Aquaculture research, tập 42, số 7, 2011

E-mail: fanqixue@ mail.hzau.edu.cn; Xiaoyin Wang, College of Science, Huazhong Agricultural University, Wuhan, Hubei, 430070, China E-mail: wxywxq@163.com Abstract Growth, nitrogenous ex

Effects of ration level on growth, nitrogenous excretion and energy budget of juvenile yellow catfish,

Lei Zhang1, Zhigang Zhao2, Dongmei Xiong1,Wei Fang1, Bo Li1, Qixue Fan1, Kai Yang1

& Xiaoyin Wang3

1 College of Fisheries, Huazhong Agricultural University,Wuhan, Hubei, China

2 The Key Laboratory of Mariculture, Ministry of Education, Ocean University of China, Qingdao, Shandong, China

3 College of Science, Huazhong Agricultural University,Wuhan, Hubei, China

Correspondence: Qixue Fan, College of Fishery, Huazhong Agricultural University, Wuhan, Hubei, 430070, China E-mail: fanqixue@ mail.hzau.edu.cn; Xiaoyin Wang, College of Science, Huazhong Agricultural University, Wuhan, Hubei, 430070, China E-mail: wxywxq@163.com

Abstract

Growth, nitrogenous excretion and energy budget of

juvenile yellow cat¢sh, Pelteobagrus fulvidraco (initial

body weight 1.17 0.28 g) at various levels (50%,

60%, 70%, 80%, 90% and 100% satiation per day)

were investigated with feeding diet containing 40%

protein Speci¢c growth rate of yellow cat¢sh

in-creased (2.79^3.34% day) signi¢cantly (Po0.05) with

ration level (RL) increasing Feed conversion

e⁄-ciency, feed protein retention e⁄ciency and feed

en-ergy retention e⁄ciency increased with the increase

in RL, peaked at 70% of satiation, and then decreased

at higher ration, with the ranges of 78.97^97.28%,

31.31^37.93% and 26.55^31.88% respectively Both

nitrogenous excretion (u, mg g 1day 1) and faecal

production (f, mg g 1day 1) increased signi¢cantly

with the increased RL, and ranged between 0.94^

1.38 and 0.69^1.24 mg g 1day 1respectively

Ap-parent digestibility coe⁄cients in dry matter, protein,

energy decreased signi¢cantly as ration increased,

with ranges of 54.42^69.64%, 78.24^89.90% and

69.66^82.07% respectively Energy budgets of

juve-nile yellow cat¢sh at satiation RL was:

100C 5 30F18U133R129G or 100A 5 54R146G

Keywords: ration level, growth, nitrogenous

ex-cretion, energy budget, yellow cat¢sh

Introduction

Yellow cat¢sh, Pelteobagrus fulvidraco (Richardson) is

a teleost ¢sh belonging to Siluriformes, Bagride,

Pel-teobagrus, is restricted to freshwater habitats, foundmostly in the east of Asia It is one of the favouritefood ¢sh in China because of its palatability and nu-tritional quality of its £esh and its high economic im-portance It has become one of the most importantfreshwater aquaculture species in South China(Han, Jia, Xia, Mao & Wang 2003; Jiang, Song,Ye, Cai,Yang & Huang 2004; Lee & Lee 2005) In addition, theproducts from yellow cat¢sh are available in overseasmarkets in recent years

With the increasing stock of cultured yellow

cat-¢sh, the large numbers of juveniles provided forlarge-scale aquaculture are mainly from arti¢cialbreeding In the arti¢cial breeding of yellow cat¢sh,especially during the period of using arti¢cially for-mulated feed, one key issue is how to optimize ration

to support fast growth and maximize feed conversionand to reduce feed waste and water pollution Suita-ble ration not only enhanced protein synthesis andenergy allocation for improving ¢sh growth and feedconversion e⁄ciency (FCE) but also minimized aqua-culture pollution and production cost A great deal ofpast research work have published on dealing withthe determination of optimum ration level (RL) of dif-ferent ¢sh species which has recently been reported

by Ahmed (2010) Thus, it is necessary to investigatethe e¡ects of RL on yellow cat¢sh growth, nitrogen-ous excretion, faecal production and energy budget.Although many aspects of yellow cat¢sh have beeninvestigated in the past, including biology (Liu 1997),culture technology (Han et al 2003), breeding(Zhang, Luo & Li 2000; Pan, Ding, Ge,Yan, Hao, Chen

& Huang 2008) and nutrition (Jiang et al 2004; Lee &

Aquaculture Research, 2011, 42, 899^905 doi:10.1111/j.1365-2109.2010.02626.x

Lee 2005); however, little is known about the

bioener-getics of this species (Yang & Yao 2006), therefore, the

present investigation was undertaken to work out the

various parameters such as growth, nitrogenous

ex-cretion, faecal production and energy budget of

juve-nile yellow cat¢sh in relation to ration

Materials and methods

Experimental ¢sh

Juvenile yellow cat¢sh, P fulvidraco for the

experi-ment were obtained from the arti¢cial breeding by

the research group at the National Research Center

of Freshwater Fisheries Engineering (Wuhan, China)

Fish with the same batch and in apparent

good health were collected from breeding ponds

About 4000 ¢sh were transferred into eight indoor

concrete ponds with about 500 ¢sh per pond

(2.1m 1.6 m  1.1m, water volume 3.0 m3) and

ac-climated for 4 weeks During this period, aeration

was provided continuously except for the feeding

time, dissolved oxygen was maintained above

6 mg L 1, water temperature was 28.5 1.3 1C, and

¢sh were subjected to a natural photoperiod regime

(20 June 2008^17 July 2008) During feeding, the ¢sh

were fed with test diet in the form of pellets (Table 1)

The acclimated ¢sh of similar body size and

show-ing normal feedshow-ing behaviour were collected from

the concrete ponds, and then randomly stocked into

a circular ¢breglass tanks (80 cm in diameter, water

volume 300 L) for further acclimation for 1 week A

preliminary feed trial was conducted to estimate the

maximal ration during this period, which would

pro-vide experimental epro-vidence for designing the RLs

used in the growth experiment

During the acclimation period, enough pelleted

feed (Table 1) was provided to satiation as judged by

visual inspection twice a day at 07:00 and 18:00

hours, respectively, as described earlier (Wang, Xie &

Ma 2001)

Experimental diet

The experimental diet contains 40% crude protein

and gross energy value of 16.70 kJ g 1 Chromium

(III)oxide was added as an inert indicator for

digest-ibility determinations Formulation and chemical

composition of the diet are shown in Table 1 The diet

was made into 1^2 mm pellets using a pellet press,

oven-dried at 60 1C and stored at 4 1C for use

Growth experimentSix RLs were tested in the growth experiment: 50%,60%, 70%, 80%, 90% and 100% of satiation, with tri-plicates for each RL containing 50 ¢sh for each group.Fish in the control group were hand-fed to apparentsatiation, 100% of satiation The daily feed suppliedwas recorded, and the uneaten feed was collectedafter active feeding for 40 min by pipetting and thenoven-dried at 70 1C Potential loss of uneaten feedwas determined by placing feed in water for 40 minand then collecting, drying and weighing The pro-portion of the feed remaining was calculated and thisvalue was used to adjust the amount of feed intake.Ration levels of ¢sh in the other ¢ve experimental

Table 1 Formulation (percentage of wet weight) and mical composition of the experimental diet

zImmune polysaccharide was from the Bide Biotechnology Co Ltd (Guangdong, China).

‰Han et al (2004), mineral premix (mg kg 1diet): NaCl, 500; MgSO 4  7H 2 O, 7500; NaH 2 PO 4  2H 2 O, 12500; KH 2 PO 4 , 16 000; Ca(H 2 PO 4 ) 2  H 2 O, 100 000; FeSO 4 , 1250; C 6 H 10 CaO 6  5H 2 O, 1750; ZnSO 4  7H 2 O, 176.5; MnSO 4  4H 2 O, 81; CuSO 4  5H 2 O, 15.5; CoSO 4  6H 2 O, 0.5; KI, 1.5; starch, 22.5.

zHan et al (2004), vitamin premix (mg kg  1 diet): thiamine, 20; ribo£avin, 20; pyridoxine, 20; cyanocobalamine, 2; folic acid, 5; calcium pantothenate, 50; inositol, 100; niacin, 100; biotin, 5; starch, 3226; ascorbic acid, 111; Vitamin A, 110; Vitamin D 3 , 20; Vitamin E ( DL -a-tocopherol acetate), 100; Vitamin K 3 (menadione sodium bisulphite), 10.

kCoated vitamin C was from the Sunhy Biology Company han, China).

(Wu-899–905

groups was determined based on the average feed

consumption of ¢sh in the control group

The trial was conducted in a recirculatory system

with running water consisting of 18 circular

¢bre-glass tanks (80 cm in diameter, 300 L, water

ex-change rate 6.0 L min 1) Fifteen to 20% water in

the system was replaced daily

Nine hundred ¢sh which had been starved for1day

were captured, blotted of excess water and was

indi-vidually weighed, then were placed into individual

experimental tanks at the start of the growth

experi-ment During the experiment, ¢sh (mean initial body

weight 1.17 0.28 g) in each tank were subjected to

the six prescribed RLs ranging from 50% of satiation

to 100% of satiation twice a day at 07:00 and 18:00

hours respectively Faeces were collected twice a day

(09:00 and 20:00 hours) throughout the whole

ex-perimental period by pipetting, oven-dried at 70 1C,

weighed, homogenized and stored at 20 1C for

bio-chemical analysis In addition, another 180 ¢sh were

sampled for measurement of initial body composition

and energy content

Ammonia-N and urea-N excretion were measured

once every10 days During the measurement

(ammo-nia of 0.31^2.28 mg L 1 and urea of 0.07^

0.52 mg L 1), the water £ow was stopped for 24 h

(Xie, Cui,Yang & Liu 1997).Water was sampled before

and after this period The ammonia-N and urea-N

were determined using the method of Chaney and

Marbach (1962) and converted into energy by

multi-plying with energy factors 24.83 kJ g 1for ammonia

and 23.03 kJ g 1for urea (Elliott 1976) Water

tem-perature, dissolved oxygen and pH were monitored

on daily basis Water temperature was 28.6 1.6 1C,

dissolved oxygen was maintained above 6.5 mg L 1

and pH was 7.1^7.6, and the experiment was

con-ducted at the natural photoperiod conditions with

si-milar light intensity for all tanks during this period

The growth experiment lasted 60 days All ¢sh were

weighed after 1 day of starvation and 10 ¢sh from

each tank were sampled randomly for biochemical

analysis

Chemical analysis

Dry matter content of diet and initial and ¢nal ¢sh

samples were determined by oven-drying at 105 1C

Nitrogen contents of ¢sh, diet and faeces were

ana-lyzed using the Kjeldahl method; and crude protein

contents were calculated (N-Kjeldahl 6.25) The

contents of crude lipid were measured for ¢sh and

diet by ether extraction (34.6 BP) Ash content of ¢shand diet were measured by combustion at 550 1C inmu¥e furnace Gross energy contents of ¢sh, faecesand diet were measured by oxygen bomb calorimeter(Phillipson microbomb calorimeter, Gentry Instru-ments, Aiken, SC, USA) The contents of Cr2O3forthe diet and faeces were determined as describedusing the method of Bolin (Bolin, King & Klosterman1952) Triplicate samples were measured for eachvariable and the mean of triplicate determinationwas taken as the result when the relative deviationwaso2%

Statistical analysisStatistical analyses were performed usingSTATISTICA6.0 for windows A one-wayANOVAfollowed by Dun-can’s multiple range tests was used for six RLs to de-termine the signi¢cant di¡erences (Po0.05) amongthe treatments Kolmogorov^Smirnov and Levene’stest were used to test the homogeneity and normality,and the arcsine transformation was used to the datawhen they were in percentage Data were expressed

as mean SE of triplicates

ResultsSpeci¢c growth rate (SGR) of juvenile yellow cat¢shincreased signi¢cantly (Po0.05) with the increase

in RLs, and ranged between 2.79% and 3.34% day(Table 2), but there were no signi¢cant di¡erences(P40.05) among 70^100% of satiation ration Feedconversion e⁄ciency (FCE), feed protein retention ef-

¢ciency (PRE) and feed energy retention e⁄ciency(ERE) of ¢sh increased with RL, and peaked at 70%

of satiation, and then decreased at higher ration,and ranged between 78.97^97.28%, 31.31^37.93%and 26.55^31.88% respectively

The contents of dry matter, protein, lipid, ash andenergy in the body of juvenile yellow cat¢sh at di¡er-ent RLs at the beginning and end of growth experi-ment are shown in Table 3.With the increase in RLs,dry matter, lipid and energy contents of ¢sh in-creased, whereas ash contents of ¢sh declined Thecontents of dry matter and energy at 100% of satia-tion ration were signi¢cantly (Po0.05) higher thanthose fed at other RLs The protein content was notsigni¢cantly (P40.05)

Both nitrogenous excretion and faecal productionincreased signi¢cantly (Po0.05) and ranged be-tween the increased RLs, with ranges of 0.94^1.38

and 0.69^1.24 mg g 1day 1respectively (Table 4).

Apparent digestibility coe⁄cients in dry matter

(ADCd), protein (ADCp), energy (ADCe) of yellow

cat¢sh decreased signi¢cantly (Po0.05) as ration

in-creased, with ranges of 54.42^69.64%,78.24^89.90%

and 69.66^82.07% respectively

Energy budgets at each RL for juvenile yellow

cat-¢sh were shown in Table 5 The proportions of energy

intake lost in faecal production and excretion were

17.93^30.34% and 7.03^7.69%, respectively, and both

proportions tended to increase with an increase in

RL As ration increased, the proportion of energy

in-take spent in metabolism decreased and was highest

at 50% of satiation ration and ranged between

33.27% and 47.11%, whereas the proportions of

en-ergy intake stored in body as growth enen-ergy

in-creased with the RL, peaked at 70% of satiation, and

then decreased at higher ration with a ranged

be-tween 26.55% and 31.88%

DiscussionRelatively high (97.33^100.00%) survival of juvenileyellow cat¢sh was obtained in the present study atthe end of the trial Many authors reported optimumfeeding regimes based on the relationship between

RL and SGR or FCE (Hogendoorn 1983; Gˇnther, vez-Hidalgo, Ulloa-Rojas, Coppoolse & Verreth 1992;Cortes & Gruber 1994; Sun, Chen, Huang, Wang &Yan 2006) Usually, growth^ration relationship has adecelerating pattern and FCE shows a domed curve

Gal-as ration increGal-ases, so growth and FCE were not imized at the same RL (Xie et al.1997; Sun et al 2006)

max-In this case, the optimum feeding strategy would bethat ¢sh are fed at an intermediate not maximum ra-tion to obtain both rapid growth and high FCE Pre-sent study indicated that, as ration increased, growthshowed a decelerating curve, but the di¡erenceswere not signi¢cant at 70%, 80%, 90% and 100% of

Table 2 Growth, survival and conversion e⁄ciency of Pelteobagrus fulvidraco fed di¡erent ration levels (RL)

Feeding rate 2.27  0.02 a 2.38  0.01 b 2.59  0.02 c 2.67  0.04 d 2.85  0.02 e 3.08  0.02 f Initial body weight 1.16  0.06 1.20  0.10 1.23  0.02 1.19  0.07 1.21  0.02 1.19  0.05 Final body weight 6.17  0.25 a 6.91  0.07 b 8.09  0.04 c 8.42  0.18 cd 8.72  0.28 cd 8.77  0.27 d SGR 2.79  0.11 a 2.93  0.14 ab 3.14  0.04 bc 3.26  0.07 c 3.29  0.03 c 3.34  0.09 c FCE 94.28  2.59 bc 90.82  1.92 b 97.28  0.82 c 89.60  2.30 b 89.01  1.79 b 78.97  4.26 a PRE 37.93  1.32 b 35.80  0.88 b 37.83  0.45 b 34.95  1.70 ab 34.30  0.58 ab 31.31  1.76 a ERE 26.55  0.92 a 28.13  0.53 ab 31.88  0.28 c 30.92  1.00 bc 30.43  0.53 bc 28.70  1.71 ab Survival 97.33  2.67 a 98.67  0.67 a 100.00  0.00 a 98.67  0.67 a 98.67  1.33 a 98.67  1.33 a

Mean values (mean  SE of three replicates) in the same row with di¡erent letters are signi¢cantly di¡erent (Po0.05).

Feeding rate (FR) 5 total feed intake  100/f[(initial body weight1¢nal body weight)/2]  feeding daysg

Speci¢c growth rate (SGR) 5 100  [ln (¢nal body weight)  ln (initial body weight)]/days of the experiment.

Feed conversion e⁄ciency (FCE) 5 100  (¢nal body weight  initial body weight)/feed intake.

Feed protein retention e⁄ciency (PRE) 5 100  [(¢nal body weight  ¢nal protein content)  (initial body weight  initial protein content)]/(feed intake  protein content).

Feed energy retention e⁄ciency (ERE) 5 100  [(¢nal body weight  ¢nal energy content)  (initial body weight  initial energy tent)]/(feed intake  energy content).

con-Table 3 Body composition (%) and energy content (kJ g 1) of Pelteobagrus fulvidraco fed di¡erent ration levels (RL, % per day)

RL Initial body composition 50 60 70 80 90 100

Dry matter 22.0  0.2 24.0  0.5 a 24.5  0.2 a 24.6  0.3 a 24.8  0.2 a 24.7  0.1 a 25.8  0.3 b Protein 13.8  0.1 15.6  0.2 15.4  0.0 15.3  0.3 15.3  0.3 15.2  0.0 15.5  0.0 Lipid 3.6  0.1 4.2  0.1 a 5.2  0.3 b 5.4  0.3 b 5.9  0.4 b 5.8  0.2 b 6.1  0.3 b Ash 3.5  0.1 3.8  0.1 b 3.7  0.1 ab 3.6  0.1 ab 3.6  0.0 ab 3.6  0.1 ab 3.5  0.0 a Energy 4.8  0.1 4.7  0.1 a 5.1  0.1 b 5.4  0.1 c 5.6  0.0 c 5.6  0.0 c 5.9  0.1 d

Mean values (mean  SE of three replicates) in the same row with di¡erent letters are signi¢cantly di¡erent (Po0.05).

899–905

satiation; however, FCE showed a domed curve and

peaked at 70% of satiation

The whole-body composition of ¢sh is often used as

an indicator of ¢sh quality Several factors, including

growth and diet are known to a¡ect the body

compo-sition of ¢sh Body compocompo-sition is also signi¢cantly

a¡ected by feeding rate (Panda, Mishra &

Samantar-ay 1999; AdebSamantar-ayo, Balogun & Fagbenro 2000; Khan,

Jafri & Chadha 2004; Ahmed 2007) The present

re-sults revealed that crude lipid content increased

when ash content decreased with increased rations,

consistent with the previous studies (Shimeno,

Shi-kata, Hosokawa, Masumoto & Kheyyali 1997;

Abdel-ghany & Ahmad 2002; Han, Xie, Lei, Zhu & Yang

2004; Cho, Lee, Park & Lee 2006) When ¢sh were

fed at higher rations, lipid accumulation was

preva-lent One of possible reason is that there is a trend

for ration to determine growth and more food energy

at high RL could be converted into ¢sh fat (Shearer,

Silverstein & Dickho¡ 1997; Han et al 2004) The

de-cline in ash content with RLs could be caused by

re-latively a lower proportion of bones compared withother tissues (Rasmussen & Ostenfeld 2000; Han

et al 2004) However, in the present studies crudeprotein content of ¢sh was not signi¢cantly a¡ected

by di¡erent RLs, which were similar to previous dies (Huisman1976; Grayton & Beamish1977; Reinitz1983; Storebakken & Austreng 1987a, b)

stu-Based on the above results, it is recommended that70% of satiation ration per day (actual 68% of satia-tion ration), about 2.59% body weight per day with10.3 g protein kg 1day 1 and 432.5 kJ digestibleenergy kg 1day 1, would be useful for optimum RLfor maximum growth and FCE, and excellent bodycomposition with the formulated feed (Table 1) Theresult is higher than the RLs reported for other cul-tured ¢sh species in terms of body weight/day whichwas recently reported byAhmed (2010)

Nitrogenous excretion and faecal production areimportant indexes, which can estimate water pollu-tion in the intensive culture In the present study,both nitrogenous excretion and faecal production of

Table 4 Nitrogenous excretion (u, mg g 1day 1), faecal production (f, mg g 1day 1) and apparent digestibility coe⁄cients

in dry matter (ADC d , %), protein (ADC p , %), energy (ADC e , %) of Pelteobagrus fulvidraco fed di¡erent ration levels (RL, % per day)

u 0.94  0.04 a 1.04  0.02 ab 0.98  0.01 ab 1.14  0.06 b 1.15  0.03 b 1.38  0.10 c

F 0.69  0.03 a 0.77  0.04 a 0.78  0.02 a 0.93  0.04 b 1.11  0.00 c 1.24  0.10 c ADC d 68.12  0.96 c 69.64  0.53 c 67.12  0.53 c 62.74  1.52 b 56.17  1.13 a 54.42  0.81 a ADC p 89.90  0.68 d 89.69  0.33 d 88.86  0.53 d 86.14  1.03 c 81.07  1.16 b 78.24  0.66 a ADC e 80.71  0.91 c 82.07  0.48 c 80.33  0.77 c 74.75  1.52 b 71.58  1.72 ab 69.66  1.30 a

Mean values (mean  SE of three replicates) in the same row with di¡erent letters are signi¢cantly di¡erent (Po0.05).

ADC d 5100  (1  Cr 2 O 3 in the diet/Cr 2 O 3 in the faeces).

ADC p 5100  (1  Cr 2 O 3 in the diet  crude protein content in the faeces/Cr 2 O 3 in the faeces  crude protein content in the diet) ADC e 5100  (1  Cr 2 O 3 in the diet  gross energy content in the faeces/Cr 2 O 3 in the faeces  gross energy content in the diet).

Table 5 Energy budgets of Pelteobagrus fulvidraco fed di¡erent ration levels (RL, % per day)

Food energy (kJ g  1 day  1 ) 0.404  0.010 0.432  0.007 0.421  0.002 0.468  0.014 0.473  0.008 0.539  0.025

As a percentage of food energy (%)

Growth energy 26.55  0.92 a 28.13  0.53 ab 31.88  0.28 c 30.92  1.00 bc 30.43  0.53 bc 28.70  1.71 ab Faeces energy 19.29  0.91 a 17.93  0.48 a 19.67  0.77 a 25.25  1.52 b 28.42  1.72 bc 30.34  1.30 c Excretion energy 7.04  0.16 a 7.27  0.09 ab 7.03  0.05 a 7.34  0.20 ab 7.36  0.08 ab 7.69  0.20 b Metabolism energy 47.11  1.62 c 46.67  0.80 c 41.41  0.78 b 36.49  1.64 a 33.79  1.64 a 33.27  1.08 a

As a percentage of assimilated energy (%)

Metabolism energy 63.93  1.56 b 62.39  0.79 b 56.49  0.51 a 54.09  1.58 a 52.56  1.32 a 53.74  2.01 a Growth energy 36.07  1.56 a 37.61  0.79 a 43.51  0.51 b 45.91  1.58 b 47.44  1.32 b 46.26  2.01 b

Mean values (mean  SE of three replicates) in the same row with di¡erent letters are signi¢cantly di¡erent (Po0.05).

C 5 F1U1R1G or A 5 R1G, where C, food energy; A, assimilated energy; F, faeces energy; U, excretion energy; R, metabolism energy, calculated by di¡erence R 5 C  F  U  G and G, growth energy.

juvenile yellow cat¢sh increased with increased

ra-tion Higher rations could be result in water pollution

because of more nitrogenous excretion and faecal

production in the yellow cat¢sh intensive culture

Some studies have reported that apparent energy

digestibility coe⁄cients showed no signi¢cant

di¡er-ences related to RL (Allen 1980; Hogendoorn1983) In

the present study, apparent digestibility coe⁄cients

in dry matter, protein, energy of yellow cat¢sh

de-creased signi¢cantly with inde-creased of ration, which

is an agreement with the values reported on African

cat¢sh and Nile tilapia by Henken, Kleingeld and

Tijs-sen (1985) and Xie et al (1997)

In the present study, the estimate of food energy

(C), growth energy (G), faeces energy (F) and

excre-tion energy (U) should be accurate Although the

es-timation of allowed errors might be a¡ected by

nutrient leaching problems for faeces energy and by

the estimation of initial energy content of

experimen-tal ¢sh from the mean values of the control ¢sh for

growth energy, and the remaining component,

meta-bolism energy (R), was estimated indirectly by

di¡er-ence R 5 C F  U  G as theoretical energy

budgets should be balanced (Sun & Chen 2009) The

proportion of food energy intake channelled into

growth showed the highest value at an intermediate

ration (70% of satiation) and was lower at below or

above RLs The food energy lost in metabolism

de-creased with increasing RLs This could be due to

the limited food resources at low RLs Fish kept

swim-ming to enhance the chances of obtaining feed,

in-creasing the energy cost for activity (Han et al

2004) When increasing RLs to 80% of satiation, ¢sh

activity decreased, and the metabolism energy were

gradually stabilize Based on the information

avail-able on energy budgets for 14 ¢sh species, Cui and

Liu (1990) calculated the following average energy

budget for ¢sh fed at satiation ration: 100A 5

60R140G In the present study, energy budget of

ju-venile yellow cat¢sh at satiation ration was:

100A 5 54R146G The proportion of assimilated

en-ergy spent in growth for juvenile yellow cat¢sh was

slightly higher than the average value, consistent

with the results for southern cat¢sh (Silurus

meridio-nalis) (Xie & Sun 1993) and Chinese longsnout cat¢sh

(Leiocassis longirostris Gˇnther) (Han et al 2004), but

di¡ers from the values reported for Nile tilapia

(Oreo-chromis niloticus) of R (76%), G (24%) (Xie et al 1997)

and Gibel carp (Carassius auratus gibelio) of R (75%), G

(25%) (Zhu, Xie & Cui 2000), which were carnivores

¢sh This seemed to suggest that yellow cat¢sh has

moderate growth e⁄ciency and metabolic

expendi-ture, and the results were accordant with the factthat the yellow cat¢sh is an omnivorous ¢sh species

AcknowledgmentsThe authors are grateful for the project support bythe National Key Technology R&D Program of China(Grant No 2007BAD37B02)

References

Abdelghany A.E & Ahmad M.H (2002) E¡ects of feeding rates on growth and production of Nile tilapia, common carp and silver carp polycultured in fertilized ponds Aquaculture Research 33, 415^423.

Adebayo O.T., Balogun A.M & Fagbenro O.A (2000) E¡ect of feeding rates on growth, body composition and economic performance of juvenile clariid cat¢sh hybrid (, Clarias gariepinus  < Heterobranchus bidorsalis) Journal of Aquaculture in theTropics 15, 109^117.

Ahmed I (2007) E¡ect of ration size on growth, body position, and energy and protein maintenance require- ment of ¢ngerling Indian major carp, Labeo rohita (Hamilton) Fish Physiology and Biochemistry 33, 203^ 212.

com-Ahmed I (2010) Response to the ration levels on growth, body composition, energy, and protein maintenance re- quirement of the Indian cat¢sh (Heteropneustes fossilis- Bloch 1974) Fish Physiology and Biochemistry 36, 1133– 1143.

Allen J.R.M (1980) The estimation of natural feeding rate of the three-spined stickleback, Gasterosteus aculeatus L (Pisces) PhD thesis, University of Wales, Aberystwyth, 341pp Bolin D.W., King R.P & Klosterman E.W (1952) A simpli¢ed method for determination of chromic oxide (Cr 2 O 3 ) when used as an inert substance Science 116, 634^635 ChaneyA.L & Marbach E.P (1962) Modi¢ed reagents for de- termination of urea and ammonia Clinical Chemistry 8, 130^132.

Cho S.H., Lee S.M., Park B.H & Lee S.M (2006) E¡ect of ing ratio on growth and body composition of juvenile ol- ive £ounder Paralichthys olivaceus fed extruded pellets during the summer season Aquaculture 251,78^84 Cortes E & Gruber S.H (1994) E¡ect of ration size on growth and gross conversion e⁄ciency of young lemon sharks, Negaprion brevirostris Journal of Fish Biology 44, 331^341 Cui Y.B & Liu J.K (1990) Comparison of energy budget among six teleosts ^ I Food consumption, faecal produc- tion and nitrogenous excretion Comparative Biochemistry and Physiology 96A, 163^171.

feed-Elliott J.M (1976) Energy losses in the waste products of brown trout (Salmo trutta L.) Journal of Animal Ecology

45, 561^580.

899–905

Grayton B.D & Beamish F.W.H (1977) E¡ect of feeding

fre-quency on food intake, growth and body composition of

rainbow trout (Salmo gairdneri) Aquaculture 11, 159^172.

Gˇnther J., Galvez-Hidalgo N., Ulloa-Rojas J., Coppoolse J &

Verreth J.A.J (1992) The e¡ect of feeding level on growth

and survival of jaguar guapote (Cichlasoma managuense)

larvae fed Artemia nauplii Aquaculture 107, 347^358.

Han D., Xie S.Q., Lei W., Zhu X.M & YangY.X (2004) E¡ect of

ration on the growth and energy budget of Chinese

long-snout cat¢sh, Leiocassis longirostris Gˇnther Aquaculture

Research 35, 866^873.

Han Q., Jia K.S., Xia W.F., Mao X.J & Wang W.B (2003) Trial

on the cage culture of Pelteobagrus fulvidraco Fisheries

Science 22, 32^33 (in Chinese with English abstract).

Henken A.M., Kleingeld D.W & Tijssen P.A.T (1985) The e¡ect

of feeding level on apparent digestibility of dietary dry

mat-ter, crude protein and gross energy in the African cat¢sh,

Clarias gariepinus (Burchell,1822) Aquaculture 51, 1^11.

Hogendoorn H (1983) Growth and production of African

cat¢sh, Clarias lazera (C and V.) III Bioenergetic relations

of body weight and feeding level Aquaculture 35, 1^17.

Huisman E.A (1976) Food conversion e⁄ciencies at

mainte-nance and production levels for carp, Cyprinus carpio L.,

and rainbow trout, Salmo gairdneri Richardson

Aquacul-ture 9, 259^276.

Jiang R., Song X.H., Ye Y.T., Cai C.F., Yang C.G & Huang H.Z.

(2004) Optimum dietary protein and energy-protein ratio

of yellow cat¢sh Pelteobagrus fulvidraco Richardson

Jour-nal of Dalian Fisheries University 19, 252^257 (in Chinese

with English abstract).

Khan M.A., Jafri A.K & Chadha N.K (2004) Growth and

body composition of rohu, Labeo rohita (Hamilton), fed

compound diet:Winter feeding and rearing to marketable

size Journal of Applied Ichthyology 20, 265^270.

Lee O.K & Lee S.M (2005) E¡ects of the dietary protein and

lipid levels on growth and body composition of bagrid

catsh, Pseudobagrus fulvidraco Aquaculture 243, 323^329.

Liu S.P (1997) A study on the biology of Pseudobagrus

fulvi-draco in Poyang lake Chinese Journal of Zoology 32, 10^16

(in Chinese with English abstract).

Pan J.L., Ding S.Y., Ge J.C., Yan W.H., Hao C., Chen J.X &

Huang Y.H (2008) Development of cryopreservation for

maintaining yellow cat¢sh Pelteobagrus fulvidraco sperm.

Aquaculture 279, 173^176.

Panda S., Mishra K & Samantaray K (1999) E¡ect of feeding

rate on the growth performance of Channa punctatus (Bloch)

fry, and protein and energy requirement for their

mainte-nance and maximum growth Aquaculture 7, 37^42.

Rasmussen R.S & Ostenfeld T.H (2000) E¡ect of growth rate

on quality traits and feed utilization of rainbow trout

(On-corhynchus mykiss) and brook trout (Salvelinus fontinalis) Aquaculture 184, 327^337.

Reinitz G (1983) Relative e¡ect of age, diet and feeding rate

on the body composition of young rainbow trout (Salmo gairdneri) Aquaculture 35, 19^27.

Shearer K.D., Silverstein J.T & Dickho¡ W.W (1997) Control

of growth and adiposity of juvenile Chinook salmon corhynchus tshawytscha) Aquaculture 157, 311^323 Shimeno S., Shikata T., Hosokawa H., Masumoto T & Kheyyali D (1997) Metabolic responses to feeding rates

(On-in common carp, Cypr(On-inus carpio Aquaculture 151, 371^ 377.

Storebakken T & Austreng E (1987a) Ration level for nids I Growth, survival, body composition and feed con- version in Atlantic salmon fry and ¢ngerlings Aquaculture 60, 189^206.

Storebakken T & Austreng E (1987b) Ration level for nids II Growth, feed intake, protein digestibility, body composition, and feed conversion in rainbow trout weighing 0.5^1.0 kg Aquaculture 60, 207^221.

salmo-Sun L.H & Chen H.R (2009) E¡ects of ration and ture on growth, fecal production, nitrogenous excretion and energy budget of juvenile cobia (Rachycentron cana- dum) Aquaculture 292, 197^206.

tempera-Sun L.H., Chen H.R., Huang L.M.,Wang Z.D & YanY (2006) Growth and energy budget of juvenile cobia (Rachycen- tron canadum) relative to ration Aquaculture 257, 214^ 220.

Wang C.F., Xie C.X & Ma J (2001) The feeding rhythm and the daily feeding rates of Pelteobagrus fulvidkraco in early life history Chinese Journal of Fisheries 14, 66^68 (in Chi- nese with English abstract).

Xie S.Q., Cui Y.B.,YangY.X & Liu J.K (1997) Energy budget of Nile tilapia (Oreochromis niloticus) in relation to ration size Aquaculture 154, 57^68.

Xie X.J & Sun R.Y (1993) Pattern of energy allocation in the southern cat¢sh (Silurus meridionalis) Journal of Fish Biol- ogy 42, 197^207.

Yang Y.O & Yao F (2006) E¡ect of ration level on nitrogen and energy budget of yellow cat¢sh Hubei Agricultural Sciences 45, 653^655 (in Chinese with English abstract) Zhang G.Q., Luo X.N & Li B (2000) Critical techniques for arti¢cial propagation of Pelteobagrus fulvidraco (Richard- son) Journal of Dalian Fisheries University 15, 293^299 (in Chinese with English abstract).

Zhu X.M., Xie S.Q & Cui Y.B (2000) E¡ect of ration size on growth and energy budget of Gibel carp Oceanologia et Limnologia Sinica 31, 471^479 (in Chinese with English abstract).

Real-time quantification of the immune gene

expression in rainbow trout fed different forms of

Akshaya Panigrahi1,2, Kiron Viswanath3& Suichi Satoh1

1 Department of Marine Biosciences,Tokyo University of Marine Science and Technology, Minato,Tokyo, Japan

2 Central Institute of Brackish Water Aquaculture, Chennai,Tamil Nadu, India

3 Faculty of Biosciences and Aquaculture, Bodo University College, Bodo, Norway

Correspondence: A Panigrahi, Central Institute of Brackish Water Aquaculture,75, Santhom High Road, R.A Puram, Chennai-28,Tamil Nadu, India E-mail: panigrahi@ciba.res.in

Abstract

Whether it is better to use viable or non-viable

pro-bionts in aquaculture is still a matter of debate In this

study, the molecular immunomodulation in rainbow

trout Oncorhynchus mykiss induced by viable or killed

forms of the probiont Lactobacillus rhamnosus JCM

1136 was investigated Three forms of this probiont:

(1) heat-killed (HK), (2) live spray (LI) and (3)

freeze-dried (FD) were incorporated into a basal (control)

diet for rainbow trout O mykiss The LI and FD diets

are referred to as viable diets A rearing trial, in

tripli-cate, was conducted for 30 days, with the control and

probiotic diets as treatments The cytokine genes

such as the tumour necrosis factor (TNF),

transform-ing growth factor (TGF-b), interferon (IFN) and

immune gene Immunoglobulin (Ig) found in tissues

from the kidney and spleen were assessed for their

expression pattern by real-time polymerase chain

re-action The tested immune genes were up-regulated

in the treatment groups, sometimes even in many

folds like in the case of the Ig gene The TNF gene

was found to be highly (Po0.05) up-regulated

(5000-fold) in groups fed both viable forms (LI, FD)

With regard to the TGF-b gene, the spleen of the HK

and FD groups showed signi¢cant up-regulation of

20- and 30-folds respectively The IFN gene was

up-regulated (Po0.05) in all treatments, but more in the

viable diet treatments Kidney and spleen tissues

showed similar expression patterns, i.e all of these

genes were up-regulated more with the viable diets

that with the control, and in most cases, the viable

diets induced a higher expression of the immune

genes than the HK diet

Keywords: lactic acid bacteria, rainbow trout,probiotics, cytokines, gene expression

IntroductionProbiotics are de¢ned as living microorganisms that,when administered in adequate amounts, confer ahealth bene¢t on the host The ever-increasing use ofprobiotics in aquaculture for alternative health man-agement makes it essential to unravel the underlyingmechanism Probiotics can keep pathogens at baythrough competitive exclusion or host immunityenhancement In aquaculture, de¢ned probiotics areused as remediation agents in a number of ways such

as enrichment of larval food, inclusion in the diet or dition to the water to maintain a balanced population.Several probiotic species including the lactic acidbacteria (LAB) have been evaluated for various can-didate aquaculture species (Gatesoupe 1991, 1994,1999; Robertson, Reijula, Jarvis,Veijalainen & Hintik-

ad-ka 1997; Ringo & Gatesoupe 1998; Robertson, Dowd &Burrells et al 2000; Vershuere, Rombaut, Sorgeloos &Verstraete 2000; Balcazar,Vendrell, De Blas, Ruiz-zar-zuela, Girones & Muzquiz 2007; Salinas, Myklebust,Esteban, Olsen, Meseguer & Ringo 2008) Probiotics

in aquaculture act as non-speci¢c lants (Anderson1992; Balcazar, De Blas, Ruiz-zarzue-

immunostimu-la, Vendrell, Calvo, Marquez, Girones & Muzquiz2007), help improve water quality (Vadstein, Oie,Olsen, Salvesen, Skjermo & Skjak-braek 1993;Skjermo & Vadstein 1999), enhance larval survivaland growth (Skjermo & Vadstein 1999) and helpsupplement nutrition (Ronnestad, Thorsen & Finn

Aquaculture Research, 2011, 42, 906^917 doi:10.1111/j.1365-2109.2010.02633.x

1999) Further, several studies (Noh, Han,Won & Choi

1994; Moriarty 1998) have reported the use of

func-tional feeds including probionts to induce growth

and feed utilization The probiotic bacteria Bacillus

sp in viable form have been proven to be bene¢cial

for the growth and survival of shrimp (Rengpipat,

Phianpak, Piyatiratitivorakul & Menasveta 1998)

Culture viability is one of the important criteria in

the standardization of probiotics A concentration of

107CFU g 1at the time of consumption is considered

to be functional (Gomes & Malcata 1999) in humans

The information available on the potential probiotic

Lactobacillus rhamnosus JCM 1136 provides

convin-cing evidence of its safety (GRAS generally regarded

as safe) This probiotic bacterium is known to

im-prove immune response and disease resistance in

rainbow trout (Nikoskelainen, Ouwehand, Bylund,

Salminen, Lilius 2003; Panigrahi, Kiron, Kobayashi,

Puangkaew, Satoh & Sugita 2004; Panigrahi, Kiron,

Puangkaew, Kobayashi, Satoh & Sugita 2005) and

ti-lapia (Pirarat, Kobayashi, Katagiri, Maita & Endo

2006) Fructooligosachharides (FOS) as a prebiotic

component in addition to the probiotic bacteria are

also known to modulate lipid metabolism and

im-prove the bioavailability of essential minerals There

is an urgent need to know the density of probiotics,

and for the same reason, it is important to

under-stand the method of administration and dosage as

that of the form of bacteria used

Probiotic and cytokine gene interaction as reported

previously (Panigrahi, Kiron, Satoh, Hirono,

Kobaya-shi, Sugita, Puangkaewa & Aoki 2007) indicates that

there is a strain-speci¢c up-regulation of interleukins

(IL-b1), tumour necrosis factor (TNF) and

transform-ing growth factor (TGF-b) in the spleen and kidney

tis-sue of probiotic fed rainbow trout However, it is

unlikely that only the viable form of probionts can be

immunostimulating Previous studies have illustrated

that the viable form exhibits a better e¡ect of elevating

the cellular and humoral immune response than that

of the heat^killed (HK) form In this study, the gene

expression has been quanti¢ed following the

interac-tion of macrophages with living and killed probionts

provided through feed

The present study aimed to investigate the immune

gene expression induced by feeding di¡erent forms of

the probiont L rhamnosus JCM 1136 to rainbow trout

The in£uence of the non-viable HK bacteria is

com-pared with that of the viable form [as viable sprayed

and freeze-dried (FD)] of L rhamnosus on the immune

gene [TNF,TGF-b, interferon (IFN) and

immunoglobu-lin (Ig)] expression in the primary immune tissues

Materials and methodsSource, maintenance and forms of bacterialstrains

The bacterium L rhamnosus JCM 1136 was obtainedfrom the Japan Collection of Microorganisms (JCM),Institute of Physical and Chemical Research (Riken),Japan, in a FD form Man, Rogosa and Share (MRS)media (De Man, Rogosa & Sharpe 1960) were used torevive the culture for experimentation and to preservefor subsequent use Following mass culture, L rham-nosus was harvested by centrifuging at 16500 g for

10 min and washing with sterile peptone water (NaCl,0.85% and polypeptone, 0.1%) to remove the media.Three di¡erent forms of bacteria were incorpo-rated into the diet either as live spray, a heat-inacti-vated form, or a FD form The viable bacteria were

HK in a continuously stirred water bath at a ture of 75 1C for 60 min The non-viability waschecked by plating on MRS agar three times The FDform was prepared by keeping the bacterial suspen-sion for 60 h at 20 1C in a REL 206 freeze-drier(Kyowa Vacuum Tech., Tokyo, Japan) The FD formwas vacuum packed before preserving at 20 1C un-til further use The FD powder and suspension formwere enumerated for bacterial number per gram andper microlitre of the product respectively

tempera-Diet formulation and probiotic incorporationThe basal diet was formulated by incorporating pre-biotic (10% FOS) components used for feeding the con-trol group of animals FOS are naturally occurringnon-digestible carbohydrates, which, in combinationwith the probionts, promote long-term health via gas-trointestinal immunity The experimental diet was for-mulated with 50% defatted ¢sh meal (DFM) as theprotein source and linseed oil as the lipid source TheDFM was autoclaved to reduce the possible microbialload For the preparation of the diets, the ingredients,

as described in one of our earlier papers (Panigrahi

et al 2005), were mixed mechanically (ACM-50 LAT,Aikohsha, Tokyo, Japan) and sterilized water wasadded before pelletizing (AEZ12 M, Shimadzu, Kyoto,Japan) The pellets were dried in a REL 206 freeze-drierand stored at 20 1C until further use

The live spray (LI) diet was prepared by gentlyspraying a bacterial suspension (100 mL kg 1feed)

of the lactic acid bacterial strain taken on the basaldiet and slowly mixing it in a drum mixer; the ¢nalmix was air dried on a clean bench for 12 h In thesame way, mixing a similar amount of a HK bacterial

Aquaculture Research, 2011, 42, 906^917 Immune gene expressions with probiotic feeding A Panigrahi et al.

suspension (identical density as in diet LI) with the

basal diet, the HK diet was prepared The FD form

was included by freeze-drying 100 mL of bacterial

suspension and mixing with the third diet (FD)

pre-pared from the basal ingredients The stock diet was

stored at 20 1C, and the daily rations were thawed

to 4 1C before feeding The viability of the

incorpo-rated LAB suspension was assessed by vortexing

10 g of diet in 90 mL of peptone water through

plat-ing on tryptic soy agar (TSA; BBL, Becton Dickinson

Cockeysville, MD, USA), MRS and occasionally on BL

blood agar (Nissui, Tokyo, Japan) The colony count

was determined after incubation at 30 1C for 48 h

The bacterial count was nearly 1011CFU g 1for LI

and FD diets, and hence referred to as viable diets

The basal diet without the incorporation of LAB was

taken as the control diet

Experimental design

The feeding experiment was conducted in triplicate

in 60 L tanks arranged in a £ow-through system

Rainbow trout (average weight 126 g), previously fed

on a commercial pelleted feed from Nippon Formula

Feed (Yokohama, Japan) during the early stages and

then on a feed from Chubu Shiryo (Shizuoka, Japan)

during the juvenile phase, was stocked After

condi-tioning with the basal diet for a period of 2 weeks,

they were fed with the control and the three probiotic

diets two times daily to satiation for a period of 30

days The water temperature throughout the

experi-ment averaged 16 1 1C

Microbiological aspects

Bacterial identi¢cation and enumeration

The bacterial identi¢cation was based primarily on

colony and cell morphology, Gram staining and

bio-chemical testing using an API 50 CH strip

(BioMer-ieux, MarcyI’Etoile, France) The inoculation of API

50 CH strips was performed according to the

manu-facturer’s instructions and incubated at 30 1C The

identi¢cation was further con¢rmed through a

spe-cies-speci¢c polymerase chain reaction (PCR)

To con¢rm whether the probiotic strain delivered

through the diet was recovered either from the

intes-tine, faeces or water, the respective cultures were

identi¢ed based on the rRNA genes as described by

Amann, Ludwig and Schleifer (1995) Speci¢c

identi-¢cation of L rhamnosus JCM 1136 was carried out by

direct ampli¢cation of 20^30 colonies randomly lected from the MRS plates within 30^300 colonies.The PCR (Fig 1) was carried out with oligonucleotideprimer, forward primer, (LU-5: CTAGCGGGTGCGACTTTGTT) and reverse primer (Rha II: GCGATGCGAATTTCTATTATT) procured from Invitrogen LifeTechnology (Tokyo, Japan), on a PC 707 thermal cy-cler (ASTEC, Fukuoka, Japan) described in our earlierpaper (Panigrahi et al 2004)

se-Antibacterial propertiesThe antibacterial properties were analysed by theinhibition test using the double-layer method as re-ported by Dopazo , Lemos, Lodeiros, Bolinches, Barja,Toranzo (1988), with some modi¢cations (Sugita, Shi-buya, Shimooka, Deguchi 1996; Sugita, Okano, Suzuki,Iwai, Mizukami, Akiyama, Matsuura 2002) Thetested bacteria were cultured in MRS broth and TSBliquid media at pH 7, and their macrocolonies wereformed on agar plates by inoculating a 5mL droplet

in a multipotent inoculator After incubation at

25 1C for 48 h, the macrocolonies were killed by posing to chloroform vapour for 20 min

ex-The target bacteria (given in Table 2) were cultured

in 1/5 PYBG soft agar (1%), which was preparedwith distilled water instead of 50% seawater (Sugita

et al 2002) and poured into chloroform-killedplates (4.5 mL plate 1

) The plates were incubated at

Figure 1 The molecular identi¢cation of Lactobacillusrhamnosus through 16s rRNA PCR, Sequencing andhomology blast for con¢rmation of their presence in gut,faecal mater, feed and water

Immune gene expressions with probiotic feeding A Panigrahi et al Aquaculture Research, 2011, 42, 906^917

25 1C for 24 h, except for Aeromonas spp., which were

incubated for 3 days at 20 1C The antibacterial

activ-ity against the pathogenic strain was con¢rmed by a

clear inhibitory zone appearing around the probiont

macrocolonies

Immunological methods

Sample preparation

Sampling was scheduled initially and 30 days after

probiotic feeding At each time interval, the sampling

was extended for 3 days From one of the triplicate

tanks of each treatment, three ¢sh were taken

ran-domly each day, and consecutively all three groups

were sampled Thus, a total of nine ¢sh were collected

per treatment The expressions of genes TNF, TGF-b,IFN and Ig were examined in the kidney and spleenisolated from each ¢sh at the end of the feeding period.RNA extraction

The tissues were sonicated in the presence of 1mLTRIzol (Invitrogen, Carlsbad, CA, USA), in order todisrupt the cells and release the RNA The resultingsuspensions were passed repeatedly through a21inch needle to break the genomic DNA It was thenplaced on ice, and 0.2 mL chloroform was added.After vigorous shaking and incubation at roomtemperature for 5 min, the samples were centrifuged

in microtubes at 1600 g at 4 1C for 15 min Thelower phase and white protein interphase werediscarded, while the clear upper phase containingthe RNA was aspirated and placed in a fresh tube

An equal volume of cold isopropanol was added,and the solution was allowed to stand at room tem-perature for 5^10 min before being centrifuged asmentioned above The supernatant was discarded, andthe pellet was washed in1mL of cold 75% diethylpyro-carbonate ethanol (DEPC; Sigma, St Louis, MO, USA)before being centrifuged for another 5 min After the

¢nal wash, the ethanol was removed, and the pelletwas air-dried for 10^15 min before being redissolved

in DEPC-treated water The extracted RNA was diluted

35 times, and 70 mL was used to determine the totalRNA concentration and purity of the samples using aGeneQuant spectrophotometer (Pharmacia Biotech,Freiburg, Germany) The RNAwas stored at 80 1C.Real-time PCR (RT-PCR) analysis

RT-PCR The levels of cytokine mRNA in the biotic fed group were analysed quantitatively usingRT-PCR as described previously (Alimuddin, Kiron,Satoh & Takeuchi 2005) Total RNA was extracted

pro-Table 1 Primers selected for the expression study of selected cytokine genes in the spleen and kidney of rainbow trout Target gene GeneBank Accession no Sequence

R:TGC CAG ATC TTC TCC ATG

R: TCG CCA ATT TTG GAC TCA GC

R: CAA TCA TAT TGG GCA ACC TGC

R: TGT ATC TCA ACT GTG GCT CC

R: CCA CTC ACT CTC ATT GAC TC

Table 2 The antibacterial activities of Lactobacillus

rhamnosus against 13 pathogens symbolically showing the

strength of the activity as observed in the cross-streak method

Pathogenic strains L rhamnosus

Vibriovulnificus RIMD 2219009

Pasteurella piscicida K-III

Listonella anguillarum syn of

V anguillarum

IFO 13266 Aeromonas hydrophila ATCC 7966

Aeromonas caviaei ATCC 15468

Aeromonas sorbia ATCC 43979

Aeromonas veronii ATCC 35624

Escherichia coli HB 101

Test was performed by streaking probionts, followed by test

strain on TSA plates.

Symbols: if the antibacterial ring (gap between the colony and

antibacterial hallow) 5 o1.0 mm, 1 51.0  1o2.0 mm,

11 5 2.0  11o3.0 mm, 111 5  3.0 mm.

Aquaculture Research, 2011, 42, 906^917 Immune gene expressions with probiotic feeding A Panigrahi et al.

from di¡erent kinds of freshly collected tissues cDNA

synthesis was performed as described

(Boonanun-tanasarn, Yoshizaki, Takeuchi, Morita & Takeuchi

2002) Ampli¢cation of the cDNA samples was

per-formed using the iQ SYBR Green Supermix RTPCR

Kit (Bio-Rad Laboratories, Hercules, CA, USA) in an

iCycler Real-Time Detection System (Bio-Rad

Labora-tories) according to the manufacturer’s instructions

DNA Master SYBR Green I mix (containing Taq DNA

polymerase, dNTP, MgCl2 and SYBR Green I dye;

Roche Molecular Biochemicals,Tokyo, Japan) was

in-cubated with theTaqStart antibody for 5 min at room

temperature before adding the primers and the cDNA

template (Table 1) Each reaction (20mL) contained

2mL of the respective cDNA dilution, primers at

0.4mmol L 1and MgCl2atmmol L 1 The

ampli¢ca-tion program consisted of 1 cycle at 95 1C with a 60-s

hold (hot start), followed by 40 cycles at 95 1C with a

0-s hold, speci¢ed annealing temperature (55^60 1C)

with a 5-s hold, a 72 1C cycle with an 18-s hold and

¢nally a speci¢ed acquisition temperature with a 2-s

hold A negative control without a cDNA template

was run with every assay to assess the overall

speci¢-city The primer sequences for the cytokine gene are

given in Table 2 Theb-actin gene expression was

ana-lysed as an internal control by equally loading RNA

with a set primer ‘actin’ forward and ‘actin’ reverse as

mentioned above Data were analysed using the

com-parative cycle threshold (CT) method (Pfa¥ 2001),

where the CT is de¢ned as the cycle number at which

£uorescence reaches a set threshold value

Statistical analysis

Statistical analyses were conducted using the SPSS

11.0 microcomputer software package (SPSS, Chicago,

IL, USA) Analysis of variance was performed using a

one-wayANOVA, followed by Duncan’s test The level of

signi¢cance was set at Po0.05

Results

Identi¢cation and characterization of the probionts

collected from feed, gut and faecal matter was carried

out to attribute any e¡ect to this probiotic bacteria

Isolation and identi¢cation

The biochemical test of Api using API 50 CH strip

(BioMerieux, MarcyI’Etoile, France) and the

molecu-lar identi¢cation con¢rmed that the dietary probiotic

bacterium L rhamnosus JCM 1136 was traced in theintestine and constituted the major proportion ofthe intestinal bacterial load Similarly, this strainwas also isolated from the rearing water as well asthe faecal content

Antibacterial propertiesThe antibacterial property of this probiont againstknown ¢sh pathogens was quanti¢ed by the diameter

of hallow zone (Table 2) The strength of the dal activity was measured as the di¡erence (mm) indiameter between the clear zone and the macrocolony.This probiotic bacterium was found to have antipatho-genic properties against a number of pathogens,which were expressed more against Aeromonas spp

bacterici-Molecular immune response

In primary immune related tissues kidney and spleen,the expression pattern of these cytokines and immunegenes showed di¡erential expression pattern for di¡er-ent forms of probionts as described below

Tumour necrosis factor gene expressionThe TNF gene was found to be highly (Po0.05) up-regulated in both viable form fed groups (LI, FD),whereas there was no di¡erence in the kidney orspleen expression patterns in the HK group (Figs 2

probiont Lactobacillus rhamnosus fed groups as mined by real-time polymerase chain reaction (PCR) CO,control feeding with no probiotic fed group; HK, heat-killed form of probiotic fed group; LI, live spray probioticfed group; FD, freeze-dried probiotic fed groups The PCRtemplate for theb-actin gene was diluted 100-fold Dataare means SD of three individual ¢sh performed in tri-plicate PCR Values with di¡erent letters are statisticallysigni¢cant at (Po0.05) as determined by one-wayANOVA

deter-followed by Duncan’s test

Immune gene expressions with probiotic feeding A Panigrahi et al Aquaculture Research, 2011, 42, 906^917

and 3) In the kidney, there was more than a

5000-fold and a 1500-5000-fold increase in the case of the viable

FD and LI group, respectively, whereas the

up-regula-tion was much lower in the spleen

TGF-b gene expression

The TGF-b gene expression showed a signi¢cantly

higher (Po0.05) up-regulation in case of both viable

form fed groups in the kidney tissue sample (Fig 4)

The HK group even showed signi¢cant up-regulation

(20-fold) in the spleen similar to the FD group

(30-fold); this was not the case for the LI group (Fig 5)

Interferon gene expression

The IFN gene was up-regulated (Po0.05) in all three

treatment groups; although the viable probiont fed

groups showed a strikingly higher up-regulation (up

to 20-fold ) when compared with that of the HK

non-viable form fed group (Figs 6 and 7)

Immunoglobulin gene expressionWith regard to the Ig gene expression pattern, the HKfed group of ¢sh did not show any up-regulationwhen compared with the control group as observed

in both the kidney and the spleen However, both the

Figure 3 Spleen TNFa mRNA level in three forms of

pro-biont Lactobacillus rhamnosus fed groups as determined by

real-time polymerase chain reaction Other details are as

in legend of Fig 2

Figure 4 Kidney TGFb gene mRNA level in three forms

of probiont Lactobacillus rhamnosus fed groups as

deter-mined by real-time polymerase chain reaction (PCR)

Other details are as in legend of Fig 2

Figure 5 SpleenTGFb gene mRNA level in three forms ofprobiont Lactobacillus rhamnosus fed groups as deter-mined by real-time polymerase chain reaction Other de-tails are as in legend of Fig 2

Figure 6 Kidney interferon mRNA level in three forms ofprobiont Lactobacillus rhamnosus fed groups as deter-mined by real-time polymerase chain reaction Other de-tails are as in legend of Fig 2

Figure 7 Spleen interferon mRNA level in three forms ofprobiont Lactobacillus rhamnosus fed groups as deter-mined by real-time PCR Other details are as in legend ofFig 2

Aquaculture Research, 2011, 42, 906^917 Immune gene expressions with probiotic feeding A Panigrahi et al.

viable probiont fed groups showed signi¢cant

up-reg-ulation of Ig gene expression in both test tissues (Figs

8 and 9), with a considerable increase

Further, the fold change in the spleen is much

higher when compared with the kidney with regard

to the TNF,TGFb and IFN gene expressions, whereas

with the Ig gene expression, there is a multifold

change in the kidney as that of the spleen

Discussion

The mechanism of action of probiotic is not clearly

elucidated in aquaculture, although; it is used

exten-sively as an alternative health management tool The

present study evaluated viable and HK LAB in terms

of their impact on the immune tissues at the

molecu-lar level Our previous ¢ndings suggest that L

rham-nosus JCM 1136 has the potential to ameliorate host

health through the modulation of its immune system

(Panigrahi et al 2004, 2005) This strain is known toinduce cellular immune responses as observed withsigni¢cantly higher phagocytic activity Similarly, thesuper oxide anion of the head kidney leucocytes inthe probiotic fed groups was found to be elevatedcompared with that of the control The serum alter-nate complement activity and plasma Ig were theother immune parameters known to be elevated

in the probiotic-fed groups and showed an in£uence

on the humoral immune response Several studiesincluding those in ¢sh showed a correlative linkbetween immune modulation and enhanced protec-tion Nikoskelainen, Ouwehand, Bylund and Salmi-nen (2001), Nikoskelainen, Salminen, Bylund andOuwehand (2001) have studied the probiotic charac-teristics of another strain: ATCC 53101 of L rhamno-sus, which was found to suppress the growth ofpathogenic in ¢sh

Probiotic LAB can signal the immune systemthrough innate cell surface pattern recognitionreceptors (PRR) or via direct lymphoid cell activation

In some cases, this action has been shown to be cient to modulate local- and systemic-level in vivoimmune responses (Cross 2002) The probiotics candirectly or indirectly in£uence the host defencemechanism The interaction between the bacteriaand the host defences can be direct and involves theattachment of gastrointestinal (GI) tract bacteria tothe receptors of host cells It was reported by Johans-son, Molin, Jeppsson, Nobaek, Ahrne and Bengmark(1993) that, when dietary protein induces a patholo-gical state of intestinal permeability in animals, theinclusion of lactobacilli in the diet can induce animmune response and thereby reestablish the inte-stinal mucosal barrier Furthermore, the LAB andother GI bacteria can also in£uence host defencefunctions in an indirect manner by enhancing thehost’s responses to short chain fatty acid and otherbacterial metabolites (Sakata, Ichikawa & Inagaki1999) Dietary carbohydrates such as FOS used inour study may be considered, in part, to be maintain-ing the equilibrium between lectins, secretory Ig andmicroorganisms in the alimentary canal, therebypositively enhancing the health status of the ¢sh.FOS can provide an e¡ective means of promoting bi¢-dobacterium and lactobacillus growth, and arrestthe growth of pathogenic microorganisms selectively(Kaur & Gupta 2002; Roller, Rechkemmer & Watzl2004; Pool-Zobel 2005;Watzl, Girrbach & Roller 2005).Apart from fortifying the host defence mechanism,gut micro£ora plays a key role in the expression ofhost genes that regulate metabolic and physiological

su⁄-Figure 8 Kidney immunoglobulin mRNA level in three

forms of probiont Lactobacillus rhamnosus fed groups as

determined by real-time polymerase chain reaction

Other details are as in legend of Fig 2

Figure 9 Spleen immunoglobulin mRNA level in three

forms of probiont Lactobacillus rhamnosus fed groups as

determined by real-time polymerase chain reaction

Other details are as in legend of Fig 2

Immune gene expressions with probiotic feeding A Panigrahi et al Aquaculture Research, 2011, 42, 906^917

mechanisms Earlier ¢ndings (Panigrahi et al 2007)

on the forms of probionts and the associated

di¡eren-tial cellular and humoral immune response put

for-ward a testable hypothesis that at the molecular

level, the cytokine and immune gene expression

could be a¡ected in the background

Examining the probiont viability as a factor, it has

been demonstrated that viable cultures are more

ef-fective in enhancing certain aspects of immune

func-tion than the dead cultures (Vesely, Negri, Bianchi

Salvadori, Lavezzari & Simone1985; De Simone,

Bian-chi Salvadori, Negri, Ferrazzi, Baldinelli & Vesely

1986) The practical use of probiotics emphasizes that

the microorganisms should be viable for long periods

under storage as well as ¢eld conditions (Fuller 1989)

However, it has also been reported that non-viable

bacteria are able to adhere to tissue culture cells, thus

indicating adhesion without viability (Hood &

Zotto-la 1988; Coconier, Bernet, Chauvie're & Servin 1993)

Adhesion is one of the most important selection

cri-teria for probiotic baccri-teria because it is considered to

be a prerequisite for colonization (Beachey 1981)

Dif-ferent cell wall components like peptidoglycans (PG),

lipopolysaccharides (LPS), b-glucans of Gram

posi-tive, Gram negative and yeast, respectively, are found

to in£uence the immune response in vertebrate and

non-vertebrate systems The innate immune system

is independent of prior exposure to any particular

or-ganism and uses germline-encoded PRR that

recog-nize and bind these conserved pathogen-associated

molecular patterns shared by the major groups of

pathogenic microorganisms (Bols, Brubacher,

Ga-nassin & Lee 2001; Ellis 2001; Medzhitov & Janeway

Jr 2002; Elward & Gasque 2003) but are absent on the

surface of multicellular organisms in eukaryotic

cells For the development, maintenance and proper

functioning of the immune system and thus for

nor-mal health, a snor-mall amount of LPS or PG derived from

the intestinal £ora might be indispensable It is

ra-tional to presume that LPS and PG may be well

throughtout as bacterial-derived vitamins, which

provide a continuous signal to the immune system

to be alert (Rietschel, Brade, Schade, Seydel,

Zhrin-ger, Kusumoto & Brade 1988) Therefore, even though

non-viable in form, they exhibit a biological e¡ect as

re£ected in the up-regulation of the immune genes

studied, especially the TGF-b gene, which is an

important cytokine, responsible for several cellular

immune functions Cell wall PG can be produced

after digestion by lysozyme, present in saliva and

produced by cells in the intestine Itami, Asano,

Tokushige, Kubono, Nakagawa and Takeno et al

(1998) demonstrated the disease-resistant istics of this compound in tiger shrimp These PGfragments can be produced in situ and induce adju-vant activities at the mucosal surface (Link-Amster,Rochat, Saudan, Mignot & Aeschlimann 1994).Whilestudying the immunotherapy of tumours, Bloskma,

character-De Heer,Van Dijk and Willers (1979) observed that able L plantarum exclusively stimulated the delayedhypersensitivity, whereas the HK bacteria had anadjuvant e¡ect on antibody formation Coconnier,Bernet, Kerneis, Chauviere, Fournial, Servin (1993)demonstrated that the HK Lactobacillus acidophilusstrain LB inhibits the cell association with and inva-sion within human intestinal epithelial Caco-2 cells

vi-by enterovirulent bacteria The presence of a speci¢creceptor for peptodoglycan, a cell wall component ofLAB, on lymphocytes and macrophages (Dziarski1991) and the ability of PG to induce the secretion ofIL-1, IL-6 and TNF-a by monocytes (Bhakdi, Klo-nisch, Nuber & Fischer 1991; Tufano, De Lero Cipol-laro, Inneillo, Galdiero & Galdiero 1991; Heumann,Barras, Severin, Glauser & Tomasz 1994) led to simi-lar inferences

Lesser erythrocytes and kidney macrophages,smaller proportions of dead macrophages and fewerphagocytic and lysozyme activities were observedfollowing the use of the formalized preparationscompared with the use of live cells for14 days (Irianto

& Austin 2002) However, the levels of these factorswere still greater than those of the controls It wassuggested that cellular rather than humoral immu-nity is a factor in explaining the bene¢t of these inac-tivated bacterial cell preparations (Irianto & Austin2003) The results of Zhan and Cheers (1995) showedthat killed Listeriae were less e⁄cient than viableones at stimulating TNF-a and IL-1b production.Marked di¡erences were observed in the abilities oflive and HK Brucella abortus and Listeria monocyto-genes organisms to induce the production of TNF-a

by in vitro cultured macrophages Live wild-type teriae induced about 1000-fold-more IL-1b than thekilled Listeriae did while little production of IL-1bwas induced by either live or dead Brucellae In thecurrent study, the live probiont forms induced a mul-tifold up-regulation in the Ig gene in the kidney andspleen and the TNF gene in the spleen, whereas rela-tively lower up-regulation was obtained in the case ofother genes like TGF-b or INF

Lis-These observations are corroborated by a reportthat live but not HK Leishmania amastigotes triggeredTNF-a production from IFN-g-activated macro-phages (Green, Crawford, Hockmeyer, Meltzer &

Aquaculture Research, 2011, 42, 906^917 Immune gene expressions with probiotic feeding A Panigrahi et al.

Nacy 1990) Miettinen, Vuopio-verkila and Varkila

1996 demonstrated that several live LAB but not

glutaraldehyde-¢xed LAB are potent inducers of

proin£amatory cytokine released from human PBMC

and suggested live non-pathogenic LAB as a vaccine

vector or probiotics for the purpose of stimulating

non-speci¢c immunity De Simone et al (1986) have

observed that live bacteria are more potent inducers

of cytokines than ¢xed or dead bacteria This is

further corroborated by an earlier study (Panigrahi

et al 2005) that immune parameters like phagocytic

activity and complements are better induced by live

bacteria The secretions of regulatory cytokines that

in£uence host defences are modulated by bacteria

with host responses depending on the type and

pro-portions of the secreted cytokines (Chen, Isomaki,

Rimpilainen & Toivanen 1999) TNF-a is a key

cytokine in acquired cell-mediated immunity to

in-tracellular bacteria It has an important role in

anti-bacterial resistance (Havell 1989) This di¡erential

induction of cytokines by viable and dead bacteria

may well be the key to their di¡erences in the ability

to induce an immune response

In line with our ¢ndings, the immunological

activ-ity of both live and dead spores from di¡erent strains

of streptomycetes was demonstrated in mouse

macrophages and in human A549 cells (Jussila,

Ruotsalainen, Komulainen, Savolainen, Nevalainen

& Hirvonen 1999) These bacteria also produce a vast

variety of bioactive compounds as their secondary

metabolites, which may have an in£uence on the

im-mune response Although various mechanisms of

ac-tion are proposed, the elaboraac-tion on these aspects

are beyond the scope of our discussion However,

looking at the molecular level, it is clear from our

study that the viable LAB are more e⁄cient in

enhan-cing the immune response in terms of immune gene

expressions in rainbow trout when compared with

the non-viable HK form

Conclusion

The enhanced immunity conferred by L rhamnosus

JCM 1136 at the molecular level, together with

evi-dence from previous reports (Panigrahi et al 2005,

2007) of increased cellular and humoral immune

re-sponse, con¢rms the e¡ectiveness of these probiotics

as a biotherapeutic means This result leads us to

conclude that viable probionts are better than the

non-viable, HK probionts in inducing up-regulation

of immune genes like the TGF-b, TNF-a, IFN and

Ig-genes in the kidney and spleen of rainbow trout.The immunological advantage of the viable formexplains the immune mechanism involved in theprobiotic-induced immunomodulation However, atleast in some cases, even the non-viable, HK formshowed better up-regulation of immune genes.Among all three forms, it appears that the FD probio-tic has the advantage of yielding a better immuneresponse as re£ected by the up-regulation of theimmune genes The mechanisms of microbes asimmune modulators need to be further understood,

so that the microbial components, instead of viablebacteria, could be used as e¡ective probiotics

AcknowledgmentsThis work was partly supported by the JapaneseMinistry of Education, Culture, Sports, Science andTechnology (MEXT) and internal research fundsfor advanced research from the Tokyo University ofMarine Science and Technology

References

Alimuddin, Yoshizaki G., Kiron V., Satoh S & Takeuchi T (2005) Enhancement of EPA and DHA biosynthesis by over-expression of masu salmon D6-desaturase-like gene

in zebra¢sh Transgenic Research 14, 159–165.

Amann R.L., LudwigW & Schleifer K.H (1995) Phylogenetic identi¢cation and in situ detection of individual microbial cells without cultivation Microbiological Reviews 59, 143^169.

Anderson D.P (1992) Immunostimulants, adjuvants and vaccine carriers in ¢sh: application to aquaculture Annual Review of Fish Diseases 2, 281^307.

Balcazar J.L., de Blas I., Ruiz-Zarzuela I., Vendrell D., Calvo A.C., Marquez I., Girones O & Muzquiz J.L (2007) Changes

in intestinal microbiota and humoral immune response following probiotic administration in brown trout (Salmo trutta) British Journal of Nutrition 97, 522^527.

Balcazar J.L.,Vendrell D., de Blas I., Ruiz-Zarzuela I., Girones

O & Muzquiz J.L (2007) In vitro competitive adhesion and production of antagonistic compounds by lactic acid bac- teria against ¢sh pathogens.Veterinary Microbiology 122, 373^380.

Beachey E.H (1981) Bacterial adherence: adhesion-receptor interactions mediating the attachment of bacteria to mu- cosal surfaces Journal of Infectious Diseases 143, 325^345 Bhakdi S., Klonisch T., Nuber P & Fischer W (1991) Stimula- tion of monokine production by lipoteichoic acids Infec- tion and Immunity 59, 4614^4620.

Bloskma N., de Heer E., Van Dijk H & Willers J.M (1979) Adjuvanticity of lactobacilli Di¡erential e¡ect of viable Immune gene expressions with probiotic feeding A Panigrahi et al Aquaculture Research, 2011, 42, 906^917

and killed bacteria Clinical Experimental Immunology 37,

367^375.

Bols N.C., Brubacher J.L., Ganassin R.C & Lee L.E.J (2001)

Ecotoxicology and innate immunity in ¢sh

Developmen-tal and Comparative Immunology 25, 853^873.

Boonanuntanasarn S.,Yoshizaki G.,Takeuchi Y., Morita T &

Takeuchi T (2002) Gene knock-down in rainbow trout

embryos using antisense morpholino

phosphoro-diamidate oligonucleotides Marine Biotechnology 4, 256^

266.

Chen T., Isomaki P., Rimpilainen M & Toivanen P (1999)

Hu-man cytokine responses induced by gram positive cell

walls of normal intestinal microbiota Clinical and

Experi-mental Immunology 118, 261^267.

Coconier M.-H., Bernet M.-F., Chauvie're G & Servin A.L.

(1993) Adhering heat-killed human lactobacillus

acidophi-lus, strain LB, inhibits the process of pathogenicity of

diar-rhoeagenic bacteria in cultured human intestinal cells.

Journal of Diarrhoeal Diseases Research 11, 235^242.

Coconnier M.H., Bernet M.F., Kerneis S., Chauviere G.,

Four-nial J & Servin A.L (1993) Inhibition of adhesion of

enter-oinvasive pathogens to human intestinal Caco-2 cells by

Lactobacillus acidophilus strain LB decreases bacterial

in-vasion FEMS Microbial Letters 110, 299^305.

Cross M.L (2002) Immunoregulation by probiotic

lactoba-cilli: pro-Th1 signals and their relevance to human

health Clinical and Applied Immunology Reviews 3, 115^

125.

De Man J.C., Rogosa M & Sharpe M.E (1960) A medium for

the cultivation of lactobacilli Journal Applied Bacteriology

23, 130^135.

De Simone C., Bianchi Salvadori B., Negri R., Ferrazzi M.,

Baldinelli L & Vesely R (1986) The adjuvant e¡ect of

yo-gurt on production of a-interferon by Con A-stimulated

human peripheral blood lymphocytes Nutrition Reports

International 33, 419^433.

Dopazo C.P., Lemos M.L., Lodeiros C., Bolinches J., Barja J.L.

& Toranzo A.E (1988) Inhibitory activity of

antibiotic-pro-ducing marine bacteria against ¢sh pathogens Journal of

Applied Bacteriology 65, 97^101.

Dziarski R (1991) Demonstration of peptidoglycan binding sites

on lymphocytes and macrophages by photo a⁄nity

cross-linking Journal of Biological Chemistry 266, 4713^4718.

Ellis A.E (2001) Innate host defense mechanisms of ¢sh

against viruses and bacteria Developmental and

Compara-tive Immunology 25, 827^839.

Elward K & Gasque P (2003) ‘Eat me’ and ‘don’t eat me’

sig-nals govern the innate immune response and tissue

re-pair in the CNS: emphasis on the critical role of the

complement system Molecular Immunology 40, 85^94.

Fuller R (1989) Probiotics in man and animals Journal of

Applied Bacteriology 66, 365^378.

Gatesoupe F.J (1991) The e¡ect of three strains of lactic acid

bacteria on the production rate of rotifers, Brachionus

pli-catilis and their dietary value for for larval turbot,

Scopthalmus maximus Aquaculture 96, 335^342.

Gatesoupe F.J (1994) Lactic acid bacteria increase the tance of turbot larvae, Scopthalmus maximus against pathogenicVibrio Aquatic Living Resources 7, 182^277 Gatesoupe F.J (1999) The use of probiotics in aquaculture Aquaculture 180, 147^165.

resis-Gomes A.M.P & Malcata F.X (1999) Bi¢dobacteria spp and Lactobacillus acidophilus biological, biochemical, techno- logical and therapeutic properties relevant for use as pro- biotics Trends Food Science and Technology 10, 139^157 Green S.J., Crawford R.M., Hockmeyer J.T., Meltzer M.S & Nacy C.A (1990) Leishmania major amastigotes initiate the L-arginine-dependent killing mechanism in IFN-g- stimulated macrophages by induction of tumor necrosis factor-a Journal of Immunology 145, 4290^4297 Havell E.A (1989) Evidence that tumour necrosis factor has

an important role in antibacterial resistance Journal of Immunology 143, 2894–2899.

Heumann D., Barras C., Severin A., Glauser M.P & Tomasz

A (1994) Gram-positive cell walls stimulate synthesis of tumor necrosis factor alpha and interleukin-6 by human monocytes Infection and Immunity 62, 2715^2721 Hood S.K & Zottola E.A (1988) E¡ect of low pH on the ability of Lactobacillus acidophilus to survive and adhere

to human intestinal cells Journal of Food Science 53, 1514^1516.

Irianto A & Austin B (2002) Probiotics in aquaculture nal of Fish Diseases 25, 633^642.

Jour-Irianto A & Austin B (2003) Use of dead probiotic cells to control furunculosis in rainbow trout, Oncorhynchus my- kiss (Walbaum) Journal of Fish Diseases 26, 59^62 Itami T., Asano M., Tokushige K., Kubono K., Nakagawa A., Takeno N., Nishimura H., Maeda M., Kondo M & Takahashi Y (1998) Enhancement of disease resistance of kuruma shrimp, Penaeus japonicus, after oral administra- tion of peptidoglycan derived from Bi¢dobacterium thermophilium Aquaculture 164, 277–288.

Johansson M.L., Molin G., Jeppsson B., Nobaek S., Ahrne S & Bengmark S (1993) Administration of di¡erent Lactobacil- lus strains in formented oatmeal soup: in vivo coloniza- tion of human intestinal mucosa and e¡ect on the indigenous £ora Applied and Environmental Microbiology

59, 15^20.

Jussila J., Ruotsalainen M., Komulainen H., Savolainen K., Nevalainen A & Hirvonen M.R (1999) Streptomyces anulatus from indoor air of moldy houses induce NO and IL-6 production in a human alveolar epithelial cell-line Environmental Toxicology and Pharmacology 7, 261^266.

Kaur N & Gupta A.K (2002) Applications of inulin and gofructose in health and nutrition Journal of Bioscience 27,703^714.

oli-Link-Amster H., Rochat F., Saudan K.Y., Mignot O & mann J.M (1994) Modulation of a speci¢c humoral im- mune response and changes in intestinal £ora mediated through fermented milk intake FEMS Immunology and Medical Microbiology 10, 55^64.

Aeschli-Aquaculture Research, 2011, 42, 906^917 Immune gene expressions with probiotic feeding A Panigrahi et al.

Medzhitov R & Janeway C.A Jr (2002) Decoding the

pat-terns of self and nonself by the innate immune system.

Science 296, 298^300.

Miettinen M., Vuopio-verkila J & Varkila K (1996)

Produc-tion of human Tumor Necrosis Factor Alpha,

Interleu-kin-6 and Interleukin-10 is induced by lactic acid

bacteria I Infection and Immunity 64, 5403^5405.

Moriarty D.J.W (1998) Control of luminous Vibrio species in

penaeid aquaculture ponds Aquaculture 164, 351^358.

Nikoskelainen S., Ouwehand A.C., Bylund G & Salminen S.

(2001) Protection of rainbow trout (Oncorhynchus mykiss)

from furunculosis by Lactobacillus rhamnosus

Aquacul-ture 198, 229^236.

Nikoskelainen S., Salminen S., Bylund G & Ouwehand A.C.

(2001) Characterization of the properties of human and

dairy derived probiotics for prevention of infectious

dis-eases in ¢sh Applied and Environmental Microbiology 67,

2430^2435.

Nikoskelainen S., Ouwehand A.C., Bylund G., Salminen S &

Lilius E (2003) Immune enhancement in rainbow trout

(Oncorhynchus mykiss) by potential probiotic bacteria

(Lactobacillus rhamnosus) Fish Shell¢sh Immunology 15,

443^452.

Noh S.H., Han K., Won T.H & Choi Y.J (1994) E¡ect of

anti-biotics, enzyme, yeast culture and probiotics on growth

performance of Israeli carp Korean Journal of Animal

Science 36, 480^486.

Panigrahi A., Kiron V., Kobayashi T., Puangkaew J., Satoh S.

& Sugita H (2004) Immune responses in rainbow trout

Oncorhynchus mykiss induced by a potential probiotic

bacteria Lactobacillus rhamnosus JCM 1136.Veterinary

Im-munology and Immunopathology 102, 379^388.

Panigrahi A., Kiron V., Puangkaew J., Kobayashi T., Satoh S.

& Sugita H (2005) The viability of probiotic bacteria as a

factor in£uencing the immune response in rainbow trout

Oncorhynchus mykiss Aquaculture 243, 241^254.

Panigrahi A., Kiron V., Satoh S., Hirono I., Kobayashi T.,

Sugita H., Puangkaewa J & Aoki T (2007) Immune

modulation 947 and expression of cytokine genes in

rainbow trout Oncorhynchus mykiss upon probiotic

feed-ing Developmental and Comparative Immunology 31,

372^382.

Pfa¥ M.W (2001) A new mathematical model for relative

quanti¢cation in real-time RT-PCR Nucleic Acids Research

29, 36^45.

Pirarat N., Kobayashi T., Katagiri T., Maita M & Endo M.

(2006) Protective e¡ects and mechanisms of a probiotic

bacterium Lactobacillus rhamnosus against experimental

Edwardsiella tarda infection in tilapia (Oreochromis

niloti-cus) Veterinary Immunology and Immunopathology 113,

339^347.

Pool-Zobel B.L (2005) Inulin-type fructans and reduction in

colon cancer risk: review of experimental and human

data British Journal of Nutrition 93,73^90.

Rengpipat S., PhianpakW., Piyatiratitivorakul S &

Menasve-ta P (1998) E¡ect of a probiotic bacterium on black tiger

shrimp Penaeus monodon survival and growth ture 167, 301^313.

Aquacul-Rietschel E.T., Brade L., Schade U., Seydel U., Zhringer U., Kusumoto S & Brade H (1988) Bacterial endotoxin: Prop- erties and structure of biologically active domains In: Surface Structure of Microorganisms and Their Interaction with Mammalian Host (ed by E Schriner, M.H Richmon,

G Seibert & U Schwarz), pp 1^41 Verlag Chemie, heim, Germany.

Wein-Ringo E & Gatesoupe F.J (1998) Lactic acid bacteria in ¢sh: a review Aquaculture 160, 177^203.

Robertson N.M., Reijula K., Jarvis B.B.,Veijalainen P & tikka E.L (1997) E¡ects of intranasal exposure to spores of Stachybotrys atra in mice Fundamental and Applied Toxi- cology 35, 182^188.

Hin-Robertson P.A.W., O’Dowd C., Burrells C.,Williams P & tin B (2000) Use of 972 Carnobacterium sp as a probiotic for Atlantic salmon 973 Salmo salar L and rainbow trout Oncorhynchus mykiss Aquaculture 185, 235–243 Roller M., Rechkemmer G & Watzl B (2004) Prebiotic inulin enriched with oligofructose in combination with the pro- biotics Lactobacillus rhamnosus and Bi¢dobacterium lactis modulates intestinal immune functions in rats Nutri- tional Immunology 134, 153^156.

Aus-Ronnestad I., Thorsen A & Finn R.N (1999) Fish larval trition: a review of recent advances in the roles of amino acids Aquaculture 177, 201^216.

nu-Sakata T., Ichikawa H & Inagaki A (1999) In£uence of lactic acid, uccinic acid and ammonia on epithelial cell prolif- eration and motility of the large bowel Asia Paci¢cJournal

of Clinical Nutrition 8, 9^13.

Salinas I., Myklebust R., Esteban M.A., Olsen R.E., Meseguer J & Ringo E (2008) In vitro studies of Lactoba- cillus delbrueckii subsp lactis in Atlantic salmon (Salmo salar L.) foregut: tissue responses and evidence of pro- tection against Aeromonas salmonicida subsp salmoni- cida epithelial damage Veterinary Microbiology 128, 167^177.

Skjermo J & Vadstein O (1999) Techniques for microbial control in the intensive rearing of marine larvae Aquacul- ture 177, 333^343.

Sugita H., Shibuya K., Shimooka H & Deguchi Y (1996) Antibacterial abilities of intestinal bacteria in freshwater cultured ¢sh Aquaculture 145, 195^203.

Sugita H., Okano R., Suzuki Y., Iwai D., Mizukami M.,

Akiya-ma N & Matsuura S (2002) Antibacterial abilities of testinal bacteria from larval and juvenile Japanese

in-£ounder against ¢sh pathogens Fisheries Science 68, 1004^1011.

Tufano M.A., De Lero Cipollaro G., Inneillo R., Galdiero M & Galdiero F (1991) Protein A and other surface component

of Staphylococcus aureus stimulate production of 1a,

IL-4, IL-6 and TNF and IFN-g European Cytokine Network 2, 361^366.

Vadstein O., Oie G., Olsen Y., Salvesen I., Skjermo J & Braek G (1993) A strategy to obtain microbial control dur- Immune gene expressions with probiotic feeding A Panigrahi et al Aquaculture Research, 2011, 42, 906^917

Skjak-ing larval development of marine ¢sh In: ProceedSkjak-ings of the

First International Conference on Fish Farming Technology

(ed by H Reinertsen, L.A Dahle, L Jrgensen & K

Tvin-nereim), pp.69^75 Balkema, Rotterdam, the Netherlands.

Vershuere L., Rombaut G., Sorgeloos P & Verstraete W.

(2000) Probiotic bacteria as biological control agents in

Aquaculture Microbiology and Molecular Biology Review

64, 655^671.

Vesely R., Negri R., Bianchi Salvadori B.B., Lavezzari D.D &

Simone C (1985) In£uence of a diet additioned with

yogurt on the immune system EOS Journal of Immunology and Immunopharmacology 1, 30^35.

Watzl B., Girrbach S & Roller M (2005) Inulin, oligofructose and immunomodulation British Journal of Nutrition 93, 49^55.

Zhan Y & Cheers C (1995) Di¡erential induction of phage-derived cytokines by live and dead intracellular bacteria in vitro Infection and Immunity 63,720^723 Aquaculture Research, 2011, 42, 906^917 Immune gene expressions with probiotic feeding A Panigrahi et al.

macro-Water quality in a serial-use raceway and its effect on

Linnaeus, 1758

Matthew A Naylor, Horst Kaiser & Cli¡ord L W Jones

Department of Ichthyology and Fisheries Science, Rhodes University, Grahamstown, South Africa

Correspondence: H Kaiser, Department of Ichthyology and Fisheries Science, Rhodes University, PO Box 94, Grahamstown, 6140, South Africa E-mail: h.kaiser@ru.ac.za

Abstract

To quantify the e¡ects of serial-use of water on

aba-lone growth and feed conversion, this study describes

water quality in a serial-use raceway with seven

passes A £ow index of 7.2^9.0 L h 1kg 1was

esti-mated as the minimum value at which to grow

60^70 mm Haliotis midae, as weight gain (analysis

of variance; F6,14513.9, Po0.0001) and feed

conver-sion ratio (Kruskal^Wallis test; H6, 21516.3, P 5 0.012)

were signi¢cantly reduced at lower values pH and

dissolved oxygen concentration were positively

correlated with the £ow index (pH, r250.99;

Po0.001; dissolved oxygen, r2

50.99; Po0.001),while free ammonia nitrogen (FAN) and nitrite

were negatively correlated with the £ow index (FAN,

r250.99, Po0.001; Nitrite, r250.93, Po0.001) The

concentration of nitrite increased throughout the

experiment and may re£ect the colonization of

Nitrosomonas bacteria as water re-use increased

Based on comparisons with growth and toxicity tests,

it is suggested that low pH combined with

growth-lim-iting levels of FAN were the ¢rst variables limgrowth-lim-iting

aba-lone growth in the serial-use raceway

Keywords: £ow index, ammonia, pH, nitri¢cation,

e¡ective concentration

Introduction

Most abalone farmers use £ow-through systems to

provide oxygen and dilute growth-reducing

metabo-lites To intensify production from the available water

(Piedrahita 2003; Schuenho¡, Shpigel, Lupatsch,Ashkenazi, Msuya & Neori 2003), farms may use re-circulating systems or implement serial-use of water

In a serial-use raceway, the water £ow remains stant while the cumulative biomass of the farmed an-imal increases with increasing water re-use Thebuild-up of metabolic waste degrades water quality

con-as it pcon-asses from one tank to the next (Morrison & per1988; Colt & Orwicz1991a; Summerfelt, Davidson,Waldrop, Tsukuda & Bebak-Williams 2004) A serial-use system can provide an opportunity to re-usewater as long as the water quality does not reducegrowth or compromise the health of the aquatic ani-mal (Poxton & Allouse 1982; Colt, Watten & Rust2009) From a research perspective, the gradients inwater quality that develop between tanks in seriesprovide an opportunity to monitor growth under arange of di¡erent water quality conditions

Pi-Water quality in intensive aquaculture is enced by the population density of the aquatic spe-cies and water £ow (Harris, Maguire & Hindrum1998; Colt et al 2009) Thus, the e¡ect of the £ow in-dex (L h 1kg 1) on the growth and health of aquaticspecies in serial-use systems needs to be determined

in£u-in order to improve production e⁄ciency (Fivelstad1988; Colt & Orwicz 1991a) This has not been donefor the South African species Haliotis midae underfarm conditions Low £ow indices may cause growthand health problems, while high £ow indices implyine⁄cient use of water Determining the ¢rst limitingwater quality variable for growth and health for aparticular system is imperative for re¢ning its designand management and therefore increasing its water

Aquaculture Research, 2011, 42, 918^930 doi:10.1111/j.1365-2109.2010.02660.x

re-use capabilities In general, oxygen has been

con-sidered to be the ¢rst limiting variable in serial-use

raceways (Colt & Orwicz 1991a) However, with

im-provements in aeration systems and the

supplemen-tation of pure oxygen, high ammonia and carbon

dioxide concentrations may become the primary

lim-iting variables (Colt et al 2009) First limlim-iting

vari-ables may also be system speci¢c, i.e., due to aspects

such as aeration type (Colt & Orwicz 1991b; Eshchar,

Mozes & Fediuk 2003), weir type (Colt & Orwicz

1991b; Colt et al 2009), feed (Vergara, Lopez-Calero,

Robaina, Caballero, Montero, Izquierdo & Aksnes

1999) and the species cultured (Qian,Wu & Ni 2001)

There is a paucity of information on the e¡ects of

se-rial-use of water on abalone growth, feed conversion

and health The tolerance of abalone to selected water

quality characteristics has been determined in toxicity

experiments and in studies quantifying the e¡ective

concentration (EC5) of a water quality variable that

re-duces growth by 5% For example, the e¡ect of chronic

exposure to free ammonia nitrogen (FAN) (NH3) has

been studied in Haliotis tuberculata L (Basuyaux &

Mathieu 1999), Haliotis laevigata (Donovan) (EC55

41mg L 1and EC505158mg L 1respectively) (Harris

et al 1998), Haliotis rubra (Leach) (EC554^6mg L 1)

(Huchette, Koh & Day 2003) and H midae (36 h LC50

in 1^2 cm juveniles 5 9.8, 12.9mg L 1in 5^8 cm

cock-tail size abalone and 16.4mg L 1in 10^15 cm abalone

broodstock) (Reddy-Lopata, Auerswald & Cook 2006)

The growth-reducing e¡ects of low dissolved

oxygen levels were determined for H laevigata

(EC557.36 mg L 1and EC5055.91mg L 1) (Harris,

Maguire, Edwards & Johns1999) and Haliotis

diversico-lor supertexta (Reeve) (Cheng, Li & Chen 2004), and the

importance of nitrite (NO2 ) was assessed in H

tuber-culata (safe concentration o5 mg L 1

) (Basuyaux &

Mathieu 1999) and H laevigata (67% reduction in the

speci¢c growth rate at40.56 mg NO2-N L 1) (Harris,

Maguire, Edwards & Hindrum 1997) The e¡ect of pH

on H laevigata (EC557.78 and EC5057.39) and H

ru-bra (EC557.93 and EC5057.37) has been determined

by Harris, Maguire, Edwards and Hindrum (1999)

Most research addressed the e¡ects of single factors

while maintaining acceptable water quality In

produc-tion systems, however, water quality changes in a

mul-tifactorial way as a result of the metabolic activity of

the animal and bacterial activity in the water Thus,

¢ndings from toxicity tests may not be applicable when

estimating the e¡ect of water quality on the growth

and health of aquatic animals (Colt 2006) As

applic-able data are not readily availapplic-able, water quality

known to a¡ect abalone can be used as a guideline for

system design, but predictions should be tested inon-farm growth experiments This study describesthe changes in the water quality between tanks in aserial-use raceway system with seven passes in relation

to the cumulative biomass and water £ow andestimates the £ow index (L h 1kg 1) at whichgrowth is reduced in 60^70 mm H midae By compar-ing the water quality in tanks where abalone growthwas reduced with data from growth and toxicity ex-periments on this and other abalone species, this studyidenti¢es and estimates the ¢rst limiting water qualityvariable in the serial-use system

Methods and materialsThe study was conducted at HIK Abalone Farm inHermanus, on the south west coast of South Africa(34126004.3500S; 19113012.5100E), from September toDecember 2008 The abalone were approximately 3years old and 64.2 2.6 mm in length at the begin-ning of the experiment They had been spawned inthe hatchery and weaned onto the arti¢cial diet Ab-feeds(Marifeed, Hermanus, South Africa) They hadbeen reared in well-aerated, £ow-through tanks atambient temperatures within the temperature range

of 12^20 1C, which has been identi¢ed as suitable forthis species (Britz, Hecht & Mangold 1997)

Experimental systemThree serial-use raceways each containing seventanks in series were used (Fig 1) The tanks(0.9 0.6  0.6 m) were set up at varying heights toallow for gravitational water £ow Filtered seawater(100mm) entered the top tank of each raceway (i.e.,position 1) supplied from a 0.3 m3header tank and

£owed through the remaining six tanks (i.e., tions 2^7, Table 2) The header tank allowed for aconstant water £ow to each raceway, as the tidalheight could have a¡ected the volume of waterpumped from the sea Polyvinylchloride (PVC) piping(40 mm outer diameter) was used for in£uent and ef-

posi-£uent pipes between tanks During a 5-day tion period, additional piping provided seawater at anindustry standard £ow rate of 116 L h 1to each indi-vidual tank Each tank held one ‘oyster-mesh’ basket,which contained vertical plastic racks providing asurface area of 3.2 m2 Feed was placed onto a horizon-tal, plastic feeder plate on top of each vertical rack, po-sitioned 10 cm below the water surface Aeration wasprovided to each tank via 20 mm PVC airlines placed

acclima-Aquaculture Research, 2011, 42, 918^930 Water quality in an abalone serial-use raceway M A Naylor et al.

horizontally 50 mm above the tank bottom The three

serial-use raceways were shaded from direct sunlight

using a densely meshed shade cloth, and the presence

of the feeder plate at the water surface

Experimental design

As the total abalone biomass per unit water £ow

in-creased with an increase in serial use, cumulative

biomass was related to the £ow rate to obtain a £ow

index in L h 1kg 1(Table1) At the beginning of the

experiment, 50 abalone per tank were weighed

indi-vidually to the nearest 0.01g using an electronic

bal-ance (Snowrex BBA-600, Snowrex International,

Taipei, Taiwan, R.O.C) and measured to the nearest

0.1mm from photographs using the software

pro-gramme,SIGMASCANSPRO 5(Systat Software, San Jose,

CA, USA) Length measurements were taken along

the long shell axis and photographs were calibrated

based on length measurements using Vernier

calli-pers Photographing the abalone reduced the

hand-ling time and the duration for which abalone were

out of the water Each of the 21baskets was ¢lled with

8.1 0.1kg (the industry stocking density for this

size class chosen by farmers) of abalone and allocated

randomly to a tank position The number of abalone

in each basket ranged from 163 to 169 There was no

signi¢cant di¡erence in the initial average weight tween all baskets (F20,102951.38, P 5 0.12) Abalonewere fed Abfeed S34s(composition in g kg 1: 340 gprotein, 530 g carbohydrates, 1.2 g lipids) to apparentsatiation at 16:00 hours daily from a bucket assigned

be-to each tank Food was given be-to tanks that hado15pellets left on the feeder plate using a cup that con-tained 64 2.8 g of food This feeding method wasdeveloped by farm management using research byBritz et al (1997) to be practical and minimize foodwaste The amount of feed given over a 2-week periodwas calculated by subtracting the end weight (bucketand remaining food) from the full bucket weight atthe start of that period Tanks were cleaned every 14days This involved moving the baskets to an emptyfarm tank, after which the three serial-use racewayscould be drained, scrubbed clean and re¢lled Thebaskets were then returned to the same position inthe serial-use raceways The abalone were grown inthe serial-use raceways for 101 days (normal gradinginterval for this size class), after which the weighingand measuring process was repeated to determinethe average wet weight gain, length gain and condi-tion factor (CF) of abalone in each tank Conditionfactor was determined using the equation:

CF 5 5575 (weigh length 2.99) (Britz 1996) As eaten food was not removed each day, the apparentfeed conversion ratio (FCR) for baskets in each tank

un-Figure 1 The three serial-use raceways used to grow South African abalone, Haliotis midae, each consisting of seventanks in series

Water quality in an abalone serial-use raceway M A Naylor et al Aquaculture Research, 2011, 42, 918^930

position was calculated as: FCR 5 (dry weight feed

gi-ven/biomass gain), where biomass gain is the

di¡er-ence in the wet weight of abalone within a basket

between the start and the end of the experimental

period

Water quality analysis

Temperature, water pH, dissolved oxygen

concentra-tion, percentage oxygen saturation and the

concen-tration of total ammonia nitrogen (TAN), suspended

solid and nitrite-N (NO2-N) were monitored at the

out£ow of each tank twice a week Samples were

taken between 09:00 and 10:00 hours, to achieve

the most reliable estimate of daily means, based on

the results obtained by Yearsley (2008) for diurnal

water quality conditions in abalone tanks on this

farm Water £ow measurements were performed in

duplicate for each raceway at each sampling time,

using a 2 L graduated container and a stopwatch

Water £ow into the ¢rst tank of each raceway

averaged 288.8 1.45 L h 1 Temperature and pH

were measured using a pH meter (YSI Model # 60/

10 FT; Yellow Springs, OH, USA) The pH meter was

calibrated once a week using bu¡er solutions of pH

4, 7 and 10 Dissolved oxygen concentration and

oxy-gen saturation were measured using an oxyoxy-gen

me-ter (YSI Model # 55D) The oxygen meme-ter was

calibrated before each recording in air-saturated

seawater, and the membrane of the electrode was

re-placed every 3 weeks TAN and nitrite concentrations

were determined using the method of Solorzano

(1969) and the Merck Nitrite Test Kit (Cat no

1.14776.0001, Merck, Modderfontein, South Africa),

respectively, with colour absorbance read through a

spectrophotometer (Prim Light, Secomam, Ales,

France) Absorbance was converted into the

concen-tration of TAN or nitrite using the coe⁄cients derivedfrom least-square linear regression standard curves(TAN, n 515, r250.998 using ammonium chloride

to make up the concentration range, and nitrite,

n 510, r250.994, using sodium nitrite) Total nia nitrogen and nitrite samples were collected inacid-washed glassware, processed immediately andkept dark after adding the reagents The concentra-tion of FAN was calculated using the values for TAN,temperature, pH and salinity of the respective watersamples (Bower & Bidwell 1978) The concentration

ammo-of suspended solids 410 mm was determined by

¢ltering a 2 L water sample through a 10mm ¢ltermesh The ¢ltrate was washed with distilled wateronto a pre-dried and weighed ¢lter paper [Munktell,Filtrak 1291, 2^5mm typical pore size (Brenstein,Germany)] The ¢lter papers were re-dried at 60 1Cfor 24 h and weighed to the nearest 0.0001g using amicro-balance (Denver Instruments, NY, USA) Theconcentration of suspended solids (mg L 1) was cal-culated as the dry weight gain of the ¢lter paper (mg)divide\ by the volume of the water sample (L)

Statistical analysisAnalysis of variance (ANOVA) was used to compare theaverages of the dependent variables wet weight gain,length gain, FCR and CF between tank positions, i.e.,between the levels of water re-use Assumptions ofequality of variance and normality of residuals werechecked using Levene’s test (Levene 1960) and theShapiro^Wilk test (Shapiro & Wilk 1965)respectively.Post hoc di¡erences between means of the dependentvariables were compared between tank positionsusing Tukey’s HSD test The non-parametricKruskall^Wallis test was used when assumptions forANOVAtesting could not be met Water quality and

Table 1 Initial and ¢nal cumulative biomass (kg), initial and ¢nal £ow indices (L h 1kg 1) and the mean water £ow rate (L h  1 ) at each tank position in three abalone serial-use raceways

Tank positions

Initial cumulative biomass (kg) 8.2  0.17 16.2  0.15 24.3  0.17 32.3  0.15 40.4  0.1 48.6  0.21 56.6  0.15 Final cumulative biomass (kg) 10.3  0.41 20.1  0.45 30.2  0.31 40.1  0.41 49.7  0.45 59.2  0.17 68.4  0.49 Mean water flow (L h  1 ) 288.8  1.45 288.8  1.45 288.8  1.45 288.8  1.45 288.8  1.45 288.8  1.45 288.8  1.45 Initial flow index (L h  1 kg  1 ) 34.3  1.38 17.8  0.2 11.9  0.11 8.9  0.1 7.1  0.06 6.0  0.06 5.1 Final flow index (L h  1 kg  1 ) 28.0  1.01 14.4  0.3 9.6  0.11 7.2  0.1 5.8 4.9 4.2

Tank position represents the degree of water re-use, i.e., 1 5the ¢rst tank and 7 5 the last tank in series Values are averages  dard deviation (n 5 3).

stan-Aquaculture Research, 2011, 42, 918^930 Water quality in an abalone serial-use raceway M A Naylor et al.

growth data were modelled as a function of tank

position and the £ow index using least-square

regres-sion analysis When making comparisons between

possible regression models, the highest t-statistic,

the P-value to test for the signi¢cance of coe⁄cients,

F-statistic and the distribution of residuals were used

as criteria to decide on the preferred model An error

level of 5% (Po0.05) was used for all the tests.Values

in this paper are reported as mean standard

devia-tion (Table 2)

Results

The average wet weight gain (g abalone 1) di¡ered

depending on the level of serial-use (ANOVA;

F6,14513.9, P 5 0.00003), i.e., between some tank

positions in the raceway.Wet weight gain was

signi¢-cantly lower in tank positions 6 and 7 than that in

positions 1^4 (post hoc test) Abalone in position 1

had a signi¢cantly higher average weight gain than

those in position 5 Wet weight gain decreased with

increasing serial use and was positively correlatedwith the mean £ow index (L h 1kg 1) (least-squareregression analysis, r250.72; Po0.001; Fig 2) Nosigni¢cant di¡erences were observed betweentank positions in the mean length gain (mm)

Tank position 4

y = 2.6x−0.29

Figure 2 Mean weight gain (g abalone 1) and food conversion ratio in three replicate serial-use raceways as a function

of the tank position and the average £ow index

Table 2 Average concentration (  standard deviation) of water quality variables sampled from the seawater entering the serial use abalone culture system via a header tank Water quality variable Average Range

Dissolved oxygen (mg L  1 ) 8.45  0.35 7.89–9.31 Percentage saturation of oxygen 104  3.7 97.3–115.3 Total ammonia nitrogen

[TAN; NH 41(mg L 1)]

8.73  5.14 0.60–18.6 Free ammonia nitrogen

[FAN; NH 3 (mg L 1)]

0.19  0.14 0.01–0.45 Nitrite, NO 2 -N (mg L 1) 6.11  2.09 2.10–11.40 Suspended solids (mg L  1 ) 6.06  2.97 1.34–15.63

Water quality in an abalone serial-use raceway M A Naylor et al Aquaculture Research, 2011, 42, 918^930

(ANOVA; F6,1451.84, P 5 0.164) In position 1, length

gain averaged 6.7 0.7 and 5.1  0.5 mm in position

7 No dead abalone were found in any of the tanks

The average CF was signi¢cantly lower in tank

po-sitions 6 and 7 than that in popo-sitions 1^4 (ANOVA;

F6,1457.19, P 5 0.0012) The average FCR di¡ered

sig-ni¢cantly between tank positions (Kruskal^Wallis

test; H6, 21516.3, P 5 0.012) Feed conversion ratio

in-creased as a function of tank position (r250.60,

Po0.001) (Fig 2), with an average FCR of 1.1  0.2

in position 1 and 1.8 0.5 in position 7 Feed

conver-sion ratio was negatively correlated with the £ow

in-dex (r250.50, Po0.001) (Fig 2)

For the ¢rst 2 weeks of the experiment, the average

feed consumption in position 2 (209 2.6 g) was 29%

lower than the average for all tanks (306 33.9 g),

and was signi¢cantly lower (Kruskal^Wallis test:

H6, 21514.2, P 5 0.027) than that for tanks in position

3 (348 6.1g) From week 3 onwards, there were no

more signi¢cant di¡erences in bi-weekly feed

con-sumption between tank positions (period 2: H 512.1,

P 5 0.059; period 3: H 5 8.7, P 5 0.192; period 4:

H 5 4.3, P 5 0.630; period 5: H 5 3.9, P 5 0.693; period

6: H 510.0, P 5 0.125; period 7: H 5 8.1, P 5 0.230;

period 8: H 5 9.1, P 5 0.166)

The results suggest that temperature was linearly

related to the tank position (linear regression,

r250.71, Po0.001).The average temperature in

posi-tion1was15.2 1.2 1C, increasing to15.5  1.3 1C in

position 7 The suspended solid concentration

in-creased from 4.2 2.0 mg L 1 in position 1 to4.8 2.5 mg L 1in position 7

Repeated water use, i.e decreased £ow index duced water pH (r250.99, Po0.001; Fig 3) The pHaveraged 7.86 (range: 8.02^7.73) in position 1 and7.60 (range:7.75^7.45) in position 7 (Table 3) The dis-solved oxygen concentration was positively corre-lated with the £ow index (r250.99, Po0.001; Fig 3).Percentage oxygen saturation displayed a trend simi-lar to dissolved oxygen, being positively correlated tothe £ow index (r250.99, Po0.001) The average per-centage oxygen saturation decreased from 99.42.5 in position 1 to 91.6 3.1 in position 7 The aver-age TAN and FAN concentrations were positively cor-related with the tank position (r250.99, Po0.001;Fig 3) and negatively correlated with the £owindex (r250.99, Po0.001) The mean TAN was40.7 20.58 mg L 1 in position 1 and 298.1138.9mg L 1in position 7 The FAN concentrationwas the highest in position 7 (2.69 1.32 mg L 1).The mean NO2-N concentration increased exponen-tially from the in£ow to the out£ow of the serial-useraceways (r250.94, Po0.001) and was negativelycorrelated with the £ow index (r250.93, Po0.001;Fig 3) Nitrite concentrations increased with time,with this increase being most pronounced in the low-

re-er tank positions (Fig 4) Ovre-er the course of the periment, there were cyclical variations in the pH,dissolved oxygen concentration and the concentra-tion of FAN (Fig 4)

ex-Table 3 Basic statistics for selected water quality variables from water samples taken in tank positions 1^4 and 5^7 Positions Mean  SD Range

Averages were calculated on the basis that abalone growth was signi¢cantly reduced from tank position 5 onwards.

Aquaculture Research, 2011, 42, 918^930 Water quality in an abalone serial-use raceway M A Naylor et al.

Abalone growth

Water quality in the serial-use raceway became

de-graded progressively due to the metabolic waste from

the abalone, thereby reducing weight gain in the

bas-kets at the end of the serial-use system Atlantic

sal-mon (Salmo salar L.) ¢ngerlings raised in a six-tier

serial-use raceway showed a signi¢cant reduction ingrowth after three serial uses (Morrison & Piper1988) Fivelstad, Bergheim and Tyvold (1991) andHuchette et al (2003) suggested establishing the £owrate per unit biomass at which growth reduction oc-curs in S salar and H rubrarespectively As growthwas not signi¢cantly reduced within the ¢rst four se-rial passes, 7.2^9.0 L h 1kg 1was the lowest £ow

Tank position 7.2

race-Water quality in an abalone serial-use raceway M A Naylor et al Aquaculture Research, 2011, 42, 918^930

index required to prevent a growth reduction in 50^

60 g H midae (60^70 mm) in the serial-use raceway

used in this study Yearsley (2008) recommended

£ow indices in commercial £ow-through tanks of9.7^14.5 L h 1kg 1 to avoid growth reduction in20^30 g H midae The di¡erence in the suggested

£ow indices between this study and Yearsley (2008)may be due to di¡erences in the characteristics ofthe systems and the relatively high sensitivity ofsmall abalone to poor water quality (Fallu 1991; Red-dy-Lopata et al 2006)

The weight gain of abalone in tank position 2 waslow, possibly as a result of inconsistency in manage-ment These abalone may have been subjected tomore handling stress than the abalone from the othertreatments, due to a support wire, which made pla-cing the baskets into this tier of the raceways di⁄cult.The wire was subsequently removed This may ex-plain why the feed consumption of these animalswas signi¢cantly lower for the ¢rst 2 weeks of the ex-periment, thus delaying their start to normal feedingbehaviour Abalone growth in tank positions 1, 3 and

4 was consistent with that of abalone from the samebatch grown in farm tanks at high £ow indices (16.7^13.9 L h 1kg 1) Abalone in tank positions 5^7 hadhigher FCR values, and it is hypothesized that theseabalone required proportionally more energy togrow and maintain physiological functions such asrespiration, excretion and osmoregulation due to thedeteriorated water quality

Water qualityAmbient air temperature, coupled with the e¡ect ofretention time of water in the system, caused theaverage water temperature to increase by 0.3 1C be-tween the ¢rst and the last tanks in the series.Although the growth rate in H midae was positivelycorrelated with the temperature within the range of12^20 1C (Britz et al.1997), this small temperature in-crease is unlikely to have confounded the growth re-sults Using a model provided by Britz et al (1997), theincrease in temperature down the raceways shouldhave only increased weight gain by 0.19^0.54 g in po-sitions 5^7 Subtracting these values from the re-corded di¡erences of 3.4, 5.1 and 6.2 g betweenpositions 1 and 5^7 would not have altered any ofthe above-mentioned outcomes

Time (weeks) 0

Figure 4 Average weekly concentrations with 95%

con-¢dence levels of dissolved oxygen, pH, free ammonia gen and nitrite-N concentrations over the course of the101-day experimental period, for tank positions 1^7 of aserial-use raceway used for the production of the SouthAfrican abalone, Haliotis midae

nitro-Aquaculture Research, 2011, 42, 918^930 Water quality in an abalone serial-use raceway M A Naylor et al.

Similarly, the low concentrations of suspended

so-lids reported from this study and the small di¡erences

between the ¢rst (4.2 2.0 mg L 1) and the last

(4.8 2.5 mg L 1) tanks in series are unlikely to

have contributed to the di¡erences in abalone growth

between tank positions These concentrations were

similar to those reported by Park, Kim, Kim and Jo

(2008) for the culture of Haliotis dicus hannai (Ino) in

a recirculating system with ba¥ed tanks, where the

total suspended solids concentration averaged

2.7 mg L 1, with a maximum of 7.9 mg L 1 In

race-ways used for ¢sh farming, the movement of the ¢sh

can disturb bottom sediments, thereby increasing

the suspended solid concentration As the abalone

baskets were suspended within the tank, food

parti-cles and faeces on the tank bottom could not be

stir-red up by the abalone In juvenile green grouper,

Epinephelus coioides (Hamilton), exposure to a

sus-pended solid concentration of 50 mg L 1for 6 weeks

caused only minor respiratory distress (Au, Pollino,

Wu, Shin, Lau & Tang 2004) In the mussel, Perna

vir-idis L., exposure to suspended solid concentrations of

0,180, 300, 440 and 600 mg L 1for 14 days, followed

by 14 days of recovery resulted in no signi¢cant

dif-ferences in oxygen consumption or growth (Shin,

Yau, Chow,Tai & Cheung 2002) A low dissolved

oxy-gen concentration in the culture water is a primary

limiting factor for growth in aquatic organisms (Fry

1971) and the ¢rst limiting variable in £ow-through

systems (Colt & Orwicz 1991a) In this study,

di¡er-ences in weight gain were larger than EC5

predic-tions for dissolved oxygen in other species,

suggesting that dissolved oxygen was not the ¢rst

limiting water quality variable in this serial-use

sys-tem In abalone, a low dissolved oxygen

concentra-tion can limit growth by reducing the capacity for

aerobic metabolism (Harris, Maguire, Edwards &

Hindrum 1999; Harris, Maguire, Edwards & Johns

1999), but serial-use resulted in a small decrease in

the dissolved oxygen concentration and oxygen

sa-turation between the ¢rst (8.1mg L 1 O2; 99.4%)

and the last tank (7.4 mg L 1O2; 91.6%) In juvenile

H laevigata (10.8 g abalone 1), weight gain was

re-duced by 5% and 50% at 7.36 and 5.91mg L 1O2

,re-spectively, while there were no signi¢cant di¡erences

in shell growth in the range of 8.9^6.2 mg O2L 1

(Harris, Maguire, Edwards & Hindrum 1999; Harris,

Maguire, Edwards & Johns 1999) In the current trial,

the average dissolved oxygen concentration

re-mained above the EC50suggested by Harris, Maguire,

Edwards and Hindrum (1999); Harris, Maguire,

Ed-wards and Johns (1999); however, values below the

EC5were recorded in 36 samples (12% of samples ken) for tanks 1^4 and 78 samples (35% of samplestaken) for tanks 5^7 For commercial £ow-throughtanks, Yearsley (2008) observed good growth in 20^

ta-30 g H midae at median dissolved oxygen tions of 7.0^7.2 mg L 1O2 In H laevigata and H di-versicolor supertexta, the respiration rates start todecline at 6.2 mg L 1 O2 (80% saturation) and5.5 mg L 1O2respectively (Jan & Chang 1983; Har-ris, Maguire, Edwards & Hindrum 1999) In tank posi-tion 7, dissolved oxygen concentrations andpercentage oxygen saturation were similar to the

concentra-EC5for H laevigata, but there was a 52% decrease inweight gain relative to the values for position 1; thus,

it is hypothesized that other factors made a largercontribution to the observed reduction in abalonegrowth

Based on a comparison of the ¢ndings from thisstudy and a review of the literature, it is suggestedthat low pH contributed to the reductions in growthrecorded in the lower tank positions For example,there was a reduction in pH from 7.86 to 7.6 from po-sitions 1 to 7 The metabolite CO2 combines withwater to form carbonic acid; the carbonic acid thendissociates into bicarbonate ions, resulting in the re-lease of H1ions (Sanni & Forsberg 1996) A lower pHmay be bene¢cial in that it reduces the percentage ofFAN, the toxic form of total ammonia (Bower & Bid-well 1978; Thurston, Russo & Vinogradov 1981), butthe e¡ect of pH on abalone health requires furtherstudies (Harris, Maguire, Edwards & Hindrum 1999;Harris, Maguire, Edwards & Johns 1999), due to theinteractions between pH, dissolved CO2and alkali-nity of seawater (Poxton & Allouse 1982) While pHvalues of 6.5^8.5 are considered to be safe for ¢sh cul-ture (Poxton & Allouse 1982; Colt & Orwicz 1991a), in-creased carbon dioxide concentrations present at alow pH reduced the growth and survival of animalswith calcite shells, such as echinoderms and gastro-pods (Shirayama 2002) In the mussel, Mytilus gallo-provincialis (Lamarck), reducing the water pH to 7.3using CO2gas caused signi¢cant reductions in thegrowth and dissolution of the shell compared with acontrol treatment kept at pH 8.05 (Michaelidis, Ou-zounis, Paleras & P˛rtner 2005) Many abalone inthe current study exhibited signs of a ‘shiny shell’,which is the dissolution of the outer calcite shelllayer, resulting in the exposure of patches of shinynacre The number of abalone with these signs wasnot quanti¢ed In juvenile greenlip abalone, H laevi-gata ( 2.3 g abalone 1), a signi¢cant reduction inweight and length gain occurred in animals exposed

Water quality in an abalone serial-use raceway M A Naylor et al Aquaculture Research, 2011, 42, 918^930

to pH 7.46 over 50^68 days, compared with animals

exposed to pH 7.76 and 8.27 (Harris, Maguire,

Ed-wards & Hindrum 1999; Harris, Maguire, EdEd-wards &

Johns 1999) The authors estimated the EC5and EC50

at 7.78 and 7.39 respectively In the current study,195

pH samples (60%) from tanks 1 to 4 and all pH

sam-ples taken from tanks 5 to 7 (Table 3) were below the

EC5threshold for H laevigata Signi¢cant di¡erences

in weight gain occurred from tank position 5

on-wards, where pH values averaged 7.63,7.61and 7.6

re-spectively These averages fall within the EC5 and

EC50 for growth in both H laevigata and H rubra,

thus further supporting the hypothesis that low pH

values as a result of carbon dioxide excretion

contrib-uted to the reduction in abalone weight gain

However, combined changes in the concentrations

of carbon dioxide and free ammonia nitrogen, FAN,

may lead to growth-limiting conditions in serial-use

systems (Fivelstad et al 1991; Colt et al 2009) The

¢ndings from this study suggest that the average

con-centration and peaks of FAN limited abalone growth

under the experimental conditions but the

growth-reducing e¡ect of FAN may have been exacerbated

by low pH values

The excretion of ammonia by the abalone (Barkai

& Gri⁄ths 1987) and bacterial decomposition of

un-eaten food and faeces (Yearsley, Jones & Britz 2009)

are hypothesized to be the main causes of the linear

increase in total ammonia (TAN) concentrations

with increasing water use The linearity of the TAN

increase suggests that the net production of TAN

was similar in all tanks In a similarly designed study

using H rubra, Huchette et al (2003) found that the

TAN concentration was positively correlated with

cu-mulative biomass A high ammonia concentration in

the culture water can lead to elevated ammonia

con-centrations in the body of aquatic organisms (Randall

& Tsui 2002; Miron, Moraes, Becker, Crestani,

Spane-vello, Loro & Baldisserotto 2008) Some ¢sh species

are, however, tolerant of high ammonia

concentra-tions in the body and have developed strategies to

minimize ammonia toxicity (Randall & Tsui 2002)

Such strategies have not been documented for

aba-lone Russo and Thurston (1991) suggested that

re-duced growth and feed uptake were recorded at

concentrations of 2^150mg L 1FAN In the range of

40^400mg L 1FAN, ¢sh exhibited anaemia, gill and

kidney degeneration and reduced disease resistance

Abalone seem to be very sensitive to FAN (Harris et al

1998; Huchette et al 2003; Reddy-Lopata et al 2006)

The EC5for weight gain at pH of 7.95^8.17 was

esti-mated at 41mg FAN L 1for H laevigata (32 mm) and

4^6mg FAN L 1for H rubra (15^65 mm) (Harris et al.1998; Huchette et al 2003) In tank positions 1^4, nosamples exceeded the estimated EC5of 4mg L 1for

H rubra However, in tank positions 5^7, values tween 4 and 6mg L 1FAN were recorded 38 times(15% of samples), occurring at least three times ineach tank Reddy-Lopata et al (2006) showed thatthe 36 h LC50 for 50^80 mm H midae was12.9mg FAN L 1 The authors also recorded a 59%decrease in speci¢c weight gain in 10^25 mm H mid-

be-ae after chronic exposure to 7.4mg FAN L 1at pH 7.8and stated that the EC5for H midae is likely to be low-

er than that estimated for H rubra (EC5, 4^6mg L 1).Percentage weight gain was 52% lower in tank posi-tion 7 than that in tank position 1, where the FANconcentrations averaged 2.7mg FAN L 1 Althoughthe concentrations of FAN experienced by abalone

in the lower tank positions are slightly lower thanthose presented by Harris et al 1998; Huchette et al

2003 and Reddy-Lopata et al (2006), these tanks hadlower pH values (pH 7.63^7.60) than the top tanks.The 96-h LC50concentrations for FAN in rainbowtrout (Oncorhynchus mykiss Walbaum), fathead min-nows (Pimephales promelas Ra¢nesque) (Thurston

et al 1981) and silver cat¢sh (Rhamdia quelen Quoy &Gaimard) (Miron et al 2008) were lower at a lower en-vironmental pH It is hypothesized that EC5and EC50concentrations would be a¡ected similarly and thatthe presence of H1ions may increase the toxicity orthe bioavailability of the NH3molecule Peaks in theFAN concentration in weeks 4, 8 and 11 are likely tohave been caused by high ambient temperatures, re-sulting in increased TAN production by the abalone,and during the breakdown of food and faecal wastes

As tanks were cleaned once every 2 weeks, the peaks

in FAN during weeks 4 and 8 would have coincidedwith a period when accumulated feed and faecalwastes were at their maximum Yearsley et al (2009)showed that the presence of accumulated sludge onthe tank bottom resulted in TAN concentrations44% higher then tanks containing no sludge.Although the nitrite concentration increasedthroughout the experimental period, being the high-est in tank positions 5^7, nitrite was the least likelycause for the reduction in abalone growth Nitrifyingbacteria are slow-growing, autotrophic bacteria (Colt

et al 2009) and require ammonia as an energysource Nitrifying bacteria were not counted or iden-ti¢ed in this study However, it is hypothesized thatthe availability of ammonia initiated and maintainedbacterial growth, leading to ammonia being oxidized

to nitrite Nitrifying bacteria are sessile and may have

Aquaculture Research, 2011, 42, 918^930 Water quality in an abalone serial-use raceway M A Naylor et al.

been attached to the abalone shells, the basket, rack

and feeder plate, which were merely moved during

tank cleaning and never scrubbed or sprayed clean

Establishing nitrifying bacteria within the tank

be-fore the introduction of abalone may help reduce

am-monia concentrations in the lower tank positions of

the serial-use raceway, in particular, when operating

at low £ow indices In freshwater ¢sh species, the

transport of nitrite occurs via the chloride cells and

can result in blood plasma nitrite concentrations

above the concentrations in the water (Lewis &

Mor-ris 1986; Russo & Thurston 1991; Tomasso & Grosell

2005) Because of the high chloride concentration in

sea water, marine ¢sh developed an opposite osmotic

gradient to freshwater ¢sh, thereby preventing nitrite

uptake via this pathway (Jensen 2003) However, the

high concentration of chloride ions in marine

sys-tems may not protect all invertebrates (Tomasso

1994) Nitrite may enter marine organisms through

the drinking of seawater and subsequent uptake

across the intestinal epithelium, or via di¡usion of

ni-trous acid across epithelial cells (Jensen 2003) Nitrite

is generally less toxic in seawater, especially at the

naturally occurring high pH levels (Wedemeyer &

Ya-sutake 1978) There is a paucity of data on the chronic

toxicity of nitrite to abalone Nitrite oxidizes the

re-spiratory pigment haemocyanin to form

methaemo-cyanin, thereby reducing oxygen transport In H

laevigata, Harris et al (1997) recorded signi¢cant and

uniform reductions in growth regardless of

concen-tration in the range of 0.56^7.8 mg NO2-N L 1

rela-tive to a control treatment (0.024 mg NO2 -N L 1)

The average nitrite concentrations in the serial-use

raceways were similar to those of the control used

by Harris et al (1997)

The variations in in£uent dissolved oxygen and pH

over time were probably caused by weather patterns

such as upwelling, discharge of acidic river waters,

rough seas and high temperatures

Assuming a farm with a standing stock of 100 t,

costs of South African Rand (ZAR) 0.29 m 3 of

water pumped and a feed cost of ZAR 17 kg 1,

oper-ating at the minimum prescribed £ow index would

reduce pump costs by 30^32% when compared with

the current £ow indices used on commercial abalone

farms in South Africa Because of the high value of

abalone, and the slightly better production at high

£ow indices, the minimum £ow index suggested here

is not the most pro¢table However, in South Africa,

the electricity price is expected to rise by 25% per

year for the next 3 years By the end of the third year,

a £ow index of 7.2^9.0 L h 1kg 1 is expected to

generate more pro¢t due to lower pumping coststhan the currently used £ow indices Thus, based onthese estimates, there is a justi¢cation to design fu-ture studies to conduct a comprehensive economicanalysis of serial-use systems

ConclusionsWater £ow was set low in order to determine theminimum required £ow indices, and to understandhow the concentrations of dissolved oxygen, pH, am-monia and nitrite a¡ect growth in a commerciallyoperated system Deterioration in water quality due

to serial-use negatively a¡ected the wet weight gainand FCR of 60^70 mm H midae, but not their survi-val The minimum £ow per unit biomass was esti-mated to range from 7.2 to 9.0 L h 1kg 1to preventsigni¢cant growth reduction The model equationsfor weight gain and FCR can be used by farmers tocalculate the most economic £ow indices within thesafe range suggested here, depending on the cost ofpumping and the stocking strategy used By compar-ing the results of this experiment with those by otherresearchers, it is suggested that low pH was the ¢rstgrowth-limiting variable in the serial-use racewaysthrough its interaction with FAN, which becomesmore toxic as the pH decreases

AcknowledgmentsThe ¢nancial support of HIK Abalone Farm, AquafarmDevelopment and Roman Bay Sea Farm is gratefullyacknowledged The ¢nancial assistance of the NationalResearch Foundation (NRF) towards this research ishereby acknowledged The opinions expressed andthe conclusions arrived at are those of the author andare not necessarily to be attributed to the NRF Wethank Rowan Yearsley, Ernst Thompson and MorenaKhashane for assistance during the trial period

References

Au D.W T., Pollino C.A., Wu R.S.S., Shin P.K.S., Lau S.T.F & Tang J.Y.M (2004) Chronic e¡ects of suspended solids on gill structure, osmoregulation, growth, and triiodothyro- nine in juvenile green grouper Epinephelus coioides Mar- ine Ecology Progress Series 266, 255^264.

Barkai R & Gri⁄ths L (1987) Consumption, absorption ciency, respiration and excretion in the South African abalone Haliotis midae South African Journal of Marine Science 5, 523^529.

e⁄-Water quality in an abalone serial-use raceway M A Naylor et al Aquaculture Research, 2011, 42, 918^930

Basuyaux O & Mathieu M (1999) Inorganic nitrogen and its

e¡ect on growth of the abalone Haliotis tuberculata

Lin-naeus and the sea urchin Paracentrotus lividus Lamarck.

Aquaculture 174, 95^107.

Bower C.E & Bidwell J.P (1978) Ionization of ammonia in

seawater: e¡ects of temperature, pH and salinity Journal

of Fisheries Research Board Canada 35, 1012^1016.

Britz P.J (1996) The suitability of selected protein sources for

inclusion in formulated diets for the South African

aba-lone, Haliotis midae Aquaculture 140, 63^73.

Britz P.J., Hecht T & Mangold S (1997) E¡ect of temperature

on growth, feed consumption and nutritional indices of

Haliotis midae fed a formulated diet Aquaculture 152,

191^203.

ChengW., Li C & Chen J (2004) E¡ect of dissolved oxygen on

the immune response of Haliotis diversicolor supertexta

and its susceptibility to Vibrio parahaemolyticus

Aquacul-ture 232, 103^115.

Colt J (2006) Water quality requirements for reuse systems.

Aquacultural Engineering 34, 143^156.

Colt J & Orwicz K (1991a) Modeling production capacity of

aquatic culture systems under freshwater conditions.

Aquacultural Engineering 10, 1^29.

Colt J & Orwicz K (1991b) Aeration in intensive culture In:

Aquaculture andWater Quality (ed by D.E Brune & J.R

To-masso), pp 199^271 World Aquaculture Society, Baton

Rouge, LA, USA.

Colt J.,Watten B & Rust M (2009) Modeling carbon dioxide,

pH, and un-ionized ammonia relationships in serial reuse

systems Aquacultural Engineering 40, 28^44.

Eshchar M., Mozes N & Fediuk M (2003) Carbon dioxide

re-moval rate by aeration devices in marine ¢sh tanks The

Israeli Journal of Aquaculture 55,79^85.

Fallu R (1991) Abalone Farming Fishing News Books,

Ox-ford, UK, 195pp.

Fivelstad S (1988) Water£ow requirements for salmonids in

single-pass and semi-closed land-based seawater and

freshwater systems Aquacultural Engineering 7, 183^200.

Fivelstad S., Bergheim A & Tyvold T (1991) Studies of

limit-ing factors governlimit-ing the water£ow requirement for

Atlantic salmon (Salmo salar L.) in land-based seawater

systems Aquacultural Engineering 10, 237^249.

Fry F.E.J (1971) The e¡ect of environmental factors on the

phy-siology of ¢sh In: Fish Phyphy-siology, 9 (ed by W.S Hoar & D.J.

Randall), pp l^98 Academic Press, New York, NY, USA.

Harris J.O., Maguire G.B., Edwards S.J & Hindrum S.M.

(1997) E¡ect of nitrite on growth and oxygen

consump-tion for juvenile greenlip abalone Haliotis laevigata

Dono-van Journal of Shell¢sh Research 16, 395^401.

Harris J.O., Maguire G.B & Hindrum S.M (1998) E¡ect of

ammonia on the growth rate and oxygen consumption

of juvenile greenlip abalone, Haliotis laevigata Donovan.

Aquaculture 160, 259^272.

Harris J.O., Maguire G.B., Edwards S.J & Hindrum S.M.

(1999) E¡ect of pH on growth rate, oxygen consumption

rate, and histopathology of gill and kidney tissue for

juve-nile greenlip abalone, Haliotis laevigata Donovan and blacklip abalone, Haliotis rubra Leach Journal of Shell¢sh Research 18, 611^619.

Harris J.O., Maguire G.B., Edwards S.J & Johns D.R (1999) Low dissolved oxygen reduces growth rate and oxygen consumption rate of juvenile greenlip abalone, Haliotis laevigata Donovan Aquaculture 174, 265^278.

Huchette S.M.H., Koh C.S & Day R.W (2003) Growth of juvenile blacklip abalone (Haliotis rubra) in aquaculture tanks: e¡ects

of density and ammonia Aquaculture 219, 457^470 Jan R & Chang K (1983) The oxygen consumption by for- mosan abalone, Haliotis diversicolor supertexta Lishke, during decline of ambient oxygen Bulletin of Institutional Zoology, Academia Sinica 22, 43^48.

Jensen F.B (2003) Nitrite disrupts multiple physiological functions in aquatic animals Comparative Biochemistry and Physiology Part A 135, 9^24.

Levene H (1960) Robust tests for equality of variances In: Contributions to Probability and Statistics (ed by I Olkin

& H Hotelling), pp 278–292 Stanford University Press, Stanford, CA, USA.

Lewis W.M & Morris D.P (1986) Toxicity of nitrite to ¢sh: a review Transactions of the American Fisheries Society 115, 183^195.

Michaelidis B., Ouzounis C., Paleras A & P˛rtner H.O (2005) E¡ects of long-term moderate hypercapnia on acid^base balance and growth rate in marine mussels Mytilus gallo- provincialis Marine Ecology Progress Series 293, 109^118 Miron D.S., Moraes B., Becker A.G., Crestani M., Spanevello R., Loro V.L & Baldisserotto B (2008) Ammonia and pH e¡ects on some metabolic parameters and gill histology

of silver cat¢sh, Rhamdia quelen (Heptapteridae) ture 277, 192^196.

Aquacul-Morrison J.K & Piper R.G (1988) E¡ect of reused water

on Atlantic salmon The Progressive Fish-Culturist 50, 110^112.

Park J., Kim H., Kim P & Jo J (2008) The growth of disk lone, Haliotis discus hannai at di¡erent culture densities in a pilot-scale recirculating aquaculture system with a ba¥ed culture tank Aquacultural Engineering 38, 161^170 Piedrahita R.H (2003) Reducing the potential environmen- tal impact of tank aquaculture e¥uents through intensi-

aba-¢cation and recirculation Aquaculture 226, 35^44 Poxton M.G & Allouse S.B (1982) Water quality criteria for marine ¢sheries Aquacultural Engineering 1, 153^191 Qian P.,Wu M & Ni I (2001) Comparison of nutrients release among some maricultured animals Aquaculture 200, 305^316.

Randall D.J & Tsui T.K.N (2002) Ammonia toxicity in ¢sh Marine Pollution Bulletin 45, 17^23.

Reddy-Lopata K., Auerswald L & Cook P (2006) Ammonia toxicity and its e¡ect on the growth of the South African abalone Haliotis midae Linnaeus Aquaculture

261, 678^687.

Russo R.C & Thurston R.V (1991) Toxicity of ammonia, nitrite and nitrate to ¢shes In: Aquaculture and Water Aquaculture Research, 2011, 42, 918^930 Water quality in an abalone serial-use raceway M A Naylor et al.

Quality (ed by D.E Brune & J.R Tomasso), pp 58^89.

World Aquaculture Society, Baton Rouge, LA, USA.

Sanni S & Forsberg O.I (1996) Modelling pH and carbon

di-oxide in single-pass sea-water aquaculture systems.

Aquacultural Engineering 15, 91^110.

Schuenho¡ A., Shpigel M., Lupatsch I., Ashkenazi A.,

Msuya F.E & Neori A (2003) A semi-recirculating,

integrated system for the culture of ¢sh and seaweed.

Aquaculture 221, 167^181.

Shapiro S.S & Wilk M.B (1965) An analysis of variance test

for normality (complete samples) Biometrika 52,591^611.

Shin P.K.S,Yau F.N., Chow S.H.,Tai K.K & Cheung S.G (2002)

Responses of the green-lipped mussel Perna viridis (L.) to

sus-pended solids Marine Pollution Bulletin 45, 157^162.

ShirayamaY (2002) Towards comprehensive understanding

of impacts on marine organisms due to raised CO 2

con-centration In: Proceedings of the 5th International

Sympo-sium on CO 2 Fixation and E⁄cient Utilization of Energy, pp.

177–181 Tokyo Institute of Technology, Tokyo, Japan.

Solorzano L (1969) Determination of ammonia in natural

waters by the phenolhypochlorite method Limnology

and Oceanography 14,799^801.

Summerfelt S.T., Davidson J.W.,Waldrop T.B.,Tsukuda S.M &

Bebak-Williams J (2004) A partial-reuse system for

cold-water aquaculture Aquacultural Engineering 31, 157^181.

Thurston R.V., Russo R.C & Vinogradov G.A (1981)

Ammo-nia toxicity to ¢sh: e¡ect of pH on the toxicity of the

un-ionized ammonia species Environmental Science and nology 15, 837^840.

Tech-Tomasso J.R (1994) Toxicity of nitrogenous wastes to culture animals Reviews in Fisheries Science 2, 291^314 Tomasso J.R & Grosell M (2005) Physiological basis for large di¡erences in resistance to nitrite among freshwater and freshwater-acclimated euryhaline ¢shes Environmental Science and Technology 39, 98^102.

aqua-Vergara J.M., Lopez-Calero G., Robaina L., Caballero M.J., Montero D., Izquierdo M.S & Aksnes A (1999) Growth, feed utilization and body lipid content of gilthead seabream Sparus aurata fed increasing lipid levels and ¢sh meals of di¡erent quality Aquaculture 179, 35^44.

Wedemeyer G.A & Yasutake W.T (1978) Prevention and treatment of nitrite toxicity in juvenile steelhead trout (Salmo gairdneri) Journal of Fisheries Research Board Cana-

da 35, 822^827.

Yearsley R.D (2008) Water quality, abalone growth and the tential for integrated mariculture on a South African abalone Haliotis midae L farm MSc thesis, Rhodes University, Gra- hamstown, South Africa,77pp.

po-Yearsley R.D., Jones C.L.W & Britz P (2009) E¡ect of settled sludge on dissolved ammonia concentration in tanks used

to grow abalone (Haliotis midae L.) fed a formulated diet Aquaculture Research 40, 166^171.

Water quality in an abalone serial-use raceway M A Naylor et al Aquaculture Research, 2011, 42, 918^930

Aquaculture development and scenarios of change

in fish trade and market access for the poor in

Cambodia

Martin L van Brakel1& Lindsay G Ross2

1 World Fish Center, Penang, Malaysia

2 Institute of Aquaculture, University of Stirling, Stirling, UK

Correspondence: Present address: M L van Brakel, Challenge Program on Water and Food, 127 Sunil Mawatha, Pellawatta mulla, Sri Lanka E-mail: m.vanbrakel@cgiar.org

Battara-Abstract

Aquaculture holds considerable potential to

contri-bute to poverty alleviation, if it provides poor people

with opportunities other than as primary producers

Integration of aquaculture into poverty reduction

programmes provides means to diversify production

systems and reduce food insecurity but also needs

improved markets in locations where aquaculture

can o¡er sustainable livelihoods to poor farming

households This study reviews the current

con-straints that poor people face in accessing markets

in Cambodia and analyses its implications for

pro-poor domestic aquaculture development We use

a Geographic Information System-based spatial

Bayesian probability model to simulate market

acces-sibility and estimate the numbers of poor people who

could potentially bene¢t from improved market

access under four di¡erent scenarios Analysis of

secondary data con¢rms that the potential for

poor aquaculture producers to interact with urban

markets in Cambodia is currently low The potential

of aquaculture to interact with rural markets is,

how-ever, high It is concluded that the development

of aquaculture has considerable potential to reduce

the transaction costs in domestic ¢sh trade by

improved access of poor producers and consumers

to rural markets in Cambodia An aquaculture

devel-opment strategy that improves rural market access

could include bene¢ts for up to 1 million poor aquatic

resource users

Keywords: aquaculture, market access, poverty,

scenarios of change

IntroductionAquaculture and poverty alleviationAquaculture and improved aquatic resource man-agement hold considerable potential for alleviation

of poverty if poor people are engaged through tunities other than as primary producers Poor peo-ple need access to appropriate and a¡ordable low-riskaquaculture technologies, markets, access and con-trol over common property resources and rights toparticipate e¡ectively in aquaculture developmentplanning Bene¢ts from aquaculture can be directthrough target groups engaging in aquaculture pro-duction, indirect through secondary opportunitiesassociated with aquaculture and inclusive throughbroader bene¢ts from aquaculture that eventuallyreach target groups (Muir 1999) This study aims toidentify current limitations to the development ofpro-poor aquaculture in Cambodia and uses simula-tion of change in market access in order to emulatehow such changes could a¡ect opportunities for thedevelopment of pro-poor aquaculture in Cambodia inthe foreseeable future The current situation of ¢shsupply and demand in Cambodia provides a baselinefor simulation of di¡erential market access by poorand better-o¡ producers and consumers The simula-tions provide estimates of the numbers of poor peoplewho potentially could bene¢t from improved aqua-culture value chains and market access

oppor-Employment in aquaculture has important tial to contribute to poverty reduction in developingcountries such as Cambodia and the Philippines,where labour supply is still abundant (Ahmed &

poten-Aquaculture Research, 2011, 42, 931^942 doi:10.1111/j.1365-2109.2010.02661.x

Lorica 2002; Irz, Stevenson, Tanoy, Villarante &

Morissens 2007) Adoption of aquaculture by

agricul-tural households has had positive income,

consump-tion and employment e¡ects in countries such as

Bangladesh, China, India, Indonesia, Thailand and

Vietnam Incorporation of aquaculture into existing

agricultural farming systems has also contributed to

improved productivity and diversi¢cation of

small-scale farms In £ood-prone ecosystems in

Bangla-desh, signi¢cant increases in farm and household

income were reported following interventions that

targeted aquaculture in order to increase on-farm

production of resource poor households (Gupta,

Mazid, Islam, Rahman & Hussain 1999) Compared

with non-producing households and the national

average consumption, the per capita consumption of

¢sh in rural areas of Bangladesh was also noticeably

higher for households adopting aquaculture (Dey,

Bimbao,Yong, Regaspi, Kohinoor, Pongthana &

Para-guas 2000) Phong, Udo, van Mensvoort, Bosma,

Tri, Nhan and van der Zijpp (2008) concluded that

Integrated Agriculture-Aquaculture systems in the

Mekong Delta of Vietnam have provided an adequate

response to threats and opportunities arising from

rapid agricultural development, related market

£uc-tuations and changes in policy Poorer farmers

tended towards diversi¢cation of their farming

sys-tems with aquaculture in order to avoid risks and

safeguard their livelihood, while better-o¡ farmers

tended towards specialization and intensi¢cation of

their farming practices Similar opportunities

poten-tially exist in Cambodia, given the importance of

aquatic resources to the rural poor in the country

and Cambodia’s rapid integration into regional and

global markets since the mid-1990s, which came

ac-companied by considerable demand volatility The

prices of ¢sh and other food commodities have

in-creased signi¢cantly since May 2007 (CDRI 2008)

This has provided opportunities, but also obstacles

for agriculture and wider economic development,

and has undermined poverty reduction in Cambodia

(Sovannarith 2009) Fishing communities are among

those most severely a¡ected Fish catches and the

average daily income of ¢shing households declined,

while their daily expenditure increased Fresh ¢sh

prices have increased relatively modestly, by about

20%, in comparison with rice, which doubled in price

(CDRI 2008) In a survey held in June 2008 on the

ef-fects of high food prices among 2235 households in14

villages in Cambodia, 98% of people in ¢shing villages

indicated that they did not have enough money to buy

food or cover essential expenses (Sophal 2009)

Aquatic resources and poverty in CambodiaThe Cambodian poor still accounted for 30% of thepopulation or 4 million people in 2008 (CDRI 2008).The highest incidence of poverty in Cambodia isfound among households where agriculture is theprimary source of income In 2004, 90% of the poorlived in rural areas (Royal Government of Cambodia2005) In rural Cambodia, poverty is closely linked tofood insecurity According to Sophal (2009), morethan 1.5 million people in rural areas and more than

150 000 of the urban population are food insecure.Sixty-¢ve per cent of rural households are eitherlandless or land poor (owning one hectare or less).The majority of rural residents are net buyers of food

In the poorest two quintiles, food consumptionamounts to 70% of the total household expenditure.Rice, ¢sh and ¢sh products, supplemented by seaso-nal fruits and vegetables, are the staple foods forthese people Capture ¢sheries contribute signi¢-cantly to food security, nutrition and income genera-tion On a seasonal basis, ¢shing is a part-timeactivity for virtually all farming households Percapita inland ¢sheries production amounts to28.2 kg/year and ranks ¢rst worldwide (FAO 2003).Cambodia’s ¢sheries sector is undergoing majorreforms towards a more poverty-focused approach.Despite this focus, Cambodia’s National Poverty Re-duction Strategy (Council for Social Development2002) acknowledges that wild ¢sh supplies will not

be enough to meet the future demand of Cambodia’srapidly increasing population Recent increases inthe development of pond-based aquaculture aremainly a response to declining supplies of wild ¢sh.Promotion of small-scale aquaculture and commu-nity-based ¢sheries are integral components of thecountry’s strategy for equitable agricultural develop-ment that prioritizes poverty alleviation and foodsecurity Aquaculture development through pond-based and rice-¢sh aquaculture has been receivingincreasing attention because of its potentially impor-tant role in providing food security and rural incomegeneration (Ministry of Planning, Royal Government

of Combodia, United Nations World Food Programme

& United Nations Development Program 2001).Aquaculture is currently practiced in virtually all

of Cambodia (Viseth & Pengbun 2005) Its tion to food ¢sh supply has increased by 28.6% from

contribu-an estimated 20 675 tonnes in 2004 to 34 200 tonnes

in 2006 (FAO 2009) Uno⁄cial estimates, ever, suggest that the aquaculture output in Cambo-dia currently could be as high as 80 000 tonnes

how-Aquaculture and scenarios of change in Cambodia M L van Brakel & L G Ross Aquaculture Research, 2011, 42, 931^942

annually (Nam & Leap 2007) River cat¢sh

Pangasia-nodon hypophthalmus (Sauvage) is the most widely

cultured species in ponds It is cultured around

Phnom Penh and major provincial towns in the

country, where feed inputs are more readily available

Cage and pen culture are still the principal systems of

inland aquaculture in provinces around the Great

Lake (Tonle Sap), where it is especially developed in

Kampong Chhnang province (Viseth & Pengbun

2005) These types of aquaculture have evolved from

the activity of stocking surplus catch from the highly

seasonal inland ¢shery in £oating cages and bamboo

¢shpens, in order to keep the ¢sh alive for sale in the

o¡-season (DFID-SEA Aquatic Resources

Manage-ment Programme 2000) The two major species

cultured in cages, river cat¢sh and giant snakehead

Channa micropeltes (Cuvier), are high-value species

Cage culture is highly dependent on feed ¢sh from

the wild, and concerns have been raised over the

impacts of expanding feed ¢sh demand on the

avail-ability of ¢sh for poorer consumers and on the

environment (Phillips 2002; Sverdrup-Jensen 2002)

The availability and price of feed ¢sh vary during the

year Feed ¢sh is readily available during December to

March, when small ¢sh are in abundance During June

to September, when the ¢shing season is closed, the

availability of feed ¢sh is low and the prices are high

According to Thuok and Viseth (2004) cited in Viseth

and Pengbun (2005), the production of river cat¢sh

and giant snakehead in both ponds and cages has

de-clined steadily due to the shortage of wild seed supply,

shortage of feed ¢sh and deterioration of water quality

Small-scale aquaculture production may be the

only option for poor farmers to derive direct bene¢ts

from aquaculture, especially in areas with limited

ac-cess to capture ¢sheries resources During the last

decade, considerable e¡ort has been dedicated to

pro-mote small-scale aquaculture in earthen ponds The

¢sh are fed on a variety of farm waste products that

are cheap and easily available all year round

Rice-¢sh culture, an extensive form of aquaculture where

¢sh are released into £ooded rice paddies, has been

introduced particularly in Prey Veng and Svay Rieng

provinces, and more recently in Takeo province

Land-based pond aquaculture and rice-¢sh culture

systems are most appropriate for poor farmers in

in-land rural areas, but many poor households do not

own su⁄cient land to practice culture in a

land-based system For landless farmers, water-land-based

sys-tems such as cages, pens or enhanced ¢sheries in

large communal water bodies may be the only option

to culture ¢sh, but this type of culture is usually

beyond the reach of poor and marginal farmers ashigh investment levels are required (Edwards 1999;Friend & Funge-Smith 2002)

Markets and fish trade in CambodiaOur hypothesis is that bene¢ts from aquaculture de-velopment to the poor in Cambodia will be mostly in-direct through improved trade e⁄ciencies andreduced transaction costs along the domestic ¢shmarketing value chain Freshwater ¢sh are amongCambodia’s most important traded commodities(Chea & McKenney 2003a) Fresh and processed

¢sh are widely traded domestically, and exported insigni¢cant quantities to neighbouring countries,principally Thailand and Vietnam Cambodia is a netexporter of ¢sh and the sector has been targeted asimportant for export promotion Nonetheless, theavailable statistics indicate that export revenues varywidely from month to month and between years Rev-enues declined almost by half from USD 10 million in

2005 to USD 5.1million in 2006 (source: CDRI).Whileexport promotion is intended to spur ‘pro-poor’ trade,market integration and trade e⁄ciencies also need to

be improved domestically

Fish trade in Cambodia is in£uenced by a complexset of factors Currently, ¢sh supply to domestic mar-kets is stable, but the market prices of ¢sh are variableand are in£uenced by imports from neighbouringcountries and the availability of wild-caught ¢sh.Cambodia has a small and fragmented internal mar-ket (Ward 2002) Domestic ¢sh trade statistics arescant, but where available, con¢rm the wide varia-tion in fresh ¢sh prices between provinces andbetween seasons Exports may deprive domesticusers of ¢sh, and shortages in supply may occur sea-sonally in areas with limited access to markets andcapture ¢sheries resources In rural areas locatedclose to Phnom Penh, marketable ¢sh may be di-verted away to urban markets, reducing its localavailability, whereas the costs of transportation tothese markets become prohibitive in areas distantfrom Phnom Penh Markets and distribution centresare characterized by ine⁄ciencies, which increasecosts Traders and distributors complain about highdistribution centre fees and lack of services Onlyone distribution centre serves Phnom Penh for eachmajor trade route and ¢sh distribution facilities areabsent in Poi Pet, the only o⁄cial export border cross-ing with Thailand Spoilage and weight loss along themarketing chain result in considerable value losses,

Aquaculture Research, 2011, 42, 931^942 Aquaculture and scenarios of change in Cambodia M L van Brakel & L G Ross

which are compounded by poor retail marketing

practices such as displaying ¢sh for sale without ice

Analysis of the Tonle Sap ¢sh trade o¡ers valuable

insights into the current market constraints

asso-ciated with ¢sh trade in Cambodia The majority of

traded ¢sh in Cambodia originates from the Tonle

Sap (Chea & McKenney 2003a, b) Marketing of this

¢sh involves a number of transactions, of which sales

from ¢shers to traders, from traders to retailers via

distributors and from retailers to customers are the

most important ones Tied relations are the norm in

this system Distributors play a dominant role in the

market structure They are the ¢nanciers of ¢sh trade

and lend capital to traders to support ¢sh purchases

Traders in turn re-lend some of this capital to ¢shers

for gear purchases and other expenses Under the

terms of these loans, ¢shers are obliged to sell all ¢sh

to their creditor, the trader, and traders in turn must

sell all ¢sh through their distributor and pay the

associated commission fees The problem of credit

dependency places ¢shers at a disadvantage Most

¢shers are in debt to a trader and in a weak position

for negotiating the sale of their catch Pervasive fee

charges along the trade route also reduce pro¢t

mar-gins in ¢sh exports Many of these fees are charged

without a legal basis (Chea & McKenney, 2003b)

Seng (2006) reports that informal fees charged on

¢sh exports via waterways to Vietnam are about

twice as high as formal fees Informal fees also tend

to increase in relation to the distance between the

landing site and the market Fee charges absorb a

large proportion of the potential earnings of

expor-ters and add more than 50% to the costs of exporting

¢sh Hence, high transaction costs and tied trade

re-lationships are the principal constraints to be

ad-dressed in order to secure bene¢ts from ¢sh trade to

the rural poor in Cambodia

Methodology

Materials

This study makes use of data on poverty headcount

and spatial population distribution in Cambodia to

simulate possible scenarios for pro-poor aquaculture

market development In the absence of data on

poverty disaggregated by speci¢c indicators, let

alone aquaculture speci¢c poverty indicators, the

simulation is supported by anecdotal information

about poverty and aquatic resource use in Cambodia

The IDRISI AndesrGeographic Information System

is used to estimate the numbers of poor people who

could potentially bene¢t from small-scale ture development under four di¡erent scenarios ofmarket access A global population distribution mod-

aquacul-el at 1km resolution, known as LandScanTM (OakRidge National Laboratory, Oakridge, TN, USA), pro-vides a proxy indicator of aquaculture developmentpotential under these di¡erent market scenarios.The proxy is a distance decay function of populationdistribution and marketing constraints, expressed in

‘distance to potential markets’ Commune-level lation and poverty counts from the Cambodia povertyanalysis in the year 2000, provided by the vulnerabil-ity analysis and mapping unit of the World Food Pro-gramme (WFP) Cambodia o⁄ce are redistributed onthe basis of this proxy While the scenarios can holdtrue for rural market development in general, theyare assumed particularly sensitive to the ¢sheries sec-tor because trade constraints in the sector are exacer-bated by distance and limited market access,characterized by high transaction costs and tied rela-tionships and associated with a lack of storage andpreservation facilities

popu-Population distributionLandScan is a population distribution model that com-bines four primary geospatial datasets that are key in-dicators of population distribution, namely land cover,roads, slope and night-time lights The model is based

on best available census counts and represents an bient population distribution over a 24-h period Itdoes not represent residential population density butinstead estimates the average number of people en-countered per grid cell over 24 h This feature empha-sizes hotspots of economic activity, which is important

am-to this study The LandScanTMGlobal Population tabase (2002) was used because the time lag betweenthe 2002 release and population data from Cambodia’s

Da-1998 population census was considered to be ciently short to justify the use of the two datasets inconjunction Moreover, the LandScanTMGlobal Popu-lation Database (2002) incorporates important re¢ne-ments over earlier versions The IDRISI regional datadisaggregation tool was used to redistribute the com-mune-level population averages available from thecensus according to the LandScan model This stepprovided a more realistic population distribution grid

su⁄-at a nominal 1km resolution The populsu⁄-ation countswere log-transformed (ln) in order to normalize themover a continuous scale for better visualization (Fig.1a) and were then classi¢ed into discrete classes of

Aquaculture and scenarios of change in Cambodia M L van Brakel & L G Ross Aquaculture Research, 2011, 42, 931^942

population density (Fig 1b; Table 1) This grid was

overlaid (multiplied) with the WFP commune-level

poverty headcount expressed as a fraction of the

po-pulation ranging between 0 (no poor) and 1 (100%

poor) The result is a commune-level poverty

distribu-tion grid at 1km resoludistribu-tion

Market access

A spatial Bayesian probability model was applied to

the redistributed population counts in order to

esti-mate the likelihood of access of poor people to either

urban or rural markets Bayesian probability theory

establishes the probability that an entity belongs to

any of a number of di¡erent classes or states In this

case, the two states were ‘urban market access’ (Pu)

and ‘rural market access’ (Pr) The underlying

as-sumption was that the probability of encounteringmore than 5000 people per km2at any given moment

is an indicator of the presence of urban centres andmarkets A probability of encountering 500^5000people per km2is an indicator of peri-urban centresand was assumed to be equivalent to rural markets.The population density at any point of origin divided

by its distance to these potential markets provided aproxy for the accessibility of markets from that point,expressed as the ‘interaction potential’ Vibetweenthat point and its nearest market The method as-sumes a concentration of rural population in theproximity of rural towns and major urban centres,implying that rural population densities are higher

in the vicinity of such centres than in areas distantfrom urban centres The basis of this approach is theconcept of social gravitation Urban centres are re-garded as physical masses, where the magnitude ofattraction or interaction is proportional to the popu-lation size and inversely proportional to some form ofspatial friction (Deichmann & Eklundh 1991), in thiscase market access

For a given point i, the interaction potentialViof itspopulation with any urban centre is therefore

Vi¼Xn j¼1POPDj= dð Þij b ð1Þ

where POPDjis the population density of market jand dijis the distance between points i and j The ex-

Ln(pop) (a)

TS

P KA PV SV TK

Figure 1 (a) Natural logarithm of population density [ln (pop) km 2]; (b) population density redistributed into discreteclasses KA, Kandal province; P, Phnom Penh (capital); PV, Prey Veng province; SR, Siem Reap province; SV, Svay Riengprovince; TK,Takeo province; TS,Tonle Sap lake

Table 1 Population density (POPD) classi¢cation

ln (pop) (Fig.1a) Inhabitants km 2(Fig.1b) POPD class

Aquaculture Research, 2011, 42, 931^942 Aquaculture and scenarios of change in Cambodia M L van Brakel & L G Ross

ponent b is a distance weight that determines the

structure of the distance decay In this study, the

popu-lation density at any point was assumed to be ‘known’

Rather than being interested in POPDj, which is

al-ready ‘known’to be a market on the basis of population

density, the population density POPDiat any potential

supply point i to market j was used as an indicator of

market access from that point as a function of the

line-ar distance dij.The distance decay exponent b was set to

equal 1, representing the baseline situation of ‘current’

market access Because there is no sharp boundary

be-tween access or no access to market j fuzzy set logic

was applied to assign any given supply point i a

prob-ability between 0 and 1 of proximity to urban markets

(Pu) and rural markets (Pr) The probability of market

access from each point was modelled as a J-shaped

membership function under the assumption that 0

(no market access) is only reached at in¢nity The

inter-action potential Vibetween any point i and market j

was calculated as a function of the linear distance

(dij)1 for urban (POPDj45000) and rural markets

(POPDjin the range 500^5000) separately (Table 2)

The membership function approached full market

ac-cess (P 51) atVi5 20 and poor market access atVi5 5

The two models of interaction potential between

urban (u) and rural (r) markets and their respective

hinterland were used as conditional probability input

images Puand Prin the Bayesian probability model

The output consists of posterior probabilities of

ur-ban vs rural market access Pu 0.5 assumes urban

market access and Puo0.5, Pr40.5 assumes rural

market access At equal posterior probabilities

(P 5 0.5) of access to urban markets u and rural

mar-kets r, it was assumed that access to urban marmar-kets is

preferred over access to rural markets.‘No’ market

ac-cess is approached at Pro0.5

Di¡erential market access of poor producers

The Bayesian probability analysis was repeated in

or-der to simulate three alternative scenarios of access

to markets In order to simulate di¡erential marketaccess of poor people, the distance exponent b, whichdetermines the structure of the distance decay, wasadjusted according to three di¡erent assumptionsabout the e¡ect that marketing constraints wouldhave on their access The ¢rst assumption was thatmarket access is relatively more constrained for poorproducers and consumers than for the better-o¡ Un-der this assumption, the distance exponent b is set to1.5 and the interaction potential Vibecomes a func-tion of POPDi/(dij)1.5 The second assumption exploredaccess to urban markets as a relatively higher con-straint to poor producers and consumers, while ac-cess to rural markets would be less of a constraint.The assumption here was that rural markets aremore numerous and widespread, and thus easier toaccess for poor and better-o¡ people alike The dis-tance exponent b was set to 1.5 for urban markets,and lowered to 0.75 for rural markets under this as-sumption The third assumption was one of severeconstraints resulting in high transaction costs to thepoor, simulated by a distance exponent b equalling 2.The absolute number of poor people (producers andconsumers) with no access, rural market access orurban market access under the four di¡erent scenar-ios (including the baseline assumption of a linear dis-tance decay with no di¡erential access between thepoor and the better-o¡) was estimated using theIDRISI extraction tool for summary statistics On thebasis of these estimates and empirical evidence on

¢sh marketing in Cambodia, the implications of ferential market access for pro-poor aquaculture de-velopment in Cambodia are discussed

dif-ResultsPopulation and poverty distributionThe Cambodian National Institute of Statistics cal-culated the country’s total national population at

12132172 in1998 The sum of redistributed populationcounts for that year (Fig 1a) results in a population

of 10 881878, thus representing an underestimate ofabout 10% The highest urban population density is inthe capital, Phnom Penh (P), and the highest rural po-pulation densities are found in provinces in the south-east of the country, in the vicinity of the capital Thisappears to con¢rm the hypothesis that rural popula-tion tends to concentrate in the proximity of ruraltowns and major urban centres The distribution ofpoor population re£ects the overall population distri-bution over Cambodia, but poverty is particularly high

Table 2 Market access from supply point i (V i  20, exp.

in Siem Reap province (SR) north of Tonle Sap lake (TS),

with a second concentration of poor people in the

south-eastern provinces, particularly Prey Veng

vince (PV) (Fig 2a) Poverty in the south-eastern

pro-vinces is predominantly rural (Fig 2b) but the context

is one of high rural population density (Fig.1b) and

re-latively close proximity to Phnom Penh (P) The Capital

is relatively a¥uent, with a poverty headcount index of

ca 12% (Ministry of Planning, Royal Government of

Combodia, United Nations World Food Programme &

United Nations Development Program 2001)

Market access

Under the baseline scenario (a), estimated as a

func-tion of linear distance, urban market access (Vu) is

generally good in Kandal province (KA) This vince lies well within the sphere of in£uence of theCapital, Phnom Penh (P) Isolated spots of good ur-ban market access are found around some larger pro-vincial centres in the rest of the country Overall,urban market access is limited, with only 3% of thecountry having good access under the baseline sce-nario (Table 3; Fig 3a) Under scenario (b), a higheroverall distance constraint for poor producers andconsumers results in a decrease in area of urban andrural market access (Fig 3b) Under scenario (c),where rural market access is relatively less con-strained for the poor than urban market access, some

pro-of the more remote areas in the northeast pro-of thecountry shift from ‘no’ access to rural market access(Fig 3c) Under this assumption, about two-thirds of

KA TK

SR

TS

SV Poor inh_ / km2

Figure 2 (a) Natural logarithm of poverty density [ln (pov) km 2]; (b) poverty density redistributed into discrete classes

KA, Kandal province; P, Phnom Penh (capital); PV, Prey Veng province; SR, Siem Reap province; SV, Svay Rieng province;TK,Takeo province; TS,Tonle Sap lake

Table 3 Area (%) of accessibility and number of poor (#) having ‘no’ access, rural market access or urban market access under di¡erent scenarios