Characterization of Volatile Organic Compounds in ‘Rossa di Tropea’ Onion by Means of Headspace Solid-Phase Microextraction Gas Chromatography–Mass Spectrometry (HS/SPME GC–MS) and Sensory Analysis

: Background: Plant viral infections induce changes in the host plant, which can potentially impact composition, organoleptic properties, and storability characteristics of plant products. In particular, onion odor and ﬂavor are determined mainly by volatile organic compounds, and changes upon infection with onion yellow dwarf virus may deeply inﬂuence these characters. Methods: A time-course study of volatile organic compounds in onion yellow dwarf virus-infected versus healthy ‘Rossa di Tropea’ onion bulbs was performed using headspace solid-phase microextraction gas chromatography–mass spectrometry; sensory analysis performed at marketability stage of onion production was used to correlate such changes to the taste characteristics perceived by consumers. Results: Volatile organic compounds regulated in infection conditions were identiﬁed, mainly belonging to mono- and poly-sulﬁdes classes. The most abundant compounds in the analyzed samples were propyl disulﬁde, allyl-isopropyl disulﬁde, and propanethiol; signiﬁcantly different concentrations were observed for 7 out of 11 VOCs in virus-infected compared to healthy bulbs. Statistical analysis based on a partial least squares discriminant analysis model and hierarchical cluster analysis allowed us to cluster samples based on phytosanitary status and storage time and to identify the most responsible compounds for such classiﬁcation. Conclusions: Onion yellow dwarf virus infection induces changes in volatile organic compounds in onion during storage. The impact of such regulated compounds on ‘Rossa di Tropea’ onion odor and ﬂavor and correlation with sensory analysis are discussed.


Introduction
In recent years, globalization and international trade have led to market saturation, particularly for agricultural food products, with a flattening of varieties and diversity due to the demanding global scale intensive productions [1]. Nevertheless, the recently growing consumers' awareness of typical and local products represents a promising opportunity for relaunching and developing niche agricultural sectors [2], particularly in countries such as Italy, characterized by wide biodiversity in terms of landscapes, production areas, and local varieties/landraces [3]. The differentiation of local products includes several variables, such as quality, taste extolling, freshness, "rural" practices, and area of production, leading to recovering and requalification of old cultivations that previously could not face the global market [4,5]. Among them, onion, Allium cepa L., member of genus Allium, one of the most important and representative genera of the Amaryllidaceae family and widely This study aims to investigate the effect of a viral pathogen such as OYDV on the overall organoleptic properties in 'Rossa di Tropea' onion, mainly focusing on the determination of VOCs profile and their variations briefly after the inoculation and in postharvest storage.

Plant Materials and Experimental Setup
Plant material (i.e., onion bulbs) was obtained from an experimental trial established as follows. Seeds of 'Rossa di Tropea' onion (biotype locally named Mezza campana), collected from healthy plants as assessed by specific OYDV Real-Time RT-PCR test [33], were used as starting material. All the procedures performed during the different trial phases were carried out in an insect-proof screenhouse at "Mediterranea" University of Reggio Calabria, Italy. The experimental design was based on randomized blocks including two treatments (healthy-h and infected-i), each applied in three parcels. Overall, the trial was composed of 180 plants (each plant growing in one pot), 30 pots for each parcel, and six parcels, labeled from A to F. The cultivation practices mimicked those commonly performed in the open fields by farmers in 'Rossa di Tropea' onion production area. In particular, water and nutrient supplies were assured by daily irrigation, one round every 12 h (20 L/m 2 per h), 100 g/pot of diammonium phosphate in granules, respectively. Additionally, fungicide treatments were performed by Rizolex ® and Signum ® application, one pre-transplant (3 g/m 2 ) and two post-transplant, in a range of 10 days (15 g/L), respectively. Three months after the transplant, an artificial inoculation was carried out in parcels A, D, and E, starting from fresh tissue material infected by an OYDV isolate, following previously described procedures [21]. For this study, three analytical time points were sampled, mirroring the 'Rossa di Tropea' onion bulb production steps: at harvesting time (t 0 ), i.e., at the complete development of plants and bulbs; an experimental check timepoint (t 1 ) during postharvest storage, i.e., 1 month after harvesting; and after storage (t 2 ), i.e., 3 months after harvesting, when the bulbs reach the marketable stage, corresponding to 2, 3, and 5 months post-inoculation, respectively. It has to be underlined that the middle sampling time t 1 was then observed to have no biological nor statistical significance for the scope of this work; hence, only results of the samples at harvesting and at the end of postharvest storage (i.e., t 0 and t 2 ) are reported. Each time point was represented by collecting three biological replicates (Table 1) sharing the same morphological features as bulb dimension (i.e., diameter) and weight (data collected only at t 0 for bulbs sampled at t 0 , both at t 0 and t 2 for bulbs sampled at t 2 ).

Phytosanitary Status Assessment
To ascertain sanitary status of infected (i) and healthy (h) plants after inoculation, leaf samples collected from all the 180 plants in the experimental trial were assayed by DAS-ELISA (BIOREBA, Reinach, Switzerland) and RealTime RT-PCR [33]. The sampling and analyses were performed 21 days post inoculum (d.p.i.). The infection rate was about 80% of inoculated plants; only bulbs from confirmed healthy and OYDV-infected plants in the respective parcels were harvested and sampled for downstream analyses.

VOCs Analysis by HS/SPME GC-MS
A relatively recent technique called headspace solid-phase microextraction (HS-SPME) has been used to evaluate VOCs changes induced by OYDV infection. The HS-SPME technique is solvent-free and time-efficient, and it consists of immersing a thin fused-silica fiber in the headspace of a liquid or solid sample.
This fiber, coated with an extracting phase, can absorb several analytes (the class of metabolites absorbed depends on the fiber used), which are successively desorbed into the gas chromatography-mass spectrometry (GC-MS) apparatus for the analysis [34].
The HS/SPME GC-MS analysis was carried out as previously reported by Araniti et al. [35] with some modifications.
Briefly, 1 g of plant material, per sample and replicate (n = 3), was sealed in a 20 mL vial and allowed to equilibrate for 20 min at room temperature. Successively, the SPME grey fiber (StableFlex, divinylbenzene/Carboxen on polydimethylsiloxane coating; 50/30 µm coating, Supelco-Merck, Darmstadt, Germany) was exposed to plant VOCs for 20 min at room temperature.
The VOCs were identified using a gas chromatograph apparatus (G-Trace 1310 Thermo Fisher, Milan, Italy) coupled with a single quadrupole mass spectrometer (ISQ LT Thermo Fisher, Milan, Italy). The capillary column was a TG-5MS 30 m × 0.25 mm × 0.25 µm, and helium (purity 6.0) was used as carrier gas with a flow of 1 mL/min. Samples were injected in a split mode with a split ratio of 60. Injector and source were set at the temperature of 200 • C and 260 • C, respectively. The temperature ramp used was settled as follows: 7 min 45 • C, 45-80 • C rate 10 • C per min, 80-200 • C rate 20 • C per min, 3 min 200 • C. Mass spectra were recorded in electronic impact (EI) mode at 70 eV, scanning the 45-500 m/z range. Compound identification was carried out, comparing the relative retention time and mass spectra of the molecules with commercial libraries (NIST 2017). Moreover, relative quantitation of these compounds were also accomplished by evaluating the relative percentage for each peak [peak area/total ion chromatogram (TIC) area] ( Table 2).

Statistical Analysis
HS/SPME GC-MS quantitation data for each identified compound were subjected to Student t-test to establish the statistical significance of differences between healthy and infected groups at each sampling time. For multivariate modeling, the collected HS/SPME GC-MS quantitation data were assembled in a data matrix containing samples in rows and features (amounts of each VOC) in columns. It was then normalized by constant sum, mean-centered, and divided by the standard deviation of each variable (auto-scaling) to be used as input for statistical analysis. No data filtering was applied, and no missing values were present in the matrix. The reduced and normalized data were then analyzed through the Biostaflow web application (biostatflow.org, v.2.9.2, access date 14 February 2020). Partial least squares discriminant analysis (PLS-DA) [36] was used for classification and relating the variation of VOCs profile to phytopathological status and sampling time. The optimal number of components was determined by evaluating the variance captured and statistics, considering R2, Q2, and accuracy parameters. The obtained model was based on two components and was validated by the leave-one-out cross-validation method (LOOCV) [37]. Furthermore, a permutation test was applied calculating the ratio of the between sum of squares to the within sum of squares (B/W-ratio) for the class assignment prediction ability of the model with a permutation number of 1000 [38,39]; the obtained model was correctly observed to be an outlier of the distribution of the 1000 randomly permuted models.
For the PLS-DA model obtained, variable important for projection (VIP) classifier was applied; the limit of significance value was placed equal to 1.5.
Hierarchical clustering analysis (HCA) was also applied on the same dataset in order to better elucidate the clustering of samples based on similarity and observe its consistency with phytopathological status/sampling time [40,41]. HCA was performed using Metaboanalyst software version 5.0 [42] considering Euclidean distance measure as a definition of similarity and Ward clustering algorithm.

Sensory Analysis
The sensory analysis was performed by 10 panelists (seven women and three men, aged between 28 and 50 years old) trained according to ISO 8586:2012 guidelines [43] and selected on the basis of their familiarity with onion. Onion samples (infected and healthy) were cut into slices and coded with random three digit numbers. Onion samples were placed in plastic plates and given to the panelists in individual booths in an airconditioned sensory evaluation room at 20 • C under incandescent white lighting. Water and unsalted bread were used to clean out tastes among samples. The onion sensory profile was expressed into four general attributes of acceptability (odor rating, taste rating, texture rating, and general acceptability) and in 12 specific attributes: color, pungency, peculiar odor, typical odor, sweetness, spiciness, savoriness, bitterness, typical aroma, peculiar aroma, pulp texture, and crunchiness. During the tasting evaluation, the consumers were asked to rate the samples for overall liking on a 9-point hedonic scale, where 0 = extremely low score and 9 = extremely high score [44]. The values of attributes are presented as means for infected and healthy samples in the form of a spider web graph.

Results and Discussion
The aroma of onion is mainly characterized by volatile sulfur compounds produced by the rearrangement of thiosulphinates; VOCs profile of fresh onion has been extensively studied, and its changes due to onion variety, soil characteristics (e.g., sulfur and nitrogen fertility), biotic stressors, and postharvest storage have been reported in the literature [17,[45][46][47][48][49].
In this work, 11 VOCs were identified and quantified using HS/SPME GC-MS analysis in healthy and OYDV-infected onion bulbs at two sampling times; moreover, using a panel test, the sensory attributes of onion bulbs at the marketable stage (t 2 ) were investigated, aiming at highlighting likely differences between healthy and infected samples.

VOCs Analysis by HS/SPME GC-MS
VOCs profile was obtained by HS/SPME GC-MS analysis to investigate any difference in VOCs in response to virus infection and storage time. The HS/SPME GC-MS Agronomy 2021, 11, 874 6 of 14 analysis allowed us to identify 11 VOCs (Table 2) belonging to different chemical classes: monosulfides (compounds 3, 5, 10), polysulfides (1, 2, 4, 6, 8, 9, 11), and organic acids (7). The relative quantitation of these compounds was evaluated as the relative percentage for each peak at two sampling times t 0 and t 2 for healthy and OYDV-infected samples; the relative percentages of peak area average for each sample group are reported in Table 3. VOC ratio was also calculated for each compound as the ratio of average amounts in OYDV-infected onions relative to healthy onions, irrespective of the sampling time. Values are expressed as mean ± standard error for three samples each group; Student's t test performed between healthy and infected at each sampling time: * significant difference at p < 0.05; ** significant difference at p < 0.01.
The most abundant compound in the analyzed samples (ca. 60% of the TIC), as reported in Table 3, was generally propyl disulfide (11), in accordance with previous literature [50][51][52]. In the genus Allium, polysulfides are produced by the rearrangement of thiosulfinates, which are, in turn, generated through alk(en)yl sulfenic acids (lachrymatory factor). Differences among ACSOs specific moieties (methyl, allyl, prop(en)yl) are found in the Allium genus, strongly depending on the species. Onion shows high levels of Sprop(en)yl l-cysteine-S-oxide (isoalliin); hence, thiosulfinates and polysulfides found in onion also have the same moieties of their precursor as major residues; conversely, high levels of allyl polysulfides are found in garlic. The kind and amount of such moieties have also been correlated with the peculiar kind of aroma typical of each of the species in the Allium genus: propyl sulfides have also been correlated with "fresh onion" taste, while allyl radical containing compounds are reported to cause pungent, radish-or garlic-like aroma [50].
Other well-represented VOCs in samples (ca. 13% and 12% of the TIC, respectively) were allyl-isopropyl disulfide (1) and propanethiol (10); they are widely reported in the literature as main components of VOCs profile and important sources of flavor in fresh onion [15,25]. Allyl-isopropyl disulfide is generated as described above for propyl disulfide, according to the different radical present in the corresponding ACSO. Propanethiol is widely reported in the literature to be present in small amounts compared to propyl disulfide [48,51]; proton-transfer reaction-mass spectrometry (PTR-MS) and GC, coupled with sulfur chemiluminescence detection (GC-SCD) experiments, elucidated that propanethiol undergoes oxidation and dimerization to propyl disulfide during GC-MS sampling and analysis [52]. Hence, propanethiol is likely to be present at higher amounts in the original sample, a fraction of the recorded amount of propyl disulfide being an artefact. Additionally, another eight compounds were identified in all sample groups and quantified at different amounts; most of them (mono-and poly-sulfides) were previously reported in VOCs profile of fresh onion [9,53]. To our knowledge, acetic acid, which was detected in all samples, has never been reported in raw onions. On the contrary, its formation joined with other organic acids was reported in cooked onions as a consequence of thermal degradation [28]. In this view, acetic acid in samples is likely to represent an artefact resulting Agronomy 2021, 11, 874 7 of 14 from thermal degradation occurring during GC analysis. Moreover, its amounts did not change with storage time nor upon virus infection. Due to these reasons, quantitation data on acetic acid were not included in Table 3 and discussed.
The HS/SPME GC-MS data on the identified VOCs were compared in healthy and OYDV-infected samples at each sampling time to highlight possible differences in some specific VOCs in response to OYDV infection and/or storage time. Student's t-test showed significant differences in concentration of 7 out of 11 VOCs in OYDV-infected vs. healthy bulbs at different sampling times ( Table 3). Five of them were regulated at t 2 , and the other two compounds were regulated at t 0 ; no compound was found regulated at both sampling times, indicating a huge significance of postharvest storage in the evolution of VOCs profile. In detail, methyl-propyl disulfide (9) and propanethiol (10) were found regulated by OYDV infection at t 0 , the first being depleted and the second enriched in infected bulbs compared to healthy control. At t 2 , two compounds were recorded at higher levels in infected samples, i.e., 3,5-diethyl 1,2,4-trithiolane (2) and 2,4-dimethyl thiophene (5), while three were depleted upon infection, i.e., allyl-isopropyl disulfide (1), 1-propyl 2,4-thiohept-2-en-5-yl disulfide (4), and 2-mercapto-3,4-dimethyl 2,3-dihydrothiophene (6). All of them are sulfur compounds identified in onion as VOCs formed by rearrangements of thiosulfinates produced by the enzymatic action of alliinase and are reported to play a role in flavor development [15,28]. These five compounds are relevant to the actual difference in VOCs profile between healthy and OYDV-infected 'Rossa di Tropea' onions when they reach the market and the consumers.
Conversely, it has to be underlined that the two VOCs potentially regulated upon infection at t 0 , but not at t 2 (9 and 10), should not be considered crucial for the VOCs profile of onion at marketable stage, nor to be correlated to sensory analysis, which was performed only at t 2 . Nonetheless, their regulation can provide an insight into the early plant response to OYDV infection shortly after inoculation. In detail, propanethiol (10) was upregulated in infected samples at t 0 with an overall VOC ratio of ca. 2, indicating a twofold increased emission in OYDV-infected samples. As mentioned above, propanethiol is one of the most represented VOCs identified in analyzed samples; it was previously indicated as the main source of the characteristic onion odor [52], and its increase was previously correlated with fungi infection in onion [32]. Methyl-propyl disulfide (9) was found in lower amounts in the infected group at t 0 , recording an overall VOC ratio of 0.26, indicating a fourfold decreased emission in OYDV infected samples. Such "mixed" sulfides are produced when two or more alliin homologs are found in the same tissue, giving rise, upon rupture, to asymmetrical sulfenic acid and, hence, to disulfides with two different moieties.
VOCs profile changes at t 2 in function of OYDV infection was even more important, indicating a likely impact on the flavor of the edible product at the marketable stage and correlation to panel test results. Compounds 2 and 5 recorded a higher emission in infected samples, with an overall VOC ratio of 2.24 and 1.48, respectively. Compound 2 (3,5-diethyl 1,2,4-trithiolane) is reported to be synthesized from acetaldehyde and hydrogen sulfide [54] and known to represent an intermediate in the formation of dimethylthiophene (5) [55]. The coherent increase of the two compounds in infected samples at the same sampling time (t 2 ) might reflect a general upregulation of their biosynthethic pathway; this may be due to an increase of one or more of their precursors. In accordance, acetaldehyde was already found to be around six-fold upregulated in diseased onion infected by Fusarium oxysporum [32]. These results may have an impact on the consumption and production of 'Rossa di Tropea' onion: OYDV infection is here demonstrated to alter VOCs profile in 'Rossa di Tropea' onion bulbs, especially at the marketable stage; this may also influence the control measures that are meant to manage these crops and protect them from virus infection. Compounds 1, 4, and 6 were instead emitted in lower amounts by OYDV-infected onions at t 2 . Allyl-isopropyl disulfide (1) was recently found in onion for the first time [56] and represented another example of "mixed" disulfide typical of the Allium genus. Being one of the most abundant VOCs in the analyzed samples, and yet recognized as one of the main ones responsible for flavor in onion, it is supposed to play a key role also in Agronomy 2021, 11, 874 8 of 14 modulating samples' flavor as a consequence of the infection. Allyl moiety has been correlated with pungent, radish-like odor, while n-propyl residue is usually associated with typical onion odor [50]; to our knowledge, no correlation to a specific odor has been observed relative to isopropyl residues in Allium sulfides.
Compound 6 was reported to arise from pyrolysis of bis (1-propenyl) disulfide [57]. It is considered a non-enzymatically produced volatile sulfur compound, having been reported to be over-synthesized in onions where lachrymatory factor synthase was suppressed [58]. This precursor is reported as a major VOC in onion, being the derivative of S-propenyl-L-cysteine-S-oxide, which is the most abundant ACSO in onion [15]. Compound 6 has also been recorded in Meliaceae by Liu et al. [59] but is not considered important in the aroma formation in Amaryllidaceae due to the presence of more pungent molecules.
It has also been proposed that 6 can rearrange to 3,4 dimethylthiophene, an isomer of (5), allowing us to hypothesize that the decrease of the former may be correlated to the increase of the latter at t 2 .

Statistical Analysis
In order to look over the collected data, a PLS-DA model was built, allowing us to reduce the dimensionality of such complex datasets, thus, enabling an easier visualization of any clustering of samples in function of phytopathological status and/or sampling time and to classify samples according to the pattern of quantified compounds. Figure 1a shows the score plot for the score matrix of the two-components model obtained by PLS-DA: PC1 accounts for 12.9% of the captured variance in the model, while PC2 accounts for 23.8%. The distribution obtained shows that VOCs profile variables partially discriminated groups of samples, according to phytopathological status and sampling time.  Table 1).
Loading plot (Figure 1b) was useful for identifying which variables (i.e., quantified VOCs) have the most significant effect on each component in the model. High positive or negative loading values for a variable indicate that the variable strongly influences the component; loadings close to 0 indicate that the variable has a weak influence on the component. In particular, three compounds (1, 6, and 9) are found at high positive values of PC2 (i.e., >20); hence, their content is suggested to strongly influence PC2, which, as reported for the score plot, rules the discrimination between healthy and infected samples; the other two VOCs (2 and 10) have high negative values of PC2 (i.e., <15), indicating that the sustained increase in their content in infected samples with respect to healthy is another factor influencing PC2, thus, driving group discrimination based on phytosanitary status. These results are in accordance with the direct comparison of amounts performed in the previous section for each quantified compound.  Table 1).
Loading plot (Figure 1b) was useful for identifying which variables (i.e., quantified VOCs) have the most significant effect on each component in the model. High positive or negative loading values for a variable indicate that the variable strongly influences the component; loadings close to 0 indicate that the variable has a weak influence on the component. In particular, three compounds (1, 6, and 9) are found at high positive values of PC2 (i.e., >20); hence, their content is suggested to strongly influence PC2, which, as reported for the score plot, rules the discrimination between healthy and infected samples; Agronomy 2021, 11, 874 9 of 14 the other two VOCs (2 and 10) have high negative values of PC2 (i.e., <15), indicating that the sustained increase in their content in infected samples with respect to healthy is another factor influencing PC2, thus, driving group discrimination based on phytosanitary status. These results are in accordance with the direct comparison of amounts performed in the previous section for each quantified compound. Figure 2 reports the results of the variable important for projection (VIP) classification for the PLS-DA model, identifying the variables (compounds) on which the built model mostly relies; overall, seven compounds were reported to have a high score (>1.5). Three of them were upregulated upon infection (5, 2, and 1), three were downregulated (6, 4, and  9), and one did not show any regulation correlated with virus infection with respect to healthy control (8). The most important variables were represented by compounds 6 and 5, in accordance with literature and previous discussion.
Loading plot (Figure 1b) was useful for identifying which variables (i.e., quantified VOCs) have the most significant effect on each component in the model. High positive or negative loading values for a variable indicate that the variable strongly influences the component; loadings close to 0 indicate that the variable has a weak influence on the component. In particular, three compounds (1, 6, and 9) are found at high positive values of PC2 (i.e., >20); hence, their content is suggested to strongly influence PC2, which, as reported for the score plot, rules the discrimination between healthy and infected samples; the other two VOCs (2 and 10) have high negative values of PC2 (i.e., <15), indicating that the sustained increase in their content in infected samples with respect to healthy is another factor influencing PC2, thus, driving group discrimination based on phytosanitary status. These results are in accordance with the direct comparison of amounts performed in the previous section for each quantified compound. Figure 2 reports the results of the variable important for projection (VIP) classification for the PLS-DA model, identifying the variables (compounds) on which the built model mostly relies; overall, seven compounds were reported to have a high score (>1.5). Three of them were upregulated upon infection (5, 2, and 1), three were downregulated (6, 4, and 9), and one did not show any regulation correlated with virus infection with respect to healthy control (8). The most important variables were represented by compounds 6 and 5, in accordance with literature and previous discussion.  Given the only partial group separation obtained by the PLS-DA model, HCA was also performed to fully understand hidden patterns and allow a better visualization of the actually similar samples. The dendrogram obtained by HCA is reported in Figure 3: it shows a clear separation of clusters referring to healthy and infected samples, respectively, at t 2 sampling time. Instead, healthy and infected samples collected at t 0 showed some similarity: bulbs with different sanitary status were observed to be quite mixed in the dendrogram and did not account for a distinct pattern according to phytosanitary status. This result indicates an effect of OYDV infection on the VOCs profile of onions at the marketable stage (t 2 ), and again, it underlines the huge importance of postharvest storage in the development of the taste and aroma perceived by consumers; this evolution has been observed previously in onion regarding other important compounds for taste and aroma, such as flavonoids, sugars, and aminoacids [60,61]. the dendrogram and did not account for a distinct pattern according to phytosanitary sta-tus. This result indicates an effect of OYDV infection on the VOCs profile of onions at the marketable stage (t2), and again, it underlines the huge importance of postharvest storage in the development of the taste and aroma perceived by consumers; this evolution has been observed previously in onion regarding other important compounds for taste and aroma, such as flavonoids, sugars, and aminoacids [60,61].

Sensory Analysis
The sensory analysis of onion samples was reported in a spider graph ( Figure 4) that allowed us to compare the perceived attributes for infected and healthy 'Rossa di Tropea' onions at t2 sampling time, when onions reach the marketable stage after postharvest storage. The infected onions manifested higher scores for color, pungency, spiciness, and bitterness than healthy ones, which manifested instead higher scores in the other sensory descriptors, such as sweetness, pulp texture, and crunchiness. Very low peculiar odor (ranging from 0.67 to 2.65) and aroma (ranging from 0 to 1.89) were evidenced in both

Sensory Analysis
The sensory analysis of onion samples was reported in a spider graph (Figure 4) that allowed us to compare the perceived attributes for infected and healthy 'Rossa di Tropea' onions at t 2 sampling time, when onions reach the marketable stage after postharvest storage. The infected onions manifested higher scores for color, pungency, spiciness, and bitterness than healthy ones, which manifested instead higher scores in the other sensory descriptors, such as sweetness, pulp texture, and crunchiness. Very low peculiar odor (ranging from 0.67 to 2.65) and aroma (ranging from 0 to 1.89) were evidenced in both samples, whereas typical aroma and odor resulted in high scores (5.1 and 4.8, respectively). Among the four general attributes of acceptability, texture rating was higher in healthy onions (5.20) than infected ones (4.86), and it can be correlated with some effect of infection to the vegetable tissues since viral infections are known to induce structural alteration starting from the cell wall [62,63]. The evaluation of general acceptability obtained a good score on both infected and healthy onion samples with values of 5.30 and 5.23, respectively.
The statistical data elaboration evidenced significant differences only for few attributes between healthy and infected samples. The taste attributes accounting for the major differences were bitterness and sweetness, suggesting an influence of virus infection on these characteristics. In fact, infected onion samples showed a lower sweetness (5.42) compared to healthy onions (6.09); higher bitterness (2.89) and spiciness (4.34) for the first ones were also recorded. As panel test results show an enhanced pungency, spiciness, and bitterness in the infected group, it is likely that the allyl moiety in compound 1 is not crucial in defining these characters, as compound 1 is underrepresented in infected onions. tively). Among the four general attributes of acceptability, texture rating was higher in healthy onions (5.20) than infected ones (4.86), and it can be correlated with some effect of infection to the vegetable tissues since viral infections are known to induce structural alteration starting from the cell wall [62,63]. The evaluation of general acceptability obtained a good score on both infected and healthy onion samples with values of 5.30 and 5.23, respectively. The statistical data elaboration evidenced significant differences only for few attributes between healthy and infected samples. The taste attributes accounting for the major differences were bitterness and sweetness, suggesting an influence of virus infection on these characteristics. In fact, infected onion samples showed a lower sweetness (5.42) compared to healthy onions (6.09); higher bitterness (2.89) and spiciness (4.34) for the first ones were also recorded. As panel test results show an enhanced pungency, spiciness, and bitterness in the infected group, it is likely that the allyl moiety in compound 1 is not crucial in defining these characters, as compound 1 is underrepresented in infected onions.

Conclusions
In this study, the quantitation of VOCs in 'Rossa di Tropea' onion bulbs by means of HS-SPME GC-MS indicated significant differences between healthy and OYDV-infected samples in the content of seven VOCs at harvest and after postharvest storage. All of them belong to the sulfur compounds class, which is well known to be crucial in the development of odor and taste in Allium species.
Multivariate statistical analysis based on this quantitation discriminated between samples according to their sanitary status at t2 sampling time, indicating the significance of virus infection during postharvest storage in the development of VOCs profile evolving to the taste and aroma perceived by consumers. The VOCs mostly accounting for such clustering were also identified as 2-mercapto-3,4-dimethyl-2,3 dihydrothiophene (6), 2,4dimethylthiophene (5), and 3,5.diethyl-1,2,4-trithiolane (2). The sensory analysis also accounted for differences in some sensory attributes, and in one general attribute of acceptability (i.e., texture), between healthy and infected samples at the marketable stage. All these results suggest an impact of OYDV infection on the taste of this particular onion

Conclusions
In this study, the quantitation of VOCs in 'Rossa di Tropea' onion bulbs by means of HS-SPME GC-MS indicated significant differences between healthy and OYDV-infected samples in the content of seven VOCs at harvest and after postharvest storage. All of them belong to the sulfur compounds class, which is well known to be crucial in the development of odor and taste in Allium species.
Multivariate statistical analysis based on this quantitation discriminated between samples according to their sanitary status at t 2 sampling time, indicating the significance of virus infection during postharvest storage in the development of VOCs profile evolving to the taste and aroma perceived by consumers. The VOCs mostly accounting for such clustering were also identified as 2-mercapto-3,4-dimethyl-2,3 dihydrothiophene (6), 2,4dimethylthiophene (5), and 3,5.diethyl-1,2,4-trithiolane (2). The sensory analysis also accounted for differences in some sensory attributes, and in one general attribute of acceptability (i.e., texture), between healthy and infected samples at the marketable stage. All these results suggest an impact of OYDV infection on the taste of this particular onion variety; this change involves several compounds, and it might indicate their importance in postharvest storage and the development of taste and aroma in 'Rossa di Tropea' onion.