3.1. Identification of aroma-active compounds in control bagels
A combined total of 40 aroma compounds were identified in the three differently processed bagels (i.e. control; 24 h cold fermented bagels; and bagels produced from 48 h cold fermented dough mass). Among these compounds, 10 aldehydes, 9 alcohols, 7 acids, 6 ketones, 5 heterocyclic compounds and 3 esters were positively identified (Table 2). To reveal the differences between the flavors of the bagels, the volatile fractions of their crumbs were subjected to aroma extracts dilution analysis (AEDA). In the control bagels, 40 aroma compounds were detected in the FD factor range of 2 to 256 respectively (Table 2). Furthermore, the results revealed 2-acetyl-1-pyrroline (roasty), methional (baked potato-like), vanillin (vanilla-like), 2,3-butanedione (buttery) and 4-hydroxy-2,5-dimethyl-3(2H)-furanone (HDMF) as compounds with the highest FD values in the control bagel. These aroma-active compounds exhibited high FD factors (128–256) (Fig. 1). Other important aroma compounds in the control bagels were butanoic acid (sweaty), acetoin (buttery), benzaldehyde (almond-like), furfural (bread-like), 2/3-methyl butanoic acid (sweaty), acetyl pyrazine (toasty), phenyl acetaldehyde (honey-like), 2-phenylethanol (honey-like), octanoic acid (fatty, soapy), 4-vinyl-2-methoxyphenol (smoky), acetic acid (sour), 3-methyl butanol (malty), and 2-methylpyrazine (nutty) all of which exhibited FD factors ranging from 16 to 32 (Fig. 1).
3.2. Aroma-active compounds in long, cold fermented bagels
Application of long, cold fermentation (5 oC, 24 h & 48 h) produced bagels that exhibited a wider range of FD factors (4- 1024) than the control bagel (Table 2). For instance, the FD factor of 2,3-butanedione in the 24 h fermented bagels increased by almost (x4) times the value obtained in the control bagel. Other compounds exhibiting higher FD factors in the long, cold fermented bagels were; acetic acid (sweaty), 2/3-methylbutanal (malty), 2,3-butanedione (buttery), propionic acid (sweaty/pungent), butanoic acid (sweaty), acetoin (buttery), 3-methylbutanol (malty), furfural (bread-like), 2-methyl pyrazine (nutty), 2/3-methyl butanoic acid (sweaty), methional (baked potato-like)l, 2-acetyl-1-pyrroline (roasty), benzaldehyde (almond-like), (Z)-4-heptenal (biscuit-like), acetyl-pyrazine (toasty), 4-HDMF (sweet/caramel), benzyl alcohol (sweet/flowery), phenyl acetaldehyde (rose-like), 2-phenyl ethanol (honey-like), octanoic acid (fatty), 4-vinyl-2-methoxyphenol (smoky) and vanillin (vanilla-like) all of which exhibited FD factors from16 to 1024.
The influence of fermentation temperatures on the formation of volatile compounds in bread crust and crumb has been well documented [47–49]. While high fermentation temperatures (≥ 27 oC) are more suitable for generating more complete volatile profiles, most bread industries are more favorable to employing more fermentation time or using sourdough that needs time to ferment. For instance, Zehentbauer & Grosch  observed that when bread is prepared from dough subjected to an initial 2 h of fermentation at 22 oC and an additional 18 h of fermentation at 4 oC, the resulting bread exhibited similar amounts of Strecker aldehydes (i.e. 2-methylpropanal, 2-methylbutanal and 3-methylbutanal) as obtained with the artisanal process. This observation is probably due to a longer proteolysis which leads to the formation of amino acids that participates in the Strecker reactions as well as the Ehrlich pathway to produce the aldehydes. It is worthy of note that both 2,3-butanedione and HDMF which exhibited the highest FD factors in the cold fermented bagels as well as many other key aroma compounds such as; 2/3-methylbutanal, acetoin, 3-methylbutanol, furfural, 2-methyl pyrazine, isoamyl acetate, methional, 2-acetyl-1-pyrroline, benzaldehyde, (Z)-4-heptenal, acetyl pyrazine, phenyl acetaldehyde and vanillin have been identified in the crumb of wheat bread [3, 11 & 47]. Also, various acids such as acetic acid, butanoic acid, 2/3-methyl butanoic acid and octanoic acid which exhibited high FD factors ≥ 16 in the cold fermented bagels have been reported in bread [50, 51].
3.3. Quantitation and odour-activity values (OAVs) of aroma-active compounds in bagels
To have an insight into the contribution of each compound to the overall aroma of bagels, 22 aroma-active compounds with FD factors ≥ 16 were selected for further investigation. For each of the selected compound, a stable isotopologue (Table 1) was employed as an internal standard to quantify it. As expected the long cold fermented bagels produced compounds with significantly (p < 0.05) high concentrations (Table 3). The highest concentrations (1126 µg kg − 1- 12950 µg kg − 1) were determined for 2,3-butanedione, 2-phenylethanol, 3-methylbutanal and acetoin respectively (Table 3). The least concentrations (17 µg kg − 1- 43 µg kg − 1) were obtained for phenyl acetaldehyde, methional and 2-acetyl-1-pyrroline respectively. A comparative analysis of the aroma potencies between the three differently produced bagels revealed some differences. Cold fermented bagels showed more potencies for the buttery smelling 2,3-butanedione, baked potato-like methional and the toasty-like 2-acetyl-1-pyrroline as revealed by their respective high odour-activity values (Table 3). For example, 2-acetyl-1-pyrroline exceeded its threshold by factors of 2603 and 2466 in the 24 h and 48 h cold fermented bagels respectively. 2-Acetyl-1-pyrroline only exceeded its threshold by a factor of 2329 in the control bagels. Similarly, 2,3-butanedione exceeded its threshold by factors of 1815 and 1992 in the 24 h and 48 h cold fermented bagels respectively. On the other hand 2,3-butanedione only exceeded its threshold by a factor of 109 in the control bagel. Similar trend was noticed with the methional, acetyl pyrazine, HDMF, 4-vinyl-2-methoxyphenol, vanillin, 2/3-methylbutanal, 2-phenyl ethanol, butanoic acid, 3-methylbutanol and benzaldehyde. However, acetic acid, phenyl acetaldehyde had OAVs below 1.
While some of the bagel aroma compounds were already present in the wheat flour and were thus transferred into the bagel. Others such as 3-methylbutanol, 2-phenyl ethanol and 2,3-butanedione were probably formed during biochemical reactions in the yeast metabolism during the dough fermentation . On the other hand the nitrogen-containing compounds such as the roasty 2-acetyl-1-pyrroline and acetyl pyrazine were formed via the reaction of free amino acids L-ornithine or L-proline with dihydroxyacetone phosphate . In addition to the nitrogen-containing compounds, aldehydes, such as 2/3-methylbutanal (malty) phenyl acetaldehyde (rose/floral) and methional (baked potato-like) were formed by the Strecker degradation of valine, isoleucine, leucine, phenylalanine and methionine respectively . Moreover the caramel-like 4-Hydroxy-2,5-dimethyl-3(2H)-furanone (HDMF) can be formed by the Maillard reaction . 4-Hydroxy-2,5-dimethyl-3(2H)-furanone is mainly formed via Maillard reaction of pentoses with the amino acids glycine and alanine, respectively. Alternatively, 4-hydroxy-2,5-dimethyl-3(2H)-furanone can also be produced without the direct interaction of glycine . Furthermore, certain aldehydes such as (E,E)-2,4-decadienal, (E)-2-nonenal, and (E)-4,5-epoxy-(E)-2-decenal were formed by autoxidation and thermal degradation of fatty acids respectively .
3.4. Aroma profile analysis and aroma simulation model
The results of sensory evaluation of the different bagels (i.e. control, 24 h fermented and 48 h fermented) are shown in (Fig. 2A) (Table 4). The aroma profiles of the cold fermented bagels were characterized as roasty, biscuit-like, malty, smoky and buttery. The control bagel exhibited similar but less intense aroma notes as compared to the cold fermented bagels. However, the 24 h and 48 h bagels flavor profiles were similar with the exception of the biscuit-like aroma note (Table 4). The statistical analysis results (Table 4) showed that the six attributes (roasty, malty, buttery, biscuit-like, smoky and baked potato like) with different superscripts provided a clearer explanation of the aroma characteristics of the different bagels. To confirm this observation, recombination experiments were carried out by mixing solutions of the pure reference compounds in the same amounts as indicated for both 24 h and 48 h bagels respectively (Table 5). A parallel evaluation of the recombination models of the freshly baked 24 h and 48 h bagels was conducted. Results showed that the recombinant model imitated well the flavor of the freshly baked bagels (Fig. 2B and C) (Table 4). The aroma of the recombination models had good similarities for all the odor notes such as roasty, baked potato-like, smoky and biscuit-like. The roasty and biscuit-like aroma notes were perceived as equally intense in the aroma models as well as in the bagels.
3.5. Omission tests
The contributions of some key aroma compounds to the flavor of the bagels, was evaluated by omission tests. Omission test is used to assess the contribution of individual compound to the overall aroma of a given food . Eleven (11) aroma omission models (M1-M11), containing of either single or a group of compounds, were prepared. Each of the omission models was analyzed in triangular experiments with two complete recombination models (Table 6). Results showed that, the omission of the entire group of acids (M1) from the complete recombination model could be distinguished by 9 out of the 10 assessors. This shows that these acids (i.e. acetic acid, butanoic acid and 3-methyl butanoic acid) play an important role in the overall aroma of the long, cold fermented bagels. In the second group, the ketones (2,3-butanedione and acetoin) with characteristic buttery nuance were omitted. Acetoin was included in this group because of its high concentration. Result of the omission of the entire ketones from the complete recombination model showed that all 10 assessors could detect between the omission model and the complete recombination models. This shows that 2,3-butanedione and acetoin greatly influence the overall aroma of the bagel. When the aldehydes (M3) (2,3-methyl butanal, methional, benzaldehyde, (Z)-4-heptenal, phenyl acetaldehyde and vanillin) were omitted, only 8 assessors were able to detect the difference (p < 0.01). Similar trend was observed when the entire group of alcohols (M4) was omitted. In model 5, 4-vinyl-2-methoxyphenol was omitted because of its high concentration and the result showed that only 7 assessors were able to detect the difference between the omission model and the complete recombination models. In model 6, 4-hydroxy-2,5-dimethyl-3(2H)-furanone was omitted and this resulted in significant (p ≤ 0.001) reduction in the characteristic aroma of the bagels. In addition, 9 of the assessors were able to distinguish its omission from the complete recombination models. Similar observation was obtained when other single compounds such as 2-phenyl ethanol, methional, (Z)-4-heptenal, 5-methyl-2-furanmethanol and 2-acetyl-1-pyrroline were omitted from the complete recombination models respectively. However, the omission of 5-methyl-2-furanmethanol and 2-acetyl-1-pyrroline was detected by all 10 assessors.