Valorization of apricot, melon, and watermelon by-products by extracting vegetable oils from their seeds and formulating margarine ☆

– Vegetable oils extracted from seeds and oleaginous fruits are a substantial source of bioactive compounds. In this study, oils of some fruit by-products were investigated and their composition and properties were compared. Apricot (ASO), melon (MSO), and watermelon (WSO) seed oils were extracted by cold pressing. The physico-chemical parameters and the contents of pigments were assessed using standard methods. The values of the physico-chemical parameters revealed the purity of the oils and it was recorded that the WSO has the best contents of chlorophylls and carotenoids, which were 12.43± 0.71 mg/kg of oil, and 1.35 ± 0.02 mg equivalent of b -carotene/g oil, respectively. In addition, the oils were analyzed by gaschromatographyandtheirmajorfattyacidswerelinoleic,oleic,palmitic,andstearic.TheASOrevealedthe highestantioxidantactivityinthequenchingof1,1-diphenyl-2-picrylhydrazyl(DPPH)withinhibition percentageof89.2±2.3after30minutesofcontact.Likewise,theoilswereexploredfortheforti ﬁ cation of margarine. The physicochemical parameters of the formulated margarines comply with the standards. The Rancimat test showed that the highest induction time (16.54 h) was assigned to margarine enriched with 150 m g/g of WSO. Hence, this oil can has numerous applications in other food industries.


Introduction
Currently, the generation of food wastes is a critical difficulty, and their exploitation as food components for the elaboration of novel functional foods is necessary since customers request for foods that include ingredients with improved nutritional characteristics has augmented (Silva et al., 2019). The usage of agri-industry wastes to conceive and elaborate novel foods with supplemented benefit is very critical for sustainability, considering this as resolution to diminish food losing, and ecological influence (Silva et al., 2020).
Apricot (Prunus armeniaca L.) from Rosaceae family is prominent fruit since it is delightful and nutritive. Its seed (kernel) is a precious waste generated after consumption and processing (Al-Juhaimi et al., 2021). Melon (Cucumis melo L.), and watermelon (Citrullus lanatus L.) from the Cucurbitaceae family are a famous fruits, consumed worldwide, that comprise considerable quantities of seeds (Petchsomrit et al., 2020;Rabadán et al., 2020).
The by-products of the selected fruits are an excellence source of oils with interesting composition and biological properties. Previous studies carried out on some of these oils confirmed the interest in use them as a possible functional ingredient (Siddeeg and Xia, 2015;Ramadan, 2019;Silva et al., 2022).
Apricot seeds possess a considerable oil yield and this oil is rich in unsaturated fatty acids, sterols, tocochromanols and squalene (Ramadan, 2019). Additionally, ASO contains phenolic substances that play a favourable role in regards to the oxidative stability of oil and to the antioxidant capacity and health benefiting activities as cardiovascular diseases, cancer, tumors and ulcers (Ramadan, 2019;Al-Juhaimi et al., 2021).
With regards with MSO, it have an appreciable quantity of sterols, tocopherols, and phospholipids which reinforce the relevance with its advantageous impact on the population where it assists in diminishing the hazard of cardiovascular pathologies by inhibiting the oxidation of polyunsaturated fatty acids (PUFA) (Ramadan, 2019).
As for WSO, it was stated as an excellent origin of essential fatty acids, carotenoids, tocopherols, thiamine, and phenolic compounds. Furthermore, WSO exhibited different biological properties mainly antioxidant, anti-inflammatory, cardioprotective and antimicrobial ones (Petchsomrit et al., 2020).
Algeria possesses a broad diversity of flora, which can provide oils with various bioactivities. Interestingly, Algeria is among the top five apricot manufacturers worldwide (Ramadan, 2019). Nevertheless, there is few data on the composition and exploitation of the oil seeds.
Controlling oxidation is essential to manage the evolution of biological systems in their complexity, in particular in the case of foods whose degradation can have consequences for food safety. Among these food products, margarine, a plastic emulsion consisting essentially of two phases namely fatty and aqueous, also contains 2% of water and fat-soluble additives. Its composition is represented at 82% by a mixture of oils: the first target of oxidation (Fruehwirth et al., 2021).
Lipid oxidation is a substantial reason of degradation of margarine over its fabrication that is mainly perceptible by the appearance development of disagreeable fragrance. This does not just denote oxidative deterioration products with possibly toxic features but conduct to the non-acceptance of the merchandises by customers (Himed and Barkat, 2014;Fruehwirth et al., 2021). So, to ensure a long shelf life of dietary products such as margarine, natural antioxidants are widely employed. This study focuses beforehand on the comparison of the physico-chemical properties, the fatty acid composition, and bioactivity of ASO, MSO and WSO then on the impact of the substitution of the synthetic additive by these oils, as well as the determination of their capacity to protect margarines from accelerated peroxidation.
There are a number of papers communicating the fortification of margarine by natural products (Himed and Barkat, 2014;Chougui et al., 2015;Kaanin-Boudraa et al., 2021). However, the except of some reports dealed with the valorization of ASO, MSO and WSO by their use in other food products apart from margarine, papers describing the elaboration of functional foods using these oil by-products remain scarce. To the best of our knowledge, there are no references in the literature reporting the enrichment and preservation of margarine from oxidative deterioration using the chosen vegetable oils.

Chemicals and materials
All chemicals were acquired from Biochem Chemopharma (Montreal, Canada), Sigma-Aldrich (St. Louis, MO, USA) or VWR International GmbH (Vienna, Austria). All used solvents are of analytical grade.
Seeds of apricot (Prunus armeniaca L.) of variety "pêche de Nancy", of melon (Cucumis melo L.) of variety "inodorus", and of watermelon (Citrullus lanatus L.) of variety "grey bell" were used. These oilseed plants were grown in Northeast Algeria in the Bejaia department.

Oils extraction
Oils were extracted from sampled seeds by the cold pressing, which is achieved by directly pressing raw/dried seeds on a continuous screw press at low temperature. After 24-hour sedimentation, the extracted oil was separated from the sediment by decantation. Then, the oils were stored in dark glass bottles at a temperature of 4°C, and then subjected to analyses (Brahmi et al., 2020).

Determination of oil parameters
Physical characteristics of the oils were determined by measuring the refractive index using a refractometer RL 4 (PZO, Poland) at 40°C. Absorbances in UV were measured by spectrophotometer (UV-Vis Spectrophotometer, spectro scan 50 Shimadzu, Kyoto, Japan) at 232 and 270 nm, respectively. Density of the oils was measured with respect to water density at a temperature of 20°C.
Quality parameters (peroxide value (PV) and acid value (AV)) of cold-pressed oil were assessed according to the methods described by Duru et al. (2019).

Determination of chlorophyll and carotenoid contents
Measurement of the content of chlorophyll was performed according to AOCS method as described in our previous study (Brahmi et al., 2020) and its content was expressed in mg of chlorophyll/kg of oil.
Carotenoid content in the oils was determined by using calibration curve of b-carotene (0.1-5 mg/mL) dissolved in nhexane and recording the absorbance at 440 nm. N-Hexane was used for oil dilution to obtain the absorbances in a range of the calibration curve. The carotenoid contents were expressed in mg equivalent of b-carotene/g oil (Brahmi et al., 2020).

Determination of the radical scavenging activity of the oils
Radical scavenging activity of vegetable oils was assessed by reduction of DPPH radical in toluene as reported by Ramadan and Moersel (2006). Toluenic solution of DPPH • was freshly prepared at a concentration of 10 À4 M.
For determination, 0.1 mg of each oil/mL toluene was mixed with 390 mL toluenic solution of DPPH • and the mixture was vortexed for 10 s at ambient temperature (25°C). Against a blank of pure toluene without DPPH, the decrease in absorption at 515 nm was measured after 1, 30 and 60 min of mixing using UV-Vis Spectrophotometer (spectro scan 50, Shimadzu, Kyoto, Japan). Antiradical action toward DPPH • was estimated from the difference in absorbance with or without sample (control) and the percent of inhibition was calculated from the following equation: Absorbance control Â 100:

Determination of fatty acid composition of the oils
The analysis was carried out in the technical platform for physico-chemical analyzes of Bejaia, Algeria. The fatty acids were determined according to the method adopted by IOC with slight modifications. This method permits the determination of fatty acid methyl esters (FAMEs) from C12 to C24. In fact, the slight medications that we mentioned in this paper concern some operating conditions including the used gas carrier and detector temperature. In this regard, the gas carrier used in the IOC method is hydrogen although we used nitrogen, which is also an inert gas. In addition, the temperature of the FID detector was 270°C although it was 250°C in the IOC method. 0.1 g of the oil was purified through Silica-SPE cartridge then eluted with an admixture hexane: ether (50:50, v/v). The collected fraction was evaporated then dissolved in heptane. Triglycerides hydrolysis and transesterification, in basic medium, into fatty acids methyl esters (FAMEs) were performed by adding 0.5 mL of 2 N methanolic KOH. The upper part containing the FAMEs was recovered for GC injection. The fatty acids determination was performed with a Shimadzu-Nexis GC-2030 equipped with a flame ionization detector (FID). 1 mL of analytes were injected then separated on a high polar column Supelco SP-2380 (poly 90% biscyanopropyl-10% cyanopropylphenyl siloxane), 60 m length Â 0.25 mm internal diameter Â 0.20 mm film thickness (Sigma-Aldrich Co. LLC, St. Louis, MO, USA). A flow rate of 1.5 mL/min of nitrogen as a carrier gas and a split injection (50:1 split ratio) were applied. An oven program temperature had begun at 165°C for 8 min then risen at 2°C/min to reach a final temperature of 210°C. Consequently, the run time was 30.5 min. The auxiliary gases were hydrogen at 32 mL/min, air at 200 mL/min, and nitrogen as a makeup gas at 24 mL/min. The identification was carried out using a standard chromatogram including FAMEs from C12 to C24 and using the chromatogram standard reported in the IOC referential method by comparing the obtained retention times (RT) to that reported in IOC. Each fatty acid was determined in percentage (%) represented by the area of the corresponding peak relative to the sum of the areas of all the peaks. The analysis was performed in triplicate for all samples from purification to injection.

Preparation of margarine samples
Oils at 50, 100, and 150 mg/g were employed to prepare three samples of margarine at CEVITAL agri-food industry (Bejaia, Algeria). A control margarine containing no oils but vitamin E was also formulated. After weighing the two phases and their ingredients, the mixture was poured into a stainless container where the emulsification took place with stirring for 20 min. At this point, the stability of the emulsion was incomplete; it had to undergo crystallization, which was carried out in a vessel containing ice-cold water. Stirring was carried out until a homogeneous margarine was obtained. The produced margarines were packaged in trays of 250 g each and stored at 6°C.
After that, the margarines have been analyzed by determining humidity, pH, salt content, and melting point (Kaanin-Boudraa et al., 2021).

Oxidative stability determined by Rancimat method
The oxidative stability of the margarines was evaluated by Rancimat method as reported in our previous work (Kaanin-Boudraa et al., 2021). The oxidation induction time (OIT, in hours) was determined with the Rancimat apparatus (Metrohm 743, Herisau, Switzerland). For all analyses, 20 g of margarine were melted in an oven (DRY-Line ® , VWR International GmbH) at 80°C for 25 min. Then, Rancimat vessels containing 3 g margarine were used for the analysis and an air rate of 10 L/h was applied. The OIT was evaluated using a temperature of 120°C.

Statistical analysis
Results were presented as means ± standard deviation from three replicates of each experiment. A significance level P < 0.05 was employed to denote significant differences between mean values determined by the analysis of variance, post-hock Tukey tests (ANOVA) using Statistica 10 software.

Physicochemical and quality parameters of the oils
The physicochemical and quality parameters of oils are illustrated in the Table 1. In addition to the characteristics listed in the table, the determination of the color of oils is a common and crucial factor for the production of margarine. Visual inspection of the extracted oils revealed that they had more or less dark yellow color; this indicates the occurrence of yellow pigments such as carotenoid compounds (Duru et al., 2019), which can be beneficial for margarine fortification since they can act as natural antioxidants.
Peroxide value and free fatty acids are amongst the most prominent properties to assess the quality of edible oil samples. Acid value (AV) is an indication of liberated fatty acids by oil chemical breakdown but peroxide value (PV) is employed to measure the compounds generated from primary oxidation (Ok and Yilmaz, 2019). Indeed, the elevated AV is linked to the appearance of Free Fatty Acids (FFAs) that are more vulnerable to oxidation comparatively to the fatty acids found in the triacylglycerols (Rezig et al., 2019).
Discrepancies were noticed among the studied oils in their acidity and peroxide values (Tab. 1). ASO had high value of acidity (4.40 ± 0.21 mg/g oil) followed by the WSO (1.10 ± 0.01 mg/g oil) and, MSO (0.80 ± 0.02 mg/g oil).
The extent of edibility of a fat is usually perceived to be conversely proportionate to the total level of FFAs. The low acidity of the investigated oils revealed that they are comestible and might have a prolonged storage life.
The PV of the investigated oils was in the range of 0.20-0.26 meq O 2 /kg of oil. According to the Codex Alimentarius Commission, the peroxide value should not exceed 10 meq of peroxide oxygen/kg of oil.
The values revealed in this investigation are lower than those found in the literature for the ASO, MSO and WSO from different origins. PV of different apricot varieties from Pakistan ranged from 1.0 to 2.32 meq O 2 /kg (Manzoor et al., 2012). MSO and WSO are among the cucurbit seed oils, according to Mariod et al. (2009) these oils have PV, which varied from 2.3 to 4.1 meq O 2 /kg. Melon grown in region of Plovdiv, southern Bulgaria provided seed oils with substantially elevated AV (1.5-2.1 mg KOH/g), however with bass PV (1.1-3.4 meq O 2 /kg (Petkova and Antova, 2015). Seed oils extracted from Cucumis melo L. growing in west Algeria possess much higher acidity and peroxide values compared to those found in this present study which are respectively 4.01 mg KOH/g and 2.25 meq O 2 /kg (Mulengi et al., 2016). Oil from watermelon from Romania showed a moderate AV of 1.9 mg KOH/g and very high PV of 7.5 meq O 2 /kg (Dumitru and Tutunea, 2017). AV and PV of seed oils extracted from some Cucurbitaceae including melon and watermelon ranged from 8.95 ± 0.84 to 9.60 ± 0.87 mg KOH/g oil and from 1.04 ± 0.11 to 9.56 ± 0.97 meq O 2 /kg oil, respectively (Rezig et al., 2019). In one study, the AV of the seed oils of melon and watermelon from Nigeria were high and respectively of 3.029 and 3.010 mg KOH/g, likewise the recorded peroxide indices were considerable with values of 26 and 27 meq O 2 /kg, respectively (Duru et al., 2019).
Nevertheless, kernel oil acquired from the seeds of some common sweet and bitter apricots from India were in perfect accordance with those reported in this study. The AV and PV varied from 0.2 (Maraghe) to 0.6 mg KOH per g of oil (Osku bitter), and 0.35 (Maraghe bitter) to 1.9 meq O 2 per kg of oil (Osku bitter), respectively (Shariatifar et al., 2017).
Besides, the AVof WSO found in this current study is in the range of the contents of the oils obtained from the two varieties of Moroccan watermelon seeds using various extraction methods (1.40 ± 0.03 and 2.80 ± 0.06 mg KOH/g oil). Nevertheless, these authors observed greater PV (6.00 ± 0.08, 4.0 ± 0.67 and 3.80 ± 0.20 meq O 2 /kg oil) (Ouassor et al., 2020).
Vegetable oils density is conditioned by their composition in fatty acids, minor constituents and temperature (Neagu et al., 2013). The density values recorded in this study were For ASO, the density value was lower than that established by Manzoor et al. (2012) which ranges between 0.87 and 0.93 mg/mL. The density of MSO is also lower compared to that found by Mulengi et al. (2016) which was 0.897. The refractive index increases with unsaturation and the presence on fatty chains of secondary functions. So, a high refractive index allows concluding the occurrence of double bonds. In this current study, the measured refractive indexes were 1.4638, 1.4666, and 1.4668 for ASO, MSO and WMS, respectively.
Results of this study concur with the literature. By assessing the physico-chemical properties of seed oils extracted from various apricot varieties from Pakistan, Manzoor et al. (2012) recorded refractive index values ranged from 1.4655 to 1.4790. With a view to improve the effectiveness of oil extraction from wild apricot kernels from India by employing enzymes, refractive index fluctuate from 1.468 to 1.471 (Bisht et al., 2015). Gayas et al. (2020) demonstrated that the refractive index was almost similar for all extraction methods studied for apricot kernels oil. It was 1.47 ± 0.00 using mechanical extraction, 1.45 ± 0.00 using Soxhlet and 1.46 ± 0.00 using ultrasound assisted.
The refractive indices and relative densities of the cucurbit seed oils ranged from 1.334-1.442 and 0.874-0.920 g/cm 3 , respectively (Mariod et al., 2009). Refractive index of MSO from Algeria was 1.470 (Mulengi et al., 2016). The seeds oil of watermelon from Romania showed density of 945 and refraction index of 1.4731 (Dumitru and Tutunea, 2017). Refractive index value of the cold-pressed watermelon seed oil was 1.4696 (Ok and Yilmaz, 2019) and those extracted with n-hexane by Soxhlet has refractive index of 0.998 (Oragwu, 2020). The refractive indices were 1.46 and 1.47 Brix for MSO and WSO, respectively (Duru et al., 2019).
Oxidation of unsaturated fatty acids leads to the formation of conjugated fatty acids, which could absorb UV light at 232 and 270 nm. Linoleic hydroperoxide can be measured at 232 nm and diketones and unsaturated ketones which can be quantified at 270 nm. So, the specific extinctions at 232 nm and 270 nm of oils can indicate their oxidation state (Loukou et al., 2013). WSO possessed the highest K232 and K270 of 0.3 and 0.2930, respectively. The seed oil of the melon has an extinction of 0.30 at 232 nm and 0.2901 at 270 nm. For ASO, the K232 nm was 0.2942 and the K270 was 0.0874. These values are much lower than those of seed oils of Pakistan apricot varieties, which were 2.30-3.42 and 0.82-1.04, respectively (Manzoor et al., 2012).
Several factors can have an impact on the parameters of oil as found by studies carried out on ASO, MSO and WMS. Kiralan et al. (2018) demonstrated that the storage period during 12 days under oxidation circumstances affects significantly the quality parameters of apricot kernel cold-pressed oils from Istanbul (Turkey). PV values of AKO achieved the highest value (54.5 meq O 2 /kg) after 10 days of storage. Similarly, the maximum K232 (conjugated dienes) was attained (10.91).
In the study of Bisht et al. (2015), the usage of enzymes demonstrated that the different handling resulted in extracted oil from apricot kernels from India with AV ranging from 3.77 to 5.30 KOH/g. The method of extraction adopted have also importance, the oil of seed apricot grown in Hebei Province (China) parameters were impacted by cold pressing, heat pressing, and refining of sun-dried and baked apricot kernels process. AV were 0.36-1.40 mg KOH/g, PV were 2.09-5.62 mmol O 2 /kg, and absorbance values at 232 and 268 nm were of 0.70-0.85 and 0.20-0.38, respectively (Zhou et al., 2016). Similarly, seed oil extracted from apricot growing in India country using ultrasound method showed AV, which rise from 2.27 to 2.69 mg KOH/g with temperature augmentation (Gayas et al., 2017). Debitterizing caused also the rising of the PV and AVof the ASO and hence affected the oils quality by oxidative rancidity. AV varied from 1.43 ± 0.22 (blanched apricot kernels) to 2.22 ± 0.22 (debitterized kernels (60°C, 6 h) mg KOH/g whereas PV from 3.37 ± 0.14 (blanched apricot kernels) to 4.57 ± 0.25 meq O 2 /kg (debitterized kernels (50°C, 8 h) (Song et al., 2018).
The oil extracted from apricot seeds from India by hexane adopting different extraction methods showed an acidity of 2.71 ± 0.01 mg KOH/g oil by mechanical, 2.86 ± 0.03 mg KOH/g oil by Soxhlet and 2.73 ± 0.01 mg KOH/g oil by ultrasound assisted extraction. Regarding the PV, the oil obtained by mechanical extraction have the lowest PV (5.03 ± 0.02 meq O 2 /kg oil), followed by oil from Soxhlet (5.09 ± 0.01 meq O 2 /kg oil) but the oil extracted by ultrasound assisted technique recorded the highest PV (5.17 ± 0.05 meq O 2 /kg oil) (Gayas et al., 2020). Nevertheless, the oils extracted by SC-CO 2 and cold-pressed process from sweet apricot kernel seed cultivar from Croatia has the same PV (0.96-0.98 meq O 2 /kg oil) (Pavlović et al., 2018).
By investigating the impact of the extraction method on the oil extracted from golden melon seeds from Shanghai (China), the acid and peroxide amounts varied from 0.69-0.79 mg/g and 5.17-5.79 mmol/kg, respectively (Chen et al., 2021).
In the same context, oils from melon of Tomelloso, Spain origin resulted from hydraulic press were of moderately improved quality with acidity of 0.30 ± 0.04% of oleic acid comparatively to oils resulted the screw press whose acidity was 0.41 ± 0.05% of oleic acid (Rabadán et al., 2020).
The generated cold-pressed oils from seeds of watermelon from Mardin (Turkey), that were formerly processed with seed boiling and roasting exhibited almost the same acid value than those analyzed in this study (1.1 ± 0.0 and 1.2 ± 0.1 mg KOH/g oil for roasted and boiled oil seeds, respectively) (Ok and Yilmaz, 2019).
Solvent-extracted seed oil from watermelon of Uli (Nigeria) origin, displayed acidity and peroxide values of 30.80 mg NaOH/g oil, and 10 mg equivalent/g oil, respectively (Oragwu, 2020).
Moreover, ASO oil extracted by three techniques revealed a considerable content of unsaturated fatty acids (linoleic acid and oleic acid) and rather reduced amount of saturated fatty acids (Gayas et al., 2020). Other authors demonstrated always the prevalence of the same fatty acids (palmitic, oleic and linoleic) in ASO (Shariatifar et al., 2017;Jin et al., 2018;Pavlović et al., 2018;Song et al., 2018;Stryjecka et al., 2019).
Similarly, in WSO oil, the predominant fatty acids were palmitic, stearic, oleic and linoleic acids (Dumitru and Tutunea, 2017;Ok and Yilmaz, 2019;Petchsomrit et al., 2020). The same trend was observed by Ouassor et al. (2020) when studying two varieties of watermelon. According to Zarifikhosroshahi and Ergun (Zarifikhosroshahi and Ergun, 2021), the order of importance of fatty acid contents in WSO is as follows: palmitic, oleic, stearic, and linoleic acids. The main fatty acids were linoleic, oleic and palmitic acids for both varieties. However, myristic, linolenic and stearic acids were found in modest quantities. Linoleic, oleic, and palmitic acids in WSO were formerly mentioned to be extended from 45.1 to 76.2%, 0.33-33.66%, and 4.30-16.2%, respectively (Biswas et al., 2017).
Numerous factors can influence the composition and fatty acid content of vegetable oils, among these conditions seed treatment. So, the fatty acid content of apricot kernel oil is dependent on the roasted procedure. Palmitic acid amounts varied from 4.38 (oven-roasted) to 4.76% (microwave roasted); oleic acid amounts were between 65.73% (ovenroasted) and 66.15% (control) and linoleic acid quantities ranged from 26.55 (control) to 27.12% (oven-roasted) (Al- Juhaimi et al., 2021). The microwave power used for roasting apricot kernels affected also the composition of ASO's oils. The fatty acid profiling of kernel roasted at 360, 540 and 720 W were highly changed by roasting procedure comparatively to raw matter (Juhaimi et al., 2018). The solid-state fermentation procedure has induced reductions of the palmitic and stearic Table 2. Fatty acid composition of apricot seed oil (ASO), melon seeds oils (MSO), and watermelon seeds oil (WSO).
The results with different letters for each fatty acid contents indicate a statistically significant difference (P < 0.05, a > b).
acids, and a considerable augmentation in the quantity of linoleic and oleic acids (Dulf et al., 2017). Oleic acid quantities of ASO augment regarding maturation at the harvest periods but palmitic and linoleic acids amounts decline (Matthaus et al., 2016). The contents also depend on the variety, Kabaası variety showed the best content of palmitic acid (6.78%) then Çatalo glu variety (5.87%) (Juhaimi et al., 2018). However, no significant changes were recorded according to the extraction methods (ultrasonic, mechanical and solvent) in ASO (Gayas et al., 2020).
The fatty acid compositions of the MSO obtained by several techniques namely hot-pressing, cold pressing and ultrasound-assisted aqueous enzymatic extraction were distinct (Chen et al., 2021). Oils resulted from the various melon cultivars reveled an elevated fluctuation in the concentrations of linoleic (51-69%) and oleic (15-34%) acids (Rabadán et al., 2020).
The extraction method as well as the variety of watermelon showed an effect on the linoleic and oleic acids level. The best content of linoleic acid was attributed to the oil resulted from Soxhlet (67.43%) for the variety "lanatus". Otherwise, the best oleic acid level (19.08%) was attributed to the oil of variety "citroides" extracted by cold pressing (Ouassor et al., 2020). The method of preservation of watermelon fruit (room temperature, at þ 4°C for 12 months and newly harvested seeds) has shown an effect on the fatty acid contents of the oils of its seeds. The content of linoleic acid was elevated in samples stored at 4°C. Besides, the ratio of polyunsaturated fatty acid to saturated fatty acids was determined as 1.16, 1.20, and 1.27 for the samples stored at room temperature, 4°C, and newly harvested, respectively (Zarifikhosroshahi and Ergun, 2021).

Pigment content of the oils
Chlorophyll and carotenoid pigments are indicators of the quality of seed oils. According to obtained results, watermelon seeds oil gave the highest content of chlorophyll (12.43 ± 0.71 mg/kg oil) while the oil of melon seeds exhibited a moderate amount of 4.63 ± 0.04 mg/kg oil and apricot seeds oil recorded a very low level (0.10 ± 0.01 mg/kg). The same trend was noticed regarding carotenoid contents of the investigate oils. Watermelon seed oil contains the best concentration (1.35 ± 0.02 mg equivalent of b-carotene/g oil) followed by melon seeds oil (0.41 ± 0.01 mg equivalent of b-carotene/g oil), while apricot seeds oil was poor in this pigment (0.12 ± 0.01 mg equivalent of b-carotene/g oil) (Tab. 3).
Chlorophylls are linked to oxidative phenomena by their catalytic actions as pro-oxidants in the presence of light and antioxidants in the dark (Li et al., 2019). Carotenoids are also involved in the mechanisms of auto-oxidation and photooxidation (Ouassor et al., 2020). Indeed, carotenoids are very effective inhibitors of photoxidation induced by chlorophyll pigments (Grati Kammoun et al., 1999).
In the same line of our results, the carotenoids content of the kernels oil of 15 apricot genotypes from India were low and ranged from 1.5 to 5.3 mg/kg of oil (Gayas et al., 2017).
b-carotene levels in seeds of five apricot cultivars grown in Poland ranged from 42.3 to 66.8 mg/g (Stryjecka et al., 2019). Górnas et al. (2017) established in their study that lutein, zeaxanthin, b-cryptoxanthin and b-carotene were the principal (76-94% of the total carotenoids) carotenoids in ASO, they also reported that the compositions are related to the genotypes.
The carotenoid content of the WSO depends on the variety and the method of extraction. Oil obtained by cold pressing from the variety var. citroides exhibited the best amount (73.95 ± 0.20 mg/kg), the oil resulted by Soxhlet from the variety lanatus comes second (24.78 ± 0.14 mg/kg) while the oil extracted using sonotrode ultrasound assisted extraction from citroides var showed the lowest content of 18.31 ± 0.22 mg/kg (Ouassor et al., 2020).

Radical scavenging activity of the oils
The capacity to neutralize the DPPH • by the three oils studied increases as a function of time (Fig. 1). It's obvious that ASO has higher anti-free radical activity, followed by WSO, and MSO gave low effect.
The anti-DPPH • capacity of the oil extract of roasted apricot kernels varied between 2.55% (oven) and 19.34% The results with different letters for each pigment contents indicate a statistically significant difference (P < 0.05, a > b). (microwave-roasted). Additionally, radical scavenging effect increased with roasting (Al- Juhaimi et al., 2021).
The anti-DPPH • activity of the melon seeds oil from Sudan was good with IC 50 of 25.25 mg/mL (Azhari et al., 2014).
WSO at 1 g/mL from Romania exhibited a moderate antiradical activity (46%) (Dumitru and Tutunea, 2017). Ouassor et al. (2020) assessed the antioxidant activity by the free radical DPPH of the methanol fraction of different WSO. They demonstrated that the technique of obtaining the oil and the variety of watermelon considered have a great impact on the activity. The results found oscillate between 51.1 ± 0.1% for the cold-pressed oil sample of var. lanatus and 84.8 ± 0.04% for Soxhlet var. citroides.

Physico-chemical analysis of the formulated margarines
Given the interesting characteristics of the oils explored, they are used to enrich table margarine. The results of the physico-chemical analysis of the formulated margarines as compared to the control are depicted in Table 4.
We noticed that there is no difference between the moisture content of the control margarine (M0) and the three other margarines elaborated (MASO, MMSO, and MWSO) with the vegetable oils studied, the levels were 15.97 ± 0.01, 15.30 ± 0.67, 15.23 ± 0.74, and 15.87 ± 0.10%, respectively. This is compatible with the initial formulation of the margarine, which is constituted of 82-84% of fatty phase and 16-18% of the aqueous phase. The addition of salt to margarine is intended to improve its organoleptic characteristics and also to inhibit the growth of certain bacteria, which prolongs its shelf life. In the present study, the salt content (NaCl) of the three margarines produced were 0.37 ± 0.04, 0.35 ± 0.02 and 0.37 ± 0.04%, respectively. The control margarine had a salt amount of 0.33 ± 0.01%.
The melting point of the formulated margarines were 35.4 ± 0.1, 35.3 ± 0.3 and, 35.1 ± 0.1°C for MMSO, MWSO and MASO without significant difference with M0 (35°C). This characteristic is related to the fatty acid composition, long chain saturated fatty acids have a higher melting point than the short chain unsaturated fatty acids.
As regarding the pH values, which provides information on the state of freshness of the sample, they varied between 4.10 and 4.80. This means that the amounts of acetic acid added to margarines were respected. The pH, between 4 to 5.5, is intended to make margarine taste good and prevent microbial growth (Karleskind and Wolff, 1992).

Oxidative stability of the formulated margarines
The association of considerable oleic and linoleic acid amount in oil is of benefit, as they are regarded stable and nutritious oils, which aid avoids different pathologies (Gayas et al., 2020).
Considering that all the investigated oils were a substantial source of both oleic and linoleic acids, they might be employed as cooking or salad oils or might alike be employed for margarine formulation (Rezig et al., 2019).
Oils are very important in human diet because of the high contents of essential fatty acids and antioxidants. They may be a good substitute for synthetic products for food applications.
Margarine is an emulsion prone to oxidation, and it is well known that vegetable oils are rich in antioxidants; their addition during margarine formulation could have an advantageous impact.
According to the results of this study (Fig. 2), the induction times of the three formulated margarines with 150 mg/g of ASO, WSO and MSO were 15.36 h, 16.54 h and 14.66 h, respectively; these results prove the oxidative stability generated by the investigated oils. Nevertheless, commercial margarine containing vitamin E gave the best induction time of 19.47 h. The margarine enriched with WSO has a higher induction time, followed by a margarine prepared with ASO but the induction time given by a margarine prepared with MSO was the lowest.
These oils can improve antioxidant capacity by preventing oxidation of PUFA thanks to their active components such as carotenoids and phenolics.
Unfortunately, there was no study in the literature for comparison. The oils studied have instead been applied in the production of other foods. As application of the apricot kernels oil, Saini et al. (2021) reported in their review study its use in macaroon paste and for enrichment of noodles. MSO has been mixed with peanut oil and the findings revealed that this can help to enhance some properties of the oil (nutritional and functional) (Siddeeg and Xia, 2015). MAKO: margarine formulated with apricot seed oil; MMSO: margarine formulated with melon seed oil; MWSO: margarine formulated with watermelon seed oil. The results with different letters for each parameter indicate a statistically significant difference (P < 0.05, a > b).

Conclusion
The seeds of apricot, melon, and watermelon were valued in this study by the extraction of their oils and their use in protecting a food product from oxidation, which is margarine.
The oils analysis showed that their physicochemical and quality parameters were are up to standard and of satisfactory quality. Furthermore, according to GC-FID analysis of these oils, oleic (C18:1 w9), linoleic (C18:2), palmitic (C16:0) and stearic acids (C18:0) were the main fatty acids in the three oils. It should also be noted that the oils studied exerted a significant scavenging effect with respect to the DPPH • . Oils at 150 mg/g added to table margarine have helped in protecting it from oxidation and watermelon seed oil showed better effect without affecting the properties and quality of the margarine.
The oils studied can be used to improve the nutritional quality of margarine due to their richness in fatty acids and to protect it from oxidation because they are a natural source of antioxidants.