The olive is one of the most ancient domesticated tree crops (Besnard et al., 2013). Olive oil consumption is considered a hallmark of the traditional Mediterranean diet, which has been associated with reduced mortality caused by cardiovascular diseases (Psaltopoulou et al., 2004). The beneficial nutritional properties of olive oil were traditionally associated with its high content in monounsaturated oleic acid, in contrast to higher poly-unsaturation levels in seed oils until comparative studies with high-oleic seed oils suggested an additional role of minor constituents (Pérez-Jiménez et al., 1995). Minor components in olive oil associated with health-promoting effects include polar phenolic compounds, squalene, tocopherols, and sterols (Boskou, 2009).

The necessity of breeding efforts in olive has not been as evident as in other fruit species (Byrne, 2012). In fact, systematic breeding programs have been scarce until very recently, producing only a limited number of cultivars so far (Bellini et al., 2008). The main objectives in olive breeding are early bearing, high productivity, resistance to biotic and abiotic stresses, suitability to different growing systems, mechanical harvesting aptitude, and oil content and quality (León et al., 2008; Fabbri et al., 2009). At the first stages of olive breeding, the focus is mainly placed on vigor and fruit traits. Fruit and oil quality traits are generally considered only in the last selection steps. Several authors have reported improved cultivars/advanced selections with enhanced oleic acid, tocopherol, and/or total phenolic contents (Manaï et al., 2007; Baccouri et al., 2007; Ripa et al., 2008; León et al., 2011; De la Rosa et al., 2013). Practically no breeding research has been conducted on other oil quality traits such as squalene and phytosterols.

Squalene is one of the most biologically active constituents of olive oil due to its powerful antioxidant and anti-carcenogenic properties (Sotiroudis and Kyrtopoulos, 2008). Chemically, squalene is a terpenoid hydrocarbon synthesized as a biochemical intermediate at the initial steps of the phytosterol biosynthesis. Virgin olive oil is the richest vegetable source of squalene (Boskou, 2009). Phytosterols are compounds that play an important nutritional role by reducing cholesterol absorption due to their similar chemical structure and biological function as cholesterol (Piironen et al., 2000). From a technological perspective, phytosterols such as Δ⁵-avenasterol that contain an ethylidene group in their structure confer higher antioxidant activity at frying temperatures than other sterols (Rossell, 2001). The presence of this sterol in olive oil has been associated with retarded polymerization in heated triacylglycerols (Boskou, 2011). Tocopherols are compounds with free radical scavenging activity, both in vivo (vitamin E) as well as in vitro. α-tocopherol, which is the tocopherol homologue with the highest vitamin E activity (Eitenmiller and Lee, 2004), is the predominant tocopherol form in olive oil (Boskou, 2009). Total tocopherol content in olive oil is much lower than in seed oils, including sunflower oil that also contains primarily α-tocopherol (Gunstone and Harwood, 2007).

One of the main constraints for the selection of oil quality traits is the scarce information about their genetic control and heritability (Fabbri et al., 2009). Most of the studies have focused on the fatty acid profile, with several studies reporting a great genotypic effect for the concentration of the major fatty acids and subsequent feasibility of selection for this trait (Ayton et al., 2001; León et al., 2008; Ripa et al., 2008; Rjiba et al., 2010; De la Rosa et al., 2013). For total phenolic content, the effect of the genotype has been found to be non-significant (Ripa et al., 2008) or of very low magnitude (El Riachy et al., 2011). Conversely, the genotypic effect was reported to be slightly higher than the year effect for tocopherol content, with low genotype × year interaction (Beltrán et al., 2010). Temperature and rainfall are the main environmental factors which influence olive oil quality (Aparicio and Luna, 2002). Beltrán et al. (2005, 2010) reported higher tocopherol content associated with lower rainfall, though a similar effect has not been observed for different irrigation levels (Gómez-Rico et al., 2007). Irrigation has been found to significantly influence the squalene content of olive oil (Martinelli et al., 2012). The altitude at which the trees are grown and the degree of maturity of the fruits are also two important factors affecting olive oil quality (Aparicio and Luna, 2002).

Seven olive breeding seedlings coming from crosses between the cultivars ‘Arbequina’ and ‘Picual’ were initially selected for their early cropping and oil content (León et al., 2004). The selections have been previously evaluated for fruit traits and fatty acid composition (De la Rosa et al., 2013). The objective of this research was to evaluate the variability for squalene content and tocopherol and phytosterol contents and profiles in those selections.

The Spanish cultivars ‘Picual’ and ‘Arbequina’ and seven breeding selections derived from crosses between them, namely UC-I-22-90, UC-I-2-35, UC-I-32-78, UC-I-36-41, UC-I-36-43, UC-I-37-69, and UC-I-42-48 were used. A detailed pedigree of the selections was previously reported (De la Rosa et al., 2013). A field trial was established in an open field in Cabra (Córdoba, Spain) in June 2007 at 6×5 m spacing using plants propagated by semi-hardwood stem cuttings. A random sample of around 500 g of fruits was collected from three trees per genotype (replications) in mid-November of 2010 and 2011. Average maximum, mean and minimum temperature and rainfall for both years are shown in Table 1. The ripening index at the time of harvest, averaged over the two years, was 3.5 for ‘Picual’ and 2.8 for ‘Arbequina’. The ripening index for the selections ranged from 2.0 (UC-I-36-43) to 3.8 (UC-I-37-69). Considering all the genotypes, the ripening index was higher in 2010 (3.45) than in 2011 (2.48). More detailed information on the ripening index of individual selections was reported by De la Rosa et al. (2013).

A subsample of around 30 g was stored at −80 °C shortly after harvest and lyophilized. The stones were then removed and the flesh was milled in a laboratory ball mill. The ground samples were then stored at −20 °C and analyzed in duplicate for squalene content, phytosterol content and profile, and tocopherol content and profile following procedures previously reported (Velasco et al., 2014).

Data were analyzed by the General Linear Model procedure of IBM SPSS Statistics version 20.0 (IBM Corp., Armonk, NY, USA) using the following model:

Yij=μ+Gi+Yj+GYij+εij

Where: Yij is the observation of genotype i in year j; μ is the general mean; Gi and Yj are effects of genotype i and year j, respectively; GYij is genotype × year interaction of genotype i with year j, and εijr is the residual error of genotype i in year j. Genotypes and years were considered as random variables. Pearson's correlation coefficients among oil quality traits were computed using IBM SPSS Statistics 20.0.

The analysis of variance revealed a significant effect of genotype for all the traits, whereas the year effect was only significant for tocopherol content and campesterol and γ-tocopherol concentrations (Table 2). The case of tocopherol content was particularly relevant, as the effect of year was much higher than the genotypic effect. The average tocopherol contents were 123 mg·kg⁻¹ in 2010 and 180 mg·kg⁻¹ in 2011. The genotype × year effect was significant for most of the traits, except for squalene content and concentrations of campesterol and stigmasterol in the phytosterol fraction. When significant, the genotype × year effect was in all cases of low magnitude compared to the main effects.

The cultivars ‘Arbequina’ and ‘Picual’ showed very contrasting levels for most of the traits, particularly for squalene contents (1800 and 5207 mg·kg⁻¹ dry flesh weight, respectively), phytosterol contents (1485 and 1109 mg·kg⁻¹ dry flesh weight, respectively), phytosterol profile (e.g. Δ⁵-avenasterol concentration of 22.1 and 10.2% of total phytosterols, respectively), and tocopherol profile (e.g. α-tocopherol concentration of 99.1 and 92.2% of total tocopherols, respectively) (Table 3). The breeding selections derived from them showed great variability for all the traits, covering in most cases the ranges of variation between the parents and presenting for some of the traits lower or higher levels than the parents, e.g. for phytosterol content, concentrations of stigmasterol, β-sitosterol and Δ⁵-avenasterol, and total tocopherol content (Table 3). The selection UC-I-36-41 showed a significantly lower phytosterol content than both parents, whereas no selection with significantly higher phytosterol content was identified (Table 3). For phytosterol profile, the selection UC-I-2-35 had a lower β-sitosterol concentration (62.6% of total sterols) and a higher Δ⁵-avenasterol concentration (30.7%) than both parents, whereas UC-I-36-41 showed a higher stigmasterol concentration (Table 3). The selection UC-I-37-69 exhibited a higher tocopherol content (263.2 mg·kg⁻¹ fruit flesh) than both parents (125.0 mg·kg⁻¹ in ‘Picual’ and 147.9 mg·kg⁻¹ in ‘Arbequina’). Phenotypes with higher levels than the parents for the mentioned traits were observed for both years. This is shown in Figure 1 for squalene, phytosterol and tocopherol contents, and the concentration of Δ⁵-avenasterol. No significant correlations among traits were observed in the average values of the seven selections, except for the expected negative correlation between the major components of phytosterol and tocopherol profiles, i.e r=−0.96 between the concentrations of β-sitosterol and Δ⁵-avenasterol, and r=−0.99 between the concentrations of α- and γ-tocopherol. Two-year average values for the parents and selections are shown in Table 3. It is important to emphasize that no significant correlations were detected between phytosterol contents and the levels of their common precursor, squalene.

The analysis of variance revealed high genotype effects, non-significant or low-magnitude year effects, and genotype x year interactions of low magnitude. The main exception was total tocopherol content, for which a strong year effect was detected. Previous studies have associated total tocopherol content in olives with annual rainfall, with high annual rainfall reducing the tocopherol content (Beltrán et al., 2005; 2010). A similar effect has been observed in the present research; tocopherol contents in the genotypes included in the study averaged 123.3 mg·kg⁻¹ in 2010, which was a very wet year with an annual rainfall of 1058 mm, whereas tocopherol content averaged 180.3 mg·kg⁻¹ in 2011, a much dryer year with an annual rainfall of 491 mm (Table 1).

The cultivars ‘Picual’ and ‘Arbequina’ have contrasting oil qualities, particularly for fatty acid profile, squalene content, phytosterol content and profile, and total phenolic content (Allouche et al., 2007). Both cultivars differed largely for squalene content and phytosterol content and profile in the present study. One of the most relevant results of this research was that the progenies from crosses between both cultivars showed segregation for some traits that surpassed the values found in the parents, e.g. for total phytosterol and tocopherol contents, and for the concentrations of stigmasterol, β-sitosterol and Δ⁵-avenasterol. Broad segregation in olive progenies has been reported previously for reproductive traits (Ben Sadok et al., 2013) as well as for oil quality traits such as fatty acid profile, tocopherol content and profile, polyphenols, and carotenoids (León et al., 2011). The most interesting variation identified in the present research consisted of increased levels of Δ⁵-avenasterol (30.7% of total sterols) in the selection UC-I-2-35 and tocopherols (263.2 mg·kg⁻¹ fruit flesh) in the selection UC-I-37-69 (Table 3).

The typical range of variation of Δ⁵-avenasterol in olive oil is between 5 and 20% of total sterols (Boskou et al., 2006). Values up to 36% have been reported exceptionally in some Greek oils (Boskou et al., 2006), probably reflecting particular environmental and/or processing conditions. Sterol composition in olive oil is affected by cultivar, crop year, geographic factors, degree of fruit ripeness, storage time of fruits prior to oil extraction, and processing (Boskou et al., 2006). The cultivars with the highest Δ⁵-avenasterol content are Koroneiki (23.9%) (Vekiari et al., 2010), Arbequina (23.2%) (Gracia et al., 2009), and Leccino (21.4%) (Gül and Seker, 2006). Accordingly, the high Δ⁵-avenasterol levels in the selection UC-I-2-35, observed in both years of the experiment (Figure 1), are exceptional within olive oil cultivars. Additional research should determine their effect on the oxidative stability of the olive oil.

Increasing tocopherol content is an important breeding objective for improving the nutritional quality of olive oil. The tocopherol content in virgin olive oil typically ranges from 100 to 250 mg·kg⁻¹ oil (Boskou, 2009), although higher values of up to 510 mg·kg⁻¹ oil have been reported for some cultivars in single environments (Psomiadou et al., 2000; Beltrán et al., 2010). Tocopherol content is influenced by the same factors mentioned above for Δ⁵-avenasterol, i.e. cultivar, crop year, geographic factors, degree of fruit ripeness, storage time of fruits prior to oil extraction, and processing (Boskou et al., 2006). Nitrogen fertilization has also been found to influence tocopherol content in olive oil, with high fertilization levels decreasing tocopherol contents (Fernández-Escobar et al., 2006). The maximum average tocopherol content identified in the present research, 263.2 mg·kg⁻¹ fruit flesh in selection UC-I-37-69, was exceptionally high compared to the values found in both parents (148 mg·kg⁻¹ in ‘Arbequina’ and 125 mg·kg⁻¹ in ‘Picual’).

Virgin olive oil is the richest vegetable source of squalene, with typical levels between 200 and 7500 mg·kg⁻¹ (Boskou, 2009). The squalene content in olive oil mainly depends on the cultivar and the extraction and refining technologies (Wiesman, 2009; Nergiz and Çelikkale, 2011). The squalene content of the extracted oil has been reported to decrease during maturation (Sakohui et al., 2011; Ben Mansour et al., 2015a). However, Fernández-Cuesta et al. (2013) found that, at the fruit pulp level, squalene content was scarcely related to maturation. Squalene content has also been reported to be affected by the geographic area (Ben Mansour et al., 2015b) and agronomic practices such as irrigation (Martinelli et al., 2012). In the present research, one of the selections (UC-I-42-48) showed high squalene content. Although the squalene content of this selection was not significantly higher than the best parent, ‘Picual’, the values in the fruits of the selection were higher than in the parent in both years, i.e. 5871 mg·kg⁻¹ in UC-I-42-48 compared to 5434 mg·kg⁻¹ in ‘Picual’ in 2010, and 5552 mg·kg⁻¹ in UC-I-42-48 compared to 4980 mg·kg⁻¹ in ‘Picual’ in 2011 (Figure 1).

The genotypes included in this study are advanced selections of a breeding program in which the main initial selection criteria have been early bearing and high oil content. They have also been characterized for fatty acid profile and fruit traits (De la Rosa et al., 2013). The most promising selections in that research, UC-I-42-48 and UC-I-2-35, show other additional interesting oil quality attributes. For example, UC-I-42-48 showed very high squalene content, whereas UC-I-2-35 showed very high Δ⁵-avenasterol content. Interestingly, the latter selection is also characterized by high oleic acid content (De la Rosa et al., 2013), which is one of the main factors contributing to oil thermo-stability (Allouche et al., 2007). Further research is required to investigate the contribution of Δ⁵-avenasterol to the thermo-stability of the oil as well as alleged synergistic effects between high oleic acid and high Δ⁵-avenasterol contents.

The present research identified great genetic variability for squalene and tocopherol contents and phytosterol content and profile in advanced olive selections, including phenotypes with higher levels than the parents for tocopherol content and Δ⁵-avenasterol concentration. These selections may play an important role in improving olive fruit and oil quality for specific market niches.