3.2.1. Odor
The volatile compounds in each fish sauce were analyzed using GC-MS, and the obtained peaks with areas larger than 100,000 were compared. For
, the numbers of
S. melanostictus,
O. lacepedii,
T. japonicus,
K. punctatus, and their collective mixture were 63, 55, 44, 57, and 63, respectively. For
, the numbers of peaks were 42, 37, 43, 35, and 37 for
S. melanostictus,
O. lacepedii,
T. japonicus,
K. punctatus, and their collective mixture, respectively. The numbers of peaks commonly observed for each fish sauce are shown in
Figure 3. For example, 16 peaks observed in
prepared from
S. melanostictus were also detected in the three types of
. For
prepared from S, O, T, and M, the number of volatile compounds contained in the sample was greater than 17. The peaks common to all
were not specific to
, and were also common to all or some of the
. The ratio of the number of peaks specific to one fish species or common to two fish species was 47.5% in
, which was higher than that in
(35.0%). The ratio of the peaks specific to
was calculated as 66.3%. Therefore, the results suggest that
contains fish-specific odor components, while
enables the production of fish sauce with a reduced fish-specific odor.
The peak similarity score, which indicates how many fish species the volatile compounds present in each
and
were common to on average, was calculated. Fish sauces with higher scores indicated higher similarity in volatile compounds among fish sauces. The score for
was 3.27 and that for
was 2.77. Thus, the fish sauce preparation methods with higher scores indicated a higher similarity of volatile compounds among the fish sauces, indicating that the volatile compound profiles of
were more similar than those of
. This result was consistent with the observation that the unique odor of the fish species was suppressed, and the common odor of fish sauce was a soup stock-like odor in
(
Table 3). A 16% increase in the percentage of peaks commonly present in
was observed in the unused fish mixture, indicating that the types of volatile compounds tended to be similar. The odor characteristic of
showed a decreased fish-specific odor and an increased soup stock-like odor. This result is consistent with those observed in the sensory analysis in this study.
Table 5 shows the odor attributes with significant differences (
p < 0.05) and significant trends (
p < 0.11) in
by sensory evaluation. The quantification of the odor impressions is shown in
Table 3. In
, putrefactive and rancid odors were reduced, preferred odors were enhanced regardless of the fish species, and the soup stock-like scent tended to be enhanced. In addition, when comparing the
S. melanosticus and the unused fish mixtures, similar differences were observed. These trends were in agreement with those of some 2-month fish sauces [
6], and the longer fermentation period of 3 months in the present study tended to improve the odor of
. This suggests that a fermentation period of 3 months or longer is preferable for the production of
in terms of odor. Furthermore, all panelists distinguished between
and
prepared from
S. melanostictus and an unused fish mixture. However, more than half of the panelists could not distinguish between
prepared from a mixture of
S. melanostictus and the unused fish mixture. The results of both GC-MS and sensory analyses demonstrated that the
exhibited a comparable odor profile, suggesting that pCO₂ can be employed to generate fish sauces with a similar odor profile, even when utilizing unspecified fish as the raw material.
3.2.2. Taste
The taste analyzers recognized three types of first tastes (salty, umami, bitterness and miscellaneous), and two types of aftertastes (bitter and umami).
Figure 4 shows a radar plot comparing the tastes detected in
and
. The radar shape varied with
, and the bitterness of miscellaneous tastes commonly decreased in
.This tendency was more pronounced in 3-month fish sauce compared than in 2-month fish sauce [
6].
Figure 4 shows the PCA of the tastes detected in
and
. The distribution of
is scattered, whereas that of
is similar. These results suggest that the taste of
reflects the characteristics of each raw fish, whereas in
, regardless of the raw fish, the tastes converged to the same family, characterized by enhanced umami richness and reduced bitterness.
Table 6 shows the amount of free amino acids in each fish sauce sample. The total amounts of free amino acids in
prepared from
S. melanostictus,
O. lacepedii, T. japonicus,
K. punctatus, and their collective mixture, respectively, were 1.9, 2.2, 1.3, 1.7, and 1.4 times and significantly (
p < 0.05) higher than those in
. Furthermore, a significant (
p < 0.05) increase in total free amino acid content was observed in the
prepared from the unused fish mixture. Free amino acids are produced by endogenous proteases and the contaminating bacteria [
1]. Endogenous acid proteases can effectively hydrolyze fish proteins in a pCO
2 environment because the number of contaminating bacteria may be reduced under such conditions.
Therefore, the enrichment of total free amino acid content is a universal phenomenon in
regardless of fish type.
Figure 5 shows the principal component analysis of the proportions of free amino acids in each fish sauce. The free amino acid composition appeared to be divided into two groups, depending on PC1 between
and
. Glutamic acid, a typical umami amino acid, does not always constitute this group. The PCA results indicate that the free amino acid composition did not provide sufficient evidence for enhanced umami taste and umami richness. Imai et al. [
28] reported that the bitter taste was not caused mainly by free amino acids but by peptides such as carnosine and glutathione. Kuroda et al. [
29] reported that a tripeptide, glutamylvalylglycine, detected in fermented foods such as fish sauce, has umami richness.
Jung et al. [
30] and Ohshima et al. [
1] reported that microbial metabolism during fermentation plays an important role in flavor formation in fermented fish-based foods. Wang et al. [
31] reported that the peptides in fish sauce are formed by the hydrolysis of fish proteins by microbial proteases. Therefore, the bacterial flora of each fish sauce was analyzed using 16S rRNA amplicon analysis and presented as relative ratios of the bacterial flora (
Figure 6). In all raw fish, the bacterial flora changed with the different methods of fish sauce preparation. However, the dominant species in both the
and
collective mixture were staphylococci. The diversity of microflora appeared to increase in CO
2 fish sauces. The bacterial flora in fish sauce could not be clearly characterized by
and
. It is difficult to infer which bacteria contribute to the common flavor observed in
. The bacterial flora in fish sauce changes during fermentation [
8]. The fermented foods contained dead cells, which were also detected using 16S rRNA amplicon analysis. In addition, 16S rRNA amplicon analysis could not quantify the absolute number of viable cells. Therefore, it is necessary to examine viable cells quantitatively and over time to identify the bacterial species that have a significant impact on flavor similarity.
The organic acid composition was similar for all types of
, and a high lactic acid content was commonly observed in
(
Table 7). The types of organic acids used varied depending on the raw material used in
. The average lactic acid content of
tended to be higher (
p = 0.07) than that of
. Lactic acid has a peculiar soft acidity and a low-acid smell. Acidity was not detected in
(
Table 3 and
Figure 4), suggesting that the organic acids had no significant effect on the flavor of
. Lactic acid in fish sauce is produced by lactic acid bacteria or rigorous mortis [
32,
33]. However, positive evidence that lactic acid bacteria accumulated lactic acid at high concentrations was not obtained from the microflora analysis (
Figure 6). In addition, production via the rigor mortis process provides a clear explanation for the difference in lactic acid content between
and
prepared from
O. lacepedii and
T. japonicus.