Total flavonoid contents were determined according to the procedure of Kim et al. (2003). An aliquot of sample (200µL) was mixed with distillated water (800 µL). A volume of 60 µL of 5% NaNO₂ was added to the flask. After 5 min, 60 µL of 10% AlCl₃ were added. At 6 min, 40 µL of NaOH (1 M) were added to the mixture. Immediately, the contents of the reaction flask were diluted with the addition of 480 µL of water and thoroughly mixed. Absorbance of the mixture was measured at 510 nm. Catechin was used as standard and the amount of flavonoids was calculated as milligrams of catechin equivalents (CE) per 100 g of dry weight.

The ferric reducing antioxidant power (FRAP) was applied as described by Benzie and Strain (1996). A volume of acetate buffer (160 µL) was added to 20 µL of sample. The mixture was placed in a well plate and put in the spectrophotometer (Thermo Scientific, Madrid, Spain) at 37 °C, then mixed during 2min. After that, 40 µL of FeCl₃ (10 mM) and 40 µL of 2,4,6-tripyridyl-s-triazine TPTZ (33%) were added, and the absorbance was recorded at 593 nm after 10 min. The results were calculated and related to a Fe⁺² standard solution tested in parallel and expressed in micromol of FeSO₄ per gram of dry weight (µmol of Fe⁺²·g^–1 dw).

The hydrophilic oxygen radical absorbance capacity (ORAC) assay is limited to the measurement of the hydrophilic chain breaking against peroxyl radicals. A further dilution of the olive extract was made with phosphate buffer. A portion of 25 µL of the diluted sample was added to a well in a 48-well microplate. A volume of 180 µL of fluorescein solution (0.45 mg·mL^–1) and 75 µL of the 2,2′-azobis (2-amidino-propane) dihydrochloride AAPH (60 mg·mL^–1) were added to the assay mixture. ORAC values were calculated and expressed as micromol of trolox equivalents per gram of dry weight (µmol of TE·g^–1 dw) (Prior et al., 2003).

Antioxidant activity was estimated according to the procedure described by Velioglu et al. (1998). A β-carotene solution was prepared in chloroform (0.12 mg·mL^–1). Next, 3 mL were taken and added to a flask containing 40 mg of linoleic acid in 400 mg of tween 20. The chloroform was removed in a vacuum evaporator, and then 100 mL of hydrogen peroxide (30%) were added. After thorough mixing, 3 mL of the emulsion were added to 0.5 mL of extract (without washing with hexane).

The oxidation of β-carotene emulsion was monitored at 470 nm after incubation at 50 °C (120 min). The antioxidant activity was expressed as percent inhibition relative to the control using the following equation:

AA = (R_control–R_sample)×100/R_control

Where R_control and R_sample were the bleaching rates of β-carotene in the reactant mixture without antioxidants and with olive extracts, respectively.

The results showed significant differences (P<0.05) in the total flavonoid contents among the studied cultivars (Table 1). The total flavonoid concentration of cultivars ranged between a mean value of 872 (Aberkane and Abelout) and 1537 mg CE·100g^–1dw (Azeradj) in fresh olives. These contents are higher than those obtained by Brahmi et al. (2013) who used methanol as the extraction solvent. However, the flavonoid amounts are comprised only between 394 (Abelout) and 1272 mg CE·100 g^–1dw (Azeradj) in salted olives. Consequently, the dry salting caused a decrease in flavonoid contents with a loss rate ranging from 22% (Azeradj) to a mean value of 55% (Abelout and Bouchouk). This decrease can be explained by the diffusion of these compounds under the action of salt and/or their oxidation during salting. Furthermore, the variability of the decrease noted among the studied cultivars can be related to the characteristics of each cultivar such as diameter of fruit, since the decrease is related to the diffusion of such compounds (Bianchi, 2003). In addition, the difference of the polarity of each flavonoid compound can also influence their diffusion (Tomás-Barberán and Gil, 2008).

The analysis of the studied cultivars showed considerable quantitative differences (P<0.05) in individual flavonoids (Table 2). Four flavonoids were identified: rutin, luteolin-7-O-glucoside, cyanidin-3-O-glucoside and cyanidin-3-O-rutinoside. The presence of rutin and luteolin-7-glucoside in olive fruits was always reported (Brenes et al., 1995; Morello et al., 2005; Savarese et al., 2007) except that the fresh olives in the present study did not contain the luteolin-7-glucoside until they were treated. Also, Romero et al. (2002a) identified cyanidin-3-O-glucoside and the cyanidin-3-O-rutinoside as the main pigments in seven natural black olive cultivars.

The amount of rutin in fresh olives varied between 227 (Abelout) and 1857 mg·Kg^–1dw (Azeradj). These contents are similar to those found by Sousa et al. (2014) and Morello et al. (2005) except for the Azeradj cultivar which contain a high concentration. This finding is similar to that reported by Garrido-Fernández et al. (1997) who consider that the flavonoid composition is useful for the biochemical characterization of olive cultivars.

The content of cyanidin-3-glucoside in fresh olives varied from a mean value of 16 (Bouchouk and Atefah) to 2588 mg·Kg^–1dw (Azeradj), whereas, the concentration of cyanidin-3-rutinoside is comprised between 27 (Atefah) and 4358 mg·Kg^–1dw (Azeradj). These amounts are relatively higher than those obtained by Romero et al. (2002a). The olive cultivars of the present study had the characteristic that the cyanidin-3-glucoside amount is higher than that of cyanidin-3-rutinoside. This could represent a useful tool for a phytochemical characterization of the cultivars.

After processing, we noted a decrease in individual flavonoid contents except for luteolin-7-glucoside. The content of luteolin-7-glucoside ranged from 30 (Abelout) to 370 mg·Kg^–1dw (Azeradj). These amounts are higher than the luteolin content reported by Dimitrios (2006) for Greek-style naturally black olives (25–75 mg·Kg^–1dw), although the rutin content ranged between 24 (Abelout) and 474 mg·Kg^–1dw (Azeradj). Piscopo et al. (2014) reported a decrease in the quercetin amount after the drying of green olives. Also, Brenes et al. (1995) noted that the rutin content of the olive flesh decreased with the alkaline treatment, and practically disappeared after the washing step. In the adopted method, the content of this compound decreased but it did not disappear. This reduction can be attributed to the glycosidic bond breaking during salting.

The obtained results indicate that the effect of dry salting is dependent on the individual flavonoids; it can induce a decrease (rutin) or an increase (luteolin-7-glucoside). This is in agreement with the data reported by Rice-Evans and Packer (2003), since salt can generate sodium adducts from flavonol-3-glucoside (rutin), and consequently, the content of the latter decreases. By contrast, these adducts are not obtained from flavone glucoside (luteolin-7-glucoside).

The dry salting significantly affects (P<0.05) the content of olive pigments: the cyanidin-3-glucoside disappeared, but the cyanidin-3-rutinoside is detected only in three cultivars with concentrations of 7 (Aberkane), 11 (Abelout) and 16 mg/Kg dw (Azeradj). This can be explained by the fact that anthocyanins are water-soluble compounds which diffused from the olive to the surrounding medium during dry salting. These substances can also be either transformed or degraded during processing. According to Garrido-Fernández et al. (1997), the anthocyanin contents may be strongly influenced by the processing and the cultivar; the total content can decrease to below 50% of its initial value.

The cultivar had a significant effect on the observed changes in the flavonoid composition of studied olives. Among the investigated cultivars, for both fresh and salted olives, Azeradj showed the highest flavonoid levels and even higher than that of other studies on olive cultivars (Morello et al., 2005; Damak et al., 2008).

The antioxidant capacity of olive extracts is determined by the ability of the antioxidants to reduce ferric to ferrous iron. The statistical analysis showed significant differences (P<0.05) in the FRAP values of the studied olive cultivars. They varied from 126 (Bouchouk) to 353 µmol of Fe⁺²·g^–1dw (Azeradj) for fresh olives, while these activities ranged between 127 (Bouchouk) and 209 µmol of Fe⁺²·g^–1dw (Azeradj) for salted ones (Figure 1). These data reveal that both fresh and salted olive extracts showed a marked capacity for iron reduction with values higher than those obtained by Ziogas et al. (2010). The ferric reducing antioxidant power of both fresh and salted olive cultivars decreased in the following order: Azeradj>Abelout=Aberkane>Atefah>Bouchouk. The processing had a significant effect (P<0.05) on this activity. Overall, we noted a decrease in ferric reducing capacity except for Bouchouk cultivar, which showed a stable activity. The decrease varied between a mean value of 16% (Abelout and Aberkane) and 41% (Azeradj). This result could be explained by the low contents of flavonoids and/or other reducing agents in salted olives, since, the antioxidant capacity of flavonoids has been attributed to their electron-donating ability (Morales-Soto et al., 2014). Also, Prior et al. (2005), consider that the reducing capacity is related to the degree of hydroxylation which is a characteristic of flavonoid compounds. The present study demonstrated that other factors such as the structure of antioxidant could also influence this activity, since Azeradj cultivar which had the lowest loss rate of flavonoids, lost almost half of its activity, while inverse effect was noted for the Abelout cultivar.

Figure 2 shows significant differences in antioxidant activity among the studied cultivars (P<0.05): ORAC_H values ranged between 201 (Bouchouk) and 551 µmol TE·g^–1dw (Azeradj) in fresh olives; these values varied from 199 (Atefah) to 418 µmol TE·g^–1dw (Aberkane) in salted ones.

The obtained results indicate that dry salting had a significant effect on the antioxidant capacity (P<0.05); it leads to a decrease for the studied cultivars except for Bouchouk which showed an increase of 37%. In fact, the loss rate varied from 13 (Aberkane) to 47% (Azeradj). The scavenging effects of fresh olive cultivars against the oxygen radical decreased in the following order: Azeradj > Aberkane > Abelout > Atefah > Bouchouk. After processing, this order changed to the following: Aberkane > Abelout ≥ Bouchouk > Azeradj > Atefah. This variation could be related to the specific changes in phenolic composition of each cultivar after processing, since the ORAC assay cannot be based only on the flavonoids which are effective electron-donors, but also includes the hydrogen donor compounds. On the other hand, the increase noted for Bouchouk cultivar can be explained by the possible synergistic effect between the antioxidant compounds detected after processing.

On the other hand, the results reveal that the cultivar containing high levels of rutin did not show necessarily low ORAC_H values. This osbervation is different than that of Ou et al. (2002). This can be related to the specific antioxidant composition of the olive cultivar and the possible synergistic effect of other antioxidants with rutin in the total ORAC assay.

The antioxidant activity of olive extracts measured by the bleaching of β-carotene is shown in Figure 3. The statistical analysis (P<0.05) revealed that the fresh olive extracts showed an inhibition rate which ranged between 32 (Aberkane) and 54% (Bouchouk).

Concerning the treated olives, the inhibition activity varied from 17 (Atefah) to a mean value of 33% (Aberkane and Abelout). Dry salting significantly affected the β-carotene bleaching activity of the studied olives except for the Aberkane cultivar; it induced a loss with a rate comprised between 9% (Abelout) and 52% (Atefah). We note that Azeradj cultivar, which showed high values of ORAC and FRAP, had a relatively low inhibition effect of β-carotene oxidation. This can be related to its low content in lipophilic antioxidants, which may play a major role in this activity since the extracts in the β-carotene bleaching assay are not defatted and contain both lipophilic and hydrophilic antioxidants. This observation is similar to that of Alu'datt et al. (2013) who demonstrated that full-fat olive extracts had higher antioxidant activity than de-fatted extracts.

Because multiple reaction characteristics and mechanisms are likely involved, each method only provides an estimate of antioxidant capacity which is subjective to its conditions and reagents.