Processing practices and quality of crude palm oil produced on a small scale in Valença, Bahia, Brazil

Data collection

The study was conducted at ‘smallholders’ processors in the municipality of Valença, a Brazilian city located on the coast of the state of Bahia, northeast region of Brazil.

An introductory meeting was held with the Association of Palm Oil Producers in Valença, the government representatives and the manufacturers of CPO extraction machinery. The latter were surveyed. The mill owners were selected according to their interest in participating in this study. Data were collected through verbal interviews with the producers and manufactures of CPO. The processes were observed and the workers were also interviewed during CPO extraction. Observations and interviews were conducted to gather pertinent data about the company's real conditions, products and existing technologies, equipment operation, degree and type of mechanization, safety and environmental issues associated with CPO extraction. All participants provided informed consent, which had previously been approved by the Ethics Committee of the School of Nutrition of the Federal University of Bahia (Protocol 2.027.540).

Sample collection

Nineteen smallholder oil palm fruit processing mills were visited (4 T-CPOE, 8 LM-CPOE and 7 SM-CPOE) in Valença during the palm fruit season. Specifically, oil sampling was performed from December 2017 to March 2018. Three visits were conducted to each mill, in intervals of 30 days, during which CPO samples were collected. The number of samples studied were 57 (12 T-CPOE, 24 LM-CPOE and 21 SM-CPOE). On the day of each visit, the researchers followed the entire extraction process and 1 L of bottled CPO was randomly selected the following day from each type of extraction according to the local procedure. This procedure aimed to collect cold samples. The oils were wrapped in aluminum foil to protect them from light. After collection, the samples were taken to the laboratory in thermal containers. Then, the samples were transferred to amber bottles, flushed with nitrogen and sealed. The bottles were stored at -20 °C prior to analysis.

Analytical methods

Statistical analysis

Data were analyzed using IBM SPSS software package version 29.0 (Armonk, NY: IBM Corp). One-way ANOVA was applied. Multiple comparisons were made using Duncan’s test or the non-parametric Gomes-Howell’s test when the Levene’s test for equality of variances was not fulfilled. Spearman’s correlation was performed to assess the relationship between the different physicochemical parameters. Significance was considered at the 5% level.

Processing practices in Valença-Bahia for CPO extraction

A flow chart diagram illustrating the operation units involved in each CPO extraction process is presented in Figure 1. Generally, the CPO extraction encompasses several stages, including splitting (shredding) and/or chopping of fresh fruit bunches (FFB), storage of cut bunches for fruit loosening, threshing to separate the fruit from bunches, sterilization, crushing, oil separation and clarification. These producers use vehicles or animals to transport the fruit bunches to the processing unit. All three methods share the common practice of processing fruits within a time frame that typically varies from three to six days after fruit harvesting. This relatively long time facilitates the separation of fruit from bunches, but favors and intensifies the oil acidity (De Almeida et al., 2013De AlmeidaDT, NunesIL, CondePL, RosaRPS, RogérioWF, MachadoER. 2013. A quality assessment of crude palm oil marketed in Bahia, Brazil. Grasas Aceites64, 387–394. 10.3989/gya.118412; Nchanji et al., 2013NchanjiYK, TatawO, NkonghoRN, LevangP. 2013. Artisanal Milling of Palm Oil in Cameroon. Working Paper 128. Bogor, Indonesia: CIFOR.). In the T-CPOE method, threshing is carried out manually. In LM-CPOE, an axe or machete is employed for threshing, while in SM-CPOE, a mechanical thresher (4m x 0.8m x 1.6m) is utilized. This mechanical thresher employs either rotation or vibration to effectively separate the fruit from the bunches (Figure 1). The T-CPOE is normally performed by individuals who select the ripe fruits, remove the damaged ones, and manually wash them.

Figure 1 Schematic representation of the crude palm oil extraction (CPOE) methods. T-CPOE, traditional; LM-CPOE, little-mechanized; SM-CPOE, semi-mechanized 

The sterilization process can be carried out using boiling water or steam, and it serves a dual purpose. Firstly, it softens the fruit, making oil extraction easier. Secondly, it deactivates lipolytic enzymes that can cause oil hydrolysis and oxidation, ultimately leading to a significant reduction in oil quality (Nchanji et al., 2013NchanjiYK, TatawO, NkonghoRN, LevangP. 2013. Artisanal Milling of Palm Oil in Cameroon. Working Paper 128. Bogor, Indonesia: CIFOR.). In the T-CPOE, the fruit (about 20 kg per batch) is cooked or sterilized in boiling water for 2 h using an aluminum pan (20 L). An open fire fueled by empty fruit bunches and fiber is employed for this purpose. In the LM-CPOE, a higher amount of fruit is processed per batch (about 200 kg) and the sterilization is also conducted in boiling water, but the duration ranges from 3 to 6 h, and iron drums (200 L) are utilized.

After this procedure, the water is poured onto the ground, and the fruits are cooled to room temperature. In the SM-CPOE, the FFB (2 tons) are sterilized with steam for 2.5-3 h using an iron tank (6 m x 3 m x 1.5 m) equipped with a discharge port. Then, an employee removes the cooked bunches from the tank using an iron hook and takes them to the mechanical thresher, which is driven by a 7.5-hp electric motor capable of processing 6 tons of fruit per hour, effectively separating the fruit from the bunches at high speed (Figure 1). In this method, the empty bunches slide down the front of the thresher and onto the floor, where they are subsequently removed by an employee using a wheelbarrow.

The palm fruit separated from the bunches proceeds to the digestion process, which involves both crushing and maceration to extract the oil (Nwakodo et al., 2020NwakodoCS, ChukwuMN, UdensiES. 2020. Effect of Processing Methods and Storage Time on Physical Characteristics of Palm Oil. SSRN Electronic Journal. 10.2139/ssrn.3517478). The LM-CPOE and SM-CPOE methods employ mechanized digesters. In the LM-CPOE process, a horizontal iron crusher measuring 2.4 m in length, equipped with 40 small iron blades (10cm x 5cm), is utilized. This crusher is powered by a 5-hp engine and has a processing capacity of 4 tons of fruit per hour. In the SM-CPOE, the digester is a 3-meter-long vertical metal tube containing 7 metal blades (30 cm x 7.5 cm). This digester is driven by a 5-hp engine coupled with a speed reducer. It can process up to 6 tons of fruit per hour. In the T-CPOE method, this is performed either by pounding the cooked fruit in large wooden mortars with wooden pestles or by foot trampling on the cooked, albeit cold, fruit (Figure 1).

The digested mash is transferred to a 20 L aluminum pan in the T-CPOE or to a cement tank with a capacity of 2 to 3 thousand liters in the LM-CPOE process. In both cases, oil separation is achieved through the addition of water. The volume of water added is twice the volume of the digested mash. Then, the mixture containing oil, seeds, fibers, and water is vigorously shaken by hand or using long wooden spoons. This agitation process continues until the oil separates from the mixture due to differences in density, rising to the surface, while the remaining components are decanted (Figure 1). In the T-CPOE method, the oil is methodically removed by hand, while in LM-CPOE, plastic bowls are employed for this purpose. The collected oil is then deposited into an aluminum pan until a thin oil layer remains atop the aqueous medium. The decanted material is squeezed by hand and the fibers and nuts are separated. In the case of LM-CPOE, the tank is equipped with a faucet for the convenient separation of water.

In the SM-CPOE process, the oil is extracted by pressing. Thus, the digested fruit is conveyed by a screw conveyor (2.2 m x 50 cm x 50 cm) to a hydraulic press. After pressing, the extracted oil flows into a continuous tank (2 m x 1 m x 1 m) and then it is pumped into a nearby cooking tank (3 m x 1.6 m x 1.2 m) for oil clarification. The remaining intact seeds combined with pulp fibers (pulp cake) passes through a mechanical shredder (4 m x 90 cm x 20 cm). This shredder employs vibration and suction extraction to effectively separate the pulp fibers from the seeds.

The remaining water in the CPO is removed through a process known as oil clarification. This step essentially involves a drying process, where the oil is heated to facilitate the evaporation of water. Local processors employing the T-CPOE use 20 L aluminum pans for the oil clarification, whereas those using the LM-CPOE heat the oil in 200 L drums. It is worth noting that some T-CPOE producers introduce herbs, such as Ocimum gratissimum L., during this operation for flavor enhancement, primarily to mask off-flavors (Nwakodo et al., 2020NwakodoCS, ChukwuMN, UdensiES. 2020. Effect of Processing Methods and Storage Time on Physical Characteristics of Palm Oil. SSRN Electronic Journal. 10.2139/ssrn.3517478). The CPO is deemed adequately clarified when it ceases to boil. The duration of this process depends on factors such as the quantity of oil being clarified, and the efficiency of the wood used as fuel. Typically, the process takes between 1 and 2 h. The clarified oil is allowed to cool in 1 L glass or PET bottles (LM-CPOE) or 5 L cans (SM-CPOE) for commercialization. The dark orange-colored sludge, mixed with the fiber residues and settled at the bottom of the containers, is sometimes used in the production of soap or animal feed (Taiwo et al., 2000TaiwoKA, OwolarafeOK, SanniLA, JejeJO, AdeloyeK, AjibolaOO. 2000. Technological assessment of palm oil production in Osun and Ondo states of Nigeria. Technovation20, 215–223. 10.1016/S0166-4972(99)00110-8). In the SM-CPOE, the oil is heated within a closed tank (6 m x 3 m x 1.5 m) for about 2-3 h. This tank features a small central opening at the top to facilitate water evaporation. Then the oil is pumped into storage tanks.

In all of these processes, the residual water generated in the sterilization and clarification stages is typically discarded and discharged onto the ground. On the other hand, the solid waste generated, including fibers, shells, and empty fruit bunches, is used as fuel in the boilers to generate the steam and energy required for the mill operation, or as animal feed (Robins, 2018RobinsJE.2018. Smallholders and machines in the West African palm oil industry, 1850–1950. Afr. Econ. Hist.46, 69–103. 10.1353/aeh.2018.0002,). The seeds are spread onto the ground adjacent to the mill and left to dry for approximately two weeks for later commercialization.

CPO characterization

The statistical analysis of the data did not show significant differences among the three collections throughout processing time, except for the DOBI and the C* color parameter when the LM-CPOE and SM-CPOE samples were respectively analyzed (Table 1). Consequently, the data from the various collections were analyzed as a whole to evaluate significant differences among the extraction processes.

The PV in the oils, ranging from 3.1 to 7.4 meq O₂/kg oil, was consistently below the recommended limit for CPO, as set by the Codex Alimentarius, 2023, which is 15 meq O₂/kg oil (Table 1). No significant differences in PV were found among the three extraction methods (Table 1). Ruswanto et al. (2020RuswantoA, RamelanAH, PraseptianggaD, ParthaIBB. 2020. Effects of ripening level and processing delay on the characteristics of oil palm fruit bunches. Int. J. Adv. Sci. Eng. Inf. Technol.10, 389–394. 10.18517/ijaseit.10.1.10987) demonstrated that fruit ripeness and storage duration can affect the PV, with oils from overripe fruit and prolonged storage showing higher values, even reaching 15 meq O2/kg. In the present study, the constant PVs throughout the season can be attributed to the use of the same fruit variety, sub spontaneous Dura (De Almeida et al., 2013De AlmeidaDT, NunesIL, CondePL, RosaRPS, RogérioWF, MachadoER. 2013. A quality assessment of crude palm oil marketed in Bahia, Brazil. Grasas Aceites64, 387–394. 10.3989/gya.118412), and a standardized harvesting practice. Smallholders visually identify ripe FFBs based on specific criteria, including the color of detached fruitlets that have fallen to the ground (Lai et al., 2023LaiJW, RamliHR, IsmailLI, HasanWZW. 2023. Oil Palm Fresh Fruit Bunch Ripeness Detection Methods: A Systematic Review. Agriculture13, 156. 10.3390/agriculture13010156). Additionally, uniform fermentation periods (Figure 1) may contribute to the observed consistent PV.

The fermentation process is utilized to improve fruit loosening and decrease the time required for spikelet processing, which may result in the production of crude palm oil (CPO) with high free fatty acid (FFA) content. This could be avoided with adequate equipment for fruit processing. In this regard, the objective of this research is precisely to demonstrate the need for modernization of the CPO extraction industry in this region. The formation of peroxides in CPO often results from processing methods by which hot oil absorbs oxygen from the environment and the process is catalyzed by the presence of certain metals (e.g. iron) and steam. The values tend to be higher with longer periods of fermentation and storage.

The TC contents in the oils ranged between 375 and 598 ppm, with no significant differences observed among the extraction processes (Table 1). It is noteworthy that the Codex Alimentarius (2023Codex Alimentarius2023. Standard for Named Vegetable Oils. Codex Stan 210. Revised in 2001, 2003, 2009, 2017, 2019. Amended in 2005, 2011, 2013, 2015, 2019, 2021, 2022, last updated01/02/2023.[Online]. https://www.fao.org/fao-who-codexalimentarius/sh-proxy/fr/?lnk=1&url=https%253A%252F%252Fworkspace.fao.org%252Fsites%252Fcodex%252FStandards%252FCXS%2B210-1999%252FCXS_210e.pdf) sets a broader range for TC values in CPO, ranging from 500 to 2000 ppm. Comparatively, TC values reported for Bahia CPO have been in the vicinity of those of this study, with reports of 585 ppm (Feitosa et al., 2019FeitosaS, BoffoEF, BatistaCSC, VelascoJ, SilvaCS, BonfimR, Almeida, DT. 2019. A real case study on the physicochemical changes in crude palm oil (Elaeis guineensis) during the deep-frying of akara, traditional cowpea-paste balls, in Brazil. Grasas Aceites70 (2), e305. 10.3989/gya.0703182) and 545-578 ppm (De Almeida et al., 2013De AlmeidaDT, NunesIL, CondePL, RosaRPS, RogérioWF, MachadoER. 2013. A quality assessment of crude palm oil marketed in Bahia, Brazil. Grasas Aceites64, 387–394. 10.3989/gya.118412). The relatively lower TC levels observed in Valença's oils may be attributed to potential losses of carotenes during processing. Factors such as high temperatures sustained for extended periods during sterilization and oil clarification, as well as light exposure during the clarification process, can contribute to the oxidation of carotenoids.

The CIELab coordinates did not reveal significant differences among the different CPO extraction processes (Table 1). All CPO samples were situated within the first quadrant of the CIELab color space, showing positive values for the a* and b* parameters. However, slight but significant differences were observed among the extraction processes for the b* and C* parameters, with both being slightly lower for the SM-CPOE samples. It was observed that the hue values (h_ab^o) consistently placed the oils in the reddish-orange color zone (Table 1).

The DOBI is calculated as the ratio between the levels of TC and compounds that absorb UV light at 269 nm, specifically secondary lipid oxidation products containing conjugated triene structures (Basyuni et al., 2017BasyuniM, AmriN, PutriLAP, SyahputraI, ArifiyantoD.2017. Characteristics of fresh fruit bunch yield and the physicochemical qualities of palm oil during storage in North Sumatra, Indonesia. Indones J. Chem. 17, 182-190. 10.22146/ijc.24910). Even though no significant differences were found in the TC contents among the three extraction processes, distinct variations were observed in the DOBI. Notably, the oils obtained through the T-CPOE method exhibited higher DOBIs, with values within the range of fair quality CPOs (2.31-2.92) (Table 1). Conversely, the DOBIs in the oils from the LM-CPOE and SM-CPOE processes were characteristic as poor-quality oils (1.68-2.30) (Gee, 1999GeePT. 1999. Use of the deterioration of bleachability index (DOBI) to characterise the quality of crude palm oil Masai, Malaysia. http://www.innoleague.com/Deterioration_Of_Bleachability.pdf). Several factors were observed across all processes that could have contributed to the reduction in TC levels and an increase in secondary oxidation products, consequently leading to low DOBI values (Basyuni et al., 2017BasyuniM, AmriN, PutriLAP, SyahputraI, ArifiyantoD.2017. Characteristics of fresh fruit bunch yield and the physicochemical qualities of palm oil during storage in North Sumatra, Indonesia. Indones J. Chem. 17, 182-190. 10.22146/ijc.24910). These factors include delays in fruit processing, variations in sterilization time and temperature, contamination from sterilizer condensate water, exposure to badly oxidized sludge, and potential hygiene issues in the mills.

The OSI values for the oils were extremely low compared to reported values for CPOs produced in the same region (2.04-4.66 h) (De Almeida et al., 2013De AlmeidaDT, NunesIL, CondePL, RosaRPS, RogérioWF, MachadoER. 2013. A quality assessment of crude palm oil marketed in Bahia, Brazil. Grasas Aceites64, 387–394. 10.3989/gya.118412). The OSI of the oils obtained by the LM-CPOE was lower (p < 0.05) compared to those obtained by the T-CPOE and SM-CPOE (Table 1). This clearly shows that the LM-CPOE oil had poor quality and suggests that the LM-CPOE method was the least suitable in terms of oil quality.

Similar to data in Table 1, no significant differences were found in the levels of TPC among the different collections, except for the polymer content when analyzing the LM-CPOE and SM-CPOE samples. Therefore, regardless of the collection, the data were analyzed to assess variances between methods. Across all CPO extraction processes, the levels of TPC ranged from 8.7 to 18.5 % (Table 2). Notably, the total contents of TPC were significantly lower in the T-CPOE samples (8.7-9.6%) compared to those found in the LM-CPOE (14.2-18.5%) and SM-CPOE (11.9-13.6%) oils (Table 2).

The total fraction of polar compounds in the CPOs primarily consisted of products from hydrolytic degradation, i.e. DAG and FFA. In descending order of concentration across all extraction methods, TPCs ranked as DAG ≈ FFA > oxTGM > (TGP + TGD) (Table 2). These results confirm the prevalence of DAG and FFA in fresh CPO (Nizam and Mahmud, 2021NizamAFA, MahmudMS. 2021. Food quality assurance of crude palm oil: A review on toxic ester feedstock. OCL - Oilseeds and fats, Crops and Lipids28, 23. 10.1051/ocl/2021011). DAG values for LM-CPOE and SM-CPOE were within those observed in the studies by Feitosa et al. (2019FeitosaS, BoffoEF, BatistaCSC, VelascoJ, SilvaCS, BonfimR, Almeida, DT. 2019. A real case study on the physicochemical changes in crude palm oil (Elaeis guineensis) during the deep-frying of akara, traditional cowpea-paste balls, in Brazil. Grasas Aceites70 (2), e305. 10.3989/gya.0703182), who reported values of 8.7%. The DAG and FFA values were significantly lower in the oils obtained by the T-CPOE. The oils produced by the LM-CPOE and SM-CPOE methods can be regarded as inadequate CPOs because the FFA contents in the oils were higher than 5%, except the SM-CPOE sample in collection 3, which was slightly below, but still close to the limit (Codex Alimentarius, 2023Codex Alimentarius2023. Standard for Named Vegetable Oils. Codex Stan 210. Revised in 2001, 2003, 2009, 2017, 2019. Amended in 2005, 2011, 2013, 2015, 2019, 2021, 2022, last updated01/02/2023.[Online]. https://www.fao.org/fao-who-codexalimentarius/sh-proxy/fr/?lnk=1&amp;url=https%253A%252F%252Fworkspace.fao.org%252Fsites%252Fcodex%252FStandards%252FCXS%2B210-1999%252FCXS_210e.pdf).

The higher hydrolysis of the LM-CPOE and SM-CPOE samples is related to the processing of damaged fruits, in which enzymatic hydrolytic reactions are initiated by the contact of natural lipases with the oil (Ruswanto et al., 2020RuswantoA, RamelanAH, PraseptianggaD, ParthaIBB. 2020. Effects of ripening level and processing delay on the characteristics of oil palm fruit bunches. Int. J. Adv. Sci. Eng. Inf. Technol.10, 389–394. 10.18517/ijaseit.10.1.10987). The meticulous fruit handling practiced in the T-CPOE process is crucial in preventing hydrolysis in CPO and ensuring the production of high-quality oils. Additionally, the absence of equipment made from iron material in this process further contributes positively. Thermal and oxidative reactions take place during oil extraction due to the high temperatures employed, as well as to the presence of polyunsaturated fatty acids (linoleic) that produce oxTGMs and, in advanced stages of oxidation, polymerized compounds (TGPs). Basically, the conversion of TAG to oxTGM is considered the beginning of oxidation. Similar to PV, no significant differences were observed in the levels of oxTGM among the three extraction methods (Table 2). Feitosa et al. (2019FeitosaS, BoffoEF, BatistaCSC, VelascoJ, SilvaCS, BonfimR, Almeida, DT. 2019. A real case study on the physicochemical changes in crude palm oil (Elaeis guineensis) during the deep-frying of akara, traditional cowpea-paste balls, in Brazil. Grasas Aceites70 (2), e305. 10.3989/gya.0703182) detected 1.3% of oxTGM for CPO produced in Valença. CPO extraction includes several heating steps, which give rise to the formation of hydroperoxides (primary oxidation compounds) and aldehydes and ketones, among other secondary oxidation products. Both primary and secondary oxidation products form the group of oxidized triacylglycerols (oxTGM). The significant differences found in the DOBI between the T-CPOE and the other two extraction processes (Table 1) were not observed for the oxTGM (Table 2). The TPC analysis is an accurate method to evaluate used frying oils, which normally present considerably high thermoxidative degradation. When applied to oils that have not been heated or oils with low oxidation, the quantification of oxTGM is subject to a considerably high error due to incomplete separation between the non-polar and polar fractions. It is very common that trace levels of the non-polar triacylglycerols are not completely separate and they are collected in the polar fraction. Even when these levels are very low, they contribute significantly to increasing the low levels of oxTGM, as they (TAG and oxTGM) coelute and so are determined together in the HPSEC analysis. In used frying oils, which normally comprise relatively high levels of oxTGM, the contribution of non-separated TAG is negligible.

The levels of polymers found in the present study were, as expected, very low (0.1-0.5%) (Table 2) and in agreement with those reported by Khor et al. (2019KhorYP., HewKS., AbasF, LaiOM, CheongLZ, NehdiIA., SbihiHM, GewikMM, TanCP. 2019. Oxidation and polymerization of triacylglycerols: In-depth investigations towards the impact of heating profiles. Foods8, 1–15. 10.3390/foods8100475), who reported 0.5% in palm oil and 0.1-0.8% in pure olein (Table 2). Triacylglycerol polymers (TGP+TGD) are indicative of thermal degradation at elevated temperatures. Sterilization and oil clarification are thermal treatments in the CPO extraction processes. However, the temperature does not surpass the water boiling temperature (100 ºC) in the sterilization step. In contrast, if the time taken in the oil clarification is not well controlled, temperatures higher than 100 ºC can be reached after water evaporation, resulting in polymerization (Feitosa et al., 2019FeitosaS, BoffoEF, BatistaCSC, VelascoJ, SilvaCS, BonfimR, Almeida, DT. 2019. A real case study on the physicochemical changes in crude palm oil (Elaeis guineensis) during the deep-frying of akara, traditional cowpea-paste balls, in Brazil. Grasas Aceites70 (2), e305. 10.3989/gya.0703182).

Linear correlations between the physicochemical parameters

The entire dataset obtained in this study was analyzed to examine possible linear correlations between the different physicochemical parameters. As expected, a negative correlation was observed between the DOBI and the oxidation indicators, namely, DOBI with TPC (r = -0.634, p < 0.01), TGP+TGD (-0.755, p < 0.01) and oxTGM (-0.439; p < 0.01) (Table 3). Additionally, a positive correlation was found between DOBI and TC (0.634, p < 0.01). The decrease in DOBI indicates a reduction in the levels of carotenoids and/or an increase in oxidation. Similar to the DOBI, an inverse correlation was observed between carotenoids and the oxidation compounds, i.e. (TGP +TGD) (r = -0.347; p < 0.01) and the PV (r = -0.366; p < 0.01) (Table 3), which could partially explain the lower TC contents found in the oils. An inverse correlation of TC with L* (r = -0.514, p < 0.05), b= (r = - 0.482, p < 0.01) and h_ab^o = (r = -0.840, p < 0.01) was also observed, and a positive correlation with a* (0.811; p < 0.01) (Table 3), indicating that the lower the carotenoid content, the more transparent, the less orange and the more yellow the CPO becomes. De Almeida et al. (2018De AlmeidaDT., CurveloFM, CostaMM, VianaTV, de LimaPC. 2018. Oxidative stability of crude palm oil after deep frying akara (Fried bean paste). Food Sci. Technol. (Brazil)38, 142–147. 10.1590/1678-457X.02217) showed changes in the color of CPO samples stored at 26-32 °C and 20-25 °C after 9 months of storage with increases in L*, b* and h_ab and a decrease in a*, indicating that oxidation led to a loss in orange color in the oils.

Positive correlations were found among oxTGM, DAG, FFA and PV (Table 3). High concentrations of FFA in CPO indicate a deterioration of oil quality, as these compounds are known to promote oxidative reactions. Notably, TPC, TGP+TGD and the DOBI were correlated with the FFA content, regardless of the extraction conditions, as indicated by significant (p < 0.05) Sperman’s correlation coefficients of 0.896, 0.542 and -0.526, respectively (Table 3).