International Journal of Food Microbiology 46 (1999) 219–229

Characterization of lactic acid bacteria isolated from a Thai low-salt

fermented fish product and the role of garlic as substrate for
fermentation

¨ *, Hans Henrik Huss, Lone Gram

Christine Paludan-Muller
Danish Institute for Fisheries Research, Department of Seafood Research, Technical University of Denmark, Building 221,
DK-2800 Lyngby, Denmark

Received 22 July 1998; received in revised form 18 September 1998; accepted 24 November 1998

Abstract

Lactic acid bacteria (LAB) isolated from raw materials (fish, rice, garlic and banana leaves) and processed som-fak (a Thai
low-salt fermented fish product) were characterized by API 50-CH and other phenotypic criteria. Lactococcus lactis subsp.
lactis and Leuconostoc citreum were specifically associated with fish fillet and minced fish, Lactobacillus paracasei subsp.
paracasei with boiled rice and Weisella confusa with garlic mix and banana leaves. In addition, Lactobacillus plantarum,
Lactobacillus pentosus and Pediococcus pentosaceus were isolated from raw materials. A succession of aciduric,
homofermentative lactobacillus species, dominated by Lb. plantarum /pentosus, was found during fermentation. In total, 9%
of the strains fermented starch and 19% fermented garlic, the two main carbohydrate components in som-fak. The ability to
ferment garlic was paralleled by a capacity to ferment inulin. An increased percentage of garlic fermenting strains was found
during fermentation of som-fak, from 8% at day 1 to 40% at day 5. No starch fermenting strains were isolated during
fermentation. Three mixed LAB cultures, composed of either starch fermenting Lc. lactis subsp. lactis and Lb. paracasei
subsp. paracasei, or garlic fermenting Lb. plantarum and Pd. pentosaceus, or a combination of these strains were inoculated
into laboratory prepared som-fak with or without garlic. In som-fak without garlic, pH was above 4.8 after three days,
irrespective of addition of mixed LAB cultures. The starch fermenting LAB were unable to ferment som-fak and sensory
spoilage occurred after three days. Fermentation with the combined mix of starch and garlic fermenting strains led to
production of 2.5% acid and a decrease in pH to 4.5 in two days. The fermentation was slightly slower with the garlic
fermenting strains alone. This is the first report describing the role of garlic as carbohydrate source for LAB in fermented
fish products.  1999 Elsevier Science B.V. All rights reserved.

Keywords: Low-salt fermented fish products; Som-fak; Lactic acid bacteria; Garlic; Inulin; Starch

1. Introduction

*Corresponding author. Tel.: 1 45-4588-3322, Fax: 1 45- Traditional lightly salted fermented fish products
4588-4774, E-mail: cpm@dfu.min.dk are widespread in South-East Asia (Ishige, 1993;

0168-1605 / 99 / $ – see front matter  1999 Elsevier Science B.V. All rights reserved.
PII: S0168-1605( 98 )00204-9
220 ¨
C. Paludan-Muller et al. / International Journal of Food Microbiology 46 (1999) 219 – 229
Adams et al., 1985). They are typically composed of
freshwater fish species, salt (2–7%), a carbohydrate
source (boiled / roasted rice, millet, sugar or fruit)
and spices (garlic, ginger, chilli, pepper). Som-fak is
a Thai product composed of minced fish fillet, salt
(2–5%), ground boiled rice (2–12%) and minced
garlic (4%). The mixture is tightly packed in banana
leaves or plastic bags and left to ferment for two to
five days at 308C (Saisithi et al., 1986). Som-fak can
be served raw or cooked either as a main course with
vegetables or as a snack. A meat analogue is found
in nham, a Thai fermented pork sausage (Steinkraus,
1996).
In som-fak, rapid growth of lactic acid bacteria
(LAB) causing pH to decrease below 4.5 in two days
is essential to prevent spoilage (Østergaard et al.,
1998a) and to ensure safety of the product. Pediococ-
cus sp. and Lactobacillus sp. have been identified as
the dominating LAB genera in commercial samples
of som-fak and those prepared in the laboratory
(Kimhamanon, 1994; Tanasupawat et al., 1993;
Saisithi et al., 1986). However, a detailed study of
the succession of LAB species in som-fak and their
fermenting properties has to our knowledge not been
published.
The lightly salted fermented fish products always
contain a carbohydrate source and in som-fak, rice-
starch is assumed to be the substrate for fermentation
by LAB. However, increasing the percentage of rice
in som-fak from 5 to 20% caused only a slight extra
decrease in pH (Saisithi et al., 1986). It has been
suggested, that the main role of rice is to reduce the
high buffering capacity of the fish in order to obtain
a rapid decrease in pH (Owens and Mendoza, 1985).
Since fish contains very little carbohydrate ( , 0.5%)
other ingredients may act as substrates for fermen-
tation. Garlic is added to som-fak and many other
fermented fish products as a flavouring agent, often
in high levels (2–6%). Garlic is believed to affect
the microbial flora in two ways; (1) by acting as an
antimicrobial agent, in particular against Gram-nega-
tive bacteria, due to allicin (Feldberg et al., 1988;
Beuchat, 1994); (2) by stimulating the growth of
LAB (Nes and Skjelkvale, ˚ 1982; Zaika and Kiss-
inger, 1984). In the Korean fermented fish product
gajami sikhae an increase of the garlic content from
2.5 to 5% resulted in a lower final pH of the product
(Souane et al., 1990). Garlic contains approximately
30% of fructo-oligosaccharides, mainly in the form
of inulin (Van Loo et al., 1995), which is a polymer
of D-fructose linked by b(2-1) bonds with an a-(1-2)
linked D-glucose at the terminal end of the molecule.
Microbial formation of fructose from inulin involves
a single enzymatic step requiring inulinase, whereas
the degradation of starch needs two enzymatic steps
to glucose and at least three steps to fructose
(Vandamme and Derycke, 1983). Our hypothesis is
therefore, that garlic may be directly involved in the
fermentation process of som-fak by supplying the
LAB with fermentable carbohydrates.
The aim of this study was to characterize the LAB
on raw materials and during fermentation of som-fak
with the main purpose of identifying the species
essential for fermentation. The importance of garlic
and rice as substrates of fermentation was also
investigated.
2. Materials and methods
2.1. Origin of bacterial isolates
LAB were isolated from samples of raw materials
(fish fillet, minced fish, boiled rice, garlic mix and
banana leaves) and from som-fak at days 0, 1, 2, 4
and 5. The samples were obtained on the day of
production from a factory in Bangkok, Thailand. The
processing of som-fak, sampling and microbiological
analyses have been described by Østergaard et al.
(1998a). LAB were isolated and purified on MRS
agar (CM361, Oxoid, Hampshire, England) and
stored frozen at 2 808C in a freezing media with
glycerol (Gibson and Khoury, 1986). For identifica-
tion and fermentation experiments, isolates were
cultured at 258C in MRS-broth (1,10661, Merck,
Darmstadt, Germany) for 24 h or on MRS agar for
two to three days in anaerobic jars containing 90%
N 2 and 10% CO 2 .
2.2. Phenotypic characterization of lactic acid
bacteria
All isolates were initially tested for colony mor-
phology, Gram reaction by the KOH method (Gre-
gersen, 1978), catalase reaction by the 3% H 2 O 2
method, cytochrome oxidase by DrySlide Oxidase
strips (Difco Labs.), cell morphology using phase
contrast microscopy, oxidative / fermentative utiliza-
 ¨
C. Paludan-Muller et al. / International Journal of Food Microbiology 46 (1999) 219 – 229
tion of glucose (Hugh and Leifson, 1953), NH 3
production from arginine with glucose added in final
concentrations of 0.1% and 2% (w / v) (Leisner,
1992), gas production from glucose, and final pH
after growth in modified MRS-broth (Shaw and
Harding, 1984).
LAB isolates were further characterized by their
carbohydrate fermentation pattern using the API-50
´
CH system (bioMerieux, France). The type strain of
Lb. plantarum (DSM 20174) (Deutche Sammlung
von Mikroorganismen und Zellkulturen,
Braunschweig, Germany) was included for every
identification of 25 isolates and 10 strains were
identified twice as a control of test reproducibility.
2.2.1. Capacity of LAB strains to ferment starch,
garlic and inulin in model substrates
All isolates were tested for their capacity to
ferment the carbohydrate components in som-fak.
The fermentation substrate used was modified MRS-
broth without glucose, with chlorophenol red as
indicator and pH adjusted to 7. The carbohydrate
sources were added individually in final concen-
trations of 10% soluble potato starch (Sigma, St.
Louis, MO, USA) or 1.3% inulin (from Chicory root,
Sigma) before sterilization of the substrate (1158C,
10 min). Garlic was added after sterilization as a
20% extract prepared as described by Kyung et al.
(1996), except that the garlic was homogenized with
sterile distilled water using an ultra Turrax homogen-
izer. Fifty ml of extract was added aseptically to 200
ml of media to give a final concentration of 4%
garlic. The media were dispensed into 4-ml tubes
and inoculated with isolates grown on MRS-agar-
plates. The tubes were covered with paraffin oil and
incubated at 308C for up to five days.
2.2.2. Analysis of phenotypic data
The NTSYS-pc version 2.1 (Exeter software, New
York, USA) was used to analyse data based on 40
tests: 32 of the API-tests (17 tests were either
positive or negative for all isolates tested, and thus
not included in the numerical taxonomy analysis),
cell morphology (rods / coccoid rods or coccoids), gas
production from glucose, final pH after growth in
modified MRS-broth (pH # 3.7; 3.7 , pH # 4; pH .
4.0), fermentation of garlic and starch. Similarities
were assessed using the simple matching coefficient
(Ssm ) and the similarity matrices were clustered using
221
complete linkage and average linkage (UPGMA).
The type strains of Lactobacillus plantarum (DSM
20174), Lactobacillus casei (DSM 20011), Lac-
tobacillus paracasei subsp. paracasei (DSM 5622),
Lactobacillus pentosus (DSM 20314), Lactobacillus
brevis (DSM 20054), Lactobacillus curvatus (DSM
20019), Leuconostoc citreum (5577), Leuconostoc
mesenteroides subsp. mesenteroides (DSM 20343),
Pediococcus pentosaceus (DSM 20336) and Weisella
confusa (DSM 20196) were included as references.
2.3. Fermentation of som-fak with and without the
addition of garlic
Danish trout (Salmo trutta) was filleted and
minced. Three kg of fish mince was mixed with 0.12
kg of salt (3%), 0.45 kg of boiled, minced rice
(12%) and 0.15 kg of peeled, minced garlic (4%). In
another portion, 3 kg of fish mince was mixed with
0.11 kg of salt (3%) and 0.43 kg of boiled, minced
rice (12%). Three different mixed LAB cultures
were prepared using strains isolated from raw materi-
als and processed som-fak. Each mix was composed
of five strains with a capacity to ferment either
garlic / inulin (mix G), starch (mix S) or both garlic /
inulin and starch (mix G / S) (Table 1). The strains
were grown individually in MRS broth, mixed,
harvested (6500 g, 10 min, 258C), and diluted in
sterile 0.9% NaCl (w / v) and 0.1% peptone saline
(w / v). 0.5 ml of LAB mix was added per 50 g of
som-fak to an initial level of approximately 10 7
cfu / g. Uninoculated som-fak with and without garlic
was included as controls. All samples were vacuum-
packed in 50-g bags (Riloten / X / 40 / 50 Conovac,
Otto Nielsen Emballage, Lyngby, Denmark) and
stored at 308C. Samples in duplicates were with-
drawn daily (for six days) for chemical and mi-
crobiological analysis.
2.3.1. Chemical analysis
The pH was measured with an autocal pH meter
(Radiometer, Copenhagen, Denmark) in 5 g of
sample homogenised with 5 ml of deionized water.
Using the same homogenate, the amount of total
titratable acid was determined by titrating against 0.1
M standard NaOH to a final pH of 8.0. The % (w / w)
acid in the sample was calculated by multiplying the
volume of alkali (ml) by the factor 0.09.
222 ¨
C. Paludan-Muller et al. / International Journal of Food Microbiology 46 (1999) 219 – 229

Table 1
Strains isolated from raw materials and during fermentation of som-fak used in mixed LAB starter cultures for the fermentation of som-fak
with and without the addition of 4% garlic
Mixture of LAB Strain API-50 CH identification
strains designation Species ID (%)
Mix G 314 Pediococcus pentosaceus 99.9
(garlic)a 455 Lactobacillus plantarum 98.3
506 Lactobacillus plantarum 79.6
509 Lactobacillus plantarum 99.9
516 Lactobacillus plantarum 92.8

Mix S 302 Lactococcus lactis lactis 99.3

(starch)b 303 Lactococcus lactis lactis 99.9
344 Lactobacillus para. paracasei 97.2
383 Lactococcus lactis lactis 99.4
396 Lactococcus lactis lactis 99.9

Mix G / S 314 Pediococcus pentosaceus 99.9

(garlic and starch) 455 Lactobacillus plantarum 98.3
509 Lactobacillus plantarum 99.9
303 Lactococcus lactis lactis 99.9
383 Lactococcus lactis lactis 99.4
a
4% (w / w) minced garlic.
b
12% (w / w) minced rice.

2.3.2. Microbiological analysis
Ten g of sample was homogenised in 90 ml of
0.9% (w / v) NaCl and 0.1% (w / v) peptone for 60 s in
a stomacher 400 Lab Blender (A.J. Seward, Bury St.
Edmunds, UK). LAB were determined by spread-
plating on MRS-agar and plates were incubated in
anaerobic jars containing approximately 10% CO 2
and 90% N 2 for three days at 258C. Aerobic counts
were determined by spread-plating on Tryptone Soy
Agar (TSA) (CM131, Oxoid, Hampshire, UK) and
plates were incubated for three to four days at 258C.
Direct microscopy was carried out using the primary
dilution from all samples.
2.4. Comparison of the capacity of LAB mixed
cultures to ferment 0, 2, 4, 6 and 8% garlic in
model substrates
Garlic extract was prepared as described in Sec-
tion 2.2.1 above and 25, 50, 75 and 100 ml was
added aseptically to 225, 200, 175 and 150 ml
modified MRS-broth without glucose, containing the
same amount of MRS components, to give final
concentrations of 2, 4, 6 and 8% garlic. Substrate
without extract was included as control. No indicator
was added to the substrate. Strains were pre-cultured
individually in MRS-broth for 16 h and 500 ml of
each mix (100 ml of each strain) was inoculated into
10 ml of substrate corresponding to an initial level of
approximately 5 ? 10 7 cfu / g. Optical density at
OD 600 (Pharmacia LKB, Novaspec II) and pH was
measured daily for up to five days.
3. Results
3.1. Characterization of lactic acid bacteria
Of 202 representative and dominant strains iso-
lated from MRS-agar, 193 (96%) were tentatively
identified as LAB. Rods or coccoid rods were the
dominant LAB on raw materials accounting for 63–
100% of the isolated strains (Table 2). The LAB of
vegetable origin were dominated by obligate or
facultative heterofermentative and aciduric (end-
pH # 4 in MRS-broth) Lactobacillus species, where-
as the LAB isolated from fish raw material were
dominated by Lc. lactis subsp. lactis strains with
end-pH . 4 in MRS-broth. An increase in aciduric,
homofermentative Lactobacillus spp. was found
 ¨
C. Paludan-Muller et al. / International Journal of Food Microbiology 46 (1999) 219 – 229 223

Table 2
Characterization of LAB isolated from raw materials used for production of Thai som-fak and from som-fak during fermentation
% Positive isolates
Raw materials Days of fermentation
Discriminative Boiled Banana Garlic Fish Fish Day 0 Day 1 Day 2 Day 4 Day 5
tests rice leaves mix fillet mince
Rods / coccoid rods 100 63 67 100 87 95 100 87 93 100
NH 3 from arginine 100 88 87 100 73 55 56 53 15 43
CO 2 from glucose 25 63 60 9 27 50 16 20 0 0
End-pH in MRS # 4.0 17 88 93 18 27 45 52 100 100 100
Starch fermentation 30 0 0 21 38 28 0 0 0 0
Garlic fermentation 0 38 33 0 7 20 8 21 21 40
No. of isolates tested 12 8 15 11 15 22 25 30 27 28

during fermentation (Table 2). Starch fermenting

strains were isolated from fish fillets (21%), fish
mince (38%) and boiled rice (30%) (Table 2). In
som-fak, starch fermenting strains were only isolated
at day 0 (28%), before the start of fermentation.
Garlic fermenting strains were isolated from garlic
mix (33%), banana leaves (38%), fish mince (7%)
and throughout fermentation of som-fak (8–40%)
(Table 2). The estimated level of garlic fermenting
strains increased from approximately 10 6 cfu / g to
10 7 cfu / g during the first day of fermentation and
reached 10 8 cfu / g at the end of fermentation. All
garlic fermenting strains, except one strain of Lb.
coprophilus and three strains of Lb. plantarum, also
fermented inulin. The relationship of 185 strains of
the 193 strains identified (nine strains were un-
clustered) is illustrated in a simplified dendogram
(Fig. 1).
Cluster A (83% S) consisted of 27 strains, iden-
tified as Lc. lactis lactis, isolated from fish fillet, fish
mince, day 0 and day 1 during fermentation (Table
3). Cluster B (87% S) consisted of seven strains,
identified as Lb. paracasei subsp. paracasei, isolated Fig. 1. Simplified dendrogram showing relationship among 185
from boiled rice. The cluster differed from the type LAB strains isolated from raw materials and during fermentation
strain of Lb. paracasei subsp. paracasei by the of som-fak. The horizontal scale represents similarities. For
capacity of strains to ferment glycerol, D-turanose description of clusters see Table 3.
and, for five of the strains, starch (Table 3). Starch
fermenting strains were only included in these two
clusters (Table 3). Cluster C (78% S) consisted of 42 subsp. delbrueckii, although the similarity level was
strains, all isolated during fermentation. The type very low (50%). Cluster D (89% S) consisted of 18
strains of Lb. casei and Lb. paracasei casei were in strains isolated from garlic mix, banana leaves, day 0
this cluster at a 85% and 84% similarity level, and day 1 during fermentation. These strains were
respectively. However, by API 50-CH the strains identified as ‘‘Lb. coprophilus’’ (Weisella confusa,
were identified as Lb. acidophilus or Lb. delbrueckii Collins et al., 1994) and 22% of the strains fer-
224 ¨
C. Paludan-Muller et al. / International Journal of Food Microbiology 46 (1999) 219 – 229

Table 3
Origin and fermenting capacity of LAB strains, isolated from raw materials and during fermentation of som-fak, divided in clusters by
numerical analysis using UPGMA clustering
Cluster No. of Similarity API-50 CH Origin % Fermenting strains
isolates identification Garlic Starch
A 27 0.83 Lc. lactis lactis Fish fillet 0 44
Minced fish
Days 0 and 1
B 7 0.87 Lb. para. paracasei Boiled rice 0 71
C 42 0.78 Lb. casei Days 1, 2, 4 and 5 0 0
D 18 0.89 W. confusa Garlic mix 22 0
Banana leaves
Days 0 and 1
E 8 0.87 Leuc. citreum Minced fish 0 0
Days 0 and 2
F 8 0.75 Leuc. mesen. Days 1 and 2 0 0
G 7 0.81 Lb. curvatus Days 0, 1, 2 and 5 0 0
H 15 0.78 Lb. pentosus /Lb. plantarum Fish fillet 13 0
Garlic mix, (rice)
Days 0, 1, 2, 4 and 5
I 32 0.83 Lb. plantarum /Lb. brevis Days 0, 1, 2, 4 and 5 59 0
J 5 0.86 Pd. pentosaceus Minced fish 100 0
Garlic mix

mented garlic. Cluster E (87% S) consisted of eight
strains, identified as Leuc. citreum, isolated from
minced fish, day 0 and day 2 during fermentation.
Cluster F (75% S) consisted of eight strains iden-
tified as Leuc. mesenteroides subsp. mesenteroides,
isolated at day 1 and day 2 during fermentation.
Cluster G (81% S) consisted of seven strains iden-
tified as Lb. curvatus, all isolated during fermen-
tation. Garlic fermenting strains were found in
cluster H, I and J, which were similar at a 70% level
(Fig. 1). Cluster H (78% S) consisted of 15 strains,
identified as Lb. pentosus /plantarum, isolated from
fish fillet, garlic mix and during fermentation. Cluster
I (83% S) consisted of 32 strains, identified as Lb.
plantarum, all isolated during fermentation. The type
strain of Lb. plantarum was included in a subcluster
containing 10 strains at a similarity level of 85%
(results not shown). More than 80% of the strains not
included in this subcluster differed from the type
strain by their capacity to ferment garlic / inulin.
Cluster J (86% S) consisted of five strains, identified
as Pd. pentosaceus, isolated from garlic mix and
minced fish.
3.2. Fermentation of som-fak with and without the
addition of garlic
Aerobic counts (not including LAB) reached 10 8
cfu / g after one day in som-fak without garlic,
whereas in som-fak with garlic, the counts were
5 ? 10 6 cfu / g (results not shown). After two days, the
total counts were 2 ? 10 9 cfu / g in som-fak without
garlic and 8 ? 10 8 in som-fak with garlic. LAB counts
increased from 10 3 cfu / g to 5 ? 10 8 in two days
irrespective of garlic addition.
When the three mixtures of LAB strains were
added, the numbers increased from an initial level of
10 7 –5 ? 10 7 cfu / g to approximately 10 9 cfu / g in two
days (results not shown). In som-fak without added
LAB strains or with the addition of starch fermen-
ters, the pH was above 4.8 after three days of storage
(Figs. 2 and 3). In contrast, the mixed cultures of
garlic or garlic and starch fermenting strains were
able to ferment som-fak, but only when garlic was
added (Figs. 2 and 3). Garlic and starch fermenting
strains produced 1.6% acid after two days corre-
sponding to a decrease in pH to 4.5 (Fig. 3). After
 ¨
C. Paludan-Muller et al. / International Journal of Food Microbiology 46 (1999) 219 – 229 225

Fig. 2. Changes in content of (a) total titratable acid (%) and (b) pH during fermentation of laboratory prepared som-fak without garlic.
Inoculation with mixtures of starch fermenting LAB (mix S), garlic fermenting LAB (mix G), or garlic and starch fermenting LAB (mix
G / S).

Fig. 3. Changes in content of (a) total titratable acid (%) and (b) pH during fermentation of laboratory prepared som-fak with 4% garlic.
Inoculation with mixtures of starch fermenting LAB (mix S), garlic fermenting LAB (mix G), or garlic and starch fermenting LAB (mix
G / S).
226 ¨
C. Paludan-Muller et al. / International Journal of Food Microbiology 46 (1999) 219 – 229
six days, the amount of acid was 2.5% and pH was
about 4.3 (Fig. 3). With the garlic fermenting strains
alone, acid was produced at a slower rate, but the
final amount of acid was higher (3%) and pH
slightly lower (4%) (Fig. 3).
After two days of fermentation, the odour of all
samples of som-fak was described as cod roe. Som-
fak with garlic was characterized by a strong odour
of garlic throughout the fermentation period. The
odour of som-fak without added LAB strains was
described as fishy, rotten and sour after three days.
With the addition of starch fermenting isolates
similar odours were developed after six days. Som-
fak to which the mix of garlic fermenters or garlic
Fig. 4. Changes in pH during fermentation of modified MRS
without glucose containing 0, 2, 4, 6 and 8% garlic by (a) a
mixture of garlic fermenting LAB (mix G) and (b) a mixture of
garlic and starch fermenting LAB (mix G / S) inoculated to an
initial level of approximately 5 ? 10 7 cfu / g.
and starch fermenters had been added was described
as sour after three days of fermentation.
The mix of garlic fermenters and the mix of garlic
and starch fermenters were able to ferment 2, 4, 6
and 8% garlic in model substrates (Fig. 4). Two
percent garlic was fermented within one day with a
decrease to a final pH of 4.5. Garlic fermenting
strains fermented 4, 6 and 8% garlic at a slower rate,
but a lower final pH of 4.1 was reached after five
days. Garlic and starch fermenting strains fermented
4% garlic in two days, with a decrease in pH to 4.1,
whereas with 6 and 8% garlic a similar pH was
reached after five days (Fig. 4b). The decrease in pH
corresponded to the growth of the mixed cultures
measured as optical density (results not shown). In
substrates with 2% garlic, the cell density of both
garlic and garlic combined with starch fermenting
strains reached an OD 600 of 1.4 in one day. When a
concentration 4% garlic was used, the cell density of
the garlic fermenting strains did not exceed an OD 600
of 0.6, whereas the garlic and starch fermenting
strains reached an OD 600 of 1.8 (results not shown).
4. Discussion
One-hundred and eighty-five LAB, isolated from
raw materials and during fermentation of som-fak,
were identified and clearly separated in clusters
according to their origin and species identification by
phenotypic characterization. At the start of fermen-
tation, Leuconostoc spp. Lb. brevis and Lc. lactis
subsp. lactis were dominant, followed by more acid
tolerant species of Lb. curvatus, Lb. casei, Lb.
pentosus and Lb. plantarum, the latter dominating
the LAB microflora towards the end of fermentation.
This sequence of LAB growth is similar to vegetable
fermentations (e.g., sauerkraut, brined cucumbers
and korean kimchi) (Lee, 1997). An initial growth of
LAB producing mixtures of lactic and acetic acid is
suggested to be important for a rapid decrease in pH
(Adams and Hall, 1988). Acetic acid is also found to
be more effective than lactic acid against spoilage
bacteria in fish silage (Nygaard and Mjelde, 1987).
An association of heterofermentative Lactobacillus
sp. (W. confusa) and Pd. pentosaceus from the raw
materials of plant origin, garlic mix and banana
leaves, was found in this study in accordance with
Daeschel et al. (1987). In contrast, the LAB mi-
 ¨
C. Paludan-Muller et al. / International Journal of Food Microbiology 46 (1999) 219 – 229
croflora associated with rice and fish fillet were
dominated by homofermentative Lactobacillus sp.
(Lb. paracasei subsp. paracasei and Lb. pentosus)
and on fish fillet in addition by Lc. lactis subsp.
lactis. These strains were also less aciduric than the
strains isolated from plant material. The fermenting
properties of strains were in general associated with
the main carbohydrate substrate of their origin. Thus,
starch fermenting strains originated from boiled rice
and garlic or inulin fermenting strains from garlic
and banana leaves. Fish fillet contained starch fer-
menting strains and minced fish both garlic and
starch fermenting strains. Starch fermenting LAB
have been isolated from fresh tropical fish (Nair,
unpublished results) and from fermented fish prod-
ucts (Olympia et al., 1992). However, a starch
fermenting Lc. lactis subsp. lactis strain isolated
from the Thai product plaa-som was unable to
ferment laboratory prepared som-fak (Østergaard et
al., 1998b). Inulin fermenting Pd. pentosaceus and
Pediococcus sp. have been isolated from fermented
Thai foods, including fish products (Tanasupawat
and Daengsubha, 1983). In our study the inulin and
garlic fermenting strains were found among Lb.
plantarum, Lb. pentosus, Pd. pentosaceus (faculta-
tive heterofermentative) and W. confusa (obligate
heterofermentative). Lb. plantarum has recently been
used as starter culture for the fermentation of blan-
ched garlic with lactic acid as the major end-product
(de Castro et al., 1998). Otherwise the fermentation
of fructo-oligosaccharides by LAB has been found
among Lb. paracasei subsp. paracasei, Lb. plan-
tarum, Lb. brevis and Pd. pentosaceus ( , 1% of the
¨
species) naturally occurring on grasses (Muller and
Lier, 1994) and among oral streptococci (Hartemink
et al., 1995). In som-fak we found a succession of
garlic fermenting Lb. plantarum strains during fer-
mentation, suggesting that garlic may be more
important than rice-starch as a carbohydrate source
for fermentation. The results indicate that an initial
presence and growth of LAB with a capacity to
ferment garlic is essential for a rapid decrease in pH.
The addition of garlic or garlic and starch fermenting
LAB starter cultures to som-fak confirmed this, since
a decrease in pH similar to factory som-fak was
obtained. Model substrate containing 4% garlic as
the only carbohydrate source was fermented at a
similar rate as som-fak with garlic. The more rapid
decrease in pH obtained by the mix of garlic and
227
starch fermenting strains compared to the garlic
fermenters alone was found to correlate with a higher
growth rate of the mixed culture in garlic substrate.
Some fermented fish products that do not contain
garlic are prepared using substrates prefermented
with yeasts and moulds providing amylases for the
degradation of rice starch, such as red burong isda
(Sakai et al., 1983), plaa-paeng-daeng and plaa-
chao (Phithakpol et al., 1995). Yeasts are found in
levels of 10 6 –10 8 cfu / g, whereas in som-fak, yeasts
are found in lower levels of 10 2 –10 5 cfu / g (Saisithi
et al., 1986). In another Thai product, a plaa-som
variant of Southern Thailand, which is prepared from
fish, salt, palm sugar and sometimes roasted rice,
yeasts were also found in high levels together with
LAB (Paludan-Muller ¨ and Madsen, unpublished
results). A growth inhibitory effect of garlic has been
found against different types of industrial and food
spoilage yeasts (Conner and Beuchat, 1984) and
garlic is also found to inhibit food pathogens, such as
Staphylococcus aureus, Salmonella typhi, Es-
cherichia coli and Listeria monocytogenes (Kumar
and Berwal, 1998). In this study, the aerobic counts
of bacteria (not including LAB) were lower during
the first two days of fermentation in som-fak with
garlic as compared to som-fak prepared without
garlic. Thus, garlic may play a dual role in fermented
fish products by inhibiting Gram-negative bacteria
and yeasts and providing LAB less sensitive to garlic
with a carbohydrate source for growth. The viability
of the parasite Cysterticus cellulosae in nham, Thai
fermented pork sausage, was found to be prolonged,
when 3% garlic was used as compared to lower
levels of garlic (Keittivuti et al., 1986). The effect
was explained by a slower decrease in pH due to
inhibition of LAB by garlic, although the samples
did also contain higher levels of salt, which may also
delay the fermentation process. Garlic may be an
important selective factor for the sequences of LAB
growth during fermentation of som-fak and since the
capacity of LAB strains to ferment and tolerate garlic
is variable, a succession of specific strains for use in
starter cultures could improve safety and prevent
sensory spoilage. The growth of individual LAB
strains within the mixed cultures as well as the
selection of LAB strains during fermentation will be
further studied and the type of organic acids pro-
duced by LAB in fermented fish products containing
garlic and starch as carbohydrate sources will be
228 ¨
C. Paludan-Muller et al. / International Journal of Food Microbiology 46 (1999) 219 – 229
investigated in order to determine the preservative
effect of the fermentation process.
Acknowledgements
We thank Lonnie Hedberg for technical assistance
and Martin R. Adams, University of Surrey for
comments on the manuscript. The project was partly
financed by the Life Sciences and Technology for
Developing Countries (STD3) programme of the
Commission of the European Communities.
References
Adams, M.R., Cooke, R.D., Rattagool, P., 1985. Fermented fish
products of South East Asia. Trop. Sci. 25, 61–73.
Adams, M.R., Hall, C.J., 1988. Growth inhibition of food-borne
pathogens by lactic and acetic acids and their mixtures. Int. J.
Food Sci. Technol. 23, 287–292.
Beuchat, L.R., 1994. Antimicrobial properties of spices and their
essential oils. In: Dillon, V.M., Board, R.G. (Eds.), Natural
Antimicrobial Systems and Food Preservation. CAB Interna-
tional, Wallingford, pp. 167–179.
de Castro, A., Montano,˜ A., Sanchez,
´ H.A., Rejano, L., 1998.
Lactic acid fermentation and storage of blanched garlic. Int. J.
Food Microbiol. 39, 205–211.
Collins, M.D., Samelis, J., Metaxopoulos, J., Wallbanks, S., 1994.
Taxonomic studies on some leuconostoc-like organisms from
fermented sausages: description of a new genus Weisella for
the Leuconostoc paramesenteroides group of species. J. Appl.
Bacteriol. 75, 595–603.
Conner, D.E., Beuchat, R., 1984. Effects of essential oils from
plants on growth of food spoilage yeasts. J. Food Sci. 49,
429–434.
Daeschel, M.A., Andersson, R.E., Fleming, H.P., 1987. Microbial
ecology of fermenting plant materials. FEMS Microbiol. Rev.
46, 357–367.
Feldberg, R.S., Chang, S.C., Kotik, A.N., Nadler, M., Neuwirth,
Z., Sundstrom, D.C., Thompson, N.H., 1988. In vitro mecha-
nism of inhibition of bacterial cell growth by allicin. Anti-
microb. Agents Chemother. 32, 1763–1768.
Gibson, L.F., Khoury, J.T., 1986. Storage and survival of bacteria
by ultra-freeze. Lett. Appl. Microbiol. 3, 127–129.
Gregersen, T., 1978. Rapid method for distinction of Gram-
negative from Gram-positive bacteria. Eur. J. Appl. Microbiol.
5, 123–127.
Hartemink, R., Quataert, M.C.J., van Laere, K.M.J., Nout, M.J.R.,
Rombouts, F.M., 1995. Degradation and fermentation of
fructo-oligosaccharides by oral streptococci. J. Appl. Bacteriol.
79, 551–557.
Hugh, R., Leifson, E., 1953. The taxonomic significance of
fermentative versus oxidative Gram-negative bacteria. J. Bac-
teriol. 66, 24–26.
Ishige, N., 1993. Cultural aspects of fermented fish products in
Asia, In: Lee, C.-H., Steinkraus, K.H., Reilly, P.J.A. (Eds.),
Fish Fermentation Technology. United Nations University
Press, Tokyo, pp. 13–32.
Keittivuti, A., Keirrivuti, B., Tohaditep, P., 1986. The viability of
Cysticercus cellulosae in Thai native food (nham). Int. J.
Zoon. 13, 266–272.
Kimhamanon, T., 1994. Microbial changes during production of
fermented fish (som-fag). Master of Science Thesis in Food
Technology, Prince of Songkla University 2538 (in Thai).
Kumar, M., Berwal, J.S., 1998. Sensitivity of food pathogens to
garlic (Allium sativum). J. Appl. Microbiol. 84, 213–215.
Kyung, H.K., Park, K.S., Kim, Y.S., 1996. Isolation and charac-
terization of bacteria resistant to the antimicrobial activity of
garlic. J. Food Sci. 61, 226–229.
Lee, C.-H., 1997. Lactic acid fermented foods and their benefits in
Asia. Food Control 8, 259–269.
Leisner, J.J., 1992. Characterization of lactic acid bacteria isolated
from lightly preserved fish products and their ability to
metabolize various carbohydrates and amino acids. Ph.D.
Thesis. Royal Veterinary and Agricultural University,
Copenhagen, Denmark.
¨
Muller, M., Lier, D., 1994. Fermentation of fructans by epiphytic
lactic acid bacteria. J. Appl. Bacteriol. 76, 406–411.
˚
Nes, I.F., Skjelkvale, R., 1982. Effect of natural spices and
oleoresins on Lactobacillus plantarum in the fermentation of
dry sausage. J. Food Sci. 47, 1618–1625.
Nygaard, H., Mjelde, A., 1987. Lactic Acid Bacteria – Friend or
Enemy? Report from Sildolje-og Sildemelindustriens
Forskningsinstitut (SSF) No. 3, pp. 13–19 (in Norwegian).
Olympia, M., Ono, H., Shinmyo, A., Takano, M., 1992. Lactic
acid bacteria in a fermented fishery product, ‘‘Burong ban-
gus’’. J. Ferment. Bioeng. 73, 193–197.
Østergaard, A., Ben Embarek, P.K., Yamprayoon, J., Wedel-Neer-
gaard, C., Huss, H.H., Gram, L., 1998. Fermentation and
spoilage of som-fak, a Thai low-salt fish product. Trop. Sci.
38, 105–112.
Østergaard, A., Ben Embarek, P.K., Wedel-Neergaard, C., Huss,
H.H., Gram, L., 1998. Characterization of anti-listerial lactic
acid bacteria isolated from Thai fermented fish products. Food
Microbiol. 15, 223–233.
Owens, J.D., Mendoza, L.S., 1985. Enzymatically hydrolysed and
bacterially fermented fishery products. J. Food Technol. 20,
273–293.
Phithakpol, B., Varanyanond, W., Reungmaneepaitoon, S., Wood,
H., 1995. The Traditional Fermented Foods of Thailand.
Institute of Food Research and Product development, Kasetsart
University, Bangkok, pp. 12–20.
Saisithi, P., Yongmanitchai, P., Chimanage, P., Wongkhalaung, C.,
Boonyaratanakornkit, Maleehuan, S., 1986. Improvement of a
Thai Traditional Fermented Fish Product: Som-fug. Institute of
Food Research and Product Development, Kasertsart Universi-
ty, Bangkok, FAO report.
Sakai, H., Caldo, G.A., Kozaki, M., 1983. Yeast-flora in red
Burong isda, a fermented fish food in the Philippines. J. Agric.
Sci. 28, 181–185.
Shaw, B.G., Harding, C.D., 1984. A numerical taxonomic study of
lactic acid bacteria from vacuum-packed beef, pork, lamb and
bacon. J. Appl. Bacteriol. 56, 25–40.
 ¨
C. Paludan-Muller et al. / International Journal of Food Microbiology 46 (1999) 219 – 229
Souane, M., Kim, Y.B., Lee, C.-H., 1990. Microbial characteriza-
tion of gajami sik-hae fermentation. In: Reilly, P.J.A., Parry,
R.W.H., Barile, L.E. (Eds.), Post Harvest Technology, Pre-
servation and Quality of Fish in South-East Asia. Echamis
Press, Manila, pp. 35–49.
Steinkraus, K.H., 1996. Handbook of Indigenous Fermented
Foods. Marcel Dekker, New York, pp. 271–274.
Tanasupawat, S., Daengsubha, W., 1983. Pediococcus species and
related bacteria found in fermented foods and related material
in Thailand. J. Gen. Appl. Microbiol. 29, 487–506.
Tanasupawat, S., Okada, S., Suzuki, K., Kozaki, M., Komagata,
K., 1993. Lactic acid bacteria, particularly heterofermentative
229
Lactobacilli, found in fermented foods in Thailand. Bull. Jpn.
Fed. Culture Collection 9, 65–78.
Van Loo, J., Coussement, P., De Leenheer, L., Hoebregs, H.,
Smits, G., 1995. On the presence of inulin and oligofructose as
natural ingredients in the western diet. Crit. Rev. Food Sci.
Technol. 35, 525–552.
Vandamme, E.J., Derycke, D.G., 1983. Microbial inulinases:
fermentation process, properties and applications. Adv. Appl.
Microbiol. 29, 139–176.
Zaika, L.L., Kissinger, J.C., 1984. Fermentation enhancement by
spices: identification of active component. J. Food Sci. 49,
5–9.