Chemicals such as: 2,2-Diphenyl-1-picrylhdrazyl (DPPH), methanol, ethanol, sodium bicarbonate, vanillin, gallic acid, catechin, Folin-Ciocalteau reagent, potassium ferricyanide, and sodium hydroxide were purchased from Sigma-Aldrich Chemical Co. (St. Louis, MO, U.S.A.). Aluminum chloride, iron (III) chloride, nitrite de sodium, phosphate buffer, and trichloro acetic were purchased from Merck (Germany).

The seed samples of A. boehmii (vernacular name: Umushindwi) were harvested from three sites of three different eco-climatic zones of Burundi, namely: eastern depression (1200 - 1500 m of altitude), central trays (1400 - 2000 m of altitude) and foothills of Mumirwa (1000 - 1500 m of altitude). The seed of P. angolensis (vernacular name: umusurura) were harvested in the eco-climatic zone of the foothills of Mumirwa from different commune (Vyanda, Vugizo and Musigati). Harvest was carried out in July for P. anglolensis and October for A. boehmii. The identification of the plant species was performed at the herbarium of the University of Burundi (BJA) and the herbarium of the Burundian Office for the Protection of the Environment. The ripe fruits, identifiable by their respective color, were harvested manually. Nine trees were sampled on each specie and three samples were collected per site. The fruits were dried at room temperature in the Microbiology Laboratory of bioengineering’s faculty at University of Burundi. After drying, the seeds were hulled manually.

The first step was the extraction of the oil from the seeds to have oilcake. Seeds were crushed using a Moulinex blinder (France) and then the extraction was performed with hexane as solvent in a Soxhlet apparatus under reflux for 8 h. Thus, we found 24% and 69% of oil respectively from A. boehmii and P. angolensis. Then, to 10 g of completely defatted cake, 50 ml of ethanol (80%) was added and homogenized for 30 minutes on a magnetic stirrer. Subsequently, the phase separation was done by centrifugation at the 4000 rpm for 20 minutes. The supernatant (ethanolic extract) was recovered in a flask and the pellet was again reextracted three times. The three ethanolic extracts thus obtained were mixed and evaporated to dryness under reduced pressure. The extract was recovered in a known amount of ethanol 80%.

Chemical composition of the samples namely ash, and crude protein of oilcake were determined according to the Association of Official Analytical Chemist [33] methods. Carbohydrate content in the samples was estimated using formula [100%-ash-protein].

Since Folin Ciocalteu can react with other compounds, applying this method does not give exact values of polyphenols, rather estimates. For this, some researchers tend to replace the designation of total polyphenol contents by the Folin-Ciocalteu reagent assay measures sample reducing capacity [34] while others continue to use it as estimators of total polyphenol contents [35] . Thus, the total phenol content of the samples was determined using the Folin-Ciocalteau reagent [36] . Briefly, an aliquot of 100 μl of ethanol extracts was added to 1.0 ml of distilled water and 0.5 ml of Folin-Ciocaleu reagent (1/10 v/v). After mixing, 1.5 ml of 2% sodium bicarbonate was added to the mixture. The absorbance was read using a spectrophotometer at 760 nm after 30 min of incubation in the dark. The total content of phenolic compounds was expressed as mg gallic acid equivalent (GAE) of extract/100 g dry matter of oilcake (DM).

Total flavonoids were assessed using a colorimetric test as described by [37] . 250 μl of the extract and 1 ml of distilled water were successively introduced into a tube. At the initial time (0 minutes), 75% µl of a Na NO₂ solution (5%) were added, after five minutes 75% µl of AlCl₃ (10%) were successively added to the mixture. The absorbance of the mixture obtained was directly measured with a UV-visible spectrophotometer at 510 nm against the blank.

Condensed tannin concentrations were determined by a modified method of [38] [38] . Ten microliters of samples were mixed with 200 µl of vanillin-HCl reagent (4% vanillin in methanol and 8% concentrated HCl in methanol). After 15 min, the absorbance of the mixture was determined at 500 nm against a blank solution. The condensed tannin content was expressed as catechin equivalents (CE) in milligrams per 100 gram (mg/g) of dray matter of oilcake (DM).

Qualitative LC-MS analysis was carried out using a Dionex UltiMate 3000 RSLC system coupled to a TSQ Endura triple quadrupole mass spectrometer, equipped with an H-ESI source working in negative mode. Mass spectra were acquired in profile mode with a setting of 30,000 resolutions at m/z 400. Operation parameters were as follows: source voltage, 4 kV; sheath gas, 60 (arbitrary units); auxiliary gas, 20 (arbitrary units); sweep gas, 2 (arbitrary units); and capillary temperature, 275˚C. Extract samples were analyzed in full scan mode at a resolving power of 30,000 at m/z 400 and data-dependent MS/MS events acquired at a resolving power of 15,000. The most intense ions detected during full scan MS triggered data-dependent scanning. Ions that were not intense enough for a data-dependent scan were analyzed in MSn mode with the Orbitrap resolution also set at 15,000 at m/z 400. An isolation width of 100 amu was used and precursors were fragmented by collision induced dissociation C-trap (CID) with a normalized collision energy of 40 V and an activation time of 10 ms. The mass range in FTMS mode was from m/z 100 to 1000. The data analysis was achieved using XCalibur software v4.0.27.42 (Thermo Fisher Scientific).

Phenolic acid separation was performed on a reverse phase C18 column (4.6 × 250 mm) and the compounds elution were monitored with diode array detector. The mobile phase used was 2.5% acetic acid (Solvent A) and acetonitrile (Solvent B. The following gradient was applied: initial 3% B; 9% B for 5 min; 5 - 15 min, 16% B; and 15 - 50 min, 50% B.

The DPPH free radical scavenging test was measured as described by [39] . A volume of 0.1 ml of each of the solutions of the ethanolic extracts at different concentrations was mixed with 3.9 ml of methanolic solution (80%) of DPPH⁺ (0.004%). The decrease in absorbance was determined at 515 nm at 0 min, and every 5 min until a steady state was reached. The inhibition of free radicals in percentage (I%) was calculated using the following formula: I% = [1 − (Abs test/Abs control)] × 100; where Abs test is the absorbance of the sample and Abs control is the absorbance negative control. The concentration of each extract and ascorbic acid necessary to decrease the initial DPPH concentration by 50% (Efficient Concentration = EC₅₀) was calculated graphically. The antiradical activity was expressed as EC₅₀ (mg/ml). The kinetics of the reaction were evaluated according to [40] by calculating the time needed to reach a steady state with an antioxidant concentration corresponding to EC₅₀ (TEC₅₀) and effectiveness antiradical (EA). The TEC₅₀ was determined graphically. The AE, which involves the potency (1/EC₅₀) and the reaction time (TEC₅₀), was calculated using the following formula: AE = 1/EC₅₀ × TEC₅₀. The lower the EC₅₀, the lower the TEC₅₀ and the higher the AE. All tests were performed in triplicate.

The reducing power was measured according to [41] . A volume of 0.5 ml of the sample was homogenized with 1.25 ml of phosphate buffer (0.2 M, pH 6.6) and 1.25 ml of potassium ferricyanide [K₃Fe (CN)₆] (1%). After incubation in a water bath (50˚C/20 min), 1.25 ml of trichloroacetic acid (10%) was added to the mixture. which was then centrifuged at 2000 rpm for 20 min. The upper layer of the solution (1.25 ml) was mixed with distilled water (1.25 ml). The absorbance was read at 700 nm after the addition of 0.25 ml of Iron (III) chloride (1%). The increase in absorbance indicates a high reducing power [42] and results were expressed as antioxidant gallic acid equivalents in mg per 100 g of product (mg GAE/100 g).

Data analysis was performed using IBM SPSS statistic 20. Results were analyzed using one-way analysis of variance (ANOVA) followed by Tukey’s multiple comparison test. Correlation between various parameters was also investigated. Significance was determined at p < 0.05 and the results were expressed as mean values and standard error (SE) of the means. Comparisons were made between the results of the same species for three sites and another comparison between the results of these two species.

Results on proximate components of A. boehmie and P. angolensis oilcakes are illustrated in Table 1 . The Overall trends of compound contents compared to dry matter in both oilcake species were found to be similar. The contents of carbohydrates, proteins, and ash were, respectively 70.08% ± 4%, 26.81% ± 3.2% and 3.92% ± 0.3% for A. boehmii and 66.52% ± 2.4%, 28.67% ± 2.2%, and 4.82% ± 0.6% for P. angolens. Furthermore, proximate compounds from A. boehmii oilcake, calculated including oil content (24%), were found to be similar to those reported on the same species by [43] .

Thus, carbohydrate, crude protein, and ash contents were, 56.5%, 21.6%, and 2.5%, respectively. For P. angolensis, because of its high oil content (69%), the calculation includes the latter has remarkably lowered the proximate compound content at very low values: carbohydrate (39.36%), crude protein (16.96%), and ash (2.85%). Moreover, large discrepancies were observed for the levels of carbohydrates and proteins reported in the previous study on P. angolensis [16] . Regarding origin influences on yields, significant differences (p < 0.05) were observed in the results of protein contents (foothills > depressions > trays) recorded for the A. boehmii oilcake extracts and ashcontents (Vyanda > Musigati > Vyanda) determined in P. angolensis oilcake extracts.

The ethanolic extraction yields of A. boehmii and P. angolensis are mentioned in Figure 1(I) and are expressed as percentage (%) of dry matter from oilcake. The amount of TPC in samples was reported as mg of gallic acid equivalent (GAE) per 100 g. Figure 1(II) showed that P. angolensis exhibited the highest TPC content (1089.89 ± 293.40 GAE/100g) compared to A. boehmii (874.97 ± 20.45 GAE/100g). However, their differences were no significant at p > 0.05. TPC of A. boehmii oilcake extracts from different sites were also found not to be statistically significant (p > 0.05). These contents varied from 905.65 ± 37 mg GAE/100g obtained in the central trays region to about 850.44 ± 57.58 mg GAE/100g) in the seeds collected in the Mumirwa zone. Significant differences (p < 0.05) were observed for P. angolensis between TPC from Musigati (1530 ± 397.74 mg GAE/100g) > Vugizo (995.12 ± 59.35 mg GAE/100g) > Vyanda 744.56 ± 113.59 mg GAE/100g DM). Although A. boehmii seeds originated from climatically different regions (depressions, central trays, and Mumirwa), it seems not to have influenced the phenolic contents of this specie. Other studies reported a great environmental influence on TPC [44] .

For A. boehmii, yields obtained from seeds originating from different regions were found not to be significantly different (p > 0.05). But, for P. angolensis, significant differences (p < 0.05) were observed in this order: Vyanda > Vugizo > Musigati. Overall extraction averages of A. boehmii (14.80% ± 0.18%) and P. angolensis (17.50% ± 2.35%) showed no significant differences (p > 0.05).

Trends in TPC and extraction yield of ethanolic soluble substances observed between the two species were reversed in TFC and CTC. The results ( Figure 1(III) and Figure 1(IV) ) showed that TFC and CTC values were higher in A. boehmii oilcake extracts compared to these of P. angolensis oilcake extracts with significant difference at p < 0.05. Furthermore, significant differences were observed in A. boehmii oilcake extracts only. It was obvious that the environmental factor has influenced the accumulation of these molecules. The highest TFC was found to be in the seeds from Mumirwa (236.14 ± 43.22 mg CE/100g) followed by those from depressions (188.36 ± 10.43 mg CE/100g) and lowest were registered in those from central trays (132.88 ± 8.03 mg CE/100g) with significant difference at p < 0.05.

For P. angolensis, the highest were obtained in the seeds harvested from Vyanda commune site (113.59 ± 19.97 mg CE/100g) and the lowest these from Vugizo sites (95.69 ± 18.47 mg CE/100g). Results on CTC showed that the highest content (341.203 ± 40.15 mg CE/100g) determined in A. boehmii was twice as high than the highest obtained in the P. angolensis extracts (150.59 ± 26.34 g). The comparison of TFC and CTC with those of total polyphenols showed that it had no interdependence. Statistically, no correlation ( Table 2 ) was recorded between the contents of these two compounds (TFC and CTC) and those of TPC. Several studies, also reported that high content of polyphenols does not necessarily correlate with high content of flavonoids or condensed tannins [41] [45] .

* The correlation is significant at the 0.05 level (bilateral). ** The correlation is significant at the 0.01 level (bilateral).

The identification of the phenolic compounds revealed that the extract of A. boehmii contained important amounts of tannins ( Table 3 and Figure 2(a) ). Ellagitannins, the major tannins detected in A. boehmii extracts, are characterized by their hexahydroxydiphenoyl (HHDP) group which is released by acid hydrolysis and spontaneously lactonizes to ellagic acid [46] [47] . The sequential losses of galloyl (m/z = 152), gallate (m/z = 170) and HHDP (m/z = 301) residues of ellagitannin family [48] allowed to distinguish five compounds. The Hexahydroxydiphenoyl (HHDP)-hexoside (Peak 1), bis-HHDP-glucose isomer (Peak 2), galloyl-hexoside (Peak 3), HHDP-galloylglucose (Peak 5), galloyl-bis-HHDP-glucose (Peak 6), were identified according to other studies [49] - [52] .

Ellagic and gallic acid are phenolic acids which are indispensable compounds in the tannin structures [48] [53] . Therefore, the identification of tannins must be accompanied by that of gallic acids and ellagic acids. Thus, peak 10 was suggested as ellagic acid due to its mass spectrometry at m/z (MS m/z = 301) and its fragments with mass spectrometry at m/z (MS/MS m/z = 257, 229) as were reported by [54] . The Compound of peak 4 was proposed as gallic acid (MS m/z = 169, MS/MS m/z = 125) after being compared with a commercial standard and furthermore, a similar fragmentation pattern (Peak 4) was previously described [53] , [55] .

Chromatogram analysis showed also that A. boehmii contained also flavonoids, but less important than tannins. whereas peak 9 with MS at m/z 477 and fragments at m/z 315 and 151, was suggested as Isorhamnetin hexoside [54] . Peak 13 was proposed to be Quercetin 3-O-rhamnoside (MS m/z = 447, MS/MS m/z = 301, 179, 151). Previously, such features (MS and MS/MS) have been described [56] on raisins muscadine phenolic compound. Peak 11 was identified as Quercétine diméthyl éther 3-hydroxyferuoyl-glucosylglucoside-7-O-rhamnosid (MS m/z = 991, MS/MS m/z = 845, 653, 639, 301) whereas peak 12 was proposed to be monoglycosylated derivative of isorhamnetin (MS m/z = 447, MS/MS m/z = 314, 315).

The phenolic composition of P. angolensis extract was different from those of A. boehmie. For P. angolensis, flavonoids were the most prevalent than condensed tannins ( Table 4 and Figure 2(b) ). However, the colorimetric analyses showed that the two families of polyphenols have almost the similar content ( Figure 2 ).

In the present study, 7 flavonoids were observed at absorbance around 366 nm. Compound of peak 2 was presented as pelargonidin-3, 5-diglucoside (MS m/z = 595, MS/MS m/z = 433, 271) by referring on [49] while that of peak 3 was proposed to be proanthocyanidin B1(MS m/z = 595, MS/MS m/z = 433, 271) as describe by [54] Compounds shown by peak 4 and 8 were suggested as kaempferol 3-glucuronide (MS m/z = 463, MS/MS m/z = 287) and cyanidin hexoside (MS m/z = 449, MS/MS m/z = 327, 287) respectively. The examination of the chromatograms and data of [56] led to the identification of peak 9 as myricetin rhamnoside compound (MS m/z = 464, MS/MS m/z = 317). Peaks 7, due to its fragments observed, was proposed to be Apigenin 6-C-arabinosyl-8-C-glucoside (MS m/z = 563, MS/MS m/z = 503, 473, 383, 353). According to [57] in their study performed on determining of Duchesnea indica phenolic compound showed that neutral losses of 60, 90 and 180 corresponding to apigenin (270) + hexose (162) + pentose (132). Peak 13 was suggested as Procyanidin dimer (MSm/z = 577, MS/MS m/z = 451, 425, 407, 289) [47] . Regarding tannins, two compounds HHDP-hexoside (Peak 10) and HHDP-glucose (Peak 14) were tentatively identified basing on the procedure used for those of A. boehmie and the literature data [49] [58] . Others compounds detected were peak 5 and 6 identified as ellagic acid (MS m/z = 301, MS/MS m/z = 301) and 1 as citric acid derivative (MS m/z = 391, MS/MS m/z = 217, 191, 373) [49] [59] .

The scavenging effect of A. boehmii, P. angolensis oilcake extracts against the DPPH radical was evaluated using a spectrophotometric method. Thus, the evaluation of the antioxidant activity was performed in two approaches: the determination of the amount of antioxidant necessary to reduce 50% of DPPH^• after 60 minutes ( Table 5 ) and the time needed to reach a steady state for a given concentration of extract ( Figure 3 ).

Results are mean of three replicates (from 3 sites for samples) with standard errors (Mean ± S.E, n = 3, -, not determined). In each column different letters mean significant difference (p < 0.05): the letters decreasing in value, a > b > c, were used to compare the regions for one specie while u > v > w were used to compare the global values of the two species between them one hand, and between that of the ascorbic acid on the other hand.

The overall results revealed that the activity in A. boehmii extract (0.08 ± 0.00 mg/ml) was higher compared to P. angolensis EC₅₀ (1.29 ± 0.15 mg/ml) and almost similar to that of the ascorbic acid (0.09 ± 0.00 mg/ml). However, no significant difference (p < 0.05) was observed between EC₅₀ from A. boehmii extract and that from ascorbic acid. Regarding EC₅₀ calculated based on total polyphenols, results on A. boehmii and P. angolensis were the same at 0.01 mg GAE/ml and they were very significantly (p < 0.05) higher than that of ascorbic acid. This showed that both species contain significant amounts of bioactive molecules. According to our TEC₅₀ results, A. boehmii extract (11.8 ± 1.2 minutes) reached a steady state at a time that was not significantly different (p > 0.05) from that of P. angolensis (17.03 ± 1.9 minutes) while ascorbic acid (5 ± 0.6 minutes) was very significantly effective than both species. Results obtained on EA, by combining EC₅₀ and TEC₅₀ parameters, were in the following order: A. boehmii > ascorbic acid > P. angolensis. The statistical test revealed significant difference (p < 0.05) between them. Furthermore, a good correlation was found between EC₅₀ and EA (p < 0 0.01, r² = 0.6306). However, TEC₅₀ was not found to be correlated with EA (R² = 0.066).

Significant differences were also observed between the results of extracts from seeds originating from different regions. Thus, EA of A. boehemii harvested from foothills showed the highest activity, followed by those from depression then those from trays with significant differences at p < 0.05. The same trend of difference significations observed above was also observed between EC₅₀ determined in extracts: foothills > depression > trays. There was no significant difference at p < 0.05 between TEC₅₀ and EC₅₀ calculated in reference to the polyphenol concentrations.

Concerning P. angolensis, EA determined in oilcake from Vugizo was found to be significantly higher than those from other sites; order being Vugizo > Musigati > Vyanda (p < 0.05). EC₅₀ analyzed in oilcake extracts showed highest activity in oilcake from foothills in comparison of the other sites and followed the order as: foothills > trays > depression and significantly different at p < 0.05. Value EC₅₀ calculated in reference to polyphenol contents and TEC₅₀ were almost the same in their respective sites. No significant differences (p > 0.05) between these sites for each of these parameters. After the involvement of all parameters in DPPH assay, A. boehmii extract was found to be the most powerful antioxidant than that P. angolensis with EA of 146.75 ± 34.67 ml/µg·min while P. angolensis was 13.3 ± 0.5 ml/µg·mim. This is probably due to the high tannin content of A. boehmii. Particularly ellagitannins who is reputed to have important antioxidant, antiviral, antimicrobial, immunomodulatory, antitumor, and hepatoprotective activities [60] [61] .

Antioxidant activities of a crude extract may be due to different compounds acting through different mechanism to prevent the oxidation of sensitive organic substances. Ferric reducing antioxidant power (FRAP) assay was as complement to the DPPH assay. FRAP assay is also reported to have his high sensitivity and reliability [62] .

Through the results of Table 3 , the same trend observed in DPPH assay has been reproduced in FRAP assay where A. boehmii oilcake extract exhibited highest reducing power (3160.3 ± 415.0 mg GAE/100g) compared to P. angolensis oilcake extract (993.0 ± 108.1 mg GAE/100g). Furthermore, a highly significant correlation was observed between FRAP and TFC (p < 0.01, r² = 0.8446), and between FRAP and CTC (p < 0.01, r² = 0.7132). Furthermore, in both specie, significant difference between zones was recorded. For A. boehmii, results recorded on oilcake from foothills were found to be the highest Foothills > Trays > Depression with significant difference at p < 0.05. FRAP result of P. angolensis extracts from sites were also significantly different (p < 0.05) in the following descending order: Vugizo > Musigati > Vyanda. There is no research which is already performed on the evaluation of the antioxidant activity of these wild species using the FRAP method.

The results obtained using these two methods revealed that both species content valuable potent antioxidants. Probably that was due to the high content of the polyphenols. It has been reported that antioxidant activity is correlated strongly with polyphenol content [63] [64] , especially with flavonoid phenolic acids, and tannins contents [65] [66] . In this study, flavonoids represent 21.33 and 10% of the total polyphenols, respectively, in A. boehmii oilcake extract and in P. angolensis oilcake extract while condensed tannins had occupied, in the same order, 39.00% and 15.31%. However, there are lack of studies performed on ethnobotanical and biochemical activity of A. boehmii.