The triangular discriminative test was used to select tasters with the ability to recognize differences in the flavor intensity of beverages. The samples consisted of commercial strawberry aroma solutions (502223C, Firmenich®) at two concentrations (2.5 and 5.0 mL of aroma in 1 L distilled water, respectively). The test was performed in individual booths under fluorescent light. Each tester received three samples (two equal and one different) and was instructed to indicate on the analysis chart, the different sample from the three presented. The samples, coded with a three-digit number according to a complete block diagram, were served in capped tulip-type cups. The testers were asked to smell the solutions when removing the cap after slow stirring. The selected sensory team was composed only by the testers who were able to identify the difference in aroma intensity among the samples, with at least 75% of correct answers in the four repetitions given.

Mango and guava juices, respectively, were presented to the selected tasters at two different concentrations, to develop and train their olfactory memory in relation to the strong and weak intensity of the aromas, characteristic of these fruits. The two intensities offered were strong (200 mL water to 100 g fruit) and weak (50 mL of the above prepared juice diluted in 200 mL distilled water), and corresponded to the extremes of the 9 cm unstructured hedonic scale (Fig. 1).

The trained tasters were instructed to evaluate the aroma intensity characteristic of the aqueous fractions collected during the juice concentration process (Table 1), by indicating the intensity of the characteristic aroma of mango and guava in each fraction on a 9 cm unstructured scale (section 2.3.2; Fig. 1). The samples were composed of absorbent paper strips containing an aliquot (0.1 mL) of each analyzed aqueous fraction. The strips were coded with three-digit numbers and evaluated in individual booths.

The form of presentation was monadic and randomized according to a complete block diagram, to avoid first-order effects [24Meilgaard M, Civille GV, Carr BT. Sensory Evaluation Techiniques 1988; 281., 25Stone H, Sidell JL. Sensory Evaluation Practices 1985; 310.]. The time between the application of the aliquot of the aqueous fraction on the paper tape and the presentation to the tester was duly standardized. The testers assessed each of the samples in three replicates.

The testers who had at least 75% of correct answers in the triangular discriminative test were selected as the sensory team. Among the 120 testers participating in the selection process, only 12 were chosen. Of these, 2 were 100% correct, and 10 were 75% correct, and 11 evaluated the intensity of the mango and guava aromas in the aqueous fractions of the juices.

Fig. (2) Intensity of characteristic mango aroma of (A: 650 mmHg and B: 680 mmHg) and guava (C: 650 mmHg and D: 680 mmHg) in the aqueous fractions of the first collection made in different temperature conditions.

Fig. (3) (A) Comparison of the TIC of the assays 1 —, 2 —, 3 —, 4 —, 5 —, 6 —, 7 — ; (B) natural juice —, tests 1 —, 2 — , 3 — ; (C) natural juice —, tests 4 —, 5 —, 6 — , 7 —

Table 1 presents the intensity of the characteristic aromas of mango and guava in the aqueous fractions, expressed on the unstructured 9 cm scale. The maximum intensity of the mango aroma in all the trials was observed in the first fraction, with no significant difference between the vacuum and temperature conditions used in the process. It was also found that in all the tests, the samples for the first fraction significantly differed from the samples from the third fraction, indicating that there was a reduction of the volatile compounds as a function of the concentration time. In the experiments carried out at 650 mmHg, the second fraction did not differ significantly from the first one, suggesting that this fraction also carried a considerable amount of volatiles characteristic of the fruit aroma (Table 1). Also, the intensity of the aroma in the two the first fractions was not affected by the increase in process temperature. At 680 mmHg, the second fraction did not differ from the first at the operating temperatures of 50 and 65 °C, but at 45 and 55 °C, this behavior was not observed (Table 1).

Considering the concentration of guava juice, the maximum intensity of the fruit aroma was perceived in the first aqueous fraction under all the vacuum and temperature conditions evaluated. At 650 mmHg and 55 and 65 °C, there was a significant reduction of the guava aroma intensity in the second and third fractions collected (Table 1), showing a decrease in the concentration of guava volatiles, dependent on the time of the concentration process. However, at 680 mmHg and 50, 55 and 65 °C, the significant difference emerged only in relation to the first fraction, indicating that at the higher vacuum, the volatile carrier was mainly present in this first fraction (Table 1).

The influence of temperature on the aroma intensity characteristic of mango and guava in the first fractions of each assay can best be seen in Fig. 2. The intensity of the characteristic mango aroma was highest at 650 mmHg and 55 °C (Fig. 2A), and at 680 mmHg at 45 and 55 °C (Fig. 2B). For guava, the aqueous fraction with the highest fruit aroma intensity also appeared at 55 °C and 650 mmHg (Fig. 2C). However, when the vacuum was increased (680 mmHg) the aqueous fraction with the highest aroma intensity was obtained at 65 °C, but this difference was not significant (Table 1). Nonetheless, when recovering the aroma of fractions from fruit juice concentration, this information becomes relevant since the higher the quantity of volatiles in the aqueous fraction, the easier it is to try and separate them, by employing other methods.

The results of the sensorial analysis highlighted the importance of the process parameters since for mango juice, the highest intensity of the fruit aroma presented in the first two fractions, and, for guava juice, the first fraction concentrated the characteristic volatiles of the fruit. These findings direct the steps of collecting the volatiles and the subsequent incorporation of the natural aroma of the fruits into juices.

(Fig. 3A) compares the TIC of the natural mango juice with that of the aqueous fractions from the concentration assays, under the different operating conditions. In the amplification presented, it seemed most of the compounds present in the natural juice were not in the aqueous fractions (Figs. 3B and 3C). However, when integrating the peaks, it was evident that the components present in the natural juice were also present in these fractions, albeit at lower intensity when comparing the area percentages of the peaks with those of the juice (Table 2).

α-Pinene and 3-carene were the major components of both the natural juice and the aqueous fractions, with retention times of 6.93 and 9.29 min, respectively, (Fig. 3). For a better visualization of the TIC, (Fig. 4) displays each of the peaks on a magnified scale.

The components observed between peaks 4 and 8 may be related to contaminants of the solvent, despite its chromatographic grade (Fig. 4). Peak 9 confirmed the presence of the monoterpenes α- and β-pinene, β-myrcene, α-phellandrene, 3-carene, limonene, and γ- and α-terpinene, which are characteristic of volatiles present in mangoes of several species [28Bender RJ, Brecht JK. Aroma volatiles of mature-green and tree-ripe “Tommy Atkins” mangoes after controlled atmosphere vs. air storage. HortScience 2000; 35(4): 684-6.
[http://dx.doi.org/10.21273/HORTSCI.35.4.684] -30Pino JA, Mesa J, Muñoz Y, Martí MP, Marbot R. Volatile components from mango (Mangifera indica L.) cultivars. J Agric Food Chem 2005; 53(6): 2213-23.
[http://dx.doi.org/10.1021/jf0402633] [PMID: 15769159] ]. According to Bender et al. [28Bender RJ, Brecht JK. Aroma volatiles of mature-green and tree-ripe “Tommy Atkins” mangoes after controlled atmosphere vs. air storage. HortScience 2000; 35(4): 684-6.
[http://dx.doi.org/10.21273/HORTSCI.35.4.684] ], 3-carene is the main monoterpene present in Tommy Atkins mango, as also evidenced in the current work. Two ketones (peaks 21 and 22) and two sesquiterpenes (peaks 24 and 25) were also noted (Table 2).

The percentage of the total relative area of the peaks listed in Table 2 for the volatiles identified in the natural mango juice represented approximately 82% of the total components present. The sum of the percentage of these areas in the aqueous fraction fell considerably to between ~22 and 31%.

Considering only the percentage of the relative area of the peaks, since the concentration was not determined, it was noticed that the main components in the natural mango juice complemented those in the aqueous fractions. Among these constituents, were the monoterpenes, 3-carene, α-pinene, γ-terpinene and limonene. The data also revealed a considerable increase in the intensity of the ketone peaks (21 and 22) in the aqueous fraction. Considering that these compounds have higher molecular weight than monoterpenes, and if all volatiles have lowered boiling temperatures at reduced pressures, these ketones are proposed to be volatiles derived from the degradation of some component present in the juice during the concentration (Table 2).

Pandit et al. [16Pandit SS, Chidley HG, Kulkarni RS, Pujari KH, Giri AP, Gupta VS. Cultivar relationships in mango based on fruit volatile profiles. Food Chem 2009; 114(1): 363-72.
[http://dx.doi.org/10.1016/j.foodchem.2008.09.107] ] characterized the major chemical classes in mango as alcohols, aldehydes, monoterpene hydrocarbons, oxygenated monoterpenes, sesquiterpene hydrocarbons, oxygenated sesquiterpenes, lactones, ketones and non-terpene hydrocarbons. In a study conducted by San et al. [31San AT, Joyce DC, Hofman PJ, et al. Stable isotope dilution assay (SIDA) and HS-SPME-GCMS quantification of key aroma volatiles for fruit and sap of Australian mango cultivars. Food Chem 2017; 221: 613-9.
[http://dx.doi.org/10.1016/j.foodchem.2016.11.130] [PMID: 2797 9249] ], the main volatile aroma compounds in the ripe fruit of two different mango varieties (B74 and Kensington Pride) were detected at mean concentrations (μg/L) of 0.6 and 0.5 (ethylene octanoate), 6 and 54 (hexanal), 12 and 14 (p-cymene), 23 and 43 (α-terpinene), 29 and 31 (2-carene), 216 and 264 (limonene), 929 and 383 (3-carene), and 11,272 and 19,719 (α-terpinolene), respectively. Furthermore, α-terpinolene, 3-carene, p-cymene and limonene were demonstrated to be odor-active volatiles in all the mangoes studied, and α-terpinolene, 3-carene and hexanal were the most odor-active.

Table 4
Vitamin C present in mango and guava juices before and after concentration and its loss percentage at each assay.

Benevides et al. [32Benevides CMJ, Bezerra MA, Pereira PAP, Andrade JB. HS-SPME/GC–MS analysis of VOC and multivariate techniques applied to the discrimination of Brazilian varieties of mango. Am J Anal Chem 2014; 5(3): 157-64.
[http://dx.doi.org/10.4236/ajac.2014.53019] ] determined the volatile compounds in three mango varieties (Tommy Atkins, Rosa and Espada) and found that the main classes of compounds were esters and terpenes. Among the identified compounds, only α-pinene, β-myrcene and caryophyllene were common to all three varieties. The relative abundance of the volatile organic compounds varied among the samples of the three varieties. For instance, ethyl butanoate (17.7% relative abundance) was recorded only in the Tommy Atkins mango. 3-Carene was detected in the Tommy Atkins (51% relative abundance) and Espada (17% relative abundance) varieties but not Rosa, and the relative amounts of α-pinene were 2.8% (Espada), 17.9% (Tommy Atkins) and 31.3% (Rosa).

(Fig. 5) compares the TIC of the natural guava juice with those of the aqueous fractions from trials, 2, 3 and 4 (Fig. 5A), and 5, 6 and 7 (Fig. 5B). Only when viewing the integrated peaks of the TIC (Table 3), was it apparent that the compounds present in the natural juice also appeared in these fractions, albeit at lower intensity when comparing the peak area percentages. According to the TIC of the aqueous fractions collected in the tests (Figs. 4 and 5), the most abundant volatile compounds were common to all the samples.

To better illustrate the preliminary identification of the compounds detected in guava juice and their correlation with the compounds found in the obtained aqueous fractions, Fig. 6 shows the augmented TIC of guava juice. The GC/MS identification of the compounds corresponding to each of these numbered peaks, the area percentage of the same compounds in all assays, their respective retention times and their Kovats indices, are shown in Table 3.

Among the volatiles present in both the juice and aqueous fractions, carbonyl compounds, alcohols, sesquiterpenes, esters and ketones, all characteristic of the volatiles known to occur in various guava species, were observed [33Jordán MJ, Margaría CA, Shaw PE, Goodner KL. Volatile components and aroma active compounds in aqueous essence and fresh pink guava fruit puree (Psidium guajava L.) by GC-MS and multidimensional GC/GC-O. J Agric Food Chem 2003; 51(5): 1421-6.
[http://dx.doi.org/10.1021/jf020765l] [PMID: 12590492] -35Yen GC, Lin HT. Changes in volatile flavor components of guava juice with high-pressure treatment and heat processing and during storage. J Agric Food Chem 1999; 47(5): 2082-7.
[http://dx.doi.org/10.1021/jf9810057] [PMID: 10552500] ].

As seen in the mango analysis, the compounds abundant in high proportions in the guava juice, such as n-hexanal or eucalyptol, also migrated at high proportions into the aqueous fractions. Other volatiles, such as (E)-2-hexenal, (Z)-3-hexen-1-ol, 1-hexanol, 3,4-dimethylacetophenone and 2,5-dimethylacetophenone were noted in higher proportions in the aqueous than juice fraction, demonstrating that such compounds are entrained within the water vapor and thereby enriched in the aqueous fraction. If these volatiles are key components in the constitution of the juice aroma, any decrease in their quantities during concentration of the juice will impair its flavor (taste and odor).

Interestingly, the sesquiterpenes, such as β- and γ-caryophyllene, important in the characterization of tropical fruit flavor, migrated to the aqueous fraction, but were detected in smaller proportions than those present in the original juice. The relative percentage of the total peaks (Table 3) for the volatiles detected in the natural juice of guava, represented about 89% of the total volatiles captured on Porapak-Q. The sum of the percentage of these areas in the aqueous fraction remained close, at approximately 71–88%. Considering only the percentages of the relative areas of the peaks, since the concentration was not determined, it was confirmed that the components found in greater intensity in the natural juice of guava correlated with those found in the aqueous fractions.

Soares et al. [8Soares FD, Pereira T, Marques MOM, Monteiro AR. Volatile and non-volatile chemical composition of the white guavafruit (Psidium guajava) at different stages of maturity. Food Chem 2007; 100(1): 15-21.
[http://dx.doi.org/10.1016/j.foodchem.2005.07.061] ] found the esters cis-3-hexenyl acetate (21.78%) and trans-3-hexenyl acetate (17.80%) were the most abundant in mature white guavas (P. guajava) of cultivar Cortibel. Other important compounds were the sesquiterpenes caryophyllene (12.96%), α-humulene (7.85%) and β-bisabolene (5.46%). Moon et al. [36Moon P, Fu Y, Bai J, Plotto A, Crane J, Chambers A. Assessment of fruit aroma for twenty-seven guava (Psidium guajava) accessions through three fruit developmental stages. Sci Hortic (Amsterdam) 2018; 238: 375-83.
[http://dx.doi.org/10.1016/j.scienta.2018.04.067] ] observed that hexane, (E)-2-hexenal and (E)-caryophyllene, were common at all maturation stages of guavas analyzed, and (Z)-3-hexenyl acetate, ethyl butanoate and ethyl octanoate predominated in mature guavas.