The trials were carried out on Algerian extra monovarietal extra virgin olive oils of ten cultivars (Limli, Ferkani, Chamlel, Abani, Zeletni, Souidi, Bouricha, Aalah, Rougette de Mitidja and Mekki), obtained from a collection of the Technical Institute of Arboriculture of Fruits and Vine (TIAFV), located in the Soummam Valley (Sidi-Aïch, Bejaia, Algeria), central-eastern region of Algeria about 1 km from the national road 26.

The olive trees were grown under identical agronomic and pedoclimatic conditions. The selected olive trees were over 60 years old planted at 10×10 m and harvested by hand during the 2014/2015 and 2015/2016 seasons, from late December to early January, due to the different maturation periods of the different cultivars [maturation index (MI) ranging from 2.01 to 3.30]. Olive trees were subjected to non-irrigated conditions throughout the year, except during July and August, when trees were irrigated (50 L/m² per tree) once a week. Monthly precipitation from 2014 to 2016 for the study area is shown in Fig. 1. In the spring, foliar fertilizers were applied to stimulate growth shoot.

After harvesting, the olives were immediately transported to the laboratory in plastic crates. The olive oil was extracted separately from each cultivar using an oleodoseur Siol 20240 (Ghisonaccia, France) equipped with a centrifugal divider (48000 rpm) with a capacity of 1.5 kg. The olives were crushed using a hammer mill and the resulting paste was mixed in an olive paste blender at room temperature for 40 min. The olive oil was then separated by centrifugation without the addition of hot water using a two-phase decanter. The resulting oils were packaged in dark glass bottles and stored at -20 °C until analysis.

In this work, ultrapure water from a Milli-Q system (Millipore, Milford, MA) was used to prepare all solutions. All reagents and standard stock solutions used were from Merck (Darmstadt, Germany). For digestion preparation, nitric acid (65%) and hydrogen peroxide (30%) were employed. To reduce the risk of contamination, all laboratory glassware was soaked in 10% v/v HNO₃ for at least 24 h and then rinsed copiously with ultrapure water before use.

An Agilent 240FS AA Fast Sequential atomic absorption spectrometer (FAAS) was used for the determination of Zn, Fe, K, Na and Mg. As, Co and Cu were analyzed by GFAAS, an Agilent 240Z AA Zeeman graphite furnace atomic absorption spectrometer was used for graphite furnace measurements. The operating conditions for FAAS and GFAAS are presented in Table 1 and Table 2, respectively. The samples were digested using a microwave-assisted digestion (Milestone Ethos MicroSYNTH) oven with programmable power control (10W increments, maximum power 1000 W) with segmented rotor MPR-600 (35 bar of maximum operating pressure and 260°C of maximum operating temperature) and 10 reaction vessels.

Triplicate samples of olive oil samples from each cultivar were weighed (0.5 g) and added into the digestion vessel with 7 mL of nitric acid and 1 mL of hydrogen peroxide. Then, the samples were digested in a microwave digester using a two-step temperature program. In the first step, the temperature was linearly increased to 200 °C over 10 min; the maximum power of the rotating magnetron was 1000 W. In the second step, the temperature was maintained at 200 °C for 15 min. After digestion and cooling at room temperature, all the digestion liquors were diluted with ultrapure water in a 20 mL volumetric flask. Prior to analysis by FAAS and GFAAS each sample was vigorously shaken.

For the quantitative analysis of oils, all calibration curves were built from standards at five different concentrations. The element standards solutions were prepared by diluting stock standards for each element (1000 mg/L) in a nitric acid solution (1% v/v). Calibration ranges were selected according to the expected concentrations of the elements of interest.

The results of the mineral composition were expressed as a mean value ± standard deviation. The statistical analysis was carried out using SPSS 21 statistical package program. Two-way ANOVA was employed to find the significant differences of mineral elements contents in ten cultivars of olive oils with respect to cultivar and year of harvest. The significant differences were set at p<0.05. Mineral elements’ data was analyzed through principal component analysis technique (PCA), using XLSTAT version 19.3.01.

Macro minerals play an important role in various bodily functions. Sodium is the main cation in extracellular fluid in the body, the recommended daily intake for Na is 2 g/day (equivalent to 5 g salt/day) (WHO, 2007). Na concentrations in our samples were high in the 2014/2015 season samples. This trend was particularly relevant for the cultivar Rougette de Mitidja, which had the highest value at 104.720 µg g^-1. The Na levels in olive oil samples were higher than in previous reports (Mendil et al., 2009; Zeiner et al., 2010). However, concentrations were lower than these literature values 73.401-390.699 µg g^-1 (Fuqha et al., 2015), 110-320 µg g^-1 (Gouvinhas et al., 2015) and 150-200 µg g^-1 (Gouvinhas et al., 2016).

Relevant to the Na concentrations, Mg levels for the 2014/2015 season (1.040-2.560 µg g^-1) were also higher than those for 2015/2016 season (0.120-0.560 µg g^-1). In other works by Zeiner et al. (2005) and Fuqha et al. (2015), the concentration of Mg ranged 2.910-3.620 and 294.738-782.968 µg g^-1 respectively.

In contrast to Na and Mg, the concentrations of K in our samples from 2015/2016 season were relatively higher than those of the corresponding samples from the 2014/2015 season. The highest value (117.133 µg g^-1) was observed in cv. Aalah. WHO (2012) suggests a K intake of at least 90 mmol/day (3510 mg/day) for adults. The values of K from the present study were higher than those reported by Mendil et al. (2009) and Zeiner et al. (2010), with the exception of the Fuqha et al. (2015), who reported a higher value of 168.883 µg g^-1.

In fact, the year of harvest had a significant influence on Na, Mg and K levels. The increase in Na and Mg concentrations and the decrease in K concentrations obtained in the 2014/2015 season, whereas the opposite was observed in the 2015/2016 season, is probably the result of the interactions between minerals. These interactions can be modulated by environmental and agronomic conditions such as precipitation, irrigation in summer and foliar fertilizers; salts in water can have an influence on plant growth and, therefore, on the efficiency in nutrient use. It is well known that K, Ca and Mg interact in the soil exchange complex, sometimes inducing deficiencies in one element through an excess of another. Interactions also occur at the plant level, e.g. a deficiency or excess of one element may affect the uptake or utilization of another (Fernández-Escobar et al., 2019). The concentration of NaCl in the soil can interfere with the absorption of cations such as K+ (Tester & Davenport, 2003) and excessive amounts of K reduce Mg absorption (Guo et al., 2016). Lower K concentrations compared to Na concentrations have been reported in olive root and fruit (Bedbabis et al., 2014).

Elemental ions such as Cu, Fe and Zn are involved in many crucial biological processes and are necessary for the survival of any living organism. Although, many reports describe the deleterious effect of trace amounts of elements on the oxidative stability of oils, Fe, Ni and Co in particular, significantly reduce the oxidative stability of unsaturated fats and accelerate the development of rancidity in edible oils and fats when their concentration is a few ppm (Angerosa & Di Giacinto, 1993).

Regarding the elemental analysis, the quantities of Cu detected ranged from 1.383 to 3.337µg g¹. Oils harvested in the second year had higher Cu concentrations than those harvested in the 2014/2015 crop year. With the exception of ˈZeletniˈ, which has the highest value in the first crop year. Our values were in agreement with some publications (Fuqha et al., 2015). However, any sample we analyzed exceeded the maximum limit of 0.1 mg/kg set by the International Oil Council (IOC, 2015) in virgin olive oils.

In addition, our results showed a high variability in Fe content in the first and second crop years. For all cultivars, levels of Fe were higher in the first year than in the second, except for ˈBourichaˈ, ˈAbaniˈ, ˈFerkani and ˈSouidiˈ. Our results (1.640-13.213 μg g¹) were higher than those obtained by Pehlivan et al. (2008) (0.0125-0.0295 μg g¹) and by Zeiner et al. (2010) (0.05-4.61 µg g-¹), but close to those of Mendil et al. (2009) (52.0-291.0 µg g-¹) and Zhu et al. (2011) (32.8-34.1 µg g-¹).

In accordance with Fe, large variations were also observed for Zn between cultivars for the same crop year and between both crop years. The highest concentrations were present in the samples of olive oils harvested in the 2015/2016 crop (2.666-32.866 µg g-¹). Our results were higher than those of other studies (Beltrán et al., 2015) but close to those of Fuqha et al. (2015) which were about 13.743 and 116.658 µg g¹.

The variation in Cu, Fe and Zn contents was influenced by both type of cultivar and year of harvest. In relation to the type cultivar, the olive trees are grown in the same area, so they have been exposed to the same soil and climatic conditions, and extraction conditions were the same for all cultivars. However, in the same crop season mineral contents differences in olive oil were observed, which could be explained by the genotype. The genotype can influence growth rate directly, or indirectly through its influence on nutrient utilization efficiency (Chatzistathis et al., 2012); for example, according to Nergiz & Engez (2000), cultivar properties influence the iron levels in olive fruit. Seasonal differences could be explained by environmental and agronomic factors; the presence of discharging waste from industrial and domestic sources into the Soummam River, which is not far from the study area, appears to have effect on Cu, Fe and Zn concentrations. Mouni et al. (2009) showed significant concentrations on Fe in water samples from the Soummam River. Indeed, the work of Gouvinhas et al. (2016) on olive oils from three major Portuguese cultivars (‘Galega’, ‘Cobrançosa’ and ‘Picual’) at different stages of ripening showed that the main variations in mineral content affect Fe, Zn and Cu. In addition, Angioni et al. (2006) showed high variability within cultivars and with the first and second harvests for Zn.

The FAO/WHO (1999) has set a limit for elements intakes based on body weight. For an average human adult (60 kg body weight), the provisional daily intake (PTDI) for Cu, Zn and Fe are 3 mg, 60 mg, and 48 mg, respectively. Regarding the bioavailability of trace elements, rats fed with an olive-oil supplemented diet had significantly lower levels of serum iron and transfertin saturation than control (Chetty et al., 1999), while Zn and Mg concentrations in mice tissue remained unchanged (Milin et al., 2001).

The determination of trace and toxic heavy metal concentrations in food is a major concern in public health studies. As far as Co is concerned, this element is beneficial to humans because it is part of vitamin B12, which is essential for maintaining human health. Cobalt (0.16-1.0 mg cobalt/kg body weight) was also used as a treatment for anemia (Agency for the Toxic Substances and Disease Registry (ATSDR, 2004). Nevertheless, based on animal data on acute and chronic exposure to Co, the International Agency for Research on Cancer (IARC) has determined that it is possibly carcinogenic to humans (ATSDR, 2004). The Co content of the samples ranged from 0.006 to 0.146 μg g¹. These values are lower than those previously reported by Mendil et al. (2009) (1. 3 μg g¹).

Inorganic forms of As such as arsenite and arsenate are more dangerous to human health. They are highly carcinogenic and can cause lung, liver, bladder and skin cancer. Humans are exposed to arsenic through air, food and water (Jaishankar et al., 2014). The highest As content in our oil samples was found in Limli (0.051 µg g1), the presence of this element could be due to environmental contamination. It should be noted that the study area is located 1 km from National Highway 26, which is considered one of the busiest traffic roads in Algeria, with vehicular traffic polluting the road environment and neighboring agricultural land with a range of contaminants. Heavy metals are found in fuels, fuel tanks, engines and other vehicle components, catalytic converters, tires and brake pads, and road surfacing materials (Zehetner et al., 2009). According to Yan et al. (2013), the rank of risk contribution to environments among the eight heavy metals studied was Cd > As > Ni > Pb > Pb > Pb > Cu > Co > Zn > Cr. Indeed, Madejón et al. (2006) observed the presence of As in olives produced in the Guadiamar Valley (southwest Spain), which was polluted by a spill from an open pit pyrite mine.

Chemometric tools were applied to all data. PCA made it possible to group the variables on the basis of mutual correlations and to group the samples according to their similarities. The mean values (n = 3) of the elements detected (Fe, Zn, Cu, Na, K, Mg, As, Co) were used as variables to perform an exploratory analysis using PCA. The first three principal components were selected; they accounted for more than 76.25% of the total variation. Principal components 1, 2, and 3 explained 46.91, 17.42 and 11.92% of the total variance, respectively; they effectively characterize the mineral elements in the cultivars. The variables with the highest absolute load value on the principal component 1 were Mg, Na, Co and As, while for principal component 2, Fe and K had a higher load, and for the principal component 3, Zn are most important. The interrelations between the elements are most evident in the projections of components 1 and 2, and 1 and 3 (Fig. 2A and Fig. 2B, respectively).

A loading plot for the PC1 and PC2 plane and the PC1 and PC3 plane (Fig. 2A and Fig. 2B) allowed a good separation of oils obtained from the olives during the 2014/2015 and 2015/2016 seasons, which is mainly determined by the concentration of the elements Mg, Na, Co and As.

According to the Fig. 2A, there is a negative correlation between the elements Mg, Na, Co and the toxic metal As. It is clear that the concentration of these elements is strongly linked to the agro-climatic conditions and consequently to the specific season on which the olives were harvested. In the score plot generated by combining principal components 1, 2 (Fig. 2A), Fe was the element that contributed most to the categorization with distinct cultivars included in this study. PC2 therefore reflects the particularity of the ˈLimliˈ cultivar, in terms to its composition, which is strongly composed of Fe during the two harvest years. In PC1 and PC3 (Fig. 2B), the high Zn concentrations of ˈBourichaˈ and ˈZeletniˈ cultivars are responsible for the high score of the third component.

For the first time, this study reports mineral composition of Algerian monovarietal olive oils, obtained from two consecutive crop seasons, 2014/2015 and 2015/2016. The obtained results demonstrate that the combination of experimental data and chemometric technique can be used to distinguish olive oils samples according to cultivar and year of harvest. Fe is the element that contributes most to the categorization of cultivars, while Na, Mg, Co and As are the major elements that are strongly related to local agricultural and environmental conditions and therefore determine the crop season.