectively; Q-Array voltage 0 V DC and 150 V RF; detector voltage 1.5 kV. Some operating parameters (interface voltage and Q-Array voltage) were then modified in order to obtain a moderate fragmentation of the de-protonated molecular ions: an interface voltage of 5 kV and a Q-Array voltage of −50 V DC and 150 V RF were used (Savarese et al., 2007).

The Ferric Reducing Power of the extracts was measured as reported by Zhan et al. (2006) using ferric chloride. The absorbance was then measured at 700 nm and the reducing power was expressed as mg quercetin, butylated hydroxyanisole and butylated hydroxytoluene equivalents per 100 g Dry Weight (mg QE, BHAEq and BHTEq·100 g⁻¹ DW).

The procedure reported by Boskou et al., (2006) was adapted. An aliquot of the appropriate dilution of the extract (0.5 mL) was added to a 1,1-Diphenyl-2-picryl-hydrazyl (DPPH) solution (2 mL) and kept in the dark for 30 min. The absorbance was measured at 515 nm and the antiradical activity was expressed as mg quercetin and trolox equivalents per 100 g Dry Weight (mgQEq andTEq·100 g⁻¹ DW).

The hydrogen peroxide radical scavenging activity was determined according to Hemalatha et al. (2013), the absorbance was measured at 230 nm after 10 min of incubation at 37 °C. The percentage of hydrogen peroxide scavenging is calculated as follows:

Where: A_c is the absorbance of the control and A_T is the absorbance of the test.

The effect of extracts on the β-carotene- linoleic acid emulsion was determined by applying the method reported by Nsimba et al. (2008). The absorbance of the tested samples was repeatedly measured every15 min at 470 nm. The total antioxidant activity was calculated based on the following equation: % A A = [ 1 - ( A 0 - A t A 00 - A 0 t ) ] * 100

Where: AA% is antioxidant activity, A₀ and A_t; emulsion absorbance at t=0 and after incubation time t (t=105 min), A₀₀ and A_0t: absorbance values for negative control at t=0 and after the same incubation time.

Phenolic contents (Table 1) differ significantly (p<0.05) among the investigated olive cultivars, and between fresh and processed olives for the same cultivar. Their content are higher in fresh olives (346 mg·100 g⁻¹ in Bouchouk to 2406 mg·100 g⁻¹ in Azzeradj from Seddouk), than in processed ones (124 mg·100 g⁻¹ in Azzeradj from Seddouk to 1688 mg·100 g⁻¹ in Sigoise).

The results obtained for fresh olives are similar to those obtained for Portuguese cultivars (Pereira et al. 2006), Tunisian cultivars (Ben Othman et al. 2009), and Italian cultivars Piscopo et al. (2014). However, processed olives contained higher amounts of phenolic compounds than those studied by Blekas et al. (2002) and similar contents to those reported by Ben Othman et al. (2009).

A drastic decrease in phenolic content was noted after processing. Losses in those compounds differ largely among the cultivars, from 12.25% (Sigoise) to 94.80% (Azzeradj from Seddouk). This trend confirms that processing influence phenolic behavior differently between cultivars according to their phenolic profile. In fact, the diffusion of phenols from olive pulp to brine depends on cultivar characteristics, fruit skin, permeability, type of phenols present in the olives and their ability to diffuse outside the fruit (Kiai and Hafidi, 2014).

The HPLC analysis of phenolic compounds (Table 1) showed that olive flesh revealed different phenolic compound compositions for the seven analyzed cultivars with a drastic difference between fresh and processed olives.

The main phenolic compound detected in fresh olives was oleuropein; its proportion ranged from 25.43% (Sevilla) to 39.10% (Gordal). The oleuropein level was variety dependant and Azzeradj from Seddouk presented the highest amount (994.27 mg·100 g⁻¹), while in Bouchouk this compound was not detected. These significant differences could be explained by the cultivated variety, cultivars with large fruit size (data not shown) have higher amounts of oleuropein (Azzeradj from Seddouk and Gordal) than the small size ones (Sigoise and Bouchouk) as reported previously by Amiot et al. (1990). The analyzed cultivars showed higher amounts of oleuropein compared to Tunisian cultivars (Ben Othman et al., 2009) and Italian cultivars (Piscopo et al., 2014). The results confirmed the effect of cultivar and geographical origin on the phenolic content of olives.

Verbascoside was the second most abundant phenolic compound; its values ranged from 391.08 to 407.548 mg·100 g⁻¹ for Azzeradj from Seddouk and Sigoise, respectively. The amounts of ligstroside vary between 65.30 mg·100 g⁻¹ (Bouchouk) and 334.89 mg·100 g⁻¹ (Azzeradj from Seddouk).

Similarly to the evolution of oleuropeine, verbascoside and ligstroside contents showed a decline in processed olives; their contents were under quantification limits in processed olives for the majority of cultivars, excepted for Sigoise. The results obtained confirm previous data (Boskou et al., 2006; Kiai and Hafidi, 2014) which reported a decrease in oleuropein and verbascoside contents, similarly to an increase in hydroxytyrosol. Processing according to Spanish style causes hydrolysis of phenol complexes due to NaOH treatment, leading to the liberation of simple phenols. Tyrosol showed a slight decrease as well after processing. This observation was not in agreement with the findings of Sahan et al. (2013) who observed an increase in the tyrosol content arising from ligstroside hydrolysis. In addition, fresh olives are rich in quercetin -3 galactoside (109.47 mg·100 g⁻¹ in Teffahi to 356.65 mg·100 g⁻¹ in Gordal). These values are much higher than the amount reported by Piscopo et al. (2014) in Italian cultivar; with the exception of Bouchouk (0.00 mg·100 g⁻¹).

In processed olives, the first major phenolic compound was hydroxytyrosol. The highest content was recorded for Sevilla (545.42 mg·100 g⁻¹ DW equivalent to 109 mg·100 g⁻¹ FW), this result is higher than that of Sahan et al., (2013) (26.45 mg·100 g⁻¹ FW). The lowest level was observed in Teffahi (14.49 mg·100 g⁻¹). It is known that hydroxytyrosol derives from the hydrolysis of oleuropein and verbascoside. An increase in the content of this compound was observed only in Sevilla and Bouchouk varieties. In contrast, for other varieties, a decrease was noted. This might be explained by the diffusion of this polar compound to the brine, or its oxidation during de-bittering (Pasqualone et al., 2014).

Caffeic acid, which is derived from the hydrolysis of verbascoside, was not detected in processed olives. Rodriguez et al. (2008) demonstrated that phenolic acids as caffeic, p-coumaric and ferulic acids are metabolized by L-plantarum, which contain a phenolic acid decarboxylase, to their corresponding vinyl derivatives. Romero et al. (2004) showed that simple phenolic compounds are converted by lactic acid bacteria and de-polymerized.

The ferric reducing power of the phenolic extracts of fresh and processed olives are represented in Table 2 (The three standards used recorded the same level of significance).The fresh olives of Azzeradj from Seddouk exhibited the strongest reducing capacity (4531.90 mg QEq/100 g; 7784.4 mg BHAEq·100 g⁻¹; 9180.82 mg BHTEq·100 g⁻¹) while Bouchouk recorded the lowest one (522.45 mg QEq·100 g⁻¹; 887.40 mg BHAEq·100 g⁻¹; 1058.66 mg BHTEq·100 g⁻¹).

Processing caused a decrease in the reducing capacity of extracts. The values ranged from 283.36 QEq·100 g⁻¹, 486.72 BHAEq·100 g⁻¹, 574.85 mg BHTEq·100 g⁻¹ (Azzeradj from Seddouk) to2316.28 mg QEq·100 g⁻¹, 3978.64 mg BHAEq·100 g⁻¹, 4687.47 mg BHTEq·100 g⁻¹ (Sigoise). So that, decreases in reducing power after processing were estimated to 37.78% (Sigoise), 52.42% (Sevilla), 55.25% (Azzeradj from Tazmalt), 68.39% (Teffahi), 76.09% (Gordal) and 93.98% (Azzeradj from Seddouk).

The results showed that a reduction in reducing power is related to a loss in phenolic content. A significante correlation (r=0.98) (Table 3) was obtained between the two parameters. Azzeradj from Seddouk recorded the highest phenol content loss (9.80%) and showed the highest reducing capacity loss (93.98%). In contrast, Sigoise, in which the phenolic content was more preserved, showed the lowest reducing power reduction (37%). No change is noted in reducing power for Bouchouk, which is probably due to its preservation of phenolic compounds.

The potential anti-oxidant index (PAOXI) values of extracts are given in Table 4. Lower PAOXI indicates a lower phenolic content and a better efficiency (Sun and Tanumihardjo, 2007). Our results indicated that all the samples (fresh and processed olives) are efficient because of their PAOXI values of <1. However, the differences noted among PAOXI values might be related to the differences in the phenolic profile of each cultivar. The five varieties processed according to the Spanish style exhibited a better efficiency than their corresponding fresh olives. They are classified as: Azzeradj from Tazmalt=Azzeradj from Seddouk>Sevilla>Teffahi>Bouchouk. This indicates that the processing had a positive effect on phenolic profiles (in some cultivars), which permits the transformation of glycoside to aglycone forms, which is more effective.

The scavenging capacity of the extracts (Table 2) showed no statistically significant difference between Gordal, Azzeradj from Seddouk and Sigoise for fresh olives. These cultivars exhibited the highest scavenging capacities. Differences noted among the other varieties should certainly be related to the differences in the phenolic profiles of cultivars.

The effect of processing on scavenging capacity differs among the cultivars; as a consequence of the decline in phenolic content, a reduction was noted for six cultivars with remarkable differences in loss percentages: 92.75% (Azzeradj from Seddouk), 72.82% (Gordal), 53.28% (Teffahi), 48.60% (Azzeradj from Tazmalt), 46.1% (Sevilla) and only13.12% for Sigoise, which was the best efficient extract among processed olives. This in agreement with the results of Sahan et al. (2013) for the Gemlik cultivar. On the contrary, Bouchouk showed an increase (33.57%) in antioxidant activity after processing, which can be attributed to the increase in cinnamic acid derivatives (100%), hydroxytyrosol (76.93%), benzoic acid derivatives (49.96%) and tyrosol (19.35%), confirming that the antioxidant effect does not depend only on the phenolic content but also on the phenolic profile. On the other hand, Brenes and de Castro (1998) claimed that the antioxidant activity of hydroxytyrosol is higher than that of oleuropein. Velkov et al. (2007) ranked the phenols of green olives according to their scavenging activity as: dihydrocaffeic acid>hydroxytyrosol>caffeic acid>oleuropein.

According to the Folin Ciocalteu assay, Gordal showed a higher phenolic content than Teffahi, while it presented the lowest antioxidant capacity; these contradictory results reflect the influence of the nature of phenolic compound on antioxidant activity.

A graphic representation% DPPH inhibition=f (C (mg·mL⁻¹)) revealed a perfect linearity for all samples (0.993≤r≤0.997) (data not shown) which indicates that the extract scavenging effect on DPPH radical increases with increasing concentrations. The scavenging capacity of fresh olives according to the effective concentrations (EC₅₀) (Table 2) followed the order: Azzeradj from Seddouk>Sevilla>Sigoise>Gordal>Teffahi>Azzeradj from Tazmalt>Bouchouk. For the processed olives the order was: Sigoise>Sevilla>Gordal>Teffahi>Azzeradj from Tazmalt>Bouchouk>Azzeradj from Seddouk. The results obtained for fresh olives (except for Bouchouk and Azzeradj from Tazmalt) are in good agreement with those of Arslan and Ozcan (2011) and Malheiro et al. (2011). Otherwise, Ilias et al. (2011) revealed an effective concentration two times lower for Sigoise from Tlemcen; this can be related to differences in composition linked to the geographical origin (Vinha et al., 2005).

The effective concentration (EC₅₀) of extracts exhibited an inverse relationship with phenol contents, showing a significant negative correlation (r=−0.772). Phenolic compounds of olive extracts are good hydrogen donors.

Sigoise processed olives were more effective in scavenging DPPH radical, showing smaller EC₅₀ values (0.38 mg/mL) than those found by Sousa et al. (2008) for the Portuguese Alcaparra variety. These results confirm the superiority of Sigoise processed cultivar in antioxidant activity, which can be related to its higher quercetine-3 galactosid and caffeic acid contents than other processed olives. The high EC₅₀values obtained for Bouchouk and Azzeradj from Seddouk were justified by their polyphenol contents.

Inhibition percentages of hydrogen peroxide in fresh and processed green olives (Table 2) at a concentration of 3.571 mg·mL⁻¹ showed significant differences (p<0.05). Fresh olives of the Gordal and Sevilla varieties exhibited the strongest scavenging activities (88.89 and 88.67%) similar to that of the Chemlal cultivar analyzed by Nadour et al. (2012) but at a concentration of 0.25 mg·mL⁻¹. Bouchouk exerts the lowest one (10.68%).

Losses in scavenging activities after processing were estimated to be 97.94, 97.19, 72.69, 68.8 and 59.45%, respectively, for Azzeradj from Seddouk, Gordal, Sevilla, Sigoise and Bouchouk. After processing, Teffahi recorded statistically the same activity and Azzeradj from Tazmalt showed an increase of 35.0%. This may be related to the formation of more effective phenols after processing, and mainly to the generation of phenolic acids (133.94 mg·100 g⁻¹) (hydroxycinnamic and hydroxybenzoic acid derivatives).

The antioxidant activity of flavonoids against H₂O₂ was not significant (r=0.404) (Table 3). However, if we consider separately fresh samples and processed samples, this coefficient was 0.814 and 0.361, respectively. Thus, the loss in this coefficient estimated at 2.25 explains the negative effect of processing by Spanish style on the flavonoid contents of green olives.

Potential antioxidant index (PAOXI) values of the extracts (Table 4) confirmed losses in scavenging activities by the increase of PAOXI values for the cultivars Sevilla, Bouchouk, Azzeradj from seddouk, and Sigoise. Although Gordal showed a very low efficiency of its phenolic compounds, requiring the implication of more than two phenolic fractions to enrich by one scavenging activity unity. The PAOXI value obtained for Azzeradj from Tazmalt is in agreement with the increase in scavenging activity, recording the lowest PAOXI value among the studied cultivars.

The percentages of inhibition bleaching of β-carotene (IBBC) for fresh and processed olive extracts at the concentration of 7.14 mg·mL⁻¹ are given in Table 2. No significant differences (p≤0.05) were noted between fresh and processed olives except for Azzeradj from Seddouk and Sevilla. This indicates that the phenolic compounds of green olives (fresh and processed) react strongly in an organic medium, probably due to their partition coefficient.

Losses in antioxidant activity after processing varied according to the cultivar: Azzeradj from Seddouk (51.66%), Sevilla (38.5%), Teffahi (20.04%), Bouchouk (13.01%), Azzerad from Tazmalt (9.56%), Sigoise (7.26%). However, an increase was recorded for Gordal which is probably due to the increase in phenolic acids, such as in cinnamic acid derivatives. Sroka and Cisowski (2003) postulated that phenolic acids are the weakest inhibitors in lipid peroxidation. Otherwise, contrary to the findings of Han et al. (2012), who reported a synergism action between flavonoids and β-carotene accordingly at the water/lipid interfaces, our results show a decline in flavonoid content and similar antioxidant activity after processing. A moderate correlation was established between β-carotene bleaching assay and total phenolics (r=0.67) (Table 3), this means that lipid peroxidation inhibitory activity could be partially correlated to the phenolic content.

The PAOXI values obtained (Table 4) clearly showed that the majority of extracts had better activities in an emulsion medium than in an aqueous one. This may be explained by the “Polar paradox” phenomenon characterized by the accumulation of a polar antioxidant in the oil-water interface, thus protecting the lipids from oxidation (Hayes et al., 2011).

To get a complete picture of the ranking of the antioxidant capacities of table olives, a relative antioxidant capacity index (RACI) was calculated by integrating the antioxidant capacity values generated from the different tests. RACI is the mean value of standard scores transformed from the initial data generated with different methods (Sun and Tanumihardjo, 2007). Results of the classification of samples (Figure 1) revealed the superiority of the fresh olives of Azzeradj from Seddouk in the totality of tests, giving a RACI value of +1.53. The processed olives of the same variety present the lowest RACI value (−1.41). The results showed that among the processed olives, only Sigoise denoted a positive value of RACI.