Peach (P) and nectarine (N) fruits were collected from trees growing wild in two Tunisian regions, Morneg (Mg) and Gabes (Gb), in July 2014. Morneg is located in the area of Ben Arous in the northeast of Tunisia, at 36°44’N latitude and 10°13’E longitude. The altitude above sea level is 12 meters. The station of Gabes is located in the southeast of Tunisia, at 33°52’N latitude and 10°05’E longitude. The altitude sea level is nine meters. From each station, fruits from each variety were collected, the mesocarp and stony endocarp were separated, and the kernels from the endocarps were mixed and placed in an oven at 60 °C until dried and kept at a constant weight.

The dried samples were reduced in powder using a manual mortar. The oils were extracted by a Soxhlet apparatus for four hours using petroleum ether as solvent. This was evaporated under reduced pressure, using a rotary evaporator at 50 °C. The obtained oil was dried with a stream of nitrogen, weighed and stored in dark glass bottles at 4 °C until analysis. Samples from each crop were taken in triplicate.

The lipid fractions were determined by dissolving 6 mg of oil in 500µL of chloroform. The resulting solution was fractionated in a Lichrolut 0.5 g silica gel cartridge (Merck) using a vacuum manifold and then equilibrated with 2 mL of chloroform (Nash and Frankel, 1986). The solution of total lipids was loaded onto the column which was then washed with another 15 mL of chloroform to elute neutral lipids from the column. Subsequently, the column was washed with 10 mL of methanol to recover the polar lipids quantitatively.

The fatty acids composition was determined by derivatizing 6 mg of the oil fractions to their corresponding fatty acid methyl esters for 1 h at 80 °C with 2 mL methanol/toluene/sulphuric acid (85/15/2.5, v/v/v) (Garcés and Mancha, 1993). Fatty acid methyl esters were extracted with 2 mL heptane and analyzed by gas chromatography (GC) on an Agilent 6890 gas chromatography system (Palo Alto, CA) according to the method used by Bootello et al. (2016).

The analysis of TAGs of the different oil fractions was carried out by injecting 1μL aliquots of 5 mg of oil dissolved in 1.8 mL of heptane into an Agilent 7890 gas chromatograph (Palo Alto, CA) using the same chromatographic conditions as reported by Bootello et al. (2016).

For the positional analysis of TAG sn-2 fatty acids, 10 mg of purified TAGs were hydrolyzed with 2 mg of pancreatic lipase in 1 mL of a 1 M Tris-HCl buffer (pH8), 0.1 mL CaCl₂ (22%), and 0.25 mL deoxycholate (0.1%). The reaction was stopped when approximately 60% of the TAGs were hydrolyzed (1–2 min) by adding 0.5 mL of 6 N HCl. The lipids were extracted three times with 1.5-mL aliquots of ethyl ether, and the reaction products were separated by TLC. Free fatty acids and sn-2-monoacylglycerol bands representing the positions sn-1,3 and sn-2 of TAGs were scraped off the plate, trans-methylated and analyzed by GC. The validity of the procedure was confirmed by comparing the fatty acids composition of the original TAGs and those remaining after the partial hydrolysis. The distribution of saturated fatty acids between the sn-1 and sn-3 external positions of TAGs was calculated using the coefficient of asymmetry α as the quotient between subclasses Saturated-Unsaturated-Saturated and Saturated-Unsaturated-Unsaturated (α SUS/SUU) (Martínez-Force et al., 2009). Thus, α can range between 0 and 0.5; α = 0.5 indicates a symmetrical distribution of saturated fatty acids in accordance with the Vander Wal theory (Vander Wal, 1960).

The melting and crystallization profiles of the different oils were determined by differential scanning calorimetry (DSC) using a Q2000 V23.5 scanner (TA instruments, New Castle, DE, USA). The results were processed using the TA analysis software provided by the manufacturer. This instrument was calibrated prior to use with indium, azobenzene, and undecane, purchased from Sigma-Aldrich (Madrid, Spain). Nitrogen was used to purge the system. Samples were prepared by pipetting 6-8 mg of the oils into aluminium pans, and weighing them using a Sartorius M2P electronic microbalance (Sartorius AG, Goettingen, Germany). The pans were then sealed and balanced calorimetrically, using an empty sealed capsule as the reference. The melting profile of the oils was determined by heating the samples at 90 °C and holding for 5 min (in order to delete the thermal memory), then cooling to -80 °C at a rate of 10 ºC/min and holding for 20 min. Finally, a ramp of 5º C/min up to 90 °C was applied to obtain data from the melting thermogram. The crystallization profile was obtained by completely melting the oils at 90 °C for 5 min, and decreasing the temperature to -80 °C at a rate of 10 °C /min.

The study of the oil content (% dry weight) of the two varieties of P. persica in the two locations, Morneg (Mg) and Gabes (Gb), revealed slight differences between the samples. Depending on their growth location, an oil content ranging from 51.4±0.2% for nectarine to 53.7±0.3% for peach was found in the Mg cultivar; and 49.4±0.1% and 50.5±0.3%, respectively, in the Gb cultivar. In both locations, the peach kernels exhibited the highest levels of oil content. As shown in Table 1, these results are within the values previously reported in the literature for the oil content of peach and nectarine dried kernels, ranging from 42.2% in Canadian fruits (Kamel and Kakuda, 1992) to 54.5% in Egyptian fruits (Rahma and Abd El-Aal, 1988). Furthermore, the oil contents were comparable with those of other oil crops of commercial interest, such as sunflower and rapeseed. The silica column fractionation of kernel oils from both P. Persica varieties showed that more than 98% of total lipids were in the form of neutral lipids and only a very small amount corresponded to the polar fraction.

The fatty acids profile, total saturated fatty acids (SFA), monounsaturated fatty acids (MUFA) and polyunsaturated fatty acids (PUFA) of nectarine and peach kernel oils can be found in Table 2. Six fatty acid species were identified in all the samples. Qualitatively, the fatty acids compositions in both varieties and provenances were similar. Both kernel oils are characterized by the predominance of MUFA, reaching its highest level (76.6%) for those grown in Mg. Oleic acid was the major FA in all the samples, accounting for 67.7% (N Gb) to 75.0% (N Mg), followed by linoleic acid and saturated fatty acids (palmitic and stearic). Thus, the location of cultivars influenced the fatty acids profile of P. persica kernel: nectarine and peach kernel oils from Mg contained more oleic acid than kernel oils from Gb. By contrast, higher levels of linoleic acid and palmitic acid were found for nectarine and peach from the Gb station.

In fact, these samples of kernel oils showed levels of oleic acid similar to high oleic sunflower oil (87.2%). Thus, P. persica kernel oils can be considered as high oleic oils. These results are in agreement with those reported in the literature (Table 1) such as Spanish P. persica kernels that showed a high content of oleic acid (74.6%), followed by linoleic acid (15.7%), and palmitic acid (6.0%) (Sánchez-Vicente et al., 2009).

The analysis of the TAG composition showed variability according to the species and locations of the studied cultivars (Table 3). Fourteen molecular species of TAG were detected, with the predominant ones being those containing oleic, linoleic and palmitic acids. The results showed that OOO (triolein) was the main TAG species, followed by OOL, POO, OLL and POL in all P. persica kernel oils. Regarding the influence of the provenance, Table 3 shows that the nectarine oils grown in Mg exhibited the highest value of OOO and POO, according to the increasing oleic acid levels. When comparing both P. persica varieties, peaches vs nectarines, it was found that in the Mg location the peach kernel oil showed the highest content of TAGs species OOL, POO, OLL and POL compared to nectarine, which showed only an increase in OOO. For the Gb location, smaller differences were detected due to the very similar fatty acid compositions of their oils (Table 2). As observed in Table 3, P. persica kernel oils presented some similarities with high oleic sunflower oil, being OOO (69.2%) and POO (9.2%) the main TAG species. The TAG compositions of the P. persica oils were also similar to some rosaceae kernel oils, such as apricot (P. armeniaca), plum (P. domestica) and peach (P. persica) (Hassanein, 1999), which showed average contents of OOO (40.7%) and OOL (24.8%) as the major TAGs components.

The contents of the TAG subclasses di-saturated (SUS), mono-saturated (SUU), and tri-unsaturated (UUU) were obtained from the compositions of the TAG species and the coefficient α (Martínez-Force et al., 2009) was used to calculate the asymmetry of saturated fatty acids between the sn-1 and sn-3 TAG positions (Table 3). The coefficient α varied between 0.10 and 0.12 (P Mg and N Mg, respectively) and between 10 and 12% saturated fatty acids were found in one external position and between 88 and 90% in the other position. This asymmetry is due to the enzymatic machinery responsible for the esterification of saturated fatty acids in the external positions, mainly glycerol-3-P acyltransferase and diacylglycerol acyltransferase. These results are similar to those reported for hazelnut and walnut oils, which exhibited low α coefficients (0.17 and 0.04, respectively) and, therefore present a high asymmetry in their saturated fatty acids. However, the studied P. persica oils were quite different from soybean, high-stearic sunflower or rice which showed a higher α coefficient value (0.29, 0.33, and 0.49, respectively), corresponding to more symmetrical distributions of saturated fatty acids in their TAG species (Martínez-Force et al., 2004). The distribution of fatty acids in the three positions of TAG molecules has been studied in P. persica oils in the two locations (Table 4). These results confirmed that saturated fatty acid (palmitic and stearic acid) were mainly distributed in the external position (sn-1,3) of the TAG backbone; whereas the sn-2 position mostly contained unsaturated fatty acids, of which more than 80% corresponding to oleic acid. These data, together with the low values of α coefficients can explain the strong asymmetrical distribution of the fatty acids, and hence, the absence of di-saturated species of TAGs (SUS) in these oils. Therefore, Prunus persica oils are composed exclusively of TAG species with two or more unsaturated fatty acids in their molecules (SUU and UUU), something that would be reflected in their thermal properties.

The DSC thermograms of P. persica kernel oils can be found in Figure 1: heating curves or melting profiles (Figure 1A) and cooling curves or crystallization profiles (Figure 1B). In addition, Table 5 provides thermal parameters obtained from the thermograms depicted in Figure 1. Heating curves from -80 to 80 ºC showed the melting profiles with an endothermic phase transition with a single and well-defined peak. The melting peaks in the DSC thermograms corresponding to P. Persica kernel oils exhibited an onset temperature with values between -46.9 to -33.6 ºC for the nectarine Gb and peach Gb varieties, respectively. High oleic sunflower oil started to melt at higher temperature (-21.5 ºC). The melting temperature for the peak (P_m) showed values from -14.0 ºC (nectarine Gb) to -9.4 ºC (high oleic sunflower oil). For the temperature of end melting, it showed values ranging between -3.3 and 4.7 ºC for the P Gb and N Mg varieties, respectively. Again, high oleic sunflower exhibited a higher temperature for end melting (5.2 ºC). In general, it can be observed that for varieties with higher PUFA contents, lower melting temperatures were obtained. The range of transition phase (R) for the melting curves (temperature difference between T_{onset m} and T_{end m}) was higher for kernel oils compared to high oleic sunflower oil. Kernel oils contained higher amounts of the tri-unsaturared TAG species OOL and LLL, and of the mono-saturated TAG species POL and PLL. Since these TAGs species exhibit lower melting points than other mono-saturated (StOO, OOA and OOB) and di-saturated (POSt and StOA) TAGs species, only present in high oleic sunflower oil, the melting range of kernel oil was lower. Regarding the melting enthalpy, kernel oils showed lower values compared to high oleic sunflower oil. As reported in the literature, lower melting enthalpy values indicate weaker crystalline structures. Thus, less energy would have to be absorbed to destroy the crystal network (Willie and Luton, 1966).

Cooling curves from 80 to -80 ºC, showed the crystallization profiles corresponding to an exothermic event with a single and well-defined peak for both kernel and sunflower oils. For kernel oils, the crystallization peak in the DSC thermograms showed an initial temperature of crystallization (T _{onset c}) with values from -30.2 to -26.2 ºC, whereas the peak for temperature crystallization (P_c) reached values between -51.1 and -46.2 ºC. High oleic sunflower oil exhibited higher values for T_{onset c} (-16.8 ºC) and P_c (-41 ºC). These differences for peak maximum and crystallization temperature ranges may be attributed to the higher levels of long chain saturated fatty acids (arachidic and behenic acids) and the SUS type TAGs present in high oleic sunflower oil. These di-saturated TAGs species crystallize at temperatures slightly higher than SUU and UUU TAGs, so for high oleic sunflower oil the crystallization took place at a higher interval of temperature than in kernel oils. The estimation of solid content (Figure 2) was calculated by integration of the melting thermograms using the analysis software provided by the DSC manufacturer. High oleic sunflower oil displayed a higher solid content and a sharper curve compared to kernel oils. So, at -20 ºC, kernel oils containing less OOO, displayed lower contents of solids (about 55% and 70% of solid content for N Gb and P Mg, respectively), whereas high oleic sunflower oil was 100% solid at that temperature. At -10 ºC, all kernel oils were melted except for N Mg, whose solid content was close to 20%. The kernel oil with the highest OOO content was N Mg, and was completely melted around 0 ºC (similar to high oleic sunflower oil).

In summary, the melting and crystallization behavior of both P. persica varieties from different locations showed similar shapes with a unique peak and DSC parameters with different values depending on the variety and provenance (Table 5). The small variations observed in the DSC profiles are in good agreement with the changes in TAG species found between varieties and locations (Table 3). Although the major peak observed for all samples was mainly due to the triolein content, the higher levels of linoleic acid present in the P. persica kernel oils also contributed to modifying their thermal behaviour compared with high oleic sunflower oil.