Comprehensive Volatilome Signature of Various Brassicaceae Species

To investigate in detail the volatilomes of various Brassicaceae species, landraces, and accessions, and to extract specific volatile markers, volatile aroma compounds were isolated from plant samples by headspace solid-phase microextraction and analyzed by gas chromatography/mass spectrometry (HS-SPME-GC/MS). The data obtained were subjected to uni- and multivariate statistical analysis. In general, two cabbage (Brassica oleracea L. var. capitata) landraces emitted the lowest amounts of volatiles generated in the lipoxygenase (LOX) pathway. Wild species Brassica incana Ten. and Brassica mollis Vis. were characterized by relatively high trans-2-hexenal/cis-3-hexen-1-ol ratio in relation to other investigated samples. A Savoy cabbage (Brassica oleracea L. var. sabauda) cultivar and three kale (Brassica oleracea L. var. acephala) accessions exhibited particular similarities in the composition of LOX volatiles, while the LOX volatilome fraction of B. incana and B. mollis partially coincided with that of another wild species, Diplotaxis tenuifolia L. Regarding volatiles formed in the glucosinolate (GSL) pathway, Savoy cabbage and wild species B. incana, B. mollis, and D. tenuifolia showed more intense emission of isothiocyanates than cabbage and kale. Diplotaxis tenuifolia showed a rather limited production of nitriles. The results of this study contribute to the general knowledge about volatile composition from various Brassicaceae species, which could be exploited for their better valorization. Future studies should focus on the influence of various environmental, cultivation, and post-harvest factors to obtain data with a higher level of applicability in practice.


Introduction
The Brassicaceae family includes 338 genera and 3709 species [1], among which the genus Brassica is one of the most important, since it includes many species domesticated by humans that have agricultural and economic relevance [2]. Today, Brassicaceae species are appreciated by consumers and are among the most commonly consumed vegetables globally due to their high nutritional value and abundance in various valuable bioactive compounds and elements, including carotenoids, tocopherols, ascorbic acid, glucosinolates, phenols, and others [3][4][5].
Among specialized plant metabolites in Brassicaceae plants, volatile organic compounds have many important biological properties and functions. They can act as signaling molecules within and between plants, provide protection by deterring herbivores or attracting their predators, and contribute to plant resistance to damage and stress [6][7][8][9][10][11][12][13]. Several biosynthetic pathways are responsible for their formation, the most important being the lipoxygenase (LOX), glucosinolate (GSL), and methylerythritol phosphate (MEP) pathways in which the so-called green leaf C 6 and C 5 volatiles, isothiocyanates (ITCs), and terpenoids are formed, respectively [10]. The entire profile of volatile compounds emitted by a plant that can be detected is also called the volatilome. Table 1. Amounts of volatile compounds * found in various Brassicaceae species after headspace solid-phase microextraction followed by gas chromatography/mass spectrometry (HS-SPME-GC/MS) sorted by compound class and descending Fisher F-ratio.

Volatile Compounds from the Mevalonic Acid (MVA) and Methylerythritol Phosphate (MEP) Pathways
Volatile terpenoids have important roles in plant defense against herbivores and pathogens. They attract arthropods that prey on or parasitize herbivores, have a role in reproduction by attracting pollinators and seed disseminators, and play a part in responding to various stresses [26,27]. Terpenoids originate from two main precursors, isopentenyl diphosphate (IPP) and dimethylallyl diphosphate (DMAPP), from two alternate biosynthetic pathways: the cytosolic mevalonic acid (MVA) pathway, in which mainly sesquiterpenes are produced, and the plastidial methylerythritol phosphate (MEP) pathway, which is mainly responsible for the production of monoterpenes [28]. Many enzymes are involved in the mentioned biochemical reactions. For example, the formation of IPP and DMAPP from pyruvate and D-glyceraldehyde 3-phosphate in the MEP pathway consists of seven main enzymatic steps [29].
In this study five terpenoids were identified, including only a single monoterpene, eucalyptol, and a single sesquiterpenoid, perhydrofarnesyl acetone ( Table 1). The other three compounds were norisoprenoids, formed in a parallel biosynthetic branch mainly by cleavage of carotenoid precursors [30]. Previous studies reported higher numbers of terpenes identified in cabbage [10,19], kale [31], salad rocket (E. sativa L.) [23], and other Brassica species not included in this study [10]. The highest amount of β-ionone, a C 13norisoprenoid with an odor reminiscent of violets, was found in B. mollis. Two cabbage landraces contained the lowest amounts of both β-ionone derivatives, but were, together with Savoy cabbage, most abundant in isophorone, known for its characteristic peppermintlike odor. Kale accession KAL3 and the investigated wall rocket accession emitted the highest levels of perhydrofarnesyl acetone, a volatile with a floral scent.

Volatile Compounds from the Lipoxygenase (LOX) Pathway
Volatile compounds formed in the lipoxygenase (LOX) pathway are also known as 'green leaf volatiles' since they provide vegetables with characteristic odor notes reminiscent of green leaves and grass. They are emitted in high amounts from plants under stressed conditions, such as wounding or pathogen infection, play a role in plant defense, and serve for communication with insects and other neighbor plants [25,32].
LOX volatile compounds mainly belong to the chemical classes of aldehydes, ketones, alcohols, and esters. Following the activation of the LOX pathway during plant development or by a stressor, sequential action of several enzymes is initiated. Lipoxygenases (LOX) catalyze regiospecific dioxygenation of long-chain linoleic (C 18:2 ) and linolenic (C 18:3 ) fatty acids to form corresponding hydroperoxides, which then serve as substrates to hydroperoxide lyases (HPL) to produce saturated and unsaturated volatile C 6 or C 9 aldehydes. After that, aldehydes may be converted to other isomers by isomerases, while alcohol dehydrogenases (ADH) and alcohol acyl transferases (AAT) partially transform them to alcohols and esters, respectively [25].
In plants, C 6 dominates over C 9 forms. In another branch of the pathway, hydroperoxides are transformed to C 5 analogues. Several isoforms of LOX occur in most plants, which is most probably the case for other enzymes as well. For example, six LOX isoforms were detected in Arabidopsis [33].
In this study, the most abundant LOX compounds in all the investigated Brassicaceae species were C 6 volatiles ( Table 1). Cis-3-hexen-1-ol was found in the highest amount among alcohols in all the samples, followed by 1-hexanol and trans-3-hexen-1-ol. High levels of C 6 aldehydes were also noted, primarily trans-2-hexenal, followed by hexanal. Much lower, but still relatively high amounts of other hexenals were also detected. The most abundant among C 5 compounds were alcohols cis-2-penten-1-ol and 1-penten-3-ol. Considering that two of the most abundant LOX generated volatiles were trans-2-hexenal and cis-3-hexen-1-ol, the significant activity of LOXs, isomerases, and ADHs was obvious, while the absence of esters suggested that the loads of AATs were not significant, or they were not particularly active. Previous studies reported similar relations among the major LOX volatile compounds in cabbage [10], kale [31], salad rocket (E. sativa L.) [23], and other Brassica species [10], although with particular dissimilarities.
In this study, the investigated cabbage landraces were generally the least abundant in LOX volatile compounds and this feature differentiated them well from most of the other samples (Table 1). This was in accordance with the results from Liu et al. [10], who also found lower levels of LOX volatiles in cabbage in relation to particular other Brassica species. Compounds found in cabbage in the highest amount compared to other samples were 2-ethyl-1-hexanol in both landraces and cis-3-hexenal in landrace Brgujski. Among the LOX compounds, two cabbage landraces differed among each other only with respect to cis-3-hexenal and hexanal, with higher levels of the former found in Brgujski and that of the latter in Žminjski landrace.
The average ratio of the amounts of the two major LOX volatiles trans-2-hexenal/cis-3-hexen-1-ol was by far the highest in B. mollis, 3.2, followed by B. incana with 1.7. Since both volatiles are derived from the same precursor cis-3-hexenal [25], this implied that the relative activity of isomerases in B. mollis and B. incana overpowered that of ADHs. The opposite was probable in kale accessions and Savoy cabbage, for which lower average trans-2-hexenal/cis-3-hexen-1-ol ratios of 0.5, 0.3, 0.6, and 0.4 were determined, respectively. Medium ratios of 0.9, 1.3, and 1.2 were determined for cabbage landraces Brgujski and Žminjski and for the wall rocket accession, respectively. The accumulation of cis-3-hexenal in cabbage landrace Brgujski was possibly an indicator of a slowdown of the LOX cascade at that point. High amounts of hexanal in cabbage Žminjski and in particular other samples, pointed to the possibility of higher availability of linoleic acid, the main substrate in this branch of the LOX pathway [25]. Obviously, many differences and similarities in LOX volatile composition were observed between various groups of the investigated samples, while each of them was characterized by certain particularities. Such a range of features certainly resulted from a complex interplay between various enzymes and their isoforms, additionally conditioned by the availability of substrates, which was determined by genotype and was evidently unique for each variety or accession. It was previously reported that various Brassica species significantly differ in the number and type of genes that regulate the fatty acid metabolism, i.e., the LOX pathway [10].
Hierarchical clustering analysis (HCA) was conducted on a reduced dataset which included all the samples grouped into nine groups (species, varieties, accessions) and LOX generated volatile compounds as variables. Clear clustering of the groups was obtained after including the first 15 selected variables, which is shown on a heatmap diagram in Figure 1. Particular species, varieties, and accessions were related mostly to the characteristic volatile compounds previously determined by ANOVA (Table 1). Cabbage landraces were characterized by abundance in 2-ethyl-1-hexanol (L6) but were the poorest in other LOX volatiles. They formed a separate cluster and differed most from the other species based on the LOX fraction of the volatilome. Wall rocket samples were clustered together mostly based on the highest amounts of 3,5,5-trimethyl-2-hexene (L5), 6-methyl-5-hepten-2-ol (L7), and 2-methyl-2-pentenal (L13). Brassica incana and B. mollis were clearly richer in trans-2-butenal (L4), while the latter also emitted the highest levels of trans,trans-2,4-heptadienal (L12). These two wild species were clustered together and were the closest and most similar to the wall rocket accession. Kale accessions, especially KAL3, were clustered mainly based on high cis-3-hexen-1-ol (L1) amount. Additionally, they were grouped rather close, with KAL2 and KAL3 being more similar to each other. Savoy cabbage contained mostly 3,5,5-trimethyl-3-cyclohexanone (L9) and trans-3-hexen-1-ol (L15). Judging from their relatively close grouping, the overall composition of LOX volatiles in kale accessions turned out to be more similar to that of Savoy cabbage than to those emitted by the other investigated species.   Table 1. Colors of the heatmap cells indicate a low (dark blue), medium (white), and high (dark red) abundance of a particular compound.
To better identify volatile compounds characteristic for particular species, which are therefore most useful for their mutual differentiation, the data on volatiles were subjected to partial least squares-discriminant analysis (PLS-DA). Savoy cabbage and wall rocket were excluded from PLS-DA since they were represented by a single variety or accession, respectively, while two cabbage landraces, three kale accessions, and two wild species B. incana and B. mollis were grouped into three corresponding groups, respectively. The  Table 1. Colors of the heatmap cells indicate a low (dark blue), medium (white), and high (dark red) abundance of a particular compound.
To better identify volatile compounds characteristic for particular species, which are therefore most useful for their mutual differentiation, the data on volatiles were subjected to partial least squares-discriminant analysis (PLS-DA). Savoy cabbage and wall rocket were excluded from PLS-DA since they were represented by a single variety or accession, respectively, while two cabbage landraces, three kale accessions, and two wild species B. incana and B. mollis were grouped into three corresponding groups, respectively. The PLS-DA differentiation according to species based on LOX volatiles data was very successful, as can be seen in Figure 2a. Twenty LOX volatile compounds with the highest variable importance in projection (VIP) scores are listed in Figure 2b. A compound with the highest VIP score (>2.0) was trans-2-butenal (L4), which was most abundant in two wild species and least abundant in cabbage. Several other LOX volatile compounds with relatively high VIP scores (>1.4) were confirmed to be most characteristic for wild species, such as trans,trans-2,4-heptadienal (L12), trans-2-hexenal (L20), trans-2-pentenal (L22), 1-penten-3-ol (L2), 6methyl-5-hepten-2-ol (L7), and 1-penten-3-one (L24). High amounts of 3,4-nonadiene (L11), trans,cis-2,4-heptadienal (L8), and cis-4-heptenal (L16) served for the differentiation of kale, while the main marker for cabbage among LOX compounds was 2-ethyl-1-hexanol (L6).  Table 1.

Volatile Compounds from the Glucosinolate (GSL) Pathway
Brassicaceae vegetables contain glucosinolates (GSLs), a group of sulfur-containing compounds of variable structure that have important roles in plants' defense against various stressors. After the disruption of plant cells, GSLs come into contact with endogenous myrosinases, β-D-thioglucosidases, whose activity results in the release of various hydrolysis products, such as volatile ITCs, nitriles, and epithionitriles [20,34]. Isothiocyanates are considered the most important volatile compounds deriving from the GSL pathway since, besides their key contribution to the sharp odor, bitterness, and pungency of various Brassicaceae vegetables [35], they exhibit significant beneficial effects on human health [15]. Brassicaceae vegetables can also contain epithiospecifier proteins that affect the hydrolysis of GSLs in a way that may result in releasing more nitriles or epithionitriles than ITCs [20,36].
Among volatile compounds generated in the GSL pathway, cabbage landrace Brgujski emitted the highest amounts of particular nitriles, such as 4-methylpentanenitrile and 3-(methylthio)butylnitrile, while it also had a high amount of hexanenitrile together with the other cabbage landrace Žminjski (Table 1). Only

Volatile Compounds from the Glucosinolate (GSL) Pathway
Brassicaceae vegetables contain glucosinolates (GSLs), a group of sulfur-containing compounds of variable structure that have important roles in plants' defense against various stressors. After the disruption of plant cells, GSLs come into contact with endogenous myrosinases, β-D-thioglucosidases, whose activity results in the release of various hydrolysis products, such as volatile ITCs, nitriles, and epithionitriles [20,34]. Isothiocyanates are considered the most important volatile compounds deriving from the GSL pathway since, besides their key contribution to the sharp odor, bitterness, and pungency of various Brassicaceae vegetables [35], they exhibit significant beneficial effects on human health [15]. Brassicaceae vegetables can also contain epithiospecifier proteins that affect the hydrol-ysis of GSLs in a way that may result in releasing more nitriles or epithionitriles than ITCs [20,36].
Among volatile compounds generated in the GSL pathway, cabbage landrace Brgujski emitted the highest amounts of particular nitriles, such as 4-methylpentanenitrile and 3-(methylthio)butylnitrile, while it also had a high amount of hexanenitrile together with the other cabbage landrace Žminjski (Table 1). Only 4-methylpentanenitrile and 3-(methylthio)butylnitrile were found to mutually differentiate the two cabbages, both being more abundant in Brgujski landrace. Both cabbages were characterized by a relatively low total ITCs/nitriles ratio compared to the investigated wild species, although not significantly different from that of B. incana. As the total peak area of nitriles was not higher than that found in the majority of other samples, the low ratios found could be attributed to relatively low production of ITCs, as a result of weak myrosinase activity.
All three kale accessions released relatively high amounts of methyl thiocyanate, while among them, KAL1 stood out with higher amounts of 4-methylpentanenitrile, carbon disulfide, and benzyl nitrile. For several compounds, differences were also observed between different pairs of kale accessions. Similar to cabbages, kale accessions were characterized by relatively low ITCs/nitriles ratios.
Several ITCs were found in the highest amount in Savoy cabbage, namely allyl isothiocyanate, previously reported as a major isothiocyanate in various Brassica species like cabbage [19,21] and kale [31], as well as an unidentified isothiocyanate derivative (RI 1449, m/z 99, 41, 39, 72) and 2-methylbutyl isothiocyanate, with the highest amount of phenethyl isothiocyanate shared with B. mollis.
The GSL volatiles found in higher amounts in Brassica wild species B. incana and B. mollis included 4-mercaptophenol, 2,4-pentadienenitrile, benzenepropanenitrile, 3-butenyl isothiocyanate, 2-ethylthiophene, 4-(methylthio)butylnitrile, and benzyl nitrile, although without statistical significance in relation to some particular samples in a few cases. The two species mutually differed based on the amounts of several GSL compounds, such as, for example, benzenepropanenitrile among nitriles and 3-butenyl isothiocyanate among ITCs, which were both found in the highest amount in B. mollis among all the investigated samples. Several other GSL-derived compounds were found in higher amounts in B. mollis than in B. incana, while B. incana was the most abundant in 2-ethylthiophene among all the investigated samples. B. mollis was characterized by the highest sum of both ITCs and nitriles among all the samples, mostly due to the high proportions of benzenepropanenitrile and 3-butenyl isothiocyanate, respectively, which implied that the enzymatic activity in this species was relatively high. It is worth mentioning that relatively high ITCs/nitriles ratios observed in B. incana and especially B. mollis when compared to the cabbages and kales were mainly due to higher total ITCs amounts, possibly due to stronger myronase activity. Brassica incana and B. mollis emitted rather low amounts of methyl thiocyanate, carbon disulfide, 2-methylbutyl isothiocyanate, 3-(methylthio)butylnitrile, allyl nitrile, an unidentified isothiocyanate derivative (RI 1837, m/z 99, 71, 72, 59), and dimethyl disulfide, which differentiated them well from cabbage landraces, kale accessions, and a Savoy cabbage cultivar, while they shared most of the mentioned features with wall rocket.
Wall rocket exhibited several characteristic differences with respect to the other investigated vegetables. It emitted the highest levels of 4-methylpentyl isothiocyanate, carbon disulfide, and benzyl isothiocyanate, as well as an unidentified cyclic sulfur compound (IR 1846, m/z 88, 116, 117, 60), iberverin, and erucin, which were found in traces or not at all in other samples. Benzyl isothiocyanate and erucin accounted for most of the sum of ITCs in this species. Erucin (4-(methylthio)butyl isothiocyanate), which derives from glucoerucin, one of the major GSLs found in rocket leaves, was previously reported as the main volatile ITC in rocket salad (E. sativa and D. tenuifolia) [16,17,22,23]. The investigated wall rocket accession had the highest ratio of total ITCs/nitriles. Although it contained a total amount of ITCs that was higher than those found in cabbage landraces and kale accessions, a high ratio was mostly due to the very low total amount of nitriles (Table 1). This implied a weaker influence of epithiospecifier proteins that divert the GSL pathway towards produc-ing more nitriles or epithionitriles at the expense of ITCs [20,36]. Several compounds, such as 2-methylbutyl isothiocyanate and an unidentified isothiocyanate derivative (RI 1837, m/z 99, 71, 72, 59), were not detected in the investigated wall rocket accession.
HCA was conducted on a reduced dataset which included all the samples grouped into nine groups (species, landraces, and accessions) and GSL generated volatile compounds as variables. Relatively good clustering was obtained including the first 15 selected variables, although particular landraces and accessions within particular species were not completely distinguished (Figure 3). KAL1 showed significant deviations since it was not clustered together with other kale accessions and its samples were scattered rather randomly among those of different species. Wild species B. incana and B. mollis were clustered together, mostly by high amounts of 4-mercaptophenol (G5) and 2-ethylthiophene (G14) that were more abundant in the former, and benzenepropanenitrile (G10), 3-butenyl isothiocyanate (G12), and 2,4-pentadienenitrile (G7) that were more abundant in the latter, respectively. Savoy cabbage was clearly characterized by higher amounts of an unidentified isothiocyanate derivative (RI 1449, m/z 99, 41, 39, 72) (G8), 4-methylthiazole (G2), and allyl isothiocyanate (G6), same as cabbages by the amounts of 4-methylpentanenitrile (G3) and hexanenitrile (G15). 4-Methylpentyl isothiocyanate (G4) and carbon disulfide (G13) were most characteristic for wild rocket, while methyl thiocyanate (G1) and 1-allyl-1-cyclohexanol (G9) were most responsible for clustering KAL2 and KAL3 samples together. Interestingly, and contrary to the case of LOX volatiles, wall rocket turned out to be most similar to the mentioned two kale accessions. Together with the cabbages, they formed a larger cluster distant from Savoy cabbage, and especially from wild species B. incana and B. mollis. It should be noted that particular volatiles, such as erucin (G16) and iberverin (G19) that were almost exclusively found in wild rocket, were excluded by the HCA model, since they were not useful for differentiation of the other investigated samples among each other. and carbon disulfide (G13) were most characteristic for wild rocket, while methyl thiocyanate (G1) and 1-allyl-1-cyclohexanol (G9) were most responsible for clustering KAL2 and KAL3 samples together. Interestingly, and contrary to the case of LOX volatiles, wall rocket turned out to be most similar to the mentioned two kale accessions. Together with the cabbages, they formed a larger cluster distant from Savoy cabbage, and especially from wild species B. incana and B. mollis. It should be noted that particular volatiles, such as erucin (G16) and iberverin (G19) that were almost exclusively found in wild rocket, were excluded by the HCA model, since they were not useful for differentiation of the other investigated samples among each other.  Table 1. Colors of the heatmap cells indicate low (dark blue), medium (white), and high (dark red) abundance of a particular compound.
An unidentified compound (m/z 113, 45, 73, 86) and tentatively identified N-isobutylidene cyclopropylamine were found in the highest amount in wild species, with a higher amount of the former in B. incana and the latter in B. mollis. The two wild species were abundant in trans-2-(2-pentenyl)furan, while B. mollis contained the highest level of benzeneacetaldehyde and the lowest level of 1-butoxy-2-propanol among all the investigated species.

Differentiation Based on the Overall Volatilome
To additionally comprehend the differences in the overall volatilome between cabbage, kale, and two wild species, HCA was applied on a reduced dataset which included all the samples grouped into three categories and all the identified volatile compounds as variables. The three categories were separated rather clearly by the amounts of the first 35  Table 1.
An unidentified compound (m/z 113, 45, 73, 86) and tentatively identified N-isobutylidene cyclopropylamine were found in the highest amount in wild species, with a higher amount of the former in B. incana and the latter in B. mollis. The two wild species were abundant in trans-2-(2-pentenyl)furan, while B. mollis contained the highest level of benzeneacetaldehyde and the lowest level of 1-butoxy-2-propanol among all the investigated species.

Differentiation Based on the Overall Volatilome
To additionally comprehend the differences in the overall volatilome between cabbage, kale, and two wild species, HCA was applied on a reduced dataset which included all the samples grouped into three categories and all the identified volatile compounds as variables. The three categories were separated rather clearly by the amounts of the first 35 selected volatiles ( Figure 5). Each category was characterized by different sets of compounds, although kale and wild species exhibited similarities in the amounts of volatile compounds clustered in the lower half of the heatmap (from M1 down), which were mostly LOX, MEP, and other volatile compounds, without a single compound from the GSL pathway. Cabbage was distinguished by higher amounts of isophorone (M3) from MEP, 2ethyl-1-hexanol (L6) from LOX, and other miscellaneous compounds, such as phenol (O14), 3-methylbutanal (O17), and 1-butoxy-2-propanol (O2). Wild species were distinguished and grouped based on the amounts of 4-mercaptophenol (G5) and 2,4-pentadienenitrile (G7) from GSL, as well as by the amounts of the unidentified compound (m/z 113, 45, 73, 86) (O1), together with tentatively identified N-isobutylidene cyclopropylamine (O3). Kale accessions were clustered together mostly by higher amounts of several LOX generated compounds. Judging from the compounds selected from the overall volatilomes, wild species B. incana and B. mollis were more similar and were clustered much closer to kale than to cabbage. Although significant differences between landraces or accessions of the same species were found for a number of volatile compounds, HCA once again confirmed that their volatilomes retained a significant amount of important information about species' origins.
Plants 2023, 12, x FOR PEER REVIEW 13 of 18 selected volatiles ( Figure 5). Each category was characterized by different sets of compounds, although kale and wild species exhibited similarities in the amounts of volatile compounds clustered in the lower half of the heatmap (from M1 down), which were mostly LOX, MEP, and other volatile compounds, without a single compound from the GSL pathway. Cabbage was distinguished by higher amounts of isophorone (M3) from MEP, 2-ethyl-1-hexanol (L6) from LOX, and other miscellaneous compounds, such as phenol (O14), 3-methylbutanal (O17), and 1-butoxy-2-propanol (O2). Wild species were distinguished and grouped based on the amounts of 4-mercaptophenol (G5) and 2,4-pentadienenitrile (G7) from GSL, as well as by the amounts of the unidentified compound (m/z 113, 45, 73, 86) (O1), together with tentatively identified N-isobutylidene cyclopropylamine (O3). Kale accessions were clustered together mostly by higher amounts of several LOX generated compounds. Judging from the compounds selected from the overall volatilomes, wild species B. incana and B. mollis were more similar and were clustered much closer to kale than to cabbage. Although significant differences between landraces or accessions of the same species were found for a number of volatile compounds, HCA once again confirmed that their volatilomes retained a significant amount of important information about species' origins.  Table 1. Colors of the heatmap cells indicate low (dark blue), medium (white), and high (dark red) abundance of a particular compound.  Table 1. Colors of the heatmap cells indicate low (dark blue), medium (white), and high (dark red) abundance of a particular compound.
The majority of compounds selected by PLS-DA from the overall volatilomes as those with the highest VIP scores were mostly most abundant in wild species B. incana and B. mollis and least abundant in cabbage, while kale contained medium amounts with a few exceptions ( Figure 6). Again, LOX generated trans-2-butenal (L4), characteristic for wild species, emerged as a compound with the largest potency to discriminate the three categories. It was followed by 1-butoxy-2-propanol (O2) abundant in cabbage and trans-2-(2-pentenyl)furan (O9) abundant in wild species, both from the group of other volatile compounds. The compounds with relatively high VIP scores were also trans,trans-2,4heptadienal (L12), β-ionone (M1), and two unidentified compounds with characteristic ions m/z 113, 45, 73, 86 (O1) and m/z 68, 39, 98 (O8) found in the highest level in wild species, and 2-ethyl-1-hexanol (L6) found in the highest level in cabbage.
Interestingly, LOX and other volatile compounds dominated with high VIP scores, while only a single GSL volatile, allyl nitrile (G21), was included among 20 compounds most useful for the discrimination of the three categories. The majority of compounds selected by PLS-DA from the overall volatilomes as those with the highest VIP scores were mostly most abundant in wild species B. incana and B. mollis and least abundant in cabbage, while kale contained medium amounts with a few exceptions ( Figure 6). Again, LOX generated trans-2-butenal (L4), characteristic for wild species, emerged as a compound with the largest potency to discriminate the three categories. It was followed by 1-butoxy-2-propanol (O2) abundant in cabbage and trans-2-(2-pentenyl)furan (O9) abundant in wild species, both from the group of other volatile compounds. The compounds with relatively high VIP scores were also trans,trans-2,4-heptadienal (L12), β-ionone (M1), and two unidentified compounds with characteristic ions m/z 113, 45, 73, 86 (O1) and m/z 68, 39, 98 (O8) found in the highest level in wild species, and 2-ethyl-1-hexanol (L6) found in the highest level in cabbage.
Interestingly, LOX and other volatile compounds dominated with high VIP scores, while only a single GSL volatile, allyl nitrile (G21), was included among 20 compounds most useful for the discrimination of the three categories. Codes indicating compounds correspond to those reported in Table 1.
Cabbage samples consisted of local landraces 'Brgujski kapuz' (accession IPT001, CAB1) and 'Žminjski kapuz' (accession IPT028, CAB2) maintained as a part of the gen-bank collection of the Institute of Agriculture and Tourism in Poreč, Croatia. Both cabbage landraces included in this study are commonly grown by local farmers, with landrace 'Brgujski' characterized by enhanced anthocyanin coloration and landrace 'Žminjski' by rather slight anthocyanin coloration.  Table 1.
Cabbage samples consisted of local landraces 'Brgujski kapuz' (accession IPT001, CAB1) and 'Žminjski kapuz' (accession IPT028, CAB2) maintained as a part of the gen-bank collection of the Institute of Agriculture and Tourism in Poreč, Croatia. Both cabbage landraces included in this study are commonly grown by local farmers, with landrace 'Brgujski' characterized by enhanced anthocyanin coloration and landrace 'Žminjski' by rather slight anthocyanin coloration.
Kale samples consisted of three accessions belonging to two different morphotypes, one with blistered leaves (accessions IPT408, KAL1, and IPT418, KAL2) and one with flat leaves (accession IPT379, KAL3). All three included kale accessions are local populations previously collected from farmers and are also maintained as a part of the mentioned gen-bank collection.
Seeds of wild relatives B. incana (accession IPT515, INC) and B. mollis (accession IPT517, MOL) were collected at their natural habitats at Vis and Koločep islands, Adriatic seaside, Croatia, respectively. Seeds were germinated and plants were grown in growth chambers at 24 • C under a photoperiod 16/8 (day/night) over two months. Next, plants were transferred to a greenhouse for an adaptation period, and after two months they were transferred outside of the greenhouse.
All the plants were cultivated in the experimental field of the Institute of Agriculture and Tourism. Diplotaxis tenuifolia (accession IPT551, WR) was collected from a wild population growing under very similar pedoclimatic conditions near the Institute's experimental field.
All the samples were harvested manually from 30 October to 17 November 2020. During sampling, five leaves were taken from the top portion of five fully formed, solid, and undamaged heads of the two cabbage landraces. Five samples, each consisting of two young, healthy leaves, were taken from mature plants of three kale accessions, savoy cabbage, and two wild species. Samples of D. tenuifolia consisted of 10 fully developed, undamaged leaves from mature plants.

Analysis of Volatile Compounds
Leaf samples were homogenized by a commercial blender for 30 s and 1 g of each was weighed into a glass vial of 10 mL. Four mL of deionized water was added, and the vial was sealed with a screw aluminum cap and silicone/PTFE septa. A divinylbenzene/carboxen/polydimethylsiloxane (DVB-CAR-PDMS, StableFlex, 50/30 µm, 1 cm) SPME fiber (catalog number 57328-U, Supelco, Bellafonte, PA, USA) was preconditioned above the sample for 5 min at 40 • C and then subjected to the vapors in the headspace for 15 min at 40 • C with stirring (800 rpm). Volatile compounds were desorbed in a GC/MS injector at 248 • C for 10 min (3 min splitless mode). GC/MS identification and quantification was performed using a Varian 3900 gas chromatograph (GC) connected to a Varian Saturn 2100T mass spectrometer with an ion trap analyzer (Varian Inc., Harbour City, CA, USA). The GC was equipped with a 60 m × 0.25 mm i.d. × 0.25 µm d.f. capillary column Rtx-WAX (catalog number RTK-12426, Restek, Belafonte, PA, USA). The GC column was initially heated at 40 • C, the temperature was then increased at 3.8 • C/min to 200 • C, then raised by 30 • C/min to 240 • C and kept constant for the final 15 min. The carrier gas was helium at a flow rate of 1.2 mL/min. EI mode (70 eV) was used to acquire mass spectra in the 30-350 m/z range. Volatile compounds were identified by comparison of their mass spectra with mass spectra from NIST05 library. Spectra reverse match number (RM) higher than 800 was used as a criterion, whereas in particular cases for spectra with RM < 800 the identification was performed based on the satisfactory ratio of relative intensities of a quantifier ion and the next three characteristic ions with the highest intensity. Additional confirmation of the identity of volatile compounds was achieved by comparing linear retention indices calculated relative to the retention times of C 10 to C 28 n-alkanes to those reported in literature obtained using equal or equivalent capillary columns. The average values of peak area divided by 1000 obtained after five replicate analyses (n = 5) were used in further data elaboration and reporting.

Statistical Data Elaboration
Data obtained by GC/MS were subjected to analysis of variance (one-way ANOVA) and the average values were compared by least significant difference (LSD) post hoc test at p < 0.05 level. The data were further processed by multivariate statistical analysis using hierarchical cluster analysis (HCA) and partial least squares-discriminant analysis (PLS-DA). HCA is an unsupervised multivariate statistical method in which distances between samples (cases) are calculated and samples are grouped into categories based on similar characteristics defined by the variables' (e.g., volatile compounds') values. PLS-DA is a supervised multivariate statistical method that minimizes the variance within and maximizes the variance between different categories (e.g., species) and gives information about the most useful variables (e.g., volatile compounds) in the form of variable importance in projection (VIP) scores. Statistica v. 13.2 software (StatSoft Inc., Tulsa, OK, USA) was used to perform ANOVA, whereas MetaboAnalyst v. 5.0 (http://www.metaboanalyst.ca, accessed on 24 October 2022) was applied for HCA and PLS-DA.

Conclusions
The HS-SPME-GC/MS method used in this study allowed for a detailed characterization of the volatilomes of several Brassicaceae species, landraces, and accessions. Each of them exhibited many unique features and volatile markers that differentiated the species among each other well, and also the landraces and accessions of the same species. On the other hand, many specific similarities between the volatilomes of different species were observed. In general, two cabbage landraces emitted the lowest amounts of volatiles generated in the LOX pathway. Wild species B. incana and B. mollis were characterized by relatively high trans-2-hexenal/cis-3-hexen-1-ol ratio compared to the other investigated vegetables. The investigated Savoy cabbage cultivar and three kale accessions exhibited particular similarities in the composition of LOX volatiles, while the LOX volatilome fraction in B. incana and B. mollis partially coincided with that of another wild species, D. tenuifolia. Regarding the volatiles formed in the GSL pathway, Savoy cabbage and the three wild species showed more intense emission of ITCs than cabbage landraces and kale accessions. Diplotaxis tenuifolia showed rather limited production of nitriles. Judging from their overall volatilomes, wild species B. incana and B. mollis turned out to be more similar to kale than to cabbage. The results of this study contribute to the general knowledge about volatile composition of vegetables from the Brassicaceae family and contribute to a better understanding of the potential of particular Brassicaceae species, landraces, and accessions in responding to stimuli from the environment. Additionally, the information obtained could be useful for better understanding the sensory quality of the investigated Brassicaceae species, which could be exploited for their better valorization. Future studies should focus on the influence of various environmental, cultivation, and post-harvest storage factors to obtain data with a higher level of applicability in practice.

Supplementary Materials:
The following supporting information can be downloaded at: https:// www.mdpi.com/article/10.3390/plants12010177/s1, Table S1: Morphological parameters of different Brassicaceae species.  Data Availability Statement: All data generated or analyzed during this study are included in this published article.