Tangerines Cultivated on Madeira Island—A High Throughput Natural Source of Bioactive Compounds

Tangerines (Citrus reticulata) are popular fruits worldwide, being rich in many bioactive metabolites. The setubalense variety cultivated on Madeira Island has an intense aroma easily distinguishable from other tangerines, being traditionally used to enrich several foods and beverages. Nonetheless, setubalense volatile composition has never been characterized, and we aimed to unveil the bioactive potential of peels and juices of setubalense tangerines and compare them with the murcott variety grown in Portugal mainland. Using headspace solid-phase microextraction coupled to gas chromatography mass spectrometry (HS-SPME/GC-MS), we identified a total of 128 volatile organic metabolites (VOMs) in the juice and peels, with d-limonene, γ-terpinene, β-myrcene, α- and β-pinene, o-cymene, and terpinolene, the most dominant in both cultivars. In contrast, setubalense juices are richer in terpenes, many of them associated with health protection. Discriminant analysis revealed a pool of VOMs, including β-caryophyllene and E-ocimene, with bioactive properties able to differentiate among tangerines according to variety and sample type (peel vs. juice). This is the first report on the volatile composition of setubalense tangerines grown on Madeira Island revealing that its pungent aroma is constituted by secondary metabolites with specific aroma notes and health properties. This is strong evidence of the higher nutraceutical value of such fruit for the human diet.


Introduction
Citrus fruits are one of the world's major fruit crops, being produced in many countries of the tropical, subtropical, and temperate borderlines [1]. According to the recent data released by the United States Department of Agriculture (USDA), citrus production in 2019/2020 is estimated to be around 92 million metric tons, representing oranges and tangerines 46.1 and 31.6 million tons, respectively [2]. Citrus fruits are very rich in secondary metabolites with high nutraceutical value, such as vitamin C, folate, flavonoids, coumarins, limonoids, terpenoids, and carotenoids [3]. Some of these metabolites, namely, terpenes, are volatile organic compounds (VOMs), giving the fruit a rich, pungent and distinctive aroma [4,5]. For this reason, citrus fruits and by-products have very important applications, spanning the food industry, cosmetics, and medicine [4]. Moreover, citrus VOMs also constitute a valuable tool for the identification and differentiation of cultivars, hybrids and genotypes [1,4,5]. Citrus fruits are a very heterogeneous group, with dozens of citrus species and varieties, being Citrus reticulata one of the most popular. The fruits commonly known as mandarins and tangerines are one of its varieties [5]. Tangerines combine the fresh and acidic notes of other citrus

Gas Chromatography-Mass Spectrometry Analysis (GC-MS)
The analysis was carried out on an Agilent 6890N gas chromatograph system (Agilent Technologies, Palo Alto, CA, USA), equipped with a BP-20 polar column, and coupled to an Agilent 5975 quadrupole inert mass selective detector. The full settings and procedure were described previously by Figueira, Porto-Figueira, Pereira, and Câmara [11]. Briefly, trapped VOMs were loaded in the GC-MS using a 6-min splitless injection using He (1.0 mL/min) and separated in BP20-fused silica using a temperature gradient starting at 45 • C (held for 2 min), followed by a temperature gradient (2 • C/min) up to 90 • C, held 3 min, then another increase (3 • C/min) up to 160 • C, held for 6 min, and finally from 160 to 220 • C (6 • C/min) and held at 220 • C for 15 min. The resulting chromatograms were processed using the Enhanced ChemStation software for GC-qMS (Agilent Technologies, Palo Alto, CA, USA). The identification of the VOMs involved the comparison between the GC retention times (RT) of the chromatographic peaks with those of authentic standards, when available, under the same conditions, as well as by the determination of the KI of each identified volatile using a C 8 -C 20 n-alkanes series solution. Each sample was analyzed in triplicate.

Multivariate Statistical Analysis
The volatomic data generated was processed with the MetaboAnalyst 4.0 web-based tool [13] (Xia Lab, McGill University, Montreal, QC, Canada) and according to Figueira, Porto-Figueira, Pereira, and Câmara [11]. Briefly, the raw data were normalized (IS ratio correction, sample median, data transformation by cubic root and data scaling by autoscaling). The n, variance (ANOVA, p < 0.05) and partial least square discriminant analysis (PLS-DA) were used for variable reduction. PLS-DA reduces the size of the data matrix by eliminating redundant variables (VOMs), thus defining the set of volatiles which define the best separation among the different groups analyzed. Ward's linkage algorithm and Euclidean distance analysis were used to calculate the ratio of the VOMs, and the resulting metabolic alterations (top VOMs with VIP (variable importance in projection) > 1) were subjected to used HCA analysis to depict distinct clustering patterns among the studied groups.

Volatomic Profile of Tangerines
The volatile profiles from the two tangerines varieties analyzed are quite different, being the setubalense juice clearly richer than murcott in terms of number and intensity of VOMs characterized ( Figure 1). Regarding the peels, we observe that their volatile composition is much richer (number and intensity of VOMs) than the juices, while minor differences have been observed between varieties ( Figure 1). Overall, 128 different VOMs have been identified, from which 109 were in the murcott and 117 were in the setubalense variety. The detailed list of all VOMs identified and respective experimental data, including the retention time and relative peak area is available in Supplementary Table S1. The contribution of each VOM for the total volatile fraction expressed as relative peak area was calculated according to Equation (1): Peak area of analyte /Peak area of internal standard. (1) Foods 2020, 9, x FOR PEER REVIEW 4 of 13 During sample preparation, we observed a stronger and distinctive smell for the setubalense tangerines. This agrees with the fact that the sum of the relative peak areas of the 75 VOMs identified in setubalense juice is 5 times higher than the 56 VOMs identified in murcott juices. In contrast, the sum of the relative peak areas obtained for the 89 VOMs identified in setubalense peels is only 20% higher than found for the 91 VOMs identified in murcott peels samples (Supplementary Table S1).  Supplementary Table S1. Peak number identification: 9-ethanol, 15-1-penten-3-one, 16-α-pinene, 17-α-thujene, 22-hexanal, 23-β-pinene (isomer 1), 24-β-pinene (isomer 2), 28-β-myrcene, 29-α-terpinene, 30-d-limonene, 34-γ-terpinene, 37-o-cymene, 38-terpinolene, IS-internal standard (3-octanol), 64-decanal, 71-linalool, 94-α-terpineol, 104-α-farnesene, 122-dimethyl anthranilate, 124-thymol.
During sample preparation, we observed a stronger and distinctive smell for the setubalense tangerines. This agrees with the fact that the sum of the relative peak areas of the 75 VOMs identified in setubalense juice is 5 times higher than the 56 VOMs identified in murcott juices. In contrast, the sum of the relative peak areas obtained for the 89 VOMs identified in setubalense peels is only 20% higher than found for the 91 VOMs identified in murcott peels samples (Supplementary Table S1). Monoterpenes hydrocarbons are by far the functional class more abundant in all samples analyzed, representing over 93% and 84% of the volatile fractions of tangerine juices and peels, respectively ( Figure 2). Such abundance is mainly due to d-limonene, followed by γ-terpinene, β-myrcene, β-pinene, o-cymene, α-pinene, and terpinolene, which are the most abundant VOMs identified in all tangerine samples (Supplementary Table S1). This suggests that the VOMs with lower abundance should have an important contribution for the distinct profile of setubalense juices and peels in comparison with murcott. Further olfactometry analysis would be very interesting to clarify this question.
Supplementary Table S1). Since terpenes result from the Methylerythritol 4-phosphate (MEP) and the mevalonate (MAV) pathways [14,15] (Figure 3), this result suggests a lower activity of these pathways in the murcott relative to the setubalense variety, particularly in the cytosol where the final steps of sesquiterpenes biosynthesis occur. In opposition, ethyl isobutyrate, propyl acetate, and ethyl 2butenoate were exclusively identified in murcott juices. The closely related mandarin juice contains nine volatiles that are considered as core aromatic volatiles, namely, linalool, α-terpineol, terpinen-4ol, nonanal, decanal, carvone, limonene, α-pinene, and β-myrcene [6,7]. All these VOMs were identified in the tangerine samples analyzed in this work. Beyond this, more VOMs, namely, terpenes, were identified in this study in comparison to a recent report [16].

Volatomic Profile of Tangerine Peels
When compared to the juice samples, the volatomic profile of tangerine peels is more complex, being the sum of relative peak areas 50 times higher in the murcott variety and 1.25 times higher in the setubalense variety than the respective juice samples. Regarding the number of VOMs, there was also an increment in the number of VOMs identified in the peels. Overall, 89 VOMs were identified in murcott peels (56 in the juices), whereas in setubalense variety, the variation was lower (87 vs. 75 VOMs in peels and juices, respectively; Supplementary Table S1). As shown in Figure 2, these differences are mainly due to aldehydes and oxygenated terpenes, whose number of VOMs triplicates from the juices to the peels. The variation in the number of oxygenated terpenes found in the peel samples, when compared to juices samples, can be easily explained by the interaction with the atmospheric air that favors the

Volatomic Profile of Tangerine Juices
As can be observed in Figure 2, sesquiterpene hydrocarbons and oxygenated terpenes are clearly more abundant in setubalense juices, both in terms of the number of VOMs identified, as in relative peak areas for each family of compounds. In contrast, the relative peak areas for higher alcohols and esters are much higher in murcott. Ethyl acetate, for instance, is three times more abundant in murcott comparatively to setubalense juices. Beyond the relative peak areas, 28 VOMs were exclusively identified in the setubalense variety juices (7 monoterpenes, 8 sesquiterpenes, and 10 terpenoids; Supplementary  Table S1). Since terpenes result from the Methylerythritol 4-phosphate (MEP) and the mevalonate (MAV) pathways [14,15] (Figure 3), this result suggests a lower activity of these pathways in the murcott relative to the setubalense variety, particularly in the cytosol where the final steps of sesquiterpenes biosynthesis occur. In opposition, ethyl isobutyrate, propyl acetate, and ethyl 2-butenoate were exclusively identified in murcott juices. The closely related mandarin juice contains nine volatiles that are considered as core aromatic volatiles, namely, linalool, α-terpineol, terpinen-4-ol, nonanal, decanal, carvone, limonene, α-pinene, and β-myrcene [6,7]. All these VOMs were identified in the tangerine samples analyzed in this work. Beyond this, more VOMs, namely, terpenes, were identified in this study in comparison to a recent report [16].
oxidation of many metabolites, including VOMs. Similarly, 3 ketones and 3 furans, grouped as "others" in Figure 2, exhibit a robust predominance in the volatile composition of the peels. Regarding this, 1-penten-3-one accounts for half of the relative peak area of the "others" family of the tangerine peel samples from the murcott variety. In contrast, esters almost disappear, being only ethyl acetate identified in the peels of both varieties. Overall, 33 more VOMs were identified in the murcott variety and more 12 VOMs in the setubalense variety.

Volatomic Profile of Tangerine Peels
When compared to the juice samples, the volatomic profile of tangerine peels is more complex, being the sum of relative peak areas 50 times higher in the murcott variety and 1.25 times higher in the setubalense variety than the respective juice samples. Regarding the number of VOMs, there was also an increment in the number of VOMs identified in the peels. Overall, 89 VOMs were identified in murcott peels (56 in the juices), whereas in setubalense variety, the variation was lower (87 vs. 75 VOMs in peels and juices, respectively; Supplementary Table S1). As shown in Figure 2, these differences are mainly due to aldehydes and oxygenated terpenes, whose number of VOMs triplicates from the juices to the peels.
The variation in the number of oxygenated terpenes found in the peel samples, when compared to juices samples, can be easily explained by the interaction with the atmospheric air that favors the oxidation of many metabolites, including VOMs. Similarly, 3 ketones and 3 furans, grouped as "others" in Figure 2, exhibit a robust predominance in the volatile composition of the peels. Regarding this, 1-penten-3-one accounts for half of the relative peak area of the "others" family of the tangerine peel samples from the murcott variety. In contrast, esters almost disappear, being only ethyl acetate identified in the peels of both varieties. Overall, 33 more VOMs were identified in the murcott variety and more 12 VOMs in the setubalense variety.
Overall, bioactivity is recognized not only to the major VOMs identified in the tangerines but also to other present in lower abundance, as thymol and dimethyl anthranilate, whose benefits to human health are widely known [31,32] (Table 1). Thymol and dimethyl anthranilate have also been reported as having an important contribution to the mandarin aroma, being used in synthetic tangerine flavour [7,8].
According to Ladanyia and Ladanyia [33], the distinctive notes in the aroma of the Mediterranean mandarin oil were due this two VOMs, accounting dimethyl anthranilate to 0.85% and thymol to 0.08% of the volatomic profile. In this work, dimethyl anthranilate represents 1.18% and 2.35% and thymol represents 0.87% and 0.58% of the volatile profile of the peels of setubalense and murcott varieties, respectively. Such abundance in thymol and dimethyl anthranilate supports the use of setubalense tangerine as a potential and interesting alternative to the thyme oils, which are often associated to some allergic reactions, even when it is used diluted [32]. Another interesting VOM identified in this work is cosmene. This compound represents almost 20% of the oil obtained from the roots of Eupatorium adenophorum Spreng (Asteraceae) [34]. This native plant from Central America, commonly known as Mexican devil became invasive worldwide, being used in the folk medicine for its bioactive properties (antimicrobial, antiseptic, etc. [34]). In fact, Ahluwalia, Sisodia, Walia, Sati, Kumar, and Kundu [34] reported antimicrobial, antioxidant, and phytotoxic properties for the root's oils of this plant and such effect will be certainly elicit by cosmene given its high abundance in the roots extracts. Legend: 1 VOMs peak number in the chromatograms ( Figure 1); 2 ratio between the relative peak areas obtained for setubalense and murcott tangerines; 3 potential bioactive effect indicates the type of bioactive effects reported for each of the VOMs referred in the table, mur-murcott cultivar, set-setubalense cultivar; 4 strong evidence for the indicated bioactivity, } reported bioactive effect.

Classification of Tangerine Samples
In order to evaluate the potential of the volatile fingerprints obtained and discriminate tangerine samples according to the variety and sample type (juice vs. peel), a statistical analysis of the volatomic data matrix was performed using MetaboAnalyst 4.0 web-based tool [13]. The complexity of the data was reduced through normalization, as described in the experimental section. The n univariate statistical analysis (ANOVA, p ≤ 0.05) was carried out to verify the significant statistical differences among the concentrations of VOMs in each sample group (p < 0.05; Supplementary Table S2). Multivariate statistical analysis (MVSA) using PLS-DA was then performed to assess if there were significant VOMs signatures between each group. The results obtained show four different clusters discriminating each of the groups under study (Figure 4). from the roots of Eupatorium adenophorum Spreng (Asteraceae) [34]. This native plant from Central America, commonly known as Mexican devil became invasive worldwide, being used in the folk medicine for its bioactive properties (antimicrobial, antiseptic, etc. [34]). In fact, Ahluwalia, Sisodia, Walia, Sati, Kumar, and Kundu [34] reported antimicrobial, antioxidant, and phytotoxic properties for the root's oils of this plant and such effect will be certainly elicit by cosmene given its high abundance in the roots extracts.

Classification of Tangerine Samples
In order to evaluate the potential of the volatile fingerprints obtained and discriminate tangerine samples according to the variety and sample type (juice vs. peel), a statistical analysis of the volatomic data matrix was performed using MetaboAnalyst 4.0 web-based tool [13]. The complexity of the data was reduced through normalization, as described in the experimental section. Then univariate statistical analysis (ANOVA, p ≤ 0.05) was carried out to verify the significant statistical differences among the concentrations of VOMs in each sample group (p < 0.05; Supplementary Table S2). Multivariate statistical analysis (MVSA) using PLS-DA was then performed to assess if there were significant VOMs signatures between each group. The results obtained show four different clusters discriminating each of the groups under study (Figure 4).  As Figure 4A shows, the first component explains 50.7% of the variance and separate the tangerine "juice" from "peel". The second component contributes for 24.7% of the total variance of the model and separate the tangerine variety murcott from setubalense. Figure 4B presents the variables that are most contributed for the differentiation of the tangerines by variety and type of sample. Accordingly, ethyl 2-methyl butanoate, ethyl propionate, and 2-methyl-furan are more associated to setubalense juice; ethyl acetate, β-myrcene, and β-phellandrene to murcott juice; α-farnesene, α-sinensal, and camphor more associated to setubalense peel; and cis-β-terpineol, cosmene, and thymol to murcott peel.
Hierarchical cluster analysis (HCA) was also performed using the 64 most significant VOMs identified in tangerine samples, as described in Section 2.5. This strategy allows a better identification of the inherent clustering patterns between each variety, in complementarity with the statistical analysis carried out previously. The result of this treatment can be visualized in the heatmap plot ( Figure 5).
analysis carried out previously. The result of this treatment can be visualized in the heatmap plot ( Figure 5).
Overall, the set of variables that contributed to the differentiation of the tangerines by variety have the potential to be used for product identification and authenticity. Furthermore, tangerines of the setubalense variety contain higher amounts of bioactive analytes than murcott tangerines (Table 1). Therefore, this bioactive signature supports higher health benefits for the setubalense tangerines grown in Madeira Island, as can be used to the discrimination of the two varieties analyzed.  Overall, the set of variables that contributed to the differentiation of the tangerines by variety have the potential to be used for product identification and authenticity. Furthermore, tangerines of the setubalense variety contain higher amounts of bioactive analytes than murcott tangerines ( Table 1). The refore, this bioactive signature supports higher health benefits for the setubalense tangerines grown in Madeira Island, as can be used to the discrimination of the two varieties analyzed.

Conclusions
In this study, HS-SPME-GC/MS was used to characterize for the first time the volatile composition of a tangerine variety cultivated in Madeira Island (setubalense) and compare it with the murcott variety grown in mainland. This comparison involved the juice and peels of both varieties and allowed the identification of 128 VOMs, 107 VOMs in murcott, and 115 VOMs in setubalense variety, which is considerably higher than most of the previous reports using the same methodology to analyze different tangerine varieties. To the best of our knowledge, only the tangerine hybrid 9-4 × Blood4x analyzed by Miyazaki, Plotto, Goodner, and Gmitter [5] presented a similar number of VOMs (118 VOMs) than the setubalense variety. The volatile composition of the peels is more complex than the juices, and the setubalense distinctive aroma corresponds to a higher abundance of common VOMs, as most of them have been previously reported with important bioactive properties. d-limonene, γ-terpinene, β-myrcene, β-pinene, o-cymene, α-pinene, and terpinolene are among the most abundant VOMs, but setubalense tangerines are also very rich in thymol, making this fruit a very interesting alternative to thyme oils.
The uniqueness of setubalense tangerines was shown through a statistical analysis that differentiate this variety from the common murcott grown in Portugal mainland. Accordingly, ethyl 2-butenoate, (E)-2-decenal, cycloheptane, and carvone associated to the murcott variety and α-himachalene, β-santalene, α-selinene, and α-sinensal associated to the setubalense variety, were the VOMs that most contributed to the differentiation of two tangerine varieties.

Conflicts of Interest:
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.