This study was carried out with 10 walnut genotypes from the Asni region in the high Atlas Mountains (center of Morocco), with wealthy genetic resources. These genotypes were selected after four years of observations based on morphological and homological traits, such as the general status of the plant (vigor, ramification, foliar density and appearance), lateral fruiting, and physical quality of the kernel, as appreciated by the local population. These genotypes were unique seedlings; therefore each genotype is a single tree. The best genotypes were marked and fruits were collected in winter during two consecutive years (2009–2010). The nuts were collected in lots of about 1 kg when the fruit mesocarp had fully dried and split and peduncle abscission was complete. Then the mesocarp was removed and the nuts were dried and stored at ambient temperature for three weeks. From each lot, representing each genotype, two independent samples of 30 fruits were randomly taken for analysis. After cracking, the kernels were tromped in liquid nitrogen and then ground using an electrical grinder (IKA, Staufen, Germany) to obtain a fine flour.

Oil was extracted from 5 g of ground walnut kernel using a fat extractor Soxtec during 5 h and using hexane as a solvent (AOCS, 1973). The oil content was expressed as the difference in weight of the dried kernel samples before and after extraction. The protein content was obtained indirectly by determining the total N content obtained by the Kjeldahl method (AOAC, 1995) and multiplying by a nitrogen-protein conversion factor (Kc=6.25; % protein=Kc* % total N).

FAs were extracted using hexane (AOCS, 1973) and analyzed by Gas Chromatography (Series 6890, column CP 9205 WAX MS, Agilent, Santa Clara, CA, USA) as the methyl ester derivatives (FAMEs) after trans-esterification using a methanolic solution of Boron trifluoride (BF3, 10%). These methyl esters were separated using a flame ionization detector (FID). The carrier gas was helium at a flow rate of 1 mL·min⁻¹. The temperatures of the inlet and detector were held at 220 and 275 °C, respectively. The initial column temperature was 100 °C for 3 min. The oven temperature was increased from 100 to 150 at 20 °C·min⁻¹ ramp rate for 1 min, from 150 to 200 at 15 °C·min⁻¹ ramp rate for 3 min, and from 200 to 240 at 3 °C·min⁻¹ ramp rate. The temperature was kept at 240 °C for 4 min. The injection volume was 0.5 µL. The identification of individual fatty acid was achieved by comparison with the relative retention times of previous fatty acid standard injections and was expressed as percent of total fatty acids.

Tocopherol concentrations were determined in each duplicate sample as previously described (López-Ortiz et al., 2008). Samples of 0.30 g of walnut oil were dissolved in 2 mL of 1-propanol by shaking at air temperature for 3 min. The resulting mixture was then filtered through a 0.45 µm nylon syringe membrane (Phenomenex, Torrance, CA, USA) before measurement. The final extract was kept in a dark vial in a refrigerator at 2°–4 °C until chromatographic analysis. Tocopherol homologue determinations were performed using a Kontron HPLC 360 equipped with a double piston pump (Kontron, Eching, Germany). Tocopherol homologues were detected using a Kontron 440 photodiode array detector and a Sfu 25 fluorescence detector. The chromatographic conditions finally selected for the simultaneous determination of all tocopherol homologues were as follows: 20 µL of sample were injected into the chromatography column Phenomenex Luna C18 (2) (0.5×4.6 mm, 5 µm particle size, 100Å pore size) while the mobile phase consisted of an acetonitrile:methanol (30:70) mixture at 30 °C at a flow rate of 1.2 mL·min⁻¹ and maintained at 30 °C. Tocopherol homologues were detected at 295 nm. Peak areas were measured, and retention times were compared with standards of the three homologues (Sigma-Aldrich, Madrid, Spain), with a purity of 95% for α-tocopherol, 96% for γ-tocopherol, and 90% for δ-tocopherol. The values were confirmed by their characteristic spectra using the photodiode array detector, which also confirmed their purity. To quantify all the homologues, calibration curves were drawn. Standard linearity was verified in each case by analysis of six standards in triplicate each containing 20–200 mg·kg⁻¹ oil for α-tocopherol, 0.1–8 mg·kg⁻¹ oil for γ-tocopherol and 0.05–5 mg·kg⁻¹ oil for δ-tocopherol. With this method, the recovery of the three tocopherol homologues was approximately 98%. Under these experimental conditions the limits of detection calculated from the residual error of the calibration curves were: 5.5 mg·kg⁻¹ oil for α-tocopherol, 0.2 mg·kg⁻¹ oil for γ-tocopherol and 0.1 mg·kg⁻¹ oil for δ-tocopherol. Tocopherol compositions were the mean values of three replicates (n=3) from each duplicate sample and were expressed as mg·kg⁻¹ oil.

All statistical analyses were performed with the SAS program (SAS, 2000). The analysis of variance with the PROC GLM procedure was applied to distinguish the effect of the genotype and the year. The additive linear model for the statistical analysis was:

P_ij=µ+g_i+y_j+ (g×y)_ij+ε_ij

Where P_ij is the phenotypic value of the i genotype in the j year, µ is overall mean, g_i is the genotype effect, y_i is the year effect, (g×y) _ij is the genotype and year interaction and ε _ijk is the residual effect. The mean separation was done with the Duncan test at P≤0.05. The LSMEANS option of the GLM procedure was used to calculate least-square means for genotypes in each year to observe rank changes of genotypes from year to year. The coefficients of correlation between the studied variables were obtained and the Pearson’s correlation coefficients were calculated following the PROC CORR procedure.

Oil and protein contents were highly variable among genotypes (Table 1). Protein content was significantly different between years (Table 2), with a higher mean value in 2009 (Table 1). The effect of year was not significant for oil content (Table 2), although it had been significant in three other walnut cultivars (Martinez et al., 2006). Oil content ranged between 54.4 and 67.48%, and protein between 11.58 and 14.5% in 2009, (Table 1), whereas in 2010 these ranges were between 55.29 and 66.25% for oil, and between 11.99% and 14% for protein (Table 1). Protein content agreed with previous reports (Amaral et al., 2003; Savage et al., 1999), as well as fat content, although the lowest fat value was lower than any previous report (Crews et al., 2005; Amaral et al., 2003; Martinez et al., 2006; Bada et al., 2010). The significance of the year×genotype interaction (Table 2) indicated that oil content variability was significant between years for some genotypes, as a result of the specific response by these genotypes to the environmental conditions of the year.

These oils showed a high content for PUFAs (linoleic and linolenic acids), intermediate for MUFAs (oleic acid) and low for saturated FAs (palmitic and stearic acids) as in other walnut oils (Amaral et al., 2003; Savage et al., 1999). The ranges of variability for the main FAs, as shown in Table 1, were different for the two years, but similar to those already reported in other walnut oils (Amaral et al., 2003; Martinez et al., 2006; Savage et al., 1999; Bada et al., 2010). The highest values of the walnut oil were found for PUFAs: the FAs were reported as the most effective in reducing low-density-lipoprotein cholesterol in humans, although high linoleic and linolenic acid contents may result in a poor oxidative stability and shorter shelf-life of the oils (Abbey et al., 1994). However, the results of Savage et al. (1999) showed that the most unstable walnut oils, extracted from ‘Esterhazy’ and G139, did not have higher levels of linoleic and linolenic acids than the more stable oils from ‘Vina’, G120, and ‘Stanley’. In addition, MUFAs were as effective as PUFAs in the reduction of low-density-lipoprotein cholesterol in humans (Mensik and Katan, 1989), and the most important MUFA in walnuts is oleic acid (Amaral et al., 2003). Therefore, the composition of the oil from these local Moroccan walnuts points out that these seedlings may be of great benefit to health and nutrition in the human diet due to their fatty acid profile.

The effect of genotype, year, and the interaction year×genotype were significant for all acids, except for the year effect for palmitic acid (Table 2). The year effect had been significant for all fatty acids, except for stearic and linoleic acids, in an Argentinian study (Martinez et al., 2006). The variability in the FA composition has been well studied in walnuts (Amaral et al., 2003; Martinez et al., 2006; Bada et al., 2010), concluding that this composition depended primarily on the genotype, with significant effects of the year, the environment, the geographical origin, and agricultural practices. In our study, the highest contents of linoleic and linolenic acids were obtained in 2009 and, conversely, that of oleic acid in 2010 (Table 1). These seedlings were not irrigated, and water stress may significantly reduce the PUFAs, especially linoleic and linolenic acids (Amaral et al., 2003). The climatic conditions during the period of fruit growth and kernel filling (May to October) were more humid and cooler in 2010 than in 2009 (data not shown). A general negative correlation of PUFAs and growing temperature had been established: PUFAs increased in the membrane as well as in the seed storage lipids with decreasing temperatures (Rennie and Tanner, 1989). Nitrogen fertilization reduced the content of oleic and linolenic acids, and increased that of linoleic acid (Verardo et al., 2013), but our seedlings were not fertilized. The negative correlation between the contents of oleic and linoleic acids (r=−0.58 in 2009 and r=−0.77 in 2010) and of oleic and linolenic acids (r=−0.68 in 2009 and r=−0.8 in 2010) may explain the high value of linoleic acid in 2010. Similar correlations had been previously reported (Malvolti et al., 2010; Martinez and Maestri, 2008).

The significant year×genotype interaction for all acids (Table 2) indicated a different genotype behavior in relation to the climatic conditions, as already pointed out for other cultivars and climatic conditions (Amaral et al., 2003; Martinez et al., 2006). The genotypes Asni-6 and Asni-10 showed consistently high oleic acid contents over the two years (Table 1). In the walnut, the FA metabolism is controlled by a large number of diverse genes (Dandekar et al., 2005) that may react differently to stress, since a plant is a biological entity and does not behave mechanically every year. Therefore, the evaluation of the FAs over several years in the same environment is an important criterion for identification of stable genotypes to establish commercial orchards in order to produce walnuts without fluctuation in these important quality parameters.

The tocopherol profile showed that γ-tocopherol was the major homologue, followed by α- and δ-tocopherol (Table 3), as in other genotypes (Amaral et al., 2005; Bada et al., 2010). The three tocopherol homologues showed high variability among the genotypes (Table 4), as already observed in nine cultivars, where tocopherol content was also affected by environmental factors (Amaral et al., 2005), although few significant differences were observed for the individual tocopherol homologues in the other three cultivars, where the crop year was the main variability source (Martinez et al., 2006). The mean value over the two years ranged between 188.1 and 230.7 mg·kg⁻¹ oil for γ-tocopherol (Table 3). This range agreed with previous results (Amaral et al., 2005), but the maximum values of the Moroccan seedlings were lower than those previously reported (Savage et al., 1999; Bada et al., 2010). The concentration of δ-tocopherol ranged from 23.3 to 43.4 mg·kg⁻¹ oil (Table 3), higher than in a previous report (Amaral et al., 2005), but lower than in others (Savage et al., 1999; Bada et al., 2010). The range in variability for α-tocopherol was between 8.9 and 16.57 mg·kg⁻¹ oil (Table 3), similar to that of a previous report (Amaral et al., 2005), but lower than in others (Savage et al., 1999; Bada et al., 2010). These results confirm that γ- and δ-tocopherols are the most variable homologues in walnuts.

In both years γ-tocopherol showed positive and significant correlations with oil content (r=0.66 in 2009 and r=0.71 in 2010), with linoleic acid (r=0.81 in 2009 and r=0.62 in 2010), and with linolenic acid (r=0.79 in 2009 and r=0.57 in 2010), and a negative correlation with oleic acid (r=−0.58 in 2009 and r=−0.68 in 2010), agreeing with previous studies (Malvolti et al., 2010). As a consequence, genotypes with high oil contents also show high concentrations of γ-tocopherol and of linoleic and linolenic acids. Thus, the amount of oil content could be a reliable index for the amount of PUFAs in walnut oil, and consequently a reduction in time and cost of evaluating walnut oil at large scale.

The tocopherol concentration in walnuts is affected by climatic conditions (Martinez et al., 2006; Amaral et al., 2005) and geographical origin (Amaral et al., 2005). The effect of year was statistically significant for all tocopherol homologues (Table 4). The mean value of all tocopherol homologues was higher in 2009 than in 2010 (Table 3). Tocopherol levels increase in response to a variety of abiotic stresses, considered as evidence of its protective role (Munné-Bosch and Alegre, 2002). The climatic conditions of the year, mainly temperature, affect the concentration of the different tocopherol homologhes in several nut crops (Maranz and Wiesman, 2004), indicating that these components depend on the temperature and the occurrence of drought during fruit ripening. Since heat and drought stresses were higher in the 2009 fruit growing season than in 2010 (data not shown), they could increase the concentration of all tocopherol homologues in these non-irrigated genotypes. The year×genotype interaction for all tocopherol homologues (Table 4) indicated a different genotype behavior in relation to the climatic conditions, as already pointed out (Martinez et al., 2006). The genotypes Asni-3, Asni-4 and Asni-5 showed high and consistent values for γ-tocopherol and oil contents in both years. Taking into account the information on the protective function of γ-tocopherol (Verardo et al., 2009) and the relevance of high lipid contents as a source of carbon and energy during germination and seedling growth (Chenevard et al., 1994), these genotypes could be considered as a seed source for walnut propagation in the high Atlas Mountains in Morocco to recover from forest degradation, since the choice of the seed source is considered crucial for the success of future plantings in the silvicultural management (Hemery, 2008; Callaham, 1994).