*3.3. Antioxidant Activities*

The 2,20 -azino-bis(3-ethylbenzothiazoline-6-sulfonic) acid (ABTS) radical scavenging activity was evaluated in a 96-well microplate using the method of Re et al. [57]. An ABTS radical solution was prepared by mixing 7 mM of ABTS at pH 7.4 (5 mM NaH2PO4, 5 mM Na2HPO4, and 154 mM NaCl) with 2.5 mM K2S2O8. After incubation in the dark at room temperature for 16 h, the ABTS solution was diluted with ethanol until an absorbance of 0.70 ± 0.02 units at 734 nm obtained. In each microplate well, 15 µL of the "antioxidant extract" was combined with 285 µL of freshly prepared ABTS solution. The mixtures were incubated at room temperature in the dark for 10 min and absorbance values were measured at 734 nm. ABTS activity was expressed using the linear calibration curve of trolox as g trolox equivalent/g f.w. The 2,2-diphenyl-1-picrylhydrazyl (DPPH) antioxidant activity assay was performed by spectrophotometry as described by Siddhraju & Becker [58]. Briefly, 20 µL of the "antioxidant extract" and 280 µL of freshly prepared methanolic radical DPPH solution (6 × 10–5 mol/L) were introduced into a 96-well microplate. The microplates were covered with a foil paper and left for 30 min at room temperature. The reduction in absorbance was measured at 517 nm. The DPPH activity was expressed using the linear calibration curve of trolox as g trolox equivalent/g f.w. To assess the ability of almond extracts to inhibit lipid oxidation, the β-carotene-linoleic acid bleaching assay was performed using the method described by Salleh et al. [59] with minor modifications as follows: a mixture of β-carotene and linoleic acid was prepared by adding 0.5 mg β-carotene with 1 mL chloroform (HPLC grade), 25 µL linoleic acid and 200 mg Tween 40. After complete evaporation of the chloroform under vacuum, 100 mL of water was added to the residue which was gently stirred to form a yellowish emulsion. To 50 µL of the "antioxidant abstract" were added 0.25 µL of the yellowish emulsion. After thorough mixing, the mixture was incubated in a water bath at 50 ◦C for 2 h followed by the measurement of absorbance values at 470 nm against a blank. Percentage inhibition (I%) of lipid peroxidation was calculated using the following equation:

$$\text{I\%} = \text{(A\ $-caretene after 2 h/Ainritical \$ -caretene)} \times 100;\tag{1}$$

where Aβ-carotene after 2 h is the absorbance value of β-carotene after 2 h of incubation and Ainitial β-carotene is the absorbance value of β-carotene before incubation.

#### *3.4. Fatty Acids*

Total oil was extracted from 5 g almond sample with petroleum ether in a Soxhlet apparatus [30] operating at 135 ◦C for 2 h. The extracted oil was used to prepare methyl esters of the corresponding fatty acids (FAME) according to EEC (1991) [60]. The resulting FAMEs were analyzed with a Shimadzu GC-2010 Plus gas chromatograph (Shimadzu, Kyoto, Japan) equipped with a flame ionization detector (FID-2010 Plus). Component peak separation was obtained on a DB-225MS capillary column (0.25 µm, 30 m × 0.25 mm i.d., Agilent Technologies, Wilmington, DE, USA). Helium was used as the carrier gas (200 kPa, constant flow, 1 mL/min). The temperature of the column was maintained for 10 min at 200 ◦C and then increased to 220 ◦C at 5◦C per min. The inlet and detector temperatures were set

at 270 ◦C. The split ratio was 5:1 and the injected volume was 1.0 µL. The constituent FAMEs were identified by comparison with standard FAME mixtures (FAME 37, Supelco, Bellefonte, PA, USA). The amount of each FAME was expressed as a weight percentage (%) of the total FAMEs represented in the chromatogram, with the assumption that no other major lipids and other substances were present in the almond oils. In several samples, the peaks for elaidic and oleic acids co-eluted, as well as the peaks for linoleic and linolelaidic acids. Thus, peak areas corresponding to these compounds were summed in the calculations. Fatty acid data was used for the calculation of several health lipid indices. The atherogenic index (AI) and the thrombogenic index (TI) were calculated using the equations proposed by Ulbricht and Southgate [47].

The hypocholesterolemic/hypercholesterolemic (h/H) index indicates the sample's potential to provide good over bad cholesterol and was calculated using the equation proposed by Santos-Silva et al. [48].

#### *3.5. Sensorial Analysis*

Raw, roasted, and blanched almond kernels were evaluated by a panel of 12 tasters from Departamento de Biologia e Ambiente (DeBA-ECVA), Universidade de Trás-os-Montes e Alto Douro in Portugal. The panelist was well trained and have participated in sensory tests from previous studies. The test took place between 4:00 pm and 5:00 pm in a room with regulated temperature and air pressure/flow. Almond kernels were presented to the tasters in white pyrex dishes. The testing environment and procedure/equipment were all following ISO 8589:2007. Three tasting sessions were carried out; the first, second and third sessions were done using raw, roasted, and blanched almonds respectively. A Quantitative Descriptive Analysis (QDA) was performed using 20 descriptors adapted from Civille et al. [49] and a structured scale from 1 (least intense) to 5 (most intense) for each descriptor (ISO 4121:2003).

#### *3.6. Statistical Analyses*

Data are presented as mean (f.w) of three replicates. Differences among means were determined by analysis of variance (ANOVA) using SPSS (Statistical Package for Social Sciences) 19.0 (IBM Corporation, New York, NY, USA). The fulfillment of ANOVA requirements, namely the normal distribution of the residuals and the homogeneity of variance was evaluated using the Shapiro–Wilk's test (n < 50) and the Levene's test, respectively. The Tukey test was used for the comparison of means which were considered different at a 5% significance level. A stepwise linear discriminant analysis (LDA) was performed to find the linear combination of parameters that best characterized raw, roasted, or blanched samples. The LDA involved the use of a combination of forward selection and backward elimination procedures for variable separation. Before selecting a new variable to be included in the model, it was ascertained that all previously selected variables were still significant which enabled identification of all possible significant variables. The Wilk's lambda test was applied for variable selection through verification of the significance of each canonical discriminant functions using the probabilities F = 3.84 to add and F = 2.71 to remove. To avoid overoptimistic data modulation, the model performance was assessed using a leave-one-out cross validation procedure. Principal component analysis (PCA) was performed by plotting all data in a multidimensional space using LDA significant variables as dimensions (factor scores). With this approach, the number of variables was reduced to a smaller number of newly derived variables (principal component or factors) that adequately summarized the original information, and highlighted underlying patterns in the data collected. Scree plots were used for retaining the most useful factors taking into consideration eigenvalues greater than one and internal consistencies of high Cronbach α values.

#### **4. Conclusions**

This study evaluated the effects of processing (roasting and blanching) on the levels of phenolics, flavonoids and fatty acids in four Portuguese (Casanova, Molar, Pegarinhos and Refêgo) and two

foreign (Ferragnès and Glorieta) almond cultivars, as well as on the antioxidant activities and sensorial characteristics of kernels. Antioxidant activities and levels of bioactive compounds were generally enhanced following roasting but reduced following blanching. The increased antioxidant activity upon roasting may at first seem unexpected, as heat could be envisaged to decrease (oxidize) phenolics. However, we interpret that roasting may either induce cell wall disruption allowing better antioxidant extraction, or may alternatively cause chemical alterations that result in heat-induced production of hyper-antiradical scavenging species. This might be linked to browning reaction s compounds, known to affect total phenolics quantification and antioxidant activities. Roasting and blanching reduced the ability of all cultivars to inhibit lipid oxidation. The fatty acid profiles of all cultivars were similar although raw kernels of Refêgo exhibited a high content of α-linolenic acid. Glorieta and Molar were characterized by a low and a high content of SFA respectively. Both roasting and blanching led to significant changes in the fatty acid profiles of almonds, similar for all cultivars with the exception of Refêgo, and the effect of blanching was stronger than that of roasting. Relative to health lipid indices, Pegarinhos and Molar had better responses to roasting and blanching than the other cultivars. Very few significant differences in cultivars and treatments relative to sensorial characteristics were found. The negative features bitter taste and bitter almond flavor were noticed for raw kernels of Molar and Pegarinhos. Roasting of Ferragnès led to kernels with a strong sweet almond flavor, while blanching positively affected the textural properties of Refêgo and Pegarinhos. The findings of this study shed light on the nutritive and eating qualities of raw and processed kernels from neglected Portuguese almond cultivars, and highlight the potential use of these cultivars in various food industries.

**Author Contributions:** Conceptualization, I.O.; laboratory work, I.O., S.A., P.G. and A.S. writing—Original draft preparation, I.O. writing—Review and editing, I.O., A.S.M., S.A., A.V., P.G., H.T. and B.G.; supervision, B.G., A.S.M; All authors have read and agreed to the published version of the manuscript.

**Funding:** Ivo Oliveira is grateful to FCT—Fundação para a Ciência e a Tecnologia (FCT) for the financial support by national funds FCT/MCTES in the Post-doctoral Fellowship SFRH/BPD/111005/2015. This work is supported by National Funds by FCT—Portuguese Foundation for Science and Technology, under the project UIDB/04033/2020.

**Conflicts of Interest:** The authors declare no conflict of interest.
