The plant material consisted of baobab (Adansonia digitata L.) seeds from fruits collected at three locations: seeds from Kougheul (13˚58'60" N and 14˚48'0" W), Ziguinchor (12˚33'50" N and 16˚15'50" W) and Bignona (12˚48'18" N and 16˚14'4" W), Senegal. Three (3) batches of 50 kg of seeds from each source were used for oil extraction. Each batch was divided into two equal parts: 25 kg unwashed seeds and 25 kg washed seeds ( Figure 1 ). Approximately 150 litres of water at room temperature (25˚C) were used to wash the 75 kg of seeds, which were soaked and mixed for a total of two (2) hours. After washing, the seeds were dried in an oven at 65˚C ± 1˚C for 24 hours. Six (6) samples of 25 kg of seed were then taken. The seeds from the different lots were ground separately in a mill and passed through a 1 mm mesh sieve.

The baobab oil was extracted using a mechanical press (DD85G, IBG Monforts Oekotec GmbH, Mönchengladbach, Germany). The 10 mm spinner was used throughout the extraction, and the rotation speed of 25 rpm was maintained. The outlet head temperature was also maintained at 105˚C throughout the procedure. The outlet head was previously heated to this temperature for approximately 25 minutes at the start of the extraction process. At the end of the extraction, the product obtained was a mixture of oil with rubbery impurities. The crude oil was immediately bottled in amber bottles for two days. After decanting, the oil was transferred to new bottles and centrifuged using a centrifuge (Hettich, Zentrifugen, Germany) at 4500 rpm for ten (10) minutes. The baobab oil obtained was stored at 4˚C before analysis.

In this section, all the reagents used were of analytical quality. These reagents were methanol, water and acetonitrile of HPLC quality, ammonium formate, formic acid, cyclohexane, hexane and potassium hydroxide. These reagents were also purchased from Sigma (St. Louis, MO, USA).

1) Methyl esters preparation

To convert the oils to fatty acid methyl esters, 2 mL of hexane was added to a 20 µL volume of baobab oil. To this solution 100 µL of a 3 M methanolic solution of potassium hydroxide (KOH) was added and stirred at 40 rpm for five (5) minutes. After this stirring period the mixture was allowed to stand until a clear hexane phase was obtained. Gas chromatography was used to identify and quantify the fatty acids.

2) Analysis by gas chromatography-mass spectroscopy (GC-MS)

Fatty acids were analysed using a gas chromatograph (Trace GC electron) coupled to a mass spectrometer (ISQ, Thermo Finnigan, Thermo Scientific, USA). The stationary phase is a capillary silica column SGE-BPX5 (30 m × 0.25 mm i.d., 0.25 µm thick). The mobile phase (carrier gas), consisting of helium (He), was injected at a flow rate of 1 mL·min⁻¹ in split 1/20 mode. For GC-MS detection, the ionisation energy was 70 eV in electron impact (EI) mode. The transfer and ion source temperatures were 200˚C and 300˚C, respectively. The oven temperature was set at 120˚C for 20 minutes and then increased to 275˚C at a rate of 5˚C per minute. After dilution of the prepared methyl esters (1/20, v/v, in CHCl₃), a volume of 1 µL was injected into the GC-MS chromatograph in split mode. The m/z mass spectra were determined with ratios between 45 and 600. The total retention time was 60 min. Peak integration and chromatographic data processing were performed using Thermo Xcalibur software. Computer superposition of the mass spectra obtained with those of the standards and the NIST MS Search 2.0 spectral library allowed the identification of the compounds in baobab oil.

1) LC-UV analysis

LC-UV analyses were performed on a Shimadzu LC 20AD instrument consisting of a quaternary pump, a solvent degasser, a thermostated column with a ZORBAX^® SB-Phenyl column (250 mm × 4.6 mm, 5 μm) and an autosampler connected to an SPD-M20A DAD detector. LabSolutions LCMS software (Shimadzu) was used to evaluate the chromatograms. The chromatographic and UV conditions were optimised to obtain the appropriate sensitivity for the analysis of polyphenols as tyrosol (λmax = 279 nm) or caffeic acid (λmax = 325 nm). The mobile phase, consisting of a mixture of acetonitrile (solvent A) and 3 mM formate buffer, pH 3 (solvent B), was injected at a flow rate of 1 mL·min⁻¹. Chromatographic analysis was performed at 30˚C and the volume injected into the chromatographic system was 20 µL. Table 1 shows the variation in the proportions of solvents A and B.

2) Sample preparation

The phenolic fraction of baobab oils was obtained by liquid-liquid extraction. A mass of 2.0 g of baobab oil was weighed to the nearest 0.0001 g in a test tube. A volume of 1 mL of cyclohexane was added to the tube and vortexed for 10 seconds. Three (3) mL of methanol was then added and the mixture was mechanically stirred (Stuart SB3 rotator) for 5 minutes. The tube containing the mixture was immediately uncapped and centrifuged (P. Slecta centro 8-BL) at 1000 rpm for ten (10) minutes. The methanolic phase was then collected with a glass pasteur pipette and dried under a gentle nitrogen stream at 50˚C and 50 kPa pressure. The final dry residue was taken up with 0.5 mL of mobile phase (acetonitrile/formate buffer). With gentle vortexing, the new mixture was withdrawn with a syringe and then filtered. This filtration was performed on Acrodisc (Pall GHP, Membrane Acrodisc 13 mm, Syringe Filter) 0.45 μm. To analyse the phenolic fraction, a volume of 200 µL was added to a vial with an insert.

Principal component analysis (PCA) and hierarchical classification were performed on the data for acids and phenolic compounds in oils to find the best correlations between the random variables. Analyses of variance using Fisher’s LSD test at 5% significance level were also performed to compare means. All analyses were performed using R software (version 4.3.2, 2023).

Analysis of fatty acid methyl esters by GC-MS allowed the identification and quantification of eighteen (18) fatty acids in the different baobab oils. The results obtained are presented in Table 2 . They show a variation in the fatty acid composition of baobab oils depending on the origin and/or the type of pre-treatment applied to the seeds. The results show that the main saturated fatty acids in these oils are palmitic acid (14.01% to 24.63%), stearic acid (2.32% to 3.75%) and arachidic acid (2.03% to 4.45%). The most abundant monounsaturated fatty acid in these baobab oils is oleic acid (29.98% to 40.66%), followed by cis-10-nonadecenoic acid (3.18% to 8.08%). Palmitoleic acid (0.21% to 1.17%) and margaric acid (0.21% to 0.62%) are the two least abundant monounsaturated fatty acids. The main polyunsaturated fatty acid was linoleic acid (21.59% to 26.57%). The two (2) cyclopropenoic fatty acids found in the oils are malvalic acid and sterculic acid in proportions ranging from 0.72% to 1.57% and from 0.56% to 2.65% respectively. Dihydrosterculic acid, a cyclopropanic acid, was found in proportions ranging from 0.20% to 1.01%. Comparison of our results with those reported in the literature highlights the variation in composition depending on the country in which the baobab fruit was collected. Palmitic and stearic acids are also the main saturated fatty acids in these oils. Baobab seed oil collected in Pakistan had the highest stearic acid content [7] . Its high content suggests that it should be used for dietary purposes as it is converted to oleic acid by the liver [22] . The main unsaturated fatty acids in these oils are oleic and linoleic acids. The presence of palargonic acid (0.05% ± 0.07%), myristic acid (between 0.1% and 1.01%), anisostearic acid (between 0.11% and 0.15%), linolenic acid (between 2.06% and 8.84%), linolelaidic acid (between 0.18% and 0.03%), vaccenic acid (between 0.81% and 1.36%) and elaidic acid (2.8% ± 0.00%) were found in baobab oils [7] [14] [23] - [25] ( Table 3 ).

Table 2. Quantification of fatty acids (%) identified by GC-MS in baobab (Adansonia digitata L.) oils at the beginning of storage.

R.T.: Retention Time; M.F.: Molecular Formula; WZS: Washed Ziguinchor Seeds; UZS: Unwashed Ziguinchor Seeds; WBS: Washed Bignona Seeds; UBS: Unwashed Bignona Seeds; WKS: Washed Koungheul Seeds; UKS: Unwashed Koungheul Seeds.

Of the different baobab oils obtained by cold pressing, those from the Bignona seeds are the richest in saturated fatty acids (SFA). The percentage SFA in WBS and UBS oils is 31.96% and 31.83% respectively ( Figure 2 ). The lowest values were found in Koungheul oils with 29.37% and 23.24% respectively. This difference in SFA between the Koungheul oils and those from Bignona and Ziguinchor could logically be linked to the soil and climatic conditions, phenological variability, duration and/or storage conditions of the baobab seeds, as stated by Ranalli et al. [27] and Diop et al. [4] . Nevertheless, the Koungheul oils showed the highest percentages of monounsaturated fatty acids (MUFA) and cyclic fatty acids (CFA). The percentages of MUFA and CFA were 43.35% and 49.87% for Koungheul oil and 3.95% and 2.86% respectively. The polyunsaturated fatty acids (PUFAs) found in baobab oil are linoleic acid and cis-10-nonadecenoic acid. These two (2) acids were higher in the Ziguinchor oils. Ziguinchor oil from unwashed seeds contained more linoleic acid (26.57%) while Ziguinchor oil from washed seeds contained more cis-10-nonadecenoic acid (8.08%). Razafimamonjison et al. [26] indicate that baobab oil contains malvalic acid (1.77% to 3.87%), sterculic acid (0.42 to 1.68%) and dihydrosterculic acid (1.74% to 3.86%). However, the malvalic acid content obtained is lower than the 5.52% reported by Abeer et al. [28] . These cyclic fatty acids make this crude oil unsuitable for human consumption. Andrianaivo-Rafehivola et al. [29] suggest refining the oil before consumption in order to significantly reduce the cyclic fatty acids.

Figure 3(a) and Figure 3(b) show examples of chromatograms of the phenolic fractions of some baobab oils obtained by HPLC-UV analysis.

HPLC-UV chromatograms at 279 nm show variations in the phenolic compound content of different baobab (Adansonia digitata L.) oils extracted by cold pressing. Controls and literature data were used to identify and quantify the peaks. However, the other peaks detected at the end of the chromatogram between 12.5 and 20 min could not be identified due to the absence of specific controls. Phenolic compounds (mg·L⁻¹) were analysed by HPLC-UV. The levels of tyrosol, hydroxytyrosol, quercetin, caffeic acid and gallic acid are shown in Table 4 . Analysis of the phenolic fractions of the different oils extracted by pressing did not reveal the presence of gallic acid or quercetin. Salih and Yahia [30] found quercetin in the seeds (27.82 ± 0.74 mg·100 g DM, Dry Matter). However, simple phenolic compounds such as tyrosol and hydroxytyrosol are present in these oils. These two molecules are known in the pharmaceutical field for their antioxidant, anticarcinogenic, neuroprotective, antibacterial and anti-inflammatory properties [31] - [35] . The results show significant variations in tyrosol, hydroxytyrosol and caffeic acid content for the six (6) baobab oil samples. The determined tyrosol contents ranged from 0.397 ± 0.243 to 2.416 ± 2.408 mg·L⁻¹.

Table 4. Levels of phenolic compounds (mg·L⁻¹) in different baobab oils at the start of storage.

In the same column, means with the same letter are not significantly different at the 5% threshold. WZS: Washed Ziguinchor Seeds; UZS: Unwashed Ziguinchor Seeds; WBS: Washed Bignona Seeds; UBS: Unwashed Bignona Seeds; WKS: Washed Koungheul Seeds; UKS: Unwashed Koungheul Seeds; LOQ: Limit of quantification; N.D.: Not Detected.

Oils from seeds collected in Ziguinchor had the highest tyrosol content. These were 0.992 ± 0.879 mg·L⁻¹ for WZS oil and 2.416 ± 2.408 mg·L⁻¹ for UZS oil. The highest levels were also obtained in oils extracted from unwashed seeds. This difference between the two categories of oil is probably due to the contribution of tyrosol molecules from the pulp. In fact, the unwashed seeds still contained a small quantity of pulp despite the pulping operation. Studies by Vertuani et al. [36] detected the presence of tyrosol, hydroxytyrosol, caffeic acid and gallic acid in baobab pulp.

The hydroxytyrosol content was also higher in oils extracted from unwashed seeds. These hydroxytyrosol contents were 0.730, 0.585 and 3.595 mg·L⁻¹ respectively for oils from UZS, UBS and UKS. For WBS and WKS oils, these contents were 0.660 and 5.257 mg·L⁻¹, respectively. The hydroxytyrosol content of oil extracted from UKS is close to that of olive oil (1.4 - 5.6 mg·L⁻¹) reported by Montedoro et al. [37] . However, hydroxytyrosol was not detected in WZS oils. Like tyrosol and hydroxytyrosol content, caffeic acid content varies according to the origin of the seeds and the type of pre-treatment applied. In the case of seeds collected in Koungheul, the caffeic acid content of the extracted oils was 0.251 and 0.690 mg·L⁻¹ for washed and unwashed seeds respectively. In addition, the caffeic acid content of oils extracted from seeds collected in Bignona and Ziguinchor was very close to the limit of quantification (LOQ). Gallic acid was not detected in any of the oils. These results are consistent with those of Salih and Yahia [30] who did not detect the presence of gallic acid in baobab seeds. It should be noted that Vertuani et al. [36] found very low levels of gallic acid in the pulp. Therefore, the absence of gallic acid and quercetin could be explained by their absence or very low content in baobab seed shells. In short, the seeds from Koungheul provided the richest oils in tyrosol, hydroxytyrosol and caffeic acid. This richness in phenolic compounds could be explained by the greater quantity of pulp in these seeds. As a result, this UKS oil is thought to have more therapeutic properties than the other baobab oils extracted.

Principal component analysis (PCA) was used to assess the impact of baobab seed pre-treatment and fruit origin on the fatty acid and polyphenol content of cold-pressed oils. The first dimension (Dim 1) contributed 51.23% and the second (Dim 2) 22.21%. In addition, these first two dimensions (Dim 1 and Dim 2) had the highest eigenvalues (10.76 and 4.66). On the other hand, the third dimension (Dim 3), the fourth dimension (Dim 4) and the fifth dimension (Dim 5) had contributions of 17.01%, 6.97% and 2.58% respectively, and eigenvalues of 3.57%, 1.46% and 0.54%. Consequently, the first two dimensions (Dim 1 and Dim 2) selected express 73.44% of the total variance. The variables cerotic acid (0.959), sterculic acid (0.940), margaric acid (0.922), lignoceric acid (0.919), heptadecenoic acid (0.909), tricosylic acid (0.833) and hyenic acid (0.828) are positively well correlated with the first dimension, while stearic acid (−0.976) is negatively correlated with it. The variables hydroxytyrosol (0.931) and oleic acid (0.817) are positively correlated with the second dimension (Dim 2), while malvalic acid (−0.845) is negatively correlated with it. Baobab oils extracted by cold pressing were classified into three classes ( Figure 4 and Figure 5 ). Class 1, represented by Washed Bignona Seeds, Unwashed Bignona Seeds and Unwashed Ziguinchor Seed (WBS, UBS, UZS) oils, is characterised by a high content of stearic acid (V.test = 2.023; p = 0.043) and low levels of dihydrosterculic acid (V.test = -2.030; p = 0.0423), cerotic acid (V.test = −2.062; p = 0.0392) and lignoceric acid (V.test = −2.069; p = 0.0385). Class 2 is represented by Washed Ziguinchor Seed oils (WZS). Class 3, represented by Washed Koungheul Seeds and Unwashed Koungheul Seeds (WKS, UKS) oils, has a high content of hydroxytyrosol (V.test = 2.148; p = 0.0317) and a

low content of malvalic acid (V.test = −2.073; p = 0.0381). These results suggest that baobab oil extracted from Koungheul seeds would bear more biological effects as it contains less cyclic fatty acids. The presence of significant content in hydroxytyrosol molecules may offer prospects in the treatment of diseases associated with full or low-grade inflammation status.