J. Sci. Fd Agric. 1975, 26, 887-894

Some Flavouring Constituents of Fermented Fish Sauces

James Dougan and George E. Howard

Tropical Products Institute, 56-62 Gray's Inn Road, London WCl X 8L U

(Manuscript received 10 September 1974 and accepted 22 January 1975)

The aroma of fermented fish sauce comprises three distinct notes, cheesy, meaty and ammoniacal. Analysis showed that the cheesy odour was produced by lower fatty acids and the ammoniacal odour by ammonia and amines. The meatyaroma was much more complicated and was not analysed, but it was shown that it could be produced by atmospheric oxidation of precursors that were still present in mature sauces. Individual fatty acids were determined in sauces at various stages of fermentation and an hypothesis explaining the origin of the acids is deduced from the results.

1. Introduction

Fish sauces are traditional products that are used extensively in south-east Asia and some other parts of the Far East. They are prepared by mixing small, uneviscerated fish with salt in the ratio of 1 part of salt to 1.5 parts to 3 parts by weight of fish. The mixture is placed in a concrete vat where it is allowed to ferment for about 9 months at temperatures between 30 and 35 "C during which time the fish disintegrates to give a brown liquid. The liquid is run off and allowed to mature for about 3 months and the residue is leached two or three times with brine to give sauces of a much lower quality which are usually blended with various amounts of primary extract. The final residue, which consists mainly of bones, is used as a fertiliser.

Fish sauces are clear brown liquids similar in appearance to soy sauces but perhaps a little paler with a characteristic aroma compounded of ammoniacal, meaty and cheesy notes. They contain about 20 g of nitrogen/litre, of which about 16 g are in the form of amino acids. According to Amano,l the daily consumption of 40 ml of sauce would represent about 7.5 % of an individual's intake of protein. This assumes a protein intake of about 54 g per day which seems reasonable for that part of the world.2 Olley3 concludes that fish sauce could be consumed in sufficient quantity to provide all of the necessary protein in spite of the high salt content. The lysine content of fish muscle is about twice that of pork, so that 40 ml of fish sauce could supplement the lysine deficiency of several grams of rice protein provided that the availability of the lysine has not been reduced significantly.

In recent years, at least two of the principal producing countries (Thailand and the Philippines) have shown considerable interest in the development of rapid methods of manufacture that could also make use of types of fish that were previously considered to be unsuitable for the traditional process. The rapid processes mostly consist of the

887

888 J. Dougan and G. E. Howard

addition of proteolytic enzymes, such as papain, but increased temperatures have been used. The products of these processes generally have normal nitrogen contents but their flavour is often unsatisfactory. In order to overcome this difficulty, it was necessary to know more about the flavouring constituents and their formation.

A literature search revealed several publications on various aspects of the flavour of fish sauce. Van Veen49 5 stated that the aroma of fish sauce was due to the presence of methyl ketones and that the volatile fatty acids were relatively unimportant, while Truong-van-ChomG considered that the aroma was due to volatile fatty acids and other workers were noncommittal.', * Truong-van-Chom identified formic acid, acetic acid, propionic acid and n-butyric acid while Saisithi et al.8 obtained similar results except that they identified iso-butyric acid instead of n-butyric acid. Saisithi et al. tested for carbonyl compounds but found none.

Since the literature contained contradictory statements about the relative importance of carbonyl compounds and fatty acids and since there was also disagreement about the identity and the concentration of the acids present, i t was clear that the aroma of fish sauces needed to be re-examined, especially as none of the authors had used the most modern techniques of flavour chemistry based on gas chromatography.

2. Experimental

2.1. Preliminary investigation

A portion of Hong Kong fish sauce purchased in London was distilled under reduced pressure (2 Torr) to avoid any changes that might occur at temperatures above 35 "C. Portions of the distillate were tested with alkali, acid, hydroxylamine hydrochloride, hydroxylamine with alkali, phenyl mercury acetate, chloramine-T and mercuric acetate to determine the effect of these reagents upon the odour. The results indicated that carbonyls, esters and sulphur compounds played no important part in the aroma of the distillate while bases and acids were evidently important.

A more detailed examination of the commercial Hong Kong product in which basic, neutral and acidic fractions were distilled from the sauce confirmed that the neutral fraction was odourless, although chemical tests indicated that some carbonyl com- pounds were present. Samples of fish sauce from China, Hong Kong, the Philippines and Thailand were analysed for neutral monocarbonyl compounds which were isolated in the form of 2,4-dinitrophenyl hydrazones and estimated photometrically. The values obtained ranged from 40-130 pmol/litre. Analysis of the derivatives by t.1.c. revealed only the lower, less odorous members of the class that were tentatively identified as acetone and acetaldehyde with traces of butan-2-one and valeraldehyde. Apart from the basic fraction with its normal ammoniacal odours, the only odorous compounds that could be detected by chemical tests were fatty acids and at that stage it seemed that they alone were responsible for the odour of fish sauce.

This conclusion was shown to be incorrect, however, when it was noticed that the odourless residue that remained when fish sauce was distilled to dryness developed a strong meaty odour on the addition of water. Clearly the meaty odour was an inde- pendent note in the aroma of fish sauce. The presence of the strong meaty odour in

Flavours in fermented fish sauces 889

the residue seemed to showthat the substances responsible for it were not steam-volatile, yet paradoxically, the distillate was not noticeably lacking in meaty odour.

Attempts were made to isolate the meaty odour by solvent extraction and it was found that iso-propanol was not miscible with fish sauce and could extract the meaty odour very efficiently.

A litre of iso-propanol extract was prepared by repeated extraction of one litre of fish sauce, but, as one would expect, it contained a considerable amount of water which was not readily removed by anhydrous magnesium sulphate. The yellow extract was concentrated to about 30ml by distillation under reduced pressure to give a viscous, brown liquid with an extremely strong meaty odour. Light petroleum (b.p. 60-80 "C) was added to a portion of the extract and two layers were formed. The upper layer, which consisted mainly of light petroleum, was colourless and contained the sub- stances responsible for the meaty odour, while the lower layer (which was found to be miscible with water) was odourless. On exposure to air for a minute or two on the paper smelling strip, however, these lower layers developed a meaty odour and it was suspected that it was being developed by atmospheric oxidation. Oxygen was passed through a portion of the odourless liquid which developed a strong meaty odour within a few minutes.

This explains why the residue from the vacuum distillation of fish sauce developed an odour on the addition of water which contained dissolved oxygen, while the distil- late had an odour of fish sauce not noticeably lacking the meaty note. The substances responsible for the meaty odour had been distilled with the other volatile flavouring constituents and a further quantity had been formed on the addition of water which contained dissolved air. The failure to detect anything other than acids, lower carbonyl compounds and bases in the distillate was due to the fact that the meaty note could be detected and recognised organoleptically at extreme dilution.

Oxygen was passed through a portion of fish sauce that had been extracted with iso-propanol but only a faint odour was produced. Hence the meaty odour was produced by oxidation of a substance which can be extracted entirely from fish sauce with iso-propanol.

The light petroleum solution of the meaty aroma was examined by gas chromato- graphy on a 6 ft column packed with 10% of SE 30 silicone on "Diatoport S" (SO/lOO mesh) programmed after an initial period of 18 min at 50-200 "C at 4 "C/min and produced a chromatogram containing 50 peaks. The chromatograph was fitted with a stream-splitter between the end of the column and the detector so that the effluent could be smelt. Meaty, burnt, nauseating and other odours were found to coincide with different peaks in the chromatogram.

2.2. Identification of the fatty acids A portion of Hong Kong fish sauce was acidified with phosphoric acid and was distilled under reduced pressure. The distillate was made alkaline with sodium hydroxide and was evaporated to small volume on a water bath. This residue was acidified with phosphoric acid and was analysed by gas chromatography under the conditions described in detail below. Acetic acid, propionic acid, iso-butyric acid, n-butyric acid and iso-valeric acid were identified by comparing their retention times

890 J. Dougan and G. E. Howard

with those of authentic specimens. These identifications were confirmed by gas chromatography on columns containing poly-propylene sebacate and “Free Fatty Acid Phase”. Similar experiments were also carried out on commercial sauces from Thailand and the Philippines and the same results were obtained. 1x0-valeric acid has not previously been identified in fish sauces.

As the gas chromatograph used in these experiments had a flame-ionisation detector which is insensitive to formic acid an instrument with a thermal conductivity detector was used to determine formic acid in two samples of Thai fish sauce using a 6 f t column containing “Chromosorb 101”. Formic acid was identified, but both sauces contained only 50 mg/100 ml. Although it has an acutely irritant effect at high concen- trations, formic acid has no odour and it could make no contribution to the aroma of fish sauce at this concentration. Consequently, no attempt was made to identify it or determine it in any other samples.

2.3. Determination of volatile fatty acids Fish sauce (10ml) was acidified to pH 2.5 with phosphoric acid and was steam distilled in a Markham still. It was found that it was necessary to collect 150 ml of distillate to ensure that all of the volatile acids had been distilled. The distillate was titrated with 0.1 N sodium hydroxide solution using thymolphthalein as indicator and 1.5 ml of the alkali was added in excess. n-Valeric acid (3 ml of a solution contain- ing 1.194 mg/ml), was then added for use as an internal standard.

The distillate was then evaporated to small volume on a hot-plate and was trans- ferred to an oven at 40 “C where it was allowed to evaporate to dryness. As much as possible of the dry residue was transferred to a I ml graduated flask and the remainder was washed into it with 0.1 N phosphoric acid and the solution was made up to 1 ml with the dilute acid.

Portions (0.4 1.1) of this solution were injected into a gas chromatograph with a 9 ft x 1/8 in glass column packed with “Chromosorb 101” (80-100 mesh) at 200 “C and a carrier gas flow rate of 30 ml/min. Peak areas were determined by triangulation.

Calibration factors were determined using synthetic solutions prepared by mixing various volumes of standard aqueous solutions of the fatty acids. A series of standards was prepared in which the concentration of each acid varied rather more widely than in fish sauces, and most of the standards contained all the acids. This procedure was adopted in preference to using mixtures of individual acids with the internal standard because it was thought that the calibration for propionic acid might be affected by the proximity of the very much larger acetic acid peak and similarly iso-butyric acid and iso-valeric acid might be affected by the relatively large n-butyric acid peak.

A series of experiments to determine the recovery of individual acids added to a fish sauce of known composition showed that the analytical procedure recovered 99% of the acetic acid and 8 5 % of the other acids. These values were used in the calculation of the acid concentrations to correct for the losses which were shown to be caused by incomplete recovery by the steam distillation.

Samples at different stages of fermentation were collected by one of the authors during visits to four factories in Thailand. The samples were analysed by the methods described above and the results are given in Table 1. The Thai names of the fish and

Tab

le 1

. Vol

atile

fatty

aci

ds in

Tha

i fis

h sa

uces

Sam

ple

Aci

d

Ace

tic

Prop

ioni

c is

o-B

utyr

ic

n-B

utyr

ic

iso-

Val

eric

T

otal

N

o.

Tim

e Sp

ecie

s Fa

ctor

y (m

g/l0

0 ml)

(mg/

100 m

l) (m

g/10

0 ml)

(mg/

100 m

l) (m

g/10

0 ml)

(mE

q/lit

re)

17

15

16 9 12 1 10

18

13

19 3 8 22

28 6 7 4

0 St

olep

horu

s 1

wee

k St

olep

horu

s 2

mon

ths

Stol

epho

rus

3 m

onth

s St

olep

horu

s 3

mon

ths

Sard

inel

la

6 m

onth

s St

olep

horu

s 6

mon

ths

Stol

epho

rus

6 m

onth

s St

olep

horu

s 7

mon

ths

Sard

inel

la

12 m

onth

s St

olep

horu

s 15

mon

ths

Stol

epho

rus

48 m

onth

s St

olep

horu

s -

Bes

t ble

nd

-

Bes

t ble

nd

-

Firs

t lea

chin

g -

Seco

nd a

nd th

ird

leac

hing

6

mon

ths

Scum

from

sur

face

D

D

D

C

C

A

C

D

C

D

A

B

C

E A

A

A

30a

1 oa

50a

420

490

250

600

250

1400

21

0 47

0 53

0 42

0 30

0 43

0 29

430

-

17

18

28

67 5

260 11

33

47

27

19

54

3 56

- 1 -

1 2 -

-

-

-

36

26

42 3 1 37 6 16

31

43

11

120 12

121

-

-

- 2 9 8

130 T 7 30 3 6 4 4 11

9 -

6 2 8 76

88

51

1 20 42

28

0 40

85

98

78

54

93

95 6.6

The

aci

ds w

ere

isol

ated

by

the

Con

way

diff

usio

n m

etho

d an

d th

e re

sults

are

pro

babl

y lo

w b

y a

fact

or o

f ab

out

2.

T=

a tr

ace.

892 J. Dougan and G. E. Howard

the fermentation times were given by the owners of the factories. As it is customary to keep different kinds of fish separate, the identification of the genera is quite reliable. The shorter fermentation times were probably accurate to a day or two, the medium ones to the nearest month and the longer ones perhaps rather less accurate.

3. Results and discussion 3.1. Fatty acid contents The total fatty acid concents of the good quality commercial fish sauces purchased in Thailand were similar to those found in Vietnamese products by Nguyen-An-Cu and Vialard-Goudou' but the highest ratio of n-butyric acid to acetic acid was 1 : 3.3 compared with their maximum of 1 : 1. The ratio of the mean values was 1 : 19 which is close to their value of 1 : 20 for the best quality. It is interesting to note that Truong-van-Chom6 also found approximately equal amounts of acetic acid and n-butyric acid in Vietnamese fish sauces. This suggests that there might be a charac- teristic difference between Vietnamese sauces and those of other countries.

Saisithi et al. obtained somewhat lower values for the total volatile acids of Thai fish sauce than we did but they used the Conway diffusion method which we found to give low results.

The Chinese and Hong Kong fish sauces that were purchased in London were more ammoniacal and less cheesy than the Thai sauces and had very low volatile acid contents (7-37 mEq/litre, mean 18 mEq/litre). The fatty acid content of a commercial sample obtained from the Philippines fell within the range of values obtained from the Thai products.

3.2. Development of fatty acids None of the previous work on fish sauce had produced conclusive evidence about the mechanism by which the aroma was produced but the general opinion was that it was the result of bacterial activity. The present work shows that the aroma consists of three notes, at least two of which are due to the presence of distinct classes of compounds, namely volatile bases and volatile fatty acids. Consequently it is possible that each note is produced by a different agent such as bacterial enzymes, fish enzymes or atmospheric oxidation.

Saisithi et al. determined the total volatile acid content of Thai sauces taken at the first, third, sixth, ninth and twelfth months of fermentation. His results appear to suggest that a slight increase occurred over this time, but his results are too scattered to give a significant regression coefficient.

It appears from the results given in Table 1 that most of the volatile fatty acids were formed during the third month, after which the results were rather scattered and there was no discernible trend. From this one can conclude that if there was any increase in the fatty acid content after the third month it was less than the random variation between different vats and different factories and was of no practical import- ance.

The volatile fatty acids were probably derived either from the fat or from the amino acids of the fish protein. If the fats were the source, they must have been produced

Flavours in fermented fish sauces 893

by oxidation which could have been brought about either by microbiological activity, or by fish enzymes or by chemical reaction between the fat and atmospheric oxygen. Globules of fat released by the disintegration of the fish would tend to rise to the surface where they would be exposed to air. This hypothesis would explain the lag in the production of acids because the fish do not disintegrate much during the first few weeks.

The determination of individual fatty acids showed that with one exception the sauces contained appreciable amounts of straight chain acids and very little of the branched chain acids. That is precisely what one would expect if the acids were derived from the fat because fish fats themselves contain hardly any branched fatty acids. A sample of scum taken from the surface of the liquor in one vat contained small amounts of C14 and CIS saturated acids, some monounsaturated C I ~ acid, and traces of still higher acids in addition to the lower acids normally found in fish sauce. Another sample contained traces of hexanoic and heptanoic acids. The presence of these higher acids is also consistent with their having been derived from the fat.

The extremely low aldehyde content of fish sauces might appear to cast doubt upon the atmospheric oxidation hypothesis since the oxidation of fats usually begins with the formation of hydroperoxides which are subsequently converted into alde- hydes some of which are oxidised further to give acids. Free aldehydes react with free amino acids in the well-known non-enzymic browning reaction and would be expected to disappear as a result of this reaction and further oxidation to acids. In any case, they would only need to be present for a short time and we did not test for them in the early samples. n-Butyric acid is known to occur in oxidised fats.

Although the formation of fatty acids can be explained in terms of the oxidation of lipids, one must also consider the possibility that they might have been formed from amino acids by certain well-known enzymic reactions that could conceivably operate either during the autolysis of fish tissues or as a result of the microbiological fermentation of amino acids.

Oxidative deamination of protein amino acids9 would give branched chain acids, not the straight chain acids that were found to be predominant in fish sauces. iso- Leucine can be degraded into one molecule of acetic acid and one of propionic acid by a reaction involving coenzyme A10 but that would not explain the presence of n-butyric acid. Reductive deamination by the Strickland reaction11 could also explain the formation of acetic and propionic acids but not of either n-butyric or iso-butyric acid.

In order to explain the formation of n-butyric acid directly from the amino acids that occur in proteins, it is necessary to postulate either the reduction of the hydroxyl and amino groups of threonine, or the replacement of the methylthio group and amino group of methionine by hydrogen.

Saisithi et al. showed that some of the bacteria isolated from fish sauce could produce fatty acids from amino acids but they did not identify the acids produced. They also showed that the total viable count decreased steadily during the fermenta- tion, falling to a few hundredlml after one year.

The hypothesis that the formation of fatty acids is independent of the liberation of amino acids and of microbiological activity is supported by their sudden appearance

894 J. Dougan and G. E. Howard

at an early stage of the fermentation and by the fact that the first leaching of the residue contained amounts of fatty acids similar to those in the primary extract although the nitrogen content was very much lower and microbiological activity had declined to a low level. It seems likely that the sludge at the bottom of the vat contained some fat which was released by the leaching and subsequently oxidised by air.

4. Conclusion

The aroma of fish sauce consists of three notes, ammoniacal, meaty and cheesy. Our examination of the nitrogen contents of fish sauces produced results similar to those published by Rose12 and are not, therefore, given here.

The meaty note is very complicated and is similar to that of meat extract or yeast hydrolysate. The cheesy note is caused by volatile fatty acids the most important of which is n-butyric acid. The evidence available at present indicates that the fatty acids are more likely to have been formed by the oxidation of fat than by any other mechanism that has been considered.

References 1.

2.

3. 4. 5.

6. 7. 8. 9.

10. 11.

12.

Amano, K. Fish in Nutrition 1962, p. 180 (Heen, H., Kreuzer, R., Eds) London, Fishing News (Books) Ltd. Rao, K. K. P. N. Fish in Nutrition 1962, p. 237 (Heen, H., Kreuzer, R., Eds) London, Fishing News (Books) Ltd. Olley, J. CSIRO Food Research Quarterly 1972, 32, 27. Van Veen, A. G. Natuurwetenschappelijk Tijdschrifi uoor Nederlandsch Indie 1941, 101, 147. Van Veen, A. G. Fish as Food 1969, vol. 3, p. 229 (Borgstrom, G., Ed.) Academic Press, New York and London. Truong-van-Chom 9th Pacific Science Congress 1957, 5, 135. Nguyen-an-Cu; Vialard-Goudou, A. C . r. hebd. Sianc. Acad. Sci., Paris 1953,23SA, 2128. Praseart Saisithi; Bung-orn Kasemsarn; Liston, J., Dollar, A. M. J . Fd Sci. 1966, 31, 105. Mahler, H. R., Cordes, E. H. Biological Chemistry 1966, p. 686, Harper International Edition. Mahler, H. R.; Cordes, E. G. Bio/ogical Chemistry 1966, p. 691, Harper International Edition. Truton, J. S . ; Simmonds, S . General Biochemistry 1958,2nd ed., p. 757, New York, John Wiley and Sons Inc. Rose, E. Bull. economique Indochine 1918, NSZO, 155.