The Gustavia macarenensis fruits were collected from the Amazon rainforest in Pastaza, Ecuador in October 2017 and identified by the botanist Dr. David Neil in the Herbarium of the Universidad Estatal Amazónica (ECUAMZ), Puyo, Ecuador. A total of 20 randomly chosen fruits (about 5.2kg of ripe fruits ripe enough for consumption, physically undamaged and uniformly shaped, well-sized with good color) were collected and the fruit pulp (1.1kg) was manually separated. Before the extraction process, the fruit pulp was dried in a stove at a temperature of 45 °C for 48 hours. Later, it was ground in a mill until particle sizes of 1mm were achieved, in accordance with the ASTM-E1757-01 (2007) using a set of Tyler mesh sieves.

The fruit pulp was characterised using proximate analysis; while the physical-chemical properties were determined for the pulp oil. The information regarding these parameters is very useful in determining the quality of the raw material as a potential source of vegetable oil in the future. Physical-chemical determinations were performed according to the standardized methods proposed by the Official Methods of Analysis of AOAC International. The standards used in the study are summarized in Table 1.

The Soxhlet Method was selected as our extraction technique, using n-hexane as solvent, and was carried out according to the method reported by Pereira et al., (2017). For each 30g extraction of fruit pulp, the cellulose was wrapped in cartridges and placed in the Soxhlet apparatus. 250 mL of solvent were added, and the system was heated to boiling point. Reflux was held for 4 hrs. After extraction, the solvent was eliminated in a rotary vacuum evaporator (YAMATO, Model: RE200A, Japan) at 60 °C and 500 mm Hg. The extraction yield was calculated as the ratio between the mass of oil obtained and the mass of raw material used. Oil samples were stored at −5 °C ± 2 °C and protected from light prior to analysis.

The fatty acid profile in the oil sample was determined by High Performance Liquid Chromatography (HPLC), using a Waters instrument, model: Alliance 2695, equipped with a column symmetry C18, 5μm, 250 × 4.6 mm. In this analysis, the MMQ-HPLC- 09 standard was employed.

The extraction of the phenolic compounds from oil samples was carried out by liquid – liquid extraction using methanol as a solvent, according to the procedure described by Baiano et al., (2009) with some modifications. 2 mL of a methanol:water solution (70:30, v/v) and 2 mL of hexane were added to 3 g of oil and mixed with a Vortex for 10 min. The hydroalcoholic phase containing phenolics was separated from the oily phase by centrifugation (6000 rpm, 4 °C, 10 min). The hydroalcoholic phases were collected and submitted to another centrifugation (13000 rpm, at room temperature, 4 min). Finally, hydroalcoholic extracts were recovered with a syringe and then filtered through a 0.45 μm nylon filter before analysis.

The TPC was determined using the Folin Ciocalteu’s reagent method (Singleton et al., 1999). Approximately 0.1 mL of the oil sample was mixed with 0.5 mL of Folin-Ciocalteau reagent and subsequently 7 mL of distilled water were added. The mixture was left standing in the dark for 5 min at room temperature. Afterwards, 1.5 mL of sodium carbonate solution was added and the mixture was left at room temperature for another 2 hrs. The absorbance was measured at a wavelength of 765 nm in a Genesys 10 UV Scanning spectrophotometer model. Gallic acid was used as a standard to produce the calibration curve and TPC was expressed as mg of Gallic acid equivalents (GAE) per Kg of oil (mg GAE/Kg of oil).

Lipophilic antioxidant compounds were determined using a HPLC system (Shimadzu Corp., Kyoto, Japan) equipped with a Waters 600 controller and a Waters 996 photodiode array (PDA) detector set at an absorbance of 294 nm and equipped with a C18 column (150 × 4.6 mm, 3 μm; Phenomenex) as the stationary phase. Elution was performed with acetonitrile–methanol (90:10, v/v) as eluent mixture at a flow rate of 0.8 mL/min and the run time was 35 min. Samples (five milligrams) were diluted with 200 μL of chloroform and 790 μL of an acetonitrile–methanol (50:50, v/v) mixture and injected manually into the HPLC system. α- tocopherols (1 – 40 μg/mL, y = 2.347x+4.8353, R² = 0.9821) in acetone were used to produce the calibration curve. For total chlorophyll and carotenoid determination, a solution of 5% of the oil in acetone (80%) was prepared and estimated spectrophotometrically at two different wavelengths (464 nm for carotenoids and 669 nm for chlorophylls) using a Synergy™ 2 Multi-Detection Microplate. β-carotene (0.25 – 4 mg/mL, y = 0.1145× + 0.0235, R² = 0.9978) in acetone to produce the calibration curve, while the quantification of total chlorophylls was calculated as previously described (Porra, 2002).

Radical scavenging activity was studied using 1,1-diphenyl-2-picrylhydrazyl free radical (DPPH) as previously reported by Thaipong et al., (2006). 1.5 mL of oil extract was mixed with 1.5 mL of a 0.2 m Mmethanolic DPPH solution. After an incubation period of 30 min at 25 °C, the absorbance at 515 nm was measured. A Trolox calibration curve was used and data were expressed in Trolox equivalents (mmol of TE/Kg). DPPH scavenging activity (%) = [1 - ((Ab_sol - B₂)/B₁)] × 100 where Ab_sol, B₁ and B₂ are the absorbance of the sample, sample blank and oil extract blank, respectively. The values are expressed as the mean of three IC₅₀ values (concentration causing 50% inhibition) of each sample determined graphically.

Statistical analyses were performed using IBM SPSS Statistics v. 24. Data are reported as a mean value ± standard deviation (SD) for three measurements (n=3). In addition, a cluster analysis was carried out in order to classify the G. macarenensis fruit oil according to its fatty acid profile. Ward’s method was used for the formation of the clusters and the squared Euclidean distance for the calculation of the intervals between the groups. Then, a discriminant analysis was applied for the validation of the cluster. In this study, the fatty acid composition obtained from the mesocarp of 35 fruits previously reported by Chow (2007) were used. (See supplementary material).

The outer appearance of the G. macarenensis fruit and the appearance of its different parts are shown in Figure 1.

The fruit (G. macarenensis) is of medium size and the collected samples had an average weight of 258 g. The rind is tough and thick and is lilac to brownish in color mixed with some green shades. When ripe, the fruit is easily split and usually contains four to six seeds embedded in a yellowish edible pulp which is soft and sweet with an exquisite odor much liked by the local people of the Ecuadorian Amazonian basin.

The proximate analysis of the fruit pulp and the physical-chemical properties ofthe oil extracted from the mesocarp of G. macarenensis are summarized in Table 2.

As can observed in Table 2, the G. macarenensis fruit mesocarp contains a high value for total lipids (53.57%), which is uncommon for fruit pulp, as it usually contains very low levels of lipid material (0.1 to 1.0%). A few notable exceptions are avocado, palm fruit and olives, which range from 6.5–25.5%, 8–74% and 35–70%, respectively (Kamel and Kakuda, 2007). These are recognized as being sources of natural oils and as having several applications. In addition, G. macarenensis fruit mesocarp’s fat, crude fiber and crude protein values are higher than those reported by Devalaraja et al., (2011) for durian (Durio zibethinus) fruit pulp (fat (5.33%), crude fiber (3.1%), protein (1.47%)), which is considered to be a fruit with high nutritional value. The carbohydrate content is similar to that which was reported by Andrade et al., (2001) for the pulp of the Amazonian Annona squamosa fruit (20.95%) and lower than that of durian (27%).

Most of the results from the physical-chemical analyses performed on fruit pulp oils comply with the Codex Alimentarius International Food Standards established for virgin palm oils, which are found amongst the most commercialized vegetable oils worldwide (Commission, 2008). The acid value was determined at 3.56 mg KOH/g oil, which is below the maximum value suggested by CODEX STAN 210 as an essential quality parameter (10.0 mg KOH/g). This parameter is related to oil degradation caused by hydrolysis (Pereira et al., 2017). Moreover, the values for saponification, unsaponifiable matter and density are reasonably inside the range defined for palm oil by Codex Stan 210: 190–209 mg KOH/g oil; ≤ 1.2% and 0.891–0.899 g/mL, respectively. The iodine value of the G. macarenensis oil was lower than 50–55, which is the range reported for palm oil. According to Chow (2007), the iodine value is proportional to the degree of unsaturation, which is in accordance with the unsaturation content of mesocarp oil of the M. macarenensis fruit, similar to palm oil.

As can be observed in the analysis of the fatty acid profile (Table 3), the oil extracted from G. macarenensis mesocarp has palmitic and oleic acid as its main components. This explains the fact that it has a similar iodine value to that of palm oil (Elaeis guineensis), which usually has a range of 36 to 44% of oleic acid amongst its major unsaturated acids (Commission, 2008). In third place, one finds linoleic acid, although in lower quantities than the two aforementioned acids, whilst stearic, α-linolenic, eicosatrienoic, palmitoleic and margaric acids were present in small quantities of less than 4%. As described above, the statistical classification method known as the cluster analysis was used for making quantitative comparisons between the fatty acid compositions of 35 fruits and the composition of G. macarenensis oil. Figure 2 shows the dendrogram obtained from the cluster analysis of 36 fruits by Ward’s Method based on the fatty acid composition data, in which one can see the formation of 7 groups clearly defined based on their profiles. Notably, G. macarenensis was related to two more fruits: Durio zibethinus (Durian) and Buchanania lanzan (Chironji fruit).

The durian is an important economic fruit crop in several Southeast Asian countries, where it is known as the king of fruits (Ho and Bhat, 2015). In a few years, it has gone from being a curiosity to an important horticultural crop beyond the regional market, mainly due to the fact that it can be used in so many areas including the medical field and food industry. The fruit of this tree is considered a good source of fatty acids which has multiple benefits for human health (Mariod et al., 2017).

On the other hand, although less studied, is the Chironji fruit, whose kernel oil is used as a substitute for olive and almond oils, while the kernels are commonly used in sweet meats as a substitute for almond kernels (Hemavathy and Prabhakar, 1988). Other studies have reported on the immuno-stimulant and astringent properties of the dried fruits of B. Lanzan (Puri et al., 2000). These fruits are also known to possess antioxidant and anti-inflammatory activities (Shailasree et al., 2012). Pattnaik et al., (2013) carried out pharmacological studies on B. lanzan in wound healing and concluded that it has significant potential. However, it is too early to know if the G. macarenensis has some of these properties and further studies into its pharmacological properties and a phytochemical analysis are necessary.

Based on the application of the discriminant analysis for validating the cluster, it was found that 97.2% were classified correctly. Seven Fisher linear discriminant functions were obtained from the discriminant analysis. These models allow one to classify new fruits on the basis of their fatty acid profile and to determine which of the seven groups they belong to. In addition, it was concluded that of the seven fatty acids considered, 5 were highly significant with a p-value of < 0.01 (16: 0, 16: 1, 18: 1, 18: 2, 18: 3). 14: 0 was significant (p-value < 0.05), while 18: 0 was not significant in the formation of groups with a p-value of 0.619, much greater than 0.05.

The linear models can be expressed as follow:

where:

D_x: discriminant functions.

D_x:D: composition of the corresponding fatty acid (%).

Table 4 presents the total polyphenol contents, lipophilic antioxidant compounds and total antioxidant capacity of the G. macarenensis fruit pulp oil. The TPC value obtained for the G. macarenensis fruit pulp oil was higher than that determined by Tanilgan et al., (2007) in the olive oil varieties of Gemlik, Kilis, Uslu, Tirilye and Ayvalik (from 22.5 to 97.1 mg GAE/Kg). Furthermore, the phenolic contente stimated for the G. macarenensis oil is higher than the values reported by Ribeiro et al., (2008) for Brazilian mango seeds and peels (Mangifera indica L. variety Ubá), which were 82.54 and 57.24 mg GAE/Kg, respectively. According to the same authors, these are 4.6 and 7.3 times higher than those found in the Ubá pulp, respectively.

However, the values obtained in this study for carotenoids are similar to the contents reported for the Brazilian mango varieties Tommy Atkins, Palmer and Ubá (from 2.5 to 2.6 mg/Kg) (Rocha Ribeiro et al., 2007). Carotenoids have been recognized as a powerful antioxidant (Nwaichi et al., 2015). Dietary carotenoids are associated with health benefits, such as reducing the risk of disease, particularly skin cancer and inflammation along with fighting others cancers, cardiovascular and related disorders and those diseases related to oxidative stress (Johnson, 2002). The antioxidant activity by DPPH assay was determined for the oil from G. macarenensis pulp (see table 4). This value is low in comparison to other fruits. This could be due to different causes, for instance degree of fruit maturity, storage and the conditions of the technique employed.