Flavour Volatiles of Fermented Vegetable and Fruit Substrates: A Review

Health, environmental and ethical concerns have resulted in a dramatic increase in demand for plant-based dairy analogues. While the volatile organic compounds (VOCs) responsible for the characteristic flavours of dairy-based products have been extensively studied, little is known about how to reproduce such flavours using only plant-based substrates. As a first step in their development, this review provides an overview of the VOCs associated with fermented (bacteria and/or fungi/yeast) vegetable and fruit substrates. Following PRISMA guidelines and using two English databases (Web of Science and Scopus), thirty-five suitable research papers were identified. The number of fermentation-derived VOCs detected ranged from 32 to 118 (across 30 papers), while 5 papers detected fewer (10 to 25). Bacteria, including lactic acid bacteria (LAB), fungi, and yeast were the micro-organisms used, with LAB being the most commonly reported. Ten studies used a single species, 21 studies used a single type (bacteria, fungi or yeast) of micro-organisms and four studies used mixed fermentation. The nature of the fermentation-derived VOCs detected (alcohols, aldehydes, esters, ketones, acids, terpenes and norisoprenoids, phenols, furans, sulphur compounds, alkenes, alkanes, and benzene derivatives) was dependent on the composition of the vegetable/fruit matrix, the micro-organisms involved, and the fermentation conditions.


Introduction
The demand for plant-based foods is rapidly increasing. Consumers' reasons for eating more plant-based foods are varied but mainly revolve around a desire to enhance their health by reducing their risk of diseases such as heart disease, cancer, and type 2 diabetes [1], concerns around the sustainability of meat and dairy-based food production systems [2], and/or a desire to move away from the exploitation of animals [3]. Accordingly, the global market for plant-based foods is predicted to reach USD 480.43 billion by 2024 with a predicted CAGR (compound annual growth rate) of 13.82% between 2019 and 2024 [4]. Despite plant-based diets becoming increasingly popular, many consumers still prefer to prepare their meals from foods that are familiar to them in terms of their flavour, appearance, preparation, and cooking method [3]. In response to these needs, the number and range of dairy analogues available on the market are rapidly increasing [5]. While currently available dairy analogues generally have a realistic texture and appearance, their flavour is often uncharacteristic of the dairy products they are attempting to mimic.
Flavour is an important sensory attribute of food, and it is a multimodal sensory experience with the major modality contributors being taste and smell (retronasal). A wide range of non-volatile organic compounds (NVOCs) along with multiple odour-active volatile organic compounds (VOCs) contribute to the taste, aroma, and overall perceived flavour of a food [6,7]. VOCs are typically small compounds (up to C 20 ), which have a low molecular weight (<300-400 Da), and a relatively high vapour pressure at ambient temperature. Such compounds can be easily transferred into the vapour phase, which Extracted and analysed using HS-SPME-GC-MS Acids (14) Alcohols (10) Aldehydes (19) Esters (8) Ketones (12) [20] 5 Apple juice L. acidophilus L. rhamnosus L. casei L. plantarum Inoculum: 7.0 log CFU/mL pH adjusted to 6.0 using food grade Na 2 CO 3 at 37 • C for 80 h Aroma profile analysed by electronic nose system and characterized by HS-SPME/GC-MS.

Fermented Flavours of Vegetables and Fruits
Twenty-three of the thirty-five papers described VOCs originating from the fermentation of fruit juices (with 2 of the 23 on goji/wolfberry juice, 2 on jujube juice/pulp, 3 on cashew apple juice, and 3 on apple juice). Nine of the papers described VOCs originating from the fermentation of vegetable substrates (with 2 of the 9 on okara (soybean pulp)), with one investigating tomato and red pepper pomace. Three articles investigated the VOCs generated from a mixture of vegetables and mixtures of vegetables and fruits, with carrot juice being a component in all three studies and apple juice in two.
A wide range of micro-organisms were reported as being used, encompassing bacteria, LAB, fungi, and yeast. Thirty-three papers described the use of LAB, with twenty-nine studies only referring to LAB strains, one paper referring to LAB combined with other bacteria, and three papers referring to LAB and yeasts. In the remaining two studies, one study referred to the use of fungi, while the other referred to fungi in combination with yeast.  The most commonly mentioned LAB was Lactiplantibacillus plantarum (L. plantarum) (28 papers), followed by Lacticaseibacillus casei (L. casei) (12 papers), Lactobacillus acidophilus (L. acidophilus) (9 papers), Lacticaseibacillus rhamnosus (L. rhamnosus) (9 papers), Lactobacillus helveticus (L. helveticus) (4 papers), and Streptococcus thermophilus (S. thermophilus) (4 papers). Only four of the papers investigated the use of yeast in combination with LAB and fungi.
The most common fermentation temperature and time combination was 37 • C for 48 h (10 papers), as shown in Table 2. As additional examples, 9 papers described studies carried out at 37 • C with fermentation times (excluding 48 h) ranging from 20 to 120 h. Six papers described studies carried out at 30 • C with fermentation times ranging from 24 to 120 h. In 4 papers where multiple LAB strains were investigated, fermentation temperatures of 37 • C and 30 • C were used at different times.
Thirty-three of the papers isolated VOCs using headspace solid-phase microextraction (HS-SPME), and one study using a purge and trap method, while other one using a static headspace technique. HS-SPME is a simple, rapid, solvent-free method that can extract a diverse mixture of VOCs using a fibre coated with an adsorbent resin. The fibre most frequently stated as being used was 50/30 µm DVB/CAR/PDMS (Divinylbenzene-Carboxen/polydimethylsiloxane) (17 papers), followed by 75 µm CAR/PDMS (6 papers), 75 µm DVB/CAR/PDMS (1 paper), and 50/30 µm DVB/CAR (2 papers). For SPME, the most frequently reported adsorption time was 30 min (19 papers) for temperatures ranging from 35 to 85 • C, with 9 papers using 30 min at 40 • C. The remaining papers reported the use of a wide range of time and temperature combinations, ranging from 7 to 60 min at 40 to 80 • C. Gas chromatography-mass spectrometry (GC-MS) was used in 34 papers to detect the extracted VOCs, and the remaining paper reporting using gas chromatography-ion mobility spectrometry (GC-IMS).
Ethanol is synthesised from sugars naturally present in plants; LAB utilise sugar the phosphoketolase (PK) pathway, and yeast utilise sugars through the Embd
Ethanol is synthesised from sugars naturally present in plants; LAB utilise sugars via the phosphoketolase (PK) pathway, and yeast utilise sugars through the Embden-Meyerhof-Parnas (EMP) pathway [13]. In mango slurry, fermentation involving the yeast Saccharomyces cerevisiae (S. cerevisiae) generated 30-100 times more ethanol than LAB fermentation [41]. When Williopsis saturnus var. saturnus (W. saturnus) yeast was combined with LAB for fermentation, there was also a marked increase in the ethanol concentration (six-fold increase) compared to LAB alone in durian pulp [30]. However, ethanol can be generated by heterofermentative LAB, which possess the alcohol dehydrogenase enzyme that converts acetaldehyde into ethanol [26]. In kiwifruit juice (cultivars of Actinidia deliciosa cv. Xuxiang and Actinidia chinensis cv. Hongyang), the ethanol concentration was 10,316.5, 17,249.2, and 8652.7 ng/mL in Xuxiang cultivar juice fermented by either L. acidophilus, L. helveticus, or L. plantarum, respectively, compared to 6242.9 ng/mL in the unfermented juice. However, the ethanol concentration was 13,042.5, 7004.2, and 9551.9 ng/mL in Hongyang cultivar juice fermented by either L. acidophilus, L. helveticus, or L. plantarum, respectively, compared to 19,642.6 ng/mL in the unfermented juice [49]. The different ethanol concentrations produced from the two kiwifruit cultivars after LAB fermentation could be a result of the different substrate compositions of the cultivars, which are subjected to various metabolic pathways by LAB. After fermentation of orange pomace by L. rhamnosus, 0.3 µg/mL of ethanol was detected in a distillate prepared using vacuum distillation to extract VOCs. However, ethanol was not detected in a fermented orange pomace distillate prepared using a simple distillation method. In the same study, the ethanol concentration detected in distillates prepared from L. rhamnosus fermented melon by-product using vacuum distillation was 6.5 µg/mL, compared to 1.3 µg/mL in the distillate from the unfermented melon by-products, while ethanol was not detected in distillates prepared from the same samples using simple distillation [44]. In papaya juice, the ethanol concentration was significantly (p < 0.05) increased by 7 and 11 times after fermentation by either L. plantarum or L. acidophilus, respectively, compared to the concentration in the unfermented juice [23]. On the other hand, only small changes in ethanol were observed in two studies: (1) Ricci et al. [29] found that LAB fermented cherry juice had an ethanol concentration of 4.1-8.5 ng/mL, compared to 3.1-3.7 ng/mL in the unfermented juice, and (2) in watermelon juice fermented by either L. rhamnosus, L. plantarum, L. casei, or Pediococcus pentosaceus (P. pentosaceus), the ethanol concentration was 16.8, 15.2, 15.1, and 15 ng/mL, respectively, compared to 14.6 ng/mL in the unfermented juice; however, after fermentation by Levilactobacillus brevis (L. brevis), the ethanol concentration was 13.9 ng/mL [48]. Six studies reported that LAB fermentation reduced the ethanol concentration in fermented fruit and vegetable juices, compared to the unfermented juices, with the synthesis of various esters speculated to have caused this decrease: (1) In unfermented Chinese wolfberry juice, the ethanol concentration was 5501.3 µg/mL, compared to 1364.7 µg/mL in the L. acidophilus fermented juice, where it was not detected in the juice fermented by other LAB strains [50]; (2) in two varieties of unfermented jujube (Muzao, and Hetain) juices, the ethanol concentration was 6850, and 6130 ng/mL, respectively, compared to 5740, 5100, and 1530 ng/mL in Muzao fermented with either L. helveticus, L. casei, or L. plantarum, respectively, and 5380, 4400, 2660, and 2410 ng/mL in Hetain fermented with either L. casei, L. acidophilus, L. plantarum, or L. helveticus, respectively [20]; (3) in okara, the initial ethanol concentration was 44 µg/g which reduced to 20.4 and 13.8 µg/g after fermentation with LAB monoculture (L. rhamnosus or Pediococcus acidilactic (P. acidilactic), respectively) and to 19.6 µg/g after co-culture fermentation (L. acidophilus, L. rhamnosus, and P. acidilactic). However, in okara fermented with an L. acidophilus monoculture, the ethanol concentration increased from 44 to 57.4 µg/g [46]; (4) in unfermented apple juice, the ethanol concentration was 188.4 ng/mL, compared to 83.4-123.5 ng/mL after fermentation with various LAB strains [19]; (5) in non-pH-adjusted (2.7) sea buckthorn juice, the ethanol concentration was 170.7 ng/mL, which reduced after fermentation for 36 and 72 h by L. plantarum to 165.4 and 152 ng/mL, respectively. However, if the pH of the juice was adjusted to pH 3.5, the initial ethanol concentration of 166.3 ng/mL increased after L. plantarum fermentation for 36 and 72 h to 194.6 and 206.5 ng/mL, respectively [42]; and (6) in tomato juice, the ethanol concentration after fermentation with either L. plantarum or L. casei was 2.7 and 1.2 times lower, respectively, compared to its concentration in the unfermented juice [32].
1-Octanol is a fatty alcohol produced by micro-organisms utilizing glucose as a substrate through a fatty acid synthesis pathway using various enzymes [53]. The 1-octanol concentration was generally reported to increase after LAB fermentation in the 5 studies reviewed: (1) In Chinese wolfberry juice fermented by either L. plantarum, L. casei, Lacticaseibacillus paracasei (L. paracasei), L. acidophilus, L. helveticus, or Bifidobacterium Lactis (B. lactis), the 1-octanol concentration was 172.9, 119.1, 137.2, 209.3, 131.4, and 142.4 µg/mL, where it was not detected in the unfermented juice [50]; (2) the 1-octanol concentration in a distillate prepared using vacuum distillation from orange pomace fermented by L. rhamnosus was 1.5 µg/mL, compared to 0.1 µg/mL in the distillate from unfermented pomace; in distillates prepared using a simple distillation method, the 1-octanol concentration in the fermented orange pomace distillate was 2.1 µg/mL, compared to 1.8 µg/mL in the unfermented pomace distillate. Interestingly in the same study, using the simple distillation method, the 1-octanol concentration in the fermented melon by-product distillate was 21.5 µg/mL, compared to 7.1 µg/mL in the unfermented by-product distillate, whereas in extracts prepared by vacuum distillation, the 1-octanol concentration in the distillate from melon by-product fermented by L. rhamnosus was 1.1 µg/mL, compared to 0.2 µg/mL in the unfermented by-product distillate [44]; (3) in kiwifruit juice (Xuxiang and Hongyang cultivars), the 1-octanol concentration was 285.5 and 325.5 ng/mL in Xuxiang cultivar juice fermented by either L. helveticus or L. plantarum, respectively, compared to 146.4 ng/mL in the unfermented juice, where it was not detected in Xuxiang cultivar juice fermented by L. acidophilus. Interestingly, with the Hongyang cultivar juice, 1-octanol was not detected in the unfermented juice or in any of the LAB fermented juices. [49]; (4) in cherry juice fermented by either L. plantarum, L. rhamnosus, or L. paracasei, the 1-octanol concentration was 4.5-7.8, 8.4, and 5.2 ng/mL, respectively, compared to 3.4-3.6 ng/mL in the unfermented juice [29]; and (5) in apple juice fermented by either L. acidophilus, L. rhamnosus, L. casei or L. plantarum, the 1-octanol concentration was 4.2, 3.5, 3.8, and 4.0 ng/g, respectively, compared to 1.0 ng/g in the unfermented juice [21]. However, in okara fermented by LAB monoculture of either L. rhamnosus, P. acidilactic or co-culture (L. acidophilus, L. rhamnosus, and P. acidilactic), the 1-octanol concentration was 11.7, 17.8, and 1.7 µg/g, respectively, compared to 30.0 µg/g in the unfermented okara, whereas okara fermented with L. acidophilus had an 1-octanol concentration of 34.7 µg/g [46].
In 10 studies, the concentration of 2-phenylethyl alcohol, which has a flowery smell, was increased or it was detected after fermentation: (1) In Chinese wolfberry juice fermented by L. plantarum, the 2-phenylethyl alcohol concentration was 246.4 µg/mL, where it was not detected in the unfermented juice or the fermented juice by other LAB strains [50]; (2) in non-pH-adjusted (2.65) bog bilberry juice fermented by two strains of L. plantarum, the 2phenylethyl alcohol concentration was 1731.2 and 1775.8 ng/mL, compared to 663.5 ng/mL in the unfermented juice. However, if the pH of the juice was adjusted to pH 3.5, the initial 2-phenylethyl alcohol concentration of 617.5 ng/mL was decreased after fermentation by two strains of L. plantarum to 459.7 and 463.4 ng/mL [34]; (3) in grape juice fermented by LAB, the 2-phenylethyl alcohol concentration was 40.6 ng/mL, where it was not detected in the unfermented juice [43]; (4) in horse gram sprouts fermented by two L. plantarum strains, the 2-phenylethyl alcohol concentration was 1290 and 780 ng/g, compared to 40 ng/g in raw seed [37]; (5) in goji juice fermented by different combinations of bacterial strains (either L. plantarum, L. rhamnosus, L. reuteri, B. velezensis, or B. licheniformis), the 2-phenylethyl alcohol concentration ranged from 362.6 to 494.0 ng/g, compared to 103.1 ng/g in the unfermented juice [25]; (6) in jujube pulp fermented by a mixture of L. plantarum, L. rhamnosus, and S. thermophilus, the 2-phenylethyl alcohol concentration was 283 ng/g, where it was not detected in the unfermented juice [47]; (7) in okara fermented with a monoculture of Rhizopus oligosporus (R. oligosporus) fungi, the 2-phenylethyl alcohol concentration increased by 20 times, compared to its concentration in the unfermented okara, whereas with a mixed culture of R. oligosporus fungi and Yarrowia lipolytica (Y. lipolytica) yeast, the concentration increased by 8.5 times [38]; (8) in mango slurry fermented by yeast S. cerevisiae, the 2-phenylethyl alcohol concentration was 4 to 23 times higher, compared the concentration after LAB fermentation, where it was not detected in the unfermented mango slurry [41]; (9) in papaya juice fermented by L. plantarum, the 2-phenylethyl alcohol concentration was doubled, compared to the concentration in L. acidophilus fermented juice, where it was not detected in the unfermented juice [23]; and (10) in durian pulp fermented by L. casei mixed with the yeast W. saturnus, 2-phenylethyl alcohol was detected, where it was not detected in the unfermented pulp, or in the pulp fermented by a L. casei monoculture [30].
Furthermore, compared to unfermented juice, one study reported that almost half of the alcohols detected decreased in L. casei fermented tomato juice, likewise in L. plantarum fermented tomato juice, most of the alcohols detected decreased. However, due to the generation of new alcohols, the relative peak area (RPA) for total alcohols increased to 59.9% and 49.7% in juice fermented by either L. casei or L. plantarum, respectively, compared to a 49.3% RPA in the unfermented juice [32]. On the other hand, LAB fermentation increased the overall combined alcohol concentration of fruit and vegetable juices in 4 studies: (1) LAB fermentation of apple juice increased the overall combined alcohol concentration by 10 times compared to its concentration in the unfermented juice, demonstrating that most of the alcohols were produced during fermentation [21]; (2) LAB fermentation of grape juice increased the total combined alcohol concentration by 102.4% [43]; (3) LAB fermentation of kiwifruit juice (Xuxiang and Hongyang cultivars) increased the total combined alcohol concentration by 39, 107, and 56% in Xuxiang cultivar juice fermented by either L. acidophilus, L. helveticus, or L. plantarum, respectively, and by 25, 30, and 26% in Hongyang cultivar juice fermented by either L. acidophilus, L. helveticus or L. plantarum, respectively [49]; and (4) LAB fermentation increased the total combined alcohol concentration of jujube juice (Varieties of Muzao and Hetian) by 66.5% in L. acidophilus fermented Muzao juice and 33.7% in L. casei fermented Hetian juice [20]. In another study, the total combined alcohol concentration of mango slurry fermented with yeast strains was nearly 10 times higher compared to mango slurry fermented with LAB strains [41].

Esters
Esters, which have sweet and fruity notes, are formed when carboxylic acids linked with coenzyme-A (CoA) are esterified with alcohols [60]. Volatile esters were found in fermented fruit and vegetable juices in 27 of the 35 papers reviewed in this report. The sensory detection threshold for esters is lower than that of the corresponding alcohol or acid [61]. The majority of esters reported were either ethyl esters or acetate esters. Ethanol and acyl-CoA derivatives of fatty acids combine to form ethyl esters. Acetyl-CoA and alcohols, such as ethanol or higher alcohols, produced from amino acid metabolism, combine to form acetate esters [62]. Eleven of the papers stated that ethyl acetate, which is formed by alcohol acetyltransferases from the reaction between acetyl Co-A and ethanol [20], was primarily responsible for the fruity flavour of fermented fruits and vegetable juices. A variety of LAB strains [19][20][21][22][23]33,44,45,50] were used in all studies except two; one used LAB combined with a yeast [30], and the other used fungi combined with a yeast [38]. In these studies, it was reported that for Chinese wolfberry juice fermented by either L. plantarum, L. paracasei, L. acidophilus, or L. helveticus, the ethyl acetate concentration was 6931, 4827.1, 4925.4, and 7323.3 µg/mL, respectively, compared to 774.5 µg/mL in the unfermented juice, where it was not detected in the juice fermented by L. casei or B. lactis [50]. In Muzao jujube juice fermented by either L. plantarum or L. acidophilus, the ethyl acetate concentration was 111.7 and 64.2 µg/mL, respectively, where it was not detected in the unfermented juice or juice fermented by other LAB [20]. Further, in durian pulp fermented by L. casei combined with a yeast W. saturnus, ethyl acetate was detected, whereas it was not detected in sole L. casei fermentation or unfermented pulp [30]. However, Liu et al. [32] found that in tomato juice, prior to fermentation, ethyl acetate was detected, where it was not detected after fermentation by LAB, and in mango slurry fermented by either yeast or LAB, the ethyl acetate concentration was reduced by 1.2-1.8 times, compared to its concentration in the unfermented slurry [41].
In 3 experiments, ethyl butanoate, which is formed by a reaction between ethanol and butyryl-CoA during LAB fermentation [21], was the next most common ester compound: (1) The ethyl butanoate concentration in a distillate prepared using simple distillation of L. rhamnosus fermented melon by-product was 8760 ng/mL, compared to 700 ng/mL in the unfermented melon by-product distillate, and the ethyl butanoate concentrations of both the fermented and unfermented by-product distillates produced using vacuum distillation were at least 10 times lower than in the simple distillation distillates [44]; (2) in apple juice fermented by different LAB, the ethyl butanoate concentration ranged from 16.3 to 23.1 ng/g, compared to 2.1 ng/g in the unfermented juice [21]; and (3) in papaya juice fermented by LAB, ethyl butanoate was detected, where it was not detected in the unfermented juice [23].
2-Phenylethyl acetate, which is formed by a reaction between 2-phenylethyl alcohol and acetyl CoA, was the third most commonly reported (3 studies) ester in juices after fermentation: (1) In horse gram sprouts fermented by L. plantarum, the 2-phenylethyl acetate concentration was 220 ng/g, where it was not detected in raw seeds [37]; (2) in grape juice fermented by a mixed culture of L. plantarum and L. brevis, the 2-phenylethyl acetate concentration was 3.5 ng/mL, compared to its concentration in the unfermented juice (1.6 ng/mL) [43]; and (3) in durian pulp fermented by a combination of L. casei and W. saturnus yeast, 2-phenylethyl acetate was detected, whereas it was not detected during L. casei only fermentation or in the unfermented pulp [30].
Propyl acetate, which is formed by a reaction between propanol and acetyl-CoA, was detected in cherry juice fermented by either L. plantarum, L. rhamnosus, or L. paracasei, where the propyl acetate concentration was 18.5-83.4, 1186.7, and 201.6 ng/mL, respectively, compared to about 0.01 ng/mL in the unfermented juice. In this study, the formation of propyl acetate during fermentation appeared to correlate with acetic acid production. As there was a low concentration of acetic acid after fermentation by L. rhamnosus or L. paracasei, this was taken as evidence of the conversion of acetic acid to the corresponding ester. In contrast, in the same study, fermentation by L. plantarum resulted in a high concentration of acetic acid and a lower concentration of propyl acetate [29].
Overall the total combined ester concentration increased in 6 studies after fermentation due to the availability of alcohol precursors [52]: (1) In Muzao jujube juice fermented by either L. acidophilus or L. plantarum, the total combined ester concentration was 65.4 and 156.7 µg/mL, respectively, compared to 4.7 µg/mL in the unfermented juice [20]; (2) in mixed juices (apple juice, orange juice, carrot juice, and Chinese jujube juice) fermented by LAB mixed culture (L. plantarum, Bifidobacterium breve (B. breve), and S. thermophilus), the total combined ester concentration was 415 ng/mL, compared to 239 ng/mL in the unfermented juice [35]; (3) in apple juice fermented by different LAB, the total combined ester concentration ranged from 81.8 to 92.9 ng/g, compared to 33.7 ng/g in the unfermented juice [21]; (4) in grape juice fermented by LAB, the total combined ester concentration increased by 83.76%, compared to its concentration in the unfermented juice [43]; (5) in mango slurry fermented by LAB and yeast, the total combined ester concentration increased, compared to the unfermented slurry, with the yeast S. cerevisiae generating a significantly (p < 0.05) higher number of esters present at a high concentration than LAB [41]; and (6) in pomegranate juice fermented by LAB, the total combined ester concentration increased, compared to the concentration in the unfermented juice [24].
On the other hand, 3 studies reported a reduction in the total combined ester concentration after fermentation, possibly due to hydrolysis into their corresponding acids and alcohols [47]: (1) In apple juice fermented by either L. plantarum, L. helveticus, L. casei, L. paracasei, L. acidophilus or B. lactis, the total combined ester concentration was 1090, 1279.6, 787.5, 695.9, 702.3, and 643.1 ng/mL, respectively, compared to 1410.7 ng/mL in the unfermented juice [19]; (2) in jujube juice fermented by a mixture of L. plantarum, L. rhamnosus, and S. thermophilus, the total combined ester concentration was 1541 ng/g, compared to 5814 ng/g in the unfermented juice [47]; and (3) in tomato juice fermented by either L. casei or L. plantarum, the total combined ester concentration was 1.6 times and 7 times lower, respectively, compared to the concentration in the unfermented juice [32].
Overall, esters such as ethyl acetate (11 papers), ethyl butanoate (3 papers), 2-phenylethyl acetate (3 papers), and propyl acetate (1 paper) have been reported to increase or were only detected after fermentation of fruit and vegetable juices, mainly by LAB. However, the concentration of ethyl acetate did decrease after fermentation in 2 studies.
3-Hydroxy-2-butanone (acetoin), which imparts a creamy/buttery note, was the most frequently detected ketone produced during the fermentation of vegetables and fruits. Citrate in vegetable and fruit juices can be directly converted to acetoin (Figure 3) by some LAB strains exhibiting citrate permease and citrate lyase activities. Citrate can be converted by LAB to pyruvate via oxaloacetate, then to acetaldehyde-thiamine pyrophosphate (TPP) through a decarboxylation process, and finally to acetaldehyde-TPP through an enzymatic reaction involving α-acetolactate synthase, resulting in the synthesis of α-acetolactate. α-Acetolactate synthase has a low affinity for pyruvate; therefore, an excess of pyruvate is required to favour this reaction. In the presence of citrate and sugars, homofermentative LAB will convert pyruvate directly to α-acetolactate when less NADH is generated than pyruvate. Heterofermentative LAB will, however, accumulate pyruvate at low pH when citrate is the sole carbon source. Further, due to the instability of α-acetolactate, it is readily decarboxylated enzymatically or chemically to yield acetoin. Acetoin can also be synthesised from diacetyl via the enzyme diacetyl acetoin reductase [63][64][65][66]. In addition, when the pH of the medium is between 5 and 8, Lactococcus lactis can also produce acetoin from the catabolism of aspartic amino acid [67] (Figure 4). The acetoin concentration increased after LAB fermentation of fruit and vegetable juices in 8 studies: (1) In Chinese wolfberry juice fermented by either L. plantarum, L. paracasei, L. helveticus, or B. lactis, the acetoin concentration was 346.3, 267.4, 528.1, and 422.1 µg/mL, respectively, compared to 29.1 µg/mL in the unfermented juice, where it was not detected in the juice fermented by L. casei or L. acidophilus [50]; (2) in Muzao jujube juice fermented either L. plantarum or L. helveticus, the acetoin concentration was 29.9 and 30.8 µg/mL, respectively, compared to 17.5 µg/mL in the unfermented juice. However, the acetoin concentration was reduced to 10.5 µg/mL in L. acidophilus fermented juice and acetoin was not detected in L. casei fermented juice [20]; (3) in kiwifruit juice (Xuxiang, and Hongyang cultivars), the acetoin concentration was 2621.6 and 1348.9 ng/mL in the Xuxiang cultivar juice fermented by either L. helveticus or L. plantarum, respectively, where it was not detected in the L. acidophilus fermented juice or in the unfermented juice. The acetoin concentration was 8431.7 and 4390.6 ng/mL in Hongyang cultivar juice fermented by either L. helveticus or L. plantarum, respectively, where it was not detected in the L. acidophilus fermented juice or in the unfermented juice [49]; (4) in elderberry juice fermented by either L. plantarum, L. casei, or L. rhamnosus, the acetoin concentration was 83.1-496.7, 90.7-314.5, and 41.4-456.2 ng/mL, respectively, compared to 1.4-22.1 ng/mL in the unfermented juice [22]; (5) in cherry juice fermented by either L. rhamnosus or L. paracasei, the acetoin concentration was 260.7 and 5.9 ng/mL, respectively, compared to 0.001 ng/mL in the unfermented juice, and in cherry juice fermented by different L. plantarum strains, the acetoin concentration ranged from 44 to 287.9 ng/mL, compared to 0.002 ng/mL in the unfermented juice [29]; (6) the acetoin concentration in a distillate prepared using vacuum distillation from orange pomace fermented by L. rhamnosus was 450 ng/mL, compared to 110 ng/mL in the unfermented pomace distillate, where acetoin was not detected in distillates from either fermented or unfermented pomace using the simple distillation method [44]; (7) in mung beans fermented by two L. plantarum strains, the acetoin concentration was 2.8 and 7.5 times higher, compared to the concentration in the unfermented mung beans [45]; and (8) in papaya juice fermented by either L. plantarum or L. acidophilus, the acetoin concentration was 2.2 and 3.7 times higher, respectively, compared to the unfermented juice [23]. In 5 studies, acetoin was only detected after fermentation of juices: (1) In okara fermented by a LAB co-culture (L. acidophilus, L. rhamnosus, and P. acidilactici), the acetoin concentration was 166.3 µg/g, where it was not detected in LAB monocultures [46]; (2) in horse gram sprouts fermented by L. plantarum, the acetoin concentration was 440 ng/g [37]; (3) in goji juice fermented by a bacterial mixture of L. rhamnosus, L. reuteri, and B. velezensis, the acetoin concentration was 87.2 ng/g juice [25]; (4) in mango slurry fermented by S. thermophilus, acetoin was detected, where it was not detected after fermentation by yeast S. cerevisiae or other LAB [41]; and (5) in durian pulp fermented by L. casei in sequential co-culture with W. saturnus yeast, acetoin was detected, where it was not detected in L. casei monoculture [30]. Note, that in apple juice fermented by either L. acidophilus, L. helveticus, or L. paracasei, the acetoin concentration was reported to have decreased to 0.8 ng/mL, 1.2 ng/mL, and 4.3 ng/mL, respectively, compared to 5.4 ng/mL in the unfermented juice, where it was not detected in L. plantarum, L. casei, or B. lactis fermented apple juice [19].
The second most commonly reported ketone was 2,3-butanedione (diacetyl), which imparts creamy/buttery notes. It is produced by LAB from citrate present in juice ( Figure 3). As discussed for the acetoin production pathway, α-acetolactate can be directly converted into diacetyl through nonenzymatic oxidative carboxylation in the presence of molecular oxygen [63,65]. Diacetyl was reported to be increased by the presence of some LAB during fermentation of juices in 6 studies: (1) In Chinese wolfberry juice fermented by either L. plantarum, L. casei, L. paracasei, or L. acidophilus, the diacetyl concentration was 45.1, 51, 71.9, and 28.1 µg/mL, respectively, where it was not detected in the unfermented juice nor the juice fermented by L. helveticus or B. lactis [50]; (2) in elderberry juice fermented by either L. rhamnosus, L. plantarum, or L. casei strains, the diacetyl concentration ranged from 37-586.8, 16.2-400.7, and 221.9-276.6 ng/mL, respectively, compared to 3.3-12.2 ng/mL in the unfermented juice [22]; (3) in kiwifruit juice (Xuxiang, and Hongyang cultivars), the diacetyl concentration was 261.1 ng/mL in Xuxiang cultivar juice fermented by L. helveticus, where it was not detected in Xuxiang cultivar juice fermented by other LAB or in the unfermented juice. Interestingly, with the Hongyang cultivar juice, diacetyl was not detected in the unfermented juice or in any of the LAB fermented juices [49]; (4) in watermelon juice fermented by either L. plantarum, L. brevis, L. casei, or L. rhamnosus, the diacetyl concentration was 1.46, 1.47, 62.5, and 85.7 ng/mL, respectively, where it was not detected in the P. pentosaceus fermented juice and the unfermented juice [48]; (5) in mango slurry fermented by L. casei, diacetyl was detected, whereas it was not detected in other LAB or yeast fermentations [41]; and (6) in pomegranate juice fermented by L. plantarum strains, the diacetyl concentration increased, compared to the concentration in the unfermented juice [24]. However, in tomato and pepper pomace fermented by either Trichoderma atroviride (T. atroviride) or Aspergillus sojae (A. sojae), the diacetyl concentration was reduced compared to the concentration present in the unfermented pomace [40].
unfermented juice, whereas in mango slurry fermented by the yeast S. cerevisiae, the total combined ketone concentration was 1.4-2.7 times lower compared to the concentration in the unfermented juice [41].
Overall, ketones, which are key contributors to dairy notes, such as acetoin (13 papers) and diacetyl (6 papers) have been reported to increase or were only detected after the fermentation of fruit and vegetable juices. However, in two studies, the concentration of acetoin (1 paper), and diacetyl (1 paper) decreased after fermentation by LAB and fungi, respectively.
In 8 studies, L. plantarum was the main LAB producing acetoin in fermented juices, followed by L. helveticus (3 studies) and L. rhamnosus (3 studies). The two main LAB that produced high diacetyl concentrations in fermented juices were L. plantarum (4 papers) and L. casei (4 papers). Overall, for the papers reviewed, L. plantarum produced more of the creamy flavours of acetoin and diacetyl compared to other LAB strains studied.

Aldehydes
Aldehydes were present at lower concentrations after the fermentation of fruit and vegetables in 26 out of 35 studies. During the fermentation process, aldehydes are generated via alcohol oxidation or acid decarboxylation. The main aldehyde compounds detected after fermentation included ethanal (acetaldehyde), phenyl methanal (benzaldehyde), 2-methyl butanal, 3-methyl butanal (isovaleraldehyde), pentanal (valeraldehyde), Overall, after LAB fermentation, the total combined ketone concentration increased in 5 studies: (1) In okara fermented by LAB co-culture with L. acidophilus, L. rhamnosus, and P. acidilactici, the total combined ketone concentration was 2355.6 µg/g, compared to 116.1 µg/g in the unfermented okara; however, in okara fermented by monocultures of either L. acidophilus, L. rhamnosus, or P. acidilactici, the total combined ketone concentration was 98.8, 64.3, and 57.8 µg/g, respectively. During a monoculture fermentation, unstable aldehydes and ketones may be reduced to primary and secondary alcohols, whereas in a co-culture fermentation, synergic interactions between strains may instead result in the production of higher levels of ketones, which could be linked to the oxidation of alcohols [46]; (2) in cherry juice fermented by either L. rhamnosus or L. paracasei, the total combined ketone concentration was 285.1 and 11.3 ng/mL, respectively, compared to 7.2 ng/mL in the unfermented juice. In the cherry juice fermented by different L. plantarum strains, the total combined ketone concentration ranged from 48.3 to 292.4 ng/mL, compared to 6.9 ng/mL in the unfermented juice [29]; (3) in apple juice fermented by either L. plantarum, L. helveticus, L. casei, L. paracasei, L. acidophilus, or B. lactis, the total combined ketone concentration was 17.0, 27.6, 29.6, 56.6, 22.5, and 26.4 ng/mL, respectively, compared to 16.6 ng/mL in the unfermented juice [19]; (4) in another study, using apple juice fermented by different LAB strains, the total combined ketone concentration ranged from 10.1 to 11.7 ng/g, compared to 2.6 ng/g in the unfermented juice [21]; and (5) LAB fermentation of kiwifruit juice (Xuxiang, and Hongyang cultivars) increased the total combined ketone concentration by 2.6, 5.2, and 2.6 times in Xuxiang cultivar juice fermented by either L. acidophilus, L. helveticus, or L. plantarum, respectively, and by 6.3, 75, and 37 times in Hongyang cultivar juice fermented by either L. acidophilus, L. helveticus, or L. plantarum, respectively [49]. Moreover, in mango slurry fermented by LAB, the total combined ketone concentration was 1.2-1.8 times higher compared to the concentration in the unfermented juice, whereas in mango slurry fermented by the yeast S. cerevisiae, the total combined ketone concentration was 1.4-2.7 times lower compared to the concentration in the unfermented juice [41].
Overall, ketones, which are key contributors to dairy notes, such as acetoin (13 papers) and diacetyl (6 papers) have been reported to increase or were only detected after the fermentation of fruit and vegetable juices. However, in two studies, the concentration of acetoin (1 paper), and diacetyl (1 paper) decreased after fermentation by LAB and fungi, respectively.
In 8 studies, L. plantarum was the main LAB producing acetoin in fermented juices, followed by L. helveticus (3 studies) and L. rhamnosus (3 studies). The two main LAB that produced high diacetyl concentrations in fermented juices were L. plantarum (4 papers) and L. casei (4 papers). Overall, for the papers reviewed, L. plantarum produced more of the creamy flavours of acetoin and diacetyl compared to other LAB strains studied.
Acetaldehyde provides fermented juices their distinct flavour, and it is produced by LAB from the amino acid threonine [68] or from sugars via the PK (phosphoketolase) pathway, and by yeast from sugars via the EMP (Embden-Meyerhof-Parnas) pathway [13]. At lower concentrations, acetaldehyde improves the flavour of fermented juice; however, at higher concentrations (200 µg/g or 200 µg/mL or above) [20,21], it may negatively influence the flavour of fermented juices. Acetaldehyde was detected in 6 studies after LAB fermentation: (1) In Muzao jujube juice fermented by L. acidophilus, the acetaldehyde concentration was 19.9 µg/mL, compared to 1.5 µg/mL in the unfermented juice, with other LAB strains generating slightly higher or lower acetaldehyde concentrations compared to the unfermented juice [20]; (2) in kiwifruit juice (Xuxiang and Hongyang cultivars), Xuxiang cultivar juice fermented by either L. plantarum, L. acidophilus, or L. helveticus the acetaldehyde concentration was 1013.1, 136.1, and 124.3 ng/mL, respectively, compared to 109.5 ng/mL in the unfermented juice, whereas in the Hongyang cultivar juice fermented by either L. plantarum or L. acidophilus, the acetaldehyde concentration was 1075.6 and 159.7 ng/mL, respectively, compared to 95.7 and 25.5 ng/mL in the unfermented juice and fermented juice by L. helveticus, respectively [49]; (3) in apple juice fermented by either L. casei, L. rhamnosus, L. plantarum, or L. acidophilus, the acetaldehyde concentration was 40.4, 15.0, 27.5, and 21.9 ng/g, respectively, whereas it was not detected in the unfermented juice [21]; (4) in apple juice fermented by either L. plantarum, L. helveticus, L. casei, L. paracasei, L. acidophilus, or B. lactis, the acetaldehyde concentration was 5.4, 2.1, 4.5, 2.4, 1.9, and 3.0 ng/mL, respectively, whereas it was not detected in the unfermented juice [19]; (5) in non-pH-adjusted (2.7) sea buckthorn juice, the acetaldehyde concentration was 10.6 ng/mL, which increased after fermentation for 36 and 72 h by L. plantarum to 13.9 and 15.8 ng/mL, respectively. However, if the pH of the juice was adjusted to pH 3.5, the initial acetaldehyde concentration of 10.8 ng/mL decreased after L. plantarum fermentation for 36 and 72 h to 1.1 and 1.2 ng/mL, respectively [42]; and (6) in watermelon juice fermented by either L. plantarum or P. pentosaceus, the acetaldehyde concentration was 4.6 and 3.2 ng/mL, respectively, compared to 2.3, 2.0, 0.5, and 0.5 ng/mL in the unfermented juice, and L. brevis, L. casei, or L. rhamnosus fermented juices, respectively [48]. The concentration of acetaldehyde in 6 studies reported here was still well below the concentration that has been reported to adversely affect flavour, indicating that acetaldehyde may have a positive impact on the overall flavour profile of fermented juices if it is above the minimum concentration required for perception. Note, in Chinese wolfberry juice fermented by either L. plantarum, L. casei, L. paracasei, L. helveticus, or B. lactis, the acetaldehyde concentration was reduced to 52.8, 124.2, 123.3, 23.4, and 13.3 µg/mL, respectively, compared to its concentration in the unfermented juice (155.9 µg/mL), where in the juice fermented by L. acidophilus, the acetaldehyde concentration was 188.2 µg/mL [50]. Further, the initial acetaldehyde concentration in a distillate prepared using simple distillation from unfermented melon byproduct was 1320 ng/mL, which was reduced to 470 ng/mL in the L. rhamnosus fermented melon by-product distillate. Moreover, when using the vacuum distillation method, in the unfermented melon by-product distillate, the acetaldehyde concentration was 160 ng/mL, where it was only 20 ng/mL in the distillate from L. rhamnosus fermented melon by-product. Some LAB can convert acetaldehyde to ethanol and acetic acid, which could explain the decrease in acetaldehyde concentration in some fermentations [44].
Another important aldehyde from a flavour perspective as it imparts a pleasant aroma to fermented juices is benzaldehyde, which is generated by LAB from the amino acid phenylalanine. The conversion of phenylalanine to benzaldehyde by LAB is initiated by the aminotransferase enzyme. The resulting phenyl pyruvic acid is chemically converted to benzaldehyde in the presence of oxygen and manganese [58,69]. The benzaldehyde concentration increased after fermentation of vegetable and fruit juices in 7 studies: (1) In Chinese wolfberry juice fermented by either L. plantarum, L. paracasei, or L. acidophilus, the benzaldehyde concentration was 117.2, 68.1, and 40.7 µg/mL, respectively, where it was not detected in the unfermented juice nor in juice fermented by other LAB strains [50]; (2) in kiwifruit juice (Xuxiang and Hongyang cultivars), Xuxiang cultivar juice fermented by L. acidophilus, the benzaldehyde concentration was 490.3 ng/mL, compared to 369.7 ng/mL in the unfermented juice. In contrast, it was not detected in the unfermented Hongyang cultivar juice nor in the Xuxiang cultivar juice fermented by L. helveticus or L. plantarum and all LAB fermented Hongyang cultivar juices [49]; (3) in bog bilberry juice fermented by two L. plantarum strains, the benzaldehyde concentration was 55.5 and 62.3 ng/mL, compared to 41.8 ng/mL in the unfermented juice [34]; (4) in non-pH-adjusted (2.7) sea buckthorn juice, the benzaldehyde concentration was 2.7 ng/mL, which increased after fermentation for 36 and 72 h by L. plantarum to 5.4 and 7.9 ng/mL, respectively. However, if the pH of the juice was adjusted to pH 3.5, the initial benzaldehyde concentration of 2.3 ng/mL decreased after L. plantarum fermentation for 36 and 72 h to 1.1 and 1.7 ng/mL, respectively [42]; (5) in goji juice fermented by different combinations of bacterial strains (either L. plantarum, L. rhamnosus, L. reuteri, B. velezensis, or B. licheniformis), the benzaldehyde concentration ranged from 55.5 to 101.4 ng/g, compared to 46.3 ng/g in the unfermented juice [25]; (6) in durian pulp fermented by L. casei monoculture, the benzaldehyde concentration was 2.9 times higher, compared to its concentration in the sequential co-culture with yeast W. saturnus, and it was not detected in the unfermented pulp. This difference is because LAB can convert phenylalanine amino acid to benzaldehyde; however, yeast preferentially convert phenylalanine amino acid to phenylethyl alcohol via the Ehrlich pathway, resulting in a higher quantity of benzaldehyde in LAB fermentations [30]; and (7) in papaya juice fermented by L. plantarum, the benzaldehyde concentration was 2 times higher, compared to the concentration after the L. acidophilus fermentation or in the unfermented juice [23]. Though the benzaldehyde concentration increased after LAB fermentation, it also reduced in 4 studies: (1) in Hetain jujube juice fermented by LAB, the benzaldehyde concentration ranged from 22.1 to 29.7 µg/mL, compared to 33 µg/mL in the unfermented juice [20]; (2) in jujube pulp fermented by a mixture of L. plantarum, L. rhamnosus, and S. thermophilus, the benzaldehyde concentration was 3516 ng/g, compared to 4672 ng/g in the unfermented pulp [47]; (3) in cherry juice fermented by various L. plantarum strains, the benzaldehyde concentration ranged from 15.3 to 33.4 ng/mL, compared to 100.5 ng/mL in the unfermented juice, and in cherry juice fermented by L. rhamnosus, the benzaldehyde concentration was 76 ng/mL, compared to 90.5 ng/mL in the unfermented juice [29]; and (4) in okara fermented by LAB monocultures of L. rhamnosus or P. acidilactici and a co-culture (L. acidophilus, L. rhamnosus, and P. acidilactici), the benzaldehyde concentration was 22.4, 45, and 46.2 µg/g, respectively, compared to 114.6 µg/g in the unfermented okara, and in okara fermented by L. acidophilus, the benzaldehyde concentration was only slightly reduced to 109.9 µg/g [46].
Linalool, an important flavour terpene, was detected after fermentation of juices in 7 studies: (1) The linalool concentration of a distillate prepared using simple distillation from unfermented orange pomace was 1.6 µg/mL and after fermentation by L. rhamnosus of the orange pomace, the concentration in the resulting distillate was 2.3 µg/mL. Similarly, in distillates prepared using vacuum distillation, the linalool concentration in the unfermented orange pomace distillate was lower than that in the distillate from L. rhamnosus fermented orange pomace, at 0.1 µg/mL and 1.1 µg/mL, respectively. In the same study, in the unfermented melon by-product distillate prepared using simple distillation, the linalool concentration was 0.5 µg/mL and in the L. rhamnosus fermented melon by-product distillate, the concentration was 4.4 µg/mL. However, linalool was not detected in the distillates of fermented and unfermented melon by-products prepared using vacuum distillation [44]; (2) in elderberry juice fermented by either L. plantarum or L. rhamnosus, the linalool concentration was 328.5 and 339.7 ng/mL, respectively, compared to 79.3 and 189.3 ng/mL in the unfermented juices, respectively [22]; (3) in cherry juice fermented by either L. rhamnosus, L. paracasei, or L. plantarum, the linalool concentration was 39.6, 21.5, and 19.6-29 ng/mL, respectively, compared to 13.7-15.7 ng/mL in the unfermented juice [29]; (4) in goji juice fermented by a bacterial mixture (L. plantarum, L. rhamnosus, B. velezensis, and B. licheniformis), the linalool concentration was 22.3 ng/g, where it was not detected in the unfermented juice [25]; (5) in apple juice fermented by different LAB, the linalool concentration ranged from 3.0 to 4.5 ng/g, compared to 2.0 ng/g in the unfermented juice [21]; (6) in yam juice fermented by L. plantarum, the linalool concentration was 21 times higher compared to the concentration in the juice fermented by a combination of L. plantarum and S. thermophilus, and it was not detected in the unfermented yam juice [33]; and (7) in mango slurry fermented by LAB, linalool was detected, where it was not detected in yeast fermented or unfermented slurry [41]. Though the linalool concentration increased in these studies after fermentation, it was also reduced after fermentation in 2 studies: (1) In papaya juice fermented by either L. acidophilus or L. plantarum, the linalool concentration was 1.6 and 1.2 times lower, respectively, compared to the concentration in the unfermented juice [23]; and (2) in Momordica charantia juice fermented by L. plantarum, the linalool concentration was reduced compared to the concentration in the unfermented juice [18].
(1) In apple juice fermented by either L. plantarum or L. acidophilus, the phenol concentration was 3.2 and 2.2 ng/g, respectively, compared to 0.1 ng/g in the unfermented juice [21]; (2) in papaya juice fermented by L. acidophilus, the phenol concentration was 1.6 times higher compared to the concentration in L. plantarum fermented juice, where it was not detected in the unfermented juice [23]; and (3) in okara fermented by the fungi R. oligosporus in combination with the yeast Y. lipolytica, the phenol concentration was 2.4 times higher compared to the concentration when fermented by R. oligosporus in monoculture [38].

Furans
Furfural, 2-ethyl-furan, 2-propyl-furan, 2-pentyl-furan, 2,5-dimethyl-furan, 2,4-dimethylfuran, trans-2-(2-pentyl) furan, 2,3-Dihydrobenzofuran, and acetyl-furan were the major furans identified in 13 studies under furans/aldehydes/others/heterocyclic compounds. Furfural was detected in 8 studies, where it was mainly classified as an aldehyde. Production of furfural is linked to Maillard reactions, and higher levels of furfural may have a negative impact on the flavour of fermented substrates. However, LAB fermentation reduces the amount of furfural, most likely as a result of the consumption of precursors such as amino acids and reducing sugars [24,77,78]. The furfural concentration was reduced after LAB fermentation in 4 studies: (1) In jujube juice fermented by a mixture of L. plantarum, L. rhamnosus, and S. thermophilus, the furfural concentration was 1886 ng/g, compared to 3873 ng/g in the unfermented juice [47]; (2) in cherry juice fermented by four different L. plantarum strains, the furfural concentration was 26.1, 36.8, 43.5, and 51.6 ng/mL, compared to 101.7 ng/mL in the unfermented juice [29]; (3) in apple juice fermented by either L. rhamnosus, L. casei, or L. acidophilus, the furfural concentration was 95.1, 106.9, and 91.8 ng/g, respectively, compared to 114.2 ng/g in the unfermented juice [21]; and (4) in pomegranate juice fermented by different L. plantarum strains, the furfural concentration was reduced compared to the concentration in the unfermented juice [24]. However, in jujube (Muzao and Hetain varieties) juice, the initial furfural concentration in the Muzao varietal juice was 2.6 µg/mL, which was increased after fermentation by different LAB to a range of 2.9-5.2 µg/mL. In the same study, in Hetain varietal juice, the initial furfural concentration of 4.2 µg/mL was increased after fermentation by LAB to a range of 5.1-5.8 µg/mL [20]. Further, in goji juice fermented by a mixed bacterial culture, the 2-pentyl furan concentration ranged from 233.8 to 422.5 ng/g, compared to 36.3 ng/g in the unfermented juice [25], and in watermelon juice fermented by either L. plantarum, L. brevis, P. pentosaceus, L. casei, or L. rhamnosus, the 2-pentyl furan concentration was 96.7, 115, 117, 112, and 99 ng/mL, respectively, compared to 79.4 ng/mL in the unfermented juice [48]. However, the concentration of 2-pentyl furan was reduced in another 2 studies [18,38].

Alkanes, Alkenes, and Benzene Derivatives
Alkanes, alkenes, and benzene derivatives were reported less frequently, with only eight of thirty-five studies mentioning them under alkanes/alkenes/benzene derivative/hydrocarbons/others.
The use of different fruit and vegetable substrates, micro-organisms, and fermentation conditions are all likely to have had an impact on the production of fermentation VOCs. With the exception of a few studies that used bacteria other than LAB, fungi, or yeast either as a monoculture or in combination with LAB, most studies used LAB either as a monoculture or in mixed cultures. The LAB most frequently used for producing desirable VOCs were L. plantarum, L. casei, L. acidophilus, L. rhamnosus, and L. helveticus. In studies that used two cultivars of a fruit, there were notable differences in the resulting fermented VOCs present and their concentrations [20,49]. Additionally, there were variations in the types and concentrations of VOCs between the pH-adjusted and non-adjusted juices of sea buckthorn [42] and bog bilberry [34]. Moreover, there were observable differences in the detected VOCs from distillates (simple or vacuum distillation) of LAB-fermented orange pomace and melon by-product, assessed by SPME [44]. The results demonstrate how VOCs detected after being produced during fermentation are greatly influenced by the substrate (species and cultivar) being fermented, the LAB strain being used, and the fermentation conditions.
Overall, in LAB-fermented fruit and vegetable juices, the concentrations of the main dairy flavour VOCs, namely acetoin and diacetyl, ranged between 0.04 and 528.1 µg/mL, and 0.01 and 71.9 µg/mL, respectively. It was apparent that LAB fermentation can yield high concentrations of acetoin and diacetyl from plant-based substrates. However, a wide variety of VOCs, including desirable and undesirable compounds, were detected in all of the studies reviewed. Once desirable dairy flavour components have been produced, extracting and purifying them from the other components present will be the next challenge. Current research is focusing on metabolic engineering techniques that involve overex-pressing rate-limiting enzymes that produce desirable VOCs or inactivating the enzymes that produce undesirable VOCs in order to improve or create new metabolic pathways in micro-organisms. For instance, pyruvate is a crucial intermediate in the synthesis of the dairy flavours acetoin and diacetyl. Pyruvate in excess can be converted to α-acetolactate by modifying the metabolic flux of pyruvate. If acetoin production is of particular interest, α-acetolactate decarboxylase can be designed to be overexpressed, whereas when diacetyl production is of interest, NADH-oxidase can be overexpressed and α-acetolactate decarboxylase expression can be inactivated [66] (see acetoin/diacetyl production pathway in the ketones section/ Figure 3). However, due to the complexity of plant matrices, the action of different metabolic processes, and the factors influencing the fermentation, such as temperature, pH, and aeration, desirable VOCs might be metabolised, or their presence masked by undesirable VOCs. To meet these challenges, research on the metabolic pathway analysis of various micro-organism(s) on complex matrices of plants is required.

Conclusions
In conclusion, differences in substrates, micro-organisms, and fermentation conditions influence the synthesis of microbial VOCs from vegetable and fruit substrates. In comparison to other bacteria, yeast, and fungi examined, LAB strains were most frequently used to ferment fruit and vegetable substrates. Among LAB strains, Lactiplantibacillus plantarum was the most frequently used species and it produced the highest concentration of VOCs. The most frequently used fermentation temperature and time combination was 37 • C for 48 h; however, in the papers reviewed, most of the papers used temperatures of 30 and 37 • C for time combinations ranging from 20 to 120 h. Acids, alcohols, aldehydes, esters, ketones, and terpenes/norisoprenoids were the most frequent VOCs reported after the fermentation of vegetable and fruit substrates, whereas sulphur compounds, phenols, furans, alkanes, alkenes, and benzene derivatives were reported less frequently. After LAB fermentation, the concentration of alcohols, esters, ketones, acids, and terpenes/norisoprenoids generally increased, whereas the concentration of aldehydes generally reduced. The fermentation of vegetable and fruit substrates by different LAB strains generates a wide range of desired VOCs, including the dairy flavours of acetoin and diacetyl. However, due to the complexity of plant matrices, fermenting conditions, and different LAB and their metabolic characteristics, producing dairy analogues with characteristic dairy flavours is still difficult. To achieve the dairy flavours of interest for dairy analogues, in-depth research is still required on the metabolic characteristics and pathways of LAB.
Author Contributions: Conceptualization, P.S. and P.B.; data curation, S.R.; writing-original draft, S.R.; writing-review and editing, S.R., P.S. and P.B. All authors have read and agreed to the published version of the manuscript.

Conflicts of Interest:
The authors declare no conflict of interest.
Sample Availability: Samples of the compounds are not available from the authors.