*3.3. Determination of TGs Using UHPLC*/*ESI-MS*

Stock solutions of LLO, LLP, OLO, PLO, OOP, PPO, PPoO, OOO, LOL, OPL, OPO and POL at the concentration of 5 mg/mL were prepared in *n*-hexane:isopropanol (1:1, *v*/*v*). These solutions were diluted with the same solvent mixture yielding 12 calibration working solutions between 1–400 μg/mL. All the oil samples were prepared in 1 mg/mL and 500 μg/mL prior to chromatographic analysis.

The chromatographic system consisted of an Agilent 1290 Infinity series UHPLC with a diode array detector, binary pump, auto-liquid sampler and thermostated column compartment (Santa Clara, CA, USA). The UHPLC conditions were: three ACQUITY UPLC BEH C18 columns (3.0 ×100 mm, 1.7 μm, Waters, Milford, MA, USA) were connected in series with the column temperature set to 30 ◦C. The eluents consisted of acetonitrile with 0.1% water (A), isopropanol with 5 mM ammonium formate (B), and both contained 0.05% formic acid. The gradient elution started at 20% B, programmed to 50% B in 45 min, and then 60% B in 35min. The flow rate was 0.3 mL/min, and 1 μL of each solution was injected in duplicate for the LC analysis.

The UHPLC instrument was coupled to an Agilent 6120 quadrupole mass spectrometer with a dual ESI and APCI interface. Both ESI and APCI in positive and negative modes were evaluated, and ESI(+) in full scan mode from 300–1000 amu was selected for the analysis of TGs. The fragmentor voltage was optimized to 140 V. The drying gas flow was 12 L/min and the nebulizer pressure was 35 psi. The drying gas temperature and vaporizer temperature were set to 325 ◦C and 250 ◦C, respectively. The capillary voltage was 4000 V and the corona current was 4.0 μA.

### *3.4. Determination of FAs Using GC*/*MS*

Stock solutions of palmitic acid, palmitoleric acid, stearic acid, oleic acid, vaccenic acid, linoleic acid and linolenic acid in the form of methyl ester were prepared in *n*-hexane to make the 10 mg/mL stock solutions. These solutions were diluted with the same solvent yielding 12 calibration working solutions between 5–1000 μg/mL. Undecanoic acid methyl ester was used as the internal standard and added to each calibration solution at a fixed concentration of 200 μg/mL.

A modified procedure based on the AOAC 996.06 method [28] and the method proposed by Ai, et al. [38] for the hydrolytic reaction and transesterification of TGs was employed. In brief, 70 μL of each oil sample was accurately weighed (57.2–67.7 mg) and mixed with 2 mL glyceryl triundecanoate (2.5 mg/mL) and 200 μL 2N potassium hydroxide, both in methanol. The resulting cloudy solution was vortexed for 2 min, and then sonicated for 45 min at 60 ◦C in a water bath until the solution became clear. Then, 2 mL *n*-hexane was added to the solution. After sonication for 30 min and centrifugation, the supernatants were taken and diluted (2, 4 and 10 times) to obtain three different concentrations for GC/MS analysis.

The FAMEs analysis was performed on an Agilent 7890B gas chromatography (Santa Clara, CA, USA) coupled with an Agilent 5977A mass spectrometer. An Agilent J&W HP-88 column (60 m × 0.25 mm × 0.20 μm) was used. Helium was used as the carrier gas at a constant flow rate of 1.2 mL/min. The oven temperature was first set at a 60 ◦C hold for 1 min, and the temperature was subsequently increased to 145 ◦C at a rate of 10 ◦C/min, then to 190 ◦C at a rate of 1 ◦C/min, and finally to 240 ◦C at a rate of 5 ◦C/min. The inlet temperature was 260 ◦C. The split ratio was 100:1 with 1 μL injection. Full scan data was acquired in the mass range of *m*/*z* 30–500 amu and the EI voltage at 70 V. The temperature of the transfer line was 260 ◦C. The temperatures of ion source and quadrupole were set to 230 ◦C and 150 ◦C, respectively.

#### *3.5. Statistical Analysis*

The raw data for both UHPLC/MS and GC/MS were pre-processed using MassHunter Profinder (version 8.0, Agilent Technologies, Santa Clara, CA, USA) for finding features. The extracted features were exported as a cef file and then imported to Mass Profiler Professional software package (version B.12.05, Agilent Technologies) and SIMCA-P software (Version 12.0, Umetrics, Umeå, Sweden) where the features were further aligned, normalized and statistically evaluated.
